diff --git "a/qwen25_0.5b_fmft_official_5e-05/complete_results.json" "b/qwen25_0.5b_fmft_official_5e-05/complete_results.json"
new file mode 100644--- /dev/null
+++ "b/qwen25_0.5b_fmft_official_5e-05/complete_results.json"
@@ -0,0 +1,40694 @@
+{
+  "test_bleu_4": 0.018908596504347546,
+  "test_exact_match": 9.833808634083981e-05,
+  "test_rouge": {
+    "rouge1": 0.13859125771277808,
+    "rouge2": 0.01957766609514052,
+    "rougeL": 0.11310167198360728,
+    "rougeLsum": 0.11317235766398888
+  },
+  "test_bertscore": {
+    "precision": 0.8390011585664932,
+    "recall": 0.84534054110683,
+    "f1": 0.8417594840843331
+  },
+  "test_meteor": 0.10851142872748486,
+  "raw_data": [
+    {
+      "prediction": "Is there a reason for this change? I think it should be `transport=grpc,java` or `transport=grpc,java,transport=grpc`",
+      "reference": "can we also test for `transport=rest`?"
+    },
+    {
+      "prediction": "I think this should be done in the constructor of the `SolrBackendFactory` and not in the `AbstractSolrBackendFactory` class. The `AbstractSolrBackendFactory` is only used by the `SolrBackend`.",
+      "reference": "If record_batch_size is not set in config.ini, this code will trigger a notice about an undefined value. I would suggest either wrapping the setPageSize() call in an `if (!empty(...)) {` check, or else providing a default value in the set call (i.e. `$config->Index->record_batch_size ?? 100`)."
+    },
+    {
+      "prediction": "I don't think this is the right place to set the SEGMENT_KEY. I think it should be in the `analytics.js` file in `app/assets/javascripts/` folder.",
+      "reference": "I didn't realize we were hardcoding this, thanks for moving it to an env value."
+    },
+    {
+      "prediction": "I think this should be `typeof file.data.size !== 'undefined' ? html`<div class=\"UppyDashboardItem-statusSize\">${prettyBytes(file.data.size)}</div>` : ''`",
+      "reference": "We are trying to support IE 10-11, so we'll need a polyfill for this one, I think."
+    },
+    {
+      "prediction": "nit: `reset_success` -> `reset_success_message`",
+      "reference": "It looks like there's a new `isNavigatingTo( url )` selector for this very purpose so let's use this here instead. This way we just need to use the one selector rather than two. Let's assign that to a similar-named variable here (e.g. `isNavigatingToPostResetURL`) rather than the prop it's used with."
+    },
+    {
+      "prediction": "I think this should be `const MatrixType&` instead of `const AbsMat&`.",
+      "reference": "I think im2col should only accommodate CPUMat."
+    },
+    {
+      "prediction": "I think we should move this to the top of the file.",
+      "reference": "alpha sort the imports"
+    },
+    {
+      "prediction": "I think this should be `sizeof(sysfs_path) - 1` instead of `sizeof(sysfs_path) + sizeof(\"DEVICEID_PATH\") - 1`",
+      "reference": "Can you explain why is this necessary? Is `snprintf()` with four integer arguments unsafe?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work on Windows 10 21H2. It's also the only way I've been able to get it to work on Ubuntu 20.04. I'm not sure if there is a better way to do this, but it's the only way I've been able to get it to work on Windows 8.1 21H2. I'm not sure if there is a better way to do this, but it's the only way I've been able to get it to work on Ubuntu 18.04. I'm not sure if there is a better way to do this, but it's the only way I've been able to get it to work on Windows 7 64bit. I'm not sure if there is a better way to do this, but it's the only way I've been able to get it to work on Ubuntu 18.04. I'm not sure if there is a better way to do this, but it's the only way I've been able to get it to work on Windows 8.1 21H2",
+      "reference": "How likely would it be that the keyboard layout for the NVDA main thread differs from the keyboard layout of the currently focused app?"
+    },
+    {
+      "prediction": "I think this should be `service = getDefaultService() || service.start();`",
+      "reference": "Would you mind removing the default parameter above? (I doubt I'll ever use defaults again since you still have to protect against callers explicitly passing `null` or `undefined`)"
+    },
+    {
+      "prediction": "I think this should be `setParameter('text', keys.join('')).`",
+      "reference": "Also update line 2205 below"
+    },
+    {
+      "prediction": "I think this should be `volumeConfig.GetSharedStorageMountValue()`",
+      "reference": "n_: It is a good practice to name the variable to indicate what they contain. In this case since `shared` is supposed to have boolean, calling it: `isShared` can help in the readability of the code."
+    },
+    {
+      "prediction": "I think we can remove the `@param` here, since we don't need to pass `column` to `freezeColumn()`.",
+      "reference": "I suppose it's a false-positive error. Maybe there is a way to configure the `eslint-*` package to accept `i.e. .... lower case` syntax"
+    },
+    {
+      "prediction": "I'm not sure why this is necessary, but it seems like it should be `activity.ParentSpanId.ToHexString() == \"0000000000000000\"`.",
+      "reference": "this maynot be a perf issue, if ToHexString() is not actually allocating a string, but returns the caches string value. to be confirmed."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `if len(updates) > 0` block above.",
+      "reference": "`zap.String(\"image-provider\", provider.Name())` should be in a same line."
+    },
+    {
+      "prediction": "I think we can remove the `|| 300` here.",
+      "reference": "Why this deletion?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it would be better to use `assertIn` to check that the cache control header is present in the remote metadata.",
+      "reference": "This change looks unrelated to the CL description?"
+    },
+    {
+      "prediction": "I think this should be `DEDUP_RESULT_CHECK` instead of `DEDUP_RESUTL_CHECK`",
+      "reference": "I'm not certain that `res`+`utl` isn't a thing."
+    },
+    {
+      "prediction": "I think this should be `Status.EXECUTION_STOPPED, Status.SUCCEEDED, Status.KILLED, Status.FAILED`",
+      "reference": "This method can cause bugs. Why don't you use list of finalize statuses and then use join by comma and that will give you correct value. Use Apache Commons Lang library List<Integer> list = Arrays.asList(1,2,3,4,5,6,7); String str = org.apache.commons.lang.StringUtils.join(list, \",\"); You can use it at multiple places shown in this diff."
+    },
+    {
+      "prediction": "This should be `return dev->type == WLR_INPUT_DEVICE_TOUCHSCREEN ? dev : NULL;`",
+      "reference": "Merge with next condition"
+    },
+    {
+      "prediction": "This is a bit of a nitpick, but I think it would be better to call this `DeleteStack` instead of `DeleteTask`.",
+      "reference": "Maybe add it when it is used."
+    },
+    {
+      "prediction": "I think this should be `scanners` instead of `non_exist_scanner`",
+      "reference": "More clear naming: NONEXISTENT_ENABLED"
+    },
+    {
+      "prediction": "I think this should be `to.equals(Types.LongType.get())`",
+      "reference": "why is this change necessary?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be moved to `org.apache.gobblin.util.SchemaUtils#getTypeFullName`",
+      "reference": "make a SchemaTypeNameConverter.getTypeNameForEmptyType() and call that here."
+    },
+    {
+      "prediction": "I think this should be `$this->locale = $request->getSession() ? $request->getSession()->getLang()->getLocale() : Lang::getDefaultLanguage();`",
+      "reference": "Use braces on your conditional structures please"
+    },
+    {
+      "prediction": "This should be `true` by default.",
+      "reference": "Great God, I have a question. This \".nebula_history\" file is used to save history commands. Is there a file size limit? When the file is large, does it take a long time to start initialization (loadHistory)? How did you think about this? Thx."
+    },
+    {
+      "prediction": "I don't think this is the right way to check for parentheses. The `isType` check should be done at the top of the function, before the `do-while` loop. The `do-while` loop should only be entered if `isType (token, TOKEN_OPEN_PAREN)` evaluates to `true`.",
+      "reference": "Isn't this the same check as two lines above?"
+    },
+    {
+      "prediction": "I think this should be named `CONFIG_PATH` instead of `CONFIG_FILE`.",
+      "reference": "Don't redefine, just import the existing one in `cmdline.py`. :)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "why not call `respond_to?(\"#{name}=\")` from the `if` line?"
+    },
+    {
+      "prediction": "I think this should be `workingDir` instead of `workspaceRootDirectory`",
+      "reference": "This was properly set above as `<target>.runfiles/<workspace_name>` (with a fallback to workspace root dir) but never used past this line."
+    },
+    {
+      "prediction": "I don't think we need this attribute. We can just use `Page` as the name of the view model.",
+      "reference": "Are all these namespaces required for this attribute?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Given this may be used and printed in other ways, isn't `debug` more appropriate?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think we should move this to `store.go` and then call it from here.",
+      "reference": "Looks like this is unused? Did you mean to call it when clearing the block journal?"
+    },
+    {
+      "prediction": "I think it would be better to pass the `alerters` to the `ExecutorManager` constructor instead of creating a new one here.",
+      "reference": "why named alters? how about this.alerts = alerts?"
+    },
+    {
+      "prediction": "I think we can remove the `Mirror` field from the `StreamInfo` struct.",
+      "reference": "not sure if this has to be here or not tbh"
+    },
+    {
+      "prediction": "I think we should be able to get rid of this line.",
+      "reference": "There's a better fix for this in #1741, which should replace this."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Use swarm.Address not []byte as type, and construct it from actual byteslice or use swarm.NewAddress if it is constructed from hex-encoded string."
+    },
+    {
+      "prediction": "I think this should be `require.NoError(t, fs.RemoveAll(dir))`",
+      "reference": "It would be worth factoring this out to a function, but you don't have to do that here."
+    },
+    {
+      "prediction": "I don't think we should have a separate method for linkifying the uri. I think we should have a single method that returns the linkified version of the uri.",
+      "reference": "There's no need for `final` in interfaces."
+    },
+    {
+      "prediction": "I think we should use `Span.SetTag` instead of `Span.SetTag` here. The reason is that `Span.SetTag` takes a `string` tag name and a `string` tag value, whereas `Span.SetTag` takes a `string[]` tag name and a `string[]` tag value. In this case, I think we should use `Span.SetTag` instead of `Span.SetTag`.",
+      "reference": "Isn't this case handled as a custom tag in `Span.SetTag()` below? This switch is only for special tags that actually set `Span` properties."
+    },
+    {
+      "prediction": "I think this line should be indented 2 spaces.",
+      "reference": "This shouldn't be changed."
+    },
+    {
+      "prediction": "I don't think this is the right place for this file.",
+      "reference": "This one is important `2: I102 copyright year is outdated, expected 2014 but got 2012`"
+    },
+    {
+      "prediction": "nit: remove unused import `java.util.function.BiConsumer;`",
+      "reference": "Are these imports needed (BiConsumer and Consumer)? If they are unused imports, then precommit will fail."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this job. I think it should go in `lib/rails/mailers/destroyer_mailer.rb`",
+      "reference": "Why are we catching and logging exceptions? By doing that we make it look like the job has succeeded and it will be removed from the queue - if we didn't do that then it would stay queued..."
+    },
+    {
+      "prediction": "I think this should be `null` instead of `[]`.",
+      "reference": "If there are `@param` tags in docblock, there should be `@return` tag also. (applies for a whole file)"
+    },
+    {
+      "prediction": "I think it would be better to call this `VALIDATOR` instead of `VALIDATORS`.",
+      "reference": "Is the concept of validators exclusive to IBFT2? I wonder if this category should be more explicitly linked to IBFT2."
+    },
+    {
+      "prediction": "I think we can remove this import as it is not used in this class.",
+      "reference": "Spotless is configured to reject star imports. Please replace with explicit imports."
+    },
+    {
+      "prediction": "I think this should be `std::numeric_limits<uint64_t>::max()` instead of `std::numeric_limits<uint32_t>::max()`",
+      "reference": "Remind me why you are using CUDA vocabulary when naming these variables."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. Feeds are fetched from the DB, and then the feed counters are updated in the DB. I think it would make more sense to update the feed counters after the feeds have been fetched from the DB, rather than after the feeds have been fetched and the counters updated in the DB. This way, the counters will always be updated correctly, regardless of how many feeds are fetched from the DB.",
+      "reference": "Please turn the iteration order around (instead of `0...size` to `size...0`). The reason is that this sometimes skips indices when removing an item. You can therefore end up with feeds that have counter 0 and are still displayed."
+    },
+    {
+      "prediction": "I think this should be `if solid_subset is not None` instead of `if not solid_subset`",
+      "reference": "this is a behavior change. solid_subset=[] represents an empty pipeline where as solid_subset=None is the full pipeline"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this warning. I think it should be in the `CustomSettings.Validate` method, since that is the place where we validate the settings.",
+      "reference": "this is a breaking change, any specific reason you want to do this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this.",
+      "reference": "Blocking: we still want read write separation. Node functions should absolutely not write to the chain store and the interface should reflect that. Only the syncer should have this capability in production code. It should be no problem to keep casting chainForTest to a read-write interface, or doing other function decomposition and automatic casting tricks."
+    },
+    {
+      "prediction": "I think this should be `1<<30 - 1` instead of `1<<31 - 1`",
+      "reference": "The PR description says the lowest on-demand request priority is `2^30`. Why the `-1` here?"
+    },
+    {
+      "prediction": "This function should be moved to the bottom of the file.",
+      "reference": "checking health again? I think it is not necessary."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "Need an `if err == nil {break}` above this line so that we don't log/sleep if the retry succeeds."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it should be in `lib/promise.js`",
+      "reference": "to reiterate my point above, this class is not the actual provided Promise, but rather something a user can provide a Promise to/with. I think a name like `PromiseProvider` is more appropriate."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "I guess we can only call this from a post-build function, but we might need to check that this target is to be built? I guess target A depends on B which has a post build. We `plz build :B` which adds C as a dep of A. Won't we queue C to be built even though it only needs to build if A needs to be built? That's kinda why I approached this the way I did. I trust you've thought about this but would be good to understand this a bit better."
+    },
+    {
+      "prediction": "I think this should be `if arch.FREEBSD or arch.DARWIN and not (pks.__class__.__name__ == 'StreamSocket')`",
+      "reference": "You should use `isinstance()` instead of comparing the class name to a string."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "Why should audio and video elements always return `false` for `isHidden`?"
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "This obviously won't work for all the other platforms."
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think it would be better to use a `Map<String, Object> state = new HashMap<String, Object>();` in the constructor of the `InjectableStepsFactory`.",
+      "reference": "Isn't he diamond operator Java7+?"
+    },
+    {
+      "prediction": "I think this should be `throw new IllegalArgumentException(message, argName, number);` instead of `throw new IllegalArgumentException(message, argName, number);`",
+      "reference": "Prefer adding a second `positive(String, double)` that delegates down to this three-param version. Using `null` in code is generally Not A Great Idea, and it looks ugly."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `plan/expr`.",
+      "reference": "This pattern, where we add a new parameter without using it, often indicates to me that we've got a leaky interface or abstraction. I see this pattern _a lot_ in this patch, so wondering you have thoughts about it."
+    },
+    {
+      "prediction": "This should be `if err = os.MkdirAll(dir, 0755); err != nil {`",
+      "reference": "This seems like an improved pattern :)"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "`open_id_authentication` is no longer the name of the function, and it's not OpenID specific"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add this, but it's the only way I could find to make it work.",
+      "reference": "This didn't require corresponding changes to the test case(s)?"
+    },
+    {
+      "prediction": "I think it would be better to call this `Env` instead of `EnvFile`.",
+      "reference": "A question! I think `string` totally works, but what do you think of `*string` for consistency?"
+    },
+    {
+      "prediction": "I think this should be `aws.cluster.sigs.k8s.io/infrastructure-ready` and `aws.cluster.sigs.k8s.io/control-plane-ready` instead of `aws.cluster.sigs.k8s.io/infrastructure-ready` and `aws.cluster.sigs.k8s.io/control-plane-ready`",
+      "reference": "Should this be: `aws.infrastructure.cluster.sigs.k8s.io` instead?"
+    },
+    {
+      "prediction": "I don't think you need to change this line.",
+      "reference": "Not clear on why this change is correct."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it should go in `IndexIO`.",
+      "reference": "Hi Simon. Whether this should return true depends on what byte buffers are used? The same applies to ByteBufferIndexInput, actually... I don't think you can generally tell whether the ByteBuffers the input operates on come from a mmap call or from somewhere else (even direct buffers don't have to be a result of mmap)."
+    },
+    {
+      "prediction": "I think this should be 10px.",
+      "reference": "All stories in this file also need to have the default padding."
+    },
+    {
+      "prediction": "I think this should be `parent` instead of `sibling`.",
+      "reference": "I don't think these asserts are necessary, because `subsurface_find_sibling` already searches in the parent. Or am I missing something?"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "Should this be `number` instead of `id`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should go in the `TrafodionTable` class, not the `BindWA` class. The `TrafodionTable` class is the base class for all Trafodion tables. The `BindWA` class is a subclass of the `TrafodionTable` class. The `TrafodionTable` class has an `isExternalTable()` method that can be used to check if an object is an external Trafodion table. The `BindWA` class has an `isExternalTable()` method that can be used to check if an object is an external BindWA table. The `TrafodionTable` class also has a `allowExternalTables()` method that can be used to check if an object is allowed to be an external Trafodion table. The `BindWA` class also has a `allowExternalTables()` method that can be used to check if an object is allowed to be an external BindWA table.",
+      "reference": "I have forgotten why we thought this restriction is necessary. A user registers a Hive/HBase table with us but is not able to use the table with the registered name in DML. Will UPDATE STATs will be on the original name too (and update stats code will find out registered name and use it for Table_uid?). This was the registration concept is almost like an internal implementation idea, at some point could even be done by some utility program for all existing Hive/hBase tables, and the user need not be aware of it? I am struggling with a user would want to register a table, if they cannot use it by that name."
+    },
+    {
+      "prediction": "I don't think this is needed. The app2sim_alloc_rx and app2sim_dealloc_rx variables are only used by the app2sim module. The sim2app_alloc_tx and sim2app_dealloc_tx variables are only used by the sim2app module. The app2sim_mmioreq_rx and app2sim_mmiorsp_tx variables are only used by the app2sim module. The app2sim_umsg_rx and app2sim_portctrl_rsp_tx variables are only used by the app2sim module. The sim2app_dealloc_tx and sim2app_portctrl_rsp_tx variables are only used by the sim2app module. The intr_event_fds[MAX_USR_INTRS] array is only used by the app2sim module.",
+      "reference": "Should most of these non-globals be static?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Please do not approve PRs without tests!!!!"
+    },
+    {
+      "prediction": "I think this should be `win_id=win_id` instead of `win_id=win_id`",
+      "reference": "Now that you handle this in `webenginetab.py` you can undo all the changes in this file."
+    },
+    {
+      "prediction": "Why not just `throw new RuntimeException(e);`?",
+      "reference": "I think we should log the exception to a log file instead of printing it. The current behavior (not catching the exception) should lead to an automatic print of the exception to the terminal/console."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "These changes to the resource helpers have some more far-reaching implications as they also affect calls made in hive-operator, I believe. It's probably OK still, though. Ideally, the resource helper would not be the one doing the logging, since it cannot know the severity, but that is well beyond something that we should tackle for this PR. I would rather have the hive-operator log warnings than have the syncsetinstance controller log errors."
+    },
+    {
+      "prediction": "I think it should be `null, null, 'Exports'`",
+      "reference": "do you really want to change the behavior on foreign key ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the state_trans is set, it means that the event is going to be processed. If the event is going to be processed, it means that the event is going to be published. If the event is going to be published, it means that the event is going to be sent to the client. If the event is going to be sent to the client, it means that the event is going to be processed. If the event is going to be processed, it means that the event is going to be published. If the event is going to be published, it means that the event is going to be sent to the client. If the event is going to be sent to the client, it means that the event is going to be processed. If the event is going to be processed, it means that the event is going to be published. If the event is going to be published, it means that the event is going to be sent to the client. If the event is going to be sent to the client, it means that the event is going to be processed. If the event is going to be processed, it means that the event is going to be published. If the event is going to be published",
+      "reference": "Looks like my bad. Thanks for fixing!"
+    },
+    {
+      "prediction": "I think this should be `--replace-variables` instead of `--no-replace-variables`.",
+      "reference": "I think \"Don't replace variables ...\" would be cleaner."
+    },
+    {
+      "prediction": "I think this should be `objects=['3d', 'polar']` instead of `objects=['3d', 'polar', None]`",
+      "reference": "Do you validate the possible strings? I've not read the code below but we should make sure if a string is supplied it is validate..."
+    },
+    {
+      "prediction": "I think this should be `dataDiskIndex + 1` instead of `dataDiskIndex`",
+      "reference": "I'd recommend using generateDataDiskName here as well -- might as well benefit from the safety that it gives to protect yourself from future changes to w.ID()."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this method. I think it would be better to put it in the `ContentFileBuilder` interface.",
+      "reference": "qq: do we want to include anything in the name to indicate that it is a position in the manifest?"
+    },
+    {
+      "prediction": "Shouldn't this be `Log.d(LOGTAG, \"Compositor resume begin\");` and `Log.d(LOGTAG, \"Compositor resume end\");` ?",
+      "reference": "These logs should probably use the `LOGTAG`"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "From this, it was clear, that we are using our version of wireguard-go. And now it looks like we are using original packages which is confusing."
+    },
+    {
+      "prediction": "I think we can remove the `Message` property from the `ResultWrapper<T>`.",
+      "reference": "Any particular reason for this? This potentially will make harder to investigate users issues"
+    },
+    {
+      "prediction": "I think this should be `||` instead of `||=`.",
+      "reference": "Do you also need to check if Blockly.localConnection_ is non-null?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good or a bad idea. I think it's fine, but I'm not sure what the motivation is for it.",
+      "reference": "Not really a container, and it's not just about non-config data, not sure of a better description, @myronmarston ?"
+    },
+    {
+      "prediction": "nit: I think it would be better to have a separate file for this. It's not really part of this PR, but it's something we need to be able to do in the future.",
+      "reference": "The desktop app will load this fine. on iOS I get `TypeError: undefined is not an object (evaluating '_$$_REQUIRE(_dependencyMap[8], \"stream\").Readable.Readable')` if I change it to `const Readable = require('stream').Readable;` or `const { Readable } = require('stream');` I get undefined errors from the stream on iOS: `[TypeError: stream.on is not a function. (In 'stream.on(\"data\", function (chunk) { return chunks.push(chunk); })', 'stream.on' is undefined)]` What am I missing?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make a copy of `ch.Data()` and store it in `m.store[ch.Address().String()]` instead of `m.store[ch.Address().String()] = b`.",
+      "reference": "this is needed since the chunk feeder shares memory across calls to the pipeline. this is in order to avoid multiple allocations. this change mimics the behavior of shed and localstore, and copies the data from the call into the in-memory store"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "wut. We can do this at the beginning of tests if we want right? Why are we depending on init?"
+    },
+    {
+      "prediction": "Why do we need to unlock `srv.mu` here?",
+      "reference": "Same trick here IMO."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "Should be `@origin = element || origin || POINTER`"
+    },
+    {
+      "prediction": "I think this should be `if len(args) != 0 { bulkCstorSPCUpgrade(cmd, args) }`",
+      "reference": "this example needs a fix?"
+    },
+    {
+      "prediction": "Shouldn't this be `wlr_xdg_toplevel_destroy`?",
+      "reference": "Is it safe to assume this surface is a toplevel?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. It seems like it should be in `pj_internal.h` but I don't know if that's the right file to put it.",
+      "reference": "we don't need that function"
+    },
+    {
+      "prediction": "I think it would be better to have a `WINDOWS10` class extending `WIN10` so that we don't have to repeat `WIN10` everywhere.",
+      "reference": "we'll also have to think about adding \"Windows Server 2014\" or whatever they come up with later."
+    },
+    {
+      "prediction": "I think this method should be renamed to `fetchActiveFlowByExecId(int execId) throws ExecutorManagerException;`",
+      "reference": "curious do we have an API to fetch an inactive flow?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with at the time of writing.",
+      "reference": "This branch is not needed anymore because every `vnode` that passes through `h` in compat or core will have the `$$typeof`-property now ."
+    },
+    {
+      "prediction": "I think this should be called `shouldContainAriaHiddenValue` instead of `ariaHiddenValue`",
+      "reference": "Did you mean to use `hasAttribute`? You shouldn't. This does not improve performance and it messes with the readability."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Please import modules and not classes (except for Qt stuff)."
+    },
+    {
+      "prediction": "I don't think we need this method. `super.setImage(image);` is enough to set the image on the `DummyJavaNode` object.",
+      "reference": "Hm... why do we have a \"DummyJavaNode\" in src/main? Looks like, this should have gone into src/test...."
+    },
+    {
+      "prediction": "Can you add a license header here?",
+      "reference": "This seems meaningless. If we cannot add any info let us remove this."
+    },
+    {
+      "prediction": "I think this should be `GetPortsForContainer`",
+      "reference": "Can you rename `Ports` to `PortsUnsafe`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this method. I think it's a bit confusing because it's not just a return to, but a return to a return to. Maybe something like `make_return_to_path` or `make_return_to_name` would be better.",
+      "reference": "This feels like it could be moved out to your first gem!"
+    },
+    {
+      "prediction": "I'm not sure this is the right abstraction for this interface. It seems like we should be able to just return a `CompletableFuture<HttpRequest>` instead of an `Optional<HttpRequest>`. This would allow us to remove the need for the `poll` method and the `clear` method, as well as the need for the `offerFirst` and `offerLast` methods.",
+      "reference": "The expected behaviour for `poll` would be to match what `Deque` does, which is: > Retrieves and removes the head of the queue represented by this deque As such, you shouldn't need the `RequestId` here."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `utils.js` since it's only used by `utils.js`.",
+      "reference": "This is super minor, but it'd be nice if this iterated over an arbitrary number of array inputs."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. The `readlink` function will return -1 on error, and we don't want to terminate the process when that happens. Instead, we should check the return value of `readlink`, and if it's non-zero, we should print an error message and return -1.",
+      "reference": "What is the problem here? The commit message has strange wording \"readlink is defined with restrict pointers\": do you mean \"restricted\"? Is that a well-defined language term? I'm not sure what it means."
+    },
+    {
+      "prediction": "I think this should be `ImmutableSet.of()` instead of `ImmutableList.of()`.",
+      "reference": "Why are we not wrapping here?"
+    },
+    {
+      "prediction": "Why do you need this `@var`?",
+      "reference": "I though that all data objects are being unified in the fashion where all default values are initialized in constructor, are not they?"
+    },
+    {
+      "prediction": "08929000000000000\",\n \t\"libflux/flux-core/src/ast/struct.rs\":                                                           \"92929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929292929",
+      "reference": "Does this file actually need to be committed to git? Having to run `make generate` is rather tedious on each PR and it conflicts easily."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "We don't need this anymore, as Google's library handles various auth mechanisms for us. We'll be less explicit about why we're failing but we gain support for several authentication methods- swings and roundabouts!"
+    },
+    {
+      "prediction": "I think this should default to `/var/lib/docker/volumes/plugins` instead of `/run/docker/plugins`",
+      "reference": "No need to speficy attribute here. Moreover in yaml it is default to use lowercase names."
+    },
+    {
+      "prediction": "I think we can remove `using System.Diagnostics.CodeAnalysis;` from `Microsoft.DotNet.Build.CloudTestTasks` as well.",
+      "reference": "In general I'm a trying to understand the reason behind this PR. Looks like if a blob (name) already exists we check if the contents are identical? In what scenarios this is not the case? When we want to publish a package/asset that has changed but still we want to use the same version?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to use `requests` to make the requests.",
+      "reference": "Can rename `req` to be `opener`, since that is the object, and it wraps `req` internally?"
+    },
+    {
+      "prediction": "I don't think it's a good idea to put breakpoint in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea to have it in textList. It's not a good idea",
+      "reference": "@MichaelDCurran, thoughts on adding a breakpoint format field attribute? It seems almost app specific, but I guess it does seem odd abusing line-prefix. Is there any more generic concept here?"
+    },
+    {
+      "prediction": "I think this should be `self._execute(Command.GET_ELEMENT_RECT, return_base64=True)`",
+      "reference": "This would return a tuple of two dictionaries. You need to combine them and return a dictionary"
+    },
+    {
+      "prediction": "nit: I think it would be better to move these imports to the top of the file.",
+      "reference": "What do you think of moving the example to the help text of the prompt? and the prompt itself can be \"What is your application's GitHub repository URL?\""
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to get the timestamp from the `Span` object itself.",
+      "reference": "Break out tests for events into their own tests rather than placing them in other ones. That makes it easier for us to figure out where problems lie and to do a TDD-driven implementation over new APIs."
+    },
+    {
+      "prediction": "I think this should be `first + 8` instead of `std::distance(first, last) > 8`",
+      "reference": "Nice use of std::distance. I'm a bit worried about first += 9 though."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think we should move this to the `prometheus-operator` namespace.",
+      "reference": "Shouldn't this be `app.kubernetes.io/managed-by` as prometheus-operator manages this resource, but this resource isn't a part of prometheus-operator?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "As long as we're being extra-vigilant about NREs, should we check that `httpMethod` and `requestUri` are not null?"
+    },
+    {
+      "prediction": "I think it would be better to move this to the `ioctl` package, since it's not specific to xRc20.",
+      "reference": "No need to have `xrc20` alias"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this method. I think it should be in the `context_cls_inst` function, since it's only used by that function.",
+      "reference": "the naming convention I'm been adopting if `_type` for instances of these classes. So maybe `context_config_type` is a better name for this fn"
+    },
+    {
+      "prediction": "I think this should be `final` instead of `private`.",
+      "reference": "Don't call it GPU mining, call it `externalMining`, here and throughout."
+    },
+    {
+      "prediction": "I think it would be better to add a `default: null` value to the `imageMaxWidth`/`imageMaxHeight`/`imageQuality` options and set it to `0` by default.",
+      "reference": "`within this with` typo, should be `within this width`"
+    },
+    {
+      "prediction": "This should be `+k8s:openapi-gen=true`",
+      "reference": "how come this is false?"
+    },
+    {
+      "prediction": "I think we can remove this function now, since we don't use it anywhere else.",
+      "reference": "How about `Object.prototype.toString`? Or using the `typeof` operator?"
+    },
+    {
+      "prediction": "Should this be `context.Background()` instead of `context.Context`?",
+      "reference": "I know @jzila suggested this, but I disagree: you're really not supposed to save a context in a struct. Can you get by with just saving the `Done()` channel instead? The `ctx` should continue to be passed around explicitly."
+    },
+    {
+      "prediction": "I think this should be `!IsWindows && !PlatformApis.IsWindows && PlatformApis.IsDarwin;`",
+      "reference": "This check isn't right though."
+    },
+    {
+      "prediction": "I think this should be `public abstract String codeGenVersion();`",
+      "reference": "maybe `toolkitVersion` instead? It doesn't necessarily have to be called that in the generated code, but in the view model classes, I think it makes it clearer that it is the version of toolkit itself."
+    },
+    {
+      "prediction": "`setFileState` and `emit` don't seem to be used in this file.",
+      "reference": "the response data was added intentionally in #612, i think we could keep the `setFileState` stuff here as a special case, at least for now"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `deadline` is not guaranteed to be known at this point in the code, so we can't be sure that `time.Until(deadline)` will be >= `time.Now`.",
+      "reference": "This isn't really easy to read."
+    },
+    {
+      "prediction": "I think this should be `error` instead of `string`",
+      "reference": "This line changed due to go formatting. Format was not there earlier."
+    },
+    {
+      "prediction": "I think it would be better to have a `With` method on the `staticUpstream` struct, so that we don't need to pass in `Without` when creating a new `staticUpstream` struct.",
+      "reference": "The name \"Without\" in code is a little nebulous. Maybe something more descriptive like TrimPrefix or StripPrefix or WithoutPathPrefix or something like that. (Thoughts?)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be moved to `TestDatasetOperations#linkDataset` and `TestDatasetOperations#unlinkDataset`.",
+      "reference": "is this test in the move tests? I see what you mean then - it works, but I wonder if we won't lose track that it's being tested here."
+    },
+    {
+      "prediction": "I think this should be `[]string` instead of `map[string]string`",
+      "reference": "I'm not sure if this wanted to be `json:\"tags\"` (which I'd set in my PR) or `json:\"tag\"` that someone elses PR had set. Given that the rest of the fields had their JSON field name set to the same as the struct field name, I opted for `tags`."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "since the logic here and in `Proposal` is exactly the same, do you think it makes sense for us to include it elsewhere? I am not opposed to duplicated code when it makes sense, but the reason I first identified this was that I was looking for code in NCR::WorkOrder that was not specific to Work Orders. Seems like fiscal year logic might be helpful for reporting for other clients. In that case, we'd want this scope on those client data classes as well. Might be wrong to include this in the FiscalYear mixin, but it does seem like it should be outside of Proposal and WorkOrder. Does that make sense?"
+    },
+    {
+      "prediction": "Shouldn't this be `self.log.info('RabbitMQ connection got closed!')`?",
+      "reference": "there is no static method as a go between -- how does this work?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `lib/`",
+      "reference": "Can you please change the name of this class to be something else?"
+    },
+    {
+      "prediction": "I don't think we should change the signature of this function.",
+      "reference": "I think we might prefer generalizing the interface rather than creating an exception. Current design of `h2o_next_token` assumes the input to be a comma-separated list, and allows the caller to specify a different separator when parsing a nested list. As I understand, what we are trying to attain in this PR is to have a way of parsing a flat list of semicolon-separated list. Assuming that is the case, I think it would be better to change the API of the function to accept two separators (i.e. inner and outer), where for our existing use-cases `outer` would be `,`."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better idea.",
+      "reference": "why not recycling? I Lru cache now recycling?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Why it is now called `PaymentVersionV3`?"
+    },
+    {
+      "prediction": "I think we should use `klog.V(5).Infof` instead of `klog.V(4).Infof` here.",
+      "reference": "changed for debug purpose?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to the base class so that it can be overridden by each specific exporter.",
+      "reference": "@cijothomas I needed a way to set the Exporter.ParentProvider when the Processor's gets set. Internal to SDK though."
+    },
+    {
+      "prediction": "I think this should be `node_based_graph.GetEdgeData(eid).flags.road_classification.IsRampClass() || (from_motorway && node_based_graph.GetEdgeData(eid).flags.road_classification.IsLinkClass())` instead of `node_based_graph.GetEdgeData(eid).flags.road_classification.IsRampClass() || (from_motorway && node_based_graph.GetEdgeData(eid).flags.road_classification.IsLinkClass())`",
+      "reference": "Talking through this scenario with @srividyacb and she's wondering if there should also be a check of `(from_trunk && node_based_graph.GetEdgeData(eid).flags.road_classification.IsLinkClass())` as highways with trunk classifications can also have this scenario."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "where does `master_certname` come from?"
+    },
+    {
+      "prediction": "`__all__` is used in `__init__.py` files to indicate what members of a module are intended to be public. In this case, we don't want people to import `hrnet` or `hrnet_v2` or `hrnet_v3` or `hrnet_v4` or `hrnet_v5` or `hrnet_v6` or `hrnet_v7` or `hrnet_v8` or `hrnet_v9` or `hrnet_v10` or `hrnet_v11` or `hrnet_v12` or `hrnet_v13` or `hrnet_v14` or `hrnet_v15` or `hrnet_v16` or `hrnet_v17` or `hrnet_v18` or `hrnet_v19` or `hrnet_v20` or `hrnet_v21` or `hrnet_v22` or `hrnet_v23` or `hrnet_v24` or `hrnet_v25` or `hrnet_v26` or `hrnet_v27` or `hrnet_v28` or `hrnet_v",
+      "reference": "The line breaking is unnecessary."
+    },
+    {
+      "prediction": "I think this should be `feature_names_ = Common::Split(first_line.c_str(), \"\\t,\");`",
+      "reference": "why not enable split by space here ?"
+    },
+    {
+      "prediction": "Shouldn't this be `FPGA_INTEGRATED_DEVICEID != value`?",
+      "reference": "Shouldn't the operator be && ?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "My best guess is that you want to use host.name in these parts to get the name of the host ?"
+    },
+    {
+      "prediction": "I think we can remove this line now, since we don't persist the root nodeType when we marshal the manifest, as a result the root nodeType information is lost on Unmarshal.",
+      "reference": "IMO this edge case handling could remain here, just instead of overwriting the `n.nodeType`, the `makeEdgeType` method of `n` should be called, so `n.nodeType = nodeTypeEdge` -> `n.makeEdge()`"
+    },
+    {
+      "prediction": "I think this should be `&& showAutoDownloadReport` instead of `&& (UserPreferences.showDownloadReport() || showAutoDownloadReport)`",
+      "reference": "Just wondering... The two kinds of notifications are now quite different (Channel, text, maybe even icon). Would it make sense to extract the auto download notification to a new class instead of handling everything in the existing `notificationManager`? I have not checked if this will lead to a lot of code duplication, though."
+    },
+    {
+      "prediction": "This is a bit of a hack, but I think it's the only way to get the features of the call object without having to call `c.ic.req.RequestFeatures()`. I'm not sure if there's a better way to do this, but it seems like it should work.",
+      "reference": "Do we need to put this on the Call? We only need the API communication between the Encoding and the Transports which is currently done through the Transport.Request. The Call is used by users right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this, but it seems like it would be a good place to start.",
+      "reference": "`ctx` is unused in GenerateAPIKey"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I could come up with.",
+      "reference": "Can you judge the type of error?"
+    },
+    {
+      "prediction": "I think this should be `data[:-1] + (np.diff(data) / 2.)` instead of `data[:-1] + np.diff(data) / 2.`",
+      "reference": "Much simpler than a weird and confusing ``convolve`` call!"
+    },
+    {
+      "prediction": "I'm not sure if this is correct, but I think it should be called twice because we have two instances of the `render` method.",
+      "reference": "Huh, this seems weird to me... Surely the suspenders render was called again in order to get the new `<div>Hello second 2</div>` output... Imma take a peak at these tests to understand what's going on"
+    },
+    {
+      "prediction": "I think this should be called `success_message` instead of `success_message_error`",
+      "reference": "Since this is just a helper function to create text, perhaps this could live in a helper? I noticed that we have a few other error/message creators in the application controller as well, but I think this type of function should be re-factored to be part of a helper."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. It's not a template, it's a single line message.",
+      "reference": "Will this mess up parsing the messages by other editors? (vscode, vim ...)"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only way I could get this to work.",
+      "reference": "Why can the pool be null?"
+    },
+    {
+      "prediction": "I think this is the only place where `ia2Web.Ia2Web` is used, so I think it should be `ia2Web` instead of `ToggleButton`.",
+      "reference": "It might be good to have a note here: > work-around for issue #7562 allowing us to differentiate presentational lists from interactive lists (such as of size greater 1 and ARIA list boxes). In firefox, this is possible by the presence of a read-only state, even in content editable."
+    },
+    {
+      "prediction": "I don't think this is the right place for this change. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used to represent a node in a directed graph. The `DigraphNode` class is used",
+      "reference": "This set of changes is the one part of this that I'm not completely sure is valid. Can someone look over this to make sure I got the E's right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `frontier_pulls` is a `std::list` and `push_back` is a `std::insert`. The latter modifies the list in-place, so the `frontier_pulls` size doesn't change. I think you need to use `std::unique_ptr` and `emplace_back`.",
+      "reference": "There doesn't seem to be a reason to copy this here."
+    },
+    {
+      "prediction": "I think this file should be moved to `h2o/common/include/h2o/ebpf.h`",
+      "reference": "We've avoided including h2o headers in BPF programs because it's a runtime dependency. However, IIRC, this is because h2olog was maintained in the separate repository so that h2olog did not know where h2o was installed. Now h2olog can use `H2O_ROOT`, we should add it to BCC's `cflags` in order to include h2o headers in BPF programs."
+    },
+    {
+      "prediction": "I think we should rename this to `AlertQueryURL` to be consistent with the rest of the ruler.",
+      "reference": "I think the docstring here should include the CLI arg `--alert.query-url` just to make it clear to users which setting this uses."
+    },
+    {
+      "prediction": "I think we can remove the `externalNodeCIDRs` field from the `ipipManager` struct.",
+      "reference": "Same change just above?"
+    },
+    {
+      "prediction": "I think this test is not necessary, as we have a test in `syncTables` that checks if notes are synced correctly.",
+      "reference": "Could you check the result content rather than just the number of search results please? For example with this test if the search engine suddenly starts returning \"latin n\" for both queries, we won't know about it."
+    },
+    {
+      "prediction": "I think this should be `agentlessID` instead of `agentID`.",
+      "reference": "nit: I think that `workloadID` should suffice for this test... that's what we'd be issuing anyways"
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think it would be better to catch `NoClassDefFoundError` instead of `ClassNotFoundException` as it is more specific.",
+      "reference": "We should have a rule to detect identical catch branches"
+    },
+    {
+      "prediction": "I think we can remove the `RetryAttempts`, `RetryDelayInSeconds`, and `MaxClients` properties from this class.",
+      "reference": "remove these from the targets file."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm also not sure about the other changes in this file.",
+      "reference": "now that we have the signature in the hive messages, it might be that this must be drastically reduced, since there are limits on the protobuf reader/writers i believe"
+    },
+    {
+      "prediction": "I think we can remove `google-auth==1.6.2` now that we are using `google-auth-httplib2==0.0.3`",
+      "reference": "I recommend that we move this to be optional, as other users might not need it. Can you look at `OPTIONAL_PACKAGES` section, around line 68?"
+    },
+    {
+      "prediction": "I don't think you need to import `*`. You can just use `Collections.emptyIterator()`, `Collections.emptySet()`, `Collections.emptyMap()`, etc.",
+      "reference": "I usually set idea to never use wildcard import such as `import java.util.*;` Now i use setting from javaslang standard."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Where is host['ips'] coming from?"
+    },
+    {
+      "prediction": "I think we can remove the `FuzzUIDMap` function now.",
+      "reference": "oh! missed a `:` here; let me fix that; also can get rid of the intermediate variable"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [122/120]",
+      "reference": "Is this necessary?"
+    },
+    {
+      "prediction": "I think we can remove the `header` and `footer` components as they are no longer used.",
+      "reference": "I'd create a different method for each render section."
+    },
+    {
+      "prediction": "nit: can we rename this to `wasReverted` or `wasRevoked` or something similar?",
+      "reference": "Instead of storing the revert flag in the tracer is it possible to use org.hyperledger.besu.ethereum.mainnet.TransactionProcessor.Result#getRevertReason? (via org.hyperledger.besu.ethereum.transaction.TransactionSimulatorResult#getResult)? If a TX reverts without a reason do we get an empty revert reason or a revert reason with zero bytes?"
+    },
+    {
+      "prediction": "I think we can remove the `SparkOrcReader::new` method and just use `readOrcSchema` in the `SparkOrcReader` constructor.",
+      "reference": "I think this should test with and without container reuse if that is implemented in this PR. Probably just make this test parameterized."
+    },
+    {
+      "prediction": "I don't think we need to do this. The `ManifestFile.SPEC_ID` field is only used in the `ManifestFile` constructor, so it should be available there.",
+      "reference": "These modifications allow BaseFile to translate into a SparkRow with the specID as a column"
+    },
+    {
+      "prediction": "I think this should be `typeof readPreference === 'object' || readPreference instanceof ReadPreference`",
+      "reference": "is this something we've been missing this whole time?"
+    },
+    {
+      "prediction": "I think this should be `*` instead of `allEvents`.",
+      "reference": "nit: this is a package-wide constant so better to narrow the scope like `allEventsSign`."
+    },
+    {
+      "prediction": "I think this should be `auth_cmd = ['thrift', '-r', '-I', '.', 'gen', 'py', auth_thrift]` instead of `auth_cmd = ['thrift', '-r', '-I', '.', 'gen', 'py', auth_thrift]`",
+      "reference": "There seems to be some repetition. Does a local function make this code shorter overall?"
+    },
+    {
+      "prediction": "I think we can remove the `isHydrating` variable now that we don't need to keep track of whether we're hydrating or not.",
+      "reference": "Might be cheaper to reuse the `null` assignment of line 313 and set `isHydrating` to null instead WDYT?"
+    },
+    {
+      "prediction": "Why was this removed?",
+      "reference": "Revert this blank line"
+    },
+    {
+      "prediction": "Why are we clearing these here?",
+      "reference": "This was introduced recently for the Hadoop 3 upgrade. Copy/paste error but definitely causing some of the new test failures."
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "I think we should only need to do this for `:approve` since we are using `authenticate_user!` for all other actions and that checks for disabled client"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this test. I think it should go in `test_webapi.py`",
+      "reference": "I wonder if we should create new tests for header checks; here for instance we're mixing querystring and headers. Thoughts?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this task. I think it should go in `lib/tasks/apply_helpers.rb`.",
+      "reference": "We should be able to add sensitive by hard-coding the parameters, same as you put into the metadata in apply_helpers."
+    },
+    {
+      "prediction": "Why do you need this?",
+      "reference": "Why do you think that it is necessary to create this class?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `aws-sdk-go-amazon-ec2` package, since that's where the `aws-sdk-go` package is.",
+      "reference": "replacing the `1` with `io.SeekCurrent` may be good here"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface. I think it should be in `gopkg.in/square/go-jose.v2/src/jose/rsa.go`",
+      "reference": "nit: perhaps this file would be better named `jwks_source.go` ?"
+    },
+    {
+      "prediction": "I think this should be `PARAM_OVERRIDE_FILE_NAME`",
+      "reference": "It will be helpful to specify the intended priority as well for the properties within this file."
+    },
+    {
+      "prediction": "I think this should be `Run func(*exec.Cmd) ([]byte, error)`",
+      "reference": "Make this private and update all the public functions to not take this argument. The variable you set below should also be private, then in the tests instead of passing the variable in to the function just update the variable. We don't want to expose the testing implementation in the public api if it can be avoided."
+    },
+    {
+      "prediction": "I think this should be `hive/tpcds/trafodion/customer/customer.dat`",
+      "reference": "Do you now why is it a specific table name is used as a default?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `if (DirExists(\"/home/DDITAdministrator/myagent/_work/_temp\"))` part of the if statement above.",
+      "reference": "If all you're doing is checking for the existence of a directory as your \"temp\" directory, why does it matter what OS you're on?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could get it to work.",
+      "reference": "nit- would be cleaner to import it once"
+    },
+    {
+      "prediction": "I don't think this is necessary. The `x-amz-*` headers should be added to the `Access-Control-Allow-Headers` header.",
+      "reference": "Looks like there is a closing parenthesis `)` missing here."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that the `error->all(FLERR, str);` is only used to print an error message. It doesn't actually change the value of `error->err` or `error->errstr`.",
+      "reference": "this one bothers me. There are many places where this one is written to for an error message and just a single callsite is changed."
+    },
+    {
+      "prediction": "I think we can remove this test now.",
+      "reference": "You can put this in `kbfsOpsInit()` in you want, so we don't have to call it in every test."
+    },
+    {
+      "prediction": "I think this should be ARM64 instead of ARM_32.",
+      "reference": "`ARM_32` is what clients define as an input to DR, and is not always defined internally: we use just `ARM` to mean AArch32."
+    },
+    {
+      "prediction": "I think this should be `originalSchema` and not `schema`.",
+      "reference": "s/for use/to use for/"
+    },
+    {
+      "prediction": "I think this should be `$event->getAvatarId() ? $event->getAvatarId()->getValue() : $event->getAvatarFilename() ? $event->getAvatarFilename()->getValue() : null;`",
+      "reference": "` $event->getAvatarFilename()` this function return `string` or `null`. In this place ` $event->getAvatarFilename()->getValue()` return `Fatal error `"
+    },
+    {
+      "prediction": "I think this should be `ErrPersistenceLimitExceeded` and not `ErrPersistenceMaxQPSReached`.",
+      "reference": "why not creating a new error type for server's own resource limit exceed error and do conversion within rpc interceptor?"
+    },
+    {
+      "prediction": "I think this should be `Propagators.DefaultTextMapPropagator` and not `Propagators.DefaultTextMapPropagator`",
+      "reference": "Could be a timing thing here. When options are created they'll copy the ref for the current default propagator. If user sets through SDK the global propagator after that, it won't be reflected. Could leave it null here and then when it is used in instrumentation do `options.Propagator ?? Propagators.DefaultTextMapPropagator`. It would be a slight perf hit but it makes it hot-swappable."
+    },
+    {
+      "prediction": "I think it's better to remove this empty line.",
+      "reference": "Unrelated file. Please remove from PR."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `else` branch of the `else` above.",
+      "reference": "Careful: you only want two newlines after the final header message. (That indicates the end of headers and the start of data.)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `if flow.request.method == 'GET'` block, since we don't want to force an upgrade if it's a GET request.",
+      "reference": "This will not work on Python 3 (to which we are transitioning) because `.content` is bytes, not a str. Can you make the pattern a bytes object as well (like so: `b\"pattern\"`)?"
+    },
+    {
+      "prediction": "`requiredOwned` and `requiredAttrs` should be renamed to `requiredAttributes` and `requiredAttrsList` respectively.",
+      "reference": "This was to allow the tests to pass when the order of the required was different. Silly, but we don't have an easy way to check for \"equal but order doesn't matter\" in chai."
+    },
+    {
+      "prediction": "This should be `static const CostMethodHbaseDelete *m = NULL;` instead of `static THREAD_P CostMethodHbaseDelete *m = NULL;`",
+      "reference": "maybe say \"returns the default cost method that returns an object of cost 1\"."
+    },
+    {
+      "prediction": "I think we can drop this rule now, since we don't need to support VXLAN traffic anymore.",
+      "reference": "I believe these rules will be enforced both (1) on egress from a local workload, and (2) on ingress **to** a local workload. Right? I understand that we definitely want (1), but do we really want to enforce (2) as well?"
+    },
+    {
+      "prediction": "I think this should be `callbackContext.error(response.asString() + \"\\n\" + response.fullResponseAsJSONObject());`",
+      "reference": "Use `response.asJsonObject()` instead. Also, use `put()` instead of `putOpt()`, `null` as a value is fine."
+    },
+    {
+      "prediction": "`expect(editor.is(':visible')).toBe(true);` should be `expect(editor.is(':visible')).toBe(true);`",
+      "reference": "Are you sure? We've changed it during an introduction of the IME support."
+    },
+    {
+      "prediction": "I think it would be better to use `null !== $dqlFilter` instead of `empty($dqlFilter)`.",
+      "reference": "I think here we want the use of `empty()` to take care of empty strings. If you put `dql_filter: ''` in your YAML config ... this will add `->andWhere('')` and it will fail, right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `if p2pMsgLatency > 0` block, but I'm not sure if that's the right place to do it.",
+      "reference": "this is golang's named return, err is defined, `return` is equivalent to `return err` the current code has no problem"
+    },
+    {
+      "prediction": "I think it would be better to use a `char *` instead of a `char * (out)`.",
+      "reference": "I don't think this is right; in Rust terms it's like a `&mut String`, in Java `StringBuilder` - it's not a return value from the function which is what `(out)` is for."
+    },
+    {
+      "prediction": "`>= 0` and `<= label_channels` can be replaced with `torch.clamp(labels, max=0, min=label_channels)`.",
+      "reference": "If the label is not binary, should we rename this function?"
+    },
+    {
+      "prediction": "I think it would be better to use `os.path.join` instead of concatenation.",
+      "reference": "This is very-very bad idea because of its performance impact"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "WHY U NOT SORT IMPORTS?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `sign_up_helper.rb` since it's only used for sign up.",
+      "reference": "Put a comma after the last item of a multiline hash."
+    },
+    {
+      "prediction": "I think this should be `Dimension, Dimensioned` to be consistent with the rest of the docstring.",
+      "reference": "How come we need ``Dimensioned`` in the top-level namespace?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best tags to use here, but it's the only one I could think of.",
+      "reference": "excess whitespace through plugin."
+    },
+    {
+      "prediction": "`std::size_t` should be `unsigned`",
+      "reference": "What is the problem here? I feel we used this in a lot of places."
+    },
+    {
+      "prediction": "do you need to return a list here?",
+      "reference": "Update the docstring."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "delete the empty line \"github.com/pkg/errors\" should be grouped with other third party packages same for the rest"
+    },
+    {
+      "prediction": "I think this should be `this.capabilities = capabilities == null ? new ImmutableCapabilities() : capabilities;` instead of `capabilities = capabilities == null ? new ImmutableCapabilities() : capabilities;`",
+      "reference": "Assigning capabilities to the field here is not the best idea. Semantically this field contains the capabilities returned by the browser after session start. So here we sould better amend capabilities and return them from `init` method to pass later to `startSession` method (that will assign the capabilities returned by the browser to the field)."
+    },
+    {
+      "prediction": "I'm not sure why this was here in the first place, but I think it's no longer needed.",
+      "reference": "Is this a breaking change, or is the API private?"
+    },
+    {
+      "prediction": "I think this should be `import pandas as pd; test_data = pd.read_json(\"./dumps/courseData.json\");`",
+      "reference": "I don't think we need this dependency in this project right now"
+    },
+    {
+      "prediction": "I think this should be `cloud-run-events-broker` and not `cloud-run-events`",
+      "reference": "I don't think this is necessary since `CommonLabels` is a public func and every component name is also public constant."
+    },
+    {
+      "prediction": "I think this should be `[\"rb\", \"-r\", \"puppet\"]` instead of `[\"/usr/bin/ruby\", \"-r\", \"puppet\"]`",
+      "reference": "The `_example` field should be updated to include an interpreter with an array value."
+    },
+    {
+      "prediction": "I'm not sure if we want to do this, but I think it's fine for now. We'll need to update the `supportedProtocols` variable in `caddyhttp/fastcgi/fastcgi.go` as well.",
+      "reference": "Put this in the godoc of SupportedProtocols instead. Preferably we would just use one map though. Why not just use this one?"
+    },
+    {
+      "prediction": "I think this should be `DockerID: dockerID,` and not `dockerID: dockerID`",
+      "reference": "I'd feel a little safer if the first argument were a string, not a *string unless there's a particular reason for making it a pointer."
+    },
+    {
+      "prediction": "I don't think this is the right place for this test. It should be moved to `org.sonar.plugins.java.api.tree.AssignStmtTree`.",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "I think this import should be removed.",
+      "reference": "You can safely revert changes to this file."
+    },
+    {
+      "prediction": "I think you can drop the `num_total_samples` from the `loss_single` function.",
+      "reference": "This deletion is unnecessary."
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. The problem is that the `img` tag is not an AMP element, so it doesn't have a `alt` attribute. I think we need to find a way to get the `alt` text from the `smallImage` prop and use that instead of the `alt` from the `img` tag.",
+      "reference": "I'm not a huge fan of this overloading of the prop; accepting very different types of arguments for props like this can be confusing, and wasn't part of the IB. I understand wanting to import the SVG directly, but the IB suggests using `${global._googlesitekitLegacyData.admin.assetsRoot}${userInputSuccessImage}` as the `src`/`smallImage` value here. It'd be great to refactor this to accept components directly in the future, but let's do that all-at-once rather than this hybrid approach that makes the API less clear."
+    },
+    {
+      "prediction": "nit: remove asterisk imports.",
+      "reference": "I've had this happen recently with IntelliJ, was this intentional?"
+    },
+    {
+      "prediction": "I think this should be `std::unordered_map` instead of `std::unordered_map`",
+      "reference": "Order(1) vs Order(logN) -- not sure it's measurable at the ~200 element size of the table. All of the published comparisons start at ~1000 elements."
+    },
+    {
+      "prediction": "I think this should be `location_factory` and not `location` factory.",
+      "reference": "I don't know about this aliasing and 'factory'. Previous version was rather straightforward: `location.CreateLocationResolver`. Perhaps `location.CreateResolver` would be even better? What do we actually gain here from moving DI to a separate sub-package?"
+    },
+    {
+      "prediction": "This should be `throw std::runtime_error(\"ERROR: Engine bass class \" + func + \"() called. \" + engineType + \" child class is not implementing this function\\n\");`",
+      "reference": "Use `UpperCamelCase` for function names"
+    },
+    {
+      "prediction": "I think this should be `value instanceof Date` instead of `typeof value === 'string' && !isNaN(Number(value)) && (Number(value) >= 275761 || Number(value) < 0)`",
+      "reference": "I'm not 100% sold on this idea but I like it in general. Nice compromise between using the 'Date' constructor where possible and falling back to the pre #5880 behavior when it makes sense. However, instead of `Number(value) < 0`, let's do `Number(value) < MIN_YEAR` because `new Date('-2017')` is perfectly valid in JS and I don't want to break the date behavior."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `ClusterState` class.",
+      "reference": "Remember to close SOLR-11877 after this"
+    },
+    {
+      "prediction": "I think this should be `||` instead of `||`",
+      "reference": "this change should be reverted"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "This function signature is a bit complicated, some parameters are IN type (stopLoop which is modified from outside), others are OUT (registeredEvent channel which is modified inside function) I suggest the following signature -> SubscribeToRegistrationEvent(identityAddress) returns registeredEvent chan of type (RegisteredEvent not int), Unsubscribe func(), error (in case of any error). That way caller can always cance subscription if needed and check for any errors. Also - unsubscribe function hides any internal details how subscribtion is really implemented"
+    },
+    {
+      "prediction": "I think it would be better to have a separate query for `src/**/*.wav` and `src/**/*.mp3` and `src/**/*.ico` and `src/**/*.gif` and `src/**/*.svg` and `src/**/*.jpg` and `src/**/*.png` and `src/**/*.js` and `src/**/*.css` and `src/**/*.json` and `src/**/*.min.js` and `src/**/*.min.css` and `src/**/*.min.json` and `src/**/*.min.css` and `src/**/*.min.html` and `src/**/*.min.mp3` and `src/**/*.min.webm` and `src/**/*.min.webp` and `src/**/*.min.svg` and `src/**/*.min.wmf` and `src/**/*.min.wmf.gz` and `src/**/*.min.webm` and `src/**/*.min.webp` and `src/**/*.min.mp4` and `src/**/*.min.mp3` and `src/**/*.min.webm` and `src/**/*.min.mp4` and `src/**/*.min.ts` and `src/**/*.ts` and `src/**/*.txt` and `src/**/*.css` and `src/**/*.js` and `src/**/*.html` and `src/**/*.min.js` and `src/**/*.css` and `src",
+      "reference": "`.wav`?! cannot we use something slightly more modern and compressed instead? :) I dunno, like `.mp3` or `.aac` or `.ogg`..."
+    },
+    {
+      "prediction": "I don't think this is the right place to emit this. It should be in the `swap` class, not the `client` class.",
+      "reference": "Wouldn't a single signal call be enough? (without the boolean even)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I suspect that this whole block, which is going to be repeated a number of times, should probably be in a helper. I guess it would need to be given the object and the title and could probably figure out everything else from that?"
+    },
+    {
+      "prediction": "I think this should be `attr_reader :path_resolver`",
+      "reference": "Should we expose this attribute in the Asciidoctor.js API ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `Traversable` since it's related to traversal of the multimap.",
+      "reference": "I think contains on `Multimap` was broken - it should return true if one of the values is associated with the key? It that right"
+    },
+    {
+      "prediction": "I think this should be `DEF_KV_CONCURRENCY` and `DEF_ISOLATION_READ_ENGINES` should be `DEF_KV_ISOLATION_READ_ENGINES`",
+      "reference": "delete this line"
+    },
+    {
+      "prediction": "I think this should be `conn.table_exists? \"test_client_requests\"`, not `return if conn.table_exists? \"test_client_requests\"`",
+      "reference": "Presumably we no longer need `force: true` here."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this check. I think it should be done in `hipHostAlloc` and `hipHostMalloc`.",
+      "reference": "HIP runtime is compiled using a C++ compiler. So comparison to both NULL as well as 0 does not make sense. Just comparing to NULL is sufficient."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this, but it seems like it would be a good place to start.",
+      "reference": "these urls are still in `a` tags so we should probably include these links separately for a non-HTML version"
+    },
+    {
+      "prediction": "I think this should be `s.mu.Lock()` instead of `s.mu.Unlock()`",
+      "reference": "That's fine though, meaning that you can send to this channel under the server lock. The internalSendLoop will pick up the change when the server lock is released (if loop is blocked trying to grab the server lock). Even the way you do it here (releasing the lock, sending, then reacquiring) does not guarantee that the internalSendLoop will have time to refresh in that interval."
+    },
+    {
+      "prediction": "I think this should be renamed to `ENABLE_CHROMIUM_DOWNLOADING`",
+      "reference": "The name `enableDownloading` implies this is a boolean capability. How about `downloadDir`?"
+    },
+    {
+      "prediction": "I think this method should be renamed to `getValidator` or `getValidatorForProject`.",
+      "reference": "Found one internal team is using this method. Will coordinate with them about the migration solution."
+    },
+    {
+      "prediction": "I don't think we need to add this to the docs, since it's only used in one place.",
+      "reference": "The field `HttpServerErrorCodes` should be called `HttpServerErrorStatuses`"
+    },
+    {
+      "prediction": "nit: I think we should move this to the top of the file, so that it is visible to the user of the file.",
+      "reference": "Can you please make each of these in a separate line. It will be easier to keep them sorted, and add/remove."
+    },
+    {
+      "prediction": "I think this should be `req.getParams().getBool(ConfigSetParams.CLEANUP, false)` instead of `req.getParams().getBool(ConfigSetParams.OVERWRITE, false)`",
+      "reference": "should we error instead of silently ignoring the `cleanup` param? it defaults to `false`, so someone must have explicitly set it to `true`"
+    },
+    {
+      "prediction": "I think this should be `fmt.Errorf(\"archive root directory missing: expected a 'root' directory but got %q\", archive.File[0].Name)`",
+      "reference": "There is no specific requirement that the root be called `root`, just that there is a single root directory and that it be the first entry in the zip."
+    },
+    {
+      "prediction": "I think this should be `!current_page?(user_path) && current_user` instead of `!current_page?(me_path) && current_user`",
+      "reference": "perhaps we should encapsulate this logic in a helper method w a test?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Now that we're only using this repo as a Gem you shouldn't need to `require` an expanded local file path like below."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `Timestamps.java` since it's only used in that class.",
+      "reference": "This isn't necessarily a year. It may be months, days, or hours. Can we return `intValue()` directly instead?"
+    },
+    {
+      "prediction": "I think we can remove this too.",
+      "reference": "this is redundant because it gets logged immediately on entering the handleContainerChange function"
+    },
+    {
+      "prediction": "I think this should be `e := \"EVENT_\" + v.Event; e == model.NotificationEventType_EVENT_DEPLOYMENT_WAIT_APPROVAL.String() {`",
+      "reference": "`ds.GenericDeploymentConfig.DeploymentNotification` in L128 is nullable."
+    },
+    {
+      "prediction": "I think this should be `getProtocolContext()`, not `getProtocolContext().get()`.",
+      "reference": "this shouldn't be exposed here - this class isn't a carriage for this - its used internally to determine if/how we can mine."
+    },
+    {
+      "prediction": "I think this should be renamed to `uninstallExtensionAsync`",
+      "reference": "It's fine to leave these checks in. It'll make the exception come from `FirefoxDriver`, and that's probably clearer to a user."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I've been able to get this to work.",
+      "reference": "We don't want to include the `IntegrationSpecHelper` for models...mind moving this line out of the loop to be `config.include EnvironmentSpecHelper, type: :model`?"
+    },
+    {
+      "prediction": "I think this should be `0755` instead of `0666`.",
+      "reference": "Please check the return on these."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to make sure that the tarball is created in the cache first, and only if it isn't, then add it to the local storage. Otherwise, if the tarball is already in the cache, then the local storage won't be able to find it.",
+      "reference": "Do we need `==` for true?"
+    },
+    {
+      "prediction": "I think it should be 0.0001.",
+      "reference": "Changing the default value may cause BC-breaking. Suggest changing this value in config."
+    },
+    {
+      "prediction": "I think this should be `int(slider_pos / len(FEE_ETA_TARGETS) + 0.5)`",
+      "reference": "how does that happen?"
+    },
+    {
+      "prediction": "doctests are not supposed to be run as part of the unit tests.",
+      "reference": "This is gross/clever. :-)"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Take a look at the name of this file. Definitely copy and"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Can you check with @mahebbar how to work this error. Should be different from 404."
+    },
+    {
+      "prediction": "I don't think this change is needed.",
+      "reference": "Do we still need the upper bound `<3.9`?"
+    },
+    {
+      "prediction": "I think this should be `realpath($input_path) === realpath(dirname(__DIR__, 2) . DIRECTORY_SEPARATOR . 'bin' . DIRECTORY_SEPARATOR . 'psalm')` and not `realpath($input_path) === realpath(dirname(__DIR__, 3) . DIRECTORY_SEPARATOR . 'bin' . DIRECTORY_SEPARATOR . 'psalm')`",
+      "reference": "Does this mean `vendor/bin/psalm` is not a symlink (or whatever equivalent Windows has for symlinks) on Windows?"
+    },
+    {
+      "prediction": "I think this is a leftover from the previous commit.",
+      "reference": "@lpabon having a separate package name `aws_test` allows to test the package as if the tester was an external package. If the test package name is the same as the package being tested, the test package can also use methods and variables not exposed to the eventual user."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be in `test/integration`",
+      "reference": "The exporter should still be able to test these histogram instrument kinds, right? Is there another reason to remove these?"
+    },
+    {
+      "prediction": "I think this should be `resetTxnBytes` instead of `ResetTxnBytes`.",
+      "reference": "should this increment eval.Round somehow?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a good idea of what the best way to do this is.",
+      "reference": "Although this change looks good. Maybe there are other inconsistent cases too?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this helper. It's used in `layouts/application.html.erb`, which is the main application template, but this helper is only used in `layouts/teams.html.erb`, which is the template for the teams landing page. I think it would be better to move this helper to the `application_helper.rb` file, which contains all of the common helper methods used throughout the application. What do you think?",
+      "reference": "This is a way better method name."
+    },
+    {
+      "prediction": "I think we can remove this notice altogether since we're not using it anywhere else in the codebase.",
+      "reference": "Prefer single-quoted strings when you don't need string interpolation or special symbols."
+    },
+    {
+      "prediction": "I think this should be `options = isArray(options) ? options.slice() : [];`",
+      "reference": "I was wondering if this should allow per role specification, instead of (or in addition to) a generic \"allowed everywhere\". So you could do: `{ separator: ['aria-valuenow', 'aria-valuemin', 'aria-valuemax'] }`. You could still allow the array, and you could add a wildcard option for the \"generic\" case you've got now: `{ '*': ['always-allowed'] }`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `NABasicObject` since that's where the `NABasicObject` type is defined.",
+      "reference": "I'm confused by the name. The string was a Hive string but it isn't any longer? Why should we care? (I imagine I'll find the answer later... there is some different semantic that we want downstream...)"
+    },
+    {
+      "prediction": "I don't think you need to import * here.",
+      "reference": "Could you please leave the explicit imports?"
+    },
+    {
+      "prediction": "I think we should rename this package to `proto` instead of `protoproto`",
+      "reference": "Why need alias?"
+    },
+    {
+      "prediction": "I think this should be `nil` and the `tlfIDGetter` should be `testTLFJournalConfig.GetTlfID`",
+      "reference": "Shouldn't this be a `libfs.nullIDGetter` (exported, of course)?"
+    },
+    {
+      "prediction": "I'm not sure if we should be using these functions here. I think we should be using the ones in `pkg/logicsig` and `pkg/app` instead, but I don't know if we should be using them at all.",
+      "reference": "I dislike the messages.go pattern; the strings are only used once and make more sense in context. The other two added reportErrorf() calls in clerk.go have inline strings."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `TagNameSet::add`",
+      "reference": "Seems we don't need it anymore."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to validate the `published` field. I think it should be in the `GuidingTopic` model, since that's where the `published` field is set.",
+      "reference": "This might cause problems with the weird way we publish Guidance and Groups in the UI. We will have to make sure that UAT is thorough."
+    },
+    {
+      "prediction": "I'm not sure this is the right default. I think it should be `time.Minute` or `time.Second` depending on what you want it to be.",
+      "reference": "this default needs to be on order 1 minute, I'm not sure why we defaulted to 1 hour below..."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to do this. I think it would be better to do this in the `ACMEConfig` module, since that's where the challenges are loaded.",
+      "reference": "I figure to make `{env.*}` work, right? I can't think of any other placeholder that makes sense here"
+    },
+    {
+      "prediction": "I think this should be `network::NetworkUtils::getAvailablePort()`, not `network::NetworkUtils::getAvailableLocalPort()`.",
+      "reference": "Why change the port from 0 to getAvailablePort()?"
+    },
+    {
+      "prediction": "I think this should be `30.000000000000004`",
+      "reference": "For me would be great don't repeat timeout sending in every get_locator call. Looks like it should be sent into LocatorsManager init. And possibly it should me non mandatory, something like that: `def __init__ (self, locators, timeout=60):`"
+    },
+    {
+      "prediction": "nit: I think we can remove the `NewGlobalOpts` constructor and just use the `New` method directly.",
+      "reference": "I feel like since now we don't initiate the `opts.projectName` with `opts.ProjectName()`, which means `opts.projectName` and `opts.ProjectName()` they don't necessarily equal to each other, we need to also substitute all usage of `opts.ProjectName()` within `Ask()`, `Validate()`, and `Execute()` to `opts.projectName`. Otherwise if `-p` is set, `opts.ProjectName()` will get expired."
+    },
+    {
+      "prediction": "I don't think we need to check for null here.",
+      "reference": "why do we thinkg that an empty string is ok to return from this method? I think the null should never be passed in here in the first place"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The idea here is that if there are multiple nodes with the same actual node, we want to return the one that is closest to the node we are looking for.",
+      "reference": "nit: can remove this `else` since we return from the condition above."
+    },
+    {
+      "prediction": "I think we can remove the `default(CancellationToken)` argument from the `WriteAsync` method as well.",
+      "reference": "You missed `Write<T>(...)`! I'm kidding ofc."
+    },
+    {
+      "prediction": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Was there a customer who had this set to an empty string?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in `Ncr::Approvals::ApprovalsController` since it's only used there.",
+      "reference": "I know we have the `reporter` spec below, but what about a unit test for this to explain reasoning behind logic? If I were going to update this decorator, I would assume it wasn't covered by tests because there is no unit test."
+    },
+    {
+      "prediction": "`attach` is not used in this PR, so I think we can remove it.",
+      "reference": "it's better to be : void attach(String name, Part part);"
+    },
+    {
+      "prediction": "I don't think we should change the signature of the `initDefinitionLevelsReader` method.",
+      "reference": "I didn't see where the IOException can get thrown, is this just to match the V1 reader?"
+    },
+    {
+      "prediction": "I think this should be `bool weightDistMat` instead of `bool forceRDKit)`.",
+      "reference": "We normally suppress this warning with `RDUSED_PARAM(forceRDKit)`"
+    },
+    {
+      "prediction": "This should be `auto it = mappedPlayerGuids.find(guid); if (it == mappedPlayerGuids.end()) return nullptr; return it->second;`",
+      "reference": "What if `Game::getPlayerByNameWildcard` had not been called before. It would not find a player, wouldn't it?"
+    },
+    {
+      "prediction": "I think this should be `patchRequired` instead of `requiresPatchCheck`.",
+      "reference": "Why upper case \"R\"?"
+    },
+    {
+      "prediction": "I think we can remove this function now, since it's no longer used.",
+      "reference": "Does this need to return an error or could it return `false, nil`?"
+    },
+    {
+      "prediction": "Why is this protected? It looks like it should be `public` by default.",
+      "reference": "Can you not change those scopes in public API classes? This applies here and in other places -- protected changed to package-scope for source is not really an API-compatible change."
+    },
+    {
+      "prediction": "Why do you need this?",
+      "reference": "this parameter is only used in POST, please use `getRequest()->request->get('...');` Thanks"
+    },
+    {
+      "prediction": "I think this should be `filename` instead of `filename`",
+      "reference": "Problem number 1: Scoped packages would have a `/` character here. Changing this to `validatePackage` resolves the 403."
+    },
+    {
+      "prediction": "I think this should be `except KeyError as e: deep_mapped[k] = e` instead of `except Exception as e: deep_mapped[k] = e`",
+      "reference": "I don't quite get why there would be key errors: ``deep_mapped`` is a clone of ``self`` and ``k`` comes from ``self.items()`` so why would the key ever be rejected?"
+    },
+    {
+      "prediction": "I think we can remove the `chainID` parameter from the `withdraw` method, since we don't use it anymore.",
+      "reference": "this chain ID determines only from which chain to withdraw, therefore your changes do not accomplish what you want them to accomplish. You'll need changes to `func (aps *hermesPromiseSettler) Withdraw(chainID int64, providerID identity.Identity, hermesID, beneficiary common.Address) error`. The method probably has to include two chain ids: to and from. Currently, the method internally uses: `aps.config.L1ChainID` -> the chain to withdraw to `chainID` -> the chainID that was passed as the chain that the withdrawal is originating from."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `trace` package.",
+      "reference": "If instead we had `TraceContext.extract` return a bool value as a second return value, we could avoid the byte array comparison in `TraceID.IsValid`. Did you consider that alternative?"
+    },
+    {
+      "prediction": "`paginated` should be a `const bool` parameter.",
+      "reference": "here shouldn't we use `depotFreeLimit`?"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea, but I can't think of a better way to do this.",
+      "reference": "Could we use the Rails application's secret token instead? Do we actually need our own here?"
+    },
+    {
+      "prediction": "I think it would be better to use `kdf.groupby(['a', 'b'], axis=1)` instead of `kdf.groupby('a', 'b')`",
+      "reference": "so should be fixed here also"
+    },
+    {
+      "prediction": "I think this should be `dmc->size == sizeof(dr_mcontext_t) - sizeof(dr_opmask_t) - 1`",
+      "reference": "I would just remove this assert as it's going to get un-maintainable with a long list of valid sizes. Ditto below."
+    },
+    {
+      "prediction": "I think this should be `/var/lib/ecs/data/`",
+      "reference": "`HostDataDir` is misleading. Can we rename it to something more relevant ? The constant should also be moved up and reused as necessary."
+    },
+    {
+      "prediction": "I think this should be 36. The overhead of the ESP packet is 36 bytes, including the ESP header and trailer.",
+      "reference": "I still feel like we are double-counting the outer IP header here (once in `defaultMTUGRE` and once in `ipsecESPOverhead`) but I'm not that familiar with IPsec."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `System.Diagnostics.Metrics`",
+      "reference": "Not sure what benefit we gain by extending BaseProcessor here, as this seems to be defining a new OnEnd method. Could we modify MeasurementItem to have all the things, and then MeasurementProcessor can be simply extending BaseProcessor<MeasurementItem> (not blocking. just noting some observations in the PR :) )"
+    },
+    {
+      "prediction": "I think this should be `strings.ReplaceAll`, not `strings.Replace`. The latter replaces all occurrences of a pattern with another, whereas `ReplaceAll` replaces all occurrences of a pattern with the same value as the first occurrence.",
+      "reference": "This change was not really intended but made by the linter of VS Code. And looking at other packages this looks like a best practise to place interal packages on the top and gh imports afterwards."
+    },
+    {
+      "prediction": "I think it would be better to add a note in `New` explaining that a configured `Requestlogger` will not log HealthChecks.",
+      "reference": "This seems like a more appropriate message under `ListenAndServe`. WDYT?"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in the test.",
+      "reference": "Is this not required?"
+    },
+    {
+      "prediction": "I think this should be `isset($this->entity['search']['sort']['field'] ?? $this->request->query->get('sortField'));`",
+      "reference": "We should start thinking of an object that encapsulates this information :) we might need more arguments in the future."
+    },
+    {
+      "prediction": "Why are these two error messages different?",
+      "reference": "Given that we are (conceptually) searching across multiple self-dependencies, all we can say is that we didn't have a self dependency in which *both* source and dest masks were correct. Since the spec doesn't imagine this case, the valid usage statement assume we can differentiate only wrong source from only wrong dest within a *single* self-dependency -- which we can't across multiple (at least not meaningfully)... so we always return both the source/dest VUID's if we can't find a self-dependency in with *both* are correct."
+    },
+    {
+      "prediction": "I think this is the only place where `method.response()` is used.",
+      "reference": "rename? MiningCoordinator always exists"
+    },
+    {
+      "prediction": "This should be `true` by default, not `false`.",
+      "reference": "Here would be the same case you don't need to call `loadConfigs`, but you would need to reread configs using `getConfig` on each getHost function call, not once per file."
+    },
+    {
+      "prediction": "I think this should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `d.folder.fs.renameAndDeletionLock.Lock()` and `d.folder.fs.renameAndDeletionLock.Unlock()` should be `",
+      "reference": "Unlocking with defer means that this lock is still held curing the call to forgetNode(), below, which I see attempts to acquire `f.mu`, which looks dangerous to me."
+    },
+    {
+      "prediction": "I think we can remove the `cst` and `bs` variables now.",
+      "reference": "Does anyone still use the cst out of this method? If not consider deleting"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this redirect. I think it should go in the `podcasts` app, since it's about podcasts, but I don't know for sure.",
+      "reference": "Line is too long. [97/80]"
+    },
+    {
+      "prediction": "I don't think you need to import `BytesIO` here. You can just use `io.StringIO()` or `io.BytesIO()`.",
+      "reference": "This seems unused."
+    },
+    {
+      "prediction": "I think this should be `panic(\"unsupported os-release\")` instead of `panic(\"unsupport os-release\")`",
+      "reference": "@luguanglong , Thanks for the fix, can you re-phrase it to sound better something like \"This OS version is currently un-supported by keadm\""
+    },
+    {
+      "prediction": "I think this should be `delattr(mol, 'fpInfo') if hasattr(mol, '_fpInfo') else None`",
+      "reference": "should probably be called `_DeleteFpInfoAttr` because it removes it. Cleaning gives the impression it is still there. I would also move this to the end of the function `GetAtomicWeightsForFingerprint`."
+    },
+    {
+      "prediction": "I think this should be `VUID-vkCmdBindDescriptorSets-pipelineBindPoint-00361` and not `VUID-vkCmdBindDescriptorSets-pipelineBindPoint-00361`",
+      "reference": "nit, can we use `VK_PIPELINE_BIND_POINT_RAY_TRACING_KHR` here (granted it isn't else where, maybe worth fixing here or in separate PR)"
+    },
+    {
+      "prediction": "I think this should be `params.require(:workout).permit(:edit_url, :summary, :title, :url)`",
+      "reference": "This will need to be updated in the upcase-exercises repo as well."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [121/120]",
+      "reference": "Think it's worth extracting this to `Exercise#complete?`?"
+    },
+    {
+      "prediction": "Why is this method removed?",
+      "reference": "This is still a potentially useful public method - should we remove such methods?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this method. I think it should go in the `ApiDefaultConfig` class, since that's where the rest of the `ApiConfig` class is.",
+      "reference": "I don't understand this change, what is happening here?"
+    },
+    {
+      "prediction": "I don't think we need to split here, just use `int(self.obj.excelCellObject.Application.Version.split('.')[0])`",
+      "reference": "I think its worth stating this is `versionMajor`"
+    },
+    {
+      "prediction": "I don't think we need this function at all, since we have `SetConsensus` and `GetConsensus` already.",
+      "reference": "rebuild: say loads and merges"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Why was this removed here (also in the other file)? Shouldn't we import them so that the reference in the docs below is interpreted correctly?"
+    },
+    {
+      "prediction": "I don't think we need to change this line.",
+      "reference": "I'd prefer to have a blank line between the class declaration and its first member. Is it possible?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/unit/mongodb/test-helper.js`",
+      "reference": "Is it really LEGACY_HELLO if we are using the most up to date HELLO? Perhaps `MONGOS_HELLO` would work here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be.",
+      "reference": "I am afraid this is going to cause problems on Windows where it is not uncommon to have injected threads (CTRL_SHUTDOWN, CTRL_LOGOFF, etc.) we have no privileges to suspend -- and thus retrying will just fail again, and with the new \"synchall failure is fatal and should kill the process\" approach it turns what used to work for us on Windows into process death."
+    },
+    {
+      "prediction": "Should this be `--installer-signing-identifier`?",
+      "reference": "think it should only be in create_dist"
+    },
+    {
+      "prediction": "I think we can remove the `if int(cport.HostPort) > 0` check here, since it's already handled by the `if int(cport.ContainerPort) > 0` check above.",
+      "reference": "Do you plan to support Pod spec change? Like hostPort is added/removed later after Pod creation?"
+    },
+    {
+      "prediction": "I think this should be `0` instead of `-1`",
+      "reference": "Is this behavior correct? It used to return `true`, which should map to `dokan.ExistingDir`. Was that previously a bug?"
+    },
+    {
+      "prediction": "I think this should be `osrm::util::Log() << \"No shared memory regions found. Try running osrm-datastore\";`",
+      "reference": "What about just printing an empty list in this case? That would make the output more predictable."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "I looked and couldn't find any usage of `container` in any of the editor component files (js or hbs), so I assume this was used once and didn't get removed?"
+    },
+    {
+      "prediction": "Can you add a space after the `if`?",
+      "reference": "I just updated a number of sites using older versions of Roots onto WP 3.6 and this little function change was required. Multiple search bars were displaying when I used the search widget in a widgetized sidebar. Updated the roots_get_search_form as seen in this change resolved it for me!"
+    },
+    {
+      "prediction": "I think this should be `type` instead of `typeOfFreeProduct`",
+      "reference": "Fixes test failed in some cases"
+    },
+    {
+      "prediction": "I'm not sure if we want to keep this or remove it. I think we want to keep it, but I don't know if it's safe to keep it.",
+      "reference": "what's the prod url for this? just curious. @amirsaber"
+    },
+    {
+      "prediction": "I think this should be `azkaban.job.command.args` and not `az_job_command_args`",
+      "reference": "Can you please change this to azkaban.jobs.java.opts?"
+    },
+    {
+      "prediction": "I think this should be `__( 'Source:', 'google-site-kit' ) + sprintf( '<a class=\"googlesitekit-cta-link googlesitekit-cta-link--external googlesitekit-cta-link--inherit\" href=\"%1$s\">%2$s</a>', getSiteKitAdminURL( 'googlesitekit-module-analytics' ), _x( 'Analytics', 'Service name', 'google-site-kit' ) )`",
+      "reference": "Here is another concatenation which should be updated. Even though `Source:` and the link are essentially separate, it isn't RTL friendly. This would be another good use for `createInterpolateElement` I think so that we wouldn't need to include all of the classnames in the translation string (or extract them to a placeholder). That would allow us to use the `Link` component as well, rather than using its resulting HTML as a string which will be a pain to maintain if we were to make changes to the base `Link` component."
+    },
+    {
+      "prediction": "Why is this needed? `Task.FromCanceled` returns a `Task<int>`, not a `Task<int, int>`.",
+      "reference": "I would add an overload `GetCancelledZeroTask()` which calls `GetCancelledZeroTask(CancellationToken.None)`. This is cleaner than requiring the caller to pass `default(CancellationToken)` or `CancellationToken.None`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in `PreCallRecordDestroySwapchainKHR` as well as `PreCallRecordDestroyImage`.",
+      "reference": "Would it be equivalent to check if `chain != nullptr` below? Not suggesting a change, just curious."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in `wallet::change_async` instead of `wallet::queue_work_regeneration`.",
+      "reference": "Should probably do `auto this_l (shared_from_this ());` and pass/use that instead of `this`. Same a few other places. IOW, replace both `shared_from_this()` and `this` with `this_l`"
+    },
+    {
+      "prediction": "This is a bit of a nitpick, but I think it would be better to have a separate message for the verb case, rather than a separate message for the case where the verb isn't there.",
+      "reference": "And here. Please, use `'` for apostrophes in changed strings in this file."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `ExecuteContractRead` function, but I'm not sure if that's the right place to do it.",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "I think this should be `DefaultNewBlockDataVersion()`, not `DefaultNewBlockDataVersion(c Config, holes bool)`.",
+      "reference": "Please move this to `block_types.go`."
+    },
+    {
+      "prediction": "I think it would be good to have a test for this as well.",
+      "reference": "So this is not needed anymore for all attributes?"
+    },
+    {
+      "prediction": "I think we can remove the `Message` parameter from the `Event` event, since it's now part of the exception message.",
+      "reference": "Happy to get suggestion for better messaging here. Here's what I intended to convey: Something is wrong - the Exception is a hint to what might be the cause The impact of that - tracing wont work."
+    },
+    {
+      "prediction": "`update_listener` is not used in this function, can you remove it?",
+      "reference": "I think we can remove this function, right? (duplicate with `ses_starter.py`)"
+    },
+    {
+      "prediction": "I think this should be renamed to `rename_field` or something.",
+      "reference": "With this fix, I am not even sure we need to overwrite this function."
+    },
+    {
+      "prediction": "I think this should be `use_tqdm = os.getenv('QUILT_USE_TQDM').lower() == 'true'`",
+      "reference": "Why not just import this from `data_transfer`? As a rule copying the same code twice is not a good idea. Also: please run `pylint` on all files in this PR."
+    },
+    {
+      "prediction": "_server=>\"none\"}}\n+        expect(parser.parse_args(args)).to be === {:hosts_file=>hosts_path, :options_file=>nil, :type=>\"git\", :provision=>true, :preserve_hosts=>false, :root_keys=>false, :quiet=>false, :xml=>false, :color=>true, :debug=>true, :dry_run=>false, :fail_mode=>nil, :timesync=>false, :repo_proxy=>false, :add_el_extras=>false, :consoleport=>443, :pe_dir=>\"/opt/enterprise/dists\", :pe_version_file=>\"LATEST\", :pe_version_file_win=>\"LATEST-win\", :dot_fog=>\"#{home}/.fog\", :ec2_yaml=>\"config/image_templates/ec2.yaml\", :ssh=>{:config=>false, :paranoid=>false, :timeout=>300, :auth_methods=>[\"publickey\"], :port=>22, :forward_agent=>true, :keys=>[\"#{home}/.ssh/id_rsa\"], :user_known_hosts_file=>\"#{home}/.ssh/known_hosts\"}, :install=>[\"git://github.com/puppetlabs/puppet.git#1.0\", \"git://github",
+      "reference": "You've got an absolute path here referencing your home dir :-)."
+    },
+    {
+      "prediction": "I think this should be `prefixPropertyMetaModel` instead of `prefixPostfixPropertyMetaModel`.",
+      "reference": "Maybe I should do some sorting so that these properties don't keep moving around..."
+    },
+    {
+      "prediction": "I think this test should be moved to `test/unit/mount_test.go`",
+      "reference": "Can we keep this test anyway, as it's part of the spec?"
+    },
+    {
+      "prediction": "I think it would be better to have a `MailerStrategyInterfaceCompilerPass` in the `Kernel` class itself, so that it can be used in other modules as well.",
+      "reference": "Can be removed"
+    },
+    {
+      "prediction": "I think this should be `use Config as Config;`",
+      "reference": "It's a minor quibble I know, but I like having the imports ordered by lengt"
+    },
+    {
+      "prediction": "I don't think this is the right fix. We should not be changing the content type of the response if it is a failure.",
+      "reference": "It's better to get the charset from context or other setting to let the user override it."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `v1alpha1` part of the informers, but I'm not sure if that's the right place for the informer as well.",
+      "reference": "this filename as well needs change"
+    },
+    {
+      "prediction": "I don't think we need to add `SecretAccessKeyRef` here, since we don't have a `SecretKeySelector` for `HostedZoneID`.",
+      "reference": "I think we want to remove `omitempty` here"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be using the `SignalHandler` from `runtime` instead of creating a new handler.",
+      "reference": "You cannot move this here because it breaks detach. Just call `tty.Close()` before returning the error from start"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this change.",
+      "reference": "I updated only the implicit roles who needed a update for this new rule to validate."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "> // Code generated by protoc-gen-go. DO NOT EDIT."
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "minor: skipOSConfigMetadata, otherwise, the name sounds like a boolean"
+    },
+    {
+      "prediction": "I think we can remove the `SyndTypeUtils.enclosureTypeValid` method altogether.",
+      "reference": "~~I think lines 65-67 should be moved into this else clause after `type = ...` Else, it is a bit confusing why we check the enclose type validity a second time~~ Nevermind, will do some refactoring, myself. But thanks for \"fixing\" this!"
+    },
+    {
+      "prediction": "I don't think we need this function at all. We can just use `client.http.Get(\"sessions\", ...)` directly.",
+      "reference": "`sessions := endpoints.SessionsDTO{}` and using only `sessions` should be simpler."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. It seems like it should be in the same place as dcontext->is_exiting, but I'm not sure if that's the right way to do it.",
+      "reference": "These need to be in the other order to avoid crashing when dcontext == GLOBAL_DCONTEXT (==-1)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `ConfigureKestrel` method, since it's related to the Kestrel part of the pipeline.",
+      "reference": "Why not change the other samples?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it's better to use the `getAttribute` method to get the value of the `id` attribute and then use `!!document.querySelector` to check if there is a matching element in the DOM.",
+      "reference": "Indentation is mixed up here due to spaces/tabs, I'm guessing."
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "Should we provide a little more context in the error message? Maybe throw in something about what operation was being performed?"
+    },
+    {
+      "prediction": "Shouldn't this be `array_key_exists('settings', $configuration) ? $configuration['settings'] : null`?",
+      "reference": "Project standard is spaces, not tabs. Please fix and then we can merge this."
+    },
+    {
+      "prediction": "Is this needed?",
+      "reference": "@hgtw is there anything we're worried about with regards to using this cache? Do we want to do staged testing around expeditions before we merge? Curious your thoughts here"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. It seems like it should be in `ManifestReader` since it's only used by that class, not `Manifest` itself.",
+      "reference": "The purpose of this was to allow pushing filters into the manifest files if we ever use a format other than Parquet. I think overall this is a good change because we can add it back later. Just wanted to note why the unused argument was there."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I think BSON::Decimal128 should always be serialized as Decimal128 regardless of config option."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `spec/` directory under `spec/models/bolt_spec.rb`.",
+      "reference": "We should probably just move the `require 'bolt/plan_creator'` in `Bolt::CLI` to `Bolt::Application` so it will already be loaded."
+    },
+    {
+      "prediction": "I think it would be better to have a constant for this value instead of hardcoding it.",
+      "reference": ">20 [](start = 45, length = 2) Can we override this at runtime? Maybe it should be a command line option with default."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "can just remove it ?"
+    },
+    {
+      "prediction": "I think we should move this into the `FinishMigrateRepository` function.",
+      "reference": "Better call it `CleanUpMirrorInfo`?"
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "@shannonwells is there still something missing here?"
+    },
+    {
+      "prediction": "I think this should be `add_setting :stress_test_duration_in_seconds`",
+      "reference": "This could be more indicative of what it is... e.g... `stresstest_time_limit`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to the `PDPSimple` constructor where we create the listener.",
+      "reference": "Shouldn't we do a `reader_payload_pool_.reset()` here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be in the `BeaconChainState` struct, since it's related to the beacon chain.",
+      "reference": "I'd prefer the nomenclature less tightly coupled to DRAND throughout. I played with renaming the `drand` package to `beacon` but it was too much noise in this change."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. It's not a property of the DynamicMap, it's a property of the underlying HoloMap. The DynamicMap should have a `current_key` property, not the underlying HoloMap.",
+      "reference": "I think I would prefer you declare `self._current_key=None` in the constructor and just return `self._current_key`. That way you can prevent anyone from overwriting `current_key` without needing to use `getattr` here."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/flux` since that's the package that contains the `flux` runtime package.",
+      "reference": "I guess I like the name `mode` okay. I can't think of anything better."
+    },
+    {
+      "prediction": "I think it would be better to have a method `NewStepWithDefaultTimeout` instead of duplicating the code.",
+      "reference": "I would create a new method (NewStepWithDefaultTimeout?) for this purpose because timeout<=0 looks like a hidden logic"
+    },
+    {
+      "prediction": "`limit` should be `int` not `int64`",
+      "reference": "Are these limit changes related?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `scapy.utils` or `scapy.utils6`.",
+      "reference": "Don't you mean `import scapy.consts`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we need to make sure that we don't roll back the primary stage if there are still resources left in the update.",
+      "reference": "`ctx` is unused in rollbackPrimary"
+    },
+    {
+      "prediction": "I think this should be -1235 instead of 0.",
+      "reference": "This is not what title says."
+    },
+    {
+      "prediction": "I don't think this is the right place for this interface. I think it should be in `org.pmd.lang.ast`",
+      "reference": "a fix applies several fixes? maybe this should simply be `applyToNode`"
+    },
+    {
+      "prediction": "I think this is the only place where we use fakeGetClientset and fakeListfn. I think we should remove the fakeGetClientset and use fakeListOk wherever we need it.",
+      "reference": "U1000: func `fakeGetOk` is unused (from `unused`)"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Because there is a \"k8s.io/client-go/tools/cache\" below, we can sort out all the \"imports\" here."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "Nit should be using `goimports` to format the import statements with standard libary imports first, new line, followed by non-standard library imports."
+    },
+    {
+      "prediction": "I think this should be `return errors.Wrapf(err, \"error deleting pipeline secret %s\", o.PipelineSecret)`",
+      "reference": "Do we not have tests for pipeline delete"
+    },
+    {
+      "prediction": "I think this should be `load_hadoop_cmd() + ['fs', '-appendToFile', local_path, destination]`",
+      "reference": "Good that you mention this constraint in the docstring :)"
+    },
+    {
+      "prediction": "I think you can remove this line.",
+      "reference": "remove this blank?"
+    },
+    {
+      "prediction": "I think this should be computed based on the number of active partitions, not the number of rows.",
+      "reference": "Partitions is misspelled 8 times in this commit, might make sense to fix the spelling for all of those."
+    },
+    {
+      "prediction": "I think this should be `from qutebrowser.browser import qutescheme` instead of `from qutebrowser.browser.qutescheme import qutescheme`",
+      "reference": "Please remove this blank line - those are only used to group Python/third-party/qutebrowser imports."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` and `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/traefik/traefik.go` or `cmd/tra",
+      "reference": "`insecure` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think we can remove the `setting.Repository.DisableHttpGit` check here.",
+      "reference": "Same as above (HTTP in upper-case)"
+    },
+    {
+      "prediction": "I think this test should be moved to the `YouTubeProviderTest` class, since it is specific to the YouTube provider.",
+      "reference": "This is getting repetitive  maybe you could introduce an abstract test case with a `getExpectedUrl($providerReference)` method ?"
+    },
+    {
+      "prediction": "I think this should be called `MakeFunction` instead of `MakeFunc`.",
+      "reference": "I do not think returning `Value{}` is correct here. For example, `reflect.MakeFunc(...).Kind()` would return `reflect.Invalid` instead of `reflect.Func`. Therefore, I think this should panic instead."
+    },
+    {
+      "prediction": "`substring(2)` is unnecessary here, since `name` is already two characters long.",
+      "reference": "Does this work for Custom Elements? iirc we're lacking test cases for them. /cc @andrewiggins @developit"
+    },
+    {
+      "prediction": "I think we can remove the `externalNodeCIDRs` field from the `ipipManager` struct.",
+      "reference": "Same change just above?"
+    },
+    {
+      "prediction": "I'm not sure about this one, but I think it's fine.",
+      "reference": "I suppose `apploader.js` isn't used by WebWorkers. So `self` will always be `window` here."
+    },
+    {
+      "prediction": "I think we can remove this line too.",
+      "reference": "Since this var is no longer defined we should remove the `if skip_targets.any?` bit below."
+    },
+    {
+      "prediction": "I think this should be `lambda_function_arn` instead of `lambda_function_arn`",
+      "reference": "I think we should `return arn` as a fallback at the end of this function (otherwise the `['Resource']` entry below could become `None`)."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to handle this. I think we should check if `np.socketProtect` is not nil, and if it's not nil, then we don't need to protect the connection. Otherwise, we need to protect the connection. What do you think?",
+      "reference": "Why session is started at all, if you need to shut it down e.g. DI should launch noopSession"
+    },
+    {
+      "prediction": "I think we should rename this to `DeleteWorkspace` instead of `DeleteAll` to be consistent with the rest of the SDK.",
+      "reference": "nit: can we rename the interface to `wsFileDeleter`"
+    },
+    {
+      "prediction": "I think we can remove the `FromAddresses` and `ToAddresses` fields from the `CompletedRule` struct.",
+      "reference": "Why target cannot be external endpoints?"
+    },
+    {
+      "prediction": "I think this should be `auto_` instead of `AUTO_PREFIX`.",
+      "reference": "We use a suffix \".AUTOCREATED\" for configsets, maybe we can use the same here?"
+    },
+    {
+      "prediction": "I think this should be `OpenTelemetrySdk.EnableOpenTelemetry(...)`",
+      "reference": "This one won't be disposed. Should be (something like) `using var openTelemetry = OpenTelemetrySdk.EnableOpenTelemetry(` no?"
+    },
+    {
+      "prediction": "I don't think you need to change this line.",
+      "reference": "Why this is in the diff?"
+    },
+    {
+      "prediction": "I think we should rename this to `CloudBackupsCreate` to be consistent with the naming used in `CloudBackupCreate` and `CloudBackupGroupCreate`.",
+      "reference": "How is status determined? When the user calls CloudBackupCreate( src_volume_id ) they can then call CloudBackupStatus( src_volume_id ) Is there something similar for this new API?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `diffBound`, since it's only called when the diff is happening in the DOM.",
+      "reference": "We could pass `excessDomChildren` to `options._catchError` and only do this if an error-boundary catches the error. Not entirely sure if that's better."
+    },
+    {
+      "prediction": "I think this should be `array_merge(..., array_values($this->getFilters()));`",
+      "reference": "In php8 named parameters were introduced and now it is required to match called method parameter name when setting parameters by array destructing or call_user_func_array() etc."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `b3` package, since that's where `b3` is defined.",
+      "reference": "If instead we had `B3.extractSingleHeader` and `B3.extract` return a bool value as a second return value, we could avoid the byte array comparison in `TraceID.IsValid`. Did you consider that alternative?"
+    },
+    {
+      "prediction": "Shouldn't this also be translated?",
+      "reference": "I think this is more of a semantic difference than a typo - `weiter hinzufgen` is like `continue adding` whereas `weitere hinzufgen` is like `add more`. I think it makes sense to change it though"
+    },
+    {
+      "prediction": "I think this test should be moved to a separate file, since it's not related to the rest of the tests.",
+      "reference": "nit: 200 is superfluous"
+    },
+    {
+      "prediction": "I'm not sure why this test is failing, but I think it's because of this line: `std::string query = \"CREATE TAG upper(name string, EMAIL string, age int, gender string, row_timestamp timestamp)\";`",
+      "reference": "This is testing `unreserved keyword`"
+    },
+    {
+      "prediction": "Shouldn't this be PROJ_WKT2_NAME_METHOD_QUADRILATERALIZED_SPHERICAL_CUBE?",
+      "reference": "paramsNatOrigin doesn't include sUVtoST. I would just remove that definition for now"
+    },
+    {
+      "prediction": "I think this test should be moved to `testutil/testutil.go` since it's only used by `testutil/blockRetrievalQueue_test.go` which is only used by `initBlockRetrievalQueueTest`",
+      "reference": "It feels like the test should be waiting for the prefetcher to shut down, but I don't have a great reason why other than that it might be confusing for debugging if there are still goroutines from old prefetchers lying. But I guess since `TogglePrefetcher(false)` doesn't actually set the prefetcher to nil, the queue shutdown method will still end up waiting for the shutdown to finish. If that sounds right, nevermind me."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this snippet. I think it should go in the `aws-identity-v2` package.",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think this should be `RESOURCE_NAME_PROJECTS`",
+      "reference": "As a long term thing, would it make sense to move the resource names as keys under the requirements map?"
+    },
+    {
+      "prediction": "I think this should be `if (cell.hasAttribute('headers') || tableGrid.hasAttribute('headers'))`",
+      "reference": "That doesn't fix the whole problem. The issue lays in this line right here, not the one above. There are two problems with this line: 1. It finds things that aren't cells in the table 2. if it doesn't find anything, it shouldn't return empty here, but continue down to look for row/ column headers."
+    },
+    {
+      "prediction": "Shouldn't this be `BaselineVersions.getResolvedArtifacts(getProject().getArtifacts())`?",
+      "reference": "I am pretty sure that this is infinite recursion as getAllProjects returns the project itself."
+    },
+    {
+      "prediction": "I think this should be renamed to `data-lists`",
+      "reference": "Something I didn't catch before I merged this PR, is the hyphenated constant a problem? E.g. should \"data-lists\" be \"data_lists\"? @paulkaplan"
+    },
+    {
+      "prediction": "I think this should be `if not conf.use_winpcapy and not conf.use_pcap and not conf.use_dnet:`",
+      "reference": "Can you write the test as `if not (conf.use_winpcapy or conf.use_pcap or conf.use_dnet):`"
+    },
+    {
+      "prediction": "I think this should be `ForeignAsset` instead of `ForeignAsset`?",
+      "reference": "we have to include foreign assets on the repository directly because they don't belong to a job? This seems very awkward... Is this a step towards the job-less assets on the repository? Did you consider having `build_asset_job` take in a set of foreign assets instead? I suppose we would then need to subclass it to be a more special `AssetJobDefinition` that can keep track of them."
+    },
+    {
+      "prediction": "Why is this change necessary? It looks like this function is only used in one place, and that place already uses `std::unique_ptr` for the lifetime of the `TempDir` object.",
+      "reference": "`-Wstringop-truncation` is a new kind of error detector introduced in GCC 8. As for this patch, these are two false-positives though."
+    },
+    {
+      "prediction": "nit: I think it would be better to call this \"only one credential type may be specified per source profile\"",
+      "reference": "Can we port this error msg to v2 too? This one is better, as it explicitly states what sources are allowed."
+    },
+    {
+      "prediction": "nit: can you move this import to the top of the file?",
+      "reference": "The convention so far has been to require directly from the defining file (in this case `../../read_preference') . I think the concern has been mostly about the potential for circular dependency cycles"
+    },
+    {
+      "prediction": "`color_type` is not used in this module, please remove it.",
+      "reference": "I suggest expanding dims here to simplify the formatting."
+    },
+    {
+      "prediction": "I think this should be `config/kinto.ini` instead of `config/kinto.ini`",
+      "reference": "Why do you need to specify the kinto prefix here?"
+    },
+    {
+      "prediction": "`!r` -> `r!r`",
+      "reference": "The `!r` is unnecessary."
+    },
+    {
+      "prediction": "I think we can remove the `accountKey != \"\"` check here, since it's already handled by the `if` above.",
+      "reference": "I didn't see existing tests for this `OpenBucketURL`. Not sure if it's easy to do without mocking these `opener` calls."
+    },
+    {
+      "prediction": "I don't think this should be a property. It should be a constructor parameter.",
+      "reference": "Should we default to `true` ?"
+    },
+    {
+      "prediction": "I think this file should be renamed to `consensus.go` and moved to `consensus` package.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `babel-plugin-transform-react-18-00.js` file, since that is where the presets are defined.",
+      "reference": "Shouldn't these options also include `@babel/preset-env`? Also I see you set `babelrc` to `false`, could we rely on our existing `.babelrc` file? Feels like some duplicate configuration otherwise."
+    },
+    {
+      "prediction": "I think this should be `reason` instead of `reasonCode`",
+      "reference": "Nit: We can use `%s` since `disconnectReason` is a string."
+    },
+    {
+      "prediction": "why do you need to change this?",
+      "reference": "I think it's better to create a new example"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `resolve_write` function, since it's specific to writing to a color attachment.",
+      "reference": "The stages are correct, but the more forgiving `kAttachmentRasterOrder` should be used, based on a review of the spec. That should give the same effect of suppressing the false positive conflict between the DEPTH R/W and resolve. > End-of-subpass multisample resolves are treated as color attachment writes for the purposes of synchronization. This applies to resolve operations for both color and depth/stencil attachments. That is, they are considered to execute in the VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT pipeline stage and their writes are synchronized with VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT. Synchronization between rendering within a subpass and any resolve operations at the end of the subpass occurs automatically, without need for explicit dependencies or pipeline barriers."
+    },
+    {
+      "prediction": "I think this should be `O[SD3](=O)(=O)[O-]` instead of `O[SD3](=O)(=O)[OH]`",
+      "reference": "To be consistent, I think it should actually just be a single `-`"
+    },
+    {
+      "prediction": "I don't think we should be using `flask` at all. We should be using `jinja2` and `jinja2_2` (or `jinja2_3`).",
+      "reference": "this seems extraneous."
+    },
+    {
+      "prediction": "I think we can remove this feature set from the `Suite` now.",
+      "reference": "Ah, I see here that 'LongDomain' is defined as something that contains a subdomain segment that is `maxLengthOfDomainSegment` long (which I think is 63 characters) - I don't think any public ACME servers/Let's Encrypt's staging environment has a restriction on this?   if it does, and the 'pebble' based ACME server does not, then Pebble needs modifying to also fail in these cases as it aims to replicate the ACME RFC as closely as possible"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be under `ENV['HOME'] + \"/.rvm/bin\"`, since that's where `r-` is installed.",
+      "reference": "I think you need `-y` here to prevent a user confirmation prompt"
+    },
+    {
+      "prediction": "This should be `bool dynamo_avx512_code_in_use = false;`",
+      "reference": "Repeating: I don't think this should be a core-wide global var. This should be isolated to arch/x86 or at least arch/, maybe inside getter/setters as mentioned above."
+    },
+    {
+      "prediction": "I think it would be better to use `JMeterPlugins` here, as it's the name of the plugin we're looking for. Also, I think it would be better to use `JMeterPlugins140` instead of `1.4.0` as it's the latest version of the plugin.",
+      "reference": "right URL is like in cmdrunner, using search.maven .org"
+    },
+    {
+      "prediction": "I think this should be `endpoint` instead of `endpointString`.",
+      "reference": "The link to valid syntax is not really applicable for NET2_1, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/version` since that's where the versioning code lives.",
+      "reference": "Shouldn't it be more specific, like \"gcppubsub supports all UTF-8 strings\"?"
+    },
+    {
+      "prediction": "I think we can remove the `slave` variable from the `CloudJettyRunner` constructor, since it's not used anymore.",
+      "reference": "Everything in this class is SolrCloud-related, not legacy replication"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `drbbdup_get_tls_raw_slot_val()`, but I'm not sure why it is called `merge_arith_flags()` instead of `arith_merge_flags()`. Can you explain why it is called `merge_arith_flags()` instead of `arith_merge_flags()`? I think it should be `arith_merge_flags()`, but I'm not sure why it is called `merge_arith_flags()` instead of `arith_merge_flags()`. Can you explain why it is called `merge_arith_flags()` instead of `arith_merge_flags()`?",
+      "reference": "Could just assign directly and eliminate the `newval` var."
+    },
+    {
+      "prediction": "I'm not sure if this is a good name for this. I think it's a bit confusing because it's not just about the title, but also about the pull request itself. Maybe `pullRequestTitleRegexp` would be better?",
+      "reference": "... with the name of the affected package, or \"all\", followed by a colon,..."
+    },
+    {
+      "prediction": "I think this should be `SynchronizeVersion(\"src/Datadog.Trace.ClrProfiler.Managed.Core/Datadog.Trace.ClrProfiler.Managed.Core.csproj\", NugetVersionReplace);` and not `SynchronizeVersion(\"src/Datadog.Trace.ClrProfiler.Managed.Core/Datadog.Trace.ClrProfiler.Managed.Core.csproj\", NugetVersionReplace);`",
+      "reference": "We'll want to remove this one if we make the assembly version constant."
+    },
+    {
+      "prediction": "This should be moved to the `app/Http/Redirects` folder.",
+      "reference": "Can we have a single function here that uses the same subscribers and just changes meta based on context?"
+    },
+    {
+      "prediction": "I think this should be called `load_kernels_and_setup_function_symbol_map`",
+      "reference": "where would ` static startup_kernel_loader skl;` be instantiated? if it's not instantiated anywhere should this be removed?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "I think we should move those util functions that work with drag-drop to @uppy/utils, so they can be shared (maybe later) with drag-drop plugin? Otherwise it will continue to depend on drag-drop module."
+    },
+    {
+      "prediction": "I think we can remove the `spec.VolumeLabels` and `spec.ReplicaSet.Nodes` here.",
+      "reference": "Migrate the spec.Labels to locator.Labels ?"
+    },
+    {
+      "prediction": "I think this is the only place where `wantErr` is used.",
+      "reference": "The test case already existed but it never tested properly because of the conditional in the test."
+    },
+    {
+      "prediction": "I think this should be `return Objects.equals(sourceId, that.sourceId) && name.equals(that.name) && transform.equals(that.transform);`",
+      "reference": "I'm okay with this, but I don't see a lot of benefit to removing unnecessary parens. If extra parens make something more readable (like this) or clarify order of operations even when matching the default, I would say we should keep them."
+    },
+    {
+      "prediction": "I think this should be `protected final DynamicField[] dynamicFields = new DynamicField[0];`",
+      "reference": "@sarowe why was this volatile? It's fishy to see this as the only volatile field."
+    },
+    {
+      "prediction": "I don't think this is the right package name for this class.",
+      "reference": "template is not correct?"
+    },
+    {
+      "prediction": "I think it's better to move this file to `mmdet/sampler/assigner`.",
+      "reference": "Rename the registies to `BBOX_ASSIGNERS` and `BBOX_SAMPLERS` to avoid ambiguity. There is also a registry for dataset sampler."
+    },
+    {
+      "prediction": "`FakeFS.activate!` should be moved to the top of the file.",
+      "reference": "This is no longer stubbed on every test. Perhaps it should be, and should be unstubbed in the single case that it matters."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this, but I don't have a better idea off the top of my head.",
+      "reference": "This seems much simpler! I'm not sure I follow what you mean by the threading part... Also tests are not liking this constant. Might need a require or to fully qualify this."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this check. I think it should be in `i#2144`.",
+      "reference": "Convention is \"FIXME: i#2144\" or \"XXX: i#2144\""
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this interface. I think it should be in `Datadog.Trace.ClrProfiler.Interfaces`, but I'm not sure if that's the right place to put it.",
+      "reference": "This should probably be a property instead of a method."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to check for the state to be RECONCILED, but it's the only way I could find to check for the state to be RECONCILED.",
+      "reference": "do we need to consider setting the `state` as well? If so, lot of changes might be required, like, setting to 'Reconciling' in upgrade code, setting to 'error' or 'reconciled' in volumeReconciler functions."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Line is too long. [82/80]"
+    },
+    {
+      "prediction": "I think this should be `MIT License`.",
+      "reference": "Add blank line below license header."
+    },
+    {
+      "prediction": "I think it would be better to move this logic to `TableMetadata#buildReplacement`",
+      "reference": "I'm not clear on whether this really should be the right behavior. Basically we're saying that a replace table will keep the existing location (as opposed to using defaults). I suspect we don't have create or replace with location semantics, but this is making some assumptions that a replacement is somehow the same as the old. If we were to go with id based pathing convention, this wouldn't work. I don't think this is an issue at this point, but it might make sense to push this down to the location provider."
+    },
+    {
+      "prediction": "I think this should be `github.com/temporalio/temporal/.gen/proto/persistenceblobs` and not `github.com/temporalio/temporal/.gen/proto/replication`",
+      "reference": "Just run a global replacement for all `executionproto`."
+    },
+    {
+      "prediction": "I think we can remove the `file_detector` argument from this class.",
+      "reference": "@AutomatedTester @davehunt thoughts on a new keyword argument?"
+    },
+    {
+      "prediction": "I'm not sure this is the right error code to return here. It's not an error in the sense that the caller should not have expected it, but it's also not an error in the sense that the caller should not have been expecting it. It's an error in the sense that the device driver should not be able to handle it, and it's also an error in the sense that the caller should not have expected it.",
+      "reference": "Should line 47 be OPAE_ERR?"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "Not necessary, just write `time.Time{}`"
+    },
+    {
+      "prediction": "I think this should be `tabbar-size` instead of `tabbar-height`.",
+      "reference": "As these two strings simply get concatenated for the docs, there's a space missing after the dot here."
+    },
+    {
+      "prediction": "I think we can remove the `if (IntegrationOptions<TIntegration, TTarget>.IsIntegrationEnabled)` check here, since it's already done in the `return BeginMethodHandler<TIntegration, TTarget>.Invoke(instance);` line above.",
+      "reference": "What about integrations where we don't plug ourselved on OnMethodBegin?"
+    },
+    {
+      "prediction": "I don't think this is a good name for this class. It's not just checking that an element is visible, it's checking that an element is visible and interactable. I think something like `expect_any_element_is_interactable` would be better.",
+      "reference": "shouldn't **call** return a boolean?"
+    },
+    {
+      "prediction": "I don't think this is necessary. The `Interface` interface is a superset of the `com.google.api.tools.framework.model.Interface` interface, so it should have all of the methods and properties from the `com.google.api.tools.framework.model.Interface` interface. It's not necessary to implement all of the methods and properties from the `com.google.api.tools.framework.model.Interface` interface, but it's not necessary to implement any of the methods and properties from the `com.google.api.tools.framework.model.Interface` interface.",
+      "reference": "ViewModel classes should not expose any classes from framework.model."
+    },
+    {
+      "prediction": "I think this test is no longer necessary, but I don't know why it was removed.",
+      "reference": "how would you feel about keeping these tests to be more explicit and having them `assert not _validate...`? or is your view that that is too duplicative of the tests in `test_evaluator.py`?"
+    },
+    {
+      "prediction": "I think this should be `--tags` instead of `--format`",
+      "reference": "Why do we use the `winrm_nodes` variable here and not the string `\"winrm_nodes\"` as we were using before?"
+    },
+    {
+      "prediction": "I think it's better to use `vesoft` instead of `vesoftinc`",
+      "reference": "why update Copyright ?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this check. I think it should be in `Scan`'s `bindNode` method.",
+      "reference": "Does this mean we cannot update statistics with sampling on a table with LOB columns?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in `NLog.LayoutRendererBase`",
+      "reference": "Look good, but I think a `[flags]` enum would be more future proof? Although i'm not sure if a flags-enum is working from the XML config. What do you think?"
+    },
+    {
+      "prediction": "I think this should be `if (row.treeLevel > -1 || (row.treeNode.children && row.treeNode.children.length > 0)) {`",
+      "reference": "Since I believe state will never be anything other than 'expanded' or collapse, you can rewrite this as follows: `return row.treeNode.state === 'expanded' ? 'ui-grid-icon-minus-squared' : 'ui-grid-icon-plus-squared'; `"
+    },
+    {
+      "prediction": "I'm not sure if we should remove this, but I don't think it's a good idea to change the default behavior of the `SolrClient` in the `SolrHttpClient`. I think it would be better to keep the default behavior of the `SolrClient` in the `SolrHttpClient` and let the `SolrHttpClient` handle the default behavior of the `SolrClient` in the `SolrHttpClient`.",
+      "reference": "the latter part can go but isn't the first part of this still sound -- that we assume \"id\"?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `@` and `$` are escaped by the `String.prototype.replace` method, not by the `String.prototype` method. So the `@` and `$` end up in the output. The fix should be to replace the `String.prototype.replace` method with the `String.prototype.replaceAll` method, and then remove the `@` and `$` special characters from the `value` before calling `String.prototype.join`.",
+      "reference": "Why is the escaping of @ and $ removed?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this interface. I think it should be in `Datadog.Trace.Interfaces`.",
+      "reference": "Nit: `Http` is redundant in this interface's name."
+    },
+    {
+      "prediction": "I think this should be `store.Put` instead of `store.Exists`?",
+      "reference": "what if I set SynchronizableEntitiesPrefix to \"a/b/c/d\" in config?"
+    },
+    {
+      "prediction": "I think this should be `opts.DockerfilePath != \"\"` instead of `opts.DockerfilePath != \"\" && isDir`",
+      "reference": "nit: This error message doesn't mention that the path is a directory, maybe \"Dockerfile path is a directory:%s, please provide path to file.\""
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It's not an issue with the software itself, but with the hardware. We should be looking at the hardware and not the software.",
+      "reference": "This is done automatically, you can skip it next time ;)"
+    },
+    {
+      "prediction": "I think this should be `self._elem.dispatchEvent(new Event(event))` instead of `self._elem.evaluateJavaScript(\"this.dispatchEvent(new Event('{}'))\")`",
+      "reference": "This is needed in `set_value` because of `:open-editor` (you could open an editor, close the tab, then close the editor). I don't think it makes any sense to have it here?"
+    },
+    {
+      "prediction": "I'm not sure if this should be `profiling_frequency` or `profiling_period` or `profiling_interval` or `profiling_periodic_interval` or `profiling_frequency_periodic_interval` or `profiling_periodic` or `profiling` or `profiling_interval` or `profiling_period` or `profiling` or `profiling_periodic` or `profiling_periodic_interval` or `profiling_interval` or `profiling` or `profiling_periodic` or `profiling_periodic_interval` or `profiling_interval` or `profiling` or `profiling_periodic` or `profiling_periodic_interval` or `profiling_interval` or `profiling` or `profiling_periodic` or `profiling_periodic_interval` or `profiling_interval` or `profiling` or `profiling_periodic` or `profiling_periodic_interval` or `profiling_interval` or `profiling` or `profiling_periodic` or `profiling_periodic_interval` or `profiling_interval` or `profiling` or `profiling_periodic` or `profiling_periodic_interval` or `profiling_interval` or `profiling` or `prof",
+      "reference": "Perhaps we can assume that profiling is enabled if ProfilingPort is set? And/or configure a default port and frequency, so we don't have to set three config vars every time?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this hack. I think it should go in `lib/beaker/vagrant.rb`",
+      "reference": "This nil check won't correctly handle the case where forward_ssh_agent is set to false - it won't be nil but it should still not be set."
+    },
+    {
+      "prediction": "I think this should be `protoVersion` instead of `currentVersion`",
+      "reference": "We could avoid passing proto since threshold isn't needed."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that we don't allow connections from the same IP address to the same service to the same location oracle at the same time. If we don't do that, then we'll end up with multiple connections from the same IP address to the same service to the same location oracle at the same time.",
+      "reference": "This route excluding needed for reconnect only when one connection is lost and we trying to establish another. I was thinking to make 2 separate flags for killswitch and reconnect enabling. This way we can keep killswitch working, but we don't need these routes excluded."
+    },
+    {
+      "prediction": "I think we can remove the `LOG(INFO)`.",
+      "reference": "why can this modification avoid crash?"
+    },
+    {
+      "prediction": "I think we can remove `current_user` from `@account` since it's no longer needed.",
+      "reference": "This looks good -- signing out the user if the current user is disabled"
+    },
+    {
+      "prediction": "I think this should be `The collection of the selection layer levels where the whole row was selected using the row header or the corner header.`",
+      "reference": "The line exceeds 120 characters."
+    },
+    {
+      "prediction": "I think this should be `output = output + fmt.Sprint(siteInfo)` instead of `output = output + siteInfo`",
+      "reference": "Let's go ahead and change \"Site\" to \"Project\", since that's the path we've chosen. One less thing to alter in the other issue."
+    },
+    {
+      "prediction": "I think we should remove this line too. It's not used anymore.",
+      "reference": "I am for this change iff we are sure the information is no longer helpful. I think this could still be valuable for certain scenarios, wbu?"
+    },
+    {
+      "prediction": "I think you can use `Comparator.comparing(ImageRampup::getRampupPercentage, Ordering.natural().reverse())` here instead of `Comparator.comparing(ImageRampup::getRampupPercentage, Ordering.natural().reverse())`",
+      "reference": "I thought we decided on using the latest active version and not the one which is max ramped up."
+    },
+    {
+      "prediction": "I think this should be `%s to interrupt, create new public subnets, then rerun %s and import them.`",
+      "reference": "can we update this to a `log.Warningf`?"
+    },
+    {
+      "prediction": "I think this should be `subprocess.run` instead of `subprocess.check_call` so we don't fail when trying to disconnect a container from a network that it's not connected to.",
+      "reference": "can we get something emitted in the logs on non-zero exits to trace back to for problems like this failing on the first invocation?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `AddVerticesProcessor` class.",
+      "reference": "Do we need `schema manager` at here ?"
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `C.CString(\"\")`",
+      "reference": "This won't work because of the permissions on the directory when this is included as a library. But, we may be able to do this. The key would be wrapping this command with either a script or a go binary that ensures the current directory is writable. If the current directory is not writable, it would attempt to make it writable so that we could write out the produced binary. But I'm not sure if that will influence the hash that gets produced for `go.sum` since I am not sure how that is calculated. Alternatively, we can choose a place in `/usr/local` or choose a location that is writable in the go cache and use that. I'll give that a try since it might be needed for the ARM builds if we choose to only include `x86_64` precompiled binaries in the module."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `eval` function in `pkg/repl`.",
+      "reference": "Correct me if I'm wrong, but is this necessary? Why not just use the value returned by `interpreter.Eval` and not mess with the scope? `interpreter.Eval` will return any produced side effects. This means TableObjects as well as any value resulting from any expression statements."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `alertmanager` service.",
+      "reference": "This service is global per namespace, not per Alertmanager/Prometheus/ThanosRuler instance. So this approach won't work. But maybe I missing out something, can you describe your use case here?"
+    },
+    {
+      "prediction": "I think this should be `Attestations`, not `CanReattest`.",
+      "reference": "Do we need this default? Since we aren't using a sql.NullBool or *bool, an unset column will be interpreted as `false` already... We don't set a default on our other bool fields (e.g. entry admin and downstream columns)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `ExportsHelper` since it's only used in `download_plan_page`",
+      "reference": "thanks for moving these over. makes more sense for them to be in the exports_helper"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `qutebrowser/utils/keyinput.py`",
+      "reference": "That import now isn't needed anymore"
+    },
+    {
+      "prediction": "I think this should be `path: full path to the addon that will be installed.`",
+      "reference": "Can you update the `Usage` to be an absolute path as well? Something like `/path/to/firebug.xpi`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `UAP` namespace, since it's specific to UAP.",
+      "reference": "This class is missing the `MvxPlugin` attribute"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the main frame class.",
+      "reference": "What if `wxLang` is returned `None` from `getWxLang`"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure I like the fact that we're passing a `std::vector<std::string>` to the `ingest` function. I think we should pass a `std::vector<std::string>(size_t size)` instead of a `std::vector<std::string>` to the `ingest` function. This way, we don't need to pass a `std::vector<std::string>` to the `ingest` function, and the `ingest` function doesn't have to pass a `std::vector<std::string>` to the `ingest` function.",
+      "reference": "Please don't do changing like this unless it _**really**_ cares. BTW. Please look around to infer our naming conventions."
+    },
+    {
+      "prediction": "I think we should use `log.info` instead of `log.finest` here.",
+      "reference": "This is in test code: understanding what we're doing is important in this context."
+    },
+    {
+      "prediction": "I think it would be better to say something like: `Example: if --min-gas-price is in the list and condition is that either --miner-enabled or --goquorum-compatibility-enabled should not be false, we log a warning.`",
+      "reference": "prob should still have an example that has multiple option names even if you have to make it up. Or maybe we don't need this method?"
+    },
+    {
+      "prediction": "Layout/ExtraSpacing: Unnecessary spacing detected.<br>Layout/SpaceBeforeBlockBraces: Space missing to the left of {.<br>Layout/SpaceInsideBlockBraces: Space between { and | missing.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br>Layout/SpaceInsideBlockBraces: Space missing inside |.<br",
+      "reference": "I believe respond_to whitelists the formats passed to the block so if we don't want to display html, we can just remove the line format.html..."
+    },
+    {
+      "prediction": "I think it would be better to use `fmt.Sprintf` instead of concatenating strings.",
+      "reference": "stylistic nitpick: I _feel_ like we've largely shown errors like this as \"error: \" vs. \"error=\". IMO colon/space reads better."
+    },
+    {
+      "prediction": "I think this should be #include <rtps/flowcontrol/FlowController.h>",
+      "reference": "We could put this implementation in `RTPSWriter::create_change_pool` and avoid an extra function."
+    },
+    {
+      "prediction": "I think this file should be renamed to `persistencetests_test.go`",
+      "reference": "should this file ends with _test.go?"
+    },
+    {
+      "prediction": "I think this should be `inet_pton` instead of `inet_pton`",
+      "reference": "It's because of this change. `ss` was unused and it made sense in the program. However I haven't read the IPv6 RFC so I'm unsure of what it does"
+    },
+    {
+      "prediction": "I think this should be `flattenIndex` instead of `flattenIndex`",
+      "reference": "\"Flatten\" is a verb, so I'd probably go with `flattenedIndex` as a variable name here."
+    },
+    {
+      "prediction": "I think this should be `valuesToMask.push(...)` instead of `valuesToMask.push(...valuesToMask)`",
+      "reference": "If it can only be called once, perhaps a more appropriate name is something like `setMaskables`? `addX` sounds like you can add many `X`es by calling it many times"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with that doesn't break anything.",
+      "reference": "Nodecfg should be downloading from a private bucket -- these shouldn't be for public consumption. In general these should be generic and expect environment to provide appropriate credentials and bucket."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `appModuleHandler` module, since that's where the `AppModuleHandler` class is defined.",
+      "reference": "Could you please end the file with an empty line?"
+    },
+    {
+      "prediction": "I think this should be `LOG.info(String.format(\"Detected dialect: %s\", toReturn.dialect.toString()));`",
+      "reference": "This is an incorrect change. The dialect spoken is an important part of the handshake and should be communicated to users."
+    },
+    {
+      "prediction": "I think this should be `bool` instead of `sync.Mutex`.",
+      "reference": "Can this be non-exported to avoid direct use? It would be a breaking change but using previous package version would work."
+    },
+    {
+      "prediction": "I think we can just return `gcerrors.InvalidArgument` here, since we don't care if it's an error or not.",
+      "reference": "Nit: this is just `return gcerrors.Code(err)`, isn't it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `CreatePayments` call, as it is the caller of the `CreatePayments` call that needs to wait for the payment to be mined.",
+      "reference": "I don't know where this came from, but it's not a good idea. This is actually timing out after 5 rounds, not 5 blocks. 5 consecutive null blocks won't be that uncommon. Also when testing with a short block time, this is a very short duration that can contribute to flaky tests."
+    },
+    {
+      "prediction": "I think this should be `with open(statePath, \"r\", encoding=\"utf-8\") as f:`",
+      "reference": "In Python 3 when pickling or unpickling objects, the file needs to be opened as binary so that no text encoding/decoding takes place. So for any open calls around pickle loads or dumps, the mode for reading must be rb and the mode for writing must be wb."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this validation. I think it should be in the `fetchMergedEventGroups` method.",
+      "reference": "did you remove **params.qstring.method** intentionally? if so why?"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for this value instead of hardcoding the default value.",
+      "reference": "What is the current default? Is that inconsistent across uses and that's why this is null?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `deploy` method of the `CloudFormation` provider, since that's where the `CloudFormation` provider is instantiated.",
+      "reference": "Why did we make this change? How come it wasn't an issue before"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Can you tighten scope by moving to of these variables? i.e. Move to ~4372?"
+    },
+    {
+      "prediction": "I think this should be `System.getProperty(\"dataverse.rserve.password\")` instead of `System.getProperty(\"dataverse.rserve.pwrd\")`",
+      "reference": "Nice to see this `dataverse.rserve.password` fix rolled in."
+    },
+    {
+      "prediction": "I think this should be `strnlen_s(tmp_optarg, sizeof(tmp_optarg) - 1)` instead of `strnlen_s(tmp_optarg, sizeof(tmp_optarg))`.",
+      "reference": "How about the following faster alternative that doesn't need `strlen()` at all: if (*tmp_optarg == '\\0' || *endptr != '\\0') { fprintf(...) ... This would treat the bus argument as invalid if it's either empty (`tmp_optarg` points to '\\0') or invalid (`endptr` points to something other that a '\\0'). Actually, the existing code would accept an empty argument and set the bus to `0`, so if (*endptr != '\\0') { fprintf(...) ... would be enough. The check for an empty argument doesn't hurt, though."
+    },
+    {
+      "prediction": "I think it would be better to add `bcc: self.class.bcc_email` here.",
+      "reference": "I could be missing something here, but the `from` and `cc` are the same. On the test email sent to gatewaycommunicator, these values are different."
+    },
+    {
+      "prediction": "I think this test is still a bit out of date, since it doesn't test that `cleanupFunction` is not called.",
+      "reference": "I'm honestly scared because act is now a hard requirement for useEffect which it wasn't before... This could break some tests :("
+    },
+    {
+      "prediction": "Shouldn't this be `&block_a` instead of `&block_a;`?",
+      "reference": "Could use election from `info_a.election`, or is this deliberate?"
+    },
+    {
+      "prediction": "I think this should be `LegacyFormHelper::getType('form')` instead of `[LegacyFormHelper::getType('form')]`",
+      "reference": "looks like this should be `return array(LegacyFormHelper::getType('form'));`"
+    },
+    {
+      "prediction": "I think this should be `assert(typeof opt_port !== 'function', `",
+      "reference": "Maybe it would simpler to ignore opt_port if type !== 'number'?"
+    },
+    {
+      "prediction": "I think this should be `name = categoryNames[i];` instead of `name = categoryNames[i];`",
+      "reference": "Where was `name` being declared before?"
+    },
+    {
+      "prediction": "I think this should be `UnknownDeployment` instead of `Unknown`",
+      "reference": "nit: use `corev1.ConditionUnknown`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for validity, but it's the only one I could come up with.",
+      "reference": "You also need to add a similar check on line 992, there's another place where we check `length === 24`"
+    },
+    {
+      "prediction": "I think this should be `validatorProvider.getVoteProviderAfterBlock(parentHeader).get().getVoteAfterBlock(parentHeader, localAddress);`",
+      "reference": "nit: can extract the voteProvider as a local variable"
+    },
+    {
+      "prediction": "I think this test is still valid, but I think it would be better to move it to a new test.",
+      "reference": "[0] Not suggesting you change it here, but....kindof weird that there's just not an `assertFuncNotEquals`"
+    },
+    {
+      "prediction": "I think it would be better to have a separate function for this.",
+      "reference": "why are we passing 'registryAddress' here? We probably should construct it together with bc. Registry is not something that change?"
+    },
+    {
+      "prediction": "I think this should be `RA.spreadPath`",
+      "reference": "`.` at the end of the sentence"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "wouldn't it be useful to split this into an error metric for the timing?"
+    },
+    {
+      "prediction": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Not sure if this is good enough. Each generator will have it's own unit test anyway. Ideally, I think I'd want to test that each generator in the yml is actually a valid generator..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this.",
+      "reference": "line is 138 characters"
+    },
+    {
+      "prediction": "I think it would be better to call this `cleanup` instead of `cleanupFunc`.",
+      "reference": "Nit: I would leave `Wire` out of the description; if you use this without wire you can still use it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `com.google.api.tools.codegen.viewmodel`",
+      "reference": "Nit: each field can be on a separate line to make it visually easier to read."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be in `compiler_info.json`",
+      "reference": "Do we still have these files? Shouldn't we remove these?"
+    },
+    {
+      "prediction": "I think this should be `ErrorRenewBeforeInvalid`",
+      "reference": "A lot of things to do with constants seemed to move since the original commit, so I stuck this here, Is there a better place for it?"
+    },
+    {
+      "prediction": "I think we can remove the `BlockChanSize` and `BlockSyncChanSize` and use the `ActionChanSize` and `ActionSyncChanSize` instead.",
+      "reference": "is 10 too small compared to 1000? consider sync chan is unicast only (vs block chan is broadcast + unicast), i would say use 200~400 for BlockSyncChanSize my concern is that this would potentially slowdown sync speed of full-node"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this logic into `get_plan_info` so that we don't have to call it every time we want to know if a plan is private.",
+      "reference": "A full parse of the plan here on listing the plans will be expensive from both a computation and IO perspective."
+    },
+    {
+      "prediction": "This should be `BUTTON5` and `BUTTON6` instead of `BUTTON4` and `BUTTON3`.",
+      "reference": "These constants use the `_PIN` suffix, while the other constants don't use it. I'm not sure what is best, but I would prefer to keep this consistent. Do you have an opinion on which it should be (with or without suffix)?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. This is a command-line parameter, and it should be configurable by the user.",
+      "reference": "Hmm... this feels like discovery should come first. perhaps `--discovery-dns-url`? @NicolasMassart any opinions on this or ideas on who it should be run by?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "why is this needed?"
+    },
+    {
+      "prediction": "I think it would be better to keep the `name` and `type_attributes` as `ConfigType.__cache` attributes, and just add a `key` argument to `ConfigType.__init__`. This way, we don't need to worry about clearing `ConfigType.__cache` when we add a new type, we just need to worry about clearing `ConfigType.__cache` when we remove an existing type.",
+      "reference": "I wonder if it'd be possible to autogenerate a key from the name within this function if one is not provided explicitly, rather than having all the callsites pass both the name and key (and usually as the same value)? Might give us a good place to implement a `name->key` function that isn't 1:1."
+    },
+    {
+      "prediction": "I think this should be `TYPE_I64` instead of `TYPE_STRING`.",
+      "reference": "@adriancole I do not see a change in the thrift file. Are there unit tests verifying that this manual serialization is compatible with the native Thrift serialization done by classes generated from `.thrift` IDL file?"
+    },
+    {
+      "prediction": "This should be `return RETURNVALUE_PLAYERWITHTHISNAMEISNOTONLINE;`",
+      "reference": "spelling (variable name, all three changes)"
+    },
+    {
+      "prediction": "I think this should be `keybase1.LoadTeamPlusKeys`",
+      "reference": "I believe you meant \"specify `desiredUID` and `desiredRole`\"?"
+    },
+    {
+      "prediction": "I think we should use `v1beta1` for consistency with the rest of the codebase.",
+      "reference": "Please follow the import style, move it to its similar group"
+    },
+    {
+      "prediction": "I think this should be `errors.RouterUnrecognizedEncodingError(want, name)` instead of `errors.RouterUnrecognizedEncodingError(want, string(req.Encoding))`",
+      "reference": "Do we really need these? This is expanding on an API that we're about to do work on with the error stuff @kriskowal"
+    },
+    {
+      "prediction": "This method is not used anymore, so I think it should be removed.",
+      "reference": "passwordModificationTime is no longer used - did the change drop a time check that should be restored? Or should the param get dropped from the methods?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `commandLine` can be empty, so we can't compare it to an empty string. I think we need to do something like `if (string.IsNullOrEmpty(evt.CommandLine) || processName.Equals(evt.ProcessName, StringComparison.OrdinalIgnoreCase) && pids.Contains(evt.ProcessID) && evt.CommandLine.Trim() == commandLine.Trim())`",
+      "reference": "Would it break here without trim? If so, can we do trim in Startup.cs so we don't need to add this code to every parser?"
+    },
+    {
+      "prediction": "nit: I think you can remove the `log.Printf` call here.",
+      "reference": "Since we're using Logrus, probably best to use `Infof` to avoid confusion (Logrus' Printf behaves differently to the built in one)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `createShadowRoot_` function.",
+      "reference": "Hm, this looks like a change that should go upstream as well."
+    },
+    {
+      "prediction": "I think we can remove this function and use `GetActionByActionHash` instead.",
+      "reference": "let's keep ActionByActionHash and delete getActionByActionHash"
+    },
+    {
+      "prediction": "Shouldn't this be `PricingGroupInterface`?",
+      "reference": "please use return type"
+    },
+    {
+      "prediction": "I think this should be `fine` instead of `fine`",
+      "reference": "This is deliberately at this level."
+    },
+    {
+      "prediction": "I think this should be `as msg` instead of `as %s`",
+      "reference": "I'm going to go ahead and undo this change, I don't think it was intentional."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to say it, but it's the only way I could think of.",
+      "reference": "> specialised This isn't a typo in my neck of the woods"
+    },
+    {
+      "prediction": "I think this should be `github.com/google/knative-gcp/test/cmd/storageTarget/target`",
+      "reference": "will change to `storage_target`"
+    },
+    {
+      "prediction": "I think we can remove `cls.securitycenter = securitycenter.SecurityCenterClient(version='v1')` now.",
+      "reference": "This would be better as `cls.securitycenter_client`, to match what is being instantiated."
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. I think we should be using the `sys` module to get the path of the native system32 directory, and the `os` module to get the path of the syswow64 directory if it exists on the current system.",
+      "reference": "Could we have this initialization as part of the initialize method?"
+    },
+    {
+      "prediction": "This should be `this.controllerFor('patients/edit')`",
+      "reference": "This line should be removed as editController is not used from this context anymore."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `plan/transformations` since it's only used by `plan/transformations.go` and `plan/transformations_test.go`.",
+      "reference": "This is correct in concept but the method signature doesn't really communicate the same thing. In other words, if the function produces an output table where is it? Its not on the function signature. Can you explain where the output table is created?"
+    },
+    {
+      "prediction": "`self.pretrained = init_cfg['checkpoint']` --> `self.pretrained = init_cfg.get('checkpoint', None)`",
+      "reference": "For insurance, it's best not to take it directly"
+    },
+    {
+      "prediction": "I think this method is not used anywhere, can we remove it?",
+      "reference": "change 'value' to keysToSend, here and in other references in this commit. 'value' implies the user is getting the value of the element, rather than just the keys we're sending to it."
+    },
+    {
+      "prediction": "I think this should be `engine.state.AddTaskIPAddress(taskIP, task.Arn)`",
+      "reference": "is the plan to fully migrate to boltdb and then remove the state save here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the styles.js file, since it's only used there.",
+      "reference": "This should be `output.colorBright`. Unless I'm missing some reason for adding a new theme variable."
+    },
+    {
+      "prediction": "I don't think we need this. We already have a `version.py` file in the root of the repo. You can add the following to `setup.py` in the root of the repo: `python_requires='>=3.5,<3.7'` And then you don't need the `version.py` file in the root of the repo.",
+      "reference": "nit: I wonder if these should be moved down to where version is defined, just to keep them all in the same place."
+    },
+    {
+      "prediction": "I think this should be `import java.lang.*;\\nimport java.util.*;\\nimport javax.annotation.*;\\n` and not `import java.lang.*;\\nimport java.util.*;\\nimport javax.annotation.*;\\n`",
+      "reference": "Is this supported by every implementation of Java (and Android)? Should it be conditional upon `gen_nullable`?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "@briri Thanks! I just caught this too updating my feature branch"
+    },
+    {
+      "prediction": "I think this should be `envpw != \"\"` instead of `envpw != \"\" && len(configKey) == 0`",
+      "reference": "I think you should move this block (and the `envpw := os.Getenv(\"RCLONE_CONFIG_PASS\")`) right to the start of the function, then we can remove the duplicate code below"
+    },
+    {
+      "prediction": "I think this test should be moved to `RemoteNodeTest` instead of `NodeTest`",
+      "reference": "Is this only to set the node to draining?   I think we can do that inside `Before` or something?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this logic to `AppServicesInstrumentation` since it's specific to Asp.Net.",
+      "reference": "This should have been moved up above this inner `try` because the log message in the `catch` block does not apply to this."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "Do we expect people to use PlanResult directly?"
+    },
+    {
+      "prediction": "I think this should be `return wait.ExponentialBackoff(backoff, func() (bool, error) { return condition() })`",
+      "reference": "Maybe we should call this something like `errToReturn`?"
+    },
+    {
+      "prediction": "Why do you need to check `epochNum >= fairBankEpochNum+a.foundationBonusLastEpoch`?",
+      "reference": "should calculate numEpochIn1Year = xxx and <= fairBankEpochNum+numEpochIn1Year a.foundationBonusLastEpoch just happens to be equal to 1 year now, but should not count on that"
+    },
+    {
+      "prediction": "I think this should be `contentTypes: ['embedded', 'phrasing', 'flow']`",
+      "reference": "I wasn't sure if the content type needed to be removed from both the `default` and `usemap` objects - I'm not sure how usemap is used."
+    },
+    {
+      "prediction": "I don't think you need to import `struct` here.",
+      "reference": "Duplication of line 28 Please refer your tox -e flake8 It seems that XStrField, XShortField, FieldLenField are not used"
+    },
+    {
+      "prediction": "I think this should be `self._file_view.setRowHidden(index.row(), path, hidden)` instead of `self._file_view.setRowHidden(index.row(), index.parent(), hidden)`",
+      "reference": "I don't really understand the `self._file_model.index(path)` as parent here - you use `self._root_index` for `rowCount` above, so wouldn't the parent here need to be `self._root_index` as well?"
+    },
+    {
+      "prediction": "Should this be `maxQueryLength` or `minQueryLength`?",
+      "reference": "Do we want validation somewhere that max >= min? Or simply treat the max < min case as ignoring max?"
+    },
+    {
+      "prediction": "I think this should be `ndc:topframes` instead of `${ndc:topframes=1}`",
+      "reference": ":+1: that's a easier example :)"
+    },
+    {
+      "prediction": "Why do we need this method?",
+      "reference": "If an external module decorates this interface, such a change will cause it to generate an error"
+    },
+    {
+      "prediction": "(style) 'FN' may be misspelled - perhaps 'FN'? (style) 'FN' may be misspelled - perhaps 'FN'?",
+      "reference": "Oh ha hah, FNM_PERIOD worked out nicely there. Points for co-opting a file system convention."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `github.com/elastic/beats/libbeat/cmd/elastic`",
+      "reference": "wrong grouping. As you can tell from the other files, we put system packages in the first group, the 3rd party packages in the second group, and our own packages in the third group."
+    },
+    {
+      "prediction": "I think this should be `if _, err := o.configStore.GetJob(o.appName, o.name); err != nil {`",
+      "reference": "Is `o.name` always set here?"
+    },
+    {
+      "prediction": "I think this should be `verifyStatusResolvable(target, nameMap, target.SteadyStateDependencies, onSteadyStateCanResolve)` instead of `verifyStatusResolvable(target, nameMap, target.SteadyStateDependencies, onSteadyStateCanResolve)`",
+      "reference": "Can this line be removed, as `SteadyStateDependencies` isn't updated anywhere? Also do we need to check the `TransitionDependencySet` here?"
+    },
+    {
+      "prediction": "Metrics/CyclomaticComplexity: Cyclomatic complexity for has_active_subscription is too high. [7/6]<br>Metrics/PerceivedComplexity: Perceived complexity for has_active_subscription is too high. [8/6]",
+      "reference": "`plan.present?` and `has_active_subscription?` are equivalent. We can drop this change, while leaving the one on `app/controllers/application_controller.rb`."
+    },
+    {
+      "prediction": "I think this should be `too-many-arguments`. We don't want to allow more than 29 positional arguments.",
+      "reference": "No need for this, as pylint already only turns things off for this function and it's needed for the entire function."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure we don't informer for the `alertmanager` namespace if it's already informer for the `alertmanager_config` namespace.",
+      "reference": "L171 needs to be modified too?"
+    },
+    {
+      "prediction": "(style) 'e' may be misspelled - perhaps 'it'? (style) 'e' may be misspelled - perhaps 'it'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (style) 'e' may be misspelled - perhaps 'ith'? (",
+      "reference": "Why do you compare by nodeid? Simply saying d == e should be enough. Unless we have to objects in memory with the same nodeid. If that is so, something went really wrong."
+    },
+    {
+      "prediction": "I don't think we need to add `coco_detection` here, since it is not used in RetinaNet.",
+      "reference": "configs/pvt/retinanet_pvt_t_fpn_1x_coco.py --> configs/pvt/retinanet_pvt-t_fpn_1x_coco.py"
+    },
+    {
+      "prediction": "I'm not sure if we should have a `DesiredStatus` field here. I think it's fine to have it, but I'm not sure if it makes sense to have it in the `TaskResponse` struct at all.",
+      "reference": "Nit, this could be json:\",omitempty\""
+    },
+    {
+      "prediction": "I think this should be `BuildRequiredFunc` or `BuildRequired`.",
+      "reference": "It's kinda funny to read `s.Image.Image` I wonder why it resulted to this way. What would it take to refactor to just `s.Image`?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "nit: I like the idea! can we instead change the tag to `integration-remote` to be more specific? I feel like it's possible for us to use the `deploy` build tag for possibly another scenario"
+    },
+    {
+      "prediction": "This should be `int32 param_1 = (int32)SVIV(ST(1));`",
+      "reference": "Copy and paste error here. Edit: Also missing commas here and in other Perl croaks between some parameters and not others."
+    },
+    {
+      "prediction": "I don't think we need this interface. We can just use `IPendingTxSelector` in the `PendingTxSelector.cs` file.",
+      "reference": "stateRoot or maybe ParentHeader would make more sense from API point of view?"
+    },
+    {
+      "prediction": "Why is this removed? It seems to be the only place where we are calling `callback.setupNotification(false, info);`",
+      "reference": "Are you sure that this is no longer needed?"
+    },
+    {
+      "prediction": "I think it would be better to have this list as a static field in `BaseTableScan` instead of adding it to `SCAN_COLUMNS` and `SCAN_WITH_STATS_COLUMNS`. The reason is that `BaseTableScan` is used in multiple places, and having this list as a static field ensures that it is always up-to-date.",
+      "reference": "Would it help to use SCAN_COLUMNS as base to build SCAN_WITH_STATS_COLUMNS e.g like so `SCAN_WITHSTATS_COLUMNS = ImmutableList.<String>builder().addAll(SCAN_COLUMNS).add(\"value_counts\",....).build()` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `piped/monitoring` since it's related to monitoring.",
+      "reference": "`ctx` is unused in ReportStat"
+    },
+    {
+      "prediction": "Please don't include `list` and `map` in `boost` headers. They are not standard C++ headers, and should not be used in `boost` headers. Instead, use `boost::list` and `boost::map` from the `boost` header.",
+      "reference": "We don't want to add boost as a dependency when building the Thrift compiler. Sorry."
+    },
+    {
+      "prediction": "This should be `int32_t minSpeedChange = 0;` and `int32_t maxSpeedChange = INT32_MAX;`",
+      "reference": "It's not printing the range. Also I think it's useful to print a warning above stating that the minimum change is -1000."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think we should move this to `cluster/v1beta1/cluster` or `cluster/v1beta1/internal/cluster`",
+      "reference": "if name is empty will this become enumerate ? (and cause the unmarshal to fail ?)"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [122/120]",
+      "reference": "How about including a flash message that explains the exercises are only available to subscribers of X plan?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. The problem is that we don't have a way to tell if the key was a backspace or something else, so we can't tell if we should leave the window or not. I'm not sure what the best way to do this is, so I'm not sure this is the best way to do it.",
+      "reference": "You should also call `e.accept()` and `return` so the key press isn't processed further (as we're leaving insert mode anyways)."
+    },
+    {
+      "prediction": "I think this should be `ENV.fetch(\"EMAIL_RECIPIENTS\", \"noreply.sendgrid.net\")`",
+      "reference": "Freeze mutable objects assigned to constants."
+    },
+    {
+      "prediction": "I think this should be `await expect( page ).toClick( '#googlesitekit-opt-in' );`",
+      "reference": "Hmm, this feels hacky. Maybe good enough if it makes the test more stable, but why is timing even an aspect here, since below it should wait for these two things anyway?"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to do this, but I don't have a better suggestion.",
+      "reference": "Do we want this on by default, or an opt-in (e.g. via some config while add the instrumentation)? I guess a more general question is - what's the bar for the default vs opt-in tags."
+    },
+    {
+      "prediction": "I think this should be `graph` instead of `status`",
+      "reference": "`status` is just one of the features, named as status is not suitable."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. Feeds are filtered by the user, not the system. Feeds are filtered by the system, not the user. Feeds are filtered by the user, not the system. Feeds are filtered by the system, not the user. Feeds are filtered by the user, not the system. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are filtered by the system, not the user. Feeds are",
+      "reference": "There is an option to hide the subscriptions list from the sidebar. If it is hidden, the filter text should not be displayed."
+    },
+    {
+      "prediction": "I think this should be `memoryCircuitBreakerThreshold` instead of `memoryCircuitBreakerThresholds`",
+      "reference": "I'd like for all of this to be dynamically configurable at some point, but it doesn't have to be in this PR. Can add it to the future SIP or create a separate JIRA for it, as you think would be appropriate."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "move the constructor to down the `namespace`"
+    },
+    {
+      "prediction": "I think we should rename this field to `FederatedClusterID` or `FederatedClusterRef`.",
+      "reference": "Nit: ending with a '.' looks consistent with the fields around it."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add the audit fields. I think it should be done in `svidv1.MintJWTSVIDResponse` constructor.",
+      "reference": "Audit log will not have a warning about they are using a deprecated path, is it something we must care about?"
+    },
+    {
+      "prediction": "I think this should be `contract.Visibility != TypeMemberVisibility.FamilyOrAssembly && contract.Visibility != TypeMemberVisibility.Family` instead of `contract.Visibility != TypeMemberVisibility.FamilyAndTypeMemberVisibilityOrAssembly`",
+      "reference": "I think you also want to update the condition to add ` || impl.Visibility == TypeMemberVisibility.FamilyOrAssembly`."
+    },
+    {
+      "prediction": "I think this should be `uint64(100000)`",
+      "reference": "`DepositToRewardingFundBaseGas` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think it would be better to have this be `Development` instead of `Operational` to be consistent with the rest of the CLI.",
+      "reference": "What do you think of \"operations\"?? Also what do these emojis look like on Linux??"
+    },
+    {
+      "prediction": "I think it would be good to have a link to the pricing page in the tagline. I know it's a bit of a stretch, but I think it'll make the tagline a bit more useful.",
+      "reference": "I like the word \"give\" here. Feels like I'm giving a gift to my team."
+    },
+    {
+      "prediction": "I don't think you need to import ` randint` and `uniform` here. You can just use `randint(0, 100)` and `uniform(0, 100)`.",
+      "reference": "Is it possible to use `np.random` module instead?"
+    },
+    {
+      "prediction": "I think this should be `None` by default. If you want to allow `None` then you need to add `allow_none=True` to the docstring.",
+      "reference": "``default=None`` implies ``allow_None`` so ``allow_None`` is superfluous here. As a special case, if allow_None=True (which is true by default if the parameter has a default of None when declared) then a value of None is also allowed."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `system:k3s-controller` instead of `system:k3s-controller`",
+      "reference": "Is there anything in particular that makes setting up the downstream rolebinding(s) to `system:k3s-controller` burdensome or confusing? This changes looks fine to me but it seems a shame to alias an embedded k3s controller. If we are doing this in other places that I am not aware of then we can dismiss this concern out of hand."
+    },
+    {
+      "prediction": "I'm not sure if we want to use `go-acd` or `github.com/ncw/go-acd` here.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I don't think we need this file at all. We can just use `OS.Name` and `OS.Architecture` from `System.Runtime.InteropServices`.",
+      "reference": "nit: all other types in this project have full names, so maybe a better name would be `OperatingSystem`?"
+    },
+    {
+      "prediction": "I think this should be `#include \"MolTransforms.h\"`, not `#include <GraphMol/MolTransforms/MolTransforms.h>`",
+      "reference": "Why is this needed here?"
+    },
+    {
+      "prediction": "I think this should be `throw new UnsupportedOperationException(\"Not yet implemented\");`",
+      "reference": "mmm, why a method declaration should not have the TypeVariableResolutionCapability? Is this ok?"
+    },
+    {
+      "prediction": "I think you can remove the `final` here and in the other constructors as well.",
+      "reference": "nit: any reason you swapped the ordering?"
+    },
+    {
+      "prediction": "\").dispatchEvent(new CustomEvent(\"change\", {\n             dom: page.getElementsByTagName(\"input\")[0]\n         }));\n```\nReview comment?",
+      "reference": "can you de-uglify at least this line?.. hard to tell what changed..."
+    },
+    {
+      "prediction": "I think we should rename `LoadSockopsAuto` to `LoadSockops`.",
+      "reference": "Please can you merge master in before making these changes. I just reinstated a bunch of BPF UTs. Possible that the UTs use this code."
+    },
+    {
+      "prediction": "I don't think this is a regression test.",
+      "reference": "Do you want to add additional cases for `Bla.__b()` and `self.__c()`? (Just add additional classmethods `__b` and `__c` and the calls to `__init__`)"
+    },
+    {
+      "prediction": "I think this should be `len(recids)`, not `len(get_current_user_records_that_can_be_displayed(qid))`.",
+      "reference": "the recids is never changed after. So, it should contains the exact shown results, whatever are the rights for the user (admin or simple user, restricted collections...)"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to pass in the md5sum of the file, rather than the md5sum of the entire file.",
+      "reference": "Shouldn't the docstring mention that the MD5 checksum of `p` is computed each time and verified? Also, could there be use cases where a caller might not want such a check to happen because, eg, a blob storage solution doesn't provide MD5 verification or uses another hash algorithm such as SHA256?"
+    },
+    {
+      "prediction": "I think this should be `if obj.focusRedirect and (eventName==\"gainFocus\" or eventName==\"releaseFocus\")`",
+      "reference": "focusRedirect is used in the powerpnt appModule. We might have to make sure that this does not break. Having said that, I really like this being handled on the events level!"
+    },
+    {
+      "prediction": "I think you can remove this line.",
+      "reference": "You imports are out of order"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for this.",
+      "reference": "This should not be called Flow Trigger. FlowTrigger should has its own flowTrigger job name, for instance, \"flowtrigger\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `sql/sqlite3.go`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `Program.cs`, not in `Stress.cs`.",
+      "reference": "Why do we need `\\n` if we already use `WriteLine`? (and `\\n` is not cross platform)"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for this.",
+      "reference": "What is instrumentation name used for? I'm not seeing other classes use it, only a getter, constructor, and builder."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't see any other way around it.",
+      "reference": "@kristofgazso could you review these warnings?"
+    },
+    {
+      "prediction": "This file should not be included in this PR.",
+      "reference": "will delete this."
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "~~Indent 2 lines above~~ Disregard. My fault"
+    },
+    {
+      "prediction": "I don't think this is the right place for this check. I think we should move this check to `org.mule.runtime.core.api.context.impl.DefaultMuleContext` instead of `org.mule.runtime.core.api.context.impl.DefaultMuleContextConfiguration`.",
+      "reference": "What was the bug that this is fixing?"
+    },
+    {
+      "prediction": "I think it would be better to use `urllib.quote_plus` instead of `base64.b64encode`",
+      "reference": "This will always add the authorization header to the request object. Is this the right scope for these two lines? If username/password are not defined, it will encode 'Basic :'"
+    },
+    {
+      "prediction": "I think this should be `\"/com_github_google_flatbuffers/tests/\"` instead of ` \"../com_github_google_flatbuffers/tests/\"`",
+      "reference": "Is this always how Bazel does paths? Would it be nicer to do `-DBAZEL_TEST_DATA_PATH=../com_github_google_flatbuffers/tests/` so this string is not hardcoded in the the source?"
+    },
+    {
+      "prediction": "I think this should be `date_utc` instead of `date`",
+      "reference": "Not an expert of ES, so not sure of the impact here."
+    },
+    {
+      "prediction": "I'm not sure why this change is needed.",
+      "reference": "Who deletes the memory for the granteeList? (or the roleList for that matter)"
+    },
+    {
+      "prediction": "I think this should be `ws_listener.decrease_subscription_count (subscriptions.begin (), subscriptions.end ())`",
+      "reference": "This loop is missing a lock on subscriptions_mutex."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "shouldn't clearance be setting these up for us?"
+    },
+    {
+      "prediction": "I think this should be renamed to `noop` instead of `defaultServiceNameDetector`.",
+      "reference": "Need to update the doc for `TelemetrySDK` and `Host` structs deleting references from removed functions."
+    },
+    {
+      "prediction": "I think this should be removed as well.",
+      "reference": "I think we should keep that here and add a new method `isSyntacticallyFinal` that returns `super.isFinal()` (and can be used in UnnecessaryModifier). Otherwise the contract of `isFinal` is not respected"
+    },
+    {
+      "prediction": "I think this should be `if (gridCol.colDef.enableCellEditOnFocus === true) {` instead of `if (gridCol.colDef.enableCellEditOnFocus === true) {`",
+      "reference": "Is there a reason why we wouldn't want to check this every time? Why are we only checking for the footer and scroll bar when enableCellEditOnFocus is true?"
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "You need to create an API for this class. This class should follow roughly the same dependency structure than RemoteAvailableInstallersPlayOnLinuxImplementation / RemoteAvailableInstallers. Maybe we could use a inner class here?"
+    },
+    {
+      "prediction": "I think this should be `workflow_id = VALUES(workflow_id)`.",
+      "reference": "This should not be a case, right? If `run_id` is the same `workflow_id` can't be changed. Actually surprised that it is not part of a key."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem here is that the lock conflict error 8558 is not specific to AQR, it's a general error that occurs during the update process. So disabling AQR here is not going to solve the problem. I think we need to find a way to prevent the error from occurring in the first place. One way to do this is to make sure that the lock is acquired before entering the update loop. Another way is to make sure that the lock is released before exiting the update loop.",
+      "reference": "How would AQR work for an INSERT/SELECT of one table into another where a LOB column is being copied?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be under `github.com/envoyproxy/go-control-plane/envoy/service/discovery/v2`",
+      "reference": "Since these are ultimately different endpoints, it would be nice if we could move them up one level and nuke the common `sds` directory in order to reduce path stutter"
+    },
+    {
+      "prediction": "`beforeChangeResult === false || beforeChangeResult[0] === null` -> `beforeChangeResult.length === 0 || beforeChangeResult[0] == null`",
+      "reference": "Please check also whether `null` occurs in the rest of `beforeChangeResult` array."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `transport/unary.go`",
+      "reference": "Outdated docs. There is no response, there's an ack."
+    },
+    {
+      "prediction": "I think this should be `return fmt.Errorf(\"cannot start a container that has run and is stopped\")`",
+      "reference": "I'd rather have the container process remove the FIFO after it unblocks. Then `start` can always `Exec()`, and you can catch the \"FIFO does not exist\" error and translate it to a prettier \"someone must have already started the container\"."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this check. I think it should be done in the `getTransactionFromActionInfo` function, but I'm not sure if that's the right place to do it.",
+      "reference": "can you check if there's other similar cases to add nil-check like this?"
+    },
+    {
+      "prediction": "Shouldn't this be `is_valid_commonname = commonname is not None and len(commonname) < 64`?",
+      "reference": "`<= 64`? I just picked up what you said in #3981 (\"the CN field is limited to 64 characters\") but maybe there's something I don't know where the 64th character is needed (trailing dot or whatever?) Also this sounds like something that could be beautifully unit tested. Sorry for bugging you :grin:"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "I understand that this grabs a Licenses objects but what is the autocomplete method used for? What does this do in context of the auto_completes controller?"
+    },
+    {
+      "prediction": "I think this method should be called `transactionLocation` instead of `transactionLocationByHash`",
+      "reference": "I don't really like that way of naming methods based on their arguments. But I can see that the other method names are the same ..."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. This is only used in one place, and that is in `org.apache.druid.java.util.common.Base64Encoder`. I think it would be better to move this method to `org.apache.druid.java.util.common.Base64Encoder`.",
+      "reference": "@sekmiller This method correctly strips the leading and trailing slashes (and also \".\" and \"-\"); But I thought the plan was also to replace any multiple slashes between nested folders with a single slash. For example, as implemented now, I can enter \"folder1///folder2\", and it gets saved and displayed like this, with the 3 slashes. I thought we wanted to sanitize it as \"folder1/folder2\"."
+    },
+    {
+      "prediction": "I think this should be `int64_t pid;`",
+      "reference": "Looks like the X86 inline assembly is not happy with this type. I will update that"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Tabs and a space, again."
+    },
+    {
+      "prediction": "This should be `return \"\";` instead of `std::string();`",
+      "reference": "Please just enclose if blocks with brackets; this has created issues in the past so I'd prefer we don't use them"
+    },
+    {
+      "prediction": "I'm not sure why this was here in the first place, but I'm not sure why it's gone now either.",
+      "reference": "`version` isn't need anymore because it's built into `approve_proposal_url`, yes?"
+    },
+    {
+      "prediction": "I don't think this is needed, it's already imported in `scapy.layers.dot11._dbmField`",
+      "reference": "Moved to the bottom of the file. `CtrlPDU` sits on `BTLE_DATA` so let's put it after"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "This ends up requiring the `json` gem, right? Is there any way we can do this to be implementation agnostic (e.g. to let me use `yajl-ruby` for performance reasons (or because my app already brings in a JSON library)"
+    },
+    {
+      "prediction": "I don't think this is correct. We should be returning an error if we hit an error.",
+      "reference": "This seems dangerous. What circumstances lead to the need for a retry? Can that happen more than once? I'm concerned that if we hit the pathological state, this could turn into a CPU spinner. I wonder if we could get the same benefit from just \"trying again\" and failing if it fails the second time."
+    },
+    {
+      "prediction": "I think this should be called `startingSnapShotId` instead of `startingSnapshotId`",
+      "reference": "If there is a reference to other fields or methods, please do the initialization in the constructor."
+    },
+    {
+      "prediction": "I think it would be better to have this in a separate file, so that it can be easily moved to another PR if needed.",
+      "reference": "These should be importing from `pill-container`"
+    },
+    {
+      "prediction": "This should be `player->sendCancelMessage(RETURNVALUE_NOTPOSSIBLE);`",
+      "reference": "Why this change? It will now only show on the bottom of the screen as white text, is it correct behaviour?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in `view_config.rb` since it's only used there.",
+      "reference": "Does this need to allocate a string or would a nil value (indicating no header) be a better?"
+    },
+    {
+      "prediction": "I think this should be `component` instead of `constructor`",
+      "reference": "Nit: Past tense of `reset` is also `reset`."
+    },
+    {
+      "prediction": "I think it would be better to have a `get_fee` method in the config module, and have it return the value of `config.fee_per_kb()`. This way, you don't have to pass the config value to the command, and you don't have to pass the command to the config.",
+      "reference": "Shouldn't this use `'n'` instead?"
+    },
+    {
+      "prediction": "Shouldn't this be `enumsspb.TASK_TYPE_TIMER_TASK_TIMEOUT`?",
+      "reference": "revert back to 'TASK_TYPE_ACTIVITY_TIMEOUT'"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `GetBlock<T>` overload.",
+      "reference": "Shouldn't we prefer passing a pointer by reference T*&, since these are C++ bindings?"
+    },
+    {
+      "prediction": "I think you can remove the `iFromPeers.IPBlocks = append(iFromPeers.IPBlocks, ipb...)` line here.",
+      "reference": "nit: Personally I feel that it's cleaner to let `toAntreaPeerForCRD` to take `ingressRule` as a param instead of `ingressRule.From`. It can call `processRefCGs` inside the function and do the ipBlock/AG union within the function itself. It might not be worth the refactor though."
+    },
+    {
+      "prediction": "I think this should be `api.ReconcilePending` instead of `api.ReconcileRequested`",
+      "reference": "Pending has been set only for SPC.. why? for other CRs as well, this need to be done.. right?"
+    },
+    {
+      "prediction": "I think this should be `||` instead of `===`",
+      "reference": "would use an empty string here as default and then check for emptiness in the twig template. I dont like mixing types @core23 what do you think?"
+    },
+    {
+      "prediction": "I think this should be `PAGE_TOKEN_NAME` instead of `PAGE_TOKEN_PAGE_TOKEN_NAME`",
+      "reference": "How come we need case manipulation for the nextPageToken and not for pageToken?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "why use these rather than regular named paths?"
+    },
+    {
+      "prediction": "I think this should be `format` instead of `guide`",
+      "reference": "Hm, I don't think the extra flags are doing any harm here, but it does seem like `--help` is the only flag you could *actually* use with this command. We might eventually want to separate those out."
+    },
+    {
+      "prediction": "I think this should be `similar_artists_candidate_set_df` instead of `similar_artist_candidate_set_df`.",
+      "reference": "We should make this a docstring, so that editors are able to pick it up."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `feature/requests/show.html.erb` since that is where the request is being made.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I think this should be 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024 * 1024",
+      "reference": "App cache size is now managed dynamically by the `WebView`. This statement has no effect in the new framework."
+    },
+    {
+      "prediction": "I think this should be `return typeof virtualNode.attr('alt') === 'string' && (nn === 'img' || nn === 'input' || nn === 'area')`",
+      "reference": "VirtualNode has a `hasAttr` function, any reason why you're not using it?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the `HttpAccess` class itself.",
+      "reference": "This now throws when running tests (maybe the move to java 11??). Do we want to keep it?"
+    },
+    {
+      "prediction": "I think we can drop the `[:10]` now, since it's the same as `len(sha1) > 7`",
+      "reference": "We can allow 7-char SHA, does not mean we want to show with 7-char in default, please change to `if len() > 10`, then cut."
+    },
+    {
+      "prediction": "I think this should be `noDuplicates: 'Cannot add the duplicate file %{fileName}, it already exists'`, not `noDuplicates: 'Cannot add the duplicate file: %{fileName}, it already exists'`",
+      "reference": "this one should also have quotes i guess :)"
+    },
+    {
+      "prediction": "I think this should be `t.Fatalf(\"Test %s failed, expected %v but got %v\", name, mock.expectedName, b.unstruct.Object.GetName())`",
+      "reference": "Using the variable on range scope `mock` in function literal (from `scopelint`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. If the workflow is completed, then the dataset has been created, so it should be ok. If the workflow is not completed, then the dataset has not been created, so it should be accepted.",
+      "reference": "does this result in a 200 when the dataset is still inprogress/publishing not yet finalized? Seems like 202 is the right code for that (as it was) and the test should be watching for a 202?"
+    },
+    {
+      "prediction": "I think this should be called `smartSqlOrderPath` to be consistent with `smartSql` and `smartSqlQueryType`.",
+      "reference": "Previously you could only order by the field you were searching by. But for full-text search, you can search across all indexed fields, and it didn't make sense not to have a sorting. For backward compatibility, the javascript code uses indexPath as the orderPath when no orderPath is provided."
+    },
+    {
+      "prediction": "I think this should be `|` instead of `?`",
+      "reference": "VIM didn't like your whitespace."
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "Are you sure the namespace should change?"
+    },
+    {
+      "prediction": "Why is this change necessary? I don't see any callers to `Startup`.",
+      "reference": "@martijn00 I'm not sure why we're keeping the object parameter and return type since this can be done by using MvxApplication<THint> with THint set to object"
+    },
+    {
+      "prediction": "I think it is better to add `TIME` in `TimeType` class itself, so that we don't need to create a new class `TimeType` when we want to add a new time type.",
+      "reference": "why do we create a time type here?"
+    },
+    {
+      "prediction": "This file should be renamed to `components/listview/listview.js`",
+      "reference": "Does this style actually need to get loaded in all of these components?"
+    },
+    {
+      "prediction": "I don't think we need to import `DefaultAccessLogPatternParser` and `AccessLogHandlerImpl` here.",
+      "reference": "import but not used so you did not resolve compile warnings?"
+    },
+    {
+      "prediction": "This should be `static_cast<int32_t>(std::ceil(armorValue * 0.475));`",
+      "reference": "This is so much more concise and beautiful than the previous formula"
+    },
+    {
+      "prediction": "I think this should be `final`",
+      "reference": "Is there any reason not to have the `static` modifier for the logger? _(that would be in keeping with the reference being uppercase)_"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should be in the `provider` package, since that's the package that contains the `provider.go` file.",
+      "reference": "It's not about session creation config. It's about passing consumer config parameters to underlying transport - nothing to do with session itself"
+    },
+    {
+      "prediction": "I think this should be `loggerFn`, not `loggerFnOverride`. The reason is that `loggerFn` is a function that takes no arguments and returns a logger, whereas `loggerFnOverride` is a function that takes two arguments, the prefix and an enable/disable flag.",
+      "reference": "Why is there an \"enable\" in the param name? Couldn't it just be `overrideDebug`? Here are everywhere else."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this action. I think it should be moved to `Video` controller, but I'm not sure if there is a better place to move it.",
+      "reference": "`published.recently_published_first` reads a little oddly to me. Is this the same thing as `Video.published.ordered`?"
+    },
+    {
+      "prediction": "I think this should be `throw new Error('Unable to parse content of ' + file.originalPath);` instead of `throw new Error('Unable to parse content of ' + file.originalPath.replace(extRegex, '.html'));`",
+      "reference": "Encountered this because I had a stray comma. Figured I'd tweak it a bit."
+    },
+    {
+      "prediction": "I think this should be `0.5s` instead of `8s`.",
+      "reference": "qq: Should this include the iptables6-save sim. to the inverse checks above?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to just use `reflect.DeepEqual` here.",
+      "reference": "im open to explicitly checking NotSet value here but json logic already handles it so kept it this way."
+    },
+    {
+      "prediction": "This should be moved to the top of the file.",
+      "reference": "The naming of the include file appears to be unlucky."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I don't have a better suggestion. The problem is that `legend_opts` is a `dict` and `axis` is a `matplotlib.axes.Axes` object, so you can't just pass `axis` to `legend_opts` and expect it to be a `dict`. You need to convert `axis` to a `matplotlib.axes.Axes` object before you pass it to `legend_opts`, and then you can use `axis.get_legend()` to get access to the `Legend` object.",
+      "reference": "Bit worried about this. At minimum you should make a copy of the dict here to avoid modifying a user supplied variable."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `CreateLaunchTemplate` function, since that is the place where `CreateLaunchTemplateVersion` is called.",
+      "reference": "Is it better to create one before pruning? In case creation fails we don't want to delete the previous one. We create a new one, it is tagged as latest, so the previous can be deleted. `CreateLaunchTemplateVersion` returns the version created, how about directly trying to delete the previous version? Assuming the numbering is strictly increasing."
+    },
+    {
+      "prediction": "This should be `ContainerDTO` instead of `WinePrefixContainerDTO`",
+      "reference": "Do we require the specific implementation information here?"
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this method. It's a bit confusing because we're checking if the email address contains the entire domain, not just the last part of it. Maybe something like `email_banned?(email_address.split(\"@\")[0])` would be better?",
+      "reference": "How about `exists?(['domain ~* ?', domain.downcase])` ?"
+    },
+    {
+      "prediction": "I think we can remove the `newContextWithTimeout` function now, since we don't need it anymore.",
+      "reference": "Is this getting used in other places? Can we switch all the places to use the new API you added?"
+    },
+    {
+      "prediction": "I'm not sure why this is being removed. It seems to be the only place where it is used.",
+      "reference": "This need to be checked with linux. AFAIK, we have this file in our bin directory that is a link to the system file."
+    },
+    {
+      "prediction": "I think we can remove the `/* eslint-disable indent */` from this file.",
+      "reference": "Shouldn't we import `default`? I can't get here to test."
+    },
+    {
+      "prediction": "I think this should be `unix_socket` and not `self.unix_socket`.",
+      "reference": "!!!!!!!!!!!!! @graingert, does this mean that tests haven't been running???"
+    },
+    {
+      "prediction": "Should this be `type == null` instead of `type != null`?",
+      "reference": "I've already tried the null-coalescing operator here and got a strange error, so did this the old-fashioned way."
+    },
+    {
+      "prediction": "I think it would be better to use `!parent.hasAttribute('role') || parent.hasAttribute('role') === 'list'`",
+      "reference": "This doesn't follow our spacing convention. It's also a little hard to read. Can you reformat?"
+    },
+    {
+      "prediction": "I think we can remove the `internal` prefix from the package name.",
+      "reference": "The deploy should come before template? EDIT: I see that in other files, we put a separate line and put deploy at the end. What is the reason for this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we need to add a new configuration type for aggregations that can be dropped.",
+      "reference": "Consider making a constant (e.g. `MetricStreamConfiguration.Drop`)."
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "same question: why not just switch this to range too?"
+    },
+    {
+      "prediction": "This should be `const Type type(m_IO->InquireVariableType(name));`",
+      "reference": "@chuckatkins most places used an empty string for \"no type\", but a few used `\"unknown\"`. I've converted both cases to `Type::None`. Do you know why there was a distinction before?"
+    },
+    {
+      "prediction": "I think this should be `static_cast<size_t>(std::min(train_data_->FeatureGroupSizesInByte(i), num_data_))`",
+      "reference": "should we assert for the type for `FeatureGroupData` ? I think it should be 1-Byte type."
+    },
+    {
+      "prediction": "I think this should be `np.array([[[ 0,  4,  8], [ 1,  5,  9], [ 2,  6, 10], [ 3,  7, 11]], dtype=np.int32])`",
+      "reference": "As long as you are sure this is definitely correct now... :-)"
+    },
+    {
+      "prediction": "I think it would be better to keep the text as \"Cancel\" and just add a space between \"Cancel\" and the number.",
+      "reference": "... so that all action buttons have the same height :-)"
+    },
+    {
+      "prediction": "I think this should be `DomainParticipantFactory.hpp` and `DomainParticipant.hpp` and `DataReaderQos.hpp` and `DataReader.hpp` and `DataWriter.hpp`",
+      "reference": "Use correct, alpha sorted, include order. Correct order means: 1. Header corresponding to this source (i.e. `\"LatencyTestPublisher.hpp\"`) 2. C system headers 3. C++ system headers 4. Alpha-sorted external libraries headers 5. Alpha-sorted public headers from this project 6. Alpha-sorted private headers"
+    },
+    {
+      "prediction": "I think this should be `RequestFactory` and not `OutboundOption`",
+      "reference": "I hate this name, open to suggestions."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `StoreTargetMetadata` method.",
+      "reference": "Does this log line work? We need Warningf or just warning without the format string."
+    },
+    {
+      "prediction": "Do we need to pass in the processors here? Seems like it should be possible to do this in the `Accumulator` constructor.",
+      "reference": "The pull controller would need similar updates."
+    },
+    {
+      "prediction": "I'm not sure if we want to keep the copyright year 2015-2017 or 2015-2018",
+      "reference": "ignore this.. just didn't want to litter with a commit update to satisfy license plugin"
+    },
+    {
+      "prediction": "Line is too long. [813/80]",
+      "reference": "Is this definition needed since it doesn't look like it's being used anywhere?"
+    },
+    {
+      "prediction": "I think this should be `unless fields.has_key?(self.discriminator_key)` instead of `unless fields.has_key?(self.discriminator_key) || descendants.length == 0`",
+      "reference": "Can this condition be reworded using `if` please?"
+    },
+    {
+      "prediction": "I think it would be better to check `inputs != nullptr && !inputs->hasData()` instead of `(inputs != nullptr && !inputs->hasData())`",
+      "reference": "after ||, no need inputs != nullptr"
+    },
+    {
+      "prediction": "I think this should be `IntegTestBuildspecPath string` and not `IntegTestBuildspecPath string`",
+      "reference": "Is it just preference or on purpose that using slice of structs instead of slice of pointers?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the `newVNode._children` section.",
+      "reference": "This line always confused me   anyway. Good catch!"
+    },
+    {
+      "prediction": "I think this should be done in `Shutdown` and not in `BlockServer`.",
+      "reference": "Should we add `ctx` to `BlockServer.Shutdown()` for this purpose? Would be nice, but I don't care too much."
+    },
+    {
+      "prediction": "Shouldn't this be `period.nanosec * 1000000`?",
+      "reference": "Duration_t at RTPS level must be serialized using fractions."
+    },
+    {
+      "prediction": "I think this should be called `NotRunningOutboundError` instead of `NotRunningInboundError`.",
+      "reference": "\"peer\" has its own meaning within YARPC with its own class of objects. This should probably be \"dispatcher\" too?"
+    },
+    {
+      "prediction": "I think this should be `experimental-enable-naive-ctu-analysis`",
+      "reference": "I do not like this line break before `=true`. Maybe starting the list in the next line would help to reduce the indent? This way we could keep this string together."
+    },
+    {
+      "prediction": "I think it's better to use `flag.StringSliceVarP` instead of `flag.StringVar` and `flag.StringVarP`.",
+      "reference": "_plugins should be removed too"
+    },
+    {
+      "prediction": "I think this should be `node.hasAttribute('alt') && (nn === 'img' || nn === 'input' || nn === 'area' || nn === 'amp-img')`",
+      "reference": "The rule selector will not include `<input>` and `<area>` elements. Why not make this a new, separate rule altogether?"
+    },
+    {
+      "prediction": "I think this should be `userscripts.NotFoundError` instead of `userscripts.UnsupportedError`",
+      "reference": "@The-Compiler Is this style okay or would you prefer a common userscript exception base?"
+    },
+    {
+      "prediction": "I think this should be `value.sub_type || 0x00` instead of `sub_type || 0x00`",
+      "reference": "we'd need to return a MongooseBuffer here instead of the Binary. lets add the subtype option to the buffer schema type as referenced in #1000 instead."
+    },
+    {
+      "prediction": "I don't think this change is needed.",
+      "reference": "Let's remove this change."
+    },
+    {
+      "prediction": "I think this should be `ROLE_TABLECELL, ROLE_TABLECOLUMNHEADER, ROLE_TABLEROWHEADER` instead of `ROLE_TABLECELL, ROLE_TABLECOLUMNHEADER, ROLE_TABLEROWHEADER,ROLE_TABLEROW`",
+      "reference": "Could you split this into multiple lines?"
+    },
+    {
+      "prediction": "I think this should be `__( 'How quickly your page loads, how quickly people can interact with your content, and how stable your content is.', 'google-site-kit' ) `",
+      "reference": "This technically works, however it may be safer to use both `getCurrentReferenceURL` and `getCurrentEntityURL` and then make this condition `currentReferenceURL === currentEntityURL`. Currently, this code only works based on the internals of `getCurrentReferenceURL`, so by checking both we would decouple that."
+    },
+    {
+      "prediction": "I don't think this is the best way to format the response. I think it should be something like: `respMsg = fmt.Sprintf(\"Response code: %d, response body: %s\", resp.StatusCode, resp.Body.String())`",
+      "reference": "It does not recommend capitalization"
+    },
+    {
+      "prediction": "I think we should use `expect(page).to_not have_content(\"Your credit card was declined\")` instead of `expect(page).to have_content(\"Your credit card was accepted\")`",
+      "reference": "Minor thing, but can you use 1.9 hash syntax for this?"
+    },
+    {
+      "prediction": "I think this should be `Util.format_windows_path(os.path.join(lambda_cwd, lambda_cwd_on_host))` instead of `Util.format_windows_path(lambda_cwd)`",
+      "reference": "I tried to follow the code structure already present."
+    },
+    {
+      "prediction": "I think this should be a `FileAlreadyExistsException` instead of an `IOException`.",
+      "reference": "seems reasonable to also want to include the 'to' location?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. I think you should be able to use `os.path.join` to construct the path to the nupic install directory.",
+      "reference": "Please leave 1 space char between commas here and other places in this PR."
+    },
+    {
+      "prediction": "I think this should be `0 == pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED)`",
+      "reference": "hm is this really &&? Not ||? Was it initially correct if any of these things are not set that it needs to recreated?"
+    },
+    {
+      "prediction": "I think this should be `Collections.EMPTY_LIST` instead of `new String[] {}`.",
+      "reference": "Please don't use C-style array declarations. IMO our pre-commit ought to be enhanced to not allow this"
+    },
+    {
+      "prediction": "I think this should be `return (p->data.bo.size == p1->data.bo.size);`",
+      "reference": "@rhc54 This doesn't look like a comprehensive comparison. For `PMIX_STRING` we seem to actually compare the content of the data while here we only compare meta-information which doesn't ensure that values are the same."
+    },
+    {
+      "prediction": "I think this should be `boost::system::error_code` instead of `bool`.",
+      "reference": "I think `bypass_frontier_confirmation` conveys the intention better (default false as well)"
+    },
+    {
+      "prediction": "I think this file should be renamed to `testutil.go` and moved to `pkg/test`",
+      "reference": "This looks like it goes beyond the scope of removing `BlockHeight()`, in the future please do this in a separate commit at a minimum -- separate PR is fine too."
+    },
+    {
+      "prediction": "I think this should be `self._tab.set_muted(muted)`.",
+      "reference": "You'll also need to adjust `FakeWebTabAudio` in `tests/helpers/stubs.py`."
+    },
+    {
+      "prediction": "This file should be removed.",
+      "reference": "@csrwng what should we do here? This one is from origin."
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "Follow alphabetical order."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should be in `Diagnostics` or `Tracing` folders.",
+      "reference": "As far as I can tell, these always get and set the context as an `IReadOnlyDictionary<string, string>`. Can we use that here instead of `object` and get rid of the `as IReadOnlyDictionary<string, string>`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `else` branch of the `if self._opt.text is not None` condition above.",
+      "reference": "As you only have `{}` as the format string (without anything else in it), this is the same as doing `self._doc.setHtml(html.escape(self._opt.text))`"
+    },
+    {
+      "prediction": "I think this should be `azkaban.security.credential.name`",
+      "reference": "Where would this be configured? In azkaban.properties?"
+    },
+    {
+      "prediction": "I don't think this is the right place to check for atomic operators. I think it should be in the `update` object itself.",
+      "reference": "1. Use `const` instead of `var` here. 2. Move the check to after the `if (update == null || typeof update !== 'object')`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to have a function like `struct st_h2o_multithread_queue_t *st_h2o_multithread_queue_new(void)` that takes a `struct st_h2o_multithread_queue_params_t` and returns a pointer to a `struct st_h2o_multithread_queue_t`. Then you can just call `st_h2o_multithread_queue_new()` from the `st_h2o_multithread_queue_create()` calls below.",
+      "reference": "Let's assume that on Linux `eventfd` is always available as a wrapper of the eventfd2 system call. It is supported since linux 2.6.27 + glibc 2.9, as well as from the ancient versions of bionic. In other words, it would be fair to argue that the pair would be available on practically all platforms that provide `pipe2`. Therefore switching from using `pipe2` to `eventfd` on linux unconditionally would not cause issues on our users."
+    },
+    {
+      "prediction": "I think this should be moved to the `DBPReferentialIntegrityController` class.",
+      "reference": "This field could be `final`."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "would be good to have a test for the case when `shutdown!` raises"
+    },
+    {
+      "prediction": "I think it would be better to add a `if (t == ComTdbHbaseAccess::INSERT_ || t == ComTdbHbaseAccess::UPSERT_ || t == ComTdbHbaseAccess::REPLACE_ || t == ComTdbHbaseAccess::TRUNCATE_ || t == ComTdbHbaseAccess::RENAME_ || t == ComTdbHbaseAccess::CREATE_ || t == ComTdbHbaseAccess::DROP_ || t == ComTdbHbaseAccess::REINDEX_ || t == ComTdbHbaseAccess::CREATE_INDEX_ || t == ComTdbHbaseAccess::DROP_INDEX_ || t == ComTdbHbaseAccess::CREATE_ALIAS_ || t == ComTdbHbaseAccess::DROP_ALIAS_ || t == ComTdbHbaseAccess::TRUNCATE_ALIAS_ || t == ComTdbHbaseAccess::RENAME_ALIAS_ || t == ComTdbHbaseAccess::CREATE_INDEX_ALIAS_ || t == ComTdbHbaseAccess::DROP_INDEX_ALIAS_ || t == ComTdbHbaseAccess::TRUNCATE_ALIAS_ || t == ComTdbHbaseAccess::RENAME_ALIAS_ || t == ComT",
+      "reference": "Does upsert pass this check also?"
+    },
+    {
+      "prediction": "I think this should be `WithField(\"stdout\", \"stderr\", ioStreams.Out.(*bytes.Buffer).String()).Info(\"patch command successful\")`",
+      "reference": "Apologies reviewers, I did end up pushing another change. I was starting to request SRE-P help to get the apiserver configs from some clusters before and after when I realized I can just log the stdout from the patch command to see if anything was changed or not. Much simpler to verify if my work did or did not make changes to the config during the transition. Also got the spacing issue you mentioned @abhinavdahiya Thanks all."
+    },
+    {
+      "prediction": "I think this should be `kbfsConfigNameTempDir` or `kbfsConfigNameTemp`",
+      "reference": "I'm not sure this naming is a good idea; that's the format for macOS xattr metadata files on unsupported filesystems."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Metrics/LineLength: Line is too long. [85/80]"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Please remove these imports. The project's style is to use `Assert.assertEquals` and not import static methods in general. This also caused a lot of unnecessary changes."
+    },
+    {
+      "prediction": "(style) code indent should use tabs where possible (style) please, no space before tabs",
+      "reference": "thanks. Note that returning at this point might leak memory from the allocations above, so the PR will need to take care of that too."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with right now.",
+      "reference": "If you really want to rename this, you'll also need to adjust the name in `__init__` and in other places it's used (`browser/commands.py`)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this annotation. I think it should be in the `DataFile` class.",
+      "reference": "As discussed at standup, I gave up on this. Calling `DeleteMapLayerMetadataCommand` felt cleaner anyway because there might be other cleanup that needs to happen. @scolapasta and @matthew-a-dunlap plan to discuss this."
+    },
+    {
+      "prediction": "I think this should be `fmt.Printf(\"EXP:%+v\\nACT:%+v\\n\", objExp.GetObject(), objAct.GetObject())`",
+      "reference": "Do we need these changes? If so, can you tidy up the messages? Looks like it might have been your testing/debugging changes"
+    },
+    {
+      "prediction": "I think this should be `SecretTargetLogDriver` and not `TargetLogDriver`",
+      "reference": "where is the default being set?"
+    },
+    {
+      "prediction": "I think this should be `matches[3].start` and `matches[3].length` instead of `matches[2].start` and `matches[2].length`.",
+      "reference": "these changes should likely be in the next commit instead"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these tests. I think they should be in `pkg/addonformatter/test.go` and `pkg/addontransform/test.go`.",
+      "reference": "Since you have defined this struct, I would suggest to replace all exist literal structs by this."
+    },
+    {
+      "prediction": "This should be `private static boolean mConfirmMenuDiscard = false;` instead of `private static boolean mConfirmMenuDiscard = true;`",
+      "reference": "The field name doesn't really capture what this option does. I think `mConfirmDiscardMessage` would be a better choice."
+    },
+    {
+      "prediction": "Why not using `is_string($media->getBinaryContent())` ?",
+      "reference": "what if `$media->getBinaryContent() == Symfony\\Component\\HttpFoundation\\File\\File` does `is_string()` return `true`? ping @greg0ire"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `if (field.nested_flatbuffer)` block above.",
+      "reference": "maybe rather than `bool` make this field a `StructDef *` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we need to make sure that we don't enqueue blocks to the blocksToDeleteChan more than 10ms after the last time we enqueued blocks to the blocksToDeleteChan. Otherwise, we could end up with blocks in the blocksToDeleteChan that were never actually deleted.",
+      "reference": "Shall we maybe put this in a `const` like `backgroundTaskTimeout`?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that we're trying to close the response stream before it has been sent to the client. We should be closing the response stream after it has been sent to the client.",
+      "reference": "CI is complaining about this."
+    },
+    {
+      "prediction": "I think this method should be called `determineRemovalTime` instead of `determineRemovalTimeTime`",
+      "reference": "Let's only change a file if it is really necessary. Such a change does not bring much value and makes it harder to find the original commit in which the method was introduced."
+    },
+    {
+      "prediction": "I think it would be better to use `array_key_exists('class', $attributes)` instead of `array_key_exists('class', $attributes) && $attributes['class'] !== ''`",
+      "reference": "I would prefer to set up space between classes in format pattern, `%s %s` and then use `trim()` to remove unnecessary whitespaces. This will also solve stripping whitespaces from the beginning and end of a string `$attributes['class']`"
+    },
+    {
+      "prediction": "I think it would be better to log the error message here, instead of in the `if` statement below.",
+      "reference": "when Op is illegal should return here?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "don't we need to remove the project version from installedVersions?"
+    },
+    {
+      "prediction": "I think this should be `bridge.print_page(*options)`, since it's a method on the bridge object.",
+      "reference": "the bridge here isn't defined as an accessor / reader to try mask it better. So you need to directly call the iVar `@bridge` here."
+    },
+    {
+      "prediction": "I think this should be `err` and not `err != nil`.",
+      "reference": "This is intended. Pls revert this."
+    },
+    {
+      "prediction": "I think we can remove the `T::compareTo` from `ofAll` as well.",
+      "reference": "Something wrong with cast to `(Comparator<> & Serializable)`, need to be investigated"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure it's the right thing to do.",
+      "reference": "I think we need this. Not sure, I'll test."
+    },
+    {
+      "prediction": "I'm not sure why you removed this line. It seems to be the only place where `disable_*_tables` is used.",
+      "reference": "I'm going to need to check if google compute requires these steps in this order, or if you can disable iptables after the fact."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Did go change their tags with the new version?"
+    },
+    {
+      "prediction": "I think this should be `s.config.Federation.FederatesWith`, not `s.config.Federation.FederatesWith`.",
+      "reference": "I don't recall what we decided here in terms of which source would get priority. As written, the static configuration will overwrite datastore results."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to fix this. I think it should be fixed in `checkAndReplaceLabel`.",
+      "reference": "I think it should still panic for when there are no outputs; might be nice to special-case that though so the message is more explicit."
+    },
+    {
+      "prediction": "I think this should be `final`",
+      "reference": "*NULL_DEREFERENCE:* object `null` is dereferenced by call to `meterSupplier(...)` at line 122."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do it in the same way as we do for `--op-endpoint` and `--op-endpoint-json`.",
+      "reference": "switch over to .mode"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/unit/components/handsontable/handsontable-component.spec.js`",
+      "reference": "I guess your IDE did some auto-fixing here"
+    },
+    {
+      "prediction": "I don't think we need to import `PhraseQuery` here.",
+      "reference": "This is an unused import failing recommit still."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "I'm concerned that we're conflating multiple things here. There are 2 things that Beaker should really care about: - Network transport - i.e. `ssh` vs `winrm` - Interpreter - i.e. `bash`, `cmd`, `powershell`, etc The problem is that @cowofevil is running Bitvise SSH, and he assumed we should be setting `is_cygwin: false` in node definitions. But that doesn't really accurately convey what we care about, and since Beaker doesn't appear to track network transport apart from interpreter, we end up in a strange state. `is_cygwin: false` appears to end up setting `is_powershell?` to `true`, which IMHO is not accurate or appropriate. Bitvise templates use `ssh` (like Cygwin), but the interpreter used is `cmd`, not `powershell`."
+    },
+    {
+      "prediction": "`aria-label` should be `role=\"dialog\"` instead of `role=\"dialog\"`.",
+      "reference": "I think this is ok for now. This ties into the larger issue of the site not having a consistent method for relaying form input errors. Please make sure the focus gets set on the close button when the dialog opens."
+    },
+    {
+      "prediction": "I think it would be better to add a new option `samplingRatio` instead of `autoColumnSize`.",
+      "reference": "Should these link be removed?"
+    },
+    {
+      "prediction": "`add_undo` -> `undo_on_close`?",
+      "reference": "Please make this a keyword-only argument by adding a `*` argument before `add_undo`."
+    },
+    {
+      "prediction": "I think this should be `payloadMarshaler := protocol.HandlerPayloadMarshal{...}` instead of `_ = payloadMarshaler`",
+      "reference": "didn't quite follow what this code block is doing."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `layoutManager` since it's the one that is responsible for updating the layout.",
+      "reference": "this doesn't sound pretty... I would rather we have a proper navigation for all modes - AFAIK TV remote navigation is very similar to normal keyboard, but maybe it just uses another key codes - those should be extracted in a single file and defined there depending on context then"
+    },
+    {
+      "prediction": "I think this should be `checked` instead of `checked` and `value` instead of `value`.",
+      "reference": "After diffing the children we check whether the value got out of sync, if it did we update it. We also update the `_prevValue` to prepare for the next event hitting our controlled component"
+    },
+    {
+      "prediction": "I think this should be `// Package node provides the glue-code needed in order to start a Bee node.`",
+      "reference": "It provides a type called Node which is a fully functional bee client. This package is where the dependencies are injected. It is not just a glue-code, it is concept of node."
+    },
+    {
+      "prediction": "I think this should be `_get_value_transformer('storage', 'download-directory')`",
+      "reference": "I think this will change `''` to `'%'` unconditionally, i.e. there'll be no way to set `''` anymore. This should really only be used for values which make no sense anymore."
+    },
+    {
+      "prediction": "I think this should be `Table#getRowKey()`.",
+      "reference": "Nit: this table's row key map -> this table's row keys map"
+    },
+    {
+      "prediction": "I think this should be set to `PJ_LOG_WARNING`",
+      "reference": "this change should be reverted"
+    },
+    {
+      "prediction": "I think this should be `appdata_path = os.path.join(os.path.dirname(basedir), \"misc\", \"qutebrowser.appdata.xml\")`",
+      "reference": "I'm a bit confused about the `os.path.dirname(basedir)` here - is this just to get to the parent directory? Either way, I think I'd prefer just having this in `update_version.py` as it's not needed in qutebrowser itself."
+    },
+    {
+      "prediction": "I'm not sure why this test is here, but I think it should be removed.",
+      "reference": "Revert this change and see below why..."
+    },
+    {
+      "prediction": "I think you can remove `cascade_mask_rcnn_r50_fpn` here.",
+      "reference": "Should not switch to `lvis_v1_instance` here because that base config uses ClassBalancedDataset to oversample the data."
+    },
+    {
+      "prediction": "I think we should use `LOG.info(String.format(\"Adding %s %d times\", caps, info.getMaximumSimultaneousSessions()));` instead of `LOG.finest(String.format(\"Adding %s %d times\", caps, info.getMaximumSimultaneousSessions()));`",
+      "reference": "This is an informational message that allows someone to read the console output and understand how the grid node is configured. Please leave."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in `pkg/alert/alert.go`",
+      "reference": "This doesn't make sense to me. What does RaiseSingleton mean? To me it sounds like a single object is being.. raised? Not sure."
+    },
+    {
+      "prediction": "I think this should be `AllJobExecutorTests.java` instead of `AllJobExecutorTestsTests.java`",
+      "reference": "What's the benefit of having jobExecutorTests as a member variable? What do you think about making the method static? Afterall, it is a stateless method."
+    },
+    {
+      "prediction": "I think this should be `dump` and not `check`. I don't think we should have a separate command for `check`.",
+      "reference": "This looks like it's only downloading things, not really checking their true validity. Especially for the MD object. Should we call `BareRootMetadata.IsValidAndSigned()` in `mdGet`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `setup`.",
+      "reference": "Nikoli, I believe that this addresses your concern. One question for you or Jae-Seung is if any allocation from the clone is properly cleaned up when the vector is destroyed. I believe that it should."
+    },
+    {
+      "prediction": "I think this should be `KindECSApp`",
+      "reference": "should be `ECSApp`"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could think of.",
+      "reference": "flake8 is going to fail on this having 2 lines"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `podutil` since it's only used by `podman pod delete` and `podman pod rm`.",
+      "reference": "should we make a dir named controllers/scheduler/podchaos"
+    },
+    {
+      "prediction": "nit: I think we should rename this to `KeyNodeHostname` to be consistent with the `KeyNode` variable.",
+      "reference": "there seems to be one KeyNode in kubernetes.go of PV.. would it make sense to use it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `cmd/` directory.",
+      "reference": "Im gonna be a little annoying here but did you configure your linter correctly? This empty line should not be here. Maybe go to settings and check if `goimports` is enabled?"
+    },
+    {
+      "prediction": "I think this should be `MODPRIVATECHATOUT` instead of `MODPRIVATECHAT`",
+      "reference": "Don't think this belongs in this pr"
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. It's a copy of the map of outbounds for this RPC object. The outbounds are already wrapped with middleware.",
+      "reference": "We can assert on the existence of Outbounds through ClientConfig calls right? Since this is only for tests, do we need this function?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for declaring the result. I think it should be in the `IMvxViewModel<TParameter>` interface since that's where the `TParameter` type is defined.",
+      "reference": "@martijn00 so this PR introduces a new ViewModel lifecyle method? It isn't in the PR description/any new docs"
+    },
+    {
+      "prediction": "nit: `NOTE the image size is [B, C, H, W]`",
+      "reference": "Are these modification duplicate? Or should we move it into base detector."
+    },
+    {
+      "prediction": "I think this should be `private static byte[] dagSeed(long block)`",
+      "reference": "Since this is private and single use it should be un-wrapped inside of the two-arg dagSeed method."
+    },
+    {
+      "prediction": "Shouldn't this be `array_merge_recursive_overrule`?",
+      "reference": "why the order change? Also for the next chunk. (I understand the result will be the same though)"
+    },
+    {
+      "prediction": "I think this should be `Returns` instead of `Returned`",
+      "reference": "I wonder if this should just be a single `string`? We don't have any cases where we return different types and it would make the parser simpler to just read a single identifier."
+    },
+    {
+      "prediction": "I think this should be `getInputFile(task.file().path())`",
+      "reference": "I don't think this needs to change. Can you go back to the previous implementation?"
+    },
+    {
+      "prediction": "I think this should be renamed to `Provisioner` instead of `Molecule`",
+      "reference": "Bump it 2 lines higher, 'c' > 'u'."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Please do not add author here"
+    },
+    {
+      "prediction": "I think this file should be renamed to `models.go`",
+      "reference": "Similarly, we should be able to persist action uniformly"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I think i'd prefer this override to be in the test helper instead, if that's possible?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to ensure the directory is created before we try to move the file.",
+      "reference": "This would be better with context added via wrapping or fmt.errorf()"
+    },
+    {
+      "prediction": "nit: I think this should be `// Package assert provides convenience functions that help a method or constructor check whether it was invoked correctly (whether its preconditions have been met).`",
+      "reference": "What's the purpose of this file?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the passcode length. I think it should be done in the `setPasscode` method, since that's where the user can change the passcode.",
+      "reference": "@bhariharan Why was passcode length requirement allowed to be lowered?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "This file needs a license header, too. Sorry I didn't catch that earlier."
+    },
+    {
+      "prediction": "Why do we need to pass a string for the genesis text here?",
+      "reference": "probably don't need `[:]` why not reference into `github.com/algorand/go-algorand/daemon/algod/api/server/lib` here and skip bouncing off daemon/algod/server.go ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `compute_reference` function.",
+      "reference": "Why reference is converted to table?"
+    },
+    {
+      "prediction": "I think we can remove the `else` case here and just return the `&datastore.ListRegistrationEntriesResponse{Entries: []*common.RegistrationEntry{}, Pagination: req.Pagination}`",
+      "reference": "should we implement pagination in the fake datastore so the server startup code that paginates entries for trust domain validation can be tested?"
+    },
+    {
+      "prediction": "I think this should be `if f.Vectorized != nil && f.Vectorized.Type() != nil {`",
+      "reference": "What mechanism will be exposed so the caller knows they're using the vectorized version?"
+    },
+    {
+      "prediction": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Can we remove respond_to block its not required here"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. It's only used in `graph_csv_data_source.cpp` and `load_graph.cpp`. I think it would be better to move it to `graph_csv_data_source.cpp` and `load_graph.cpp`.",
+      "reference": "Do not do that even in example, we demonstrate bad practice"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Hm, taking a second look over this, this is why we have the object inheritance structure that we do. This could be divided up by having a custom determine_if_x86_64 in the pswindows/exec hosts and then a default method in hosts.rb. That way all the custom ps windows work is in a single location."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move it to the `KestrelServerTests` assembly.",
+      "reference": "We have to `.Complete` now because of empty/null string test cases."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Is there still a way to specify no recursion?"
+    },
+    {
+      "prediction": "I think this should be a `string`.",
+      "reference": "How is this name being URI safe and being used programmatically related?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle the case where the `Dashboard` plugin is not loaded.",
+      "reference": "I think we have to keep this as a fallback for now, else it's a small breaking change :("
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it's fine, but just wanted to make sure we're on the same page.",
+      "reference": "nit: shouldn't be part of this PR."
+    },
+    {
+      "prediction": "Why did this need to be changed?",
+      "reference": "this change looks wrong to me. I think the test is trying to signal that something went wrong by throwing the `Error` here, otherwise the `catch` below will swallow it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to move this logic into the `LayoutBuilder` class itself.",
+      "reference": "I'm doubting if this should be `if (_layout.IsThreadAgnostic)`, as ThreadAgnostic stuff could be calculated on every thread. (and thus could be lazy). non-ThreadAgnostic should be calculated on the main thread."
+    },
+    {
+      "prediction": "This should be `org.slf4j.LoggerFactory.getLogger(LeftListWidgetChooser.class.getName() + \".LOGGER\")`",
+      "reference": "Please use only `Logger` instead of `org.slf4j.Logger`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `block_processor` class, since that's where the block processing logic is.",
+      "reference": "Should this be \"no limiter drop\", since this isn't an absolutely essential activity for the stability of the network?"
+    },
+    {
+      "prediction": "I think this should be `console.log('Account alias ' + process.argv[2] + ' deleted.')` instead of `console.log(\"Account alias \" + process.argv[2] + ' deleted.\");`",
+      "reference": "I updated many of the success messages, because most of these operations don't return data if they succeed. This meant many of the operations would print: `Success null` if the script ran successfully."
+    },
+    {
+      "prediction": "I think this should be `except FileNotFoundError as fnfe: return False`",
+      "reference": "Could you simply add it to the previous clause ?"
+    },
+    {
+      "prediction": "`localeCompare` is not supported in IE.",
+      "reference": "@joshyrobot, I think you can just use localeCompare with only the first argument and leave the other arguments out (here and all the other lines changed). It seems to do the right thing, and then we don't have to worry about these extra options."
+    },
+    {
+      "prediction": "I think it would be better to check if `slaOptions != null &&slaOptions.size() > 0` instead of `slaOptions.size() > 0`",
+      "reference": "@chengren311 : where does this logic move to?"
+    },
+    {
+      "prediction": "I think this should be `s.scope.AllowAdditionalRoles() || len(s.scope.ManagedMachinePool.Spec.RoleAdditionalPolicies) > 0`",
+      "reference": "Can we have a `nil` exception check at `s.scope.ManagedMachinePool`"
+    },
+    {
+      "prediction": "I think we can remove this now, since it's not used anymore.",
+      "reference": "Rather than deleting this, should you just reformat it so it's the `*` settings, so if nothing is configured in facets.ini, the existing default behavior continues to work?"
+    },
+    {
+      "prediction": "This should be `str_replace(\"_\", \" \", ucfirst($property));`",
+      "reference": "wrap, single line"
+    },
+    {
+      "prediction": "I think this should be `getGenericAwareReturnType()` instead of `getReturnType()`.",
+      "reference": "always use brackets for if statements"
+    },
+    {
+      "prediction": "Nit: I think it's better to have this as a static field rather than a class field.",
+      "reference": "`NUnitIntegration` doesn't need to cache this anymore."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we should be looking for `gz` in the `ext` field, not the `element_at` field.",
+      "reference": "Why did you change it to `>=` here?"
+    },
+    {
+      "prediction": "I think this should be `else if (tableName.startsWith(\"hive.\")) {`",
+      "reference": "Looks like this needs to be updated. There is no need to remove `hadoop.` if Hadoop tables don't use this code path."
+    },
+    {
+      "prediction": "I think this should be `digest/sha1` or `digest/sha256` instead of `sha1` or `sha256`",
+      "reference": "This require isn't need anymore, right?"
+    },
+    {
+      "prediction": "I think this is the only place where we're using the new crypto package.",
+      "reference": "a bit weird to see payments crypto package inside connection manager. There probably will be services not having any relation with payments that will use connection manager. Thats why probably ping-ping by itself can be useful without any relation to payments.."
+    },
+    {
+      "prediction": "I think this should be `BibTeX`, not `Bibliographic`.",
+      "reference": "I like the overall idea here, but I wonder if it would be more clear to define the property in the trait, and then set it in the constructors of the classes using the trait. Alternatively, at a bare minimum, it may be a good idea to add `?? 'Bibliographic'` here, to account for the possibility of the trait being used in a class that does not define the property at all."
+    },
+    {
+      "prediction": "I think this should be 10000, not 100",
+      "reference": "I would prefer to not have a magic number, but construct this value out of other constants we already have."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be under `pyramid_multiauth`, since that's where it's used.",
+      "reference": "The limitation is if somebody configure another policy with the same name, but it is an edge case we can ignore I guess."
+    },
+    {
+      "prediction": "nit: I think we can remove the `gocloud.dev/secrets/` prefix here as well.",
+      "reference": "? The package is named `hashivault`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "I don't think this should enable xwayland, because it's convenient to toggle xwayland just by setting `xwayland` to `false`"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Remove this line by accident?"
+    },
+    {
+      "prediction": "I think this should be renamed to `SourceSetContainer`",
+      "reference": "Calling `.stream()` on a `DomainObjectCollection` is pretty much always a bug, as it doesn't include objects added later (and encourages people to use afterEvaluate). I wonder if we should make this an error prone check?"
+    },
+    {
+      "prediction": "I think this should be `err != nil && !sigFound`",
+      "reference": "Thanks for paying attention to other places this might happen. This one is particularly important; I probably never should have gotten in the habit of CheckErr(), since it does a log.Panic() explicitly, which looks like something else until you look closely. It's supposed to be used places where \"can't happen\" but Things Can Happen."
+    },
+    {
+      "prediction": "I think this should be `dscp & 0x03` instead of `dscp<<2`",
+      "reference": "What is the different between nw_tos and ip_dscp? Only high 6 bits vs low 6 bits and supported version?"
+    },
+    {
+      "prediction": "I think this should be `if (cxSmiles)` instead of `if (!cxSmiles)`",
+      "reference": "I'm assuming that asSmarts & cxSmiles == asSmarts"
+    },
+    {
+      "prediction": "I don't understand why this is necessary. `super().__init__` is called by `__init__` in `__call__`, so you should be able to call `super().__init__` directly in `__init__`.",
+      "reference": "Seems like an oversight but I don't think the super would do anything else in this case unless I'm mistaken."
+    },
+    {
+      "prediction": "I think we can remove the `SampleServer` and `SampleClient` classes and just use `server` and `client` here.",
+      "reference": "Curious, do we consider this as pedantic for example code?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Should we convert the entity name to snake case using `Doctrine\\Common\\Inflector\\Inflector::tableize` or similar ? Should a `snaked_name` or whatever be part of the entity metadata in order to use it for form names, ids, & other html attributes ?"
+    },
+    {
+      "prediction": "I think this should be `GetPublicOutboundIP`",
+      "reference": "Should be `client.outboundIp`"
+    },
+    {
+      "prediction": "I think this should be `using System.IO; using System.Runtime.InteropServices;` instead of `using System.IO; using System.Runtime.InteropServices; using System.Threading;`",
+      "reference": "Must these `using` statements not go inside the `namespace` block?"
+    },
+    {
+      "prediction": "I'm not sure this is the right mutex level. I think it should be `fboHeadWriter` and `fboBlockWriter` and `fboHeadReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `fboBlockReader` and `f",
+      "reference": "Why? I thought the previous way was idiomatic. (Same with the `MetadataVer` stuff above.)"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. We should be using the `require` polyfill here.",
+      "reference": "shouldn't we `reject` here instead?"
+    },
+    {
+      "prediction": "I think it would be better to use `action.Core.GasPrice.String()` here instead of `util.StringToIOTX`",
+      "reference": "check err or use gasPriceUnitIOTX, _ := if we are sure action.Core.GasPrice is correct"
+    },
+    {
+      "prediction": "I think this should be `watcher` instead of `watcher`",
+      "reference": "Just curious, was wait not used at all before?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "nit: can we remove this?"
+    },
+    {
+      "prediction": "I think this should be `synchronized` as well.",
+      "reference": "is it not a problem when one synchronized function is called by other?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I can't think of anything better right now.",
+      "reference": "We (even as admins) shouldn't override the User Preference settings like `email_master`, `email_posts`, `email_kudos`, `email_new_followers`. These all would be set by the user of their choice. Please do remove these attributes from editing//updating. Thanks!"
+    },
+    {
+      "prediction": "This :attribute j\u00e1 est\u00e1 ligado. -> Este :attribute j\u00e1 estiver\u00e1 ligado.",
+      "reference": "\"Este :attribute j est anexado.\" sounds better."
+    },
+    {
+      "prediction": "I think this should be `if __name__ == '__main__':` instead of `if __name__ == \"__main__\":`",
+      "reference": "This line should be repeated for every module that needs testing. One cannot rely on nosetest to automatically discover all the modules. On the bright side, there is no need to change any of the other files."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "This error was unused"
+    },
+    {
+      "prediction": "I think we should rename this to `CloseShardRequestWithTasks` to make it clear that it's used to notify persistence that we're unloading a shard along with its tasks.",
+      "reference": "Personally I wish we didn't have a proto for every single little thing we do. RequestShardAction with an enum for the action type would be a lot cleaner imo, but I'm pretty sure that ship has sailed :) Just kvetching."
+    },
+    {
+      "prediction": "Do we need to do anything to the `incomplete_trail.html.erb` file to make this conditional? It looks like we don't need to do anything, but I wanted to double check.",
+      "reference": "Does this make more sense as `trails/_trail` now?"
+    },
+    {
+      "prediction": "I think this should be `OutpostARN` instead of `OutpostARNs`",
+      "reference": "synced offline. let's move this to `agent_unix.go` to make the IMDS call, since this is not supported in Windows."
+    },
+    {
+      "prediction": "I think this should be `Python::PythonSetup` instead of `Python::Install`.",
+      "reference": "Too much indentation here. Should match the lines above, which have four spaces."
+    },
+    {
+      "prediction": "I think this test should be removed as it is testing the same as the one in `test_osrm.cpp`",
+      "reference": "Same here we still need this test to make sure the fallback works."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `MvvmCross.tvOS.Views.Presenters.Attributes`",
+      "reference": "I'm wondering if we can actually base this on a `IMvxPresentationAttribute` instead of the base one."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `vnode._component` is not guaranteed to be the same as `vnode._vnode._component`. I think we need to do something like `vnode._component = vnode._vnode._component;` in `unmount`.",
+      "reference": "could we switch to `undefined` here?"
+    },
+    {
+      "prediction": "I don't think we need to do this cleanup here. The cleanup should be done in the `SmartStore`'s `dispose` method.",
+      "reference": "The cleanup call"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could think of at the time.",
+      "reference": "I think \"View artist\" is a bit more standard and expected. Or even \"Go to artist\" to take the Spotify terminology as-is."
+    },
+    {
+      "prediction": "`unless defined?(Puppet::Pops::Types::PSensitiveType::Sensitive)` -> `unless defined?(Puppet::Pops::Types::PSensitiveType) && defined?(Puppet::Pops::Types::PSensitiveType::Sensitive)`",
+      "reference": "It might make sense to refactor this later so that we use a Bolt-native type to hide sensitive values."
+    },
+    {
+      "prediction": "I think it would be better to move this to the top of the file, so that the rest of the tests don't have to jump through hoops to get to it.",
+      "reference": "We were trying to move away from this to force people to always run tests under py.test."
+    },
+    {
+      "prediction": "I think this should be done in a separate PR, as it's not related to the other changes.",
+      "reference": "1 - create browser specific options class: - chromeoptions - firefoxoptions - argsoptions"
+    },
+    {
+      "prediction": "I don't think this is the right fix. I think we need to fix the `Content-Type` header instead.",
+      "reference": "should be a case insensitive match though no? DOCTYPE and doctype are both widely used"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this action. I think it should go in `WidgetsController` since it's related to widget creation.",
+      "reference": "We must have a `fail ParamNotFound` here for cases where `@account.nil?`."
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in the PR.",
+      "reference": "How come this change was needed?"
+    },
+    {
+      "prediction": "I think these should be `ADC0 = 3`, `ADC1 = 4`, `ADC2 = 28`, `ADC3 = 29`, `ADC4 = 30`, and `ADC5 = 31`.",
+      "reference": "Why are there only 6 pins here, while below it appears to have 8 ADC inputs? Are pin 2 and 5 used for something else on this board?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to add the `secpRoot` field to `messageReceipts` as well.",
+      "reference": "I think we need an issue to track that chain sync omits BLS messages, and then to fix and test it!"
+    },
+    {
+      "prediction": "I think this should be `http://127.0.0.1:5000` and not `http://localhost:5000`",
+      "reference": "what is this default about? I am not using ngrok atm so would prefer an env var."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in `config/auth.php` or `config/database.php`",
+      "reference": "@Samuell1 Might be better to say \"Automatically check for plugin updates on login\"."
+    },
+    {
+      "prediction": "I think this should be `using System.Diagnostics.Tracing;`",
+      "reference": "Thanks for the additional Process helper! Can we also cache the first `Process.GetCurrentProcess()` result in a static field so we don't have to repeatedly call it? It means we would also need to dispose it when the static `_runtimeMetricsWriter` instance is disposed"
+    },
+    {
+      "prediction": "nit: I think this test can be moved to `TelemetrySessionTest.go` since it is testing the same functionality.",
+      "reference": "just wondering what is the result of breaking the logic we test here? it seems that in that case we will just not return immediately, but i'm not sure whether the test will fail?"
+    },
+    {
+      "prediction": "I think this should be `pctx := newTestProcessContext(filepath.Join(dir, \"src\"), nil)`",
+      "reference": "We should fail the test if this returns `err != nil`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `controllerutils.SetClusterDeploymentConditionWithChangeCheck`, but I'm not sure why it's called `provisionFailedTerminal` at all.",
+      "reference": "What is the drawback of not having this flag as a gating condition to report the metric?"
+    },
+    {
+      "prediction": "I think this should be `fmt.Sprintf(\"Graceful stop, pid=%d\", pid)`",
+      "reference": "`getProcessName()` will be less ambiguous, since Caddy has \"apps\" that it runs internally."
+    },
+    {
+      "prediction": "I think this should be `dms` instead of `dps`.",
+      "reference": "Is there any way to not hardcode this given we know the size in bytes? The public API has the flexibility of being a Span. Should we maybe not do anything printing DML? The runtime allocates the buffer, but it's a \"scratch memory\" area, The delegate gets it passed in and they decide how to use it. Also, how is DML used if the feature is for macOS support?"
+    },
+    {
+      "prediction": "I think this should be `return r.get_filter();` instead of `return r.filter_ref().get_value();`",
+      "reference": "How about just override the `buildFilter`? Passing another function here is a little weird."
+    },
+    {
+      "prediction": "I think this should be `Is this a TestRPM build with all RPMs hydrated?`.",
+      "reference": "This should just be a `.Bool()` flag I think, we don't encode anything beyond y/n here. See `$(RUN_CHECK)` and `$(STOP_ON_PKG_FAIL)` for examples of how to pass those in."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. It's not a type, it's a function.",
+      "reference": "Is this needed?"
+    },
+    {
+      "prediction": "This should be `admin.lookup.titles.deleteLookupValue`, not `admin.lookup.titles.deleteLookupValue`.",
+      "reference": "This code is passing a non localized string when it should be passing in a localized string or it should use the name of the item being deleted."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "how about moving it under L26"
+    },
+    {
+      "prediction": "I think this should be `axe._memoizedFns = []` instead of `axe._memoizedFns = [];`",
+      "reference": "I think this needs to be tested"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `Trail` model, but I don't know if it's the right place for it.",
+      "reference": "Style/MutableConstant: Freeze mutable objects assigned to constants."
+    },
+    {
+      "prediction": "I don't think we need to create a new `Deflater` instance for each compression level. We can just use the one we already have.",
+      "reference": "Just a thought, really. If it's a bug that can be probed for (and it can be - see Adrian's repro) then it could as well be a static initialization of a supplier of Deflater instances; if we probe for a buggy JVM, we return the wrapper. If we don't we return the Deflater. This way on non-affected JVMs nothing happens and if we do use the wrapper, we know the JVM is broken."
+    },
+    {
+      "prediction": "I think this should be private.",
+      "reference": "Looks like this should be publicly accessible for tests in any modules?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could get to load the `.env.local` file.",
+      "reference": "Allow putenv is necessary to be able to get environment variables with `getenv`?"
+    },
+    {
+      "prediction": "Indent the first statement one space more than the start of the previous line.",
+      "reference": "use function to check it. - don't do everything in main"
+    },
+    {
+      "prediction": "I think this should be `build_rule_book(self, global_configs)`.",
+      "reference": "Sorry I'm confused. Why not kwarg this like the others?"
+    },
+    {
+      "prediction": " mrm|vex|reqp,x,END_LIST},\n+    {OP_vgatherdpd,0x66389258,\"vgatherdpd\",Vvd,Hx,MVd,Hx,xx, mrm|vex|reqp,x,END_LIST},\n     {OP_vgatherdd,0x66389298,\"vgatherdd\",Vxd,Hx,MVd,Hx,xx, mrm|vex|reqp,x,END_LIST},\n     {OP_vgatherdq,0x66389318,\"vgatherdq\",Vxd,Hx,MVq,Hx,xx, mrm|vex|reqp,x,END_LIST},\n     {OP_vgatherqq,0x66389358,\"vgatherqq\",Vxd,Hx,MVq,Hx,xx, mrm|vex|reqp,x,END_LIST},\n```\nReview comment?",
+      "reference": "First, any changes here should be synchronized with instr_compute_VSIB_index(). Second, the original looks correct to me: the first letter of the opcode name suffix is the index size, while the second is the memory size. So \"vpgatherdq\" is a d-sized index and q-sized memory. The operand size we store for memory accesses is always the memory size, not the address size, and for VSIB we store a single memory element size. It looks like we have no inside-opnd_t storage of the index size: that's done by dispatch on opcode it seems. I have a bunch of notes of discussions on how to handle VSIB: we decided to bail on encoding too much inside the opnd_t I guess."
+    },
+    {
+      "prediction": "I think it would be better to pass the logs output as a parameter to the `run` method. This way, you don't need to pass the logs output to the `run` method every time you call the `ExternalProgramTask` and you also don't need to pass the logs output to the `build_tracking_url` method every time you call the `ExternalProgramTask`. Instead, you can just call the `build_tracking_url` method and pass the logs output as a parameter to the `build_tracking_url` method. This way, you don't need to pass the logs output to the `build_tracking_url` method every time you call the `ExternalProgramTask` and you also don't need to pass the logs output to the `build_tracking_url` method every time you call the `ExternalProgramTask`. Instead, you can just call the `build_tracking_url` method and pass the logs output as a parameter to the `build_tracking_url` method. This way, you don't need to pass the logs output to the `build_tracking_url` method every time you call the `ExternalProgramTask` and you also don't need to pass the logs output to the `build_tracking_url` method every time you call the `ExternalProgramTask`",
+      "reference": "Sorry for going back and forth. Adding docstring here would be very helpful for others to understand the need of this method."
+    },
+    {
+      "prediction": "I think this should be `return canSeeInvisibility() && creature->isInvisible();`",
+      "reference": "You introduced a bug here."
+    },
+    {
+      "prediction": "I think it would be good to have a unit test for this as well.",
+      "reference": "Should we log a warning/error if only one is specified and not the other?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in `piped_service_test.go` since this is testing the `piped_service` client.",
+      "reference": "`ctx` is unused in GetLatestEvent"
+    },
+    {
+      "prediction": "I think this should be `config.node.peering_port = network_params.network.default_peering_port;`",
+      "reference": "This looks good to me. Personally, I would have the following for maximum flexibility: * convert peering_port from uint64_t to int. * Set the default value to \"-1\" to specify the value is not set * if the value is -1 then do `config.node.peering_port = network_params.network.default_node_port` That would allow for 0 to mean, select the port number yourself and be more consistent. But I am happy with the change as is, as well."
+    },
+    {
+      "prediction": "please add license header",
+      "reference": "The blank line between the summary and detailed description is better to be kept."
+    },
+    {
+      "prediction": "I think this should be done after `a.setMetadataEndpointsKey()` and before `a.writeShapeNames()` and `a.resolveReferences()`.",
+      "reference": "Curiously, are these indents intended or should they be aligned?"
+    },
+    {
+      "prediction": "I think this should be `getSpeechTextForProperties(reason=reason,role=role,keyboardShortcut=keyboardShortcut) if config.conf[\"presentation\"][\"reportKeyboardShortcuts\"] else \"\"`",
+      "reference": "What if we changed this to this? roleText=attrs.get('roleText', lambda:getSpeechTextForProperties(reason=reason,role=role)) That will prevent the function from ever needing called in the roletext case, and removes that if."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to fix the issue. I think it's better to fix it in `libthread.c` instead of `libthread_tls.c`.",
+      "reference": "This will still result in a safe_read_tls_magic on AMD in tls_thread_preinit()."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Can you remove this change?"
+    },
+    {
+      "prediction": "I think this should be `field.Invalid(specPath.Child(\"isCA\"), isCA, \"ACME does not support CA certificates\")`",
+      "reference": "Oh, was that a bug in the validation?"
+    },
+    {
+      "prediction": "I think this should be `feedUrl = feedUrl.replaceFirst(\"((www.)?(subscribeonandroid\\\\.com)/)\", \"subscribeonandroid.com\")`",
+      "reference": "Please use the Google java code style. Basically, add more space characters like in the statements below. Next to method arguments and curly braces. This is currently not checked on CI because it is too inconsistent in the code base but I would prefer new code to be consistent."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think it would be better to have a single `SourcePropertyExpression` object that can be shared across all the tests.",
+      "reference": "Actually, I don't think we need to create object on heap."
+    },
+    {
+      "prediction": "Why did you change this? It looks like it should be the other way around.",
+      "reference": "this is a file we copy from gRPC, sp ideally it be fixed upstream as well.."
+    },
+    {
+      "prediction": "I think we can remove this import as it is no longer used.",
+      "reference": "changes like this, just strangle guava usages to be replaced by DelayLimiter into one place"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this component. I think it should live in `components` folder.",
+      "reference": "PropTypes need to be updated accordingly here. Note that the type also needs updating, not just the case. That is, these should both expect a `PropTypes.elementType` now instead of an `element`."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `ctest_resource_group_id_name` is not the same as `ctest_resource_id_name`. The former is the name of the group, while the latter is the name of the individual resource. I think the right fix is to change `ctest_resource_id_name` to be the same as `ctest_resource_group_id_name`, and then use `ctest_resource_group_id_name` in the call to `get_ctest_gpu()`.",
+      "reference": "maybe extra paren around `strncmp(...) != 0`"
+    },
+    {
+      "prediction": "I think this should be `resetOpenedWindowsCount` instead of `resetOpenedWindowsCountForPrivateWindows`",
+      "reference": "We should call `Windows.onResume `: - When the app is launched for the first time - When the app is resumed after being paused (home button and resume or device goes to sleep) - After a permission prompt is displayed"
+    },
+    {
+      "prediction": "I think it would be better to have a `BftForksSchedule<C extends BftConfigOptions>`.",
+      "reference": "Rename to ForksSchedule? Maybe in separate PR."
+    },
+    {
+      "prediction": "I think this should be `import kinto.core.storage`",
+      "reference": "Why a blank line here?"
+    },
+    {
+      "prediction": "I think this should be `active?` instead of ` deactivated_on.nil?`",
+      "reference": "I have a sense that this will actually eventually change to check the actual date so that subscriptions can be deactivated in the future. But I mention that here just to keep you in the loop. We should not make that change now, because it's not needed functionality."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `Environment` class.",
+      "reference": "What is the reasoning for using the negation of `isOption()` instead of simply using `isPositional()` to identify a positional param?"
+    },
+    {
+      "prediction": "This should be `0x01` or `0x00` depending on the vocation.",
+      "reference": "Why would you add those if client version min is set to 1100?"
+    },
+    {
+      "prediction": "I don't think we need to change this file.",
+      "reference": "You can delete this, it will now need to live in `gointerfaces` package"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Okay, I still don't understand why you have to add `html_safe` here, and it still seems like a very bad idea. It will allow html tags in the title, and keep Rails from escaping literal greater-than or less-than chars not intended as HTML tags. It ought to work to just let Rails do HTML-escaing as normal, without any manual `html_safe` or `strip_tags` or whatever. I am not clear on what problem you are trying to solve, why the straightfoward approach does not work."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `MessageAdapter` class, since that's where we set the display mode.",
+      "reference": "What is is subject when it's not the email subject. Why are we having to do this crap? What's calling this with an empty string?"
+    },
+    {
+      "prediction": "I'm not sure this is the right default. I think we want to default to AlwaysSample() if we don't specify a sampling probability.",
+      "reference": "Could you also remove the `defaultSamplingProbability` constant from `sampling.go`? It seems to became unused with this change."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in `apps/v1beta1/constants.go` or `apps/v1/constants.go`",
+      "reference": "can you delete the `generateDeploymentStrategicPatch` function as well?"
+    },
+    {
+      "prediction": "I think this should be `LOG.info(String.format(\"Connecting to %s and %s\", publishConnection, subscribeConnection));`",
+      "reference": "I'd keep this at `info` level..."
+    },
+    {
+      "prediction": "I think this should be `if (gn->limit() >= 0 && limitRows >= gn->limit()) {`",
+      "reference": "Don't use DCHECK to debug your code if it's the regular branch you need to handle."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to move this method. I think it would be better to move it to the `AntlrNode` interface itself.",
+      "reference": "You should return null here instead. Null is an acceptable default value for the image attribute."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `componentDidMount` or `componentDidUpdate` since those are the only two places where `component.__hooks` is initialized.",
+      "reference": "react bails as well?"
+    },
+    {
+      "prediction": "Why is this constant here? I don't see it used anywhere.",
+      "reference": "Can you please tell me why you did this?"
+    },
+    {
+      "prediction": "`this.opts.videoConstraints` should be `this.getOptions().videoConstraints`",
+      "reference": "Reading this again it should prob prefer the `videoConstraints.facingMode` value over `facingMode` if the former was already set"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "There's a good bit of class-level stuff that's repeated in our various product types. Think it's worth extracting a...dare I say it...module?"
+    },
+    {
+      "prediction": "I think this should be `if config.Options.WasmAbi != \"\" && config.Target.WasmAbi == \"js\"`",
+      "reference": "Please do not modify the `config.Target` field, it should contain exactly what is extracted from the JSON files. Instead, you can either: * Add a getter to the `config` variable (`*compileopts.Config`), so you can simply call `config.WasmAbi()` to get the value. * Use a local variable instead. The getter would be slightly cleaner, as it matches the pattern of other configurations (`CGO_ENABLED`, `GC`, `NeedsStackObjects`, etc)."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to start the test. I think we should start the test in the middle of the test router, so that we can make sure that the router is running before we try to start the test server.",
+      "reference": "The nil check is not necessary for these. Append checks both sides for nil."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we should be setting the attributes in the `trace` package. I'm not sure we",
+      "reference": "Shouldn't we instead have the SDK apply these, internally? I.e., I would expect to see the dctx entries included in the span as first-class distributed correlations, not as span attributes."
+    },
+    {
+      "prediction": "I think this should be `nil, errors.New(\"not implemented\")` instead of `nil, errors.New(\"not implemented yet\")`",
+      "reference": "nit: rename this in proto and update impl to `Search` to avoid stutter"
+    },
+    {
+      "prediction": "I think it would be better to use `orderBy('s.id', 'ASC')` instead of `orderBy('s.id');`",
+      "reference": "is it necessary to order by id as a second ordering? this will take place only when several new slider items are created without reordering (position is then null). Wouldn't be better to recalculate position after creating a new item? (right now items with null in position behave differently in administration and on frontend)"
+    },
+    {
+      "prediction": "I think it should be `uint64(100)` instead of `uint64(10000)`.",
+      "reference": "`ClaimFromRewardingFundBaseGas` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. It seems like it should be in the base class, but I'm not sure what the right place is.",
+      "reference": "How about modifying the NOGIL definition so that it only does anything if the thread safety flag (RDK_BUILD_THREADSAFE_SSS) is set?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should go in `pkg/p2p/p2p_interface.go` and `pkg/p2p/p2p_interface_test.go`.",
+      "reference": "blockheighter would be more clear"
+    },
+    {
+      "prediction": "I don't think we need this function at all.",
+      "reference": "This is so weird as a utility function."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that didn't make the tests flaky.",
+      "reference": "Why? What would happen if you did not?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `opt` method will return `null` if `o` is `null` or if `pathElement` is not present in `o`. You need to check if `o` is not `null` and `pathElement` is present in `o` before calling `opt` method.",
+      "reference": "Unrelated bugfix for NPE."
+    },
+    {
+      "prediction": "I think it would be better to check if the property is writable and only then check if it can be toggled.",
+      "reference": "What if some extended the `AdminController` and had a check for the old option name? Could that be a valid use case?"
+    },
+    {
+      "prediction": "`preventRequestAbort` is not used in this function. Can you remove it?",
+      "reference": "This will be only available when data-manager is enabled, is there a fallback in case data manager is disabled?"
+    },
+    {
+      "prediction": "nit: I think it would be nice to have a link to the COPILOT-1000 ticket explaining why this is the best way to do it.",
+      "reference": "Do you think we want to move the call to `generateWorkerServiceRecommendedActions` inside `RecommandedActions()`?"
+    },
+    {
+      "prediction": "This should be `VUID-vkGetQueryPoolResults-flags-02827` not `VUID-vkGetQueryPoolResults-flags-00815`",
+      "reference": "The VUIDs in this area are not great, but I think `ValidateQueryPoolStride` should probably be skipped if the query pool was created with type `VK_QUERY_TYPE_PERFORMANCE_QUERY`. VUID-02828 might be a better fit, but again, the existing VUIDs step on each other so it requires a bit of interpretation."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this method. It's not a key, it's a byte array, so maybe `ToDbKeyBytes` would be better?",
+      "reference": "Iguess you should use ToDbKey here"
+    },
+    {
+      "prediction": "I don't think this is correct. The counting index is used to count the number of blocks in the bucket, not the number of blocks in the bucket at the given height.",
+      "reference": "x.tbk is the \"index\" below, and is done in Start()"
+    },
+    {
+      "prediction": "I think this should be `private const string DD_IsAppInsightKey = \"DD_IsAppInsight\";`",
+      "reference": "Do we already have a convention for this? If not, would we consider \"DataDog.IsAppInsights\". And then use \"DataDog.\" prefix for all this settings, environment variables etc..? Such settings are, essentially, public APIs because they may conflict with customer data. Regardless of that , AppInsights has an s at the end :)"
+    },
+    {
+      "prediction": "I think this should be `ctx->log_key && ctx->time_key && ctx->time_key_format && ctx->log_key`",
+      "reference": "this assignment is not necessary if the offsetof() is used in the configmap"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `if` block above.",
+      "reference": "I'd rather leave a TODO here at least."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that the `body` is not `None` in the case of a 404.",
+      "reference": "Shouldn't this be `resource`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that `value` is initialized to `0` before the loop starts, so it will always be `0` even if the loop is not executed.",
+      "reference": "Why does this become blocking for the CPU path? Shouldn't it remain independent of the GPU path?"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this method. `configure` sounds like it's doing more than just setting up the environment.",
+      "reference": "Do we need to change occurences where `setup` was called before?"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this. I think it would be better to put it in the `resolveReferences()` function above.",
+      "reference": "Is this TODO still valid? Or are we going to put this somewhere else later?"
+    },
+    {
+      "prediction": "`timestamp` is not used in this function. Can you remove it?",
+      "reference": "We generally don't do parantheses in if conditions in Python. :) this could be better written as `if from_ts is None`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in `tm_process_req_requestregioninfo` since it's the main place where we call `HbaseTM_process_request_regions_info()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregioninfo()`, which is called from `HbaseTM_process_req_requestregion",
+      "reference": "Odd that we'd set just element 299 of a 2000-byte buffer to null. But I see that this is the way it was before."
+    },
+    {
+      "prediction": "I'm not sure if we want to add a new interface for this, but I think it would be good to have a separate interface for this. It would make it clear that we don't want this to be part of the `ClosestPeerer` interface, and it would make it easy to change the implementation if we decide we want to do it in a different way.",
+      "reference": "ScoreFunc is not used in this PR. Maybe to add it later when needed?"
+    },
+    {
+      "prediction": "I don't think we should be setting the Location header if the request is not a browser request. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the request is not a browser request, then the Location header should not be set. If the",
+      "reference": "super nit: I think this might be easier to read as `requestHeadersFromResponseWriter`"
+    },
+    {
+      "prediction": "Why remove the mutex, thread and atomic?",
+      "reference": "Need to include \"common/base/Base.h\" in front of this line"
+    },
+    {
+      "prediction": "nit: I think we can remove `matches` and `native-element-type` now since they are no longer used.",
+      "reference": "Changes in this file should go into a separate PR."
+    },
+    {
+      "prediction": "I think this should be `public static String quoted(String text) { return new String(text, StandardCharsets.UTF_8); }`",
+      "reference": "Please ensure that this will not start putting single quotes where double quotes are expected."
+    },
+    {
+      "prediction": "I think this should be `pingerPort: pingerPort` and not `pingerPort func(int, int) int`",
+      "reference": "`ProvideConfig` was changed, looks like it will not compile for windows, and should be changed too."
+    },
+    {
+      "prediction": "I don't think we need to pass `u` here, since it's already available in `ctx.User`",
+      "reference": "This does not look right, you're listing all private repositories.."
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "An OPTIONS request should respond from _any_ path? Seems weird to me..."
+    },
+    {
+      "prediction": "This should be `width` and `height` and not `width` and `height` and `max_width` and `max_height`",
+      "reference": "you need to add the content type"
+    },
+    {
+      "prediction": "I think this should be `TraceAgentProcess` and `StatsAgentProcess` instead of `traceAgentProcess` and `statsAgentProcess`.",
+      "reference": "Do we need to distinguish between these two processes? I'm thinking maybe we can have a list of processes and treat them all equally."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `get_clean_call_switch_stack_size_x64` and `get_clean_call_switch_stack_size_aarch64`.",
+      "reference": "This is used only for out-of-line -- so yes this seems right to do for x64. Inlined is aligned separately at the end of prepare_for_clean_call(). There the ifdef is x86_64 or MACOS -- no ARM, why not? Also, please add || MACOS here to match the inlined."
+    },
+    {
+      "prediction": "I'm not sure about this change. It seems to me that we should be able to use `zipkin_span_id` and `zipkin_annotation_span_id` instead of `zipkin_annotated_span_id` and `zipkin_span_id`.",
+      "reference": "guess I'm wondering if this needs to be refactored to use Schema.joinCondition() or similar?"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Align the elements of a hash literal if they span more than one line.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Does it make sense to restrict this option to `exec` only? You could add it specific to that subcommand using the `method_option`...method. There's an example of it for hosts in the `init` function."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "This is very expensive (lots of allocations, lots of data copying). Could we do something like check IndexOf('%') and bypass if no hit?"
+    },
+    {
+      "prediction": "This should be `|| client.getVar(Varbits.IN_WILDERNESS) == 1`",
+      "reference": "This should be `WorldType.isDeadmanWorld(client.getWorldType())` to be inline with the other WorldType calls."
+    },
+    {
+      "prediction": "`mlvl` is short for Multi-Level Quantization, which means Multi-level quantization.",
+      "reference": "do we also need to change `mlvl_confid` -> `mlvl_confidences`>"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "reference": "I feel slightly that InterfaceExclude is not a clear name - bearing in mind that our config names are, to some extent, an external API. From an external point of view, a clearer name might be IPVSInterfaces. Then it would obviously make sense for the value to be something like 'kube-ipvs0', and it would be a matter of Felix's internal implementation that we choose to exclude those interfaces from our monitoring. Alternatively, if we want to keep this at a level that is more general than just IPVS, perhaps 'UnmonitoredInterfaces'? WDYT?"
+    },
+    {
+      "prediction": "I think this should be `re.match(r'[A-Z][^\\\\]', path, re.IGNORECASE)`",
+      "reference": "This should be `re.search` with a `^` anchor added to the regex, as what we want here is really any path starting with something like `E:`."
+    },
+    {
+      "prediction": "I think this should be `determineVirtualService` instead of `determineVirtualServiceHealth`.",
+      "reference": "Since `IsKubernetesBuiltInResource` at L69 returns false due to lack `networking.istio.io/v1alpha3` in `builtInApiVersions` within `pkg/app/piped/cloudprovider/kubernetes/resourcekey.go`, it will never reach this point."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the otlp-metrics package since it's related to otlp-metrics.",
+      "reference": "Does codecov not run this test? Not sure how else it would not be covered."
+    },
+    {
+      "prediction": "I think it would be better to have this as a subcommand of `migrate` instead of a subcommand of `multisend`. The reason is that `migrate` is a general-purpose tool that can be used for many different things, whereas `multisend` is only used for bytecode generation. If we want to make `migrate` more reusable, we could add a `migrate` subcommand that accepts a single argument, which would be the bytecode file to be migrated. This would allow `migrate` to be used for more things, such as generating migration scripts from a database schema or generating migration scripts from a shell script.",
+      "reference": "can we add some sample of JSON_DATA format in the usage?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `ScopeManager` ctor.",
+      "reference": "Using `Tracer.Instance` in here is problematic for testing   It will likely cause some other tests to break I think - that's why we started passing in `IScopeManager` EDIT: I see you used `[TracerRestore]` - maybe that'll be enough!"
+    },
+    {
+      "prediction": "nit: I think we can remove the `@AutoValue` annotation here as well.",
+      "reference": "make all of these `static final` fields UPPER_SNAKE - they are constants."
+    },
+    {
+      "prediction": "This should be `damage * (itembonuses.DoTShielding + itembonuses.MitigateDotRune[SBIndex::MITIGATION_RUNE_PERCENT] + aabonuses.MitigateDotRune[SBIndex::MITIGATION_RUNE_PERCENT]) / 100`",
+      "reference": "No possible chance for weird overflows here, right?"
+    },
+    {
+      "prediction": "I think this should be called `DEF_EXTERNAL_COST` instead of `DEF_COST`.",
+      "reference": "Or we could use a Float object to more clearly show as user-settable via non-null?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, since the forkchoice was already updated in the forkchoiceUpdated method.",
+      "reference": "finalizedHeader should be saved in FinalizationManager when we have FinalizationBlockHash != Keccak.Zero"
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this step. I think it should be something like `data-disk-importer` or `data-disk-scratch-importer`.",
+      "reference": "Not using ExactName: true would remove the need for manually adding workflow ID as it would be added automatically by Daisy. E.g. 'disk-importer-2-import-ovf-7mn7h' was created from diskImporterDiskName above even though only 'disk-importer-2' was specified. ExactName: true should be used for resources that shouldn't include any temp prefix/suffix, such as disks that will be permanently attached to a created instance. Scratch disk is not one of them."
+    },
+    {
+      "prediction": "I don't think this is correct. The `column_scols` is a list of `ColumnSelectorExpr`s, not a single `ColumnSelectorExpr`. You need to do `sdf.select(self._internal.index_scols + [scol_for(sdf, col) for scol, col in zip(column_scols, data_columns)])`",
+      "reference": "If we support multi-index column later, we need to rename to fit the pandas' requirement."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. It's not a value in a JSON, it's a map. I think `JSONMap` might be better.",
+      "reference": "I'm not sure this is really needed. JSONValue type can be used the same as a map as far as operators go, including range. I'd leave this out for now unless there is a strong reason to keep it."
+    },
+    {
+      "prediction": "I think this should be moved to the top of the function, so that it gets executed before the loop below it.",
+      "reference": "sorry for nitpicking but I would love a new line under this line"
+    },
+    {
+      "prediction": "I think this should be `AccountingStrategy` instead of `AccountantID`",
+      "reference": "ChannelImplementation field naming is not clear, is it some kind of standard? Maybe it can be named ChannelID?"
+    },
+    {
+      "prediction": "I think this method should be called `removeAllFiles` instead of `RemoveFiles`",
+      "reference": "Looks like there is a typo: `expireSnapshots` -> `removeFiles` or whatever name we go with."
+    },
+    {
+      "prediction": "I think this should be `values[i] != null` instead of `currentSortOrder == values[i] || currentSortOrder == null`",
+      "reference": "Thanks for looking into this. I think it looks a bit strange to have this check inside the for loop. Wouldn't it also work to initialize `idxCurrentSort` with 0 instead?"
+    },
+    {
+      "prediction": "Can we have a constant for this value instead of hardcoding it?",
+      "reference": "Just a small ordering thing: Could you please bring `IMPORTANCE_HIGH` to the bottom and maybe even have one line between it and the Notification IDs?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem here is that the function `MolOps::Kekulize` doesn't throw an exception, it just returns a `const char*`. So when you try to catch this exception, you get a `const char*` instead of a `std::string`.",
+      "reference": "Should this be boost logged?"
+    },
+    {
+      "prediction": "I think this should be called `session` instead of `session_object`",
+      "reference": "What is the reason for doing expunge here? This method is purely for converting the timestamp."
+    },
+    {
+      "prediction": "I think this should be `structField.name()` instead of `structField.name().toLowerCase()`.",
+      "reference": "This calls sanitize twice if the name isn't valid."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in the `ApprovalGroup` model, since that's where the approval flow is defined.",
+      "reference": "Is there an equivalent that'd allow zero or one?"
+    },
+    {
+      "prediction": "I think this is the only place where we check if repetition is based on task object name itself. If it is, then we need to set the task object name.",
+      "reference": "Is it possible to update/re-evaluate entire metatask object for every repeat? If we decide/need to use a repeatWith resources in let us say labelSelector(bad example) then we would have to add another logic to update those properties."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could get it to work.",
+      "reference": "unnecessary conversion (from `unconvert`)"
+    },
+    {
+      "prediction": "I think you can remove this line too.",
+      "reference": "typo in commit message `openening`"
+    },
+    {
+      "prediction": "I don't think you need to include `ADIOS_CPP.h` here.",
+      "reference": "User-code doesn't see the internal ADIOS headers anymore. This whole include block should just be `#include <adios2.h>`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test_views.py`.",
+      "reference": "nit: I guess we can omit this (unless you had a reason to specify it)"
+    },
+    {
+      "prediction": "I think this should be `c.isSpokeLeafNode()`.",
+      "reference": "Why is Hub public?"
+    },
+    {
+      "prediction": "I think this is the only place where we call `appendDockerDependentCapabilities` from `ecsAgent`. I think we should move it to `appendDockerCapabilities` so that we don't have to repeat the code in `appendDockerDependentCapabilities` in `appendDockerCapabilities`.",
+      "reference": "why is this deleted?"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this. I think it would be better to put it in the `before_action :check_prerequisite` in the `Topic` model.",
+      "reference": "Maybe move that into a partial"
+    },
+    {
+      "prediction": "I think this should be `require.NoError(t, dao.Start(ctx))` instead of `require.NoError(dao.Start(ctx))`",
+      "reference": "Error return value of `dao.Stop` is not checked (from `errcheck`)"
+    },
+    {
+      "prediction": "I think this should be `Warnf` instead of `UserOut.Warnf`.",
+      "reference": "util.Warning()? Easier to say."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I'm not sure we want to log every change to the trie. I think we want to log every change to the trie once every 10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "this default is kinda superfluous (although i see it was already there before PR )"
+    },
+    {
+      "prediction": "I think we can remove the `parentNode` field from the `NodeList` interface and the `NodeList#add` and `NodeList#remove` methods can just call `setParentNode(parent)` and `parentNode = parent`.",
+      "reference": "How is this related?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to go about it, but it's the only way I could get this test to pass locally.",
+      "reference": "I'm breaking one of the rules here, it feels like the right thing to do. Open to alternative suggestions."
+    },
+    {
+      "prediction": "I think this should be `sw_vers -productVersion 2>/dev/null | cut -d . -f 2 -l 1`",
+      "reference": "The use of `#should` was triggering an rspec depracation warning for me, which is why I switched this to the rspec 3 style."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix, but it's the only thing I can come up with right now.",
+      "reference": "Why are you setting the CacheChange as REQUESTED when it is not valid (it was erased from history)?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `AnalyticsDashboardWidget` component, since that's the parent component for all of these widget types.",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "I think this should be `org.apache.avro.Schema avroSchema) {`",
+      "reference": "This needs to be public so that Avro can call it."
+    },
+    {
+      "prediction": "I think we can get rid of the `import { apiFetch } from 'GoogleComponents/data'` line.",
+      "reference": "`Google*` shouldn't be under External dependencies - seems like ESlint is not properly recognizing that these are aliases to internal dependencies."
+    },
+    {
+      "prediction": "I don't think this is correct. If `data` is a `pandas.core.series.Series`, then `data.to_dataframe()` should return a `pandas.core.series.Series` object, not a `pandas.DataFrame` object.",
+      "reference": "I think he meant the docstring in this constructor. yea we should fix"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it should be in `BinaryDictionary` since it's only used by that class.",
+      "reference": "so .. this basically follows the pattern from JapaneseTokenizer, I think. .. but somehow I don't see where we defined ResourceScheme? We're not referencing the one in kuromoji, right?"
+    },
+    {
+      "prediction": "I think we can drop the `Tls11` and `Tls12` flags, since we don't support those versions.",
+      "reference": "It would probably be better to do: const SecurityProtocolType Tls13 = (SecurityProtocolType)12288"
+    },
+    {
+      "prediction": "I don't think we should remove this.",
+      "reference": "Similar to the above, put this setting back but keep the value as `'{}'`"
+    },
+    {
+      "prediction": "I think this should be moved to the `if __name__ == '__main__'` block, so that it's only run when the script is actually being run.",
+      "reference": "As discussed in chat - we should run this function within an app context which means that we'd already have a logger configured, and a connection to timescale set up"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the `about` div and add a `#about-content` class to it so that we can use the `collapse` and `collapse-toggle` classes from the `discourse/lib/discourse.js` file.",
+      "reference": "I'm curious about why this needed to be removed."
+    },
+    {
+      "prediction": "I don't think we need to import `six` here.",
+      "reference": "why did you remove `zip`? It is used!"
+    },
+    {
+      "prediction": "I think this should be `MarkTopicReadyReason`.",
+      "reference": "Is there any reason that we don't add a similar condition like `ChannelConditionTopic` to the channel?"
+    },
+    {
+      "prediction": "This file should be renamed to `TestVendor\\Goto\\GotoTest.php`",
+      "reference": "before php8 \"goto\" was a reserved word and was not allowed as part of the namespace. Now test checks for validity of plugin namespace according to PSR-4"
+    },
+    {
+      "prediction": "I don't think this is correct. We need to support TLS.",
+      "reference": "nit: I'm not sure it can be addressed in this diff, but it seems odd that NewOutbound supports TLS but NewSingleOutbound does not. As a somewhat naive user I would expect the only difference between these two APIs is how peers are chosen."
+    },
+    {
+      "prediction": "I think this should be `yarpcerrors.Newf(yarpcerrors.CodeInternal, err.Error())` instead of `return nil, yarpcerrors.Newf(yarpcerrors.CodeInternal, err.Error())`",
+      "reference": "I would advise to enhance the test for the method `CallOneway` - we should have a test very similar to `TestCallSuccess`. We should test: - Success with response (even if it is callOneway) - Success with no response and empty payload - Errors"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think it would be better to use a regular expression that matches what we expect to be an email address, rather than a regex that matches what we don't expect to be an email address. For example, what do you think of: `^\\S+@[a-z\\d-]+(\\.[a-z\\d-]+)*$`?",
+      "reference": "We're already matching +. Not sure why this changed."
+    },
+    {
+      "prediction": "I think this should be `sonata.classification.default_category` instead of `sonata.classification.manager.category`",
+      "reference": "Please add line breaks, so the line doesn't exceed 80 chars."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `Instrumentation\\AspNetCore\\DiagSourceSubscriber`, right?",
+      "reference": "I initially thought (inccoreclty) this is a breaking change! The public api analyzer is a gift!"
+    },
+    {
+      "prediction": "I think this should be `g_hash_table_new_full (g_hash_table_hash, g_hash_table_equal, g_free, g_free);`",
+      "reference": "I'm not sure how this one didn't segfault before."
+    },
+    {
+      "prediction": "nit: I think it would be better to have a constant for the port number instead of hardcoding it.",
+      "reference": "Same points as in other test file."
+    },
+    {
+      "prediction": "These imports are not needed.",
+      "reference": "Did you run gradlew tidy? Wildcard imports shouldn't be there, hence the question."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "Shouldn't this be `ON_FPGAINFO_ERR_GOTO(res, out, \"setting type to FPGA_IFC_DFL\");`?",
+      "reference": "Is this going to restrict the output to *ONLY* ports that are bound to vfio-pci? That's not what we discussed yesterday."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. If we're going to do it here, we should do it in a way that's consistent with the rest of the verification code.",
+      "reference": "Intention is to do the lookup here, I believe (where we can cache it). This would make the call non-pure function though. I'll confirm with lotus. ID addresses are the *only* addresses I expect to actually see here."
+    },
+    {
+      "prediction": "I think this should be `emr_fs` instead of `emrfs`",
+      "reference": "electronic medical record file system (EMRFS)"
+    },
+    {
+      "prediction": "I'm not sure about this one. I think it's better to keep it as it is.",
+      "reference": "Put this higher in the list, like after \"bind\", since I guess we're considering the max request body size to be kind of a native/fundamental feature, built directly into the vhosts."
+    },
+    {
+      "prediction": "I think we can remove `ErrInvalidOrderType` and `ErrAskNotFound`, `ErrDeviceNotFound` and `ErrMinerNotFound` can be removed too.",
+      "reference": "oh come on :("
+    },
+    {
+      "prediction": "I think this should be `LOG.info(String.format(\"Detected dialect: %s\", toReturn.dialect.toString()));`",
+      "reference": "This is an incorrect change. The dialect spoken is an important part of the handshake and should be communicated to users."
+    },
+    {
+      "prediction": "I think this should be `bool: Whether or not the installer is running as a service account`",
+      "reference": "Remove as this is not needed anymore."
+    },
+    {
+      "prediction": "I'm not sure this is a good example of what we want to do here, but I'm not sure we have a better one.",
+      "reference": "This new class is for checking that #4551 works correctly with PlantUML output too."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `certificates.go` since it's only used there.",
+      "reference": "Will this cause all certificates to be re-issued?"
+    },
+    {
+      "prediction": "I think this should be `tmp = NULL;` after the `if (FPGA_OK != result)` check.",
+      "reference": "I think tmp is also leaked at the end of this loop if no error cases are encountered."
+    },
+    {
+      "prediction": "I think this should be `keyField.name()`, not `keyField.getSchema().name()`.",
+      "reference": "The previous version didn't use `keyProjection` because the entire key must be projected. If the key is a struct of multiple columns, then projecting a subset of those columns can easily introduce key collisions that aren't in the original data."
+    },
+    {
+      "prediction": "I think this should be `auto spaceId = ectx()->schemaManager()->toGraphSpaceID(spaceName);` instead of `spaceId = ectx()->schemaManager()->toGraphSpaceID(spaceName);`",
+      "reference": "Why not use StatusOr ? We can't ensure spaceId is greater than zero, especially when AdHocSchemaManager is used."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `rpc` package, but I'm not sure.",
+      "reference": "Does it seem we only use that as the json-rpc client? Do we have any other choice? It's a little weird."
+    },
+    {
+      "prediction": ".ROLE_MENUBUTTON, controlTypes.ROLE_TREEVIEW, controlTypes.ROLE_CHECKMENUITEM, controlTypes.ROLE_RADIOMENUITEM)\n \t\t\tor (role in (controlTypes.ROLE_TABLE, controlTypes.ROLE_TABLECELL, controlTypes.ROLE_TABLEROWHEADER, controlTypes.ROLE_TABLECOLUMNHEADER) and not formatConfig[\"reportTables\"])\n \t\t\tor (role in (controlTypes.ROLE_LIST, controlTypes.ROLE_LISTITEM) and controlTypes.STATE_READONLY in states and not formatConfig[\"reportLists\"])\n \t\t\tor (role in (controlTypes.ROLE_FRAME, controlTypes.ROLE_INTERNALFRAME) and not formatConfig[\"reportFrames\"])\n+\t\t\tor (role in (controlTypes.ROLE_DELETED_CONTENT,controlTypes.ROLE_INSERTED_CONTENT,controlTypes.ROLE_LINK, controlTypes.ROLE_HEADING, controlTypes.ROLE_BUTTON, controlTypes.ROLE_RADIOBUTTON, controlTypes.ROLE_CHECKBOX, controlTypes.ROLE_GRAPHIC, controlTypes.ROLE_CHART, controlTypes.ROLE_MENUITEM, controlTypes.ROLE_TAB, controlTypes.ROLE_COMBOBOX, controlTypes.ROLE_SLIDER, controlTypes.ROLE_SPINBUTTON, controlTypes.ROLE_COMBOBOX, controlTypes.ROLE_PROGRESSBAR, control",
+      "reference": "Could you split this into multiple lines?"
+    },
+    {
+      "prediction": "I think this import should be removed.",
+      "reference": "Doesn't the snapshot ID start off as null? It seems like we don't need to set it here."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "I wondered why `buffer` was passed into `LookupPath`. What is happening here? Is `buffer[:0]` equivalent to `[]V4TrieEntry{}`, and hence `buffer` isn't needed any more?"
+    },
+    {
+      "prediction": "I think we can remove the `sigValidator` and `powerTable` fields from the `Expected` struct, since they are no longer used.",
+      "reference": "nit: Ideally we would would use this abstraction everywhere we need this translation. I believe it's needed in the mining worker and the storage and market connectors."
+    },
+    {
+      "prediction": "I think this should be `UpdateHash(cStorPoolGot, common.CalculateHash(cStorPoolGot.Spec.Disks.DiskList))` and not `UpdateHash(cStorPoolGot, hash)`",
+      "reference": "need to fix the error message here"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "Unsure of how to handle this one."
+    },
+    {
+      "prediction": "I think it would be better to log the reason why the kill is being requested.",
+      "reference": "Will this.flowFinished verify with every finished state stated in Status.isStatusFinished method? If so, no need to verify isKilled."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. I think it should be something like `MaxRetryAttempts` or `MaxRetryIntervalInSeconds` or something like that.",
+      "reference": "replace \"Child Workflow\" with \"any workflow\""
+    },
+    {
+      "prediction": "I think this should be `data-order=\"-1\"`, not `data-order=\"-2\"`",
+      "reference": "Instead of showing the words \"symbolic link\" under the \"Size\" column, how about we introduce new icons for symbolic link to file and symbolic link to directory?"
+    },
+    {
+      "prediction": "I think this should be `@param string $rangeFilter The Lucene query language filter to be used for querying Solr`",
+      "reference": "Please add back the description of the parameter"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Superfluous space inserted."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to check if the node is an SVG element and only then try to get the attribute.",
+      "reference": "Didn't fix the problem. `href=\"\"` for SVG will still result in `undefined` getting returned by this function."
+    },
+    {
+      "prediction": "I think this should be `240` instead of `238`.",
+      "reference": "This seems like a strange change ... I wouldn't think that a change to access tags in the profiles would result in different bearings in this tests?"
+    },
+    {
+      "prediction": "I think this should be `api.setNavigatorObject(eventHandler.lastQueuedFocusObject, isCaret=False, isFocus=False)`",
+      "reference": "As isCaret is False by default, explicitly specifying isCaret=False here is not needed, and is perhaps a little confusing. I'd prefer that anywhere in the codebase that isCaret is only ever specified if it needs to be set to true. Mirroring that of isFocus for setNavigatorObject."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "We can't freeze this and then add something to it in the next line. The `freeze` on line 43 is sufficient. If Rubocop flags this we need to exclude it."
+    },
+    {
+      "prediction": "I think this should be `sort.Slice(aps[i].Subjects, func(i, j int) bool { return len(aps[i].Subjects[i]) > len(aps[j].Subjects[j]) })`",
+      "reference": "Why is this needed?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put it. I think it would be better to put it in `user/repo`.",
+      "reference": "Why not put two other routes after this line (221)?"
+    },
+    {
+      "prediction": "I think this should be `std::nullopt`.",
+      "reference": "If FLAGS_meta_server_addrs is not empty, in which case \"toHosts\" return empty array?"
+    },
+    {
+      "prediction": "I think this should be `Settings.JSONBoolWithQuotes(DEFAULT_JSON_USE_QUOTES);`",
+      "reference": "There is a function to output a \"JSONBool\" string. That would be a good start for finding uses. But maybe just have a look at where we decide whether it is a numerical or not, thus wrapping quotes around its value. Then you have it all I guess."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it should be `data._kdf._internal.copy(sdf=sampled, scol=scol).to_pandas()`",
+      "reference": "oops, it should be `data._scol` so that it respects the expression stored in Series. Let me fix it quick."
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "Thanks for moving this up with the rest of the variables. Much tidier :)"
+    },
+    {
+      "prediction": "I think this should be `0.0001` instead of `0.001`",
+      "reference": "I think `Rate` is a more precise and concise term for `PaymentPerTime`."
+    },
+    {
+      "prediction": "I think this should be `posEnd` instead of `posNext`.",
+      "reference": "I think this should maybe be: var limit = input.IndexOfAny(new[] { '+', '-', ']' }, **posNext**); You pretty much want to ignore any sign that is before the \"M\". But, if there is a sign somewhere before the M, and also after the M, you'd want to be able to find the sign after the M, right?"
+    },
+    {
+      "prediction": "I think this should be done in `checkValidDelivery` method, not in `checkCartNotEmpty` method.",
+      "reference": "if `checkStockNotEmpty` returns a reponse, you must return it or your script will continue its execution."
+    },
+    {
+      "prediction": "I'm not sure if we want to change the name of this package.",
+      "reference": "Any specific reason for this change? klog should work correctly here."
+    },
+    {
+      "prediction": "I think this should be `RETURNING id, as_epoch(last_modified) AS last_modified;`",
+      "reference": "we don't mention `last_modified` here?"
+    },
+    {
+      "prediction": "I think we can remove the `inputEl` variable now, since it's not used anymore.",
+      "reference": "same deal about the global state maybe interfering as in ActionBrowseTagline"
+    },
+    {
+      "prediction": "I think this should be `throw new ArgumentOutOfRangeException(\"maxCapacity\", string.Format(\"LruCache capacity cannot be less than {0}\", maxCapacity), nameof(maxCapacity));`",
+      "reference": "so sweet to see a test as the first thing"
+    },
+    {
+      "prediction": "I think this should be `config->http_proxy = getenv(\"HTTP_PROXY\");`",
+      "reference": "usually when HTTP_PROXY=\"\" (empty), proxy should be disabled. could you set it to NULL when it's empty string?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the width is null, then the height should be the same as the width, not the other way around.",
+      "reference": "Can you throw an exception if width or height are both null"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep it as it was before.",
+      "reference": "This is a breaking change. I guess these might have been intentionally exported."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the same place as `targetSplitSize(ops)`.",
+      "reference": "I don't think this is necessary. When options are used to set the split size in the Spark reader, it uses `TableProperties.SPLIT_SIZE` to pass it here. That should work for metadata tables as well, right? The situation that we need to handle in this PR is setting the default, like you had before. We just want to use a metadata split size so you can set it and not have the same split size used (by default) for both metadata and data scans."
+    },
+    {
+      "prediction": "I think it would be better to have a `String()` method on `BinaryOperator` that returns the string representation of the operator. That way, we don't have to duplicate the logic in both `BinaryOperator` and `UnaryOperator` for the string representation.",
+      "reference": "wait a sec, shouldn't this be `'*'`?"
+    },
+    {
+      "prediction": "I think this should be `varMap[\"compute_service_account\"] = oi.params.ComputeServiceAccount` instead of `varMap[\"compute_service_account\"] = oi.params.ComputeServiceAccount`",
+      "reference": "It think you'll want the var for GMI import as well: daisy_workflows/ovf_import/create_gmi.wf.json (Unfortunately there's duplication between the two :/ )"
+    },
+    {
+      "prediction": "I think it would be better to use `self._commands[\"GET_CONTEXT\"] = (\"GET\", \"/session/$sessionId/moz/context\")` instead of `self._commands[\"GET_CONTEXT\"] = ('GET', '/session/$sessionId/moz/context')`",
+      "reference": "Nit: Group this with the other commands by moving it down one line."
+    },
+    {
+      "prediction": "I think this should be `throw new UnsupportedOperationException(\"Unknown encoding: \" + dataEncoding);`",
+      "reference": "I would like to emphasize that a user can use non-vectorized reads to handle this file so maybe something like \"Cannot perform a vectorized read of ParquetV2 File with encoding %s, disable vectorized reading with $param to read this table/file\""
+    },
+    {
+      "prediction": "I think it would be better to call it `parameter_name` instead of `param_name`.",
+      "reference": "Maybe add `positional=False`. It does not need it more than the other parameters, but one must start somewhere."
+    },
+    {
+      "prediction": "I think this should be `rsWithPrimaryPath` and `rsWithoutPrimaryPath` should be `f('%s/../spec/max-staleness/ReplicaSetWithPrimary', __dirname);` and `f('%s/../spec/max-staleness/ReplicaSetNoPrimary', __dirname);`",
+      "reference": "Since we're here, can we use a template?"
+    },
+    {
+      "prediction": "I think this should be moved to the top of the file.",
+      "reference": "saw you're also doing this in `test_types.py`: `spark = _spark_context()['test'].resources['spark'].resource_fn(None)` since `_spark_context()` uses `spark_session_local` won't the above break the tests?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/testutil/testutil.go` since it's only used by `test/testutil/testutil.go` and `test/testutil/testutil.go` should be the home of unit tests that are not related to each other.",
+      "reference": "`has(default)` i think is the preferred way of doing this"
+    },
+    {
+      "prediction": "I don't think we should be using the `force` keyword here.",
+      "reference": "always put a space after a comma"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. It should be done in the `onItemLongSelected` method of the `MediaSelectionMediationActivity`",
+      "reference": "@angmas1 move the else up, next to the closing bracket of the if block. Also, there is no need for the braces as your else statement contains only a single line. Make your if-else block similar to the block in lines 277-280."
+    },
+    {
+      "prediction": "I think we can remove the `file_detector` argument from this class.",
+      "reference": "@AutomatedTester @davehunt thoughts on a new keyword argument?"
+    },
+    {
+      "prediction": "I think this should be `Integer` instead of `int`",
+      "reference": "This change shouldn't be necessary for this PR. Please remove it."
+    },
+    {
+      "prediction": "I think we can remove the `testCtx` variable now, since we don't need it for this test.",
+      "reference": "This variable is mostly unused, how about we make `NewTestCtx` return a pointer?"
+    },
+    {
+      "prediction": "I think this should be `maxPayloadLen = 32 * 1 << 20 // 32 Mi`",
+      "reference": "What is the basis of this value?"
+    },
+    {
+      "prediction": "Is this file still needed? I don't see it referenced anywhere in the codebase.",
+      "reference": "Nit: was this extra blank line added intentionally?"
+    },
+    {
+      "prediction": "I think this should be STATUS_GUARD_PAGE_VIOLATION instead of STATUS_GUARD_PAGE_VIOLATION",
+      "reference": "nit: inconsistent spacing around ="
+    },
+    {
+      "prediction": "I think this should be `Sonata\\MediaBundle\\Model\\ClassificationInterface`",
+      "reference": "We still need this import"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "We are pulling this because we removed \"groupID\" state and put it into the strategy implementations"
+    },
+    {
+      "prediction": "I think this should be `#include \"oneapi/dal/algo/kmeans_init/compute.hpp\"`.",
+      "reference": "Is this file actually related to PCA?"
+    },
+    {
+      "prediction": "I think this should be `isAdmin`, not `owner`.",
+      "reference": "See above, this should probably remain `isFirstAdmin`."
+    },
+    {
+      "prediction": "I think this should be `type Expression` instead of `type MonoType`",
+      "reference": "Why the change to make it public? The expression interface has the `TypeOf` method?"
+    },
+    {
+      "prediction": "Should this be `public void SingleArchiveFileRollsCorrectly() {`?",
+      "reference": "don't mind the snake cases in the test names. If there are in the test cases, it's fine IMO"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I would expect it to be in `Identity/identity.go` or `Identity/identity_locking.go` but not here.",
+      "reference": "`Id` defines REST resource address and should not be in payload"
+    },
+    {
+      "prediction": "`--class_wise` -> `--class-wise`",
+      "reference": "We can omit the underscore and just use `classwise`."
+    },
+    {
+      "prediction": "This should be safe as the root category has to exist for a given context but I do not like fatal errors.",
+      "reference": "Why not throw an exception instead then?"
+    },
+    {
+      "prediction": "I think this should be `getInputFile(task.file().path().toString())`",
+      "reference": "Looks like this doesn't need to change. Can you revert this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `cli/__init__.py` or `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py` in `cli/__init__.py",
+      "reference": "please import it like `from kinto.config import template`"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for `false` and `true` instead of `true` and `false`.",
+      "reference": "There are lots of default value changed, not sure if it break the old behavior."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `github.com/elastic/beats/libbeat/common`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "`addToStaticNodes` is not used anywhere, please remove it.",
+      "reference": "parameter should be called removeFromStaticNodes"
+    },
+    {
+      "prediction": "`override` should be `bool`, not `bool | False`.",
+      "reference": "The handler will have an extra property in the marker to signal the implementation is in the provider, for the cases we want to add functionality, for example, custom implementations not in moto."
+    },
+    {
+      "prediction": "Do we need this method at all? The `active?` and `deactivate`/`activate` methods seem to be sufficient to determine if a subscription is active.",
+      "reference": "Is this method actually being used anywhere? If not, I think we should remove it."
+    },
+    {
+      "prediction": "I think this should be `!= null && (bodyParameterType.ComposedProperties.Count(p => !p.IsConstant) <= Settings.Instance.PayloadFlatteningThreshold || bodyParameterType.ComposedProperties.Count(p => !p.IsReadOnly) <= Settings.Instance.PayloadFlatteningThreshold)`",
+      "reference": "Thanks! I somehow lost this between my far too many branchs."
+    },
+    {
+      "prediction": "I think this should be renamed to `SeleniumWebDriver` instead of `RemoteWebDriver`",
+      "reference": "i think this should get put into its own file. This could start to grow :)"
+    },
+    {
+      "prediction": "I think this should be `gcr.io/google_containers/pause` instead of `amazon_containers/pause`",
+      "reference": "Can you add a TODO for loading the tarball of the pause image into Docker?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Is there any value in setting the process to null like this?"
+    },
+    {
+      "prediction": "`softmax_axis_` is a member variable of SoftmaxWithLossLayer. You can access it directly.",
+      "reference": "This could be a good time to add a check that the prediction and target dimensions (except channel) agree?"
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to do, but if it is supposed to be here, it should be in a separate commit.",
+      "reference": "The requirement that OP_isb is bb-final needs to be a rule for clients, just like the rule that syscalls and interrupts must be bb-final: i.e., please add it to the dr_register_bb_event() docs."
+    },
+    {
+      "prediction": "I think this should be `Globalize.getString(\"ListPaging\").replace(\"{0}\", (totalRecordCount ? startIndex + 1 : 0) + \"-\" + recordsEnd).replace(\"{1}\", totalRecordCount);`",
+      "reference": "The translate library has a method to replace the variables."
+    },
+    {
+      "prediction": "I think this test is a bit too specific to the `fillna` method. Can we instead test `fillna` with different values? For example, `fillna(0, method='ffill')` and `fillna(0, method='bfill')`?",
+      "reference": "i don't think this test case is correct. in both cases inpalce=True returns nothing. We need to compare the ks. Also we probably need to make a copy of it. Otherwise you pollute the following \"ks\" because ks has been changed."
+    },
+    {
+      "prediction": "I think this should be `tls.TlsHandshakeHook`, not `tls.TlsHandshakeHook`.",
+      "reference": "Any proposals how to make the naming somehow include the \"completed\" idea of this hook? `TlsHandshakeCompletedHook` or similar? Or using the `...Start/End...` scheme?"
+    },
+    {
+      "prediction": "I think this should be done in the `SyncReport` class, not in the `_syncReport` class.",
+      "reference": "it introduces a lot of complexity to the state, can we flush straightaway or create a two level flush store where the questions are read form unflushed data?"
+    },
+    {
+      "prediction": "I think this should be `preg_replace('/[^a-zA-Z0-9\\-]/', '-', $node->id)`",
+      "reference": "There are two calls to getUrlFromRouteCache, but you're only urlencoding one of them. Is that intentional? Would it make more sense to do the url-encoding inside the getUrlFromRouteCache function?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `getEditedCell` method.",
+      "reference": "Maybe we can use `this.instance.getCell` with `topMost` flag to get `TD` element?"
+    },
+    {
+      "prediction": "I think this should be a debug log.",
+      "reference": "Change this to Debug Level"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but I don't have a better idea.",
+      "reference": "I refactored the Aggregate, those are hard to read"
+    },
+    {
+      "prediction": "I think this test should be moved to a separate file, since it's not related to the rest of the tests.",
+      "reference": "This is now done in the after method, so this test won't pass any more."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "What was the point of creating `user_block` if you're then not going to use it ;-)"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "@ueshin, I thought we can remove this entire file. Does that require some more works?"
+    },
+    {
+      "prediction": "I think this should be `unicode(tokens.string.strip()[1:-1])` instead of `unicode(tokens.string)[1:-1]`",
+      "reference": "Just `return tokens.string[1:-1]` is enough"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Use 2 (not 0) spaces for indentation.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "1 trailing blank lines detected."
+    },
+    {
+      "prediction": "This should be `setPrefWidth(0)`.",
+      "reference": "Isn't this the default max width? In any case I would prefer to see these definitions in the css files."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way I could get it to work.",
+      "reference": "Is this thread-safe? (If not, and if the solution is to introduce locking, let's reconsider how we do the scheme changing instead...)"
+    },
+    {
+      "prediction": "I don't think we need to change anything in this file.",
+      "reference": "Looks like this file has tabs instead of spaces."
+    },
+    {
+      "prediction": "I think this should be `seelog.Infow` instead of `seelog.Warnf`.",
+      "reference": "Hmm should we clean up container resources before task resources here? or the order does not really matter here?"
+    },
+    {
+      "prediction": "I think this should be reverted.",
+      "reference": "Please undo this :wink:"
+    },
+    {
+      "prediction": "Layout/AlignHash: Align the elements of a hash literal if they span more than one line.",
+      "reference": "thanks for adding this missing association"
+    },
+    {
+      "prediction": "I think we can remove the `new` keyword here.",
+      "reference": "Should I remove this in favor of the HttpsConnectionFilterOptions overload?"
+    },
+    {
+      "prediction": "Why not using `dba::selectMax` here?",
+      "reference": "This shode be nullable, there alway sholud be at least one status in system , if not exist that mean data problme"
+    },
+    {
+      "prediction": "I think we should move this to `app/Http/Middleware/UninstallSW` so that it can be used in other controllers as well.",
+      "reference": "Spaces not tabs"
+    },
+    {
+      "prediction": "I think this should be `fs.readdirSync(path, {withFileTypes: false}).map(e => e.name);`",
+      "reference": "To get the files inside a directory, please use `readDirStats()`."
+    },
+    {
+      "prediction": "I'm not sure if this is a good name for this. I think it's a bit confusing because we have two different Docker and Plz containers. I think we should rename it to something like `ContainerImplementation` or `ContainerType`.",
+      "reference": "might be worth to call these `Isolation` instead of containers here and when presented to the user -- docker/rkt , in addition to cgroups and namespaces, also provide image discovery and filesystem preparation; there's also the security context and probably 1-2 other small things"
+    },
+    {
+      "prediction": "Can we add a license header to this file?",
+      "reference": "File is empty, consider removing?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `auth.py` instead.",
+      "reference": "This seems to return the registry_url. The more meaningful URL is the catalog URL, which is the URL the user specifies in `quilt3.login`. The username might also be as useful here if not more useful."
+    },
+    {
+      "prediction": "I think this should be `L1StandardBridgeAddress` and `GasPriceOracleAddress` should be `L2StandardBridgeAddress` and `L2GasPriceOracleAddress`",
+      "reference": "Was the addition of `GasPriceOracleAddress` here erroneous?"
+    },
+    {
+      "prediction": "Space inside { missing.<br>Space inside } missing.",
+      "reference": "I believe that you are going to have to do some work here to get the env var support for these values to still work correctly, otherwise they will get stored as :department instead of host_tags[:department]."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it would be better to iterate over `mSessions` instead of `mStoreSubscription.getTabs()`. The reason is that `mStoreSubscription.getTabs()` is a subset of `mSessions`, but `BrowserState.getTabs()` is a superset of `BrowserState.getTabs()`. This means that `BrowserState.getTabs()` will always return all sessions, whereas `mStoreSubscription.getTabs()` may not.",
+      "reference": "How was this causing the exception?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to move it to `css` folder.",
+      "reference": "I still have issues with it loading Simplified Chinese by default over Japanese, since they also share characters and we're not sure if characters are different or not. As-is, this fixes Traditional Chinese and Simplified Chinese, but we're not sure if it'd still screw up Japanese text or not. I maintain that, in my opinion, Japanese should be the only script of this kind loaded by default (Because of Hiragana and Katakana, that aren't shared with any other script) and either variant of Chinese should be loaded only in their respective case."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be moved to `IntervalsSource.java` since it's only used by `MultiTermIntervalsSource.java`, which is also the parent of `MultiTermIntervalsSource.java`.",
+      "reference": "Isn't it better to stub AtomatonQuery and yield it here. IIRC it resolves simplest MTQ highlighting cases as nobrainer."
+    },
+    {
+      "prediction": "I think we can remove the `discount_percentage_vs_current_plan_annualized` method and just call `discount_percentage_vs_current_plan` directly.",
+      "reference": "Line is too long. [96/80]"
+    },
+    {
+      "prediction": "I think this should be `kubernetes.io/os: linux` instead of `kubernetes.io/os: [linux]`",
+      "reference": "My only concern with changing the node selector here is that someone else _could_ have built their own images for other platforms and set them to be used using the flag override on the controller, which in turn this change would break. Perhaps not changing the selector for acmesolver pods makes most sense, and then working on actually having support for Windows? wdyt?"
+    },
+    {
+      "prediction": "Why do we need this change?",
+      "reference": "Why do we need interface `InnerObjectWithSelector`, I searched the usage with this field, it seems nowhere use `GetSelectorSpecs()` methods provided by `InnerObjectWithSelector`"
+    },
+    {
+      "prediction": "I think we should move this to `init.go`",
+      "reference": "Do we need this `init` here? We have already called the `InitContext` in `StartModule` directly."
+    },
+    {
+      "prediction": "I don't think we need to check for w3c here. We can just do `self.execute(Command.MAXIMIZE_WINDOW)`.",
+      "reference": "This `if` is not necessary"
+    },
+    {
+      "prediction": "I think we can remove the `dom.value` and `dom._lastValue` vars now.",
+      "reference": "should we add a `mangle.json` mapping for this? We could reuse a property name that's only used on component or vnode objects right now, like `__s` (\"next state\")."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Please update the yard docs to indicate this new option."
+    },
+    {
+      "prediction": "Nit: I think we can remove the `Subnets` field from the `AWSClusterSpec` struct now, since it's not used in the `AWSCluster` struct anymore.",
+      "reference": "Can you provide an example in the godoc section on how to use this? From the implementation it looks like the AvailabilityZone field is required for example, otherwise it can fail / error, is that correct?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think we should be able to just do `tx.FirstOrCreate(&Migration{Version: latestSchemaVersion})` and it should work.",
+      "reference": "It would be great if this could be captured by a test."
+    },
+    {
+      "prediction": "1*time.Second),\n+\t\tESProcessorAckTimeout:             dc.GetDurationProperty(dynamicconfig.WorkerESProcessorAckTimeout, 1*time.Minute),\n \t}\n }\n \n```\nReview comment?",
+      "reference": "1000 -> 100 maybe too much, 200 ish to 500ish maybe a good option"
+    },
+    {
+      "prediction": "I think this should be `[]byte(addr.String())` instead of `[]byte(addr.GetChecksumAddress().String())`",
+      "reference": "Any reason we need to keep this package's `ChecksumAddress` type or can we get rid of it too?"
+    },
+    {
+      "prediction": "I think this should be `globalize.translate('ButtonRestart', { locale: 'en' })`",
+      "reference": "Above the one that was picked was the Button* prefix, and here it's the Header* prefix. Maybe the Button prefix is more general."
+    },
+    {
+      "prediction": "I think this should be `BestSuggestedHeader?.Hash` instead of `Head?.Hash`",
+      "reference": "I might prefer changing: public Block FindPendingBlock() => FindBlock(PendingHash, BlockTreeLookupOptions.None); public BlockHeader FindPendingHeader() => FindHeader(PendingHash, BlockTreeLookupOptions.None); in IBlockFinder, what do you think?"
+    },
+    {
+      "prediction": "I think this field should be renamed to `ApplyPolicy` or `Policy` to be consistent with the other fields in this struct.",
+      "reference": "This is necessary due to a bug in 4.7. Follow HIVE-1561 for getting rid of it."
+    },
+    {
+      "prediction": "Style/HashSyntax: Use the new Ruby 1.9 hash syntax.<br>Style/HashSyntax: Use the new Ruby 1.9 hash syntax.<br>Style/HashSyntax: Use the new Ruby 1.9 hash syntax.<br>Style/HashSyntax: Use the new Ruby 1.9 hash syntax.<br>Style/HashSyntax: Use the new Ruby 1.9 hash syntax.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `XKeysymToString()`.",
+      "reference": "Some day (tm) I will also get rid of that one. But today is not that day..."
+    },
+    {
+      "prediction": "I think this should be `if (substr($localPath, -1) === '/') {`",
+      "reference": "Add a space between if and opening parenthesis please (i.e. `if (`)"
+    },
+    {
+      "prediction": "I think this should be `else:` instead of `scol = self._scol`",
+      "reference": "nit: `rename` instead of `alias`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `BlockDevices` function.",
+      "reference": "Per AWS docs EC instances have the root mounted at /dev/sda1. This label should be skipped."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "I'm not sure it is a better way to address lose message, if wait here, edge controller cant process message, event from watching api-server will be lost yet, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `EAP` class, since it's specific to the EAP protocol.",
+      "reference": "Please keep the correct alignment, it seems broken now (at least in Github). Can you reverse the test (`if p.value_size is None`)? Also, do you want `6` when `p.value_size is None` or `x + 6`? Maybe, in that case, something like `lambda p, x: x + 6 + (0 if p.value_size is None else p.value_size)` would be easier to read. Or even easier: `lambda p, x: x + 6 + (p.value_size or 0)`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be under `NetworkConfig`, but I'm not sure if that's the right place to put it as well.",
+      "reference": "Can we make sure ECS service treats it as int64 as well?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the UI module.",
+      "reference": "Please follow the naming convention for variables, i.e. `windowText`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the engine package, but I'm not sure if it's the right package to put it in.",
+      "reference": "alias can be `cast`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the memory mmap errors. The memory mmap errors should be added in the `h2o_iovec_t` struct, not in the `h2o_iovec_t` constructor.",
+      "reference": "Let's use `%zu` instead of casting to `uint64_t` and using `PRIu64`. The alternative is to change the type of `mmap_errors` to `uint64_t`, though I'd prefer not doing that because some 32-bit platforms might not provide atomic operation support for `uint64_t`."
+    },
+    {
+      "prediction": "I'm not sure about this one. I think it's fine, but I'm not sure I agree with the wording of the warning.",
+      "reference": "I recall there being a reason we had this one set only to suggeation. @carterkozak do you remember why? or is my memory getting corrupted?"
+    },
+    {
+      "prediction": "I think this should be `flatbuffers::StripPath` not `flatbuffers::StripExtension`.",
+      "reference": "This should now be made into `if (it->second.empty())` ?"
+    },
+    {
+      "prediction": "I think this should be `self.num_of_searches -= 1` instead of `if self.num_of_searches > 0: return`",
+      "reference": "Do we need to worry about a race condition on this decrement (@The-Compiler)? I'm not sure how the python callbacks work, so this might not need to be something to worry about."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/testpropagators.go`",
+      "reference": "Should these assert that the extracted `SpanContext` is remote?"
+    },
+    {
+      "prediction": "what is the difference between `ignore_wrt_candidates` and `match_low_quality`?",
+      "reference": "typo: allow low quality matches."
+    },
+    {
+      "prediction": "I think we can remove the `cfg *cmds.Server` parameter from this function.",
+      "reference": "At some point we need to condense down and refactor our \"configs\" as we have too many in too many places and we're starting to get more illegible code. No change necessary but calling out for future reference."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `app/assets/javascripts/blacklight/breaking/javascripts/breaking-javascripts.js.erb`",
+      "reference": "Can we add the `if label` conditional to this please?"
+    },
+    {
+      "prediction": "I think this should be `mlflow==0.2.0`",
+      "reference": "@thunterdb, can we increase the minimal version to almost the latest ones? I was thinking people will mostly use the latest version of mlflow"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this hook. I think it should go in the `beforeFill` hook.",
+      "reference": "If the last argument is going to be removed, why would we add it to the API docs? Shouldn't it be private for internal use?"
+    },
+    {
+      "prediction": "I think this should be `FileSystemIntermediateManager`, `InMemoryIntermediateManager`, and `IntermediatesManager` should be renamed to `FileSystemIntermediatesManager`, `InMemoryIntermediatesManager`, and `IntermediatesManager` should be renamed to `FileSystemIntermediatesManager`.",
+      "reference": "`Intermediates` or `Intermediate`"
+    },
+    {
+      "prediction": "I don't think this is the best way to do this. I think it would be better to use `isinstance(element, dict) and 'id' in element and 'id' not in dict.get('id')`",
+      "reference": "This is not the right exception class. There is an InvalidSelectorException class that covers bad locators."
+    },
+    {
+      "prediction": "I think this should be a warning, not an info.",
+      "reference": "If this code isn't needed, it should be removed."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to `doUploadMetrics` in `metrics.go` since it's only called from there.",
+      "reference": "Should the `doRequest` func be a method on the `connection` type instead of passing state from the type?"
+    },
+    {
+      "prediction": "I think we can remove the `ECS_CLI_ENVIRONMENT_NAME` variable as well.",
+      "reference": "we'll probably tackle these e2e tests at the end"
+    },
+    {
+      "prediction": "I think we can remove the `CoreFX` category here.",
+      "reference": "what is this address pointing to? what are we measuring here? I want to have a better understanding."
+    },
+    {
+      "prediction": "I don't think you need the `or dep_status = 0 or dep_status = 4` part here. You can just use `or dep_status in (0, 4)`",
+      "reference": "Would it be better to pass the dependency status as the parameter into the SQL string? In case the enum value is changed in the future, we don't need to change the code here."
+    },
+    {
+      "prediction": "Is this needed?",
+      "reference": "Hah I had `gorename` failing without this too."
+    },
+    {
+      "prediction": "I think we can remove this now, since we don't need it for the `appsec_blocking_enabled` property.",
+      "reference": "maybe prefix these with `appsec_` for consistency?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/trace` since it's used in both `trace.go` and `trace.go`.",
+      "reference": "nit: this should be down further with the other github imports"
+    },
+    {
+      "prediction": "I think this should be `if ctypes.windll.kernel32.FreeConsole() != 0 or ctypes.windll.kernel32.GetConsoleMode() != 0:`",
+      "reference": "Why is this change necessary or related to the rest of the PR?"
+    },
+    {
+      "prediction": "I think this should be ` - see: ` instead of ` - see: `",
+      "reference": "Hm, I wonder if the `see:` prefix is even necessary?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "This should be translated."
+    },
+    {
+      "prediction": "I don't think you need to check for num_class here, since you are checking for \"boosting\" first.",
+      "reference": "I originally had this PR only changing the R package, but then ran into this error > Error: [LightGBM] [Fatal] Cannot change metric during training This is thrown even if you aren't actually CHANGING `metric`. I think the change here in `c_api` is closer to the desired behavior, only throwing an error if the parameter is being changed."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "These changes are just to make cli_spec a little more readable for VS Code users, as there's a bug with the Ruby plugin's syntax highlighting when you use multi-line regex literals."
+    },
+    {
+      "prediction": "I think we should use `controllerutils.GetHiveNamespace()` here instead of `GetHiveNamespace()` in `LoadSecretData`",
+      "reference": "Shouldn't the global pull secret by in the namespace of the hive-operator? It is part of the operand."
+    },
+    {
+      "prediction": "I think this should be `sleep(2)` to give the consumer time to start up.",
+      "reference": "kinesis-mock applies KINESIS_LATENCY to RegisterStreamConsumer actions so I added a sleep here."
+    },
+    {
+      "prediction": "I think we can remove the `LOG(ERROR)`.",
+      "reference": "Only V1 or V3 is return from `getMetaVersionFromKV`. So where do we call `updateMetaV2ToV3`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this constraint.",
+      "reference": "I think that V0 is missing here, which will prevent plugins that haven't been converted to work."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should remove the `sitemap` from the `middleware` section of the config, since it doesn't seem to be doing anything.",
+      "reference": "Should we remove this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be.",
+      "reference": "Put on top of the import."
+    },
+    {
+      "prediction": "I think this should be `analytics` instead of `anatlysis`",
+      "reference": "See above, this shouldn't be added."
+    },
+    {
+      "prediction": "This should be `ITEM_ANGUISH_ORNAMENT_KIT`.",
+      "reference": "This'll likely get fixed upstream, we should let them deal with it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `libp2p/host.go`",
+      "reference": "Somehow, this field does not seems useful for the exposed package api, only for the tests. Would you consider having an unexported field in `hostFactory func(context.Context, ...libp2p.Option) (host.Host, error)` instead to be set only by a new helper function defined in export_test.go. This is just a suggestion, not a hard requirement for the PR approval."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a function that takes the time duration in milliseconds and returns the time in seconds since the epoch. Then you can call `goBackInTime` and `goForwardInTime` on the result of this function.",
+      "reference": "As a first argument to these function, please pass the date that should go forward/backward. Also please clarify what is \"n\" (possible values, unit) and what is timeDuration (possible values, unit, as from your code it seems to be \"day\", \"hours\", etc. but from your example it seems to be milliseconds)."
+    },
+    {
+      "prediction": "I don't think you need to import `config` here. You can just use `config.conf[\"general\"][\"language\"]` directly.",
+      "reference": "Are you sure you really need this code here? that script will be certainly run in NVDA's main thread, and core would have already called setLanguage."
+    },
+    {
+      "prediction": "I think this should be 10256 instead of 10255",
+      "reference": "why is this changed ?"
+    },
+    {
+      "prediction": "I think this should say \"except to update the document's revision.\"",
+      "reference": "Should this be \"the new revision value\"?"
+    },
+    {
+      "prediction": "I think this should be `Dette feltet m\u00e5akse n\u00e5r :other er :value.`",
+      "reference": "You have deleted :attribute"
+    },
+    {
+      "prediction": "I don't think it's a good idea to add copyright notices to the source code.",
+      "reference": "The license header should be updated to the Apache one."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I would expect it to be in `IntegrationTests`, but I don't see it there.",
+      "reference": "@zacharycmontoya Is there any way to distinguish a `SocketHttpHandler` from another `HttpMessageHandler` request?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should go in `src/InstrBuilder.cpp` under the `instr_is_*` functions.",
+      "reference": "Not sure about this one -- did you check somehow?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Is there a good reason to keep this at the `pe_ver_win` name now that it's per-host?"
+    },
+    {
+      "prediction": "I think it's better to use `getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R.array.nav_drawer_items_product, getResources().getStringArray(R",
+      "reference": "Not in the order of display : Front, Ingredient, Nutrition here (which is the right thing), Actually displayed: Front, Nutrition, Ingredients"
+    },
+    {
+      "prediction": "I think we can remove the `StartTime` field from the `StressInstance` struct as well.",
+      "reference": "Why delete `StartTime`? This `StartTime` was used to avoid the PID was reused."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `after_failure` block, but I'm not sure if that's the right place to do it.",
+      "reference": "My concern here, is that if the dump_puppet_log also throws then we will lose the data about the teardown_exception."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. The `daal_knn::Parameter` needs to be a `std::vector<size_t>`, `std::vector<int>`, or `std::vector<size_t>` depending on the compiler.",
+      "reference": "Should it be `int64_t`?"
+    },
+    {
+      "prediction": "I think this should be 1000 * 60 * 60 * 5 = 3000000.",
+      "reference": "is this value related to something ?"
+    },
+    {
+      "prediction": "Shouldn't this be `(AddTxResult?,)`?",
+      "reference": "Add names to tuple elements ValueTask<(Keccak? Hash, AddTxResult? AddResult)>, should they both be nullable?"
+    },
+    {
+      "prediction": "I think this should be `caseFormat.replaceAcronyms(namePiece)`",
+      "reference": "I think it might make more sense to do this in Name.upperCamel; it is the entry point for upper camel strings."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `t` is an `ArrayType` and `s[0]` is an `ArrayType` as well. I think you need to do something like `type.ifArrayType(t -> s[0] = (ArrayType) t);`",
+      "reference": "Huh? The point is that a cast isn't necessary"
+    },
+    {
+      "prediction": "I think we should rename this to `OPENEBS_API_SERVER_INSTALL_CRD` to be consistent with the `OPENEBS_IO_INSTALL_CRD` variable.",
+      "reference": "can we name it like `InstallV1Alpha1CRDs` ?"
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "I need something to match when creating selectorsyncsets so added this label."
+    },
+    {
+      "prediction": "I'm not sure why this is here, but it seems like it should be in `spec_helper.rb`.",
+      "reference": "Isn't `nil` the default return value from an empty `rescue` clause? If that's correct, then we could just remove the `nil` line entirely as it doesn't serve a purpose."
+    },
+    {
+      "prediction": "I think we should remove this snippet. It's not clear to me what this code is supposed to accomplish.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "Why do we need to do this? I think we can just do `r.idx.Sub(r.idx.Obsolete())` in `LoadIndex`.",
+      "reference": "Wouldn't that cause the obsolete indexes to be downloaded over and over again? After all these are still stored in the repository."
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "We might want to move this logic to a Plain Old Ruby Object down the road. Not a blocker."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to move it to the `updateToolbox` function.",
+      "reference": "move `this.toolbox_.position();` to just after line 1477. Context: the if statement on line 1472 checks whether this is a toolbox with categories, and if so it populates the toolbox. Positioning the toolbox is a reasonable followup to that, and means you don't need an extra if. You may also need to call `this.flyout_.position()` after line 1483, which is the equivalent for the non-category toolbox."
+    },
+    {
+      "prediction": "Seems like this should be `if (peptideGroupDocNode != null && sequenceProteinsDict.TryGetValue(peptideDocNode.Peptide.Target.Sequence, out var sequenceProteins))`",
+      "reference": "Should the function above be .FirstOrDefault() instead? Otherwise, why check for null and tell ReSharper to ignore the fact that it can never be null?"
+    },
+    {
+      "prediction": "`STATUS_MAP` is not a constant, it is defined in `job.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which is in `jobs.js` which",
+      "reference": "Let's not create new connection, but rather user `common.db` one"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this. I think it would be better to put it in the `ImageTag` class, since that's where all the image-related logic lives.",
+      "reference": "Can this use asset_path rather than defining an image_url helper?"
+    },
+    {
+      "prediction": "This should be `const stubString = () => '';`",
+      "reference": "Hindley-Milner signatures don't use literal types. They explusively use types. It's deal with that"
+    },
+    {
+      "prediction": "I think we should rename this to OFPolicyRuleGroup as it is not specific to OpenFlow.",
+      "reference": "Could we just extend PolicyRule with more fields? It was originally designed as the struct required by openflow client, nested structs seem not helping. And for the name of the fields, initialism should be uppercase, OFID and NPName doesn't look good, then maybe FlowID, PolicyName.."
+    },
+    {
+      "prediction": "Please add a space after `json_last_error` and before `opcache_reset`.",
+      "reference": "I really dislike this alignment approach, as it makes identifying individual functions quite hard, but I kept consistent with the previous style. Since no sorting was evident, I pushed it to the end of the block. Meanwhile, is this supposed to be tested or meaningfully testable somehow?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to update the state of the `ActiveExecution`. I think it should be in `handle_event`, right?",
+      "reference": "Could also consider modifying the ActiveExecution..."
+    },
+    {
+      "prediction": "I think this should be `ValueError` instead of `ValueNotAsConst`.",
+      "reference": "Does it make sense to change that to a custom error instead of `ValueError`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way I could think of.",
+      "reference": "I guess this will need to change soon."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `http` package.",
+      "reference": "Restore this spacing - not sure why it has to be removed?"
+    },
+    {
+      "prediction": "I don't think you need to define _POSIX_C_SOURCE here.",
+      "reference": "These defines cause issues in base.h leading to failing android builds. Locale independency is now set in multiple spots, in cmake and in base.h. The base.h uses feature test macros wrong. Feature test macros can be used to indicate that a feature should be made available. They do not guarantee that it is available. On the cmake side this is done by the check symbol macros (similar to what autoconf does). On the android side not. I would suggest removing this from base.h and define locale independent in gradle. It is available since android 21 I believe. Is there a way in gradle to set locale independent only if android api >= 21?"
+    },
+    {
+      "prediction": "nit: `loopBody != getDst` can be simplified to `loopBody != getSrc`",
+      "reference": "Why not delete these codes?"
+    },
+    {
+      "prediction": "I think this should be `if (FLAGS_meta_client_retry_times > 0 && !metaClient_->waitForMetadReady(FLAGS_meta_client_retry_times))`",
+      "reference": "If pull message failed after some reties, we should ensure the storaged exit, otherwise it has risk to loss data."
+    },
+    {
+      "prediction": "I don't think this is needed, but it's a good habit to get into.",
+      "reference": "looks like you don't need to import first and count here anymore"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `engine/visibility.go`",
+      "reference": "FYI, visibility task processing is local only, meaning pushing something to local ES cluster / local archival endpoint"
+    },
+    {
+      "prediction": "I don't think you need to import `DynamicPropertyFactory` as it is not used in this class.",
+      "reference": "This import can be removed."
+    },
+    {
+      "prediction": "I think this should be `InvalidRegionError{Region: a.Region}`, and the error message should say something like \"region not allowed in ARN: fips-xxx or -fipsxxx\".",
+      "reference": "1. Nit: We can have a general helper to identify if a region is pseudo region. We might already have one somewhere. 2. Does any type of ARN support FIPS? If not, we may just move this check in arn package?"
+    },
+    {
+      "prediction": "I think this stubber should be renamed to `LambdaStubber` as well.",
+      "reference": "is this phrase completed in a next line? Or forgot to add... what?"
+    },
+    {
+      "prediction": "I think it would be better to move this to the `models` subdirectory and rename it to `mask_rcnn_resnet50_fpn_inception_v1_1x.py`",
+      "reference": "`('train', 5)` and `('train', 1)` are actually the same."
+    },
+    {
+      "prediction": "I think we should use `float64` here.",
+      "reference": "Would it be too complicated to refactor `currLimit` to be an `int64`?"
+    },
+    {
+      "prediction": "I think this should be `RSPEC_SKIP_IMPLEMENTATION` instead of `::RSpec::Core::Pending.instance_method(:skip)`.",
+      "reference": "Would it be more performant to just `include ::RSpec::Core::Pending` here? /cc @myronmarston"
+    },
+    {
+      "prediction": "This is a bit of a nitpick, but I think it would be better to have the `ChaosStatus` field be a `map[string]interface{}` rather than a `map[string]int64` so that we don't have to do `fmt.Println(obj.GetChaosStatus().(map[string]interface{}))` in the tests.",
+      "reference": "Should be podiochaos?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it's better to use `codecs.getcodec` instead of `codecs.register` and `codecs.lookup`.",
+      "reference": "Is the `codecs.register(lambda name: None)` gracefully handled by codecs? The lambda is expected to return a tuple of functions (encoder, decoder, stream_reader, stream_writer) (or a CodecInfo object), according to the docs. I've only checked python's built-in help, though."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `MeterProviderBuilderSdk` class, since it's related to the `MeterProviderBuilder` class.",
+      "reference": "in this case, it could be either instrument name or the view name, which could be invalid. might be better if we can offer very specific message. (not blocking this, we can add this as a follow up,if needed)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `switchTo` command, since that is the command that switches to a different window.",
+      "reference": "I'd rather just send the parameter twice than break encapsulation here. There's already precedence with webelement IDs"
+    },
+    {
+      "prediction": "`cascade_rcnn_mask_rcnn_r50_fpn_20e_coco.py` should be renamed to `cascade_rcnn_mask_rcnn_r50_fpn_20e_coco_backbone.py`",
+      "reference": "You may put 'open-mmlab://res2net101_v1d_26w_4s' here. MMCV will be updated later."
+    },
+    {
+      "prediction": "I think it would be better to have a constant for this, like `const PAGE_SPEED_REPORT_STRATEGIES = { mobile: STRATEGY_MOBILE, desktop: STRATEGY_DESKTOP };`",
+      "reference": "May be worth making this more specific, to avoid potential conflicts, maybe `pagespeedDashboardWidget`? Or `webVitalsDashboardWidget`?"
+    },
+    {
+      "prediction": "I think we can remove the `Error` class altogether. We can just use the `Error` function here.",
+      "reference": "I guess this assignment is now redundant?"
+    },
+    {
+      "prediction": "I think we can remove the `blockNum` variable and just use it in the `if err != nil` condition.",
+      "reference": "See above note re: blockNum synchronization."
+    },
+    {
+      "prediction": "I think we should use `backoff` from `github.com/cenkalti/backoff/v3` instead of `backoff` from `github.com/spiffe/spire/pkg/agent/backoff`",
+      "reference": "wonder if we could avoid doing named imports if we type aliased `backoff.Backoff` in the `.../agent/common/backoff` package?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "Or maybe put into alphabetical order with the other #includes."
+    },
+    {
+      "prediction": "I think this should be `ticker := time.NewTicker(1 * time.Second) defer ticker.Stop()`",
+      "reference": "This runs the entire duration of the program; this change is not needed."
+    },
+    {
+      "prediction": "I think this should be `false`",
+      "reference": "I don't feel well about the `ALWAYS_UPDATE` option   &nbsp; sounds like a workaround. I can imagine that some of the plugins want to observe all options. Maybe, in that case, we can reuse the `CONFIG_KEYS` that would return an empty array for observing all settings and `false` for disabling observing at all? I just thinking aloud - Like `CONFIG_KEYS` sounds reasonable. We could support and maintain it in the future, even after implementing the \"observability\" to the MetaManager then `ALWAYS_UPDATE` ... I don't know :)"
+    },
+    {
+      "prediction": "I think clang-format should be removed from this patch.",
+      "reference": "A \"tab difference\"? As in, there are tab characters somewhere? This disabling of clang-format for all the code here seems like overkill. Are you sure there's not some other solution? I don't really understand the problem though. Do you mean clang-format-diff?"
+    },
+    {
+      "prediction": "I don't think you need to import * here.",
+      "reference": "Could you please leave the explicit imports?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You need to check if `methodConfigMap.get(methodConfigProto.getName())` is not null before adding it to the list. If it is not null, then you can add it to the list. If it is null, then you can skip it.",
+      "reference": "Can we check for containsKey here?"
+    },
+    {
+      "prediction": "I think it would be better to use `fmt.Printf` here, rather than `fmt.Fprintln`.",
+      "reference": "Exit successfully. We served the help the user wanted."
+    },
+    {
+      "prediction": "I think this should be `InitialWorkingDirPath` and `InitialPackagePath` should be `InitialWorkingDirPath` and `InitialPackagePath` should be `InitialWorkingDirPath`.",
+      "reference": "Had to rename this because `InitialPackage()` already existed."
+    },
+    {
+      "prediction": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Can we put this in a different helper?"
+    },
+    {
+      "prediction": "I'm not sure this is correct. It seems to me that if the number of slots is 8-byte aligned, then the dstack_offs should be 8-byte aligned as well.",
+      "reference": "I'm not sure where your 31 comes from? I see 15 here and 2 above for pc and flags, for 17 total beyond the SIMD."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "If you want, you could change this to `allowCors(response` (like in \"version\", above) to allow Cross-Origin Resource Sharing. I'm sort of wondering what's returned if no API terms of use are set."
+    },
+    {
+      "prediction": "I don't think this is needed. The `onConfigurationChanged` method is only called when the user switches between portrait and landscape mode. In this case, the screen is already in portrait mode, so there's no need to call `super.onConfigurationChanged(newConfig)` again.",
+      "reference": "Why do we need to override this if all we're doing is passing it up?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the control plane.",
+      "reference": "Just wondering whether it's safe to send all CQDs to the child compiler. As this missing CQD is causing trouble, maybe some other user CQDs that now get sent could also cause the statement to fail? Also, if I have a table with a nullable unique column, should I have to set this CQD to create another table like it, or should the CREATE TABLE LIKE work without the CQD?"
+    },
+    {
+      "prediction": "I think this should be `DAGSTER_EARLIEST_RELEASE` and not `DAGSTER_FROM_SOURCE`",
+      "reference": "EARLIEST_TESTED_RELEASE might be slightly more clear"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could find that worked.",
+      "reference": "I think replacing `.*` with `\\s*` from after the hash makes sense, but not the rest of the changes."
+    },
+    {
+      "prediction": "I think we should rename this to `EnableLoggingEnabled` or `EnableLoggingFlag` to make it clear that this is a flag.",
+      "reference": "interesting .. did not know varint tag covered bool as well"
+    },
+    {
+      "prediction": "I think this should be `uint64_t` instead of `H2O_VECTOR(uint64_t)`",
+      "reference": "Could you please elaborate the reason for not using `gkc_summary` like the values above? I ask this, because I assume what we want to collect is a histogram of event loop latency, and we use `gkc_summary` for collecting histograms."
+    },
+    {
+      "prediction": "I think this is the only place where `getTxHashes` is called with an object instead of an instance.",
+      "reference": "getHash should be a public and without an `_` if used here."
+    },
+    {
+      "prediction": "I think this should be `uint64(blk.Header().Number.Uint64()) > uint64(now)`",
+      "reference": "Unrelated to these changes -- we will probably want some special handling for genesis block validation."
+    },
+    {
+      "prediction": "I think this interface is not needed. We don't need to set the disk block cache fraction in the disk block cache setter. The disk block cache setter will set the disk block cache fraction based on the disk block cache size and the number of blocks in the cache. The disk block cache size is determined by the number of blocks in the cache and the cache block size. The number of blocks in the cache is determined by the number of blocks in the cache and the cache block size. The cache block size is determined by the number of blocks in the cache and the cache block size. The cache block size is determined by the number of blocks in the cache and the cache block size. The cache block size is determined by the number of blocks in the cache and the cache block size. The cache block size is determined by the number of blocks in the cache and the cache block size. The cache block size is determined by the number of blocks in the cache and the cache block size. The cache block size is determined by the number of blocks in the cache and the cache block size. The cache block size is determined by the number of blocks in the cache and the cache block size. The cache block size is determined by the number of blocks in the cache and the cache block",
+      "reference": "This is probably overkill since we don't actually need to pass this specific interface anywhere, but I don't mind it!"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `d.p2pService.Peers()`",
+      "reference": "@janos I moved this back here, to avoid having multiple add peer calls being handled. I think it's either this or a global lock across the whole function. I think this is a better approach, especially since `hive.Broadcast` is blocking . We can remove the map entry on errors if needed. LMKWYT?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for this, but it's the only way I could get it to work.",
+      "reference": "Minor: can we put the conditional outside of the partial?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "We need to find something more efficient than this. We can probably just check the first element and assume the type list is consistent.."
+    },
+    {
+      "prediction": "I think this should be `A %s is a job which can be programmatically run, either on a schedule or via service code in response to events.`, right?",
+      "reference": "How \" via service code in response to event.\"? I don't think we'll get in the no schedule option to the release. We should probably wait for some customer feedback first to make sure we're addressing different event-triggers the best way"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Please move unrelated formatting changes to their own PR (same with the resources.py file)."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to check the state of the `window._state_before_fullscreen` and only show the normal window state if it's not focused.",
+      "reference": "Make this a public attribute (remove the leading `_`) if you want to access it from the outside."
+    },
+    {
+      "prediction": "I think this should be `if (layoutManager) {`",
+      "reference": "this conditional is useless now so it should be removed"
+    },
+    {
+      "prediction": "I think this should be `static_cast<flatbuffers::uoffset_t>(i) + 1` instead of `static_cast<flatbuffers::uoffset_t>(i+1)`.",
+      "reference": "`vecofstructs->size()` already returns a `uoffset_t`. So if you just change the type of `i`, you don't need any casts."
+    },
+    {
+      "prediction": "I don't think we should expose a command-line option for toggling proto annotations parsing. It's not something that should be configurable by the user.",
+      "reference": "Nit: Maybe merge this with the configProto check on line 208? Don't feel that strongly though, please keep separate if you prefer."
+    },
+    {
+      "prediction": "I think this should be `if (doc.ismaster || doc.hello) {` instead of `if (doc.ismaster || doc.hello) {`",
+      "reference": "I know we already did this in 4.0 so carry on, but I'm facepalming here we should've made an overrided setMessageHandler that sets ismaster to true when hello is present. Esp in 4.0 where we already have some overrides written. We can keep going here and maybe open a ticket to some effect of that solution"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "I believe we could use `link_to` here (although not super important). Also, we should internationalize this value. I'm not 100% sure what our process is for adding new UI strings these days (e.g. skip the specs that will fail if we add a new key to only one language, use Google Translate to translate the value for us, some other option)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `GetPhysicalDeviceDisplayPropertiesKHR` function, since that is the place where `pProperties` is initialized.",
+      "reference": "This was already in vkGetPhysicalDeviceDisplayProperties2KHR..."
+    },
+    {
+      "prediction": "I think we can remove this interface since we don't use it anywhere.",
+      "reference": "nit: I'd expect a public method called IsDockerEngineRunning to return a boolean yes/no, not the error message. Can we change either the return value or the name? Something like `CallDockerEngine`"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine for now, but I'm not sure we'll be able to support all dialects in the future.",
+      "reference": "If i remembered correctly, type_name is '{RESOURCE_TYPE}/{RESOURCE_NAME}', if type_name is 700 max then name cannot exceed that size"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. It seems like it should be in the same namespace as the other file_identifier functions, but I can't find where it is used.",
+      "reference": "When calling functions with keyword arguments: please provide the arguments as keyword arguments, not positional arguments."
+    },
+    {
+      "prediction": "`None` is not a valid choice for `act_cfg` and `norm_cfg`.",
+      "reference": "We may follow the argument order: conv_cfg, norm_cfg, act_cfg."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I don't have a better idea. It seems like we should be able to do something like `@capabilities[:chrome_options] = { user_data_dir: \"#{profile_json[:directory]}\", extensions: profile_json[:extensions] }`",
+      "reference": "`options['binary'] = WebDriver::Chrome.path` if set?"
+    },
+    {
+      "prediction": "Is there a reason to not include `folly/ssl/Init.h`?",
+      "reference": "folly library should be put on the top."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with.",
+      "reference": "Do we have to restrict this to SVG elements? I think setting namespaced attributes on normal dom elements (while not as common) is still valid, so perhaps we can save some bytes by removing the `isSvg` check?"
+    },
+    {
+      "prediction": "I think this should be `this.hot.hasHook('hiddenRow') && this.hot.runHooks('hiddenRow', row);`",
+      "reference": "Please add it to the `REMOVED_HOOKS` constant."
+    },
+    {
+      "prediction": "nit: can you move this to the top of the file?",
+      "reference": "This variable needs `.v1` at the end of it, or the tests below need it. There are some test failures as a result"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `Account::Access` class, since it's only used by that class.",
+      "reference": "Is this access protected so that only admins can access this route?"
+    },
+    {
+      "prediction": "This should be `0.0f` instead of `DBL_MAX`.",
+      "reference": "What happens if the minimum x coord is > 0? This will incorrectly set it as zero, right?"
+    },
+    {
+      "prediction": "I think we can remove the `if o.maxp > 0 && batchSize > o.maxp` check here, since it's already covered by the `if batchSize > o.maxp` check above.",
+      "reference": "I think you want to look at current pending + batchSize and if that is > o.maxp no?"
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `\"internal\"`?",
+      "reference": "Should this be `codes.Error`? `codes.Internal` was an error status before."
+    },
+    {
+      "prediction": "Align the elements of a hash literal if they span more than one line.",
+      "reference": "I don't think you need these lines."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. It's not a limiter, it's just a limiter of the number of blocks that can be written to the disk at the same time.",
+      "reference": "You didn't mean to export this, right?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. The AWS SDK for JavaScript (v3) doesn't support Ruby, so we can't use it in this example. The AWS SDK for Ruby doesn't support Amazon SNS either, so we can't use it in this example either.",
+      "reference": "Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "I think we should rename this to `egressNetworkConfig` to be consistent with `egressConfig`.",
+      "reference": "Do we still need this?"
+    },
+    {
+      "prediction": "I think this should be `LOG.error(\"Failed to create private marker transaction: {}\", privateMarkerTransaction.getHash());`",
+      "reference": "We do have a disconnect between the privateMarkerTransactionFactory.getSender() call and the createPrivateMarkerTransaction() call. When the plugin receives the call to create the PMT it does not know what the sender was returned by the other call. At a minimum we need to pass in the sender to the second call. I would prefer that the plugin does the locking if that is necessary. If you are using random keys you don't have to lock! Besu could signal to the plugin when a PMT was un- or successfully added to the pool using a callback. Callback could be null if this is not needed. The plugin would have to be able to get the nonce for a certain key from besu. What do you think?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `components/alert.js`",
+      "reference": "This is a bit inconsistent with the `ErrorStartHourGreaterThanEnd` in the previous file. Wouldn't `ErrorDefault``fit better here?"
+    },
+    {
+      "prediction": "I think it would be better to move this to `skipIfNetworkPolicyStatsDisabled`",
+      "reference": "I think it will revert the change @antoninbas did on purpose 05eee251c9c53400277def576f92d614ca234898"
+    },
+    {
+      "prediction": "I think this should be `ifaceName == \"tunl0\" || ifaceName == \"wireguard.cali\"` instead of `ifaceName == \"tunl0\" || ifaceName == \"wireguard.cali\"`",
+      "reference": "nit: a switch perhaps?"
+    },
+    {
+      "prediction": "I think this should be `application_name` instead of `application_name`",
+      "reference": "Would it make sense for `opensearch_description_tag` to have default values for the title and href attributes?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to check for the journal being disabled. I think it should be in `getRangeFromJournal` instead.",
+      "reference": "Just noticed this while debugging -- it's unlikely to be causing issues in practice."
+    },
+    {
+      "prediction": "I'm not sure if this is the best timeout value to use here, but I think it's fine.",
+      "reference": "That seems unrelated - did that fix a random test failure you noticed while working on this issue?"
+    },
+    {
+      "prediction": "I think we can remove the `base` parameter from `doCommit` as well.",
+      "reference": "This kind of concern should be handled in the `TableMetadata` update methods. This is similar to `metadata.rollbackTo`, which not only sets the current version, but also updates the table history. I think that all of those methods should add the current object's `file.location()` to the previous metadata location list, if it is non-null. That way, the caller never needs to remember to update it."
+    },
+    {
+      "prediction": "`mlvl_tensors` should be a list or tuple of tensors.",
+      "reference": "We need to know why all the tensors are detached in `get_bboxed`."
+    },
+    {
+      "prediction": "I don't think this is a regression.",
+      "reference": "Unrelated but shouldn't this raise `attribute-defined-outside-init`?"
+    },
+    {
+      "prediction": "I think this should be `create_no_window=True`",
+      "reference": "I would rather no have this as a `kwarg` as it encourages \"growth\" which lead to an unweildy constructor in other classes. Let's add a method or property to take care of this instead as I think it's usage is going to be quite low."
+    },
+    {
+      "prediction": "I think this should be `plan&.sections.sum { |s| s.num_answered_questions }` instead of `plan&.num_answered_questions.to_i`",
+      "reference": "You just changed my life :) I grew very attached to this safe nav approach when working with Groovy years ago `obj?.method`. Glad to see its now a part of Ruby!"
+    },
+    {
+      "prediction": "I'm not sure if this solves the issue, but I think it's a good idea to remove the `clear-fix` class.",
+      "reference": "do you mean 'clearfix' class?"
+    },
+    {
+      "prediction": "I don't think we should have a `causalConsistency` option here. It's not something that should be configurable by the client.",
+      "reference": "`causalConsistency` is already defined on `SessionOptions`. We do not need to duplicate it here."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "This should not be included since there are no code changes."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I could come up with.",
+      "reference": "If all we need just response 200, then we could use `c.Status(http.StatusOK)` here."
+    },
+    {
+      "prediction": "I think we should be able to just do `client.Get(ctx, model.ResourceKey{Kind: apiv2.KindNode, Name: hostname, Namespace: \"\"}, nil)`",
+      "reference": "Should be checking for error here"
+    },
+    {
+      "prediction": "Why do you need this?",
+      "reference": "Put a space between `(array)` and `$this` and I'll merge it"
+    },
+    {
+      "prediction": "I think this should be `func Stack(buf []byte, all bool) ([]byte, error)`",
+      "reference": "You could add this to stack.go, which has similar functions (to avoid yet another small file). None of these are likely to be implemented considering what TinyGo is designed for."
+    },
+    {
+      "prediction": "I think this class should be called `NotificationContentHolder` instead of `NotificationHolder`.",
+      "reference": "the class names `NotificationHolder` and `NotificationsHolder` are fairly difficult to quickly tell apart, particularly since they are often used close to each other"
+    },
+    {
+      "prediction": "This import is not needed anymore.",
+      "reference": "Avoid wildcard imports"
+    },
+    {
+      "prediction": "I think this should be `[]*Candidate`",
+      "reference": "Need to do this to unblock mockgen. The interface method needs to be public to be visible in another package. The right fix is to delete this test only interface method"
+    },
+    {
+      "prediction": "I think this should be `loadbalanced` instead of `loadbalancer`",
+      "reference": "same question here as I have on the other file regarding doing this check here vs L112 vs not at all"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Adding `// +build !windows` (as a separate line) will do the same for Windows. So while you are at it we might add it as well."
+    },
+    {
+      "prediction": "I'm not sure why this was removed.",
+      "reference": "this is used as important way of mocking for unit tests, FYI"
+    },
+    {
+      "prediction": "I don't think we need to set the alpha here, since it's the same as `muted_alpha` in `options.Histogram`.",
+      "reference": "Sorry, should have been clearer, I meant adding this to the ``Histogram`` options one line above."
+    },
+    {
+      "prediction": "I think this should be `C23NHe2` instead of `C23NHx2`",
+      "reference": "Just ignorance on my part, but what is the significance of the change from C23NHe2 to C23NHx2?"
+    },
+    {
+      "prediction": "Can you add a docstring for this new parameter?",
+      "reference": "What about tuples or pandas series? Do we want to support lots of different types or force a single type?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Small thing, without the check above this will throw a NoNodeException if the path doesn't exist. Maybe this can be wrapped in a try/catch just for the NoNodeException, so that the more user-friendly message used above can be thrown instead: `throw new KeeperException.NoNodeException(\"No collection found at path: \" + path);` Just a thought, I'm not even sure it's worth it."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [122/120]",
+      "reference": "I don't know why this worked before, but we should be explicit on not verifying that a blank string (allowed as a way to remove a url/download_url) is a valid url as it is not."
+    },
+    {
+      "prediction": "I think we can remove the `support will be removed in Bolt 3.0.` part of the deprecation message.",
+      "reference": "Do we want to raise an error here instead of warning? Or would it be better to just see if Bolt happens to succeed, and let it fail on it's own if it fails? I'd lean towards raising an error, but that's different from \"removing support\"."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this method. I think it would be better to put it in the `Bolt::Transport::HTTP` class, since that's where the `options` and `filter_options` methods are defined.",
+      "reference": "This is kind of confusing with the other `filter_options`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to fix this. I think it should be fixed in the `OpsCantHandleFavorite` case, since that's where the error is thrown.",
+      "reference": "Could the `deleteFromFavorites` happen when `head == nil` because it's not initialized yet somehow but not because TLF doesn't exist? I was concerned if this would make it possible in any way to have favorites seemingly deleted on a device while it's not actually happened on server."
+    },
+    {
+      "prediction": "Why not use `std::numeric_limits<int64_t>::max()`?",
+      "reference": "It doesn't look like this particular \"rowCount\" variable is used anywhere. I suppose we could delete it. The code change is harmless though."
+    },
+    {
+      "prediction": "Shouldn't this be `w.WriteHeader(http.StatusNoContent)`?",
+      "reference": "Hmm, probably we should instead write whatever `status` is returned instead of coding a special case."
+    },
+    {
+      "prediction": "I think this is the only place where we need to set the `HttpMatch` field.",
+      "reference": "Could using the pointer-to-struct approach here cause confusion later? We've had several problems of that kind in the past. What does it mean if the struct is non-nil but its fields are nil? Is that even allowed? (libcalico-go question) Are there any validation requirements for this new rule addition; should it only be allowed if `Protocol == TCP`, for example?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "why this recursive relationship definition?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Handlers` struct, since that's where the `HandlerSpec` lives.",
+      "reference": "More sentence, less list please. > Encoding of the handler. This field is optional. We don't need to mention exactly what it's used for because that list can expand (as it has already in this change)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `gocloud.dev/metrics` or `gocloud.dev/trace` or `gocloud.dev/samples`.",
+      "reference": "Not part of this PR, but should the section above say what the name of the metric is for each method?"
+    },
+    {
+      "prediction": "I'm not sure how I feel about this, but I don't think this is the right place for this. I think it should go in the `DoubleGaugeMetric` and `LongGaugeMetric` classes, since they're specific to gauge metrics.",
+      "reference": "there are changes that are upcoming to this section. For now, this works (to demo the usage of Gauge), and the changes to this section are coming as separate PRs."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `Terminal` struct itself, not in the `Terminal`/`TerminalUI` interface.",
+      "reference": "This is equivalent to stdout. Why not just replace stdin with stdout in `IsProcessBackground`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add the `compress_method` to the man pages. I think it should be in the `zlib` and `deflate` man pages, since those are the ones used by the zlib and zlibdeflate libraries.",
+      "reference": "Rather than test for nullptr, just treat the pointer as a bool, i.e. `if(p_jmsg[\"compress_method\"])`"
+    },
+    {
+      "prediction": "Shouldn't this check be done in `createShortcut`?",
+      "reference": "Can we move the remaining code of this method to a new method? I don't like the `return;` here, we could try to replace it with an `if ... else ...` syntax. What do you think?"
+    },
+    {
+      "prediction": "I don't think we need to include this capability in the API.",
+      "reference": "nitpick: _and collections_"
+    },
+    {
+      "prediction": "I don't think this is needed. The timer is already set up in the `rp_generator_t` struct.",
+      "reference": "I believe these need to be explicitly initialized in `proxy_send_prepare`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `run()` method, since that's where the actual work is being done.",
+      "reference": "This is unnecessary. The declaration of `t0` on line 179 is still in scope inside the `except` block."
+    },
+    {
+      "prediction": "I don't think this is the right place for this property. It should be in the `request` class.",
+      "reference": "The `border_color_`... properties don't support solid colors, not only this one. Maybe `@param solid_color` should be used instead of adding this note to every one of them?"
+    },
+    {
+      "prediction": "I think it would be good to have a test for this file.",
+      "reference": "We don't need this file, we can use the one that is in the grid-ui directory"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to check for an empty line. I think it would be better to check if the line is empty using `isEmpty()`.",
+      "reference": "FYI: inlined this method in the internal review."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Werid, the checksum between this file and the one on branch `develop` is the same not sure why it's showing a diff"
+    },
+    {
+      "prediction": "I think this should be `metric.Meter.Name + \".\" + metric.Name` instead of `metric.Meter.Name + metric.Name`",
+      "reference": "Prometheus doesn't have concept of Meter (like OTLP does).. Trying to see if this is a good approach to use the meter name as namespace, to avoid name collisions, when same instrument name is used across multiple instruments, from different Meter."
+    },
+    {
+      "prediction": "Shouldn't this be `&& !$this->getAuthManager()->getAuthMethod() == 'Shibboleth'`?",
+      "reference": "The formatting of this is a bit strange; I'd suggest collapsing this back to a single line, and moving the `&&` to the beginning of the second line to meet the line length restriction."
+    },
+    {
+      "prediction": "I think this should be `else if (segments != NULL)`. The `else` should only happen when `else if (segments != NULL)` is true.",
+      "reference": "One or the other should be non-NULL. If segments is NULL, the alloc above will have size zero, which we do not allow (there's no header): it should assert in debug build. So there should be asserts that one is non-NULL at the top, and if there really needs to be some kind of defensive check down here, it should cover the alloc too."
+    },
+    {
+      "prediction": "I think this should be `user.update!(first_name: user_data['first_name'], last_name: user_data['last_name'])` instead of `user.update_attributes(first_name: user_data['first_name'], last_name: user_data['last_name'])`",
+      "reference": "I'm surprised rubocop isn't picking up singe quotes?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/version.go`",
+      "reference": "Is this line still needed?"
+    },
+    {
+      "prediction": "I'm not sure if this is correct. If the initial heartbeat period is set to 0, then the initial heartbeats will never be sent. If the initial heartbeat period is set to 1, then the initial heartbeats will be sent every 1 second. If the initial heartbeat period is set to 10 seconds, then the initial heartbeats will be sent every 10 seconds.",
+      "reference": "Maybe use specific values for all the watt.times fields? Even better, have a const for it at the top of the file?"
+    },
+    {
+      "prediction": "I think this should be `tables = [PATTERNS_TABLE]`",
+      "reference": "nit: UNICODE_BRAILLE_TABLE or something might be a better name for this. IMO, braille-patterns.cti is a terrible name. This table allows Unicode braille characters to be used anywhere to produce raw dots."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Why are we removing the option to skip machine pool generation?"
+    },
+    {
+      "prediction": "I think this should be `IHttpHeadersStartLineHandler` instead of `IHttpStartLineHandler`",
+      "reference": "\"Request line\" here too."
+    },
+    {
+      "prediction": "This should be `if (entity_list.GetTypeID(ZONE_CONTROLLER))` instead of `if (entity_list.GetNPCByNPCTypeID(ZONE_CONTROLLER))`",
+      "reference": "Please remove this."
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "Weird bug. Combining those joins and group calls was triggering AREL to generate the sql cache inside of will_paginate before the final call. This is a harmless workaround, but hints that will_paginate might becoming seriously deprecated."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [121/120]",
+      "reference": "This can be simplified as discussed before."
+    },
+    {
+      "prediction": "This is a bit of a hack, but I think it's the only way to make this work. The `table` argument to `NewProcessChunkMsg` is a copy of the `table` argument to `NewProcessMsg`. Since `NewProcessChunkMsg` is a copy of `NewProcessMsg`, `NewProcessChunkMsg`'s `table` will be a copy of `table`, and so `NewProcessChunkMsg`'s `table` will be a copy of `table`, and so on and so forth.",
+      "reference": "Does `internal` or `test` in the file name actually do anything here? Or is that just to show these functions are only for tests?"
+    },
+    {
+      "prediction": "I think we should remove the `true` argument here.",
+      "reference": "does this boolean indicate unsafe serialization?"
+    },
+    {
+      "prediction": "I think this should be `static copyBundledRealmFilesToAppDefaultFolder()`.",
+      "reference": "Maybe rewrite to `Is only implemented for React Native`? (I assume that is the case)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `ProtoPageStreamingTransformer` class, since that's where the paging logic is.",
+      "reference": "Do we not need this in Java because Java *is* handling map responses?"
+    },
+    {
+      "prediction": "I think this should be `False` by default, and `True` when you want to receive all messages.",
+      "reference": "Maybe \"For configuring which scheduler messages can be received.\"?"
+    },
+    {
+      "prediction": "I think this should be `getEntityManager` instead of `getObjectManager`",
+      "reference": "It will stop working with orm, so this is not a good fix, Try using some method in a common interface"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `HttpHandlerDiagnosticListener` class, since that's where the `HttpHandler` is defined.",
+      "reference": "`s` is not necessarily the same as `sampler` - sampler could be null and this crazy lambda underneath falls back to something. So please use `s`"
+    },
+    {
+      "prediction": "Why do we need to set it to `null` here?",
+      "reference": "This code is a bit confusing. Should HttpParameterType set to the original one or it is always BODY?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `VerifyOutputs` function.",
+      "reference": "Why did this change?"
+    },
+    {
+      "prediction": "I think it would be better to move this to the `DescribeTable` function, since it's only used in that function.",
+      "reference": "nit: `status` collides with imported package named `status`"
+    },
+    {
+      "prediction": "I think this should be `l.accts.txLease` instead of `l.accts.IsWritingCatchpointFile`",
+      "reference": "\"memory pressure could be decreased\" --> \"to avoid memory pressure\""
+    },
+    {
+      "prediction": "I think this test should be moved to `test_store.py`",
+      "reference": "I think it would make sense to fetch the cache values after deleting the listens and making sure they are what we expect."
+    },
+    {
+      "prediction": "I think we can remove the `mapState` call here since it's not used anymore.",
+      "reference": "I'm thinking about covering this change with the test. Can you do that?"
+    },
+    {
+      "prediction": "I think this should be a `*net.UDPConn` instead of a `*net.UDPConn` pointer.",
+      "reference": "Not sure if it's a good idea to remove stop channel. If user cancels connection how will you stop pinger?"
+    },
+    {
+      "prediction": "I think this should be `sdkAuthConfig.User` and `sdkAuthConfig.Password` and `sdkAuthConfig.Email` and `sdkAuthConfig.ServerAddress` and not `authData.Username`, `authData.Password` and `authData.Email` and `authData.ServerAddress`",
+      "reference": "Could you please add a `TODO` here that we won't need `docker.AuthConfiguration` anymore when we migrate to SDK's pull image?"
+    },
+    {
+      "prediction": "I think this should be `BlobNames` and not `BlobNames`",
+      "reference": "Is there a scenario were we want to get an arbitrary set of blobs that don't share a common root?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it. I think we need to make sure that we don't generate the topology for any spec that doesn't match `type=regex`.",
+      "reference": "Can you leave a note about why we are skipping regex tests?"
+    },
+    {
+      "prediction": "I think this should be `if err != nil { return err }`",
+      "reference": "This was the only caller of CreateRepo, so I inlined it."
+    },
+    {
+      "prediction": "I think this should be `AuthPath string ` instead of `AuthPath string `json:\"authPath,omitempty\"`",
+      "reference": "As this is already a child of `auth`, I think we can shorten this to `path`"
+    },
+    {
+      "prediction": "I think this should be `elif position in (textInfos.POSITION_FIRSTVISIBLE, textInfos.POSITION_LASTVISIBLE, textInfos.POSITION_ALL):`",
+      "reference": "This call should use element, not 0 I think."
+    },
+    {
+      "prediction": "I think this should be `FetchUntilEmptyPageThreshold` to match the rest of the config.",
+      "reference": "This is missing the definition of `fetch_until_empty_page` - it needs to go in the config section above."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "I don't think this one needs to be changed since the only previous path is already an absolute one."
+    },
+    {
+      "prediction": "I think this should be `locationProvider` instead of `this.table.locationProvider()`.",
+      "reference": "Since we kept the original constructor we might not need this change now."
+    },
+    {
+      "prediction": "nit: remove the extra space after \"Refreshed\"",
+      "reference": "There is already a log in `BaseMetastoreTableOperations` for this. It has the location, but not the table name. Maybe just add table name to that one."
+    },
+    {
+      "prediction": "I think we should move this import to the top of the file.",
+      "reference": "This should be in the last group."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `try` block above, right?",
+      "reference": "I believe you should either remove this or rename the PR. You can't change the behavior of the tracer (even to fix a bug) in a PR named \"unit test improvements\""
+    },
+    {
+      "prediction": "I think this should be `cert-manager.k8s.io` instead of `certmanager.k8s.io`",
+      "reference": "I've gone with this group name, however I'm open to suggestions on alternatives!"
+    },
+    {
+      "prediction": "I think it would be better to have a `QuorumApi` and a `RpcApi` for the quorum api.",
+      "reference": "should this be GOQUORUM"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the validator, but I'm not sure if there is a better place to do it.",
+      "reference": "I think checkCoinbase is not skipping checking coinbase. Instead, true-> checking num(coinbase tx) = 1, false -> checking num(coinbase tx) = 0."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "same here, move it out"
+    },
+    {
+      "prediction": "I think this should be `#if defined(KOKKOS_ENABLE_LIBDL || defined(KOKKOS_TOOLS_INDEPENDENT_BUILD)`",
+      "reference": "Can you elaborate on why you need different behavior depending on Tools being built independently or not?"
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "because of the above if statement will requires both ex_save and guid_save to be true, this code will always return 'example answer'. This should be revised with `example_answer.present?` and `guidance.present?`"
+    },
+    {
+      "prediction": "I think we should use `context` from `golang.org/x/net/context` instead of `structs` from `github.com/stretchr/testify/assert`.",
+      "reference": "why not context from stdlib?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. It's not binary, it's a Thrift type.",
+      "reference": "Just saw this and wasn't able to hold myself from renaming :-)."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it would be better to move it to `src/amp-ad/amp-ad-network-adsense/amp-ad-network-adsense.js`",
+      "reference": "Can we keep the other assertion too? I think it's helpful to ensure that the uninstall function was called too"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this, but it's the only way I could come up with that didn't involve a ton of boilerplate.",
+      "reference": "`warn_with` prefixes the message with `WARNING:`, right? So this will put `WARNING:` twice. It would be good to verify all the warnings look good after this change, given how easy it is to make a simple mistake like this :(."
+    },
+    {
+      "prediction": "Why not just use `types.BlockTimer`?",
+      "reference": "FYI In the message queue I use the term \"stamp\" to refer to the time-like mark associated with each message. It's opaque to the queue/pool and should make no difference if the stamps and age limit were converted to seconds. So this could then become `Stamper` with `CurrentStamp()` method, no reference to \"blocks\" or height etc. The wrapper struct could be `StampedMessage. The interpretation as block height is deferred to the user/constructor that hooks things up, the \"business logic\". This is just an observation, take or ignore as you wish."
+    },
+    {
+      "prediction": "I think this should be `exists, code, body = is_bucket_available(bucket_name)[0]` instead of `exists, code, body = is_bucket_available(bucket_name)`",
+      "reference": "Can we remove the `TODO` statement here? (as this is actually fixed in this PR)"
+    },
+    {
+      "prediction": "I think this should be `private final DisplayPreference mDisplayPreference;`",
+      "reference": "m prefix no longer in code style"
+    },
+    {
+      "prediction": "I think we can remove this now.",
+      "reference": "need to filter the entities in buffer as well"
+    },
+    {
+      "prediction": "I think this should be `public static final String ID = \"de.danoeh.antennapod.core.glide.FastBlurTransformation\";`",
+      "reference": "Please add spaces before and after the equals sign"
+    },
+    {
+      "prediction": "Please remove this file.",
+      "reference": "This file could also be cleaned up/deleted. But lets do this separately after ensuring all the event logging has been ported to the current logmxevent_traf.cpp file completely."
+    },
+    {
+      "prediction": "I'm not sure if this is a good or a bad thing. I think it's a good thing, but I don't know if it's a good or a bad thing to have a newline at the end of the file. I think it's a good thing to have a newline at the end of the file, but I don't know if it's a good or a bad thing to have a newline at the end of the file. I also don't know if it's a good thing to have a newline at the end of the file. I think it's a good thing to have a newline at the end of the file, but I don't know if it's a good or a bad thing to have a newline at the end of the file. I also don't know if it's a good thing to have a newline at the end of the file. I think it's a good thing to have a newline at the end of the file, but I don't know if it's a good or a bad thing to have a newline at the end of the file. I also don't know if it's a good thing to have a newline at the end of the file. I think it's a good thing to have a newline at the end",
+      "reference": ":thinking:, the line number here is the line of the example (`example.location`), that's probably why I finally chose to display the example title because it's what can be found on this line, and saying that the error is from there is confusing. But presenting the actual failures is indeed better, so maybe we can get the failure actual line number from the exception backtrace and then output two distinct lines for failures like this one?"
+    },
+    {
+      "prediction": "I think this should be `chrome_options=None, service=create_no_window=False, keep_alive=True`",
+      "reference": "We shouldn't add this as a new `kwarg` here. This should all be done on the options class"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Should this be `OS_VOLUME_SUPPORT` to match the symbol key names?"
+    },
+    {
+      "prediction": "I think this should be `static flux_msg_handler_t *handlers[] = { ... };`",
+      "reference": "In C, file scope variables are already initialized to 0 so the \"= NULL\" is redundant. Not a big deal."
+    },
+    {
+      "prediction": "I think this should be `axer.commons.doc.getRootNode(node);`",
+      "reference": "on line 40, the `relevantNode` can change and therefore the `doc` might change too for the lookup on line 43. I think this code should be moved to where the `doc` is actually being used"
+    },
+    {
+      "prediction": "I think this should be `throw new ArgumentOutOfRangeException(nameof(milliseconds), milliseconds, string.Format(CultureInfo.InvariantCulture, \"milliseconds must be between {0} and {1}\", MinMilliseconds, MaxMilliseconds));`",
+      "reference": "nit: just to be similar to others, can you change to interpolation?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/apis/core/internalclientset/versioned/typed/authorization.k8s.io/v1beta1/internalclientset/versioned/typed/authorization.k8s.io/v1/internalversion`",
+      "reference": "Would this make more sense as an option on an Issuer resource (specifically on the CA issuer) rather than as a flag? It's not ideal that we have a flag that _looks_ like it could be 'global' but is actually not IMO. Also, if Istio is reading the CSR object, is it possible for it to read the CA from the CSR itself too and append the two instead? Not sure if we ever decided whether to use annotations for this purpose @JoshVanL?"
+    },
+    {
+      "prediction": "I think this should be `DATE_TIMEF` or `DATE_TIMEFORMAT` to be consistent with the rest of the codebase.",
+      "reference": "Because this acts as a coordination point between client and server, it should go in core.py instead of const.py (so eventually the server could use it to guarantee that it delivers dates in the expected format)."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "expected statement, found 'else' (and 1 more errors)"
+    },
+    {
+      "prediction": "I think this should be `return MakeRandomID(public).id[0]`",
+      "reference": "IMO this doesn't need to be in the `kbfs` repo. And I don't like setting the byte. We can always compare with the randomly generated `TlfID`."
+    },
+    {
+      "prediction": "I think we should rename this to something like `NetworkControllerModuleName` to make it clear that it's not just a device controller but also a network controller.",
+      "reference": "What's the difference from ModuleName in `pkg/apis/meta/v1alpha1/types.go`?"
+    },
+    {
+      "prediction": "I don't think we need to use `System.Security.Cryptography` at all. We can just use `System.IO` and `System.Reflection.Metadata` and `System.Reflection.PortableExecutable`.",
+      "reference": "We shouldn't be changing this task as we are trying to break our dependency on it and switch to using the shipped version."
+    },
+    {
+      "prediction": "I think it would be better to make this conditional on cci->num_simd_skip > 0",
+      "reference": "Shouldn't we have a low bar for generating out-of-line context switch, i.e., if we need save more than n (3?) simd or m (4) gprs we should go out-of-line? And it should be || instead &&. It seems the bar is still very high after this change."
+    },
+    {
+      "prediction": "I think we should move this to `pkg/metric/proxy.go`",
+      "reference": "Should the API have a dependency on the SDK?"
+    },
+    {
+      "prediction": "I think this test should be skipped because it's not relevant to the change stream functionality.",
+      "reference": "Is this a sometimes leaky test?"
+    },
+    {
+      "prediction": "I think this should be `for _, f := range ftv` instead of `for ftv.contains(f)`.",
+      "reference": "It should be possible to create a test case that enter the loop. That would be a good enough test case for me. Have a look at the extern type inference test cases that already exist. Since you can just pick the type variables the extern type use, you should be able to create a conflict that requires this loop to fix."
+    },
+    {
+      "prediction": "I think this should be `if not locale.IsAvailable(wxLang.Language):` instead of `if not locale.IsAvailable(wxLang.Language):`",
+      "reference": "It might make sense to log this."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "Line is too long. [84/80]"
+    },
+    {
+      "prediction": "I don't think this is a good name for a property. It's not a property of an option, it's a property of the ChromeOptions object itself.",
+      "reference": "nice touch since Google likes to change things every now and then."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this method. I think it should be something like `load_branding_config` or `get_branding_config` since it's not just fetching a single value but the whole thing.",
+      "reference": "Thank you this is so much nicer than the old `Rails.configuration.branding['blah']['blah']['blah']` :)"
+    },
+    {
+      "prediction": "I think this should be `SERVICE` instead of `backend_service` to be consistent with other services.",
+      "reference": "nit: Would you mind alpha-sorting this?"
+    },
+    {
+      "prediction": "I think this should be `ManifestContent.DELETE` instead of `DATA`",
+      "reference": "For my understanding, the DATA manifest & DELETE manifest could share the same read / write path so I think we could use the common reader+writer. Is there any other reason that we need to make them separate paths ?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "that change should be sufficient for avoiding the override of contact_us_url helper. The one defined at the .yml it is within a hash so no problem should arise."
+    },
+    {
+      "prediction": "Shouldn't this be \"hinzuf\u00fcgen\" instead of \"Element\"?",
+      "reference": "We actually had a discussion about this exactly in #3470, some languages seem to go for one, some for the other"
+    },
+    {
+      "prediction": "I think this should be renamed to `grpcStreamingDescriptors()`.",
+      "reference": "Is PHP the first dynamic MVVM language with grpc streaming support?"
+    },
+    {
+      "prediction": "I think this should be deleted only if tmpStream is not nullptr.",
+      "reference": "the `if` isn't necessary here; it's safe to call `delete` on `nullptr`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should go in `DataWriterImpl` since it is only used by the `DataWriterImpl::remove_all_changes` function.",
+      "reference": "Change if the method's name finally changes to `clear_history` as proposed elsewhere"
+    },
+    {
+      "prediction": "I think this should be `gui.getDocFilePath(\"userGuide.html\")` instead of `documentationUtils.getDocFilePath(\"userGuide.html\")`",
+      "reference": "can this one also be moved to the top of the file?"
+    },
+    {
+      "prediction": "I think we should rename this to `Resources` instead of `GroupMembers` to be consistent with the rest of the codebase.",
+      "reference": "So if I understand this correctly the externalEntities should be computed and stored in GroupMember, while current code usage of GroupMemberPod will be migrated to GroupMember later?"
+    },
+    {
+      "prediction": "I'm not sure if we want to keep these in the public namespace, or if they should live in `System.Diagnostics.Tracing` or `System.Diagnostics.Tracing.Internal`",
+      "reference": "Don't we have tests to change after this breaking change? probably we're not using hard coded strings in our tests when we create a `Span`?"
+    },
+    {
+      "prediction": "I think this should be `os.environ.get(\"GOOGLE_APPLICATION_CREDENTIALS\")` instead of `os.environ['GOOGLE_APPLICATION_CREDENTIALS']`",
+      "reference": "will this fail if the env variable doesn't exist? Is this the preferred behaviour?"
+    },
+    {
+      "prediction": "I don't think this is right. The tie-breaker is tied to the pooler, not the pooler to the tie-breaker. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to the pooler, not the pooler to the pooler. The tie-breaker is tied to",
+      "reference": "bad line wrap"
+    },
+    {
+      "prediction": "I think this should be `boost::system::error_code` instead of `boost::system::error_code const`.",
+      "reference": "What does the `8` here represent? Can you give it a name?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in `tests/unit_tests/test_materials.py`",
+      "reference": "`MPRester(os.environ[\"MP_API_KEY\"])` can be left simply as `MPRester()` and it will pick up the API key from the environment. What is the purpose of this file otherwise? Is it better suited for `dev_scripts`, or a test?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Looks like there may be more changes to this file than intended?"
+    },
+    {
+      "prediction": "I don't think we need to continue here.",
+      "reference": "when message type is not expected, contine to next loop instead of return to finish infinite loop"
+    },
+    {
+      "prediction": "I think this file should be renamed to `testutil.go`",
+      "reference": "This change was not really intended but made by the linter of VS Code. And looking at other packages this looks like a best practise to place interal packages on the top and gh imports afterwards."
+    },
+    {
+      "prediction": "I'm not sure if we want to do this, but I think it's fine to do it here.",
+      "reference": "instanceID should stay the same for a AWSMachine?"
+    },
+    {
+      "prediction": "I think this should be set to `True` by default.",
+      "reference": "Since this is a dict don't you want `False` or do we clean that up elsewhere?"
+    },
+    {
+      "prediction": "I don't think this is needed. The host is set in `__init__` of the scheduler.",
+      "reference": "what's the reason you had to add the `host` argument to all `get_work` calls?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be moved to `AbstractCommand` since it's not specific to a dataverse.",
+      "reference": "This will never be a dataset so we can simplify the code here"
+    },
+    {
+      "prediction": "I don't think we need to pass the skipZeroBlockRewards here.",
+      "reference": "This should be moved into the PrivacyBlockProcessor instead of adding the constructor here."
+    },
+    {
+      "prediction": "I think we should rename `insightfilestore` to `insightstore`.",
+      "reference": "nit: Remove \"insightfilestore\". I think \"insightstore\" is good enough and as I see we are using that name at other places too."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `Rspec::Core::Formatters::BisectProgressFormatter`.",
+      "reference": "It looks like you've removed `bisect_round_finished` but kept `bisect_round_started`. Seems a little odd that they aren't paired. Not sure if there's anything to do about that though..."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could come up with.",
+      "reference": "I'm not sure how to fix the lint and mypy warnings on this line. I could just ignore them, but I feel like the linter should know this exists."
+    },
+    {
+      "prediction": "I think this should be `const SKIP_SESSION = Symbol('SKIP_SESSION')`",
+      "reference": "why remove `Symbol`?"
+    },
+    {
+      "prediction": "I think this should be `\\u2000-\\u206F\\ud800\\ud801\\ud802\\ud803\\ud804\\ud805\\ud806\\ud807\\ud808\\ud809\\ud80a\\ud80b\\ud80c\\ud80d\\ud80e\\ud80f\\ud810\\ud811\\ud812\\ud813\\ud814\\ud815\\ud816\\ud817\\ud818\\ud819\\ud81a\\ud81b\\ud81c\\ud81d\\ud81e\\ud81f\\ud820\\ud821\\ud822\\ud823\\ud824\\ud825\\ud826\\ud827\\ud828\\ud829\\ud82a\\ud82b\\ud82c\\ud82d\\ud82e\\ud82",
+      "reference": "Why are currency signs considered punctuation? They technically fall under the BMP range for Currency Symbols `\\u20A0-\\u20CF` (covered in `getUnicodeNonBmpRegExp`, which is itself a misnomer)"
+    },
+    {
+      "prediction": "I think it would be better to put the maintainer in the `setup.py` and not in the `__init__.py` file.",
+      "reference": "You have forgotten a coma `,`"
+    },
+    {
+      "prediction": "`img_meta['filename']` is the filename of the original image. `img_meta['filename']` is not the same as `img_meta['filename'].replace(os.sep, '_')`, which is the filename in the output directory.",
+      "reference": "For different datasets, we may want to save the images differently. Sometimes the datasets are categorized into several subfolders, and we want to keep it when saving visualization results, and sometimes we just want to save all images in `out_dir`. An argument may be added to control the behavior."
+    },
+    {
+      "prediction": "I think this should be `return c1 == c2 || (cmp == 0 && bb.get(ia) == '\\0');`",
+      "reference": "if `other` could have a guaranteed `0` byte in it, this whole `if` could be removed, and above you could just add `|| c2 == '\\0'` to get the same effect?"
+    },
+    {
+      "prediction": "I think this should be `caseInsensitive` instead of `caseSensitive`",
+      "reference": "nit: can this be a primitive value?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `browse.go` since it's only used by that function.",
+      "reference": "This maybe not what we want. If `rewrite` occurred before, `r.URL.Path` contains the result and we should use it. Otherwise `rewrite` middleware will disfunction when combining with `browser`."
+    },
+    {
+      "prediction": "I think this should be `self.reporters + [self.aggregator]` instead of `self.reporters += [self.aggregator]`",
+      "reference": "Why change this?"
+    },
+    {
+      "prediction": "Why not use the `Symfony\\Component\\HttpKernel\\Exception\\NotLogFakeHttpExceptionsExceptionListener`?",
+      "reference": "_nitpick_ Isn't it now actually NotLogFakeHttpExceptions**Error**Listener ?"
+    },
+    {
+      "prediction": "I think this should be `JsonSchema.CreateStringEnum(resourceType, apiVersion);`",
+      "reference": "Why not just look for the \"type\" property instead of creating a ResourceType property?"
+    },
+    {
+      "prediction": "Why do we need to set the chain id here?",
+      "reference": "should use `bc.config.Blockchain.ChainID`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `ElasticsearchStorage` class.",
+      "reference": "this doesn't need to know about NativeClient"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Violation` class, under the `rule_name` field.",
+      "reference": "I know we discussed if we should add in the resource_name to the hash, and I actually don't think it's required. The violation hash was recently updated (not released) to include the violated rule, so when the next release goes out, user's findings will get updated. The resource name doesn't hurt to have here, but not required."
+    },
+    {
+      "prediction": "I think this should be `surface->texture->height_from_buffer = surface->texture->height * scale;` and `surface->texture->width_from_buffer = surface->texture->width * scale;` instead of `surface->texture->height_from_buffer = surface->texture->height / scale;` and `surface->texture->width_from_buffer = surface->texture->width / scale;`",
+      "reference": "The variable name is also misleading imo. The width from the buffer comes from `wl_shm_buffer_get_width` (or a similar EGL thing), which is what gets encoded into `surface->texture->width`."
+    },
+    {
+      "prediction": "I don't think we need to add these to the `tags` here. The `time` tag is already in the `tags` object, and the `artist_name` and `recording_msid` tags are already in the `tags` object. The `time` and `artist_name` tags should be added to the `tags` object. The `recording_msid` and `title` tags should be added to the `tags` object. The `recording_msid` and `title` tags should be added to the `tags` object. The `time` and `title` tags should be added to the `tags` object. The `time` and `title` tags should be added to the `tags` object. The `time` and `title` tags should be added to the `tags` object. The `time` and `title` tags should be added to the `tags` object. The `time` and `title` tags should be added to the `tags` object. The `time` and `title` tags should be added to the `tags` object. The `time` and `title` tags should be added to the `tags` object. The `time` and `",
+      "reference": "Why are you using these tags to carry these msids?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `flb_sds_len()` returns the number of characters in the string, not the number of bytes. This means that `strncpy()` won't copy the right number of bytes.",
+      "reference": "flb_strdup() should do the work here. note: use Fluent Bit memory wrappers"
+    },
+    {
+      "prediction": "I think this should be `go list -f '{{ .Dir }}' -m github.com/gogo/protobuf)/retrieval.proto`",
+      "reference": "This line should be separated from the package declaration by one line as it is not the package description."
+    },
+    {
+      "prediction": "why do we need to start a new sf here?",
+      "reference": "shadow: declaration of \"err\" shadows declaration at line 410 (from `govet`)"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the datasize package.",
+      "reference": "Maybe next time better to make separate PR?"
+    },
+    {
+      "prediction": "`det_bboxes` is a `List[List[Dict[str, Tensor]]]`, each of which is a `List[Dict[str, Tensor]]`, right?",
+      "reference": "line 316 also handles this case, need to simplify the logic."
+    },
+    {
+      "prediction": "I think we can remove the `bucketName == \"\"` check now.",
+      "reference": "We used to have a simple check for bucket and object names, we then decided to remove it and rely on the backend service to check for the name. Is there any reason to add an empty name check now?"
+    },
+    {
+      "prediction": "I think this is not the right way to do it. The right way is to calculate the next timeout based on the current time and the max_blocking_time_point. If the current time is greater than the max_blocking_time_point, the next timeout should be the max_blocking_time_point. Otherwise, the next timeout should be the current time plus the max_blocking_time_point.",
+      "reference": "Line too long"
+    },
+    {
+      "prediction": "I think this should be `CONST_ME_DODGE` instead of `CONST_ME_FATAL`",
+      "reference": "This isn't aligned though."
+    },
+    {
+      "prediction": "I don't think this is the right place for this.",
+      "reference": "it is mean -> it means PAAHead does not support test-time augmentation."
+    },
+    {
+      "prediction": "I don't think we need to import `plugins` here, since it's not used in this file.",
+      "reference": "@lnielsen-cern thanks. I missed it :("
+    },
+    {
+      "prediction": "I think this should be `['analytics.pagePath']` instead of `['[]']`",
+      "reference": "@tofumatt, has it been changed intentionally? It doesn't seem to be required in IB, plus we haven't had `content-drilldown` before... If it has been changed intentionally, then should we update the `AnalyticsDashboardWidget` component to be use `content-drilldown` instead of `content-pages` as well?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "nit: don't think an extra line is needed here; as retrying is an installed library and should be grouped with the rest of the installed library."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `prometheus/prometheus` package.",
+      "reference": "This should be a bool"
+    },
+    {
+      "prediction": "I think this should be `llvm::is_same<swift::WeakStorageType, swift::WeakStorageType>` instead of `llvm::isa<swift::WeakStorageType>(swift_type)`",
+      "reference": "`if (llvm::dyn_cast_or_null<swift::WeakStorageType>(GetSwiftType(compiler_type).getPointer())` ?"
+    },
+    {
+      "prediction": "I think we should rename `canaryManifests` to `canaryConfigMaps`.",
+      "reference": "The `generateVariantWorkloadManifests` function below relies on the config manifests specified in Git, but it was passed with the newly created canary's manifests where their name was updated."
+    },
+    {
+      "prediction": "I think this should be `@throws \\UnexpectedValueException`",
+      "reference": "are this docs neccesary ?"
+    },
+    {
+      "prediction": "I think this should be `private final String browserStartCommand = null;` and `private final String browserURL = null;` instead of `private String browserStartCommand` and `private String browserURL`",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "Why we need to set `clear_loss_weight` here?",
+      "reference": "This change suppose to resolve issue #2968. `loss_weight` should be of length 2, the second entry is ignored."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/unversioned/unversioned.go`",
+      "reference": "Is there an easy way to check for conflict suffix to avoid false positives here?"
+    },
+    {
+      "prediction": "`@context.get_validated_task` returns `nil` if `opts['task']` is `nil` or `undefined`.",
+      "reference": "`apply_prep.run_task` also seems like it can just use `@context.run_task`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `pageLoaded` function.",
+      "reference": "do you have any idea on how to fix it?"
+    },
+    {
+      "prediction": "I think this test is a bit too specific to this test class. Can we move it to a separate file?",
+      "reference": "You don't need to fix this (you've done so many iterations). But for next time, you can make this into a docstring so it'll have a nicer descriptive name when the tests are running."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `ENIAttachment` struct itself, not in the `ENIBaseAttachment` struct.",
+      "reference": "When does 'expiresAt' change? Do you need this?"
+    },
+    {
+      "prediction": "This should be `const Outfit*` instead of `const Outfit&`",
+      "reference": "There are enums for genders + `PLAYERSEX_LAST`."
+    },
+    {
+      "prediction": "I think this should be `uint64(dataCost) + uint64(onesCost)` instead of `zeroesCost + (96 + ones) * 16`",
+      "reference": "Should this use the constants as well? Or intentionally covering the fact that the constants may change"
+    },
+    {
+      "prediction": "I think this should be `if self._size >= 0 and self._size < len(self.best)`",
+      "reference": "The case of a negative size (= keep all elements) was not handled in this function."
+    },
+    {
+      "prediction": "I think this should be `system.infiniteTimeout` instead of `system.enableInfiniteTimeout`",
+      "reference": "I think we should keep this dynamic config knob for operations purpose."
+    },
+    {
+      "prediction": "I think this should be `self.assertEqual(ti.offsets, (4, 6))`",
+      "reference": "Was this intentional? Perhaps this could be replaced with another way to check the offsets?"
+    },
+    {
+      "prediction": "I'm not sure what this is, but it seems like it should be in the manifest.",
+      "reference": "the buckets are created via the stackset, customers don't need to provide them"
+    },
+    {
+      "prediction": "I think this should be `qt: blacklight_config.qt` and not `qt: blacklight_config.solr_path`",
+      "reference": "This doesn't have the same behavior as before. Maybe you want `reverse_merge` (and it either just works or we implement it on `Blacklight::Solr::Request`)"
+    },
+    {
+      "prediction": "I think this should be `export PATH=/usr/local/ghc/#{ghc_version}/bin:$PATH`, not `export PATH=/usr/local/ghc/#{ghc_version}/bin/:$PATH`",
+      "reference": "This should be ` /$(ghc_find #{config[:ghc]})/ ` instead of ` /#{ghc_version}/ `, I think."
+    },
+    {
+      "prediction": "I think this should be `dots-horiz` instead of `dots-horiz`",
+      "reference": "We should deprecate this at some point."
+    },
+    {
+      "prediction": "I think it would be better to add `use_sigmoid=False` here as well.",
+      "reference": "Move these two arguments before `loss_cls`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in `GenerateSelfSignedX509CA` and then call `GenerateServerX509UsingCA` with the CA as an argument.",
+      "reference": "[Super NIT]: i := serialNumber if i == 0 { i = mathrand.Int63n(...) }"
+    },
+    {
+      "prediction": "I think this should be `T` instead of `tool`.",
+      "reference": "-f short form here collided with --force. :)"
+    },
+    {
+      "prediction": "I'm not sure why this is removed, but I think it should be added back in.",
+      "reference": "I think it is better to describe in the release note that `rerender()` was removed."
+    },
+    {
+      "prediction": "I think this should be `if __name__ == \"__main__\":`",
+      "reference": "We could also test this easily with `runpy.run_module`, what do you think?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it. I think we should check if `val` is an instance of `interface{}` and then call `v.Set(reflect.ValueOf(val))`. If it is not an instance of `interface{}`, then we should just call `v.Set(reflect.ValueOf(val))`.",
+      "reference": "Is it OK to fall through if it's not assignable? I was kind of surprised when the tests passed. Maybe one of the options below ends up working? For the new tests, this panics because `**not assignable uuid.UUID to []uint8**`. `uuid.UUID` is an array, not a slice."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `BlockAPI` struct.",
+      "reference": "I know this is after the fact but imo blockAPI should be blockMiningAPI."
+    },
+    {
+      "prediction": "I think it would be better to use `preg_quote` instead of `self::SYSTEM_ATTRIBUTE_PREFIX`.",
+      "reference": "it's just a pattern as it was. SystemAttributeCode should have it's own definition and validate by it"
+    },
+    {
+      "prediction": "`cpp2` should be `cpp2`, and `LookUpIndexRequest` should be `cpp2::LookUpIndexRequest`.",
+      "reference": "Similar to the code block when we use prefix, maybe we can unify them to base?"
+    },
+    {
+      "prediction": "I think we can remove the `ValidatePodchaosSchedulerError` from the `ValidateScheduler` error list.",
+      "reference": "I think `Duration` is useful for this action and we shouldn't ignore it."
+    },
+    {
+      "prediction": "I don't think we need to return the length of `p` here, since it's the same as the length of `s.dataForWriting`.",
+      "reference": "Everything below here can be replaced with `return s.err, len(p) - len(s.dataForWriting)`."
+    },
+    {
+      "prediction": "I think this should be `KOKKOS_ENABLE_IMPL_DESUL_ATOMICS` instead of `KOKKOS_ENABLE_IMPL_DESUL_ATOMICS`",
+      "reference": "Again, we can also guard `g_host_hip_lock_arrays.atomic`, can't we?"
+    },
+    {
+      "prediction": "Why do we need to set `$this->commandBus->dispatch($ImportProductAttributesValueCommand, true);` here?",
+      "reference": "Variable name should start from lowercase latter"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this. I think it would be better to have a separate test for this.",
+      "reference": "Why has this timeout doubled?"
+    },
+    {
+      "prediction": "I think this should be `0` by default.",
+      "reference": "Should IptablesPostWriteCheckIntervalSecs be set back to its previous smaller value, if use of the iptables lock is disabled?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. This should only happen when the tab is focused, not when the tab is not focused.",
+      "reference": "You're missing a `return` here, but I'll add it :smile:"
+    },
+    {
+      "prediction": "I think this should be `wlr_gles2_get_texture(wlr_texture)`",
+      "reference": "Can you try adding back this assertion? I'd like to assert that the texture has been created in the same context as the renderer."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could think of to make it work.",
+      "reference": "This won't work with Python 3 because it will already be a string. In that case we need to use `decode_header` from utils."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `Mongoid::Fields::Validators::Macro`",
+      "reference": "Why is the respond_to check here?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "@orklah are you sure it's the same as `$chars[$i]`? It looks like `$i` is changed above, after `$char` assignment."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "I have a question? How do I construct a indexManager for testcase?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. It seems like it should be in the `xfpga_fpgaOpen` function, but I'm not sure what the proper place is.",
+      "reference": "These checks aren't needed. See pluginmgr.c : opae_plugin_mgr_load_dflt_plugins( )."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to handle the empty string and whitespaces separately.",
+      "reference": "Please take a care about prefixing."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `EvmAccountUpdater`.",
+      "reference": "this line doesn't match what the rest says"
+    },
+    {
+      "prediction": "I think this should be `+build !linux android` instead of `+build windows linux,android`",
+      "reference": "Since it's not working for linux, maybe leave the !linux in there as well?"
+    },
+    {
+      "prediction": "I think this should be `continue` instead of `continue with error`.",
+      "reference": "Should we remove the key-value pair entirely? Same for workflows in general. We have a \"sourceExists\" method that checks if a key is there if I remember correctly."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `ContentSyncManager` where we have access to the `ModificationDate` field.",
+      "reference": "Unrelated bug fix"
+    },
+    {
+      "prediction": "This should be `globalize` not `globalize.translate`",
+      "reference": "We probably want to let the translation library handle the string replacement for these as well."
+    },
+    {
+      "prediction": "I think it would be better to have a single `--list-db` flag that can be used in conjunction with `--list-memories` and `--list-members`.",
+      "reference": "should combine these 2 list (list_gossip and list_db) into just one list command, and show the discrepancy between the 2 if any."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in the `ActionPolicy` class, since it's related to actions on the proposal (approve/reject).",
+      "reference": "What do you think about prefixing the Policy method names with `can_`?"
+    },
+    {
+      "prediction": "nit: I think it would be better to use `#define AWS_CREDENTIAL_RESPONSE_EXPIRATION 3600 * 60 * 24 * 30` instead of `3600 * 60 * 24 * 30`",
+      "reference": "Based on the docs, credential process can return a key `SessionToken`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if there is a better way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if there is a better way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this",
+      "reference": "I think this if should be removed (see issue)"
+    },
+    {
+      "prediction": "I think this is the only place where we're using the ipset package.",
+      "reference": "I would avoid importing `felix/ipsets` package because this package is the windows equivalent and should be at the same level of `felix/ipsets`. We could add linux specific dependencies into `felix/ipsets` later and it will break Windows build."
+    },
+    {
+      "prediction": "This should be `Optional[Tuple[str, ...]]`",
+      "reference": "nitpick: Please move this down to the other imports, as it's a Python stdlib import."
+    },
+    {
+      "prediction": "I think it would be better to have a separate package for blobs. It's not only for GCP, but also for Azure, etc. I think it would be better to have a `blob` package for GCP, Azure, AWS, etc.",
+      "reference": "\"for getting started guides\" reads a bit weird, like it's missing punctuation. Maybe just \"for a tutorial\"?"
+    },
+    {
+      "prediction": "`@param bool $enabled is this tab enabled?` -> `@param bool $enabled is this tab enabled?`",
+      "reference": "This looks like a dead parameter -- maybe a copy and paste error. If it's truly unused, maybe you can open a separate PR to simply delete it, and then it's one less detail to worry about here."
+    },
+    {
+      "prediction": "I think this should be `bool, int64, uint64, float64, string`",
+      "reference": "Nit: Should we add support for `int` and `uint` in the `values.New()` function? That would allow us to to handle all of these types in one case. Maybe there's a good reason why we don't do that already, but I'm not sure what it is."
+    },
+    {
+      "prediction": "I think it would be better to have a `RuleViolationCache` and store all the violations in the cache. This way we can easily check if the cache is full or if there are too many violations in the cache.",
+      "reference": "if you are not dealing with cache now, please revert these changes. On their own they make little sense"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Wearing my extra-douche-bag hat I would say not spaces on a blank line. Maybe at some point we could have some automated process remove these things. For now I would not bother changing it."
+    },
+    {
+      "prediction": "Shouldn't this class be called `ShowRemainTimeUpdateListenerEvent`?",
+      "reference": "I think it would be better to use an `ItemUpdatedEvent` like for the \"prefer streaming\" preference. We already have a ton of events that need to be handled in all list fragments that just do the same everywhere. I think we could even remove some of the existing events in the future."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [131/120]",
+      "reference": "I think we have to manually write a down for this migration."
+    },
+    {
+      "prediction": "I think you can remove the `index_map = [(SPARK_INDEX_NAME_FORMAT(0), None)]` line here.",
+      "reference": "I don't think we still need this check, but I'd leave it as an assertion for now."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Since this may be the case, wondering if we should call the param `updated_at_i` or something so that we don't run into a problem distinguishing them down the road?"
+    },
+    {
+      "prediction": "I think this should be `Path.Combine(workingDir, appExe, appArgs)`",
+      "reference": "wondering why we need to join the paths here; seems evt.commandLine only takes whatever appExe is"
+    },
+    {
+      "prediction": "I don't think this is the best way to fix the issue. I think we need to find a better way to fix the issue.",
+      "reference": "1.not format code 2.failed but still continue? 3.\"UTF-8\" can changed to java.nio.charset.StandardCharsets.UTF_8.name()"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `src/python/pants/contrib/checks/brain.py`.",
+      "reference": "I'm wondering if we should add this to a `utils` module in `astroid`. Porting the util to `pylint` doesn't make sense as we would need to duplicate the globals that are being used in the function, but importing from `brain` also feels weird. @Pierre-Sassoulas Do you have an opinion?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `ReceiveAsync` method, since that's where the `ReceiveAsync` is happening.",
+      "reference": "This seems different than what we do for corrupted request headers. I would like to determine the correct behavior and consolidate this logic."
+    },
+    {
+      "prediction": "I think this should be `backend.get('resourceGroup')` instead of `backend.get('group')`",
+      "reference": "Please make the change in the CAI inventory, rather than changing the scanner."
+    },
+    {
+      "prediction": "Why is this needed? I don't see any usage of this class.",
+      "reference": "adding this `abstract` is a breaking API change. Even though I agree it makes perfect sense to have it, we can't break this API until PMD 7.0.0. We should revert this particular change. We could make a reminder ticket like we did with #463 for 6.0.0"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `spec/models/purchaseable_plan_spec.rb`",
+      "reference": "This is following the existing convention, but I'm not sure the convention is a good one. I think it would be better to have one shared example group: - It means we don't need to add new example groups twice every time. - If the shared example group looks large, that makes it more obvious that our team models have too much shared behavior. I don't want to do anything about this right now (I'm pretty sure I can wipe out this shared example group when I change team plans), but it's good to keep in mind."
+    },
+    {
+      "prediction": "I think this should be `|| nano::transport::is_ipv4_or_v4_mapped_address (new_connection->remote.address ())` instead of `|| nano::transport::is_ipv6_address (new_connection->remote.address ())`",
+      "reference": "I do not think we need to do a source code change to handle this. We could set the subnetwork to default to /32 (/128 for ipv6 ipv4-mapped)"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to go about it. I think we should be able to remove a cookie from a remote repository without having to know the name of the cookie, and without having to know the name of the remote repository.",
+      "reference": "It looks good overall! The only issue I have is that the description string is now duplicated twice right? Once in the struct and once in the parameter string? Maybe let's pass the struct to the command so that `ostree_option_context_parse` can set it as the summary? Similar to what we do in rpm-ostree."
+    },
+    {
+      "prediction": "I think this should be `ChromeDriverCommand` instead of `ChromeDriverCommandExecutor`.",
+      "reference": "This states very clearly that this is a derivative of Chrome and not Chromium. Do we need to extract an abstract `ChromiumCommandExecutor` and have both Edge and Chrome derive from that?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to check if `type0.getType() == java.lang.Enum.class` and then check if `type0.getType().isEnum() || type0.getType() == java.lang.Enum.class`",
+      "reference": "it seems to me, the issue lies on `isEnum()` itself, returning false for something that is an enum. I'd rather change it there than here."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It should be in `qute_scheme_handler.py`",
+      "reference": "Please use `{}` instead of `dict()`"
+    },
+    {
+      "prediction": "`useRef` should be used instead of `useIn` here.",
+      "reference": "I have concerns about using this in more places before #3278 - I'll take a look at that again shortly."
+    },
+    {
+      "prediction": "I think you can remove the `log`s from the `testLog` and just test the `Receipt` here.",
+      "reference": "new field needs to be asserted"
+    },
+    {
+      "prediction": "I think this should be `self->lastLogIdSent_ = std::max(self->lastLogIdSent_, self->logIdToSend_);` instead of `self->lastLogIdSent_ = self->logIdToSend_;`",
+      "reference": "when send log failed, why update the last sent log id?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. It seems like it should be in the `Appendable` interface, but I can't find where it is used.",
+      "reference": "what was the problem with the previous order? `prepend` comes conceptually before `append`, i.e. `insert before` should be before `insert after`, I think"
+    },
+    {
+      "prediction": "I think this should be a `map[string]interface{}`",
+      "reference": "Why is this necessary? Could you provide some context?"
+    },
+    {
+      "prediction": "I think you can remove the `type='text/css'` here.",
+      "reference": "Is there a recommended quote style for attributes? I see single and double here, double further down."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it seems like it should be.",
+      "reference": "Please use an atomic bool."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think we should move this to `WebSpanExtensions` or `SpanExtensions`",
+      "reference": "Consider naming this class `SpanExtensions` to follow C# conventions."
+    },
+    {
+      "prediction": "Why is this removed? I think it's still useful to know what the app domain is.",
+      "reference": "FYI @lucaspimentel and @colin-higgins since you have already approved, I wanted to point out this required change. By fixing `FrameworkDescription` to use the correct logger, it created a cycle between these two static constructors, so I'm removing this log line and delaying it to when it's actually constructed later in the Tracer. Let me know if you find that acceptable or have any suggestions."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Ah, so we are going to need to update node/host files for this to work?"
+    },
+    {
+      "prediction": "This should be `LOG.debug`",
+      "reference": "This code is new and not tested well. While we may drop the log level before we ship 4.0, right now this is extremely helpful to users."
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "used named path (`proposals_path`) instead?"
+    },
+    {
+      "prediction": "This should be `AND item_view.suppressed = false;`",
+      "reference": "Am I misreading something, or is there a mismatched parenthesis here? Please let me know whether or not this is cause for concern -- just wanted to be totally sure before merging, since I can't test this from here. Thanks!"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this validation. I think it should be done in the `response` object, not in the `transport` object. The `response` object contains the headers and the body, and the `transport` object contains the headers and the body, but the `response` object is not the same as the `transport` object. The `response` object contains the headers and the body, and the `transport` object contains the headers and the body, but the `response` object is not the same as the `transport` object. The `response` object contains the headers and the body, and the `transport` object contains the headers and the body, but the `response` object is not the same as the `transport` object. The `response` object contains the headers and the body, and the `transport` object contains the headers and the body, but the `response` object is not the same as the `transport` object. The `response` object contains the headers and the body, and the `transport` object contains the headers and the body, but the `response` object is not the same as the `transport` object. The `response` object contains the headers and the body",
+      "reference": "Might we want a hook to allow emitting metrics or logs in the case of a permitted empty service header response (or similar UpdateSpanWithErr on empty service header in response if strict enforcement is desired by the caller)?"
+    },
+    {
+      "prediction": "I think this should be `self.kwargs.setdefault(\"protocol\", \"https\")`",
+      "reference": "if using dict.get, this could just be `kwargs.get('protocol', 'https')`"
+    },
+    {
+      "prediction": "I don't think we need this method. We can just use `getGrpcStreamingApiMethodExampleName(Interface interfaze, Method method)`",
+      "reference": "Add docs to the new methods here"
+    },
+    {
+      "prediction": "I think this should be `public bool Enabled { get; set; }` instead of `public bool Enabled => true;`",
+      "reference": "what does it mean enabled?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. It would be better to do it in the `SmartStore` class.",
+      "reference": "Getting an error when calling with db.execSQL (in sqlcipher 4.3.0, pragma returns ok)."
+    },
+    {
+      "prediction": "Shouldn't this be `ctx.User.DisplayName()`?",
+      "reference": "Maybe we could remove `\" \" + ctx.Tr(\"dashboard\")` completely?"
+    },
+    {
+      "prediction": "I think you can remove the `# Changes the object's server-side encryption state during the copy operation.`",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `jsonHandler2` struct, since that's where the `Handle` method is defined.",
+      "reference": "Let's add some assertions here to ensure we don't panic. Length of results. Conditionally cast second result to error."
+    },
+    {
+      "prediction": "I think this should be 30 seconds, not 2 minutes.",
+      "reference": "Where did we take this default from?"
+    },
+    {
+      "prediction": "I think this should be `options.compareStartDate` and `options.compareEndDate` instead of `options.compareDateRanges` and `options.multiDateRange`",
+      "reference": "Let's move these down to be after all required arguments rather than in between."
+    },
+    {
+      "prediction": "I think this should be `if (_sharedStringMap == null) { _sharedStringMap = new Dictionary<string, string>(); }` instead of `if (_sharedStringMap != null) { _sharedStringMap = new Dictionary<string, string>(); }`",
+      "reference": "Should we just null the Map and let the GC handle the memory? Clearing just removes items, but not capacity, so this would leave some memory on the table."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the router.",
+      "reference": "did you mean to delete this line?"
+    },
+    {
+      "prediction": "I think it would be better to use `System.out.println` instead of `System.out.print` here.",
+      "reference": "Should we use logger instead?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix, but it's the only way I could get Travis to pass.",
+      "reference": "I think it would be preferred to do `expect(helper).to be_analytics`"
+    },
+    {
+      "prediction": "I don't think this is the right fix. We need to keep support for `pd.Index` as well as `str` and `tuple` keys.",
+      "reference": "Actually, I think it's okay to just change to `if isinstance(key, (str)) or is_list_like(key):` and `key = list(key) if is_list_like(key) else key` for simplicity for now."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "If there any other tests, I would make it optional there also."
+    },
+    {
+      "prediction": "I'm not sure if we should have a separate route for this or if it should be part of the `my-accounts` route. I think it would be better to keep it separate for now, but let me know if you want me to change that.",
+      "reference": "This can be removed too, right?"
+    },
+    {
+      "prediction": "Shouldn't this be `account.getTransportUris(0)`?",
+      "reference": "If you're going to allow people to specify multiple SMTP servers, you should also properly export/import them."
+    },
+    {
+      "prediction": "I'm not sure if we want these metrics to be exposed to the public. I think they are only used by the ASP.NET Core runtime itself.",
+      "reference": "nit: Can we prefix these variables with `AspNetCore`?"
+    },
+    {
+      "prediction": "I think this test should be `it 'transpiles a YAML plan from a path' do`",
+      "reference": "Can this also include a quick test for transpiling by name?"
+    },
+    {
+      "prediction": "I think we can remove this struct now, since we don't need to pass it to `doPostRequest` anymore.",
+      "reference": "This TODO can be removed"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should go in `MvvmCross.UIElements`",
+      "reference": "can we make tabAccessibilityIdentifier = null as default?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it's the only way I could get it to work.",
+      "reference": "Should we print this error?"
+    },
+    {
+      "prediction": "I think this should be `int(multi_scores.size(1) - 1) - 1`",
+      "reference": "For which reason do we need to convert this to int here?"
+    },
+    {
+      "prediction": "I think this should return an error.",
+      "reference": "FWIW, wondering if the error is interesting here (I actually arrived at this code initially to change `supportsFreezer()` to return a `bool`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the visibility level.",
+      "reference": "Does this still need to be a partial? is the intention down the line to move back towards a conditionally different title?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be in `integration/registry_test.go`",
+      "reference": "seems like `RegistryURL` is missing? Same for a few other places that create `archer.Environment` below."
+    },
+    {
+      "prediction": "I think it would be better to move this import to the top of the file.",
+      "reference": "Are these imports needed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. It seems like it should be in the `text.formControlValue` test suite, but I can't find where it is placed.",
+      "reference": "I think you can remove the `(isIE11 ? it.skip : it)(` above, if you're going to skip the test this way."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `copy` method.",
+      "reference": "IMO we should instead fix the above call to do `long actualChecksumIn = Long.reverseBytes(in.readLong());` to get the actual checksum value? This way the below error message would also be correct?"
+    },
+    {
+      "prediction": "I think this should be `obsoleteIndex` instead of `obsoleteIndexes`",
+      "reference": "I think we also need something similar for `rebuild-index`?"
+    },
+    {
+      "prediction": "I think this should be `defaultBundleEndpointPort = 443` instead of `defaultBundleEndpointPort = 8080`",
+      "reference": "I think convention is snake case for import naming?"
+    },
+    {
+      "prediction": "I think we can remove the `isCommonHandle` check here and just do `indexScanBuilder.addColumns(tableInfo.getColumn(col.getName()).toProto(tableInfo));`",
+      "reference": "i think haspk is false?"
+    },
+    {
+      "prediction": "I think this should be `--surefire-dir` instead of `--surefire-dir --dir`",
+      "reference": "you should add this to the cover command too (but let's have a more general convo about whether it's a flag or config option)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do this in the same way we did before.",
+      "reference": "instead of passing in pointer to empty struct, we should make it optional and support passing nil and use default cfg if it is nil."
+    },
+    {
+      "prediction": "I think this should be `obj.to_json()`, not `obj.to_csv()`.",
+      "reference": "A fair bit faster, although still not great, hence also adding a hashkey."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the progress, but it's the only way I could get to work.",
+      "reference": "If we're going to leave this in here, how about we at least pull it into partial?"
+    },
+    {
+      "prediction": "I think we should be able to get away with 20 seconds here.",
+      "reference": "This `context` is used correctly, there is nothing left TO DO."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to pass the `limit` and `skip` to the `countDocuments` function.",
+      "reference": "If `docs` is an empty array there will be an error"
+    },
+    {
+      "prediction": "I think this should be `!strings.HasPrefix(host, \"quic:\")` and not `!strings.HasPrefix(host, \"unix:\")`",
+      "reference": "Is it really necessary to have the user specify this, or can the reverse proxy infer QUIC from the upstream's Alt-Svc headers?"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in the tests.",
+      "reference": "just remove the entire variable."
+    },
+    {
+      "prediction": "I think this should be `@options[:modulepath] = modulepath.split(File::PATH_SEPARATOR).map { |moduledir| File.expand_path(moduledir) }`",
+      "reference": "Maybe we should put that in the option description?"
+    },
+    {
+      "prediction": "I think it would be better to use `Optional.ofNullable(keyFilePath)` instead of `Optional<String> keyFilePath`",
+      "reference": "Why do we need to make this optional? Isn't the idea of the builder that if you don't need this value you just don't call the `keyFilePath ` method?"
+    },
+    {
+      "prediction": "I think it would be better to have a method for this, like `metricsConfigForTable(Table table)`",
+      "reference": "Shall we do the same for ORC and Avro?"
+    },
+    {
+      "prediction": "I think this should be `MIT` instead of `Apache License, Version 2.0`.",
+      "reference": "Should be a space between line"
+    },
+    {
+      "prediction": "Why not using `SecureRandom` from `System.Security.Cryptography`?",
+      "reference": "if not behind interface then equally we can use SecureRandom.GetBytes directly, otherwise let us push it behind ISecureRandom so we can test with this class wherever used"
+    },
+    {
+      "prediction": "I think we can remove the `tile` class now, since we don't need it anymore.",
+      "reference": "Probably will undo this change for now, as I'm just targeting Weekly Iteration in this PR."
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "Unused parameter warning? How comes we did not catch that before? In any case please open another PR for this."
+    },
+    {
+      "prediction": "`use_group_of` is not used in this function, can you remove it?",
+      "reference": "I recommend move all openimage related logic to a new function"
+    },
+    {
+      "prediction": "I think this should be `ExternalCgroup bool` and not `ExternalCgroup bool ExternalCgroup bool`",
+      "reference": "We shouldn't have a bool for this. If there is a path passed, then we use it. Resource may or may not be empty."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the count, but it's the only way I could come up with that worked.",
+      "reference": "I like how you used the count on the translate call to handle this."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `getSearchHandler` method.",
+      "reference": "Do we need to do this from the outside, or is there a spot in the handler code where we can do this automatically? (I haven't studied it closely yet... just wondering if you've already thought it through or if it's worth taking a closer look)."
+    },
+    {
+      "prediction": "This is not a test. This is a test of the Sputnik library.",
+      "reference": "[Checkstyle] INFO: Using the '._' form of import should be avoided - java.util._."
+    },
+    {
+      "prediction": "I think this should be `settings['cliquet.storage_backend'] = 'cliquet.storage.redis'` instead of `settings['cliquet.storage_backend'] = 'cliquet.storage.redis'`",
+      "reference": "So, why everything is in memory but the permission backend?"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think it would be better to have the `UiSetupWizardImplementation` implement `UiSetupWizard` and have the `UiSetupWizardImplementation` call `UiSetupWizard.showBrowseStep` instead of calling `UiSetupWizard.showBrowseStep` directly.",
+      "reference": "Are you sure it should be null and not an empty List?"
+    },
+    {
+      "prediction": "I think this should be `Frameworks` and not `AdditionalDependencies`",
+      "reference": "Unfortunately I think the assumption of \"empty\" means the default dependency section is not correct. For a project.json file the default dependencies section is a shared section between all the different target frameworks and our TargetGroup being empty may map to many different target frameworks so they aren't really equated."
+    },
+    {
+      "prediction": "I'm not sure if this change is correct. I think it should be `new KeyValuePair<string, StringValues>(\"host\", new[] { \"localhost\" });` and `new KeyValuePair<string, StringValues>(\"custom\", new[] { \"value\" });`",
+      "reference": "This test was verifying the wrong thing."
+    },
+    {
+      "prediction": "I think this should be `development` instead of `brave`",
+      "reference": "pls pr a name change fro muon here too separately."
+    },
+    {
+      "prediction": "I think this should be `opts.envStore.CreateOrUpdateEnvironment(&env)` instead of `opts.envStore.CreateEnvironment(&env)`",
+      "reference": "nit: we can use `defer opts.spinner.Stop(\"Error!\")` after l.180 to make it slightly neater"
+    },
+    {
+      "prediction": "(style) 'is' may be misspelled - perhaps 'isn'? (style) 'is' may be misspelled - perhaps 'isn'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha'? (style) 'is' may be misspelled - perhaps 'isha",
+      "reference": "This is declared in the wrong block"
+    },
+    {
+      "prediction": "I think it's better to log the message itself, rather than the header and the route.",
+      "reference": "In fact it would be useful for debug, i prefer combine them into one debug level log."
+    },
+    {
+      "prediction": "I think it would be better to have a `crashType` or `crashTypeLabel` field in the `alert` object. It would make it easier to distinguish between different types of alerts.",
+      "reference": "We would need to localize them eventually too, just FYI"
+    },
+    {
+      "prediction": "I think this should be `set_type()` instead of `set_name()`.",
+      "reference": "It looks like the proto code needs to change the function name from set_name to set_type to be consistent."
+    },
+    {
+      "prediction": "I think we should remove this file.",
+      "reference": "I think we can avoid using this pattern for this use case."
+    },
+    {
+      "prediction": "I'm not sure why this is needed. It seems like it should be fine to leave out the `#include <spawn.h>` line.",
+      "reference": "Could you please change this to `#ifndef __linux__`? That's when we use our own implementation instead of `posix_spawnp`."
+    },
+    {
+      "prediction": "I think we can drop the `!= null` part here.",
+      "reference": "Nit: put the `&&` on the previous line so that a reader knows that there's more to the statement at a casual glance."
+    },
+    {
+      "prediction": "I think this should be `tlfTypeToString(t)` instead of `tlfTypeToPath(t)`",
+      "reference": "Technically the \"right\" variables to use would be the ones in `libgit/autogit_node_wrappers.go\", which define how the tlf types map to subdirs for autogit specifically. (They would have to be exported.) But this is fine too I guess, just a little dirtier."
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "Should there be a template or something associated with this email? Right now it's identical to `notification_for_approver`"
+    },
+    {
+      "prediction": "I think this should be `em` and not `$em`",
+      "reference": "i hope there is some extension in phpstan for this"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. It's not clear to me what the case is when `sep == 0` and what the case is when `sep > 0` and `sep > flagSep`. I think it would be better to move this check into the `if (sep == 0) return 0;` above.",
+      "reference": "We don't store empty dictionary entries anymore: they bring no benefits, only trouble."
+    },
+    {
+      "prediction": "I think this should be `IsEnum` instead of `opts_.scoped_enums && IsEnum(type);`",
+      "reference": "Optional: This function is C++ specific and could be declared as `static`."
+    },
+    {
+      "prediction": "I think it would be better to have a method `Roles getRoles(String domainName, String tagKey, String tagValue, Boolean members)` instead of adding new parameters to `getRoles(String domainName, Boolean members)`.",
+      "reference": "we can't remove functions as that would introduce backward compatibility issues in our java client. So we should also have a function with the original arguments: ` public Roles getRoles(String domainName, Boolean members) { return getRoles(domainName, members, null, null); } ` For the Go client we don't have a wrapper so no option but to update your client code as necessary, but for java we have the wrapper thus we always maintain backward compatibility"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to `spec_helper.rb` under `RSpec.configure do |config|` so that it can be used in other specs as well.",
+      "reference": "I'd recommend keeping this in `warn_deprecation` and not adding an override of `warn`."
+    },
+    {
+      "prediction": "I think this should be `process.argv.push(blitzPath, 'start');` instead of `process.argv.length = 1;process.argv.push(blitzPath, 'start');`",
+      "reference": "I think we should use a `blitz` bin instead"
+    },
+    {
+      "prediction": "I think this should be `if not dataset._virtual_vdims:`",
+      "reference": "Why not make the ``derived_vdims`` flag (or similar, ``validate_vdims`` maybe?) an explicit argument to ``validate``?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move it to `volume.go` since it's only used there.",
+      "reference": "was it not possible to do this at the filter or porx driver layer? We typically try to avoid driver-specific things in the SDK layer"
+    },
+    {
+      "prediction": "I don't think we need this file.",
+      "reference": "please update the year. it can be setup in intellij template."
+    },
+    {
+      "prediction": "I don't think we should have a test for `log.trace` with no args. I think we should have a test for `log.trace(message, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23, arg24, arg25, arg26, arg27, arg28, arg29, arg30, arg31, arg32, arg33, arg34, arg35, arg36, arg37, arg38, arg39, arg40, arg41, arg42, arg43, arg44, arg45, arg46, arg47, arg48, arg49, arg50, arg51, arg52, arg53, arg54, arg55, arg56, arg57, arg58, arg5",
+      "reference": "what's the use-case for this?"
+    },
+    {
+      "prediction": "I think this should be `if not isinstance(as_index, bool)`",
+      "reference": "nit: `as_index must be an boolean` -> `as_index must be a boolean`"
+    },
+    {
+      "prediction": "Shouldn't this be `updateVersion`?",
+      "reference": "What is the goal of that method ? And what are `$currentVersion` and `$newVersion` ? Are they Thelia or Module version ? :)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best version to use, but it's the one we use in the rest of the SDKs, so I'm going to go with it. Let me know what you think of this.",
+      "reference": "Is there an advantage to this vs setting `this.version` in the constructor? Cleaner this way, at the top?"
+    },
+    {
+      "prediction": "I think you can use `StringUtils.trimToEmpty` here instead of calling `url` again and again.",
+      "reference": "Fix for URISyntaxException."
+    },
+    {
+      "prediction": "I think this should be `if routeUpd.Dst != nil`",
+      "reference": "Would be good to UT this case"
+    },
+    {
+      "prediction": "I think this should be `off` instead of `warn`",
+      "reference": "Why we need to add this?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `ClusterInfo`.",
+      "reference": "This is ok to change since I believe that these were added just in main and not in public release."
+    },
+    {
+      "prediction": "I think this should be `pubsubServiceAccount` instead of `pubsubServiceAccountName`",
+      "reference": "Not needed in this PR, just want to make sure people think about this (maybe make an issue)? This is a lot of strings in a row. It will be hard/impossible for someone reading the code to see that everything is in the correct position. I recommend creating a struct instead of passing seven strings in a row. A similar problem exists on most of the `MakeFoo(...)` style methods in this change."
+    },
+    {
+      "prediction": "I think this should be done in the `OnRunCompleted` event handler of the `SarifDriver` class.",
+      "reference": "_jaw drops_ I would have sworn I wrote those lines. Good catch."
+    },
+    {
+      "prediction": "I think this should be done in a `beforeEach` callback. Otherwise, it will be called every time the module is used, which is not what we want.",
+      "reference": "This needs to be tested."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this test. I think it should go in `spec/models/building.rb`, but I'm not sure if there's a better place to put it.",
+      "reference": "I believe you're missing the comparison after 'be'"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `Get` method of the `KVStoreForTrie`",
+      "reference": "move 103~105 to in front of 97?"
+    },
+    {
+      "prediction": "Can you remove these imports? They are no longer used.",
+      "reference": "Lots of unnecessary imports left in this file."
+    },
+    {
+      "prediction": "I think this should be `descriptor->GetClass() == DescriptorClass::UniformArray`",
+      "reference": "Does this actually produce different code?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for the attribute. I think it would be better to check for `onblur` instead of `onfocus` since that is the attribute we want to check for.",
+      "reference": "This should account for whitespace. Simply putting `.trim()` on the attribute value should do."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "I don't think we need these. Simpler just to put `routetable.TargetType...` inline below."
+    },
+    {
+      "prediction": "I think this should be `method == 'PUT' or method == 'DELETE'` instead of `method == 'PUT' or method == 'DELETE' or method == 'PUT'`",
+      "reference": "Nested if statement. You can merge both statements nested together to create one"
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. It should be moved to `AbstractTest` class.",
+      "reference": "Nitpick: add a space after the comma. Also, we should consider that the format of the version could change in the future so we could get something that is not a parsable integer."
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix. I don't think we want to change the behavior of `get` in this test.",
+      "reference": "For your reference in future, I sugguest to use the `ASSERT_*` family."
+    },
+    {
+      "prediction": "This should be `#include \"fmt/format.h\"`, not `#include \"fmt/format.h\"`.",
+      "reference": "no need to import `fmt/format.h` here since the `KimInit` class is derived from `Pointers`. Any class derived from `Pointers` can assumed that `lmptype.h`, `mpi.h`, `cstddef`, `cstdio`, `string`, `utils.h` and `fmt/format.h` are already included through `pointers.h`."
+    },
+    {
+      "prediction": "I think this should be `import \"github.com/pkg/errors\"`",
+      "reference": "Can we have prefix of `OPENEBS_IO` e.g.: `OPENEBS_IO_INSTANCE_NAME`"
+    },
+    {
+      "prediction": "I think this should be `errors.New(\"GetRootNode is not supported by folderBranchOps\")`",
+      "reference": "Might as well fix these bare returns by making them `return errors.New(...` directly."
+    },
+    {
+      "prediction": "why is this needed?",
+      "reference": "we can use import here"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "thx for removing these"
+    },
+    {
+      "prediction": "Shouldn't this be `throw new RuntimeException(String.format(\"Failed to parse failures XML: %s\", sourceReport), e);`?",
+      "reference": "I snuck this one in as well as checkstyle crashing (on files from resources) caused an unfinished xml to be written, and it wasn't obvious where that XML was"
+    },
+    {
+      "prediction": "I think this test should be renamed to `TestCreateInstanceValidateMachineTypeError` to match the rest of the tests.",
+      "reference": "How is this different than \"good case\"?"
+    },
+    {
+      "prediction": "I think we can remove the `string` here and just use `services []string`",
+      "reference": "the last parameter name should be `serviceName` as is in the implementation"
+    },
+    {
+      "prediction": "I think this test should be moved to `PrettyPrintVisitorTest` since it is not related to `Javadoc` at all.",
+      "reference": "@ftomassetti - okay, this is a little meh."
+    },
+    {
+      "prediction": "This should be `ErrNotImplemented`, not `ErrNotImplemented`.",
+      "reference": "Note, changed initial capitalisation of this for consistency. Can do it the other way around too if that'd be better. :wink:"
+    },
+    {
+      "prediction": "`typeof window === 'object' ? window : this` -> `window ? window.document : this`",
+      "reference": "hey, aren't we supposed to be passing in two parameters here?"
+    },
+    {
+      "prediction": "I think this should be `bitcore.Unit = require('./lib/unit');`",
+      "reference": "please keep alphabetical ordering :)"
+    },
+    {
+      "prediction": "I don't think you need to import `InputStream` and `OutputStream` here. You can just use `Reader` and `Writer` directly.",
+      "reference": "Selenium must compile against Java 6. Revert this line."
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix. I think we need to pass in the `resourceGroup` to the `MakePullSubscription` function instead of passing in the entire `resourceGroup` to the `RunClientSet.PubsubV1alpha1().PullSubscriptions` function. This way the `MakePullSubscription` function will be able to pass in the `resourceGroup` and not have to pass in the entire `resourceGroup` to the `MakePullSubscription` function.",
+      "reference": "what is that `\"\"`, the adapterType? can you add `\"\" /* adapterType */`... in other places as well... might be cleaner if it's a pointer or some varargs at the end but don't have any strong preference.."
+    },
+    {
+      "prediction": "`pType` can be `undefined` here, right?",
+      "reference": "This check is a little odd, and will still mess up with empty strings `''` because empty string is falsy. Can we change this to `|| _this.getValue(path) == null`?"
+    },
+    {
+      "prediction": "I think this should be `if (virtualNode.hasAttribute('tabindex'))` instead of `if (virtualNode.hasAttribute('tabindex') && !isNaN(parseInt(virtualNode.getAttribute('tabindex'), 10))`",
+      "reference": "Just a minor suggestion: If you `parseInt` when you access the attribute then you shouldn't have to do it twice in the ternary."
+    },
+    {
+      "prediction": "I think this should be `body[:target]['host-key-check'] = transport == 'ssh'`",
+      "reference": "This change in particular is strange. I'm not sure why it would now be necessary. The previous default would've been true, and the default behavior without a new net-ssh version should be unchanged."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Why is that? Should be redundant as the default value exists."
+    },
+    {
+      "prediction": "I'm not sure if this should be configurable or not. If it should be configurable, then I think we should have a default of `false`. If it shouldn't be configurable, then I think it should be `true` by default.",
+      "reference": "Are all fields beginning with \"Debug\" automatically `;local` ? (I guess so, but just checking.)"
+    },
+    {
+      "prediction": "I think it would be better to call this method `createUpdateNameMapping` instead of `updateNameMapping` to be consistent with other methods in this class.",
+      "reference": "While I think it makes sense to update the mapping programmatically, I don't see much value in exposing it as part of the table API. We want to keep the Table API small, so if we can handle this by using a separate API that consumes and produces JSON, then that is preferred. Also, we may have more than one mapping in the future, so keeping this decoupled also helps out for those use cases. For now, let's build a utility class to perform these updates. We can always add this to the public API later if we choose to."
+    },
+    {
+      "prediction": "I think this should be `getIpcEnv()->getAllConnections()->waitOnAll(getStatsMergeTimeout() * 10);`",
+      "reference": "We would want ssmp to wake up every 3 seconds if there are no other requests to it and do some cleanup tasks. So, it is not clear why do you want to set this to 0."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in IRBuilder.",
+      "reference": "Is this really necessary? I would expect these types to be included with the loop below."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If we're going to do it at all, I think we should do it in a way that's consistent with the rest of the Jaeger exporter.",
+      "reference": "You could do span<byte> and it's magic of typecast to int for better efficiency"
+    },
+    {
+      "prediction": "I think this should be `cssRule.style[webkitTransform] || cssRule.style[transform] || false;`",
+      "reference": "Looks like you've covered `-webkit-transform`, but what about `-ms-transform`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `MvxBindingLog.Error`.",
+      "reference": "Is there something missing from the message here?"
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "Hmm, what does this syntax mean?"
+    },
+    {
+      "prediction": "I think we can remove the `callerTypeAPI` variable from the `getOrCreateWorkflowExecution` function.",
+      "reference": "should this be callerTypeTask?"
+    },
+    {
+      "prediction": "I think this should be `@plan.administerable_by?(@user.id, :test => true)`",
+      "reference": "Currently update is set as @plan.editable_by?(@user.id) Which one is the correct behavior? I can see a case for only owners/co-owners to be able to set visibility, test status, and other plan details"
+    },
+    {
+      "prediction": "I think this should be `MaxQueueSize = 10_000_000` and `BestKnownSearchLimit = 256_000_000`",
+      "reference": "Maybe make it configurable? Or store the actual level index in DB, making this binary search obsolete?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Nit: this file doesn't need to change. Can you revert this to avoid git conflicts?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `onClick` handler of the `HelpMenu` component.",
+      "reference": "There's no reason for this to be a callback now technically but as per our tech decision, we want all handlers to use `useCallback` now"
+    },
+    {
+      "prediction": "I think this should be called \"LO\" instead of \"LOGIC\"",
+      "reference": "This isn't strictly part of this PR, but could you move `multiSigString` from `crypto/multisig.go` into this list of `HashID` values? Now that we have other things being hashed into addresses (specifically, these new logic addresses), it's important that the hash input for multisig addrs is domain-separated from logic addresses."
+    },
+    {
+      "prediction": "I think this should be `stop_worker_pool`",
+      "reference": "nit: you could have named it just \"TestCase\" (if django's TestCase was imported differently)"
+    },
+    {
+      "prediction": "I think this should be `bridge.print_page(*options)`, since it's a method on the bridge object.",
+      "reference": "the bridge here isn't defined as an accessor / reader to try mask it better. So you need to directly call the iVar `@bridge` here."
+    },
+    {
+      "prediction": "I think this should be `lldb::eNoDynamicValues` instead of `lldb::eNoSyntheticValue`.",
+      "reference": "All the callers of `ExtractSomeIfAny`, always pass `true` to `synthetic_value`. Can we get rid of the extra argument?"
+    },
+    {
+      "prediction": "I think it would be better to move this to the top of the function, so that it's the first thing that gets executed when the script is run.",
+      "reference": "Better to use the absolute path."
+    },
+    {
+      "prediction": "I think it should be 100000000 instead of 2000.",
+      "reference": "this is 16 hours, is that enough?"
+    },
+    {
+      "prediction": "I think we should use `AddErrorToSpan` instead of `AddErrorEndSpan`.",
+      "reference": "removing boilerplate. You could also add a `StartSpan` with varargs to inline string attributes."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to store the rollup context. I think it's better to store it in the EthSyncService.",
+      "reference": "Can you replace these single letter variables with full names?"
+    },
+    {
+      "prediction": "I think this should be `Stream? Connection { get; set; }` and the `PrepareRequest` should be `Func<IFeatureCollection>`.",
+      "reference": "At first I wasn't sure, but now I think I'm sold on making PrepareRequest an action over adding a state object to the context."
+    },
+    {
+      "prediction": "This should be `preg_match('/^VuFind (\\d+\\.?\\d*)$/', $newConfig['Site']['generator'])`",
+      "reference": "Would it be too greedy to preg_replace `VuFind (\\d+\\.?)+` with `'VuFind ' . $this->to` anywhere in the string? This would update something like 'Finna (VuFind 7.1.0)' as well. Just a thought, please disregard if you'd like to keep it as is."
+    },
+    {
+      "prediction": "I think this should be `getZkStateReader().getClusterState().getCollection(collection).getConfigName();`",
+      "reference": "BTW for brevity, you can remove `getZkStateReader().` here and elsewhere since ZkController has a convenience method for the cluster state."
+    },
+    {
+      "prediction": "`manifests` is a `Iterable<ManifestFile>`, not a `CloseableIterable<ManifestFile>`.",
+      "reference": "If there are no manifests, then entries should be `CloseableIterable.empty()`, not the manifest iterable. That doesn't need to be closeable."
+    },
+    {
+      "prediction": "I think it would be better to have a separate test class for this rule. This way you don't have to repeat the same test code in every test class.",
+      "reference": "this class should be named `ArrayInitializationVerbosenessTest` to work"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this validation. I think it should be in the `logstash` gem.",
+      "reference": "Do we want to allow users to disable `console` as well? The schema currently says that it only permits a hash for `console`."
+    },
+    {
+      "prediction": "Why is this not working?",
+      "reference": "It would be very helpful to describe succinctly either inline or by linking to an issue going into depth why each test is not working. If we merge like this your knowledge of what is going on is lost and other people in the code need to do a ton of reading before understanding when/how/if we should unskip."
+    },
+    {
+      "prediction": "I think we can remove `previousDataLoaded` and `previousError` from the `currentError` and `currentDataLoaded` from the `previousError` and `previousDataLoaded` from the `currentDataLoaded` and `currentError` from the `currentDataLoaded` from the `currentDataLoaded`",
+      "reference": "As this is a `useEffect` this could be a cause for concern! Note that `useEffect` re-runs when a dependency changes **not** when a dependency is truthy (the previous code does look a bit like it's expecting that)"
+    },
+    {
+      "prediction": "`delete` is not needed here.",
+      "reference": "Should we be doing this with `make_unique` or something?"
+    },
+    {
+      "prediction": "I'm not sure if this is the ideal data to consume.",
+      "reference": "we don't want to set default values for this. Since it is the actual email, we want to make sure we are always passing in the `user` and `proposal` . we have a default value of `nil` for modifier because sometimes will update a proposal via `rails console` in which case there will be no recorded modifier."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Yes this import was in the wrong place, but should ideally be in the first list of imports (among `go/build`, `go/token`, etc.). You may move it there, or just revert this change as it's actually unrelated."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure why it was changed in the first place.",
+      "reference": "It's unnecessary to duplicate the `name`."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "this was a mistake when adding phpstan strict plugin"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems to be the only place I could think of.",
+      "reference": "Nit: I think it should be `to handle` here."
+    },
+    {
+      "prediction": "Please revert this change.",
+      "reference": "Please don't make unrelated changes or let your tools make unrelated changes."
+    },
+    {
+      "prediction": "I think this should be `uint` instead of `int` since it's an unsigned integer.",
+      "reference": "`numAccounts` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think this should be done in the `Plugin::remove()` method.",
+      "reference": "Is the ID correctly normalized at this point?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this.",
+      "reference": "if 0 is a special constant meaning \"no limit\" let's make it a constant and name it correctly :)"
+    },
+    {
+      "prediction": "I think this should be done in `runInventory()` instead of here.",
+      "reference": "Curious about the reason behind this change"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this. I think it's better to put it in the `cborStore.Put` method.",
+      "reference": "wait, i thought the cborstore was for state tree?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `set_materialized_location`",
+      "reference": "This logic is now duplicated here and in `_delete_temporary_files`... Maybe do it all at once - find temp files, delete them, and fix the logical keys?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think this should be done in the `ResponseType` class.",
+      "reference": "How about name it with NullResponseTypeProcessor"
+    },
+    {
+      "prediction": "I think this should be `if (config_.max_bin_by_feature[i] == 0)`",
+      "reference": "Maybe `Log::Fatal` here? `[LightGBM] [Fatal] Check failed: feat_is_unconstrained at ...` (and kernel death in case of Jupyter Notebook) seems to be not so informative and user-friendly."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in `decimate.py` since it's only used by `decimate.py`.",
+      "reference": "Not sure I like the name ``_apply``. Even though ``_process`` is supposed to process elements already, how about ``_process_element`` which processes elements, *excluding* Overlays/NdOverlays."
+    },
+    {
+      "prediction": "I think we can remove NodeRoleTagValue and use `NodeRole` instead.",
+      "reference": "Let's use this constant when ASG is created as well (currently hardcoded)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `objects.py` file, under the `GROUP BY` section.",
+      "reference": "Oh is `bucket` actually higher cardinality than `eventname`?"
+    },
+    {
+      "prediction": "I think this should be `from PyQt5.QtWebEngineWidgets import QWebEngineProfile  # pylint: disable=unused-import`",
+      "reference": "You can't rely on QtWebEngine being available - but why do you need to import this here at all?"
+    },
+    {
+      "prediction": "I don't think you need to do this. You should be able to do `py2exe --force-includes --onefile --no-py2exe` and it should work.",
+      "reference": "How did you determine these - and how can we be confident that nothing is missing?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this. I think it would be better to put it under the `Workshops` section since it's related to the `Workshops` section.",
+      "reference": "What is this page going to do that is different than the products index we already have?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Couldn't you just write this as an if/else or a guard clause like on line 87? Just seems a bit weird doing this conditional assignment for essentially an if/else."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "Is this intended to be here? Seems it belongs to a different PR."
+    },
+    {
+      "prediction": "Why not use `SetConsoleColor`?",
+      "reference": "These changes, while harmless, were no longer needed with newer versions of System.Console thanks to @ianhays"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this, but it's the only way I could find that works.",
+      "reference": "Umm... this isn't relevant to this PR"
+    },
+    {
+      "prediction": "I think this should be `0x8000 | 0x1000 | 0x2000 | 0x4000 | 0x4000 | 0x8000`",
+      "reference": "While All make some sense here it doesn't make a lot of sense in the SkipOnFramework context. I wonder if we really need to expose anything more here. You can just blindly use 0."
+    },
+    {
+      "prediction": "I think this should be `// Package main provides the Bee binary viper-cobra command definitions.`",
+      "reference": "Actually, this is not correct. Package `cmd/bee` doe snot have any notion of commands or even libraries used for them. It is as simple as possible, just calling the function from `cmd/bee/cmd` which actually implements commands."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this route. I think it should go in `user` folder.",
+      "reference": "This change is causing tests to fail because users sits at `/app/users`, not `/app/admin/users`"
+    },
+    {
+      "prediction": "I think this should be `null, null` instead of `null, strict` and `null, lax`",
+      "reference": "Could you provide more information on what exactly each of those three options do?"
+    },
+    {
+      "prediction": "I don't think we need to expose the perfCollect here.",
+      "reference": "Why this line break?"
+    },
+    {
+      "prediction": "@deads2k @pgavlin can you review this change?",
+      "reference": "We're importing this twice with different aliases."
+    },
+    {
+      "prediction": "I think we can remove `pandera==0.8.0` now.",
+      "reference": "We should avoid adding a pandera dependency to Dagster itself."
+    },
+    {
+      "prediction": "I don't think you need to create a new array, you can just do `this.permissions = this.permissions.concat(new ObjectPermission[] { new ObjectPermission(privilegeType, permission.getGrantor(), permission.permissions) });`",
+      "reference": "Can be replaced with `org.jkiss.utils.ArrayUtils#add`."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to do this synchronously, and then close the `ElectionContexts` stream in the `finally` block.",
+      "reference": "While we're here, this line (and a few others) should be `customThreadPool.submit(() -> IOUtils.closeQuietly(overseer);` I have no idea why we're creating a collection and a stream for a single object."
+    },
+    {
+      "prediction": "I think we can remove the `clientID` and `clientSecret` fields from the `AzureDNS01` struct as well.",
+      "reference": "Should these restrictions also be reflected in the API docs?"
+    },
+    {
+      "prediction": "Why do we need to restore the original request?",
+      "reference": "What about the request headers if someone uses `request_header`? :thinking:"
+    },
+    {
+      "prediction": "I think this should be `deployment.Spec.Template.Spec.Containers[0].Args = append(deployment.Spec.Template.Spec.Containers[0].Args, fmt.Sprintf(\"--domain-filter=%s\", instance.Spec.ManagedDomains[0]))`",
+      "reference": "Verify that external-dns allows setting this parameter multiple times (that it's not \"last one wins\")."
+    },
+    {
+      "prediction": "I think this should be `host1.example.com` instead of `host2.example.com`",
+      "reference": "Maybe 2 host/ip values could be used in the example?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move it to the `RateLimiter` test class.",
+      "reference": "We're only testing `RateLimiter.Allowed(Span)`, so I think we can use a mock `ITraceContext` instead of a real `Tracer` or `TraceContext`."
+    },
+    {
+      "prediction": "`trainer->get_name()` should be `trainer->name()`.",
+      "reference": "IMO, this is the wrong way to associate a model with a trainer... Why the string when you actually have a trainer object?"
+    },
+    {
+      "prediction": "I think this should be `ledger.store.block_exists (transaction, block_a.type (), hash);`",
+      "reference": "Wondering if we can directly put types here & for other similar ledger_processor items like rai::block_type::state"
+    },
+    {
+      "prediction": "I think we should keep the progress bar, but we should not report the errors.",
+      "reference": "As `DeleteFiles` is not only used for prune this function deserves its own file. Maybe something like `delete_files.go` or `parallel.go`?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "It should be simplified character. `` -> ``"
+    },
+    {
+      "prediction": "This should probably be `int` instead of `-1`.",
+      "reference": "I am pretty sure it breaks the visitor pattern, visitor should have no knowledge about the structure of what it is visiting or control over visiting mechanism"
+    },
+    {
+      "prediction": "`HashMap` and `TreeMap` are both `ConcurrentHashMap` implementations. I think it would be clearer to use `ConcurrentHashMap<K, V>` and `ConcurrentSkipListMap<K, V>` instead.",
+      "reference": "'A' instead of 'An' here"
+    },
+    {
+      "prediction": "I'm not sure we should change this. I think it's fine to leave it as it was, but I'm not sure what the rationale is for changing it.",
+      "reference": "Can you flip it back (here and in other places)? It is more natural to represent \"10 seconds\" as `10 * time.Second`."
+    },
+    {
+      "prediction": "I don't think you need to create a new data here, you can just use the existing one.",
+      "reference": "Why was this necessary? You should just use the `dask_classifier` defined below this. With this change, you'd only be doing the local predict on arrays each time, but we want to test on all of DataFrame, Array, and sparse matrix."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `app/views/shared/_checkouts.html.erb`",
+      "reference": "To me, this is sort of weird to have a completely separate page for this. Can we extract the different stuff (github, videos, downloads) presentation from the top of `app/views/purchases/show.html.erb` and reuse it on this page. I worry that we'll change things with fulfillment or copy and have to change it in multiple places."
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "Please revert the lines above. Not touching unrelated code keeps the git history clean."
+    },
+    {
+      "prediction": "I think this file should be removed.",
+      "reference": "`short_version` and `version_info`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Task` struct itself.",
+      "reference": "This is supposed to be `status`. Why did you change this?"
+    },
+    {
+      "prediction": "I think this is the wrong file. It should be `pkg/apis/core/internalversion.go`",
+      "reference": "I will change all of these at some point next week."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "I made this change because, while trying to debug, in Visual Studio, the `IntegrationSignatureTests`, this property getter was hit and interfered with my ability to complete the debugging."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Deployment` resource itself.",
+      "reference": "nit: `\"Unexpected error while calculating: unable to convert %T to %T: %v\"`"
+    },
+    {
+      "prediction": "I think this should be renamed to `applicationName` instead of `appName` to be consistent with the rest of the API.",
+      "reference": "I'd actually recommend not putting this in the `SampleConfig`, it's intended more for properties that are inherent of the discovery format. Add a method `getSampleAppName(String apiTypeName)` to `SampleNamer` and override it in the language specific `SampleNamer`s if needed. Then assign it in the transformer."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should go in `src/service/url-replacements-service.js`",
+      "reference": "Do we need to export these new functions? I thought the plan going forward was to keep util imports separated to make circular imports less likely and possibly do better chunk splitting. Since only new code references these files, we can scope the import to the specific files."
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "Rails will autorequire `acts_as_editable` once it encounters the constant `ActsAsEditable`. Thus if we move `ActiveRecord::Base.send :include, ActsAsEditable` from _acts_as_editable.rb_ to this file, the require line will not be needed anymore. Instead of requiring the file ourselves, we will let Rails do it for us. This fix will also remove the need to explicitly require `acts_as_editable` in _mock_active_record_base.rb_."
+    },
+    {
+      "prediction": "I think this should be `public abstract String upperCamelMethodName();`",
+      "reference": "I think we should name this `grpcMethodName` or similar. That is what this refers to, right? The fact that it is upper camel is an implementation."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this validation. I think it should be done in the `xfpga_fpgaTokenReadObject` function.",
+      "reference": "What if .. appears, but not as the first character, eg \"errors/../../../../../../../../../../../\"?"
+    },
+    {
+      "prediction": "I think this should be `owned by an organization, of which the user is a member`",
+      "reference": "Thanks, this should make things a bit easier for people who pick up the codebase but aren't a `curation center`"
+    },
+    {
+      "prediction": "I don't think we should be deprecating any of the static fields in this class.",
+      "reference": "why not removing it?"
+    },
+    {
+      "prediction": "I think it would be better to use `FlagTLSEnableHostVerification` and `FlagTLSDisableHostVerification` instead of `FlagTLSServerName` and `FlagTLSDisableHostVerification`.",
+      "reference": "same nit here: maybe word as - \"disables validation of the temporal cluster's server certificate\""
+    },
+    {
+      "prediction": "I think this should be `sql(\"SHOW TABLES\")` instead of `sql(\"show tables\")`",
+      "reference": "Nit: other statements use all caps for SQL reserved words. Should this be `SHOW TABLES`?"
+    },
+    {
+      "prediction": "I think this should be `|| c == '.' || c == ':'` instead of `|| c == '.' || c == ':' || c == '.' || c == ':'`",
+      "reference": "Do we need this? isLuaIdentifier() is used not only in extract_next_token() but also in extract_prev_toke(). I wonder whether the change for isLuaIdentifier() has an impact on extract_prev_toke() or not. If you are not sure, keep isLuaIdentifier() as is. If you are sure, could you write your conviction to the commit log."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `ConfigureDefaultMTUs` since we're setting `Wireguard.EncryptHostTraffic` and `c.Wireguard.MTU`",
+      "reference": "This line should actually be: `if c.KubernetesProvider == config.ProviderAKS && c.Wireguard.EncryptHostTraffic {` because we only need to tweak the MTU like this on AKS."
+    },
+    {
+      "prediction": "I think we should rename this to `serviceFlagDescription` to be consistent with the other flags.",
+      "reference": "Do we need to change this flag as well?"
+    },
+    {
+      "prediction": "I think this should be `module.exports = file;`",
+      "reference": "Including the source twice here makes the filesize jump to 432kb. Is there any way to minimize repeating it?"
+    },
+    {
+      "prediction": "I think this should be `setParameter('nameOrHandle', nameOrHandle)`",
+      "reference": "This should only be sent if the driver is speaking to a W3C conformant remote, so we need an if-condition check like we have in the Python bindings."
+    },
+    {
+      "prediction": "I think this should be `target['variables']` instead of `target['variables']`",
+      "reference": "Only one line of code? What a simple issue!"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `rateLimiters` section of the config.",
+      "reference": "Conceptual question: why do history and matching need throttler at all? Shouldn't throttling to be handled on FE only?"
+    },
+    {
+      "prediction": "I think we can remove the `enqueueAppliedToGroup` parameter from this function.",
+      "reference": "Would this be possible: `addCNP` has already processed the CNP to an internalNP, just hasn't added this internalNP to the `internalNetworkPolicyStore`. In this case, `reprocessCNP` will skip processing this CNP and `addCNP` will just add the \"old\" internalNP to `internalNetworkPolicyStore`."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [122/120]",
+      "reference": "Minor: you can access `current_user` in here directly - don't need to pass it in. Unless you prefer passing it explicitly?"
+    },
+    {
+      "prediction": "I think this should be `attributes = attributes&.to_h if attributes.respond_to?(:keys) && attributes.keys.present?`",
+      "reference": "What is the purpose of `&` on this line?"
+    },
+    {
+      "prediction": "I think this should be `^[a-zA-Z0-9][a-zA-Z0-9-]*$` instead of `^[a-zA-Z0-9]+$`",
+      "reference": "See above, the `isValidWebDataStreamID` implementation from before is actually correct. What we need here instead is a new `isValidMeasurementID` function."
+    },
+    {
+      "prediction": "I think this should be `if err != nil` instead of `if err != nil && c.flags.isSet(infoReceived)`",
+      "reference": "Should be an error... it is important for the admin of the server attempting to create the leafnode connection to see the error asap."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "Could you split these conditions up over multiple lines please."
+    },
+    {
+      "prediction": "I think we can remove the `tag` field from the struct now, since it's no longer used.",
+      "reference": "I think we can just change this to `Split` instead of `SplitN`. I don't think we need to limit the number of parts in the tag."
+    },
+    {
+      "prediction": "I think we can remove the `{{ .dbaport }}:{{ .dbaport }}` portion of this line.",
+      "reference": "So this is actually a mistake that I introduced without realizing it, and have known would need to be fixed when we get here  . The format of the ports directive is \"host:container\". We only want the host port to change, not the internal container ports. The ports variable probably needs to be a map which maps external -> internal ports."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "An alternate approach would be to always initialize shardInfo with a dummy (non-nil) value, which could protect against other uses before it's initialized (I couldn't find any though). But I can change it to do it that way instead."
+    },
+    {
+      "prediction": "I think this should be `NodeMetadata struct{ LastWriter keybase1.UID }` and not `NodeMetadata struct{ LastWriter keybase1.UID }`",
+      "reference": "Maybe we can just put the entire `DirEntry` in here? I can imagine it might be useful for debugging to get the block ID/refnonce, encrypted size, key gen, data version, etc. What do you think?"
+    },
+    {
+      "prediction": "`c->maximized` is a boolean, so `c->maximized_vertical` and `c->maximized_horizontal` should be `c->maximized_vertical || c->maximized_horizontal`.",
+      "reference": "For the commit message: The paragraphs seem to be out of order? The `Would not work because` refers to the stuff before, but there is a `This may seem pointless, but` in-between."
+    },
+    {
+      "prediction": "I think we can remove `influxdata/influxdb` from the `prelude`.",
+      "reference": "Why did you remove these packages from the prelude, will your new functions not work without this change?"
+    },
+    {
+      "prediction": "nit: you can use `std::vector<std::string, std::string::iterator>`.",
+      "reference": "We'd better reserve enough space before using kvKeys to avoid extra malloc."
+    },
+    {
+      "prediction": "I think we can remove this function as it's not being used anywhere.",
+      "reference": "would it make sense to return empty or error if label doesn't exists?"
+    },
+    {
+      "prediction": "I think we should use `setTimeout(attemptRm, attemptTimeout);` instead of `attemptRm();`",
+      "reference": "Don't penalize everyone with 250ms delay b/c some machines have problems."
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `\"{\"`",
+      "reference": "Probably want a space after the second %s?"
+    },
+    {
+      "prediction": "I think this should be `Unavailable` instead of `UnavailableError`",
+      "reference": "This seems like an internal error still?"
+    },
+    {
+      "prediction": "Block has too many lines. [29/25]",
+      "reference": "Line is too long. [94/80]"
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this. I think it's better to call it `AntreaClusterIdentity` or `AntreaClusterUUID` or something like that.",
+      "reference": "out of curiosity, why creating another struct to wrap it?"
+    },
+    {
+      "prediction": "Please use `std::chrono::minutes`, `std::chrono::seconds`, `std::chrono::duration_cast<std::chrono::minutes>`, `std::chrono::duration_cast<std::chrono::seconds>`, `std::chrono::duration_cast<std::chrono::hours>`, `std::chrono::duration_cast<std::chrono::seconds>`, `std::chrono::duration_cast<std::chrono::days>`, `std::chrono::duration_cast<std::chrono::hours>`, `std::chrono::duration_cast<std::chrono::days>`, `std::chrono::duration_cast<std::chrono::milliseconds>`, `std::chrono::duration_cast<std::chrono::microseconds>`, `std::chrono::duration_cast<std::chrono::milliseconds>`, `std::chrono::duration_cast<std::chrono::microseconds>`, `std::chrono::duration_cast<std::chrono::milliseconds>`, `std::chrono::duration_cast<std::chrono::microseconds>`, `std::chrono::duration_cast<std::chrono::milliseconds>`, `std::chrono::duration_cast<std::chrono::microseconds>`, `std::chrono::duration_cast<std::chrono::microseconds>`, `std::chrono::duration_cast<std::chrono::",
+      "reference": "discipline vs. Discipline? Not sure of which is correct. Also not sure if there is already an existing string const."
+    },
+    {
+      "prediction": "I think this should be `protoFile = protoFile != null ? protoFile : DEFAULT_PROTO_FILE;`",
+      "reference": "I believe this will break common resources, as they are defined in a common file, but must generate classes in service-specific namespace, so using protoFile to determine package of the generated class would not work, because common_resources namespace does not match service namespace."
+    },
+    {
+      "prediction": "I think this should be called `DEFAULT_NAME` instead of `DEFAULT_NAME_0`.",
+      "reference": "How about `SPARK_DEFAULT_SERIES_NAME`?"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for the default value of this property.",
+      "reference": "The other properties are in `commit.status-check`, not `commit.status-checks`. Could you remove the extra `s`?"
+    },
+    {
+      "prediction": "I think this should be `skip()` instead of `return`",
+      "reference": "Side note: I'd love to see more of these fields exposed on configuration."
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "In the case of accounts generated by API clients, what name gets put for the `inviter_name` or `invited_by` record?"
+    },
+    {
+      "prediction": "Why do we need to implement `getStartBlock`?",
+      "reference": "I think this class is only relevant for updating the QbftConfigOptions based on the transitions config, so probably not."
+    },
+    {
+      "prediction": "Shouldn't this be -32017 instead of -32016?",
+      "reference": "what is the number thrown by Geth?"
+    },
+    {
+      "prediction": "Why do we need to split the query here?",
+      "reference": "This would still crash on malformed database connection string like \"test\""
+    },
+    {
+      "prediction": "I think it would be better to have a `pluginNameToContext` function that takes a plugin name and returns a context.Context with the given name.",
+      "reference": "it is not an interface, maybe we can move to another package?"
+    },
+    {
+      "prediction": "`priv` is not used anywhere in this class. Can you remove it?",
+      "reference": "Same here. Can we use `this.instance.getSettings()` to get `tableMeta`?"
+    },
+    {
+      "prediction": "I think this should be `return cgroup.Resources.CpuWeight != 0 || cgroup.Resources.CpuMax != \"\"`",
+      "reference": "Not sure whether this validation is useful."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this functionality. I think it should go in `promptutils`, but I don't know if there is a better place.",
+      "reference": "Various things to note here: - Why would `question` ever be falsey (or `None`) here, i.e. why the `if question`? - Why would `question` ever not have that attribute? You set it in `Question.__init__`, so that check does nothing at all (and generally, `hasattr()` should be avoided if possible). - Why do you want to yank `None` (what does that do?) rather than e.g. show an error if there's nothing to yank?"
+    },
+    {
+      "prediction": "I think this should be `print(event['Stats'].get('Details'))` and not `print(event['Stats'].get('Details'))`",
+      "reference": "If this is needed for testing, you should use `pytest --log-cli-level=INFO` instead."
+    },
+    {
+      "prediction": "I'm not sure if this is the best URL to use, but it works for now.",
+      "reference": "We should drop ipify in all places i guess"
+    },
+    {
+      "prediction": "I think we can remove the `FilterTimeoutSmallLambdas` and `FilterTimeoutPeriod0SmallLambdas` as they are not used anymore.",
+      "reference": "Add explanations around these variables - what do they mean, how they should be configured, etc."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to move it to `app/models/osm_community_index/local_chapter.rb` since that is where the `add_to_i18n` method is.",
+      "reference": "One thing tho - I would really appreciate any advice on where to move this to, so that it's called on initialisation of the website."
+    },
+    {
+      "prediction": "This should be `@ConfigItem(position = 21, keyName = \"displayResources\", name = \"Show raw resources gathered\", description = \"Displays how much of each resource you have gathered.\", titleSection = \"resources\")`",
+      "reference": "To much tabs here (1 tab)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to set the limit. I think it should be done in the `update` method of the `ExecutorManagerDao`",
+      "reference": "This is not maintaining retention time for logs as we are deleting only 1000 entries and we might end up with huge number of rows spanning over multiple months over a period of time if the cluster generates more rows as we are restricting ourselves to delete only 24k rows/day. Like I pointed out earlier a better would be to probably run this clean up every hour with some batch_size in a loop while there are records to be deleted(i.e older than 12 weeks) and committing each batch."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I could come up with.",
+      "reference": "I don't like this helper func :-). Yeah, definitely don't like."
+    },
+    {
+      "prediction": "I think we can remove the `else if` here, since it's the same as the `else` below.",
+      "reference": "Same here. This should remain unchanged. Please, add it back."
+    },
+    {
+      "prediction": "I think this should be `if (this.s.options.ignoreUndefined && this.s.options.ignoreUndefined === true)`",
+      "reference": "@kvwalker should this method have `ignoreUndefined` supported as an `option`?"
+    },
+    {
+      "prediction": "Why do we need this change?",
+      "reference": "we dont use a `sugaredlogger` anywhere else in clutch, just curious why were using it here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we need to make sure that the queue is always populated. If it's empty, we need to wait until there are events in the queue. If there are events in the queue, we need to remove them from the queue and add them back to the queue.",
+      "reference": "Not evident from looking at this diff but the queue is already being cleared four lines below."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the `computeFeePerByte` function.",
+      "reference": "Why is this a 'transaction' and not in the block header?"
+    },
+    {
+      "prediction": "I don't think we need to expose this. It's only used by the client.",
+      "reference": "This doesn't fit the abstraction level. Should we have a property bag for extra stuff?"
+    },
+    {
+      "prediction": "I think this should be `ci/gendoc.py`",
+      "reference": "Looks like it is pretty easy to miss it when adding new files under `ci`. Hopefully, this doesn't happen too often."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to `NewBee` so that it can be used for standalone as well.",
+      "reference": "do we really need the `Ready` call if startup fails midway?"
+    },
+    {
+      "prediction": "I think this should be `std::unique_ptr<kvstore::KVStore>(TestUtils::initKV(rootPath.path(), partitions, {0, network::NetworkUtils::getAvailablePort()}))`",
+      "reference": "You can avoid changing the code if there are parameter defaults. right ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/utils.js`.",
+      "reference": "I wonder why the newline disappeared here? I guess we'll merge and see how SauceLabs fares across the supported browsers."
+    },
+    {
+      "prediction": "This should be `if (world_repop > 0)`.",
+      "reference": "I know this is legal, but I prefer we put brackets on our calls"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this sample. The `AmazonIamService` class has a `deleteServerCertificate` method, which is used to delete a server certificate. The `AmazonIamClient` class has a `getServerCertificate` method, which is used to retrieve a server certificate. The `AmazonIamPermissionsClient` class has a `getServerCertificatePermissions` method, which is used to retrieve the permissions associated with a server certificate. The `AmazonIamRoleClient` class has a `getRole` method, which is used to retrieve a role associated with a server certificate. The `AmazonIamUserClient` class has a `getUsers` method, which is used to retrieve a list of users associated with a server certificate. The `AmazonIamOrganizationClient` class has a `getOrganizations` method, which is used to retrieve a list of organizations associated with a server certificate. The `AmazonIamPolicyClient` class has a `getPolicy` method, which is used to retrieve a policy associated with a server certificate. The `AmazonIamProgramClient` class has a `getPrograms` method, which is used to retrieve a list of programs associated with a server certificate",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think this should be renamed to `NoSuchInterceptingElement` instead of `InterceptingElementException`.",
+      "reference": "We keep who wrote the code anonymous."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "Two things to make the code more error-proof and simplify it: 1. To be safe, this should be: `! isUndefined( decimal ) && ! isUndefined( decimal.value )` 2. The other clause that you changed below can be combined with that since `decimal` isn't used anywhere else, and so there's no point to re-check whether it's not undefined. In other words, this should become: `! isUndefined( decimal ) && ! isUndefined( decimal.value ) && 1000 > number`, and then you can remove the `decimal = decimal.value` assignment and replace it with line 146 (assignment of `readableNumber`)."
+    },
+    {
+      "prediction": "I think it would be better to have a separate `OutboundOptions` interface, and have the implementation of `Outbound` implement `OutboundOptions` and have the `Options()` method of `OutboundOptions` return the `Outbound` object. This way, we can still use the `Transports()` method to get all the transports used by the outbound, and we can still use the `Options()` method to get all the options used by the outbound.",
+      "reference": "Wouldn't composite outbounds compose the transport, such that it would still be represented as a single transport?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "nit: there is no need for a new line here, usually we tend to put an object creation and its deferred deletion next to each other. It helps not forgetting about the cleanup."
+    },
+    {
+      "prediction": "Why not using `Symfony\\Component\\HttpKernel\\Event\\GetResponseForRequestEvent`?",
+      "reference": "Feel free to use the full notation with my email address: `Maxime Steinhausser <maxime.steinhausser@gmail.com>` :smile:"
+    },
+    {
+      "prediction": "I think it's better to use `fmt.Sprintf` instead of `\\r\\n\\r\\n`.",
+      "reference": "This was unused across the codebase"
+    },
+    {
+      "prediction": "I'm not sure if we want to log the error here or not. If we want to log it, I think it would be better to log the error in `LookupAccount` and not `processConsumerAssignment`.",
+      "reference": "Should we put in which account we were trying to look up?"
+    },
+    {
+      "prediction": "I think this should be `InventoryTypeClass`, not `InventoryIndex`.",
+      "reference": "either SUPPORTED_TYPECLASS or supported_typeclass. Camelcase only for class names."
+    },
+    {
+      "prediction": "I think this should be `pinned INT NOT NULL DEFAULT 1`",
+      "reference": "This is not going to work. You need to add a migration to the database."
+    },
+    {
+      "prediction": "I'm not sure what this method is supposed to do. It's not clear what it does.",
+      "reference": "Should these methods be renamed in this PR?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `MeterProviderBuilderSdk` class.",
+      "reference": "we need to include the `metricStreamName` which is invalid, so users know whats causing the issue."
+    },
+    {
+      "prediction": "I'm not sure if this is the right approach. I think we should be able to load the pdb without having to specify the full path to the pdb file.",
+      "reference": "Please run Platform tests `DiaSessionTests`."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that we don't know how many of these prefixes we have, so we can't do a proper count. I think it would be better to do something like `AVX_512_PREFIXES_COUNT` or similar.",
+      "reference": "This is debug-build-only: we want a release-build notice. SYSLOG takes the name w/o the MSG and no _STRING. Args are via events.mc specifiers."
+    },
+    {
+      "prediction": "I think this should be `callback(new Error('directConnection not supported with SRV URI'))`",
+      "reference": "is it possible to have both forms here? I was hoping we wouldn't be introducing more cases where we had to check the upper and lowercase version of URI options."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this benchmark. I think it should be in `test/int64_histograms.go`",
+      "reference": "Does the name suffix hardcode something? Not clear how changing the name fixes this."
+    },
+    {
+      "prediction": "I think this should be called `getQueueProcessorThreadStates` to be consistent with the other methods.",
+      "reference": "Is undispatched same as queued? getQueuedFlows?"
+    },
+    {
+      "prediction": "I think we should keep the `connection info` tests here, but rename them to be `site info` tests. The `connection info` is a bit misleading because it's not really about the connection info itself, but about the site itself.",
+      "reference": "See above, same for all similar cases below."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move this test to `test/SingleImageImportLoggableBuilder_test.go`",
+      "reference": "Why is this done twice?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "This sync can ensure that the condition of the child node is consistent with the parent node, but I still don't understand when the child will be overwritten, and this behavior is not what we expected?"
+    },
+    {
+      "prediction": "I think this should be `DEBUG_TEST_FAILURE=true` instead of `DEBUG_TEST_FAILURE=true`",
+      "reference": "This was renamed to avoid any confusion with the more general case of debugging via `plz debug`"
+    },
+    {
+      "prediction": "I think this should be `Math.min(((String) val).length(), prefixLength)` instead of `Math.min(((byte[]) val).length, prefixLength)`",
+      "reference": "String valStr = (String)val; valStr.substring(0, Math.min(valStr.length(), prefixLength)) Make it clean."
+    },
+    {
+      "prediction": "I think it would be better to use `enum_def->type` instead of `BASE_TYPE_NONE` here.",
+      "reference": "Sorry, this still makes no sense.. the value of any union or enum is simply a name integer constant, it has nothing to do with the `BASE_TYPE_` enum. This value should be `0`. In particular: `Every union has the NONE field, which always has value 0`."
+    },
+    {
+      "prediction": "I think we should move this method to AzkabanClient.",
+      "reference": "Looks like getAzkabanURL() is added after concrete Alerter (Emailer)::getAzkabanURL(), so that Override annotation is needed."
+    },
+    {
+      "prediction": "I think this should be `if mask.dtype == torch.bool:`",
+      "reference": "The above 3 lines can be written as: `mask = segms[i].astype(bool)`"
+    },
+    {
+      "prediction": "nit: I think it would be better to move this file to `pkg/` and rename it to `address.go`",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I think this should be `logger.always_log (boost::str (boost::format (\"Node started with peering port `%1%`\").format (network.port)));`",
+      "reference": "Thought it's good to have this logged down so that we can check the used value in the logs."
+    },
+    {
+      "prediction": "I'm not sure if we want this to be optional or not. If we want it to be optional, I think we should add a `omitempty` so that it doesn't get written to the response if it's not set.",
+      "reference": "I don't think you can call this optional unless it's a pointer - remove this?"
+    },
+    {
+      "prediction": "I think you can use `listSaveProduct.isEmpty()` instead of `listSaveProduct.size() == 0`",
+      "reference": "@naofum please use `isEmpty`method instead"
+    },
+    {
+      "prediction": "I think this is the only place where `nullable_handle` is used, so I think it should be removed.",
+      "reference": "i think this is unrelated and needs to be reverted. need to discuss at sync."
+    },
+    {
+      "prediction": "I think this class should be renamed to `ProcessableBlockHeaderWithParentHash`.",
+      "reference": "Are we eventually moving those type interfaces from plugin project to datatype project? If not, then we'd have a dependency from core onto plugins, which seems a little counter-intuitive to me."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `models/index.js` file, since that's where the models are defined.",
+      "reference": "Love how we can finally use async/await in the codebase."
+    },
+    {
+      "prediction": "This should be `wallet_representative_update_existing_accounts`.",
+      "reference": "Very minor: this line could be moved to a more narrow scope, inside `if (!representative.decode_account (representative_text))` Other than that, LGTM"
+    },
+    {
+      "prediction": "I think we should rename this to `Active` or `Present` to be consistent with the other variables.",
+      "reference": "when is this consumed? I'm probably missing it but I can't find it right now"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we need to make sure we don't generate the wrong cache key for a rollup. I think we need to make sure we don't generate the wrong cache key for a rollup. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need to make sure we don't generate the wrong cache key for a groupby. I think we need",
+      "reference": "I think we also need to add the rollupGroupExprList() to the cache key. If we rebuild the list above from a ValueIdSet on line 418 above, it is probably going to be in the same order, regardless whether it was ROLLUP(a,b) or ROLLUP(b,a)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `com.google.api.tools.framework.model.Field`",
+      "reference": "can use `getParamTypeName` instead"
+    },
+    {
+      "prediction": "I think this should be `&& $this->request->query->has('property') && $this->request->query->get('property') !== ''`",
+      "reference": "Why send the parameters? They're accessible directly from `$this->request` so there's no need to inject them in the method"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `QWebPage`.",
+      "reference": "Tiny nit: I would prefer `not selected_file` or `len(selected_file) == 0`, as if choose_file starts returning, for example, tuples instead of lists, this won't break."
+    },
+    {
+      "prediction": "Why do we need to remove the name file if a duplicate node name is used?",
+      "reference": "How will this code behave with hostnames that contain hyphens?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add this column. I think it should be in the `SmallMoleculeTransitionListColumn` class.",
+      "reference": "I think this is redundant since \"NullForEmpty\" already calls \"Trim()\"."
+    },
+    {
+      "prediction": "Should this be a `CloudBackupRestoreRequest` instead?",
+      "reference": "Change this to IDs too?"
+    },
+    {
+      "prediction": "I think this should be `ImmutableSet.of(\"SCPFCCSPCF\", \"CSPFCCCSSF\", \"SCFPCSCPCF\", \"PCSFCPCSCF\", \"SCCFCPSCF\", \"SCPFCCCSSF\", \"PCSFCPCSCF\")`",
+      "reference": "private static final, and it should be located in raidsoverlay, as it's not needed in the plugin. also the name could be a bit better, DC_SCOUT_RAIDS or similiar."
+    },
+    {
+      "prediction": "I think this should be 126, not 127.",
+      "reference": "attr_value is 128 bytes?"
+    },
+    {
+      "prediction": "I think this should be `self.method = self.config.get(\"method\", \"GET\") or self.method.upper()`",
+      "reference": "if it's unpredictable - why uppercase it at all?"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "Please remove all these debug prints. It doesn't look good."
+    },
+    {
+      "prediction": "I think this should be called `SampleRandomness` and not `SampleChainRandomness`",
+      "reference": "Consumers of this interface should not be required to provide `sampleHeight`. This should be an expected consensus parameter."
+    },
+    {
+      "prediction": "I think you can remove the `if closeTime.Before(startTime)` here.",
+      "reference": "I don't believe this is the right fix. If startTime comes after closeTime, then it means some other event triggered invocation of this code path like workflow timeout. In this case we should still try to fire the cron on previous value so we should just return start the delta between startTime and closeTime immediately."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should go in `cloud/backup` since it's a specific type of backup request.",
+      "reference": "Not sure if we need tag here, it is an implementation detail in portworx that isn't exposed in openstorage"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `SolrFields` class, since that's where the `SolrField` interface is defined.",
+      "reference": "I wonder if it would be cleaner to refactor all of this similar to the SolrMarc driver, so we have a property with Summon field names associated with types, and we iterate through it in a loop... that way we don't have to repeat the loop code four times with different variable names."
+    },
+    {
+      "prediction": "I think this should be `getCoreContainer().getSolrClientCache()` instead of `getCore().getCoreContainer().getSolrClientCache()`",
+      "reference": "Need the `solrClientCache.close();` further down in the method be removed since a shared cache is now used?"
+    },
+    {
+      "prediction": "I think this should be a `uint64` instead of an `int64`.",
+      "reference": "according to the PR (schema changes?) these 2 field should be removed?"
+    },
+    {
+      "prediction": "Can we use the same name for the container name?",
+      "reference": "Panic here if `exp.Target.StressChaos.CintainerName` is nil."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the scorer iterator itself.",
+      "reference": "This allows for some hacks like returning an iterator that matches more docs than the scorer. I liked the previous approach that returned an iterator better."
+    },
+    {
+      "prediction": "nit: remove the `TODO`",
+      "reference": "Aren't there already logs for this from the underlying file system implementation?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this change.",
+      "reference": "It would be good to have a test in here that demonstrates the new munge functionality; I can help set that up if you're not sure how."
+    },
+    {
+      "prediction": "I think this should be `max_retries_to_get_pods` instead of `max_retrials_to_get_pods`",
+      "reference": "`retrials` isn't the word you intend. I think you mean `retries` and to that end, can the var name just be `max_retries`?"
+    },
+    {
+      "prediction": "I think we should add this help message to `ProductMetaTitle` as well.",
+      "reference": "Value in product should contain 255 characters or less."
+    },
+    {
+      "prediction": "I think we can remove the `config` variable and use `configs[i].Consensus.RollDPoS.Delay` and `configs[i].Consensus.RollDPoS.ProposerInterval` directly.",
+      "reference": "line is 133 characters (from `lll`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be in `custom_options_file?` and not in `file_options`.",
+      "reference": "since we're calling the file .rspec-local, I think we should rename local_options to project_options and use local_options for .rspec-local - WDYT?"
+    },
+    {
+      "prediction": "I think this is the only place where we're using `connection_config.merge(adapter: connection_config[:http_adapter])`",
+      "reference": "Will we want to refactor this when we drop rsolr 1.x support?"
+    },
+    {
+      "prediction": "I think we should rename this function to `getNodeNamedCrtFromIP` or `getNodeNamedCrtFromIPString` to be consistent with `getNodeNamedCrtFromNode` and `getNodeNamedCrtFromNodeFile`.",
+      "reference": "should this really be multiple IPs?"
+    },
+    {
+      "prediction": "I think this should be a `render_to_string` instead of a `render partial: \"proposal\"`.",
+      "reference": "Hmm, is that partial used in multiple places? Maybe we can just move that file in here."
+    },
+    {
+      "prediction": "Nit: I think we can remove this file since we don't need it anymore.",
+      "reference": "Should be 2014-2016"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "Looks like my fault for sure in trying to handle international inputs regardless of locale. I think the correct fix is actually to get rid of TrySplitColumns and use ParseDsvFields instead. It is what gets used in the end, and it already has logic for dealing with quoted fields."
+    },
+    {
+      "prediction": "I think this should be `done().then(cursors[0].close);`",
+      "reference": "similarly here, are the these cursors not being closed when the client is closed?"
+    },
+    {
+      "prediction": "I think this should be `map[string]Policy` instead of `map[string]func(string) Policy`",
+      "reference": "Or maybe `func(args ...string) Policy` to make it dynamic in case of any future policy that may require more than one args."
+    },
+    {
+      "prediction": "I think this should be done after the journal server has been created. If journal server is not created, then we won't be able to create the journal server.",
+      "reference": "Should this happen even if the above failed?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this file.",
+      "reference": "Please dont change this ."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to log the confirmed blocks. The confirmed blocks should be logged in the `confirmed` field of the `Election` object.",
+      "reference": "There is a strange $b that looks wrong."
+    },
+    {
+      "prediction": "I think this should be `containerContainsText = _(\"with %s items\") % childControlCount`",
+      "reference": "Is it still important to have `speakEntry` here? It is checked before this value `containerContainsText` is used in the \"General\" section."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `Travis::Poller` class.",
+      "reference": "It is `$JAVA_HOME/bin`, not `$JAVA_HOME`, which should be added."
+    },
+    {
+      "prediction": "This should be `#include <stdio.h>` instead of both `#include <string.h>` and `#include <stdio.h>`.",
+      "reference": "Should move this to within the `#else` block as it's only used there."
+    },
+    {
+      "prediction": "Shouldn't we wait for the page to be loaded?",
+      "reference": "is this a common rule to wait one extra second after ajax? I'm thinking about moving this extra wait into waitForAjax method."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it. I think it would be better to have the `logger` object be a property of the `SolrTask` class, so that the `logger` object can be passed into the constructor of the `SolrTask` class. This way, the `logger` object can be passed into the constructor of the `SolrTask` class without having to pass the `logger` object into the constructor of the `SolrTask` class.",
+      "reference": "Maybe you can correct the typo \"if\" too :)"
+    },
+    {
+      "prediction": "I think this should be `IdentifyForChat(ctx, uid string, assertion, reason string) (UserInfo, *keybase1.IdentifyTrackBreaks, error)`",
+      "reference": "I'd prefer a name less chat-specific, since later on we could have other app types that want the same behavior. Maybe `IdentifyAndAllowTrackBreaks`?"
+    },
+    {
+      "prediction": "I think this should be `MPI_COMM_WORLD` instead of `MPIMPIComm_Get`",
+      "reference": "I think I was involved in writing that caster, but I never fully understood the `import_mpi4py` thing. Why is importing mpi4py still necessary at that point? I would think if the user is passing a communicator from python code, they must already have imported mpi4py themselves, or does that not propagate through into the wrapping layer?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should just close the body here.",
+      "reference": "This seems like a more interesting change. Maybe we should re-name the PR"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this feature interface. I think it would be better to move it to the TransportConnection interface.",
+      "reference": "Nit: implement the methods for these interfaces explicitly in this file for consistency. It can be passthrough."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Not directly related to your PR, but I think here we should create a Type enum (MAP, ARRAY, PRIMITIVE) instead of having 3 flags."
+    },
+    {
+      "prediction": "I think it is better to use `std::this_thread::get_id()` instead of `Lng32`",
+      "reference": "Do you want this to be a stack variable? The work method might return at any time (example: when the parent queue fills up) and then get called again, which would cause this variable to become false again. I'm wondering if this variable should be a member instead?"
+    },
+    {
+      "prediction": "I think it would be better to use `reinterpret_cast<GraphSpaceID>(spaceId)` instead of `static_cast<GraphSpaceID>(spaceId)`",
+      "reference": "You could reserve a const space to save the extra computation."
+    },
+    {
+      "prediction": "I think this should be `\"/home/user/runtime/src/coreclr/runtime/src/libraries/System.Private.CoreLib/src/System/IO/..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//..//",
+      "reference": "Did you consider adding rooted paths like `C:\\ProgramData` and paths with flipped separators?"
+    },
+    {
+      "prediction": "Why do we need this defer?",
+      "reference": "does this line cause a panic if it is above the if condition ?"
+    },
+    {
+      "prediction": "I think it would be better to have this enum as a separate file. This way it's easier to see what's going on under the hood.",
+      "reference": "Maybe `LibraryDetailsPanelType` would be clearer."
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "minor nit - add \"persist-filesystem\" and \"persist-directory\" into commit message, as its something people may search on (maybe applies to a few other commit messages)"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "I verified offline with @jeremyg-lunarg this should be removed."
+    },
+    {
+      "prediction": "I think this should be `fmt.Println(\"incldued_wf name:\", i.Workflow.Name, \" timeout:\", i.Workflow.DefaultTimeout)`",
+      "reference": "This should be removed? If not, typo in incldued_wf"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could get it to work.",
+      "reference": "Sorry for the delay on this! It looks like Circle CI isn't running tests on some of the PRs, but this doesn't look like it would work   I might be missing something, but are the tests all passing locally?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/caddy/cmd/match/file.go`",
+      "reference": "Do you think the godoc should mention that all delimiters will be suffixed with `/`?"
+    },
+    {
+      "prediction": "I think this should be `aURL.startsWith(\"data:\") && SessionStore.get().isCurrentSessionPrivate() && SessionStore.PRIVATE_BROWSING_URI`",
+      "reference": "what if I want to load my own, a different, data URI in Private Browsing mode? I do this often on desktop (e.g., `data:text/html,&times;`). admittedly, I wouldn't expect this to be done by a non-developer, but this will certainly cause a minor bug."
+    },
+    {
+      "prediction": "I think this should be `long.MIN_VALUE` instead of `0L`.",
+      "reference": "I remember the sequence number 0 is kept for the data files for iceberg v1, so in theory the sequence number from delete files should start from 1. So setting it to 0 as the default value sounds correct."
+    },
+    {
+      "prediction": "I think this should be `unsigned integer` instead of `integer`",
+      "reference": "For all these MYSQL queries unsigned is required. I have tested again and without unsigned these queries are failing."
+    },
+    {
+      "prediction": "I think this should be `MAX_RECENT_HISTORY`, not `MAX_HISTORY`. Also, I think it would be better to call it something other than `stateUtils` since it's not really the responsibility of this file to own it. Maybe `stateCache` or `stateCache_` or something like that?",
+      "reference": "Are there tests for this? What happens when you get to limit?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in `data['config']['transport'] = 'remote' if target.name == 'localhost'`",
+      "reference": "Whole thing could be collapsed to `data['config']['transport'] ||= 'local' if target.name == 'localhost'`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `OpenPGPKeyDialog` class.",
+      "reference": "maybe throw in `&& mServiceConnection.isBound()`, there are cases where the connection might die"
+    },
+    {
+      "prediction": "This should be `IsEip1344Enabled => true;`",
+      "reference": "Can we add Istanbul with this first EIP please? (or as a separate issue)"
+    },
+    {
+      "prediction": "I think this should be `return usr.Name` instead of `return nil`",
+      "reference": "I don't understand this change"
+    },
+    {
+      "prediction": "This should be `self.showWarning(\"Invalid Public Key\")` instead of `self.show_warning(\"Invalid Public Key\")`",
+      "reference": "Let's allow translators for this string to be translated: `self.show_warning(_('Invalid Public key'))`"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "This needs to work with `role=form` too."
+    },
+    {
+      "prediction": "I think this should be `&& localConnection.type == 'value' && localConnection.outputType == 'output'`",
+      "reference": "This should be before the if on line 1028, which should turn into an else if"
+    },
+    {
+      "prediction": "I think we can remove the `PodSelectorMatchExp` and `NSSelectorMatchExp` fields from the `ACNPRuleAppliedToSpec` struct.",
+      "reference": "Not really introduced by this PR, but I don't recall why these have to be pointers, while `PodSelector` / `NSSelector` do not. Do you know the reason?"
+    },
+    {
+      "prediction": "I'm not sure this is correct. If the sendDelta is larger than the ackDelay, then we should be subtracting the ackDelay from the sendDelta, not adding the ackDelay to the sendDelta.",
+      "reference": "Is there a reason why we are ignoring the ackDelay if it would result in a value smaller than the min? Why not `max(sample - ackDelay, minRTT)`?"
+    },
+    {
+      "prediction": "I think we can remove the `SarifLogEqualityComparer` class and just use the `IEqualityComparer<SarifLog>` interface.",
+      "reference": "Here's an example of how the files in the `NotYetGenerated` directory drifted out of sync from the generated files. When we made the equality comparers internal, we neglected to fix this one."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `guidance_admins` policy, since it's only applicable to admin users.",
+      "reference": "nice. we should do this elsewhere too. We have a lot of repeated stuff in the policies"
+    },
+    {
+      "prediction": "I think this should be `None` instead of `(let y = None; y))`.",
+      "reference": "Small formatting thing, but can you remove the space before the semicolon?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Not new, but nit: _writeLeng*th*."
+    },
+    {
+      "prediction": "I think this should be `CheckComposeDockerCompose` to match the name of the function.",
+      "reference": "Our habit is to go ahead and put a description line (or more) in front of every function, not just non-test or exported functions."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Remove this file from the PR"
+    },
+    {
+      "prediction": "I think this should be `newName` instead of `newName.replace(Blockly.Variables.CLOUD_PREFIX, '')`",
+      "reference": "I think that for cloud variables, the default value in the prompt should be the variable name without the cloud prefix; this is slightly different from the scratch 2.0 behavior, but I think would ultimately lead to less confusion. Proposing code changes below:"
+    },
+    {
+      "prediction": "I think this should be `wlr_pointer_init(wlr_pointer, libinput_device_config_tap_get_finger_count(libinput_dev));`",
+      "reference": "I think I would rather fish this interface through than use it as the basis for heuristics."
+    },
+    {
+      "prediction": "I think this should be reverted, as it's not related to this PR.",
+      "reference": "Probably not worth making this line-ending change, since it is the only change to this file."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `decode_opnd_memx0` instead of `decode_opnd_memx1`. The reason is that `decode_opnd_memx1` is only used in the `decode_opnd_memx0` function, which is called from `decode_opnd_memx1` and `decode_opnd_memx2`. If we move `decode_opnd_memx0` to `decode_opnd_memx1` and `decode_opnd_memx2`, then the `decode_opnd_memx1` and `decode_opnd_memx2` calls should be moved to `decode_opnd_memx0` instead of `decode_opnd_memx1` and `decode_opnd_memx2` respectively.",
+      "reference": "Is this the correct place for this assert? Does this routine rely on it being 64, or was this only for testing?"
+    },
+    {
+      "prediction": "I don't think we should expose this API. It's not exposed in the protocol, and it's not used by the rest of the system.",
+      "reference": "No pb please"
+    },
+    {
+      "prediction": "I think this should be `t.integer :subscriber_id, null: false` instead of `t.bigint :subscriber_id, null: false`",
+      "reference": "We need to add indexes here on both `subscriber_id` and `changeset_id` or things will quickly collapse as we build up subscribers ;-) What I would suggest is a unique index on `[:subscriber_id, :changeset_id]` which will also make duplicate entries impossible, and an ordinary index on `[:changeset_id]` for finding the subscribers to a changeset."
+    },
+    {
+      "prediction": "I think this should be `return err` instead of `return nil`",
+      "reference": "use the context versions of everything, e.g. `PostMessageContext` and thread it through the functions. will save you a bunch of refactoring trouble later on."
+    },
+    {
+      "prediction": "I think it would be better to keep this as it was before, and just add `Tuple[Type[\"Interface\"], ...]` to the docstring.",
+      "reference": "\"not to be used elsewhere other than\""
+    },
+    {
+      "prediction": "I think it would be better to have this as a `SparkParquetWriters` instance and pass it to the `ParquetWriters.buildWriter` method.",
+      "reference": "Thanks for the contribution, @coolderli ! I also think the newPosDeleteWriter need the properties setting ..."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "These additions seem unrelated to this change?"
+    },
+    {
+      "prediction": "I don't think we should have this in the table properties. I think it should be part of the engine configuration.",
+      "reference": "Could we move this to a class for Hadoop configuration properties, like `org.apache.iceberg.hadoop.ConfigProperties`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think we need to keep the deletion of the experiment related events.",
+      "reference": "The related statements should also be deleted."
+    },
+    {
+      "prediction": "I think this should be `self.build_list_params(params, policies, 'PolicyNames.member.%d', 'PolicyName.member.%d')`",
+      "reference": "Shouldn't this line be removed?"
+    },
+    {
+      "prediction": "I think this should be `set(['TEAM', 'PUBLIC'])`",
+      "reference": "Just move this into `const.py`. It already has similar stuff. Also, make it uppercase since it's a const."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the new mode. I think it would be better to set the new mode in the `setPasscode` method of `PasscodeActivity`.",
+      "reference": "Issue is that `currentMode = newMode;` is called at the end of the method. Passcode screen will come up, but since the activity thinks current mode is biometric check we don't check passcode when submitted."
+    },
+    {
+      "prediction": "Why 0.2 instead of 1.0?",
+      "reference": "Need change it back?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. I'm not saying you have to do anything about it, but I think it's a good idea to make it public.",
+      "reference": "Didn't we delete this file a while back because it isn't used but people think that it's a security vulnerability that it's checked in?"
+    },
+    {
+      "prediction": "I think this should be `using Datadog.Trace.Extensions;`",
+      "reference": "nit: Looks like this can be removed now"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `Provision` function in `pkg/caddy/provision.go`",
+      "reference": "Can probably remove the lines above it too (L60-62), since ProvisionDefaultCA() makes sure the map isn't nil."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to do this in a way that is consistent with the rest of the config system.",
+      "reference": "Do you think it's worth erroring out? It's likely that this is a configuration error that the user might want to know about?"
+    },
+    {
+      "prediction": "I think this should be `# FIXME: find out why this is broken`",
+      "reference": "I think the functional test should not change here, this is probably a test in itself :)"
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "Why this to be a public function?"
+    },
+    {
+      "prediction": "I'm not sure why this is called `timing` but it's the only option I can think of that makes sense.",
+      "reference": "I think we can remove the timing for export env vars"
+    },
+    {
+      "prediction": "Why do we need to remove this?",
+      "reference": "wrong import grouping"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that the `PreferenceController` uses the `PreferenceActivity`'s `getResources().getStringArray` method to get the string array, and the `PreferenceActivity`'s `getStringArray` method returns a copy of the same string array. So the `PreferenceActivity`'s `getResources().getStringArray` method will always return the same string array, no matter what the `PreferenceController` uses to get the string array.",
+      "reference": "There is a doubled semicolon ;)"
+    },
+    {
+      "prediction": "Shouldn't this be `#include \"openssl/sha.h\"`?",
+      "reference": "I'm wondering why this isn't <openssl/md5.h>. Seems like one would have to copy the md5.h file into the source tree somewhere for this to compile cleanly. Maybe you meant to use angle brackets instead of quotes?"
+    },
+    {
+      "prediction": "I don't think this is a good name for this class. It's not just checking that an element is visible, it's checking that an element is visible and interactable. I think something like `expect_any_element_is_interactable` would be better.",
+      "reference": "shouldn't **call** return a boolean?"
+    },
+    {
+      "prediction": "This should be `static_cast<uint16>(casting_spell_slot)`, not `static_cast<uint16>(-1)`.",
+      "reference": "We don't need to cast to client here. (well, from spell gem not AA etc) We should also make sure the charm is a casted spell before calling SendSpellBarEnable."
+    },
+    {
+      "prediction": "I think this should be a `DateTime(timezone=True)`.",
+      "reference": "This is not an `id`. This should either be `invocation_time` or `invocated_at` to be consistent with what we are using elsewhere, and also to better rerflect the column's DateTime type. Also, within the context of this table, there is no idea of what `invocation` is. I know that `scanner` may not be future-proof, but we need to find something more descriptive. Perhaps one possibility is to call this `audit_time`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `types.go` since it's only used there.",
+      "reference": "should the 'Receiver' be the clawback address in case of clawback transaction?"
+    },
+    {
+      "prediction": "I think this should be `controllerStatusOk` and `volumeStatusOk` should be `controllerStatus` and `volumeStatus` respectively.",
+      "reference": "@ashishranjan738 -- Where are these consts used? in which pkg?"
+    },
+    {
+      "prediction": "I don't think this should be in the autodoc docs.",
+      "reference": "The only change I would suggest is moving these lines back below the `templates_path` variable to keep the diffs smaller."
+    },
+    {
+      "prediction": "I think this is the only place where `Info` is used, so I think it's fine to remove this line.",
+      "reference": "Having it above kept the block info in the cached changes, which could end up confusing things quite a bit."
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I'm pretty sure it's because we're now using the `execute_task` method to run the task.",
+      "reference": "If the execute_task returns an `ExecutionResult`, the passing of a block would be unnecessary. Instead, the `ExecutionResult` could be presented to the user in some standardized way (iterating over the result of each node, etc.)."
+    },
+    {
+      "prediction": "I think this should be `defer c.bridge.Connect(maxRetryForOFSwitch, make(chan struct{}))` instead of `if err := c.bridge.Connect(maxRetryForOFSwitch); err != nil {`",
+      "reference": "Not sure why we make a new channel here. If we don't use this channel, how about creating it inside the `Connect`?"
+    },
+    {
+      "prediction": "I think it would be better to use `raft::raftctl` to set the `lastLogId` instead of using the `raftLock_` here.",
+      "reference": "There are two question here: 1. If you only modify lastLogId, but the `iter` is not modified, I really don't know whether we can succeed. 2. We will check the if the term has changed in `replicateLogs` (if rollback really happen, the term must have been updated), so I really doubt whether the problem should be fixed in this way. Or you could describe it in more details."
+    },
+    {
+      "prediction": "Why do we need to do this?",
+      "reference": "Hello @jDolba, I have reviewed your PR and I found one problem. `$step->getFormType()` can return `FormTypeInterface`. You cannot use interface as key for an array. Can you find some better way to fix this? Thank you."
+    },
+    {
+      "prediction": "Can you add a newline at the end of this line?",
+      "reference": "i was confused by keyword \"inventory\" here. my understanding is this starts the osconfig-agent which covers inventory lookup and package-management(correct me if i am wrong)."
+    },
+    {
+      "prediction": "I don't think we need to add any new capabilities here. We already have them in the existing capabilities.",
+      "reference": "These are really meant to be the keys in the capabilities, not the keys of values within the capabilities"
+    },
+    {
+      "prediction": "Nit: `import * as sinon from 'sinon';`",
+      "reference": "This breaks tests on IE because this will import an `esm` bundle. For that reason `sinon` is available as a global in our test suite and never imported. The global is aliased to the proper `es5` file."
+    },
+    {
+      "prediction": "I think this should be `maxRTOTimeout` instead of `maxRTOTime`",
+      "reference": "Maybe move all the Loss Recovery constants to a separate file."
+    },
+    {
+      "prediction": "I think this should be `diskName`, not `diskName string`.",
+      "reference": "this field is added so that we can find disk name by device name from attachments"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Unnecessary new line"
+    },
+    {
+      "prediction": "I think this should be `if _, ok := state.NamespacePaths[nsType]; !ok {` and not `if _, ok := state.NamespacePaths[nsType]; ok {`",
+      "reference": "First determine if the namespace already exists, so it's clearer that it's handling namespaces that aren't included in the `c.config.Namespaces`"
+    },
+    {
+      "prediction": "I think we can remove the `class=\"assets-wrapper\"` and `class=\"videos-download-links\"` classes from the HTML as they are no longer needed.",
+      "reference": "What changed in the styles that means we don't need this?"
+    },
+    {
+      "prediction": "I think this should be `JObject?` instead of `JObject` since it can be null.",
+      "reference": "This pattern should be applied to the other instance where NewtonsoftEscapeJProperty is used and you can remove the NewtonsoftEscapeJProperty method."
+    },
+    {
+      "prediction": "I don't think we should be deprecating this. I think we should just warn the user.",
+      "reference": "Something to mention in the next changelog/release notes. It will be good to get histogram working consistently with everything else."
+    },
+    {
+      "prediction": "I think this should be `{{ .Volume.runNamespace }}` and not `[{{.Volume.runNamespace}}]`",
+      "reference": "What does this contain? Is it PVC namespace or is it openebs?"
+    },
+    {
+      "prediction": "I don't think we need this interface at all. We can just use `IRequestBodyMinimumDataRateFeature` directly.",
+      "reference": "Design note: Using local concrete data types on a feature abstraction makes it hard to generalize / replace. This is tolerable so long as this remains a kestrel exclusive feature, but we'd need a different design if this were ever moved to HttpAbstractions."
+    },
+    {
+      "prediction": "`call_state` is not used in this function, please remove it.",
+      "reference": "this will break R package, you should update the R files accordingly. BTW, the call state should be in the last argument."
+    },
+    {
+      "prediction": "I think this should be `array_merge($formats, array_keys($this->mediaPool->getFormatNamesByContext($mediaContext)))`",
+      "reference": "Shouldn't we throw an exception if the context is null instead ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the minimum version required by ddev, but I think it should be at least `1.10.0`.",
+      "reference": "These should both be const, not var right?"
+    },
+    {
+      "prediction": "I think it's better to put `build_match_cost` in `registry.py` and `build_from_cfg` in `build.py`.",
+      "reference": "Rename the directory, match_costers -> match_costs"
+    },
+    {
+      "prediction": "I think this test should be moved to `models_test.py` instead of `delete_model_test.py` since it's not related to the delete model functionality.",
+      "reference": "I don't think this should go in the model tests, since it is done in the resource. `PaginatedDeleteTest` seems more appropriate"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "It would be better to just change RWD to throw IllegalStateException if you attempt to execute a command after quit (unless it's a second call to quit())"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to the `prometheus` package.",
+      "reference": "Could we move this to an example test to allow the compiler to help us ensure it stays up-to-date in the future?"
+    },
+    {
+      "prediction": "I think we can remove this test now that we have `expect(cli.parse).to include(password: 'opensesame')` in the previous spec.",
+      "reference": "Maybe we could update these to use $future and then check on stderr? that way when we deprecate stdout we can not have to delete tests."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the `MinAllowedMaxFrameSize` constant and use it for both `MinAllowedMaxFrameSize` and `(_maxFrameSize * 2) + HeaderLength`.",
+      "reference": "Remove this since it always has to be set afterwards."
+    },
+    {
+      "prediction": "I think this should be `blockChangeSize(20*1024*1024)`, not `blockChangeSize(20)`",
+      "reference": "These two tests were causing us to use too many goroutines with `-race` using the default block change size, I think due to prefetching."
+    },
+    {
+      "prediction": "I don't think this is a good idea. The `requestContext` is used by other methods in this class. If we pass it here, it will be used by all the other methods in this class. I think it would be better to move the `requestContext` into the `fetch` method and pass it as an argument to that method.",
+      "reference": "Is that `try..except` block really needed? I mean, when it could fail?"
+    },
+    {
+      "prediction": "I think this should be `suspicious` instead of `suspense`",
+      "reference": "If you only want to run a specific group of tests you can always use `.only` like `it.only()` or `describe.only()`. It may not speed up the globbing process as much as this change here :)"
+    },
+    {
+      "prediction": "I think this should be `DEFAULT_BATCH_SIZE = 100`",
+      "reference": "nit: I'd probably rename this to `BATCH_SIZE_RANGES`, to use a slightly more descriptive name."
+    },
+    {
+      "prediction": "I think this should be 100, not 90.",
+      "reference": "aren't these the wrong way around?"
+    },
+    {
+      "prediction": "I think this should be `getContainerConfigService`",
+      "reference": "Lets call this getConfigSetService and maybe rename the field now (or later)"
+    },
+    {
+      "prediction": "I think we can remove `klogr` now, since it's not used anymore.",
+      "reference": "Could / should we make these constants and give them a type?"
+    },
+    {
+      "prediction": "I think it would be better to have a `platforms` map instead of a list of strings.",
+      "reference": "lets stick with lowercase for consistency."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "ordering of imports went wrong"
+    },
+    {
+      "prediction": "I think this should be `TableType.ALL_DATA_FILES` instead of `String.valueOf(MetadataTableType.ALL_DATA_FILES);`",
+      "reference": "I think these can all be .name() to get the value we are looking for"
+    },
+    {
+      "prediction": "I think this test should be moved to a new file `FunctionCallTest::testPrintOnlyStringInvalidArgument()`.",
+      "reference": "Is this a good place for these test cases?"
+    },
+    {
+      "prediction": "I think this should be `print(response_name, attr_name)`.",
+      "reference": "Did you mean to leave the print statement?"
+    },
+    {
+      "prediction": "I think this should be `500000000` and not `500000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "I would do `new BN(Interpreter.LOCKTIME_THRESHOLD)`"
+    },
+    {
+      "prediction": "I think this class is not needed anymore.",
+      "reference": "@ftomassetti I turned the tree walking algorithms into iterators so you don't have to visit every node when you only wanted a few, like for a `findFirst`."
+    },
+    {
+      "prediction": "I think we can remove the `cadenceInternal:Timeout` prefix here.",
+      "reference": "Need to add type name here to be compatible with proto string representation."
+    },
+    {
+      "prediction": "This should be `log.info`, not `log.finest`.",
+      "reference": "This is to let people know that the tooling is doing something during a build. Please leave."
+    },
+    {
+      "prediction": "I think this should be `aaedca130273574dd2bd6c3acad57f3545b662a974fa4320236f25fe474676d`",
+      "reference": "Why is the principal changing here?"
+    },
+    {
+      "prediction": "I think this should be `blob.core.*` instead of `blob.*`",
+      "reference": "Naming nit: let's use `AZURE_STORAGE_` prefix for consistency with the other ones."
+    },
+    {
+      "prediction": "I think we can remove this `install` and `uninstall` methods.",
+      "reference": "I'm a bit confused. We have an actual `ProgressBar` Uppy React wrapper component that we are testing. Why do we need a mock for it?"
+    },
+    {
+      "prediction": "I think we can remove this file, it's not used anymore.",
+      "reference": "can we use `github.com/pkg/errors`?"
+    },
+    {
+      "prediction": "I think this should be `else if actualType.Kind() == semantic.Var {`",
+      "reference": "I changed the name of the arguments here just to help make the code clearer."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `ops/dynamic_map.py`, since that's where we want to memoize the return value.",
+      "reference": "I think either 'avoiding calls to the function' or 'to avoid calling the function ...' would be read better."
+    },
+    {
+      "prediction": "I think this should be `const std::vector<std::unique_ptr<Attachment>>&` instead of `const std::vector<unique_ptr<Attachment>>&`.",
+      "reference": "@sfricke-samsung -- thanks. saves a few atomic ops which is always good :) The TODO is there to track that this object is storing *both* vectors of shared_ptr and plain pointers to the same data. The shared are to ensure scope, and the plain are for backwards compatibility with existing code that consumed a plain pointer vector (and some code that still does). That should be cleaned up."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `Nethermind.Monitoring.Metrics`",
+      "reference": "minor: typo (knowMetricsTypes -> knownMetricsTypes)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add this check. I think it would be better to add this check in `QRDescGenerator::typeSupported()` instead.",
+      "reference": "This adds a case-insensitive flag to the type in the RangeSpec, but I don't think RangeSpecs are written to handle case-insensitive comparisons. Take a look at the methods that deal with comparisons when building RangeSpecs, in file Range.cpp. So, I think you would have to do one of two things: a) disable the RangeSpec transformation for case-insensitive comparison operators (the easy way) or b) change the RangeSpec methods to handle case-insensitive comparisons."
+    },
+    {
+      "prediction": "I think this should be `public static SymbolReference<ResolvedMethodDeclaration> solveMethodInType(SymbolReference<ResolvedReferenceTypeDeclaration> typeRef, String name, List<ResolvedType> argumentsTypes, boolean staticOnly, TypeSolver typeSolver)`",
+      "reference": "I would throw an exception if the type is not solved"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "if remove `dt.skipLoad` - then this tests will run. You probably mean opposite?"
+    },
+    {
+      "prediction": "Style/MutableConstant: Freeze mutable objects assigned to constants.",
+      "reference": "I feel like we do this in a lot of places should we move that out into its own PR?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be moved to `blurhash.js`",
+      "reference": "Already asked you in Matrix, but I'll do it here again so it can be discussed publicly. Why?"
+    },
+    {
+      "prediction": "This should be indented by 4 spaces, not 2.",
+      "reference": "I think here should be a comma before remove, in all the cases"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You need to check if `stages` is set in `context` before calling `self.bbox_roi_extractor = context.bbox_roi_extractor[stages]` and `self.bbox_head = context.bbox_head[stages]`",
+      "reference": "Single quote is used by default in this project."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think we should be able to do it in the same way as we do for pods.",
+      "reference": "S1023: redundant `return` statement (from `gosimple`)"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think we need to make sure that the webhook is closed before we close the context.",
+      "reference": "`ctx` is unused in Close"
+    },
+    {
+      "prediction": "Why not use the parent class instead of creating a new one?",
+      "reference": "I'm confused. This exception is thrown when `HeurekaCategory` is not found in database. But this exception extends Http exception. Why? What has database search common with http? If the only reason is that it is the same in the whole project, then fine. But then we have even bigger problem - we don't know how to use exceptions."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this model. I think it should be in `models/`",
+      "reference": "Is it necessary to add the suffix `on_head`, because it belongs to` bbox_head`?"
+    },
+    {
+      "prediction": "This should be `AC_DEFINE([WITH_MODERN_GRUB], 1, [Define if we have a GRUB version newer than 2.02])`",
+      "reference": "Hmm, I wonder if this should just be e.g. `--with-grub-2.02` instead. (Not sure Autoconf supports periods in these switches.) Today's modern GRUB is tomorrow's ancient GRUB. :) Or maybe we should be specific about the feature this is enabling, which might be safer given that each distro carries so many patches. E.g. `--without-grub-efi-16-suffixes`?"
+    },
+    {
+      "prediction": "I think this should be `Client` instead of `daisyCompute.Client`",
+      "reference": "Add a GetSerialPortOutput method to Instance that way you don't need to access the client, also it makes the call cleaner as you don't have the odd i.Client and path.Base(i.Project), path.Base(i.Zone)"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I can't think of a better way to do this.",
+      "reference": "Line is too long. [104/80]"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it would be better to move it to `testutil/testutil.go` since it's only used by `testutil/testutil.go` and it's not clear why it should be here.",
+      "reference": "Since CD updates (assignment & power state) are now done in this controller..."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it should be in the `Validate` schema option.",
+      "reference": "Hmm I'd rather not support this option going forward - it seems like a one-off just to work around this particular issue. Would it be fine to just remove the `deepStackTrace` option? The rest of the PR looks great - I love the idea of switching to sets."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this class. I think it should be `CommentCreator` since it's the main class that creates the `Comment` objects, but I'm not sure if there's a better name.",
+      "reference": "on the one hand I like how little this is doing. On the other hand, `inbound_mail_parser` is doing most of the work here so maybe it's not as helpful as I originally thought"
+    },
+    {
+      "prediction": "I think we can remove the `eprosima::fastrtps::rtps::RTPSWriterPublishMode::ASYNCHRONOUS_WRITER` instantiation here.",
+      "reference": "This line is too long"
+    },
+    {
+      "prediction": "I think this should be `fmt.Errorf`",
+      "reference": "This isn't a fatal error, we don't want to crash just because we can't set configs, we have sane defaults set"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "I wondered why `buffer` was passed into `LookupPath`. What is happening here? Is `buffer[:0]` equivalent to `[]V4TrieEntry{}`, and hence `buffer` isn't needed any more?"
+    },
+    {
+      "prediction": "I think this should be `:href => user_url(@user_message_author, :host => SERVER_URL)` instead of `:target => \"_blank\"`",
+      "reference": "I don't really like the name of this file, especially the encoding of an implementation detail (the fact that it's a table) in the name. Can we just use `_message_body.html.erb` instead maybe?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `security.openshift.io` or `security.openshift.io/v1beta1`",
+      "reference": "Feel better not to expose our internal implementation in API/CRD, so no need to mention agent here. Consider: EnableLogging indicates whether or not to generate logs when rules are matched. Default to false."
+    },
+    {
+      "prediction": "I think this should be `status = db->Put(writeOptions, NebulaKeyUtils::dataVersionKey(), \"3.0\");`",
+      "reference": "use `NebulaKeyUtilsV3::dataVersionValue()` to replace the literal value?"
+    },
+    {
+      "prediction": "I think this should be `log.url.debug(\"URL contains characters which are not present in the current locale\")`",
+      "reference": "No need for the `\\` here, inside parentheses whitespace is ignored by Python."
+    },
+    {
+      "prediction": "I think this test is failing because it doesn't have a valid certificate.",
+      "reference": "This doesn't seem like the right expected result... it should be the same as the `clientCert` var defined above."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "need to rename this now."
+    },
+    {
+      "prediction": "I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure why this is the case, but it seems to be the case. I'm not sure",
+      "reference": "NIT: This shouldn't be a TODO. It is just a note isn't it?"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this, but I don't have a better suggestion.",
+      "reference": "I changed this to an error, that seems more appropriate than a warning here."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think we should be able to do something like: `if !l.IsEvmTransfer() && !l.IsEvmSend() && !l.IsEvmRefund() && !l.IsEthTransfer() && !l.IsEthSend() && !l.IsEthRefund() && !l.IsStakingTransfer() && !l.IsStakingSend() && !l.IsStakingRefund() && !l.IsInitiatorTransfer() && !l.IsInitiatorSend() && !l.IsInitiatorRefund() && !l.IsStakingTransfer() && !l.IsStakingSend() && !l.IsStakingRefund() && !l.IsInitiatorTransfer() && !l.IsInitiatorSend() && !l.IsInitiatorRefund() && !l.IsStakingTransfer() && !l.IsStakingSend() && !l.IsStakingRefund() && !l.IsInitiatorTransfer() && !l.IsInitiatorSend() && !l.IsInitiatorRefund() && !l.IsStakingTransfer() && !l.IsStakingSend() && !l.IsStakingRefund() && !l.IsInitiatorTransfer()",
+      "reference": "This change could mute unit test failure, but it is better to update unit tests"
+    },
+    {
+      "prediction": "Shouldn't this be `holder.notificationId` instead of `notificationId`?",
+      "reference": "All other places that access `notificationIdsInUse` are in methods with descriptive names. We should do the same here. Maybe `markNotificationIdAsFree()`?"
+    },
+    {
+      "prediction": "I think we can remove the `ErrNotFound` and `NotFoundErr` and just use `errors.New(\"tag not found\")` here.",
+      "reference": "keep as `ErrNotFound`, it is the same convention as in other packages"
+    },
+    {
+      "prediction": "`offline_mode` is not a valid api parameter. Please remove it.",
+      "reference": "Same here, we need to call `callback` in else branch"
+    },
+    {
+      "prediction": "I think this should be `certRecord.getHostName() != null ? certRecord.getHostName() : \"\";`",
+      "reference": "At this point we already check that the record is valid and so it could never be nulll"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the title. I think it would be better in the `before_action :set_default_title` in `UserController`",
+      "reference": "Would it not be more sensible just to move setting of the title to later in the method rather than duplicating it here?"
+    },
+    {
+      "prediction": "`uploadSession` is called by `MetaClient::HeartBeatThread` and `MetaClient::HeartBeatThread` is called by `MetaClient::loadUsersAndRoles`, so this is redundant.",
+      "reference": "why not do it in `reclaimExpiredSessions`? and `pushSessionToCache` can delete"
+    },
+    {
+      "prediction": "I don't think this is the right place for this model.",
+      "reference": "out_channels/num_outs/type are unnecessary"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be possible to get the length of an object's keys.",
+      "reference": "You can create an issue for it and put it on the backlog so we don't forget it."
+    },
+    {
+      "prediction": "I think this is the only place where we're using the container's HostConfig to determine whether or not to start the container in the first place. I think we should be able to get rid of that.",
+      "reference": "So now we are checking twice, both here and in `weaveCIDRsFromConfig`. Not great. I suggest changing the `ok` return of `weaveCIDRsFromConfig` to a messsage (or error?) instead, which we can then log."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `install` method of the `Package` class, since that's where the `Package` object is created.",
+      "reference": "An alternative to setting format to the default in Package.__init__ would be to set it in create_package. I think we can assume all packages are created by create_package, but not necessarily by build_package."
+    },
+    {
+      "prediction": "Why do we need a random here?",
+      "reference": "Let's add fee recipient as well."
+    },
+    {
+      "prediction": "I'm not sure what this is for, but it seems like it should be called `self._filename` instead of `self.filename`.",
+      "reference": "No need for those with `super-init-not-called`, as pylint already only turns things off for this function and it's needed for the entire function."
+    },
+    {
+      "prediction": "I think this should be called `applicationName` and not `applicationName`",
+      "reference": "I think there's another spot for this in DefaultCapabilityMatcher"
+    },
+    {
+      "prediction": "I think this should be `I18n.t(\"decorators.steps.waiting_text\", step: step.decorate.waiting_text)`",
+      "reference": "this name seems pretty vague -- thoughts on a more descriptive method name?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `System.Net.Http.Headers` since it's related to HTTP headers.",
+      "reference": "what changed in this line?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to move it to the `AdminBar` component itself. The `AdminBar` component is the top-level component of the admin bar, and it should have access to the `Data` select component.",
+      "reference": "Instead of doing this let's add another function similar to `hasZeroData` for selecting the `reportArgs` since this is all sourced from selected values. Then `hasZeroData` can use this internally, as well as the component itself. Since this function would be internal just for the purpose of avoiding duplication, we don't need to expose it on the component the same way though."
+    },
+    {
+      "prediction": "Why do you need this?",
+      "reference": "This line was useful for auto-completion, I think it should be re-added with specifying the var name (`$exception`) and simplifying the FQCN."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I'm not sure we want to block waiting for the channel to be closed. I think we want to close the channel when we're done writing to the channel.",
+      "reference": "This looks like you're not collecting everything from the error channel."
+    },
+    {
+      "prediction": "I think we can remove `to: @checkout.user_email` as we don't have any plans for this user",
+      "reference": "This currently violates the Law of Demeter. Using `checkout.plan_name` is the quick resolution."
+    },
+    {
+      "prediction": "Why did you change the type to `string`?",
+      "reference": "is there any scenario when `$temporaryFilename` can be null?"
+    },
+    {
+      "prediction": "I think we can remove the `@SuppressWarnings(\"ReferenceEquality\")` here.",
+      "reference": "this is for the `boolean isRoot = root == rowType` check, which seems to be on purpose, but maybe you could double check whether using ref. equality here is still wanted? Same for `SparkTypeToType`"
+    },
+    {
+      "prediction": "This should be done in `roots_drag_icon_update_position()`.",
+      "reference": "No need to damage after `roots_drag_icon_update_position`, this is already done in `roots_drag_icon_update_position`"
+    },
+    {
+      "prediction": "`one_img` and `one_meta` could be replaced by `image` and `meta`?",
+      "reference": "Have you tested exporting to ONNX with `device=cuda`?"
+    },
+    {
+      "prediction": "I think this should be `if o.name != \"\" {`",
+      "reference": "Do we also need to validate `envName` flag then? `appName` and `envName` are used in `initLogsSvc` which are called by `svc logs` from within `Execute()`"
+    },
+    {
+      "prediction": "I think this should be `preg_quote($request->getHost(), '')` instead of `parse_url($url, PHP_URL_HOST)`",
+      "reference": "@LukeTowers I think I would prefer that we don't force it, on the basis that: a) some people would be opting to configure their web server to protect against this kind of attack and would disable this feature in October CMS to get a small performance increase. b) it might be a bit misleading to say that `app.trustedHosts` set to `false` disables the feature entirely, just to have it be overridden here regardless."
+    },
+    {
+      "prediction": "I think this should be `c.cnpInformer.Informer().GetIndexer().ByIndex(ClusterGroupIndex, cg.Name)`",
+      "reference": "i like that you are trying to correct the naming, but its making the PR longer :( i guess in future we can do such changes in separate PRs"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea to have a test for all browsers. I think it would be better to have a test for Chrome and Firefox, and another one for Safari.",
+      "reference": "Are we marking what the browser's accessibility tree says or what screen readers with those browsers say?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "This method should behave as intended, to block until it reads an EOF from the other side."
+    },
+    {
+      "prediction": "I think it would be better to use `log.Errorf` instead of `log.Warnf`.",
+      "reference": "I prefer starts with `Failed to xxx`, how about `Failed to perform health check on xxx`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be in the same namespace as the rest of the `AzureAppServices` tags.",
+      "reference": "All these calls make me think we should refactor how traces are started and allow \"source tags\" that we initialize a trace with. I'd like to do this in a follow up."
+    },
+    {
+      "prediction": "This file should be renamed to `v1_test.go`.",
+      "reference": "This file shouldn't be in this PR"
+    },
+    {
+      "prediction": "Why is this one skipped?",
+      "reference": "This now causes `LessSpecificReturnStatement - src/somefile.php:9:32 - The type 'non-empty-list<5|string>' is more general than the declared return type 'non-empty-list<string>' for takesList`, which seems correct to me. The type `non-empty-list<5|string>` contains the type `non-empty-list<string>`. Thoughts?"
+    },
+    {
+      "prediction": "I think this should be `LOG.trace(\"Storing private transaction in enclave\", privacyUserId, maybePrivacyGroup);`",
+      "reference": "Are you going to change that back?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a macro for this.",
+      "reference": "we have a standard format for putting these together elsewhere, it does the filter check first and errors out and then sets based on the value. i think this should be adjusted to do that. there's no logical change, just an approach change."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. It's not a function, it's a property.",
+      "reference": "Note this should be `delay` instead of `scale` here. I'll fix this in my branch."
+    },
+    {
+      "prediction": "I think we can remove this now, since it's not used anymore.",
+      "reference": "`on` likely still needs a bind, similar to `this.resetProgress = this.resetProgress.bind(this)` correct?"
+    },
+    {
+      "prediction": "Why do we need to import `get_dist_info` here?",
+      "reference": "Considering the trade-off between training speed and accuracy, multi-scale training is temporarily kept. More elegant implementation will be adopted in the future."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Can you move this up along with the other builtin libraries?"
+    },
+    {
+      "prediction": "This should be `self.allowUsageStatsCheckBox` not `self.allowUsageStatsCheckBox(wx.CheckBox(self, label=_(\"Allow the NVDA project to gather NVDA usage statistics\"))`",
+      "reference": "I actually think it makes sense to reposition this checkbox after the notifyForPendingUpdateCheckBox. The current order of check boxes is a bit arbitrary now."
+    },
+    {
+      "prediction": "I think this should be `required:\"true\"`, since it's not always possible to specify a `Changes` object.",
+      "reference": "Why do we have both `before` and `after`? Naively I would expect this to work as follows: `plz query changed` with no arguments compares the current working directory state to the last commit (i.e. HEAD, i.e. a noop when directly on a git commit). `plz query changed --since [reflike]` compare the current working directory state to the given reflike. So for CI I would expect to run something like `plz query changed --since @{upstream} (or origin/master I guess)`. And on merge I'd either run everything or just `plz query changed --since HEAD^` (given our workflow) - I don't think I'd ever want to do anything else. Either way I think before and after are slightly confusing terms here."
+    },
+    {
+      "prediction": "I think this should be `self.re.search(f.server_conn.ip_address)` instead of `self.re.search(f.server_conn.ip_address[0])`",
+      "reference": "Is there a reason why we can't use `~dst`? It feels like that could be good enough.I would like to avoid extending the filter syntax unless there's an urgent need. :)"
+    },
+    {
+      "prediction": "I think this should be `DateTimeOffset` and not `LocalDateTime`",
+      "reference": "Why local and not UTC?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to check if `diagArea_` is empty before checking if it contains `-2013` and `-2012`.",
+      "reference": "2012 is a retryable error. Will avoiding rgenerating it here cause a difference in behavior in createServer() ?"
+    },
+    {
+      "prediction": "I think this file should be renamed to `factory.go` and moved to `pkg/pki/factory`",
+      "reference": "move to line 14 below"
+    },
+    {
+      "prediction": "I think this should be `public boolean intersectsTriangle(Point3D a, Point3D b, Point3D c)`",
+      "reference": "the method name should match teh docs"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `makeRootMetadata` since it's the only place where we create a new root metadata.",
+      "reference": "Won't this happen automatically when we flip the default version in the config?"
+    },
+    {
+      "prediction": "I think this should be `assert.Contains(out, \"/var/www/html/docroot\")`",
+      "reference": "This would also change back to /var/www/html/docroot if we go that way."
+    },
+    {
+      "prediction": "I think this should be `virtualNode` instead of `node`",
+      "reference": "This still needs to use the new option. We'll need tests for it too."
+    },
+    {
+      "prediction": "I think this file should be in the `appModules/` subdirectory.",
+      "reference": "We are no longer including file name in the source files."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in `ApplicationController` since it's a common thing that all controllers should have access to.",
+      "reference": "no need for this. Devise provides us with `current_user` and `user_signed_in?` helpers."
+    },
+    {
+      "prediction": "nit: can we call this method `isEth66CompatibleWithVersion`?",
+      "reference": "Would `Object.equals(capability.getName(), NAME)` would be safer, as you wouldn't have to do null checks?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to reset mocks. I think it would be better to move this to `describe '#run'` or `describe '#test'` or `describe '#spec'` or `describe '#example'` or `describe '#group'` or `describe '#feature'` or `describe '#context'` or `describe '#setup'` or `describe '#unsetup'` or `describe '#before'` or `describe '#after'` or `describe '#before_context'` or `describe '#after_context'` or `describe '#before_feature'` or `describe '#after_feature'` or `describe '#before_global_context'` or `describe '#after_global_context'` or `describe '#before_each'` or `describe '#after_each'` or `describe 'before'` or `describe 'after'` or `describe 'context'` or `describe 'setup'` or `describe 'unsetup'` or `describe 'before'` or `describe 'after'` or `describe 'before_context'` or `describe 'after_context'` or `describe 'before_each'` or `describe 'after_each'` or `describe 'context'` or",
+      "reference": "Hmm, I wonder if we should revert #1862 instead? BTW, what failure do you get w/o this line?"
+    },
+    {
+      "prediction": "I think this should be `errors.Wrapf(err, \"invalid 'apiURL' %s in Slack config\", out.APIURL)`",
+      "reference": "since there are several places where we do 1) get secret key ref and 2) validate URL, maybe it's worth having a common method? it could also trim spaces as done here (but not at the other places currently)."
+    },
+    {
+      "prediction": "I think we can drop the `gaz` part of the name here.",
+      "reference": "TestNormalize... I would also split them"
+    },
+    {
+      "prediction": "I think this should be `index = indexGen.next()` instead of `index=next(indexGen)`",
+      "reference": "All changes to this file are going to conflict with #7599. Please revert these as well. They will be addressed during the Python 3 transition."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `l.m.RLock()` and `l.m.RUnlock()`.",
+      "reference": "Just thought, using spinlock (busy wait with compare and swap) that generates 2 random numbers inside rlock. We know that collisions are rare and we don't need that \"large\" write lock really but lock namely for \"rand\". Or least use a separate Lock (Mutex), namely for \"rand\" (wrap it into method). Mutex will use some sort of spinlock as well but it may use some linked list to make it more fair."
+    },
+    {
+      "prediction": "I think this should be `from pkginfo import version` instead of `from astroid.__pkginfo__ import version`",
+      "reference": "Can we grab it directly from `__pkginfo__` as that is the source of truth for the version?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "We can revert this to reduce the diff of the PR."
+    },
+    {
+      "prediction": "I think this should be `isCloud || protoPkg.contains(\"cloud\") || protoPkg.contains(\"gcp\")`",
+      "reference": "This looks weird. `isCloud` should define if it is a cloud or no. Here it does it partially, and it can be overriden by protoPkg value (which also an argument to this function). Please make sure that isCloud completely defines the cloud thing. (i.e. it an be as straightforward as moving `protoPkg.contains(\"cloud\")` from here to the place where `assembleGoImportPath` is called."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think you should check for the flags and set the flags appropriately.",
+      "reference": "Declaring fileflags to be of type LobTgtFileFlags would eliminate the need for all this casting. And would be safer. (Not a show-stopper though.)"
+    },
+    {
+      "prediction": "I think this should be `LOG.info(String.format(\"Mapping %s to docker image %s %d times\", caps, name,```",
+      "reference": "This change prevents a user understanding how their server is configured. Best to leave at `info` level."
+    },
+    {
+      "prediction": "I'm not sure why this is necessary, but it seems like it should be removed.",
+      "reference": "Typically we would add something to the build system environment to differentiate this; also is there an alternative that can be used with older OpenSSL? Other folks who are committers will need to decide if it is worth supporting an older and likely quite vulnerable (to hacks) OpenSSL library."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these tests. I think they should be in `tests/producers_test.go` instead of `tests/read_block_producers_test.go`",
+      "reference": "`readActiveBlockProducersByHeightTests` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `updateSync` method, since it's the only place where we're saving the records to the smartstore.",
+      "reference": "Falls back on `Constants.ID` if there's no custom field set."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to skip the offset. I think it would be better to do it in the constructor.",
+      "reference": "I believe this is what we want to rearrange: this in.skip() should not be happening here, in the open method, but in the setOffset() method itself. Because we want to be able to change that offset after the initial open. The setOffset() method will need to throw an IOException, if it's called while the InputStream is still null; or if the skip() call itself results in an IOException()"
+    },
+    {
+      "prediction": "I think this should be `getMethodNameDeclaratorNode().getParent().getResultType().getChild(0).getResultType().getChild(0)` instead of `getMethodNameDeclaratorNode().getParent().getResultType().getChild(0) instanceof ASTPrimitiveType`",
+      "reference": "In case the method is `void`, there won't be any children and `getChild(0)` throws. We'll need to check with `isVoid()` for that case. I'll update this when I merge."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `yarpc` package.",
+      "reference": "If the yarpc.Interceptors wrapper will be added to any user interceptor, why not do it transitively. Can save one step for users. Same apply to filter."
+    },
+    {
+      "prediction": "I think we should rename this function to `removeEtcdDataDir` or something like that.",
+      "reference": "Is there anything we should do in the event we're unable to renaming the directory?"
+    },
+    {
+      "prediction": "I think this should be 8081?",
+      "reference": "How is this env var populated? Can this be retrieved from cluster metadata?"
+    },
+    {
+      "prediction": "I think this should be `record.Errorf` instead of `record.Warnf`.",
+      "reference": "Should probably skip the success event here, since it could be a noop."
+    },
+    {
+      "prediction": "I think this should be `imageBlurHashes: item.ImageBlurHashes,` and not `imageBlurHashes: item.ImageBlurHashes`",
+      "reference": "I added this here because people cards (which depends on cardBuilder) had this added in blurhash. Not sure when this is used though cc @JustAMan"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the URLOpener package, since that's where the URLOpener lives.",
+      "reference": "Let's keep this line in the docstring, to let end users know they shouldn't be using this directly."
+    },
+    {
+      "prediction": "I think we can remove the check for gas price here as it's already checked in the above validation.",
+      "reference": "can do the same for other validateXXX()"
+    },
+    {
+      "prediction": "Is there a reason for this change?",
+      "reference": "Why is this necessary?"
+    },
+    {
+      "prediction": "I think this should be `import datetime as pd` or `import datetime as pdt` or `from datetime import datetime as pdt` or `from datetime import datetime as pd` or `from datetime import datetime as pdt` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pdg` or `from datetime import datetime as pd",
+      "reference": "Please move this import upper to other similar imports"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "Include the key word \"translated\". Otherwise the reader will find this confusing."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to memoize this. I think it should be in the `ElasticsearchStorage` class itself.",
+      "reference": "why. is this called per request? if so this also adds a lock internally. can we not add this at the moment until analyzed as this assumption isnt carried to the several other impls. the only other consumer of this is the http api and that isnt hit often enough for specializations."
+    },
+    {
+      "prediction": "`actualNode` is not a property of `option` but of the `amp-selector` component. I think it would be better to use `option.hasAttribute('selected')` instead of `option.hasAttr('selected')`",
+      "reference": "I believe this means dependent checks can no longer operate on virtual nodes, which is something we've been pushing for lately. We may need to find a different way to solve this issue, as I don't think we want to cause a regression while fixing a bug."
+    },
+    {
+      "prediction": "I think this should be `os.path.join` instead of `os.path.abspath`",
+      "reference": "ditto about use of `abspath`"
+    },
+    {
+      "prediction": "I think this can be simplified to `Math::max`",
+      "reference": "this reduces the runtime by 50 times"
+    },
+    {
+      "prediction": "This file should be renamed to `RemoteDistributorTest.java`",
+      "reference": "Probably best not to have an empty test...."
+    },
+    {
+      "prediction": "I think this should be `final X exception = exceptionType.getConstructor().newInstance();` instead of `final X exception = exceptionType.newInstance();`",
+      "reference": "direct `newInstance` call is also deprecated now"
+    },
+    {
+      "prediction": "I think this should be `k8s.io/kubernetes/pkg/client/informers/informers.go`",
+      "reference": "We have a bit of a convention where imports are split into three blocks: first Go standard library, then imports from outside the repo, then imports from inside the repo."
+    },
+    {
+      "prediction": "I think we should keep the `protected` modifier here.",
+      "reference": "Why was this change needed?"
+    },
+    {
+      "prediction": "I think this should be `\\0\\0\\0` instead of `\\0\\0\\0\\0`",
+      "reference": "can you use this opportunity to remove `.readlines()` useless list creation? (`addresses = [l for l in fd if l.find(\"netmask\") >= 0]`)"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Why does this check differ from `plan_ssh.rb`?"
+    },
+    {
+      "prediction": "I think it's better to use `github.com/iotexproject/iotex-core/consensus` as a submodule.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be 64 instead of 3. I'm not sure how many concurrent writes we can have at the same time.",
+      "reference": "It was @lodejard who initially suggested this pattern. I think that it had something to do about prioritizing future writes even if there was a pending write operation ready to handle the newly requested write. I'm not sure I fully understood the explanation for having up to 3 pending write requests, because only having one queued write at a time still makes sense to me. I'll talk to him about it."
+    },
+    {
+      "prediction": "I don't think this is a good name for this debug flag. I would suggest something like: `--enable-debug-serialization` or `--debug-serialization` or `--enable-debug-serialization`.",
+      "reference": "We have cli --online_weight_clear in cli.cpp"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with that doesn't break anything.",
+      "reference": "So the request.Headers is a map of []string. Are you sure that the HTTP library doesn't already do this split for us?"
+    },
+    {
+      "prediction": "Shouldn't this be \"Execution stopped, executor/container crashed\"?",
+      "reference": "How is this message used? Can we remove the \"crashed executor/container\" part which is an implementation detail?"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. It's not a command, it's a property. The heap snapshot is a property of the session, so it should be under the `/session/:sessionId/heap_snapshot` path.",
+      "reference": "Instead of building in routing for a browser-specific command, could you refactor the HttpCommandExecutor to allow arbitrary commands to be registered?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the \"back\" link, but it's the only way I could come up with at the time.",
+      "reference": "Space after `=`."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the `afterCreate` handler.",
+      "reference": "if just need to do something after createInvocation just override and call super first is enough?"
+    },
+    {
+      "prediction": "I'm not sure why Travis was complaining about this one, but I think it's because of the trailing period at the end of the line.",
+      "reference": "By the way, I fixed `announce` method (missing `super` call to announce JDK version)"
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think it's better to use `Object::equals` and `Object::toString` instead.",
+      "reference": "I'm not sure about that... This line can produce <code>ClassCastException</code> if <code>Comparator</code> do not check this."
+    },
+    {
+      "prediction": "This is a breaking change. I think it would be better to just return the bare TLF in this case.",
+      "reference": "There wasn't any special reason for this to return a pointer, right?"
+    },
+    {
+      "prediction": "I think this should be a `ConcurrentHashMap<Integer, NotificationData>` instead of a `HashMap<Integer, NotificationData>`.",
+      "reference": "Why the need for a `ConcurrentHashMap`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could come up with that works.",
+      "reference": "New code should use strict equality `===`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this test. I think it should go in `spec/workers/notify_spec.rb` but I'm not sure if that's the right place to put it.",
+      "reference": "`expect(resources['Notify[local pid]']['events'][0]['desired_value']).to match(/(\\d+)/)` seems clearer."
+    },
+    {
+      "prediction": "I think we can remove the `changeMd` variable now.",
+      "reference": "Removed now-unneeded param."
+    },
+    {
+      "prediction": "`bitswap` is not a command, it's just a command-line flag.",
+      "reference": "(NON-blocking, this can be tracked in follow up issue) @anorth @mishmosh is the toplevel getting too crowded? Should we have a `network` grandparent command, or maybe a `stats` command?"
+    },
+    {
+      "prediction": "I think this should be `_x( 'Analytics', 'Service name', 'google-site-kit' )` instead of `_x( 'Analytics', 'Service name', 'google-site-kit' )`",
+      "reference": "The `_x` function needs to be imported at the top of the file (in addition to `__`)"
+    },
+    {
+      "prediction": "I think this should be `WaitForCondition(() => (DetectionsGraphController.Settings.QValueCutoff == 0.003f));`",
+      "reference": "This is functionally equivalent to the code it replaces."
+    },
+    {
+      "prediction": "I don't think we should be setting the ENIType here. We should only set it when creating the ENI.",
+      "reference": "can you change the field name here to \"InterfaceAssociationProtocol\"? same for the Config struct in agent/ecscni/types.go. i think it's better to keep the field name consistent between agent and acs payload"
+    },
+    {
+      "prediction": "This file should be removed.",
+      "reference": "See above, same for the other cases."
+    },
+    {
+      "prediction": "I don't think this method should be marked as `@since 4.2`",
+      "reference": "Shouldn't line 313 be public \\<T\\> SettableGauge\\<T\\> **settable**Gauge(String name) { ? It would also be good to call it a few times and with differnet types (Long, Integer, String) in the test as well."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. It seems like it should be in the `infer_call_result` function, but I can't find where it is used.",
+      "reference": "I would prefer to just fix the occurences of this new check rather than disable them"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it's better to have a constant for the value of `i` instead of hard-coding it here.",
+      "reference": "The tests should be fixed in #4841, so this is no longer needed"
+    },
+    {
+      "prediction": "Shouldn't this be `InvalidArgumentException` instead of `SerializationException`?",
+      "reference": "I think it should been in `SharedKernel` module."
+    },
+    {
+      "prediction": "I think it would be better to use `+optional` here.",
+      "reference": "Can we add some description for this field?"
+    },
+    {
+      "prediction": "I think this should be `DurationType` instead of `Duration`",
+      "reference": "Should its name be Time or Duration? I can see them both."
+    },
+    {
+      "prediction": "I think this should be `[]otlpconfig.GRPCOption` and not `[]*otlpconfig.GRPCOption`",
+      "reference": "Is this used anywhere?"
+    },
+    {
+      "prediction": "This is a test for `String()` and `Attributes()`, not for `Merge()`.",
+      "reference": "I believe that you can use `assert` instead of `require` here and in the following lines"
+    },
+    {
+      "prediction": "Why do we need to pass a pointer here?",
+      "reference": "In `storagemarket.go` you used `nil` for an unwanted out parameter. Do something consistent (nil seems fine if supported)."
+    },
+    {
+      "prediction": "I'm not sure we need this package. I think we can just use `Contains`, `HashBytes`, `PadBytesLeft` and `IsPotential` everywhere.",
+      "reference": "is this a new pattern we use in bee? interesting"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just return an error here and not wait for potential failure later.",
+      "reference": "If you think this PR works, i'll clean this up before resubmitting."
+    },
+    {
+      "prediction": "I think we can remove these two constants.",
+      "reference": "are you sure you can read version by this artifactid?"
+    },
+    {
+      "prediction": "I think we should revert this change.",
+      "reference": "We can get rid of this import then."
+    },
+    {
+      "prediction": "I think this should be `switch c.typ` and not `switch c.typ == CLIENT`",
+      "reference": "Should we do our own vtable?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to remove `config.exceptions_app = self.routes` from `config/environment.rb` so that `config.exceptions_app` is always `nil` when `Rails.env` is set to `development` or `test`. Otherwise, when `Rails.env` is set to `production` or `staging`, `config.exceptions_app` will be set to `Rails.application.routes.url_helpers`, which is not what we want.",
+      "reference": "This is the actual fix."
+    },
+    {
+      "prediction": "I think this should be `defaultSyncInterval = 30 * time.Second` and not `defaultSyncInterval = 60 * time.Second`",
+      "reference": "If sync has really a negative impact, this is just moving the issue from 10sec to 60sec. Wonder if you should not expose the (auto)sync params so users can decide."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in the `Decoder` interface.",
+      "reference": "should be called => will be called"
+    },
+    {
+      "prediction": "I think this is not the right place to calculate the height.",
+      "reference": "not sure whether we should do this. It may cause problem that delegates upgrade their nodes at different time, that they will have different \"last block time\", some use \"commit time\", some use \"block time\". Potential solution: Only use block time after berling, and then delete it in the next version after berling. Open to discussion."
+    },
+    {
+      "prediction": "I think it would be better to move the call to deleteCollection to the try/catch block above, and wrap the exception in the deleteCollection call. This way, if the exception is thrown by the call to deleteCollection, it will be propagated to the caller, and the caller will be able to decide what to do with the exception.",
+      "reference": "So one question I have is why is the error coming back from `buildReplicaPositions` not an `Assign.AssignmentException`? Is it because it is wrapped in a `SolrException` from the remote node?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move this test to `test_logging.py` under the `rbind` section.",
+      "reference": "`dt` won't work for some reason?"
+    },
+    {
+      "prediction": "I think this should be `\"\" != os.getenv(\"KOALAS_USAGE_LOGGER\")`",
+      "reference": "Is this because \"KOALAS_USAGE_LOGGER\" can be `None` ??"
+    },
+    {
+      "prediction": "Shouldn't this be 500 instead of 1000?",
+      "reference": "I like it but maybe the better name will be EthGasPriceConstants or EthGasPriceEstimatorConstants? but Constants not Config"
+    },
+    {
+      "prediction": "nit: I think you can remove this line.",
+      "reference": "I would inline the constructor as well"
+    },
+    {
+      "prediction": "nit: I think it would be better to use `nebula::GraphParser::semantic_type` and `nebula::GraphParser::location_type` instead of `dummyyylval` and `dummyyyloc`.",
+      "reference": "I'm not confident about this..."
+    },
+    {
+      "prediction": "I think this should be `certificates` instead of `certificateRequest`",
+      "reference": "There is also `v1alpha1.CertificateRequestKind` I think?"
+    },
+    {
+      "prediction": "I think we should rename this to `GCPBrokerSenderJob`",
+      "reference": "Can we rename it to make it less confusing? Instead of \"sender-gcpbroker\", maybe name it to \"retryable-sender\""
+    },
+    {
+      "prediction": "I think this should be `literal().value() instanceof Literal` instead of `literal().value().getClass().getName()`",
+      "reference": "So looking at the definition of the `literal()` function in this class, it seems it's possible for it to return null. I guess it's not a concern as we would get NPE on the above call at line 136 when trying to call `.to` if `literal()` returned `null` before even getting to this part that calls `literal().value()`, but something I thought I'd bring up. Perhaps something we might follow up on in another issue or possibly I just missed the workflow that makes `literal()`'s result non-null by the time this `bindLiteralOperration` is called."
+    },
+    {
+      "prediction": "I think you can remove this line.",
+      "reference": "Why did you break the line? It looks like it is under 80 characters without the break and it is inside parens so no need for backslash anyway"
+    },
+    {
+      "prediction": "I think this should be `from datatable import DataTable` instead of `from datatable import DataTable as DataTable`",
+      "reference": "@guolinke Don't you mind to rename this variable to `DATATABLE_INSTALLED`, for the consistency with other variables (for example, there are `PANDAS_INSTALLED` but not `PD_INSTALLED`). Also, `DT` is a little bit confusing: sometimes `dt` is used for `datetime`."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a helper method that does the following: 1. Finds all of the examples in the given profile 2. Finds all of the locations of those examples 3. Loops over all of those locations and formats them into a single string",
+      "reference": "We're fine with double quotes here. Just for future reference."
+    },
+    {
+      "prediction": "I don't think we need this function at all. We can just use `yarpcerrors.ErrorCode(err).String()`",
+      "reference": "can we keep this around? We should make sure we can distinguish between properly wrapped errors and \"unwrapped\" errors"
+    },
+    {
+      "prediction": "I think this should be `topology_at_rasterizer = graphicsPipelineCI.pInputAssemblyState->topology;` instead of `if (graphicsPipelineCI.pInputAssemblyState) { topology_at_rasterizer = graphicsPipelineCI.pInputAssemblyState->topology; }`",
+      "reference": "`RecordPipelineShaderStage()` might change `topology_at_rasterizer `, according to shader code, so we should check `pInputAssemblyState `first."
+    },
+    {
+      "prediction": "I think this should be `assets` not `Assets`",
+      "reference": "Should we just inline the assets here? I can't think of an advantage to having them in a separate file."
+    },
+    {
+      "prediction": "I think we can remove the `map[string]k8sproxy.Endpoint` part here.",
+      "reference": "detectStaleEndpoints to be more specific?"
+    },
+    {
+      "prediction": "I don't understand why this is here.",
+      "reference": "Please create a new issue for addressing this (if there isn't one already), so it doesn't get lost."
+    },
+    {
+      "prediction": "I think we should make this a `sync.WaitGroup` instead of a `chan interface{}`.",
+      "reference": "Nothing was ever coming out of this channel because its already being consumed. Create a new channel here and feed it in HandleNewHead method."
+    },
+    {
+      "prediction": "nit: remove `Amazon` from the name since it's no longer AWS terminology.",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think this should be `fine` instead of `fine`",
+      "reference": "This is deliberately at this level."
+    },
+    {
+      "prediction": "I think this should be `err` and not `hex.DecodeString`",
+      "reference": "payload entered on command line is in hex-encoded format should use same processing as in ioctl/cmd/action/actiontransfer.go"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to create demo pricing groups. I think we should move this logic to the `createPricingGroup` method in the `DemoPricingGroupDataFixture` class.",
+      "reference": "should this be kept in the `else` branch?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Why not put this below \"strings\"?"
+    },
+    {
+      "prediction": "I think this should be `public int UvPipeCount => _dispatchPipes.Count;`",
+      "reference": "Make it `internal` if it's just for testing."
+    },
+    {
+      "prediction": "I don't think you need to disable this one.",
+      "reference": "What is triggering this message in this file?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it. I think we need to make sure that all workers are started before we try to check the results.",
+      "reference": "Are you sure this should be removed?"
+    },
+    {
+      "prediction": "I don't think we need this constructor anymore.",
+      "reference": "What's that good for? The field is initialized in the constructor."
+    },
+    {
+      "prediction": "I think this should be `if self.wfile and self.wfile.close()`",
+      "reference": "This shouldn't be necessary (same below). Do you have a traceback for me?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "No newline needed here."
+    },
+    {
+      "prediction": "This file should be renamed to `libkbfs.go`",
+      "reference": "Why a separate block?"
+    },
+    {
+      "prediction": "I think this should be `LOG(ANNOTATIONS, 2, \"reg=%d, %s, last reg=%d\\n\", reg, get_register_name(reg), DR_REG_LAST_ENUM);`",
+      "reference": "Probably better to ask @AssadHashmi or another AArch64 expert for a review rather than me -- @AssadHashmi if you could confirm that these status registers are 64-bit despite having only a few fields?"
+    },
+    {
+      "prediction": "I think this should be `m/44'/1'/%d'/%d'/%d'` instead of `m/44'/0'/%d'/%d'`",
+      "reference": "`purpose` and `coin_type` and `account` would mimic the BIP-0044 wording, but it's up to you."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that we are setting the `Checked` property of the `rbQValue01` control to `true` when `qValue == .01f`, but we are not setting the `Checked` property of the `rbQValueCustom` control to `true` when `qValue == .01f`. The reason is that the `Checked` property of the `rbQValueCustom` control is being set to `false` when `qValue != .01f`, so the `Checked` property of the `rbQValueCustom` control is being set to `true` when `qValue != .01f` and the `Checked` property of the `rbQValueCustom` control is being set to `false` when `qValue == .01f`, so the `Checked` property of the `rbQValueCustom` control is being set to `true` when `qValue == .01f` and the `Checked` property of the `rbQValueCustom` control is being set to `false` when `qValue != .01f`, so the `Checked` property of the `rbQValueCustom`",
+      "reference": "Is this the critical change? It does seem wrong to use Select() instead of Checked = true. Not sure why that would pass sometimes and not others, though."
+    },
+    {
+      "prediction": "I think we can remove the `Spec.Platform.VSphere.NumCoresPerSocket` and `Spec.Platform.VSphere.MemoryMiB` fields from the `MachinePool` struct.",
+      "reference": "Seems a little low, but is this an installer default?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this hack. I think it should go in `structlog`, since it's used by both `kinto.core.utils` and `kinto.core`.",
+      "reference": "Maybe we could provide our own `configure` function that includes `logger._logger = None` because I don't see a use case in kinto for using `reset_logger` besides configuring the logger."
+    },
+    {
+      "prediction": "I think we can remove the `Num` and `Denom` fields as well.",
+      "reference": "This will cause the same breakage that was caused when we added Proof to Block. Be sure to let people (infra?) know ahead of time"
+    },
+    {
+      "prediction": "I think this test is still a little bit out of date, since it doesn't check that `ncs.nodes` is empty after the finalizer has been emitted.",
+      "reference": "Fixed the test; @strib want to validate that this is okay? Seems to be consistent new GC behavior."
+    },
+    {
+      "prediction": "I think this test is failing because `subdomain` is not a valid key for the `Account` struct.",
+      "reference": "Afaict, nothing cares"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `reembedBlockChanges` in `kbfs_internal.go`.",
+      "reference": "I must be missing something, but why does not having a node cache imply not neededing to re-embed the block changes, in particular? Isn't it just the fact that we don't do any writes?"
+    },
+    {
+      "prediction": "I think this should be `internal` since it's just a struct.",
+      "reference": "Are these changes from `class` to `struct` breaking if called from an older version of `Datadog.Trace.ClrProfiler.Managed`?"
+    },
+    {
+      "prediction": "I think we can remove bloom filter here.",
+      "reference": "These are just aliases now"
+    },
+    {
+      "prediction": "I think this should be safe to use plain text.",
+      "reference": "Does this mean we can remove `raw` from `_head_contents` partial?"
+    },
+    {
+      "prediction": "I think you can remove the `&& !this.splashWindow.isDestroyed` part here.",
+      "reference": "`isDestroyed` is a function, not a property. This condition will always return false."
+    },
+    {
+      "prediction": "I think this should be `Version_1_33` and `Version_1_34` instead of `Version_1_31` and `Version_1_32`",
+      "reference": "Why are we adding these versions?"
+    },
+    {
+      "prediction": "I think this should be `logger.info('imported %d listens for %s' % (len(listens), str(user)))`",
+      "reference": "fyi, you can do this by doing `.info(\"string %s %s\", formatparam, formatparam2)` instead of doing a string format with `\"str\" % (params)` the idea is that it'll only do the string interpolation if logging is enabled for this level, which theoretically is an optimisation, but in this case probably isn't important"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `before_action` callback of the `User` controller, since that's where the login is done.",
+      "reference": "Would this work, as a one liner? `setTimeout(window.location.reload, 5000);`"
+    },
+    {
+      "prediction": "I think we can remove the `cipher` package now.",
+      "reference": "Can you move `crypto/tls` into the block at the top with all the other stdlib imports?"
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "What is the motivation for this change?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this method. I think it should be in the `Subscription` interface, but I'm not sure if there is a better place to put it.",
+      "reference": "OK, so the concrete implementation will loop? But then it's important that this method waits for a while before it returns no messages. To be specific, a GCP implementation that set `ReturnImmediately` to true or false would both satisfy this requirement, but the first would cause busy-waiting."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `validatePath` function.",
+      "reference": "The data structure of `MatchClauseContext` may need to be redesigned."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we need to make sure that the `want` and `got` are the same size, and that they have the same contents.",
+      "reference": "You say on the issue that something has to be receiving while the messages are being published. But here it's possible that the publish goroutine starts, and maybe even completes, before Receive is ever called."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "how do we know that the step in question is an approval type step?"
+    },
+    {
+      "prediction": "I don't think we need to change anything in this file.",
+      "reference": "JMX offers normal APIs for this. I don't think you want the glassfish one."
+    },
+    {
+      "prediction": "I think you can remove the `requires io.vavr` line as it is already exported by the `io.vavr` package.",
+      "reference": "Is this still Java 8 compatible? @danieldietrich do we need a separate, modularized Java 9 release also?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the filter mode. The filter mode should be set by the user, not the resource type.",
+      "reference": "Should update the user input address and filter modes and pass those in fillSamplerDescriptor ?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "This object should be in `lib/music/show.rb` and you should also wrap this object in the `Faker::Music::Show`."
+    },
+    {
+      "prediction": "I think this should be `|| (raise Puppet::ParseError, 'Task parameters did not match')` instead of `|| (raise Puppet::ParseError, 'Task parameters did not match')`",
+      "reference": "I think logic will have to move to bolt since the vague discussions around bolt run plan --noop is that it would just skip any tasks that don't support_noop rather than error. This is fine until we actually elaborate that though."
+    },
+    {
+      "prediction": "please remove this empty line",
+      "reference": "File is not `goimports`-ed"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `web/testutils.py`.",
+      "reference": "why should unknown raise a 403?"
+    },
+    {
+      "prediction": "I think this should be `if (newType.getDerivedStateFromProps!=null) {`",
+      "reference": "The true clause of your condition, can't this just be c._nextState since `assign({}, c._nextState)` is equal to returning c._nextState, or am I misunderstanding something here?"
+    },
+    {
+      "prediction": "I think this should be `50000` instead of `500`",
+      "reference": "nit: align the leftmost quotes"
+    },
+    {
+      "prediction": "I think this should be `OptionalInt` instead of `OptionalInt64`",
+      "reference": "Probably \"server port\" -> \"TCP port\""
+    },
+    {
+      "prediction": "I think this should be `protected Lucene80Codec(String name, Mode mode) {`",
+      "reference": "can you use FilterCodec instead?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this setting. I think it should be moved to the `Settings` module.",
+      "reference": "I know there's only two values but I find it a bit confusing that it's true/false. Could you change it to a string enum with values \"portrait\" and \"landscape\"? Then when calling printToPdf you can just do `landscape: Setting.value('export.pdfPageOrientation') === 'landscape'`"
+    },
+    {
+      "prediction": "I'm not sure about this one. I think it's fine, but I don't want to make any assumptions.",
+      "reference": "I think this migration should fail for MDBX if it uses old table names"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Should this be negated? `!@options[:hosts_file].nil?` Otherwise it'll enter the `if` statement and `nil` will be passed to `HostsFileParser.parse_hosts_file`, which I'm assuming is bad."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the `NLog.Configuration` assembly.",
+      "reference": "I'm doubting if adding a bool here is a good idea. Maybe it should be an option object? Otherwise it's hard to extend. But an option object is a bit strange in a fluent API? What do you think @snakefoot ?"
+    },
+    {
+      "prediction": "I think this class should be private.",
+      "reference": "This is used in o.a.l.sandbox.search.PhraseWildcardQuery."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to store the protocol view.",
+      "reference": "view should be loaded on start for both statedb and factory"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure if it's related to this PR.",
+      "reference": "Why rename? Just curious..."
+    },
+    {
+      "prediction": "I think you can remove this line.",
+      "reference": "Is this needed? I think it calls the super function that already does that."
+    },
+    {
+      "prediction": "`self.num_classes` should be `num_classes` here.",
+      "reference": "This might slow down the inference time, are we sure we need to do that?"
+    },
+    {
+      "prediction": "I think it's better to move this to the top of the file, so it's clear that this is a part of the package.",
+      "reference": "This would read better if you add \"Skip code/lines\" or (exclude/hide code if you follow my earlier suggestion\")"
+    },
+    {
+      "prediction": "I think it would be better to use `org.apache.commons.io.IOUtils` instead of `IOUtils`.",
+      "reference": "Looks like glassfish went a bit nuts doing a refactor"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in the `OpenTelemetrySdk.Trace` assembly, since it's only used by that assembly.",
+      "reference": "nit: allocate the list using the number of processors."
+    },
+    {
+      "prediction": "I think this should be `javaClasses[i]` instead of `initializedClasses[i]`",
+      "reference": "Was this change triggered by baseline?"
+    },
+    {
+      "prediction": "I think we can remove this method now, since we don't need it anymore.",
+      "reference": "This code was doing nothing as far as I can tell and was making the SQL find nothing with the new version of Rails. Remove it unless someone can explain why it is there."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the browser login enabled flag. I think it should be done in the `SafeforceSDKManager` or `SafeforceSDKManagerImpl`",
+      "reference": "Setting browser based login as the default for `RestAPIExplorer`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in `DataParallelTreeLearner<TREELEARNER_T>::FindBestSplitsFromHistograms`.",
+      "reference": "maybe we can do this in Init function?"
+    },
+    {
+      "prediction": "Shouldn't this be `errRequestIDTooLong`?",
+      "reference": "How about if request ID not set, service should generate a random uuid?"
+    },
+    {
+      "prediction": "I think this is the only change that needs to be made to the test. The rest of the changes are cosmetic.",
+      "reference": "minor: this is an opportunity for you to split these long lines into multiple lines :)"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it's better to catch the exception and raise a new one with a more descriptive message.",
+      "reference": "It would be nice to have exception dumped somewhere (debug logs, info logs)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/cloud/internal/structs.go` or `pkg/cloud/internal/structs.go`.",
+      "reference": "So I don't think this is right, as discussed on #657."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this query. I think it should go in `user_groups` table.",
+      "reference": "should the group_id = \"{0}\" be group_id = %s You'd still keep the {} for the group_members_{} for the tablename but make the filter clause parameterized."
+    },
+    {
+      "prediction": "I think this should be `file_dict.get('path', path_exc)` instead of `file_dict.get(\"path\", path_exc)`.",
+      "reference": "Path can contain variable pattern in the middle. Also, `$` is not enough to tell if this is JMeter variable. We have correct variable detecting somewhere in our code."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this method. I think it should be in the `FormTestModel` class, since it's only used for testing the `FormTest` model.",
+      "reference": "@bennothommo can you revert this section? It shouldn't be required."
+    },
+    {
+      "prediction": "I think we can remove the `% 100` part here, since `FISCAL_YEARS` is an array of length 10.",
+      "reference": "Minor: maybe move the above to a `#fiscal_year` method?"
+    },
+    {
+      "prediction": "I think this should be `assert build.build_package('test_hdf5', 'generated', buildfilepath) in [os.path.join(path, 'test_hdf5'), os.path.join(path, 'generated/test_hdf5.py')]`",
+      "reference": "You should use the current directory instead - it's a temporary directory that will get removed when the test is done."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "I think we don't need synchronization here."
+    },
+    {
+      "prediction": "I don't think we should change the wording of the error message here. I think it's fine to leave it as it was.",
+      "reference": "I don't think this should be changed since on the reactive side we don't support anonymous users."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `after_save` callback.",
+      "reference": "Maybe this `if` should be on line 118?"
+    },
+    {
+      "prediction": "I think this should be `uint64` instead of `int64`.",
+      "reference": "Cool. (I thought I had fixed this already in #812.)"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine to keep the `cacert` and `token-file` defaults in the `BoltTransport` class, but I think it would be better to move the `cacert` and `token-file` defaults to the `Bolt::Transport::HTTP` class.",
+      "reference": "I intend to do this as soon as I have tests passing."
+    },
+    {
+      "prediction": "I think this should be `sys.stdin.close()` instead of `sys.stdin.read()`",
+      "reference": "Please remove the `test_*` from the filename - otherwise pytest will try to collect tests from it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `libudev-devd` since that's where the udev stuff lives.",
+      "reference": "Did you file a bug to follow up on this?"
+    },
+    {
+      "prediction": "I think it would be better to use `getOrDefault` here instead of `map` and `orElseGet`.",
+      "reference": "Looks like these 4 lines didn't actually change. Can you revert the whitespace changes here?"
+    },
+    {
+      "prediction": "Style/ImportOrder: Importing from the same module is better than importing from a different module if the import order is the same.",
+      "reference": "This can't be merged with any usage of `require 'pathname'` in lib, it's not always available and we don't want false positives for people (where they don't realise they needed to requite it). Its also unnecessary, no other formatter uses it to look up failure locations. See the exception presenter etc for how we format lines."
+    },
+    {
+      "prediction": "nit: can you move the `publishDataverse` and `publishDatasetViaLinking` assertions into their own `Then` statements?",
+      "reference": "It would be nice to fix the indentation above."
+    },
+    {
+      "prediction": "I don't think we should change this.",
+      "reference": "`short_names` is still optional, so `=` that indicates it has to stay there."
+    },
+    {
+      "prediction": "nit: I think we should rename this to `appendEFSCapabilites` to be consistent with `appendFirelensFluentbitCapabilities` and `appendFirelensLoggingDriverCapabilities`",
+      "reference": "missing calling of this method"
+    },
+    {
+      "prediction": "I think this is the only place where `deploy/cloudFormation` is used.",
+      "reference": "We could get rid of this dependency being forced on clients by changing the `StackConfiguration` interface abstraction to deal with strings instead of `cloudformation.Parameters` and `cloudformation.Tags`, and internally to the `cloudformation` package translate the input strings into the `service/cloudformation` package specific values hiding that from consumers. I don't think that should be taken on as part of this PR, but I think we should consider doing it at some point."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be under `chaos/mesh` since it's specific to the mesh.",
+      "reference": "`StressChaos` does not contains a field .spec.action"
+    },
+    {
+      "prediction": "I don't think this is the right place for this test. It should be moved to `Test\\IteratorTrait`.",
+      "reference": "Non-int literal strings are no longer coercible to int, I think this is an improvement. Without this change it now reports `InvalidArgument`."
+    },
+    {
+      "prediction": "I think this should be `ARRAY_SIZE(a)` instead of `sizeof(a) / sizeof(a[0])`",
+      "reference": "I this used anyplace?"
+    },
+    {
+      "prediction": "Why did you change this?",
+      "reference": "Why was this change required?"
+    },
+    {
+      "prediction": "I think this should be `protected final boolean missingAdminRole;`",
+      "reference": "I like that we have this attribute here. However, I think the way we use it right now might be confusing in the future. For example, when only camunda admin is checked, this exception will be instantiated with a message and `missingAdminRole` is `false`. I think that is counter-intuitive. I would expect this exception to have `missingAdminRole` set to `true`. With this explicit information, the exception message could now be generated inside this exception class with the text block that is currently used in the authorization manager. We could maybe generally reuse that text block then also for the case where a user is no admin and also has none of the required permissions. I haven't tried it myself but I think it would be great if we could streamline those things a bit more if we're working on it now. Let me know what you think."
+    },
+    {
+      "prediction": "This should be `public static int getSizePrefixedRootAsMonster(ByteBuffer _bb);`",
+      "reference": "Not sure why we're creating a new `ByteBuffer` here, ideally this refers to the existing one?"
+    },
+    {
+      "prediction": "I think this should be `DATE_RANGE_OFFSET`, not `DATE_RANGE_OFFSETS`.",
+      "reference": "Same here re: `STORE_NAME` to `MODULES_SEARCH_CONSOLE`."
+    },
+    {
+      "prediction": "I think we can remove this line now, right?",
+      "reference": "I could've sworn I got a `403` myself when I got that response locally. Is `200` what the API would send back for a `permission: false` tag?"
+    },
+    {
+      "prediction": "I think we can remove the `NewIdentityManager` function altogether. It's not used anywhere.",
+      "reference": "This attribute was intentionally private. - lets force usage of factory `NewIdentityManager()` - lets ramake to `NewIdentityManager(keydir string)` -> `NewIdentityManager(keystore keystoreManager)`"
+    },
+    {
+      "prediction": "I think this should be 100ms, not 10 * time.Millisecond.",
+      "reference": "Is this too frequent?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `models` folder.",
+      "reference": "will this not make it possible to use a zip file, e.g. on other importers who do not have to support it ?"
+    },
+    {
+      "prediction": "I don't think we need to import `SocksClientLayer` here.",
+      "reference": "Needs to be mentioned in `__all__` below."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this check. I think it should be done in the `sigfd_set` function, since that is the place where the `sigfd_table` is initialized.",
+      "reference": "What if the app's signal handler, executed at pre-syscall for epoll_pwait, executes its own epoll_pwait? For that matter: what happens natively if that happens? I would not expect the kernel to keep a stack -- does that clobber the kernel's stored pre-syscall mask?"
+    },
+    {
+      "prediction": "I think it would be better to say that the returned list is sorted in ascending order and that it is guaranteed that the returned list is sorted in ascending order. It is important that the returned list is sorted in ascending order. The returned list is sorted in ascending order. It is important that the returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order. The returned list is sorted in ascending order.",
+      "reference": "Nit: missing a space. I think we should phrase the new content a little differently. \"It is important\" isn't very clear. I think it should be \"offsets will be returned in sorted order.\""
+    },
+    {
+      "prediction": "I don't think we need to import `json` here.",
+      "reference": "Can you remove this blank line and put the added import in sorted order (likely your editor will do this)"
+    },
+    {
+      "prediction": "I think it would be better to use `SSL_CONF_UPDATE_INTERVAL` instead of `ocsp_update_interval` to be consistent with the rest of the code.",
+      "reference": "Even if `pthread_create()` for `ocsp_updater_thread()` fails, `h2o` can continue to serve. But I'm a little worried whether `h2o` should continue to serve in this case. How about you?"
+    },
+    {
+      "prediction": "I think we should move this to the top of the file, since it's not used anywhere else.",
+      "reference": "I like introduction of these constants for account/container creation, but the `ACCOUNT_CHOOSE` one isn't needed I think. This should instead be an \"empty\" value (either `''` or `0`), to indicate more clearly that this field is not filled. It should be the same for choosing a container."
+    },
+    {
+      "prediction": "This should be `0` or `-1` or `1` instead of `-1`, `0`, and `1`.",
+      "reference": "I noticed that passing in `[]` for the locales argument instead of 'undefined' seems to have the same effect. I think that's preferred over passing in `undefined` as a value, and unfortunately passing in `null` seems to throw an error."
+    },
+    {
+      "prediction": "I think you can use `DbCfg.Host != \"\"` instead of `!strings.HasSuffix(DbCfg.Host, \"]\")`",
+      "reference": "1. `0` is redundant. 2. We should save result of `strings.LastIndex(DbCfg.Host, \":\")` to a variable and reuse."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the same place as `MaxConn` and `MaxRecvMsgSize`, since they're all related.",
+      "reference": "Do you envision \"new\" servers not supporting header, or is it more a way to test mix of old and new? If so, we could make it a \"private\" (non exported) option."
+    },
+    {
+      "prediction": "I think this should be `errors.Wrap(err, fmt.Sprintf(\"failed to unmarshal connection config: %s\", config.Provider))`",
+      "reference": "For future reference. Try not to do refactoring in the same PR which solves some bug or implements some feature. Not to clutter reading. Do it in separate PR."
+    },
+    {
+      "prediction": "I think this should be `StringSliceVar` instead of `StringVar`",
+      "reference": "I'm not sure about changing the flag name. unit is more accurate, but for I think for most people asset is also a fine name to use."
+    },
+    {
+      "prediction": "I think we can remove this function now.",
+      "reference": "Which windows gets activated? A random other window?"
+    },
+    {
+      "prediction": "I think we should be able to get rid of this `if` and just do `event.Context = event.Context.WithVersion(cloudevents.VersionV03)` in the `else if r != nil` block.",
+      "reference": "Just to make sure, this is totally lossless? If I put a v2 there, then it gets upgraded to v3, but my function expects v2, then this is fine and nothing is lost?"
+    },
+    {
+      "prediction": "I think it would be better to use `WatermarkField` instead of `WatermarkFieldName`",
+      "reference": "In the past, we have 3 Flink streaming jobs (1 for each AWS region) writing to the same table. We need to write to 3 different watermark table properties (1 for each region). Watermark consumer then use the min value to determine the overall table watermark. A provider pattern similar to `WRITE_LOCATION_PROVIDER_IMPL` can work. A default impl could be a single property name. not sure if there is a simpler way to achieve this."
+    },
+    {
+      "prediction": "I think this should be `map[keybase1.UID]keybase1.NormalizedUsername`",
+      "reference": "Can you change this to `ResolvedUsersMap` and build a new map with both readers/writers? That way, we get the nice side effect of not exposing the internal map of the handle, and simplifying the callers a bit."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "I typically try to avoid nested ternary expressions. Can we rewrite this as if statements? Worst case, can we break it out into its own function?"
+    },
+    {
+      "prediction": "I think this should be `tls_start.ssl_conn.set_tlsext_host_name(tls_start.context.client.sni.decode())`",
+      "reference": "Do we still support handshakes without SNI present?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Maybe it's because I work too much on a centralized logging system but it could be great to add a \"category\" like \"attribute\", \"substitution\", \"include directive\", \"macro\", \"extension\"..."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that this function is only used in the `__init__.py` file, which is not the same as the `__init__.py` file that this function is used in. The `__init__.py` file is used by all the other files in the module, so this function should be used in the `__init__.py` file, not in the `__init__.py` file that this function is used in. The `__init__.py` file is used by all the other files in the module, so this function should be used in the `__init__.py` file, not in the `__init__.py` file that this function is used in. The `__init__.py` file is used by all the other files in the module, so this function should be used in the `__init__.py` file, not in the `__init__.py` file that this function is used in. The `__init__.py` file is used by all the other files in the module, so this function should be used in the `__init__.py` file, not in the `__init__.py` file that this function is used",
+      "reference": "it have no sense in python -> it makes no sense in python"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/portlayer/portpool.go`",
+      "reference": "maybe just do a separate implementation of the pool for openvpn case? It's a bit confusing having two methods here."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this.",
+      "reference": "how is it possible to have a `topology` that is falsey when we had to call `topology.connect` in order to get here?"
+    },
+    {
+      "prediction": "I'm not sure if we want to stub the `where` call here or not. If we don't, I think we can remove the `expect(PlanFinder).to receive(:where).with(sku: 'sku')` line below.",
+      "reference": "Do we need this expectation? If we don't call this method on `PlanFinder` there's no way for the `10` to be returned. Seems somewhat redundant. What do you think?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this change. I think it should be in `edgeconfig.go`",
+      "reference": "We need to do the same for edgecore"
+    },
+    {
+      "prediction": "I don't think we need to wait for the event to be set before calling the callback. We can call the callback right after the event is set.",
+      "reference": "This fix has nothing to do with the product change. Just a fix I am making for the test issue."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Exporter` class.",
+      "reference": "Don't know what should be the default. Lets keep BatchingProcessor for now. And revisit the area after beta."
+    },
+    {
+      "prediction": "I'm not sure if this is a good or a bad thing, but I think it's a good thing.",
+      "reference": "Ah that's how it's happening, it's probably using the `require_relative` in the optimised one and that counts as a different require to the bare one."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `IBindingListView` implementation.",
+      "reference": "What happens if there is no NIB/XIB with that name?"
+    },
+    {
+      "prediction": "I think this should be `config.Config` and not `config.API`",
+      "reference": "this is not necessary?"
+    },
+    {
+      "prediction": "I think this is the wrong place to put this. It should be in the `aws` package, not the `aws-ecs` package.",
+      "reference": "typo: `asmsecrets \"github...`"
+    },
+    {
+      "prediction": "I think this should be `wp_json_encode( $allowed, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE );` instead of `JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE`.",
+      "reference": "I think that the JSON encoding should be handled by the control itself, not the sanitization function. This should only confirm that the input is correct, not reformat it"
+    },
+    {
+      "prediction": "Is this TODO still relevant?",
+      "reference": "Should we put a TODO here to route the http.Transport through here?"
+    },
+    {
+      "prediction": "I think this should be `cmd := fmt.Sprintf(\"Get-VMSwitch -Name %s | Set-VMSwitch -AllowManagementOS $false \", networkName)`",
+      "reference": "Sorry, could you explain more about the issue? Doesn't ComputerName already default to local computer? > -ComputerName Specifies one or more Hyper-V hosts on which the virtual switch is to be configured. NetBIOS names, IP addresses, and fully qualified domain names are allowable. The default is the local computer. Use localhost or a dot (.) to specify the local computer explicitly."
+    },
+    {
+      "prediction": "I think it would be better to use `logger.Warnf` instead of `logger.Printf`.",
+      "reference": "`allowError` is a big vague. Does this mean the test case is asserting that an error should occur? If it means \"maybe an error can occur, and that's okay\" -- what's an example of this?"
+    },
+    {
+      "prediction": "I think this should be `nil, nil` instead of `nil, output.NewError`",
+      "reference": "return statements should not be cuddled if block has more than two lines (from `wsl`)"
+    },
+    {
+      "prediction": "I think this should be `public final`",
+      "reference": "Why does VirtualMachineMetrics have to implement Metric?"
+    },
+    {
+      "prediction": "I think this should be `require.NoError(m.SignVote(vote))`",
+      "reference": "line is 121 characters"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "How's that linked esbench looking with slice?"
+    },
+    {
+      "prediction": "I think this should be `files = self.conn.nlst(dirname)` instead of `files = self.conn.nlst(path)`",
+      "reference": "The split and join was probably done to ensure that this still works on Windows because os.path.dirname works differently depending on the os you're running under :(. It would probably be a little better to do `dirname, _, fn = path.rpartition('/')` if you want something cleaner than the split/join."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it would be better to put it in `SessionMapImpl` since that is where the implementation of this method is.",
+      "reference": "This is not the right approach. The `Distributor` maintains a model of the current state of the Grid. That model already contains the information about every active session. We don't need to modify `SessionMap` to expose it further."
+    },
+    {
+      "prediction": "I think this should be done in the `runLoop` loop, and the `maybeResetTimer` should be done in the `runLoop` loop.",
+      "reference": "Can you make this a member var as all the other deadlines? Then you can drop the param to resetTimer() and simplify some of the code below iiuc."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `canonicalize` is called with a `const` `TautomerData` object, and a `const` `TautomerEnumerator` object, but the `TautomerData` object has a `const` `TautomerCatalog` member. So the `TautomerData` object is passed by `const` reference to `canonicalize`, but the `TautomerData` object itself has a `const` `TautomerCatalog` member.",
+      "reference": "I don't understand the reason for the changes from unique_ptr to ROMOL_SPTR that you made in this file. The pointers aren't being shared or used elsewhere so I don't think there's any reason to make them shared. Did I miss something?"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this. I think it should be `PLUGINS` since it's a collection of plugins.",
+      "reference": "why change the variable name at all?"
+    },
+    {
+      "prediction": "`vNode` and `node` are both `ShadowRoot`s, but `vNode` is an actual child of `node` and `node` are both `ShadowRoot`s, why do we need to check if `vNode` is the same as `node`?",
+      "reference": "Since I'm still trying to keep these straight in my head, can you elaborate on what problem this solves?"
+    },
+    {
+      "prediction": "I don't think this is the right place to check this. I think it should be in the `onMessageUnboxed` method, since that's where the message view is created.",
+      "reference": "Prefer TextUtils.isEmpty() which handles getReportSpamRecipient() being null"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Couldn't you just write this as an if/else or a guard clause like on line 87? Just seems a bit weird doing this conditional assignment for essentially an if/else."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Packaging` assembly, but I'm not sure if that's the best place to put it.",
+      "reference": "Do we actually use Include anywhere yet or is this just for completion?"
+    },
+    {
+      "prediction": "I don't think we should add a new setting for this. It's not a setting that should be configurable by the user, and it's not a setting that should be sent in the `Referer` header. It's a setting that should be sent in the `X-Accel-Redirect` header.",
+      "reference": "It still bugs me this was misspelled in the standard and now the wrong spelling is the commonly used one :wink:"
+    },
+    {
+      "prediction": "I think we should keep the `secondary` class for the `listItemBodyText` and remove it from the `listItemBody` class.",
+      "reference": "seems we missed de-uglifying this one"
+    },
+    {
+      "prediction": "I don't think we need a separate type for this. We can just use the existing `NewWorkflowType` constant.",
+      "reference": "i guess these types & cron / retry specific belong to a dedicated util / struct"
+    },
+    {
+      "prediction": "I think this should be `setParameter('text', keys.join('')).`",
+      "reference": "Also update line 2205 below"
+    },
+    {
+      "prediction": "I think this should be `default: \"vertical\"`",
+      "reference": "`direction` is referenced nowhere. Do I miss something?"
+    },
+    {
+      "prediction": "I think we can remove this now that we have `deparam` imported from `npm:deparam`",
+      "reference": "The two places this was used have been removed so we can fully remove this dependency."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "This is a duplicate generation of the same file as below."
+    },
+    {
+      "prediction": "I think it's better to check `gpu_ids` in `MMDataParallel` constructor and pass it to `MMDataParallel(model.cuda(gpu_ids[0]), device_ids=gpu_ids)` in `MMDataParallel(model.cuda(gpu_ids[0]), device_ids=gpu_ids)`.",
+      "reference": "We may deprecate `gpus` if `gpu_ids` is specified."
+    },
+    {
+      "prediction": "`LOG.finest` should be used in the `finally` block, otherwise it will not be logged if an exception is thrown.",
+      "reference": "Right now this is experimental and deeply flaky. We left this at `info` to make debugging user reports a lot easier."
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. We need to do it in a way that is compatible with the rest of the codebase.",
+      "reference": "unnecessary spaces here are doubling the size of this otherwise extremely useful addition"
+    },
+    {
+      "prediction": "I think this should be `CopyDirectoryRecursively(TracerHomeDirectory, DDTracerHomeDirectory, DirectoryExistsPolicy.Merge, FileExistsPolicy.OverwriteIfNewer);`",
+      "reference": "I think we should use `FileExistsPolicy.Overwrite` instead. Files _should_ always be newer, but in the unlikely case they wouldn't be, I'm afraid some files would be overwritten and other not, leading to inconsistencies that will be hard to figure out."
+    },
+    {
+      "prediction": "I think this should be `std::cout << \"test3_5\" << std::endl;`",
+      "reference": "this should probably be removed"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/integration/test-config-daemon.go`",
+      "reference": "Nit: I would actually prefer that pretty JSON is the default, with a flag for compressed JSON. Can we acheive that easily?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `NetworkConfig` since that is the place where we configure network settings for the cluster.",
+      "reference": "Will need to think about this one. `additionalIngressRules` feels a bit opaque in terms of eventual outcome."
+    },
+    {
+      "prediction": "I think this should be `public static int fieldId(TypeDescription orcType);` instead of `static int fieldId(TypeDescription orcType);`",
+      "reference": "Can we avoid making this public with the other updates?"
+    },
+    {
+      "prediction": "I don't think this is correct. The size of `datum_vector` should be equal to the size of `transformed_blob`.",
+      "reference": "Why did you change this equal instead of less or equal?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be moved to `MutableWorldState#getMutable()`.",
+      "reference": "missing final for the blockHash field"
+    },
+    {
+      "prediction": "I'm not sure if this is a good or a bad thing. I'm not sure if we want people to be able to run `factory_bot` from the command line or not.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think it would be better to use `nn.functional.l2_norm` from nn module here.",
+      "reference": "`x_float` instead of `float_x`."
+    },
+    {
+      "prediction": "I think it's better to use `import` instead of `import swiper-bundle.esm`",
+      "reference": "Is this for `No Babel config ...` from ESLint? If so, it will be fixed in my ES6 PR."
+    },
+    {
+      "prediction": "This should be `return fmt.Errorf(\"expected state %+v to equal %+v\", state, s)`",
+      "reference": "Not important, but the code used to keep checking other cases even after one of them failed, and now it's not. Fine either way for me, just noticing."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "I think I saw this line in last PR, you can update your base branch and rebase to avoid this. And it would be awesome if you can rebase instead of merge each time you push PR to the `moby` branch, that will make the commits history clearer."
+    },
+    {
+      "prediction": "I think this should be `if (get_thread_id() == tid)` instead of `if (get_thread_id() == tid) {`",
+      "reference": "If this can change the PC of this thread, it requires handling: we can't blindly execute the syscall and lose control of the thread when the flags include CONTEXT_CONTROL. (Note that most docs imply that setting your own context this way is not supported or has undefined or unpredictable results: any idea how often that's the case, or does it generally work?) We need to handle in a similar fashion to NtContinue or setting another thread's context to ensure we retain control."
+    },
+    {
+      "prediction": "I don't think we need this method. We already know the total number of products by calling the `/count` endpoint.",
+      "reference": "@huzaifaiftikhar Changed the endpoint as suggested by Stephane in the latest commit."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/numbers.go`",
+      "reference": "There's no core.Number yet. ;)"
+    },
+    {
+      "prediction": "I think this should be `snapshot revert --volname <vol> --snapname <snap> --force`",
+      "reference": "..., ...will be posted."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `go-filecoin` command, but I'm not sure if that's the right place to do it.",
+      "reference": "Can you update this function to be `ShowHeader`? There is only one use of it at the moment in `tools/fast/series/get_head_block_height.go`."
+    },
+    {
+      "prediction": "I think this should be `DISABLE_TQDM = os.getenv('DISABLE_TQDM', False)` instead of `DISABLE_TQDM = os.getenv('QUILT_USE_TQDM', False)`",
+      "reference": "@akarve, this disables `tqdm` by default, is it intended? Also name `QUILT_USE_TQDM` might be too specific, IMHO `QUILT_INTERACTIVE` or `QUILT_PROGRESS_BARS` or something might be better."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `testAccCheckDockerEngineState`.",
+      "reference": "unrelated but just changing for consistency with code below it"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to just do `if (component.shouldComponentUpdate(props, state, context)) { component.componentWillUpdate(props, state, context); }`",
+      "reference": "This will call `componentWillUpdate()` for mounts because the else clause is removed. It should only be called for updates."
+    },
+    {
+      "prediction": "I think this should be `['read', 'write']` instead of `['write', 'read']`.",
+      "reference": "We should add write inheritance too."
+    },
+    {
+      "prediction": "I think it would be better to use `os.environ.get('JAVA_OPTS', '')` instead of `os.environ.get('JAVA_OPTS', '').strip()`",
+      "reference": "Please rename this to `LAMBDA_JAVA_OPTS`, and add a short description to the README."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this change. I think it should be in `github.com/pulumi/pulumi/pkg/util/grpc` since that's where the rest of the gRPC code lives, and this file has nothing to do with the Pulumi engine or gRPC at all.",
+      "reference": "Should be in the same import group."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this check. I think it should be in the `if pkt.tls_session.tls_version > 0x0200 and pkt.type == 23` check above.",
+      "reference": "Operator precedence is very confusing here. Care to add parentheses? ` ((version or 0x0303) >= 0x0200)` Same below."
+    },
+    {
+      "prediction": "I think you can remove this import.",
+      "reference": "javaparser-core has no dependencies and it should stay that way. So no log4j. And even then there would have been a preference for slf4j."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it would be better to put it in the `FileUtil` class itself.",
+      "reference": "Why not just retest all? it's not expensive (I think)."
+    },
+    {
+      "prediction": "I think this should be `ecr0` instead of `ecr` to match the rest of the code.",
+      "reference": "ecr and ecr0 aren't deterministically named here. This change will just cause confusion in the git history. Could you either: a) fix this and make it deterministic b) regenerate the mock until it doesn't flip definitions for ecr and ecr0"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be under `ENV['HOME'] + '/.rvm/bin'` since that's where `sudo apt-get install` is done.",
+      "reference": "This looks like bash... does this work in ruby? Or is the idea to inject the entire script into the subsequent commands?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `TraceContext` ctor.",
+      "reference": "I don't understand why this change is in this PR?"
+    },
+    {
+      "prediction": "I'm not sure if we should deprecate this behavior at all. I think it's fine to leave it as it is for now, but I'm not sure if there's a compelling reason to deprecate it. If we decide to deprecate it, I think we should add a note to the deprecation docs explaining why we decided to deprecate it.",
+      "reference": "It's a little hard to tell from the diff, but as a result of this PR, the only non-deprecated helpers are: - `has_facet_values?` (probably not long for this world) - `render_facet_partials` (perhaps part of a future component) - `render_facet_limit` (possibly obsolete once components get traction) - `facet_field_in_params` (a little tricky to deprecate at the moment) - `facet_field_presenter`"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `PersonObject` is not a subclass of `Realm.Object`, so `PersonObject.prototype` doesn't have `__proto__`. I think we need to do something like `PersonObject.__proto__ = Object.create(Realm.Object.__proto__)`",
+      "reference": "The right-hand part shouldn't use `__proto__`. Should be just `PersonObject.__proto__ = Realm.Object`."
+    },
+    {
+      "prediction": "I think this should be `panic(fmt.Sprintf(\"taskManagerImpl encountered MinTaskID >= MaxTaskID: %v vs %v\", request.MinTaskID, request.MaxTaskID))`",
+      "reference": "the == case might be legit, or not? If there is no task written since last read, but we trigger a new read, would the min == max? Or should the upper layer handle that case?"
+    },
+    {
+      "prediction": "I think this should be `new_size` instead of `old_size`.",
+      "reference": "Good catch! But is it guaranteed that `old_size < new_size` here? If not, wouldn't we want to take `min(old_size, new_size)` as the copy length?"
+    },
+    {
+      "prediction": "I think this should be marked as `MemberHiddenFromStaticInnerTypes` instead of `MemberHidesStaticFromOuterClass`",
+      "reference": "This is the only one that worries me a bit. It would be good if Brendan signed off on it."
+    },
+    {
+      "prediction": "Should this be `Optional#hasLongrunningReturnValue()`?",
+      "reference": "`LongRunningOperationDetailView` already has `isEmptyOperation`."
+    },
+    {
+      "prediction": "I think it would be better to use `urllib.quote_plus` instead of `base64.b64encode`",
+      "reference": "This will always add the authorization header to the request object. Is this the right scope for these two lines? If username/password are not defined, it will encode 'Basic :'"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Don't think a scope adds much value for us here. Also, for future reference, you don't need to use the `self.all` it is implied. Could just be: `scope :updated_at_desc, -> { order(updated_at: :desc) }` No need to change this one now though, it works."
+    },
+    {
+      "prediction": "I think this can be simplified to `return Base.rand(0, 9).round if digits == 1`",
+      "reference": "I believe `Base.` is unnecessary in this case, as the class already extends `Base`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be in `ServerConnection` since it's only used by that class.",
+      "reference": "Couldn't this result in images being scaled too small when the width is less than the height assuming the width is still what is being passed here?"
+    },
+    {
+      "prediction": "I think we should move this to the top of the file, since it's used in multiple files.",
+      "reference": "Here also, let's not cache value on top of file, but use `plugins.getConfig(\"api\").offline_mode` directly where it is needed"
+    },
+    {
+      "prediction": "I think this should be `&& $this->getAuthManager()->getAuthMethod() === 'Shibboleth'` instead of `&& ($this->getAuthManager()->getAuthMethod() == 'Shibboleth')`",
+      "reference": "You can combine the two checks into a single `!empty()` check."
+    },
+    {
+      "prediction": "I think this should be `o.MaxPayload > o.MaxPending`, since the former is the max that can be set by the user, and the latter is the max that should be set by the validator.",
+      "reference": "Should we add a warning for anything over say 8M? Just suggesting that is not recommended?"
+    },
+    {
+      "prediction": "I think this should be `currentUID` instead of `mergedMasterHead`",
+      "reference": "Looks like there are no callers left of this function."
+    },
+    {
+      "prediction": "I think this should be `attachStdin: false` instead of `attach: false`.",
+      "reference": "Should there still be a ` bool` trailing `showOutput`?"
+    },
+    {
+      "prediction": "remove this empty line?",
+      "reference": "Did we can change to v1"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do it in the `diffChildren` function.",
+      "reference": "Can you please double check that this won't break anything?"
+    },
+    {
+      "prediction": "I don't think we need to pass null here.",
+      "reference": "optionals, shouldn't need to be stated explicitly"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/version.go`",
+      "reference": "That seems quite low - why not just 0x5130ffff?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You should be able to do `__str__` on a `Gen` object and it will return the string representation of the underlying `Net` object.",
+      "reference": "Why is this needed ?"
+    },
+    {
+      "prediction": "Do we need this method? I don't see it being used anywhere in the codebase.",
+      "reference": "`Bolt::Config.user_path` returns `nil` if there's no homedir, so this will still error in that case."
+    },
+    {
+      "prediction": "I think this should be `a.Tag == protocol.ProposalPayloadTag` and not `a.Tag == protocol.ProposalPayloadTag`",
+      "reference": "nit: use %s for strings and %v for objects."
+    },
+    {
+      "prediction": "Why remove these?",
+      "reference": "Same as before, the properties already exists in parent"
+    },
+    {
+      "prediction": "I think it would be better to use `nuget_str = f\"\"\"<?xml version=\"1.0\"?>\n+    <package xmlns=\"http://schemas.microsoft.com/packaging/2013/05/nuspec.xsd\">\n     <metadata>\n         <id>LightGBM</id>\n         <version>%s</version>\n```",
+      "reference": "This fix is not quite correct. The `%s` should be replaced with `version` and the `%d` on line 39 should be replaced with `datetime.datetime.now().year`."
+    },
+    {
+      "prediction": "I think this should be `strcmp(\"gone\", type) != 0`",
+      "reference": "Maybe we should only set active = false if `strcmp(type, \"pause\") == 0`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in the `Selenium::WebDriver::Chrome` module.",
+      "reference": "Same here. and all others."
+    },
+    {
+      "prediction": "I think this should be `github.com/songgao/stacktraces/on/stacktrace`",
+      "reference": "Intentionally committed? I'm not against it, we already have a way to get goroutines without killing the process: `/keybase/.kbfs_profiles/goroutine`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do this. I think it's better to have a function that generates a new GuidPrefix_t.",
+      "reference": "I think this blank line may be removed"
+    },
+    {
+      "prediction": "I think this should be \"raise an error if it already exists\".",
+      "reference": "I think this should be a keyword-only argument, i.e. do `def add(self, url, title, *, toggle=False):` and adjust the caller to do `toggle=toggle`."
+    },
+    {
+      "prediction": "I think this should be `UINT8` instead of `UINT32`.",
+      "reference": "@aimanrosli23 Judging from the commit description, I do not know why this file got changed so much. Could you help to double confirm if you do not revert the changes brought by commits: SHA-1: 643623147a1feaddd734ddd84604e1d8e9dcebee * MdeModulePkg/SdMmcPciHcDxe: Send SEND_STATUS at lower frequency SHA-1: 49accdedf956f175041040e677163b7cbb746283 * MdeModulePkg/SdMmcPciHcDxe: Hook SwitchClockFreq after SD clock start"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "How about YAML.load_file()"
+    },
+    {
+      "prediction": "nit: can we rename this to `isContainerDispatchFail`?",
+      "reference": "Maybe make this `volatile` or atomic as this can be set/read from different threads? Also, separately you may want to check if some of the methods here need to be `synchronized`."
+    },
+    {
+      "prediction": "Shouldn't this be done in `drwrap_exit()`?",
+      "reference": "Check the return value of the drmgr ones."
+    },
+    {
+      "prediction": "I think we can remove the `data-cfasync` attribute from `js/fileupload.js` as well.",
+      "reference": "This should be `'cache'`"
+    },
+    {
+      "prediction": "I think this should be `error`, not `nil`.",
+      "reference": "The name \"old\" might cause confusion here. After installation, the \"old\" repo is at an archived path, and the new migrated repo is at the old path. Maybe something like \"target\" or \"canonical\"?"
+    },
+    {
+      "prediction": "I think this should be done in the `resolveReadPreference` function.",
+      "reference": "I think this is somewhat definitive proof that this error exists in `core` rather than `native`. We are correctly resolving the `readPreference` in the previous line, but you are able to identify that eventually the command generated in `core` is not decorated with the passed `readPreference`. Did you try to solve this there already?"
+    },
+    {
+      "prediction": "I think this should be `unwindTo = (to + s.BlockNumber) / 2 + 1`",
+      "reference": "During genesis sync it can unwind 5M blocks?"
+    },
+    {
+      "prediction": "I think this test is failing because `s.ticker()` isn't being run in a goroutine. I'm not sure if that's a bug in the engine, or if it's a bug in the test.",
+      "reference": "was it caused by the removal of removeTaskQueueManager() from this test?"
+    },
+    {
+      "prediction": "Use %i or %I for an array of symbols.",
+      "reference": "Wouldn't it just be easier to force search_session to return `with_indifferent_access`? Since you've done the hard work already, I don't think there's a problem doing it this way, but.."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `if (rp->schedule_backoff > gstate.now)` check above.",
+      "reference": "So, such tasks that could possibly not meet the deadline will never have a chance to run? I think this is not nice behavior, especially for those projects who has sometimes some very small tasks after the big one."
+    },
+    {
+      "prediction": "This should be `return 0;` instead of `return 1;`",
+      "reference": "You should add the check above this line instead. If string is empty, don't even call the function."
+    },
+    {
+      "prediction": "I think it would be better to use `item.getBoolean(KEY_CLIENT_CERT)` instead of `item.getBoolean(KEY_CLIENT_CERT)`",
+      "reference": "If clientCert attribute doesn't exist for some reason I set it to false."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `if` clause of the `if casher_branch == 'production'` block.",
+      "reference": "Do we need `app_host_flags` variable? We can directly concat with `flags`. `sh.cmd curl_cmd(unless Travis::Build.config&.ssl&.verify ? flags + ' -k' : flags, location, static_file_location), cmd_opts` right?"
+    },
+    {
+      "prediction": "This should be done in `stopEventWalk`.",
+      "reference": "Walk should actually be completed (`onWalkCompleted`) after the event is stopped.   This also makes it possible for monster to walk by smaller paths."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `Get` method of the `SimpleJoiner`.",
+      "reference": "typo in `weather` (should be `whether`)"
+    },
+    {
+      "prediction": "This flag is for testing only. Set to true to upgrade with staging windows install media.",
+      "reference": "Thoughts on having the URI as the param, instead of a boolean? The default value would be the normal prod image, and then your test would override with the staging image."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. The `fopen` and `fread` should be done in a way that doesn't leak file descriptors.",
+      "reference": "I'm surprised this is needed, but we should probably assert num_scan == 1 at least, otherwise I expect the file is pretty broken."
+    },
+    {
+      "prediction": "I think this should be `std::unique_ptr<kvstore::KVStore>(TestUtils::initKV(rootPath.path(), partitions, {0, network::NetworkUtils::getAvailablePort()})`",
+      "reference": "Could we set a default value for `partitions` and `{0, network::NetworkUtils::getAvailablePort()}` ?"
+    },
+    {
+      "prediction": "I think this should be named `RequestBodyTimeout` instead of `RequestBodyTimeouts`",
+      "reference": "Needs xml docs. The API names alone don't provide enough explanation about what these mean and how to set them. Also, we should provide some validation of inputs, such as MaxTime must be > MinTime, MinimumRate must be >= 0, etc. Consider making the properties readonly and adding a constructor that does these validations."
+    },
+    {
+      "prediction": "I think this should be `path.Join(os.TempDir(), \"testdata\", name)`",
+      "reference": "Would it make sense to create an empty file here, perhaps in a tmp dir, instead of checking empty files into Git?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I've been able to get it to work.",
+      "reference": "This change is due to `clang_format` and is not related to this PR."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in `sigaction.c` instead of `sigprocmask.c` since that is where `sigaction` is defined.",
+      "reference": "Why did you not add the signals to the call of block_all_signals_except() and instead baked them into the function? Ok, if you had a good reason for it, otherwise I would add it to the except list of the function call, since that's what it was meant for."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface. Feeds are a different thing. Feeds should be able to tell if a file has been downloaded or not. Feeds should be able to tell if a file is in the downloads list or not. Feeds should be able to tell if a file has been downloaded or not. Feeds should be able to tell if a file is in the downloads list or not. Feeds should be able to tell if a file has been downloaded or not. Feeds should be able to tell if a file is in the downloads list or not. Feeds should be able to tell if a file has been downloaded or not. Feeds should be able to tell if a file is in the downloads list or not. Feeds should be able to tell if a file has been downloaded or not. Feeds should be able to tell if a file is in the downloads list or not. Feeds should be able to tell if a file has been downloaded or not. Feeds should be able to tell if a file is in the downloads list or not. Feeds should be able to tell if a file has been downloaded or not. Feeds should be able to tell if a file is in the downloads list",
+      "reference": "Could this be done by mocking objects instead? I feel like this is changing too much of the actual logic just for the tests."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "why not just use `ManifestContent` instead of `Integer`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `TxPool.AddTx` method, since that's where the transaction is actually being added to the pool.",
+      "reference": "can we do that in TxValidator?"
+    },
+    {
+      "prediction": "I think we can remove this interface and just use the `ServiceName` field in the `Outbound` struct.",
+      "reference": "We definitely want some detailed and thoughtful docs being added to this type."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `if config.BPFEnabled && config.XDPEnabled {` block.",
+      "reference": "We would like this code to run, even when `config.XDPEnabled` is false, so that Felix can clean up its own XDP state after a restart."
+    },
+    {
+      "prediction": "I think we should move this to `space.cpp` since it is not used in `kvstore.cpp` and `kvstore_impl.cpp`.",
+      "reference": "If the status is `ERR_SPACE_NOT_FOUND` why return `SUCCEEDED` ?"
+    },
+    {
+      "prediction": "reverse (bool): If it's `True`, the order of input planes will be reversed after passing through this layer. Before passing through this layer, the order of input planes should be: (1, 3, 4, 6, 15). After this layer, the order of planes should be: (1, x, x, x, x). If it's `False`, the order of input planes will not be reversed before being passed to this layer. Otherwise, this layer does not reverse the order of input planes.",
+      "reference": "It is is more appropriate to use `downsample_first`. If `downsample_first=True`, the downsample block is the first block and it is used for ResNet. If `downsample_first=False`, the downsample block is the last block, which is used by Hourglass network."
+    },
+    {
+      "prediction": "I think this should be `service = getDefaultService() || service.start();`",
+      "reference": "Would you mind removing the default parameter above? (I doubt I'll ever use defaults again since you still have to protect against callers explicitly passing `null` or `undefined`)"
+    },
+    {
+      "prediction": "I think this should be `Consider using sys.exit()` instead of `Consider using the sys.exit()`",
+      "reference": "I don't know if there are strict conventions about this, but I think the name of the warning should be the diagnosis, not the suggested course of action. In this case, that would mean changing the name of the warning to `interactive-exit` or something like that."
+    },
+    {
+      "prediction": "I think this should be `pki.CommonNameForCertificate(crt)`",
+      "reference": "If `crt.Spec.CommonName` is not set, and `cert.Subject.CommonName` *is* set, we are not correctly handling it here."
+    },
+    {
+      "prediction": "I don't think we should be using `cmdhistory` and `fprompthistory` in tests.",
+      "reference": "This isn't a test, so it shouldn't be named `test_init`. You could name it `cmdhistory_init` or so."
+    },
+    {
+      "prediction": "I think this should be `configuration.get_config(self.spark_version, \"pickle-protocol\", default=pickle.DEFAULT_PROTOCOL)`",
+      "reference": "why is this pulling from `self.spark_version` config section rather than the `spark` config section (`py-packages` appears to pull from a config section called `spark`)"
+    },
+    {
+      "prediction": "I think this test should be moved to `test/testutil.go`",
+      "reference": "nit: Think it can be one line.."
+    },
+    {
+      "prediction": "This import is not used anymore.",
+      "reference": "please inline imports"
+    },
+    {
+      "prediction": "I think this should be `p.config.Rootless && len(p.cgroupPaths) > 0`",
+      "reference": "This check is still correct in some cases, but I guess erroring out is acceptable if someone explicitly asked for an impossible cgroup configuration (now that we could in principle nest things). I would like to see a test for this though."
+    },
+    {
+      "prediction": "I'm not sure if this is something that should be fixed in this PR, or if it should be a separate PR.",
+      "reference": "Should we update integral to operate on a single column as well?"
+    },
+    {
+      "prediction": "Why do we need to call `cuda_get_deep_copy_stream()` here?",
+      "reference": "Doesn't belong to this PR"
+    },
+    {
+      "prediction": "Nit: I think we should call this `RegisterContainerInstanceWithRegistrationToken` to be consistent with the rest of the client.",
+      "reference": "Just for my own understanding, does aws ecs API take this token as a new input attribute? Which version of the aws sdk? I did not find it in the official aws sdk doc."
+    },
+    {
+      "prediction": "I think this should be called `USER_AUTH_FEATURE_CODE`",
+      "reference": "Could we shorten this constant to maybe `FEATURE_USER_AUTH`?"
+    },
+    {
+      "prediction": "I'm not sure this is a good number to start with. If we have 1000s of workers, we could end up with 1000s of concurrent rekey requests.",
+      "reference": "Any particular reason for 8? I feel like we could probably handle more..."
+    },
+    {
+      "prediction": "I think this should be `must_not_be` instead of `must_equal`",
+      "reference": "Try to use factory association declaration for default values, so it could minimal the line length. You may write in fetch_job factory like association :repository, factory: :repository."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "Should the \"cancelled\" class be added regardless of the proposal's state?"
+    },
+    {
+      "prediction": "I think we should remove this empty line.",
+      "reference": "nit: random whitespace change, let's try to keep these out of commits unless there are other useful changes."
+    },
+    {
+      "prediction": "I think this should be `var models = model.split(' ');` instead of `var models = undefined;`",
+      "reference": "Seems kinda dangerous - what if models length is different from paths length?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in `stageSenders`, but I'm not sure if it's the right place to do it. I think it should be in `stageSenders`, but I'm not sure if it's the right place to do it. I think it should be in `stageSenders`, but I'm not sure if it's the right place to do it. I think it should be in `stageSenders`, but I'm not sure if it's the right place to do it. I think it should be in `stageSenders`, but I'm not sure if it's the right place to do it. I think it should be in `stageSenders`, but I'm not sure if it's the right place to do it. I think it should be in `stageSenders`, but I'm not sure if it's the right place to do it. I think it should be in `stageSenders`, but I'm not sure if it's the right place to do it. I think it should be in `stageSenders`, but I'm not sure if it's the right place to do it. I think it should be in `stage",
+      "reference": "set real one plz (get it from DB)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. It seems like it would be better to do this in the `fpga_uafu_irq_set_update()` function, since that is the place where the `fpga_port_uafu_irq_set_update()` function is called.",
+      "reference": "Should this be initialized to zeroes?"
+    },
+    {
+      "prediction": "I think this should be `fcontext.base.size() == 0` instead of `q == null || fcontext.base.size() == 0`",
+      "reference": "The query is already built at this point, so I don't think this particular change actually helps wrt SOLR-10732? (and the `base.size()==0` case is already trivially optimized in `SolrIndexSearcher.numDocs(Query, DocSet)`)"
+    },
+    {
+      "prediction": "I think this should be `@Getter`, not `@Slf4jGetter`",
+      "reference": "re-add the type in the annotation here"
+    },
+    {
+      "prediction": "I think this should be `stderr` and not `stderr.String()` because `stderr` is a `bytes.Buffer` and `String()` on a `bytes.Buffer` returns the string representation of the buffer, whereas `stderr.String()` on a `bytes.Buffer` returns the string representation of the underlying buffer.",
+      "reference": "Could we simpify it as `cmd.Stderr = &bytes.Buffer{}`?"
+    },
+    {
+      "prediction": "I think this should be called `X509NotBeforeExpired` or something like that.",
+      "reference": "there's enough \"nots\" in here that while it's correct by my review, I'd like to see a small unit test (just passing in an expired and non-expired cert)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/convert/svc`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I don't think this is a good name for a parameter. It's not a tolerance, it's a relative error.",
+      "reference": "The config option should probably have a more specific name. Also it's not for all gridded data but specifically for Images (and its subclasses)."
+    },
+    {
+      "prediction": "I think this is the right thing to do, but I don't understand why we need to wait for the element to become visible before we click it. Is there a reason we can't just use `clickPreference(otherTheme);` and wait for the element to become visible?",
+      "reference": "Be careful. This is not a preference but an option in a dialog."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to set the cluster name. I think it should be done in the `Deployment` struct, but I'm not sure how to do it.",
+      "reference": "this will modify the incoming object, which may not be desirable in some cases. i think we should stick with the local var, override it with deployment.ClusterName if deployment.ClusterName not empty"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in the test.",
+      "reference": "Is this not required?"
+    },
+    {
+      "prediction": "I think this should be `if (input.type === 'value') {`",
+      "reference": "You're using scare quotes on these terms instead of defining them."
+    },
+    {
+      "prediction": "I'm not sure if this is the minimum version required by ddev, but I think it should be at least `1.10.0`.",
+      "reference": "These should both be const, not var right?"
+    },
+    {
+      "prediction": "I think it would be better to use `Files.newOutputStream()` instead of `FileOutputStream`",
+      "reference": "Can you catch a more specific exception here? Thanks :-)"
+    },
+    {
+      "prediction": "I think this should be `return nil, nil` instead of `return ctrl.Result{Requeue: true}, nil`",
+      "reference": "This is a bug during handling the recover logic in `common chaos` controller, I fixed in this request when I found it."
+    },
+    {
+      "prediction": "`score_thr` is not used in this function, please remove it.",
+      "reference": "During testing, we adopt the score threshold specified in the config file. Here the threshold is only used for visualization, and the variable name `score_thr` can be misleading. Renaming it to `show_score_thr` would be better."
+    },
+    {
+      "prediction": "I think this should be wrapped in a `<div class=\"crayons-layout\">`",
+      "reference": "can you make this `nav-primary` please? i'd like to roll with this since the `<ul>` class is `nav`, and primary is the name of the navigation menu"
+    },
+    {
+      "prediction": "I think this should be `context.calendar.displayname.replace(/\\s+/g, '-').replace(/[^\\w\\-]+/g, '').replace(/\\-\\-+/g, '-').replace(/^-+/, '').replace(/-+$/, '')`",
+      "reference": "@georgehrke Just out of curiosity. Couldn't you combine at least the combine the regex for '-' and '' with groups?"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to do this, but I don't have a better suggestion.",
+      "reference": "I think this is objectively better than looking up the object by name. As a follow-up, I'm going to see whether we do this name-based lookup somewhere else and switch to your method."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `statusButtonWidth` is calculated from `flyoutWidth_ - statusButtonWidth - marginX`, so if `statusButtonWidth` is larger than `flyoutWidth_ - marginX`, then `statusButtonX` will be smaller than `marginX - this.flyoutWidth_`. I think we need to calculate `statusButtonX` based on `statusButtonWidth` and `flyoutWidth_ - statusButtonWidth - marginX`, not `flyoutWidth_ - statusButtonWidth - marginX`.",
+      "reference": "Why are you dividing by scale in the LTR case but not the RTL case?"
+    },
+    {
+      "prediction": "I don't think we need to add this to the `COMMANDS` list, since it's not really a command. It's more of a `POST` request, right?",
+      "reference": "Update after command rename"
+    },
+    {
+      "prediction": "I think this should be `googlesitekit.ModuleSettingsDetails-${ moduleSlug }` instead of `googlesitekit.ModuleSettingsDetails-${ moduleSlug }-${ moduleSlug }`",
+      "reference": "I added this change to fix a bug with the legacy settings wrapper which was preventing it from working properly with the other modules"
+    },
+    {
+      "prediction": "I think this should be 1.0, not 2.0. The 2.0 version is for Azkaban v2.0.",
+      "reference": "isn't AZKABAN_FLOW_VERSION_2_0 more explicit?"
+    },
+    {
+      "prediction": "I think this should be a warning dialog, not a crash dialog.",
+      "reference": "Please don't leave an empty default"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "This isn't directly related, but the `import React from 'react';` below should be removed. We never import this directly and any modules from it we need are imported through `@wordpress/element`. This was probably added automatically at some point, but we also provide this automatically via `ProvidePlugin`."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it would be better to use `brew install` with the `--force` and `--with-libs` flags.",
+      "reference": "You can also use `echo: false` instead. Either is fine; I'm just pointing it out."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that the `BaseInitializeLayout` is called before `InitializeLayout` is called. This means that the `Delimiter` property will be set to `CsvColumnDelimiterMode.Auto` when the `BaseInitializeLayout` is called, but it will not be set to `CsvColumnDelimiterMode.Auto` when the `InitializeLayout` is called. This means that the `Delimiter` property will be set to `CsvColumnDelimiterMode.Auto` when the `BaseInitializeLayout` is called, but it will not be set to `CsvColumnDelimiterMode.Auto` when the `InitializeLayout` is called. This means that the `Delimiter` property will be set to `CsvColumnDelimiterMode.Auto` when the `BaseInitializeLayout` is called, but it will not be set to `CsvColumnDelimiterMode.Auto` when the `InitializeLayout` is called. This means that the `Delimiter` property will be set to `CsvColumnDelimiterMode.Auto` when the `BaseInitializeLayout` is called, but it will not be set to `CsvColumnDelimiterMode.Auto` when the `InitializeLayout` is called. This means that the `Delimiter` property will be set to `CsvColumnDelimiterMode.Auto",
+      "reference": "what's the reason for this moved line?"
+    },
+    {
+      "prediction": "I think we can remove the `showTitle` variable now, since we don't use it anymore.",
+      "reference": "I don't really follow what is happening here, but it looks like this _could_ conflict with the logic on the lines above... should this be an `else if`?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `GetSequencePoints` method.",
+      "reference": "`s => s.IsHidden == false` What's the purpose of adding this?"
+    },
+    {
+      "prediction": "I think it would be better to call this `clef-signer-addresses` or `clef-signer-endpoints` to make it clear that it is an array of addresses.",
+      "reference": "I would name this `clef-ethereum-address`. We already have a bunch of addresses in Bee, and people might wrongly think that this is yet another address"
+    },
+    {
+      "prediction": "I don't think this should be skipped.",
+      "reference": "you can just remove this"
+    },
+    {
+      "prediction": "I think we should rename this to `upgrade-cstor` or `upgrade-cstor-storage` or `upgrade-cstor-volume` or something like that so that it's not confused with the storage pool upgrade task.",
+      "reference": "I think we should avoid creating a dummy CR in the case of SPC. Please see if we can avoid this since we will not be patching anything in this CR."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this. I think it would be better placed in `app/views/products/_locked_features.html.erb`",
+      "reference": "Honestly not sure myself, but do you think it makes sense to move this conditional into the partial?"
+    },
+    {
+      "prediction": "I think this should be `args.basedir = tempfile.mkdtemp(prefix='qutebrowser-prefix-') or tempfile.mkdtemp()`",
+      "reference": "As discussed in IRC (just so it doesn't get lost): This probably should be `-basedir-`, not `-prefix-`"
+    },
+    {
+      "prediction": "I think this should be `Path.Combine(directory, basePath, \"Nethermind.Db.DBOnTheRocks\")`",
+      "reference": "basepath can be absoluta path and this needs to be supported"
+    },
+    {
+      "prediction": "This should be `const ScreenPosition&`",
+      "reference": "Didn't get to the commit message"
+    },
+    {
+      "prediction": "Layout/ExtraSpacing: Unnecessary spacing detected.<br>Layout/SpaceBeforeBlockBraces: Space missing to the left of {.<br>Layout/SpaceBeforeBlockBraces: Space missing to the left of }.",
+      "reference": "What about raising `ActionView::MissingTemplate` instead? That's what HighVoltage does."
+    },
+    {
+      "prediction": "This should be `balance.second.convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std::string> (balance.second.receivable).convert_to<std",
+      "reference": "Is it kept for compatibility?"
+    },
+    {
+      "prediction": "I think this should be `m.stateViewer.AccountSignerAddress(ctx, waddr)`",
+      "reference": "I would find it pretty reasonable to add a MinerSigner method on the state view that puts these together."
+    },
+    {
+      "prediction": "I think this should be `1090001` and not `1090002`",
+      "reference": "It's stored in `PodDBAdapter.VERSION`. I usually use the expected release version code for that change. As this will be released in AntennaPod 2.2.0, the code would be `2020000`."
+    },
+    {
+      "prediction": "I think this should be `fmt.Errorf(\"error executing CQL statement: %v\" + e.Error())`",
+      "reference": "What about method name itself? Do we run this for SQL too?"
+    },
+    {
+      "prediction": "I think this should be `instanceUrl = \"https://mobilesdk.cs1.my.salesforce.com\";` instead of `instanceUrl = \"https://test.salesforce.com\";`",
+      "reference": "`instanceUrl` should be `cs1.salesforce.com`. `communityUrl` would be `mobilesdk.cs1.my.salesforce.com`."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that `name` can be `NULL` in this case. I think you should do something like: `if (name == nullptr) { name = \"\"; *size = 0; } variableBase->m_Name.copy(name, *size);`",
+      "reference": "Let's use `if(name != nullptr)` to remove ambiguity"
+    },
+    {
+      "prediction": "I don't think we should add these methods to `NameFormatterDelegator` as they are not really related to `NameFormatterDelegator`. I think they should be moved to `NameFormatter`.",
+      "reference": "I wonder if it would make sense to have a separate namer for metadata? Metadata files tend to be expressed in a different language from the repo language, and have mutually exclusive concepts. So, `PackageMetadataNamer`."
+    },
+    {
+      "prediction": "I think it would be better to have this as a constructor parameter.",
+      "reference": "This rule is quite specific to qbft and so I don't think it should be part of the common rules. Would rather the common bft code didn't know anything about contract based voting/validator governance."
+    },
+    {
+      "prediction": "I don't think this is the right place to add this method. I think it would be better to add it to `Traversable` or `Function1`.",
+      "reference": "This makes our life harder than it needs to be. A Set is unordered by definition. Please remove this method from Set and add it to SortedSet, but with no default implementation. We need to duplicate it then to LinkedHashSet but that's ok, it is the best we can do. Could you please add it also to SortedMap and LinkedHashMap? We are then able to close the original issue."
+    },
+    {
+      "prediction": "I think this should be `if response_headers is not None and len(response_headers) > 0:`",
+      "reference": "we can replace `for` with `params.update(response_headers)`"
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "reference": "Any chance this is available in .NET Core 2.0 now?"
+    },
+    {
+      "prediction": "nit: I think it would be better to use `peerstore` instead of `p2p` here.",
+      "reference": "I wonder if it's necessary to let app pass in the context. What app level context should be sent? If no, is it enough for p2p agent to compose a context with network info there?"
+    },
+    {
+      "prediction": "Why is this needed? The profiling is already done in `lbann/utils/profiling.hpp`.",
+      "reference": "Why do we need both includes here?"
+    },
+    {
+      "prediction": "This should be `cursor.getCount()` instead of `cursor.moveToFirst()`, right?",
+      "reference": "SmartStoreInspectorTest expected results in a certain order"
+    },
+    {
+      "prediction": "I think this should be `m_using_random_node.count(m_io_thread_pool->get_local_thread_id())` instead of `m_io_thread_pool != nullptr && m_using_random_node.count(m_io_thread_pool->get_local_thread_id())`",
+      "reference": "I think that you need something like `m_using_random_node.emplace(m_io_thread_pool->get_local_thread_id());`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to check this. I think it should be in the same place as `isUsedLocalVarWithoutInitializer`.",
+      "reference": "I don't think we need to maintain a separate set. The problem here is that the \"assignment\" that is killed for this variable is not really an assignment. If we just don't `assign` the variable with the non-existent value, it will not be reported. I pushed a fix."
+    },
+    {
+      "prediction": "I don't think we need this method. We can just return `null` from `#isCaseSensitive()` and it will automatically return `false` from `#targetSplitSize()`.",
+      "reference": "Let me know if this is too pervasive. It is currently a private method in `BaseTableScan`. It seems both `SparkBatchQueryScan` and `SparkMergeScan` need to know the scan-specific split size when planning tasks. Therefore, I made it open. Another approach is to move all the `planTasks` logic to scan implementations, but for the combine tasks by partition feature, it requires grouping scan tasks by partition first, instead of returning them in an iterator fashion. I'm not sure if this is OK. `SparkMergeScan` also seems to re-implemented its own plan tasks logic."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could get it to work.",
+      "reference": "Not sure what is this change doing?"
+    },
+    {
+      "prediction": "I'm not sure if we should have a default value for `mins`. If we don't have a default value, then it's possible for someone to try to use a magic number like 90 and get a ticket.",
+      "reference": "this is funky. either don't take` minutes` as keyword arg or take both `minutes` and `days` and pass all of them on to `timedelta`. i'm guessing you're aiming for backwards compatibility, but i don't think it's worth it given how confusing this is. atlernatively, make `mins=60*24*30` the default. and that brings me to another issue, `mins=None` is not a very useful default."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `AbstractRememberMeServices` since that's the only place where this method is called.",
+      "reference": "Please use a tab for indentation instead of spaces."
+    },
+    {
+      "prediction": "Why is this not `final`?",
+      "reference": "I can't believe we didn't see this typo before :))"
+    },
+    {
+      "prediction": "I think this should be `LOGGER = logging.getLogger(__name__)` instead of `LOGGER = log_util.get_logger(__name__)`",
+      "reference": "If the logger isn't used, it probably doesn't need to be added."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "These files should in `/javascript/brython`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with that doesn't break anything.",
+      "reference": "minor, you can squash these into one `if`"
+    },
+    {
+      "prediction": "I think we can remove this test now, since it's no longer needed.",
+      "reference": "Used regexes to do most of this, hence different methods of string creation of UUIDs. I plan to follow up with an additional change to remove direct references to google/pborman UUID so `uuid.New()` and `uuid.NewRandom()` will instead use our `primitives.UUID`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to make this change. I think it should be moved to the top of the `if` block above.",
+      "reference": "Add a TODO here to revisit if this was problematic."
+    },
+    {
+      "prediction": "`reg_decoded_bbox` -> `reg_decoded_bbox`",
+      "reference": "Note generally it -> It"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Is this relevant to this PR?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Can we instead ensure that `opts[:run_in_parellel]` will always be a boolean? Otherwise we'll have to account for the case when it's a non-boolean value in multiple places, such as any/every other `if` statement."
+    },
+    {
+      "prediction": "I'm not sure if we want to make this part of the `HTTP` namespace. I think it would be better to have a separate namespace for `HTTPS` and `HTTP` so that we don't have to repeat the `HTTP` namespace in the `HTTPS` case.",
+      "reference": "Is this just moved out of the other files? It doesn't seem like this was generated like the other files."
+    },
+    {
+      "prediction": "I don't think we need this interface. We can just use `HttpContent.CopyTo` and `HttpContent.CopyToAsync` directly.",
+      "reference": "I started out doing everything `async`, but it was getting in the way of debugging, so I switched back to all synchronous until I got things working. We should probably move everything back to `async` and remove the synchronous versions."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "this looks already too complicated..."
+    },
+    {
+      "prediction": "I think this test can be moved to rocksdb/test/test_rocksdb_compaction_test.cpp",
+      "reference": "As for the testings, we better to verify the actual effects of the compaction. Of course, you could do it in future."
+    },
+    {
+      "prediction": "I think we should remove the `isView` check from `loadTables` as well.",
+      "reference": "should we add a TODO here?"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "This can be inlined below as `GoogleSitekitDashboardDetails` is an unnecessary wrapper now."
+    },
+    {
+      "prediction": "I think this should be `NewVHs(versionHistory)`",
+      "reference": "NewVHS -> NewVersionHistories this one looks better"
+    },
+    {
+      "prediction": "I think this should be `@executable` instead of `executable_path`",
+      "reference": "Is this `@missing_text` ?"
+    },
+    {
+      "prediction": "I think this should be `r\"([\\w-]+/[\\w-]+)(?:/(.+))?$\"`",
+      "reference": "Minor suggestion, but wouldn't it be cleaner to simply replace PACKAGE_NAME_FORMAT to all the optional path, then check that the path is empty in validate_package_name? We might also want a helper function to pull out the package name and sub-package path."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the i18n string, but it's the only way I could get to work.",
+      "reference": "we should have a way to insert components into a i18n string. so that we could define a translation string like `'%{filename} is being edited'`, and inject the filename `<span/>` at the right place. I imagine there must be a language where the `Editing` text should come after the file name. we can think about that later tho."
+    },
+    {
+      "prediction": "I think this should be `if (pglob.gl_pathv[i] == NULL)`",
+      "reference": "Probably want break instead of continue."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get this test to pass.",
+      "reference": "This was on purpose that I would set logging (NoLog=false) and a dummy logger below. The idea is to have better code coverage and test of debug/trace statements. Any reason you have removed this?"
+    },
+    {
+      "prediction": "I think this should be `Stream.of(\"\")` instead of `Stream.<String>empty()`.",
+      "reference": "does adding a `Stream.<String>empty()` help here?"
+    },
+    {
+      "prediction": "I think this should be `{'name': 'non_exist_scanner', 'enabled': True}` instead of `{'name': 'bigquery', 'enabled': True}`",
+      "reference": "This is not being used anymore, so can be removed."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get this to work.",
+      "reference": "Is it called some where?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Please remove this extra blank line."
+    },
+    {
+      "prediction": "I think this should be `@Input` and not `@InputFiles`",
+      "reference": "this is really the key, if you don't set this then gradle doesn't wire up the tasks that produce various files inside these configurations as inputs... I think we were just getting lucky before."
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of an empty string.",
+      "reference": "Sorry if I'm missing something here, but shouldn't this message be about failing to match with an *empty* selector, and/or perhaps even an empty *tag* instead of selector? The message seems a bit disconnected from what we're testing?"
+    },
+    {
+      "prediction": "I think this should be `using System.IO;` instead of `using Microsoft.TestPlatform.VsTestConsole.TranslationLayer.Payloads;`",
+      "reference": "nit: please move these inside namespace."
+    },
+    {
+      "prediction": "I think this flag should be called `payments.enabled` instead of `payments.during-session-debug`",
+      "reference": "may users try to abuse it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `put` method of the `ByteBufferFactory`.",
+      "reference": "why is this space removed?"
+    },
+    {
+      "prediction": "Why is this file not affected by `go:build windows`?",
+      "reference": "Can we remove this other line now then?"
+    },
+    {
+      "prediction": "I think this should be `cmdhistory.py` instead of `cmdhistory.py`",
+      "reference": "There's a comma missing here at the end :wink:"
+    },
+    {
+      "prediction": "I think this should be renamed to `PublishFlatContainers` or something similar.",
+      "reference": "Why is this not required any longer?"
+    },
+    {
+      "prediction": "I don't think we need to import `github.com/Unknwon/paginater` here.",
+      "reference": "Indentation seems a bit fucked up here. Run `go fmt`."
+    },
+    {
+      "prediction": "I think this should be `nil, status.Error(codes.InvalidArgument, \"empty filter\")`",
+      "reference": "any chance in == nil? same below"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the `listen_fd` and `h2o_sysfn(listen, listen_fd, H2O_SOMAXCONN)` calls in the same order as they were before.",
+      "reference": "We would change the source code of h2o so that certain syscalls will be invoked through the `h2o_sysfn` macro."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this check. I think it should be in `GetQueueSubmitVUID`, but I'm not sure if there is a better place to put it.",
+      "reference": "the name kTimelineCannotBeSignalled is confusing. It's not that a TimelineSemphore cannot be signaled... it's the \"VK_KHR_timeline_semaphore is enabled *variant* of the \"binary cannot be signaled\" message. We should probably have a consistent naming scheme to clarify. kBinaryCannotBeSignalledAltTimeline or hide the complexity in Get*VUID functions as needed somehow (pass in the extension structure?)"
+    },
+    {
+      "prediction": "I don't think we should add a `describe` method to `pandas` at this point. It's not clear to me why we'd want to do this, and it's not clear to me why we'd want to call it `describe` at all.",
+      "reference": "Could you add a type hint? `percentiles: Optional[List[float]] = None`"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "Maybe putting ones they've used before first would be good, but this is fine for now."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "I don't think we need these. Simpler just to put `routetable.TargetType...` inline below."
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to do, but it seems like it should be removed.",
+      "reference": "Why this instead of the old way?"
+    },
+    {
+      "prediction": "This should be `if (RuleB(Pets, LivelikeBreakCharmOnInvis) || (!RuleB(Pets, LivelikeBreakCharmOnInvis) && IsInvisible(formerpet))) {`",
+      "reference": "Just combining the inner if/else blocks like this is what I originally meant (I didn't mean replacing all the invisible checks in the outer if, I realize that would have been a functionality change regarding the rule) This should now be equivalent to `if (RuleB(Pets, LivelikeBreakCharmOnInvis) || IsInvisible(formerpet))` (the second check for the rule being false after the `||` would be implied, anything after the `||` would only be evaluated if the rule was false)"
+    },
+    {
+      "prediction": "I think this should be `strings.Replace(key, \"_\", \"\", -1)` instead of `strings.ReplaceAll(key, \"_\", \"\")`",
+      "reference": "We don't want to write this back to the definition do we? Probably just want to create a local variable for it."
+    },
+    {
+      "prediction": "Shouldn't this be done in the `onPlaybackEnded` method?",
+      "reference": "Good catch! Could you please move this to `PlaybackService.this.onPlaybackEnded`? I think it's more clean if everything is in one single place."
+    },
+    {
+      "prediction": "I think this should be done in `LXQtApplication` instead of `LXQtTaskButton`. The `LXQtApplication` is the parent of `LXQtTaskButton`.",
+      "reference": "We're trying to avoid X.org specific code. So I'm not a fan of this. Can't this be achieved in display server agnostic way?"
+    },
+    {
+      "prediction": "This file should not be included in this PR.",
+      "reference": "This class can be removed?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the dispatcher.",
+      "reference": "To minimize the possibility of regret, perhaps we name this `ZapLogger`."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "This looks a little weird, but it's a feature of the gem we're using. You can give it a symbol representing the method you want called on current_user."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It should be in the `anomaly` module.",
+      "reference": "`Anomaly` imported here, but not used. Please run the changed files through pylint with nupic pylint config and fix pylint findings related to your changes."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "We needed this testutils file after all"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it's the only way I could get it to work.",
+      "reference": "line is 161 characters (from `lll`)"
+    },
+    {
+      "prediction": "Why is this file changed?",
+      "reference": "Please re-format/groupimport this line. You might need a little manual work."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `ApplyToWorkflow` and `ApplyWorkerCustomizations` methods of the `DaisyWorker` interface.",
+      "reference": "minor, you can just use `wf` instead of `worker.wf`"
+    },
+    {
+      "prediction": "I think you can remove the `\"\\n\" + \"    \\\"rampStage\\\": 2,\\n\" + \"    \\\"lastUpdatedTime\\\": 1566259437000\\n\" + \"  }\\n\"`",
+      "reference": "I would suggest putting this in the resources directory with the same package as of this test class. Then you can utilize the method azkaban.utils.TestUtils#readResource to read it as string."
+    },
+    {
+      "prediction": "I think we can remove the `upgrade` package now.",
+      "reference": "could not import github.com/openebs/maya/pkg/upgrade/templates (invalid package name: \"\") (from `typecheck`)"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a strong opinion.",
+      "reference": "a official -> an official"
+    },
+    {
+      "prediction": "I think this should be `config.KINESIS_PROVIDER in ('kinesalite', 'kinesis')` instead of `config.KINESIS_PROVIDER == 'kinesalite'`",
+      "reference": "I only want the proxy request for this to run for kinesalite."
+    },
+    {
+      "prediction": "I think this should be `this.props.dateFormat` and not `this.props.dateFormat`",
+      "reference": "White space boya?"
+    },
+    {
+      "prediction": "Nit: `cloudFormation.StackSummary` is a bit of a mouthful. Maybe `cloudFormation.StackSummary` or `cloudFormation.StackSummaryType` or something?",
+      "reference": "Maybe ditch these since `StackSummary` is not used anymore."
+    },
+    {
+      "prediction": "Why do we need to ignore index 255?",
+      "reference": "suggest to indicate the value range & meaning in docstring"
+    },
+    {
+      "prediction": "I think this is a good idea, but I'm not sure it's the best way to do it. I think it's better to have a single place where we can define the endpoints of the exporter, so that we don't have to repeat the same code for each endpoint. One option could be to have a struct that defines the endpoints of the exporter, and then have a function that creates an instance of this struct. The function could take the configuration for the exporter as input and return a new instance of the struct. This way, we could have a single place where we can define the endpoints of the exporter, so that we don't have to repeat the same code for each endpoint. One option could be to have a struct that defines the endpoints of the exporter, and then have a function that creates an instance of this struct. The function could take the configuration for the exporter as input and return a new instance of the struct. This way, we could have a single place where we can define the endpoints of the exporter, so that we don't have to repeat the same code for each endpoint. One option could be to have a struct that defines the endpoints of the exporter, and then have a function that creates an instance of this struct. The function could take",
+      "reference": "not sure if it works with prometheus by default, otherwise we will have to add this into prometheus config also"
+    },
+    {
+      "prediction": "I think this should be `r.epmm.RemoveEndPointMark(ifaceName)`",
+      "reference": "The rule renderer isn't meant to be stateful so probably best to move this out of here"
+    },
+    {
+      "prediction": "I think this should be `findNvidiaContainerRuntimes(nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil",
+      "reference": "passing in a `nil` here just so that the tests can pass in an alternative implementation seems weird, but I don't know what the convention is for doing something like this - @briandowns?"
+    },
+    {
+      "prediction": "I think we can remove the `require('html-entities').AllHtmlEntities` import.",
+      "reference": "IMO `dot` should be preferred. Remember `aXe`?"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "Not a huge deal but this will block tests, also, don't they stay open by default now?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just use the default log level for telemetry.",
+      "reference": "Could you make it a local constant for now ?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for `int` vs `str` casts. I think we should be checking for `isinstance(value, (int, six.string_types)`",
+      "reference": "What is this `collection_id` field here?"
+    },
+    {
+      "prediction": "I think this should be `if self.interactive:` since we don't want to print debug messages in interactive mode.",
+      "reference": "Is there a reason why we don't use logging for this?"
+    },
+    {
+      "prediction": "Why not using the parent class instead of adding a new one?",
+      "reference": "Exceptions should be placed in application layer -> infrastructure is aware of application - not the other way around"
+    },
+    {
+      "prediction": "I'm not sure what this is testing, but it seems like it should be `(\"message\", \"Line too long (2/2)\")`.",
+      "reference": "IIRC issue mentioned reporting _symbolic message_ - so in this case it would be `line-too-long`."
+    },
+    {
+      "prediction": "This should be `else if` instead of `else if (id == \"feedbackable\") { return spells[spell_id].reflectable; }`",
+      "reference": "Is this intended to be `spells[spell_id].reflectable` here? Edit: Should be `spells[spell_id].feedbackable`, right?"
+    },
+    {
+      "prediction": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "How should we set this for existing teams?"
+    },
+    {
+      "prediction": "I think this test should be moved to `test/install.go` since it's not related to the `InstallManager` portion of the test.",
+      "reference": "It would be nice if we could verify that this is in fact the error we expected. But that's a latent issue, something for the backlog."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "I hate then some internal process specific middleware leaks into connection manager :( todo later."
+    },
+    {
+      "prediction": "I think this should be `nupic` and not `nupic.support`",
+      "reference": "Reflects where these files were moved for `pkg_resources`."
+    },
+    {
+      "prediction": "I think this is the only thing that needs to change in this file. Everything else should remain the same.",
+      "reference": "This should be computed('i18n.locale'...."
+    },
+    {
+      "prediction": "I think this should be done in `eth20` instead of `eth0`",
+      "reference": "Can you explain to me what goes wrong here? Can the test be adjusted to set up working routing instead?"
+    },
+    {
+      "prediction": "I think we should revert this change.",
+      "reference": "We can get rid of this import then."
+    },
+    {
+      "prediction": "I think this should be `mapEventToMetric` and not `mapEventToMetric`",
+      "reference": "Not sure how useful is unlock event."
+    },
+    {
+      "prediction": "I think this should be `encoder{lowercaseFields: lowercaseFields}` instead of `e: encoder{lowercaseFields: lowercaseFields}`",
+      "reference": "Consider make the second argument a `encoderOptions` which includes the `lowercaseFields` just like opening a collection."
+    },
+    {
+      "prediction": "I think it would be better to move this to the `HiveIcebergOutputCommitter` constructor, so that it can be called from the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method, instead of the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method. This way, the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method can be called from the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method, instead of the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method, instead of the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method, instead of the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method, instead of the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method, instead of the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method, instead of the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method, instead of the `HiveIcebergOutputCommitter#commitTask(TaskAttemptContext)` method, instead of the `HiveIce",
+      "reference": "What is the value of renaming this to `ctx`? We generally prefer the longer names because they are easier to read and to type."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to fix this, but it seems to be the only place where we check if the tag is uturn.",
+      "reference": "Hm. This could be a case of left-sided driving, having u-turn lanes on the right side. Good catch, but looks good to me."
+    },
+    {
+      "prediction": "I think it would be better to have a separate class for `OK` and `Fail` instead of using `Result` as a base class.",
+      "reference": "There was a [JsonPropertyName(\"success\")] here it did serialize fine for me, so I am not sure what was the matter? Also maybe uppercase?"
+    },
+    {
+      "prediction": "I think this should be `IAuRaValidatorProcessor`, not `IAuRaAdditionalBlockProcessorFactory`.",
+      "reference": "we can just use read env here, pretty sure"
+    },
+    {
+      "prediction": "I think this hook should be called after `onAfterLocalTransformStart`",
+      "reference": "Could you replace local hook with global hook `afterModifyTransformStart`?"
+    },
+    {
+      "prediction": "I think this should be `self._files = {}`",
+      "reference": "You seem to be basing your PRs off an commit, which keeps reintroducing these changes, which makes it harder to review your PRs."
+    },
+    {
+      "prediction": "I think this should be `assert(typeof opt_port !== 'function', `",
+      "reference": "Maybe it would simpler to ignore opt_port if type !== 'number'?"
+    },
+    {
+      "prediction": "I think this should be `IEnumerable<KeyValuePair<string, object?>>`",
+      "reference": "As far as I can tell, `IDictionary<string, object>` implements `IEnumerable<KeyValuePair,string, object>>` so we are just making it more generic. From the issue, I understood that we want to maintain sequence/order. I believe `IEnumerable<>` won't fix the issue."
+    },
+    {
+      "prediction": "I think this should be `LOG.info(String.format(\"Mapping %s to docker image %s %d times\", caps, name,```",
+      "reference": "This change prevents a user understanding how their server is configured. Best to leave at `info` level."
+    },
+    {
+      "prediction": "nit: I think we can drop the `api` prefix here since it's not used in the function signature.",
+      "reference": "I will add the following to the list: - api.SpecScale - api.SpecSticky"
+    },
+    {
+      "prediction": "I think this should be 10043",
+      "reference": "As defaultSendPort is being removed, and I don't like TODOs on examples, please remove the whole line"
+    },
+    {
+      "prediction": "I think this test is still valid, but I think it should be moved to the `resources` folder.",
+      "reference": "since we are now no longer expecting an input of float to return an empty string, i have removed this test case."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to register the counter. I think it should be in the `Counter` struct.",
+      "reference": "It would be cleaner if you were to pack the name&desc in a `MetricName` and pass it to `MakeCounter`"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it in the PR description.",
+      "reference": "Do you know why do we need this?"
+    },
+    {
+      "prediction": "I think this should be `using MvvmCross.Platform.iOS.Presenters;` instead of `using MvvmCross.iOS.Views;`",
+      "reference": "I guess this using is what causes the build to fail"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `pony_unregister_thread()`.",
+      "reference": "I like the explicit unregister approach."
+    },
+    {
+      "prediction": "I think we can remove `ExponentialHistogram` and `ExponentialHistogramWithMinMax` now, since they are unused.",
+      "reference": "We should able to use 0x30 as summary, and make 0x80 reserved for future."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think we should be able to get rid of `HttpBypassHelper.UriContainsAnyOf` and just use `HttpBypassHelper.IsHttpBypassEnabled` instead.",
+      "reference": "This allows strategic exclusion of http spans."
+    },
+    {
+      "prediction": "I think it would be better to have this as an optional parameter and set it to `null` by default. This way you don't have to check for it in the constructor.",
+      "reference": "why not just `string[]` instead of nullable"
+    },
+    {
+      "prediction": "I think this should be `\"-tags=` + strings.Join(e.BuildContext.BuildTags, \",\")`",
+      "reference": "How does this work? Why isn't each tag after the first picked up as a new arg/flag?"
+    },
+    {
+      "prediction": "I think this should be Datadog-AzureAppServicesResourceId.",
+      "reference": "This key is actually yet to be determined. Meeting with the backend team and Garner to discuss."
+    },
+    {
+      "prediction": "I think this should be `assert(!!iv_state);` instead of `assert(iv_state);`",
+      "reference": "Why remove the crash protection? We're just going to get a bug filed on it."
+    },
+    {
+      "prediction": "I don't think we need to add CSP here.",
+      "reference": "Maybe be consistent and name it `FRAME_OPTIONS`."
+    },
+    {
+      "prediction": "I'm not sure why this is here, but I think it should be removed.",
+      "reference": "This check/exception is a guard for the next few lines of code. It shouldn't be necessary now. In fact, I don't think you need the if (parentConnection.type == Blockly.INPUT_VALUE) branch at all."
+    },
+    {
+      "prediction": "I think this line should be indented by 4 spaces to match the rest of the file.",
+      "reference": "Could probably move these constants too?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't have a better idea at the moment.",
+      "reference": "Maybe 'failed' instead of 'missing'. Also do we really need to write ticket numbers here?"
+    },
+    {
+      "prediction": "This import is unused.",
+      "reference": "Test used to use old bouncycastle dependency which isn't needed anymore from Hadoop. Switched to use builtin Java split."
+    },
+    {
+      "prediction": "I think this should be `v.name.encode(\"UTF-8\")` instead of `v.name.decode(\"UTF-8\")`",
+      "reference": "nit: I think this would be more readable as \"Some language names contain Unicode characters\"."
+    },
+    {
+      "prediction": "This file is covered by the GNU General Public License. See the file `COPYING` for more details.",
+      "reference": "Thanks for adding the header, but I don't think 2017 is a very accurate guess here. Could you do a quick search with git blame and change this accordingly?"
+    },
+    {
+      "prediction": "nit: you can remove the `/* eslint-env browser, mocha */` statements here.",
+      "reference": "Oh interesting - does web-test-runner not resolve package.json files?"
+    },
+    {
+      "prediction": "I think this should be `P\u00e4iv\u00e4`, not `P\u00e4iv\u00e4si`.",
+      "reference": "Is there a way to derive this value from the locale itself? I feel hardcoding the prefix in the props isn't the right approach, but I'm not sure what the locale file contains exactly."
+    },
+    {
+      "prediction": "I don't think we should warn about this.",
+      "reference": "I'd much rather see these warnings in our devtools (`debug/index.js`). Strings contribute quite a bit to our file size and moving them there would prevent bloating core."
+    },
+    {
+      "prediction": "I think this should be `SmartStore.SOUP_NAME_COL, SmartStore.PATH_COL, SmartStore.COLUMN_TYPE_COL, SmartStore.SOUP_INDEX_MAP_TABLE` instead of `SmartStore.SOUP_NAME_COL, SmartStore.PATH_COL, SmartStore.COLUMN_TYPE_COL, SmartStore.SOUP_INDEX_MAP_TABLE`",
+      "reference": "Use `String.format(Locale.US, ...) to avoid the `Lint` warning."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/apis/core/internalclientset/typed/servicemanagement/v1beta1/internalclientset/typed/servicemanagement/v1beta1/internalversion`",
+      "reference": "nit: ServiceReference represents a reference to a v1.Service."
+    },
+    {
+      "prediction": "I think this should be `[]string` instead of `[]string`",
+      "reference": "since it changes to the array, how about change the name to `DeviceNames`"
+    },
+    {
+      "prediction": "I don't think we need to have a special property for runtime. We can just use `runtime` as the name of the property.",
+      "reference": "Can we avoid this. upgrade engine code should take care of this."
+    },
+    {
+      "prediction": "I think this should be a `dict` and not a `list`",
+      "reference": "should also be alphabetized or at least grouped and alphabetized within group"
+    },
+    {
+      "prediction": "I think this should be `# pylint: disable=unused-variable`",
+      "reference": "No need for this, as pylint already only turns things off for this function and it's needed for the entire function."
+    },
+    {
+      "prediction": "I think this should be `errUnmarshalBuildOpts` as well.",
+      "reference": "nit: can we remove \"can't\" from the error message?"
+    },
+    {
+      "prediction": "Is it possible to have an empty status in the controller? If so, should we set the phase to recreate?",
+      "reference": "instead of modifying 'Status' which impacts the state diagram, how about using some annotations on CVR? This can probably help for our upgrade as well. Is this possible? cc: @AmitKumarDas"
+    },
+    {
+      "prediction": "I think we should use `pkg` instead of `yum` here.",
+      "reference": "Rename this as KubernetesGPGURL - >KubernetesUbuntuGPGURL , Modify it where ever it is used."
+    },
+    {
+      "prediction": "This should be `var DEFAULT_FEE = 0.00000000000000000000000000000000`",
+      "reference": "I'm not totally sure I like the idea of putting a default fee here, but I guess so long as we make sure to track the default fee of bitcoin core, we're good. We should really also take a look at the byte size of the transaction, and use that to determine what the fee should be. But that could also be a different PR."
+    },
+    {
+      "prediction": "small>linear</small></td><td><small>const</small></td><td><small>linear</small></td><td><small>linear</small></td><td><small>linear</small></td></tr>\n  * </tbody>\n  * </table>\n```\nReview comment?",
+      "reference": "oh yeah, `tail`, `update`, `prepend`, `append`, `subSequence`, `trim` just became effectively constant (without memory leak) :)"
+    },
+    {
+      "prediction": "I think it would be better to use `is_array` instead of `is_object` here.",
+      "reference": "Can you add back a deprecated `destroyLinks` method for backward compatibility? It can simply proxy `destroyResourceLinks`, and we can remove it in the next major release."
+    },
+    {
+      "prediction": "I think this should be `RegisterGlobalOption(\"servers\", parseServerOptions)` and not `RegisterGlobalOption(\"servers\", servers)`",
+      "reference": "A reminder that we should discuss whether to rename this to \"sockets\" or \"listeners\"."
+    },
+    {
+      "prediction": "Is this needed? I don't see it being used anywhere.",
+      "reference": "Why do you need to include this using declaration?"
+    },
+    {
+      "prediction": "I think this should be `scan < name.Length`.",
+      "reference": "@halter73 where is this file generated from?"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "What do these `#if TRUE` accomplish?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `OnNext` method of the `InstrumentationContext` class.",
+      "reference": "Probably a good optimization to have this, but I don't think it's strictly required. For \"legacy\" Activity flows we pass them through an ActivitySource for sampling. So the logic below should also catch these?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `mailer` module.",
+      "reference": "@jessieay Intentionally moving away from passing a param?"
+    },
+    {
+      "prediction": "I think we should remove the `trim` here.",
+      "reference": "I think you could condense this code considerably by getting rid of the initialization and foreach loop and simply saying: `$firstname = trim(implode(' ', $nameArr));` What do you think?"
+    },
+    {
+      "prediction": "Why remove this?",
+      "reference": "hmm, we actually should _add_ `'use strict'` everywhere"
+    },
+    {
+      "prediction": "I think this should be `logger.warn(\"Project \" + projectId + \" version \" + projectVer + \" property \" + propsName + \" is empty.\", e);`",
+      "reference": "it's not just fetching override properties here, right?"
+    },
+    {
+      "prediction": "I think it would be good to log the reason the connection was closed as well.",
+      "reference": "Is it worth logging the error?"
+    },
+    {
+      "prediction": "I don't think this is the best way to do this. I think you should be able to do something like `ns[\"foo\"] = {\"bar\": \"baz\", \"qux\": \"boo\"}` and then have `ns[\"foo\"][\"bar\"]` and `ns[\"foo\"][\"qux\"]` without needing to do `ns[\"foo\"][\"bar\"] = \"baz\"` and `ns[\"foo\"][\"qux\"] = \"boo\"`.",
+      "reference": "What's the point of this class?"
+    },
+    {
+      "prediction": "Why do we need to check for mixed content? Shouldn't it always be either a movie or a collection?",
+      "reference": "I'm being picky here, but I think I would prefer `toggle` to be kept here and just change the condition to `contentType !== 'movies' && contentType !== 'mixed'` or `!['movies', 'mixed'].includes(contentType)`."
+    },
+    {
+      "prediction": "I think we should use `github.com/syndtr/goleveldb/leveldb` instead of `sysPath`.",
+      "reference": "Please put this down with the other \"github.com\" imports."
+    },
+    {
+      "prediction": "I think this should be `push_front` instead of `push_front_preference`.",
+      "reference": "Minor but probably want `push_front_preference` to have a trailing `_a` to be consistent with the other parameter"
+    },
+    {
+      "prediction": "I think this should be `Projections.inclusive(spec, caseSensitive).project(rowFilter)` instead of `Projections.inclusive(spec).project(rowFilter)`",
+      "reference": "Nit: the formatting is a bit off. I would expect this to be 2 indentations from `this.expr = ...` but this is 8 spaces, not 4. Also, most places leave the `);` on the last line of the method arguments."
+    },
+    {
+      "prediction": "I think we can remove the `StaticLangApiMethodView` class altogether, and just use the `SimpleParamDocView` class.",
+      "reference": "Ack, @chrisdunelm snuck this class extension past me. This is not a pattern I want to have used..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this capability. I think it should be in `kinto/__init__.py` under the `kinto.views` module.",
+      "reference": "nitpick: I wonder if we should name it `flush` only (?)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to have a separate metric for the number of currently running tasks and a separate metric for the number of killed tasks. This way we can easily see how many tasks are currently running and how many have been killed.",
+      "reference": "Why gauges, not counters?"
+    },
+    {
+      "prediction": "I think we can remove the sysHashLen variable and use `sysHashLen = 6` everywhere.",
+      "reference": "Just making a note here that this may break pre GWs between pre 2.2.0 and 2.2.0 servers. Not sure, will have to experiment/dig a bit more."
+    },
+    {
+      "prediction": "I think this should be \"Pod Failed\" instead of \"Pod Failure\".",
+      "reference": "Shall we rename this status to CONTAINER_FAILED? ^^ cc: @sshardool"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this interface. I think it should be in `pkg/network/fetcher.go` since it's only used by `pkg/network/fetcher.go`.",
+      "reference": "Thinking out loud here: I believe the only reason we need `peer.ID` as a parameter to this method is because we are not persisting blocks from pubsub (instead we are refetching them). If nodes persist the blocks they received over pubsub then I think we can guarantee peers we share a connection with (i.e. that are in the peer tracker) will always have the blocks we are fetching (else how would have we gotten the pubsub message). I imagine the concept of an `initialPeer` can go away once #2962 lands since the peers in the tracker will have the block. Although I don't think that covers the case of fetching blocks our peers have deemed semantically invalid and thus not stored..."
+    },
+    {
+      "prediction": "I think it would be better to add a link to the CEL docs here, rather than the Cel spec.",
+      "reference": "I know it has no effect but my eyes can't help. Is that line not too long?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Does that mean we disable service loop prevention for packet generated by local host?"
+    },
+    {
+      "prediction": "I think this should be `raise RuntimeError, \"cygwin is not installed on #{host}\" if host.cygwin_installed?`",
+      "reference": "It seems a little odd to have both `host.is_cygwin?` *and* `host.cygwin_installed?` defined (with a possibility of having `is_cygwin?` be `true`, but `cygwin_installed?` returning `false`). Do the docs clearly explain the difference?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. If you have a v1 manifest with a sequence number 1 and an app manifest with a sequence number 2, then the app manifest will overwrite the sequence number of the v1 manifest. Is that what you want to do?",
+      "reference": "Does this mean manifests will be written with the v2 schema (i.e. with sequence numbers) even though `TableMetadata` is v1 and the manifest list is written with v1? And this should work because we do a projection on read and sequence number is optional?"
+    },
+    {
+      "prediction": "I think this should be called `rewrittenFileScanTaskSetId` instead of `rewrittenFileScanTaskSetID`.",
+      "reference": "Should we be sharing this property key with the read? Maybe it should be belong to the file-scan-task object itself?"
+    },
+    {
+      "prediction": "I think this should be `Debug` and not `DebugFields`.",
+      "reference": "Nice! Good idea to move these out of the main struct."
+    },
+    {
+      "prediction": "I think this should be `err != nil && !sigFound`",
+      "reference": "Thanks for paying attention to other places this might happen. This one is particularly important; I probably never should have gotten in the habit of CheckErr(), since it does a log.Panic() explicitly, which looks like something else until you look closely. It's supposed to be used places where \"can't happen\" but Things Can Happen."
+    },
+    {
+      "prediction": "I don't think this is a good name for this field. It's a bit confusing because IMAP and SMTP are both transport types, and IMAP is the default. I think it would be better to call this `type` or `transportType` or something similar.",
+      "reference": "Converting this to an enum makes it obvious that I combined things that don't really belong together. It would probably be better to create two enums `StoreType` and `TransportType` (in more appropriate locations). That also makes it necessary to have (at least) two `ServerSettings` classes. `IncomingServerSettings` and `OutgoingServerSettings`. Also, the naming is inconsistent. Maybe it would be better to make the string that is used for import/export explicit, e.g. WEBDAV(\"WebDAV\")."
+    },
+    {
+      "prediction": "`@memberof axe.commons.utils` should be `@memberof axe`",
+      "reference": "It's out of scope for this PR, but I don't find this utility's name to be particularly intuitive. It speaks to nothing of what it does. Does it create a token list? Process one? Get one? `tokenListToArray` would be nice."
+    },
+    {
+      "prediction": "I think this should be `this.data('hasLabel')`",
+      "reference": "The message for this check used the existence of a label to determine the output, which doesn't work with the current schema. So I updated it since the data only needed to know a label was present and not what it was."
+    },
+    {
+      "prediction": "I think this should be `app.ProjectId != claims.Role.ProjectId`",
+      "reference": "nit: \"The current project does not have requested application\""
+    },
+    {
+      "prediction": "I think this should be `public const FORBIDDEN_ATTRIBUTES = ['id'];`",
+      "reference": "Extend Unit test for this class"
+    },
+    {
+      "prediction": "I think this should be `DD_APPSEC_CUSTOM_IP_HEADER` instead of `DD_APPSEC_IPHEADER`.",
+      "reference": "This is this a copy / paste error from above."
+    },
+    {
+      "prediction": "I don't think this class is used anywhere.",
+      "reference": "this package has to be renamed to `typeinference`"
+    },
+    {
+      "prediction": "I think we should use `assertIsInstance` and `assertNotNull` instead of `TestCase.assertNotNull` and `TestCase.assertIsNotEmpty`.",
+      "reference": "This cancels the above null-check I guess."
+    },
+    {
+      "prediction": "I think this should be `servicecomb.loadbalance.stats.intervalInMillis`",
+      "reference": "change to timerIntervalInMillis"
+    },
+    {
+      "prediction": "I don't think we need this.",
+      "reference": "`any rules` should be `any roles`"
+    },
+    {
+      "prediction": "`feedItemFilter` is not used in this method.",
+      "reference": "Why does the method need to be renamed? I would just keep the old name and update the other uses (which are only tests). That way, we can reduce code duplication."
+    },
+    {
+      "prediction": "I think this should be `conf.get_bindings_for('normal').get('keystring', 'none')`",
+      "reference": "I'm not sure if this is going to work - I think it's fine to keep them hardcoded here, as the statusbar can still show `:`, `/` and `?` even if the key is rebound."
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "@tmetzke shouldn't we replace this library with the `2.12.1` instead of removing it?"
+    },
+    {
+      "prediction": "I think we can remove the `NumCommittedSectors` field from the `Miner` struct.",
+      "reference": "uint64 seems excessive here. Should we reduce to a uint32 @whyrusleeping ?"
+    },
+    {
+      "prediction": "I think this is the only place where we're using the `app` field, so I think it should be marked as `deprecated` and removed in the next major.",
+      "reference": "Hmm, I thought this was getting used. These are used elsewhere as a cached value (in `svc deploy` it's `o.targetApp`) but I guess since storage doesn't actually need to validate that the app exists, just that there are local services, we never used it."
+    },
+    {
+      "prediction": "I don't think we need to split the host address here. We can just use `r.RemoteAddr` as it is.",
+      "reference": "To keep it simple, how would you feel about just using r.RemoteAddr? Since every request comes through here I want it to be as lean as possible. Frankly I'm OK with the port showing up in the log; maybe it'd even be useful to someone."
+    },
+    {
+      "prediction": "Why do we need to check this?",
+      "reference": "grant bonus depends on both `admin{}` stored in statedb, and `P2Start/End` in local struct, which is kind of weird at Kamchatka height, we add the bonus Start/End epoch into `admin{}`, so it solely depends on `admin{}` stored in statedb"
+    },
+    {
+      "prediction": "This should be `raise ValueError(f'Invalid flipping direction \"{direction}\".')`",
+      "reference": "Use single quote to wrap the str."
+    },
+    {
+      "prediction": "I think this should be renamed to `PERMISSIONING_NOT_ENABLED` or `PERMISSIONING_NOT_SUPPORTED`.",
+      "reference": "Is the plan to rename this later?"
+    },
+    {
+      "prediction": "I think this should be `!==` instead of `!`",
+      "reference": "Is there a reason why we need a special case for Email at this point? Is the idea that other methods can turn into Email even if it's not configured as a top-level option?"
+    },
+    {
+      "prediction": "I think this should be `panic(\"dir unexpectedly empty\")` instead of `len(dir) == 0`",
+      "reference": "This wording is a bit ambiguos, I first thought it meant the directory has no entries in it. maybe \"dir\" -> \"dir string\"?"
+    },
+    {
+      "prediction": "I'm not sure this is necessary. The `m_was_*` variables are only used in the `init_async` function, which is called from the `init` function, which is called from the `main` function, which is called from the `cppunit` function, which is called from the `test` function, which is called from the `cppunit` function, which is called from the `test` function, which is called from the `cppunit` function, which is called from the `test` function, which is called from the `cppunit` function, which is called from the `test` function, which is called from the `cppunit` function, which is called from the `test` function, which is called from the `cppunit` function, which is called from the `test` function, which is called from the `cppunit` function, which is called from the `test` function, which is called from the `cppunit` function, which is called from the `test` function, which is called from the `cppunit` function, which is called from the `test` function, which is called from the `cppunit` function, which is called from the `test` function,",
+      "reference": "Why do we need both `HPX::m_hpx_initialized` and `HPX:: m_was_initialized`?"
+    },
+    {
+      "prediction": "I think this should be `info.Enabled && info.RPCAddress != \"\"`.",
+      "reference": "also check RPCName?"
+    },
+    {
+      "prediction": "I don't think you need to import `org.springframework.security.oauth2.client.web.server.WebSessionOAuth2ServerAuthorizationRequestRepository`.",
+      "reference": "Remove unused import"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could think of.",
+      "reference": "I assume there will never be a scenario where justValidate is expected to be true on a restart"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it's better to do it in the `PMDReportListener` class, since it's the only place where the cache is used.",
+      "reference": "Report listeners are synchronized, even 'though the cache is thread-safe... I've not profiled it, but it may be a cause of contingency. Any better way around this? Also, do listeners **really** need to be synchronized? Can't we just make them thread-safe?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good or a bad thing. I think it's a good thing, but I don't know if it's a good thing to do in general.",
+      "reference": "I don't think this is correct. When `compiler` is not given in `.travis.yml`, the announcement will be `--version`, which results in \"`command not found`\" (though not critical), and the cache slug will lack this information (also not critical)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `callFunction` method, since it's only used there.",
+      "reference": "Why do this rather than `args = []` in the function signature?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I think we probably *should* do validation of whether there is a suitable implementation if local-validation is true. I'm not sure how useful that actually is though"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. You need to do it in `__init__` of the target dataset.",
+      "reference": "@StrikerRUS here may need to concat two data by col."
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think it would be better to have a `NodeWithJavadoc` that has a `Node#getJavadocComment()` method that returns the string representation of the `JavadocComment` object, and then the `NodeWithJavadoc` would have a `Node#getJavadocComment()` method that returns the string representation of the `NodeWithJavadoc` object. This way, the `NodeWithJavadoc` can just call `Node#getJavadocComment()` and the `Node` will call `NodeWithJavadoc#getJavadocComment()`, and the `NodeWithJavadoc` can just call `NodeWithJavadoc#getJavadocComment()` and the `Node` will call `NodeWithJavadoc#getJavadocComment()` and the `NodeWithJavadoc` can just call `NodeWithJavadoc#getJavadocComment()` and the `Node` will call `Node#getJavadocComment()` and the `NodeWithJavadoc` will call `Node#getJavadocComment()` and the `NodeWithJavadoc` will call `NodeWithJavadoc#getJavadocComment()` and the `NodeWithJavadoc` will call `NodeWithJavadoc#getJavadocComment()` and the `",
+      "reference": "Why the space?"
+    },
+    {
+      "prediction": "I don't think we need to add a new section for this. We already have a `zalcanoTitle` property in the `ZalcanoConfig` class, which is used to set the title of the Zalcano UI. This property should be added to the `ZalcanoConfig` class, and then used to set the title of the Zalcano UI. The `zalcanoTitle` property should be added to the `ZalcanoConfig` class, and then used to set the title of the Zalcano UI. The `zalcanoTitle` property should be added to the `ZalcanoConfig` class, and then used to set the title of the Zalcano UI. The `zalcanoTitle` property should be added to the `ZalcanoConfig` class, and then used to set the title of the Zalcano UI. The `zalcanoTitle` property should be added to the `ZalcanoConfig` class, and then used to set the title of the Zalcano UI. The `zalcanoTitle` property should be added to the `ZalcanoConfig` class, and then used to set the title of the Zalcano UI. The `z",
+      "reference": "please resolve the extra indentation"
+    },
+    {
+      "prediction": "I think this should be `!ReferenceEquals(document.Settings.PeptideSettings.Libraries, previous.Settings.PeptideSettings.Libraries) && !ReferenceEquals(document.Settings.MeasuredResults, previous.Settings.MeasuredResults) && !ReferenceEquals(document.Id, previous.Id)`",
+      "reference": "More proof that this check is needed always."
+    },
+    {
+      "prediction": "I think this should be `std::unordered_map<std::string, std::unique_ptr<LBANN::DataCommunicator> >`",
+      "reference": "`m_al_comms` now contains smart pointers."
+    },
+    {
+      "prediction": "I don't think this is the right place to add this check. I think it should be in the `EnumType` constructor.",
+      "reference": "Did you do a test run with the compare script? I'm nervous about what happens on all the generators..."
+    },
+    {
+      "prediction": "I don't think we need this. We already have `nc.RemoteAddr()` in `client.go`",
+      "reference": "For a client, we store host, as string c.host. That is what we use for monitoring and statsz. Not sure if its useful here or not. Looks like probably not."
+    },
+    {
+      "prediction": "I think this should be 3, since there are 3 partitions in the original table.",
+      "reference": "This change is suspicious. Why did the number of partitions increase?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `ProposalPolicy`.",
+      "reference": "I'm still new to this area, so please forgive what may be a stupid question: When would this be false? And do we have a test for that situation?"
+    },
+    {
+      "prediction": "I think this should be `Settings.Default.TransitionListInsertPeptides = isPeptide && gridViewTransitionList.ColumnCount == 5;` or `Settings.Default.TransitionListInsertPeptides = !(!isPeptide && gridViewTransitionList.ColumnCount == 6);`",
+      "reference": "Should there be any changes to this file at all?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. It seems like it should be in `caddytls`, but I don't know if that's the right place to put it.",
+      "reference": "The only field being used is the KeyType; Instead, we can probably just make a `keyType` variable here."
+    },
+    {
+      "prediction": "I think we can remove the call to `SwaggerPetstore.baseUri = 'http://localhost:12510/api';``` here.",
+      "reference": "This should be done by code-gen change happening in the hydra repo. I am tweaking it just to get CI into a 'passing' state"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add the feature flags.",
+      "reference": "Cleaned up when I originally added support for passing `screenContext` here (and later `viewContext`) but that was removed."
+    },
+    {
+      "prediction": "I think this should be `#if defined(__MACH__) || defined(__APPLE__) || defined(__WIN32__)` instead of `#if defined(__MACH__) || defined(__APPLE__) || defined(__WIN32__) || defined(_WIN32__)`",
+      "reference": "This looks like a bunch of whitespace fixing that's unrelated. Can you revert?"
+    },
+    {
+      "prediction": "Why do you remove this?",
+      "reference": "should be removed instead"
+    },
+    {
+      "prediction": "I think this should be `caseSensitive = true;`",
+      "reference": "should use randomization ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `volume.go` file.",
+      "reference": "returning nil because we want to mayactl to exit with 0 status code."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add this change. I think it should be done in the `h2o_add_header` function, as it is the place where the headers are added to the request.",
+      "reference": "Is this change relevant to the PR? (and I believe we should use `text/plain` considering the fact that it can be displayed using web browsers...)"
+    },
+    {
+      "prediction": "This is a breaking change, right? I don't think we should be changing the schema in a breaking change like this.",
+      "reference": "`[Flags]` is now auto-generated by an argument to the `EnumHint`. (The attributes happen to come out in this order. I don't think it's worth controlling the order.) #Resolved"
+    },
+    {
+      "prediction": "I think this should be `configParams.SetUseResourceUpdates(syncerToValidator.UseResourceUpdates())` instead of `configParams.SetUseResourceUpdates(true)`",
+      "reference": "I think same here - in general we don't need to use Setters / Getters since configParams isn't a public API."
+    },
+    {
+      "prediction": "I think this should be `fmt.Sprintf(\"%s\", pdata)`.",
+      "reference": "Why using `fmt.Sprintf` where `fmt.Sprint` would do the work (notice no `f` in function name)?"
+    },
+    {
+      "prediction": "I don't think this is necessary. The `IAccessible2::attribute_current` attribute is only used by the `IAccessible` class, which doesn't have a `current` property.",
+      "reference": "Extraneous blank line."
+    },
+    {
+      "prediction": "I'm not sure about this error message. I think it would be better to say something like: \"given multisig preimage hashes to neither Sender nor AuthAddr\"",
+      "reference": "nit: syntax : neither -> nor"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `ActiveScope` or `ActiveSpanContext` classes.",
+      "reference": "Is it safe to assume that these two keys are always present? `this[string]` will throw a `KeyNotFoundException` if they are not."
+    },
+    {
+      "prediction": "`exflow.getFlowDefinitionId() > 0 ? exflow.getFlowDefinitionId() : flowId` can be simplified to `exflow.getFlowDefinitionId() > 0 ? exflow.getFlowDefinitionId() : flowId`",
+      "reference": "Is the `flowDefinitionId` sufficient to uniquely identify the flow or does it need to be the tuple `<flowId,flowDefinitionId>` ?"
+    },
+    {
+      "prediction": "I think this should be `DateTimeInterface|bool`",
+      "reference": "Shouldn't it be in the reverse order? `DateTime::createFromInterface()` returns `self` and accepts `DateTimeInterface`."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "this check is now redundant. The rest of the function downwards basically does the same thing but with safer checks."
+    },
+    {
+      "prediction": "This should be `array('id', 'id_reverse', 'alpha', 'alpha-reverse', 'alpha_reverse', 'manual', 'manual_reverse')` and not `array('id', 'id_reverse', 'alpha', 'alpha-reverse', 'alpha_reverse')`",
+      "reference": "Please could you remove the useless space."
+    },
+    {
+      "prediction": "I don't think we need to check for negative values here. If the target is out of range, it means the target is greater than the max value or it is less than 0. In either case, we should return an error.",
+      "reference": "What's the point of this block (and isn't it effectively dead code)?"
+    },
+    {
+      "prediction": "I think this should be `DiscourseSignOn::DEFAULT_SSO_SECRET`",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I'm not sure this is a good example of how to do this. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not sure it's a good example of how to do this at all. I'm not",
+      "reference": "Newline at end of file plz"
+    },
+    {
+      "prediction": "nit: I think we can remove the `goimports` import here.",
+      "reference": "Can this be refactored to avoid including the aws sdk in the types that we expose? I know it's not being exposed directly through the types we expose, but I do worry that it might make it easier to accidentally do that in the future and not realize it as easily."
+    },
+    {
+      "prediction": "This should be `time()` not `time()()`.",
+      "reference": "Hi thomas, i would propose to indicate in the name, that only the indextime is updated, otherwise somebody might think the whole items is getting updated. I would propose something like \"updateIndexTimeByItem(Item $item)\""
+    },
+    {
+      "prediction": "I don't think this is the best way to do this. I think it would be better to use `isinstance(element, dict) and 'id' in element and 'id' not in dict.get('id')`",
+      "reference": "This is not the right exception class. There is an InvalidSelectorException class that covers bad locators."
+    },
+    {
+      "prediction": "These should be `#include <Kokkos_Core_1.8.0.hpp>` and `#include <Kokkos_Core_2.0.0.hpp>`.",
+      "reference": "I assume this came from format?"
+    },
+    {
+      "prediction": "I think this should be `@var boolean default(false);` instead of `protected $clearSearchIndex = false;`",
+      "reference": "I'd suggest a default value of `false` just to make sure existing tasks are ok when they get deserialized after an update to a version containing this code."
+    },
+    {
+      "prediction": "I think this should be `dumpConfig.CDebugf(ctx, \"Trying to decrypt with keys = %s, index=%d\", dumpConfig.Keys, index)`",
+      "reference": "Do we really need this? Seems like it would pollute the logs pretty badly."
+    },
+    {
+      "prediction": "I think this should be `private static readonly int _vectorSpan = Vector<byte>.Count;`",
+      "reference": "`Vector<byte>.Count` should Jit to const when intrinsic; not sure when `Vector.IsHardwareAccelerated == false` ; however we know `readonly static int` does Jit to const."
+    },
+    {
+      "prediction": "I think this should be `if not hasattr(sqlalchemy, 'is_available'): return`",
+      "reference": "Can we use a skipIf decorator instead?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Missing version tags"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I think this should also filter out `validate_resolve_reference` hooks."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a great solution either.",
+      "reference": "unrelated, but required fix as metrics branch got broken with the latest main->metrics merge and conflict resolution."
+    },
+    {
+      "prediction": "I don't think this is the right place for this file.",
+      "reference": "Feel free to remove this line"
+    },
+    {
+      "prediction": "I think this should be `I18n.t(\"shared.subscriptions.single_user.call_to_action\")`",
+      "reference": "~~Should this match the key updated above?~~ :+1:"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "`gofmt` allows a more compact form `defer func() { _ = f.Close() }()` or just add `// nolint:errcheck` before `defer`."
+    },
+    {
+      "prediction": "I think this should be `big.Int` instead of `types.AttoFIL`",
+      "reference": "collateral is in FIL"
+    },
+    {
+      "prediction": "Why do we need to store the channel here?",
+      "reference": "We should not create channel of node which are not connected."
+    },
+    {
+      "prediction": "I think this should be `options.source` instead of `options.db`",
+      "reference": "I think the docstring should say something like \"Alias for the `source` option\" or something similar."
+    },
+    {
+      "prediction": "I think it would be better to add a `@JsonInclude(JsonInclude.Include.NON_NULL, JsonInclude.Include.ALWAYS)` annotation on these properties.",
+      "reference": "Please use properties without suffix `_fr` so they can work across different languages as @teolemon mentioned"
+    },
+    {
+      "prediction": "I think this function should be called `destroy_request(struct st_h2o_http3client_req_t *req)`.",
+      "reference": "Declaration here is `static` but the definition at the bottom is non-static?"
+    },
+    {
+      "prediction": "I'm not sure this is a good name for the option. I would have called it `md5` or `md5sum` instead of `ContentMD5` since it's not just the content md5, but also the md5sum of the content.",
+      "reference": "Hi @myml, thanks for the contribution! `blob` and `blob/driver` are both in the same module, so you shouldn't need to split this change up into multiple Pull Requests. Also, I'd like to see it working, including the implementation for `s3blob` and `gcsblob` (these should be easy, just pass-through to the provider) and `fileblob` (might be trickier...), and a test in `drivertest/drivertest.go`."
+    },
+    {
+      "prediction": "I think we can remove the `if i.nodeConfig.NodeTransportIPv4Addr != nil` check here.",
+      "reference": "Was it a bug?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "# `attr_value['string'] is not None` probably not required as `attr_value['string'] == value` check is already being done"
+    },
+    {
+      "prediction": "I think this should be `lambda: tabbed_browser._now_focused._hovered_link` instead of `tabbed_browser._now_focused._last_hovered_link`",
+      "reference": "You're accessing private variables here - `last_hovered_link` should be public in `TabData`. As for `tabbed_browser._now_focused`, I think you can use `tabbed_browser.widget.currentWidget()` instead."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this method. I think it would make more sense to move it to `BaseMailer` since it's a common method used by all the other mailers.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I don't think we need this partial class. We can just use `Microsoft.Rest.HttpClient` and `Microsoft.Extensions.Primitives` directly.",
+      "reference": "We should file a bug for this - we don't need the extensions class if there are no operations on the client"
+    },
+    {
+      "prediction": "I think we should move this to `NewMounter` instead of `NewCsiMounter` since it's not specific to csi.",
+      "reference": "Is it possible to import from k8s instead of copy in? :)"
+    },
+    {
+      "prediction": "Why did you change the type of `$start_time`?",
+      "reference": "`@param float` can be dropped here."
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "maybe we can leave this change un-reverted."
+    },
+    {
+      "prediction": "I think this class should be renamed to `ASTMethodDeclaration` instead of `ASTMethodDeclarator`.",
+      "reference": "I wouldn't make MethodDeclarator an AccessNode, nor an Annotatable. It's the MethodDeclaration that can be annotated, and has info about the modifiers, and is already an AccessNode"
+    },
+    {
+      "prediction": "I don't think this is needed. The `Player` field is already in the `PlayerHeadIconChanged` constructor.",
+      "reference": "delete this class"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix, but I can't think of a better one.",
+      "reference": "Replacing `+` (1 or more) with `*` (0 or more) would've also done the trick."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that the writer is still open when the destructor is called, so we need to close it before calling close() on dp_ostream.",
+      "reference": "I've been meaning to fix this for a while. Thanks."
+    },
+    {
+      "prediction": "I think this should be A, B, C, A, B, A, B, A, B, A, B, A, B, A, B, A, B, A, B, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A, B, A, B, A, C, A, B, A, B, A",
+      "reference": "This is adding multiple processor pipelines. I guess you wanted to add multiple processors to the same, single pipeline?"
+    },
+    {
+      "prediction": "I don't think this is a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not a good way to do it. It's not",
+      "reference": "If you need str(x[0]) here, you'll probably need str(x) the line after that I suppose. Also, shouldn't we use `raw()` here instead of `str()`?"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for this value instead of hardcoding it.",
+      "reference": "3.6 seconds? what does it stand for?"
+    },
+    {
+      "prediction": "Style/MutableConstant: Freeze mutable objects assigned to constants.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "If the command fails, is stdout nil or \"\"?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the old URL and just change the `apiVersion` parameter to `apiVersion`",
+      "reference": "Why is api-version now removed in the generated code?"
+    },
+    {
+      "prediction": "Why do we need to import `Registry` from `mmdet.utils`?",
+      "reference": "We may register all built-in optimizers of PyTorch here to simplify the builder."
+    },
+    {
+      "prediction": "I think this should be `for (auto component = ns->components.begin(); component != ns->components.end(); ++component)`",
+      "reference": "No need for this new variable."
+    },
+    {
+      "prediction": "I think this should be `if (newType.getDerivedStateFromProps != null) {`",
+      "reference": "doesn't this enqueue a double render or is that safeguarded somehow"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with.",
+      "reference": "But now you may get DATA RACE reports because c.acc is sometimes changed. I wonder if we should not rework that whole sending subs through route."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this check. I think it should be in the `else` part of the `if (type == \"HDF5Reader\")` block.",
+      "reference": "Use `ADIADIOS2_HAVE_HDF5`, not `ADIOS_HAVE_PHDF5`"
+    },
+    {
+      "prediction": "I think this should be `expect(group_value_for 'group', Object).to eq('group Object')`",
+      "reference": "Would be nice to put this in a separate context named `\"with a string and a non-string\"` (since that's what it is -- it's definitely not a non-string and a string!)."
+    },
+    {
+      "prediction": "I'm not sure about this change. It seems to me that the number of entries in the master and pertable stats tables is the same, so there's no need to increase the number of entries in the pertable stats table.",
+      "reference": "Should MAX_PERTABLE_ENTRY here be 10 ?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Should move these out of frame into a static class? Similar to reason phrases?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `if (steps[last_valid_instruction].mode == step.mode)` check above.",
+      "reference": "That means we still create `UseLane` instructions but threat them like `Suppressed`? Can we maybe change this in the pre-processing so that we will just emit `Suppress` instructions in these cases?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "no need to move here, iotex-proto is outside of iotex-core and considered same as iotex-address, go-pkgs"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `app_test.c` under the `malloc_test` function.",
+      "reference": "Looks like a clang-format style violation: should turn Travis red."
+    },
+    {
+      "prediction": "I'm not sure about this one. I think it would be better to have a setting for `:colors` instead of `:success_color`, `:pending_color` and `:failure_color`.",
+      "reference": "I think we need some YARD docs here, particularly to list all the color symbols that are valid. Otherwise users will have to look at the source to discover that."
+    },
+    {
+      "prediction": "I think we can remove this function now that we have `getSiteInfo()`.",
+      "reference": "Not sure if this needs a doc block or not since it's completely internal, but   for cleaning this up. (@felixarntz docs?)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should go in `libgtest.benchmarks.buildoptions`.",
+      "reference": "The change here is to address what exactly?"
+    },
+    {
+      "prediction": "nit: I think it would be better to use `bool` for `IsMSIEnvironment` instead of `bool IsMSIEnvironment`.",
+      "reference": "Is this really an Option? IIUC, you compute it automatically for the URLOpener case, why would we require a user to fill it in when using the constructor? I.e., can't we drop this and use `adal.MSIAvailable` instead? If that call is expensive, cache it on `bucket`, not `Options`."
+    },
+    {
+      "prediction": "Why is this needed? It's not used anywhere else in the codebase.",
+      "reference": "Was this done automatically or by hand?"
+    },
+    {
+      "prediction": "`current_group.name == desired_group` can be simplified to `current_group.name == desired_group || current_group == desired_group`",
+      "reference": "This looks like it will only work for the first child, since after that we've overridden `desired_group`. Should we have a separate variable for this?"
+    },
+    {
+      "prediction": "This should be `std::ios_base::binary` or `std::ios_base::in | std::ios_base::out` instead of `std::ios_base::binary`.",
+      "reference": "Should this be a constant somewhere?"
+    },
+    {
+      "prediction": "I think it would be better to use `OpenTelemetry.Instrumentation.Http` and `OpenTelemetry.Instrumentation.GrpcNetClient` instead of `OpenTelemetry.Instrumentation.AspNet` and `OpenTelemetry.Instrumentation.GrpcNetClient`",
+      "reference": "nit: sort alphabetically."
+    },
+    {
+      "prediction": "I think this should be `contextKey` instead of `paramsPrefix`.",
+      "reference": "Now, It's redundant, I suppose."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `go.uber.org/yarpc/peer/peerlist/v2`.",
+      "reference": "nit: The format recognized by tooling is `// Deprecated: [..]` so you likely want this to be, // Deprecated: Use \"go.uber.org/yarpc/peer/peerlist/v2\".Implementation instead."
+    },
+    {
+      "prediction": "I think this should be `log.Warnf(\"No telemetry entry found for: %s\", bootstrapID)`",
+      "reference": "nit: 1. when formatting input strings, make sure to place them in quotes so we could identify white space issues. i.e. '%s' 2. If there is untyped, non-nil error, you want to log the error string as well."
+    },
+    {
+      "prediction": "I think we can remove the `so` prefix from the `Windows` and `Linux` aliases.",
+      "reference": "I believe this can be fixed with the original code if you just omit the \".dll\" file extension so it reads `[DllImport(\"Datadog.Trace.ClrProfiler.Native\")]`. On Windows it would look for `Datadog.Trace.ClrProfiler.Native.dll` and Linux/Mac it would look for `Datadog.Trace.ClrProfiler.Native.so`."
+    },
+    {
+      "prediction": "I think this should be `if (oldVersionCode < 1070197 && prefs.getBoolean(UserPreferences.PREF_MOBILE_UPDATE_OLD, false)) {`",
+      "reference": "couldn't we just read the boolean as a string (I would assume that this return \"true\" or \"false\"), migrate if to its new value and keep using the same pref key?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `input.c` under the `json` and `jx` modules.",
+      "reference": "If these variables are only used in main function, move these to the beginning of that function. If there is a foreseeable reason to have them as global statics just let me know."
+    },
+    {
+      "prediction": "I don't think this is the right place for this change. This is an API change and should be done in a separate PR.",
+      "reference": "This one was wrong."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `BatchChanges`. The reason is that `BatchChanges` is designed to be a single place for all of the changes that need to be batched together, so it makes sense to put this logic in there.",
+      "reference": "What happens if nodes throughout a hierarchy are modified (as they would be)? It looks like we're going to `Reset` once for each `NodeID` affected."
+    },
+    {
+      "prediction": "I'm not sure this change is correct. It looks like it's just a copy-paste of the existing code.",
+      "reference": "> Note that allocation of vmheap fails upon initializing dynamoRIO 64 on WoW64 processes. Thus, we need to pass -reachable_heap to avoid having to make this allocation. This should be solved by changing the default `vmheap_size` to be much smaller than 8GB for x64 DR inside WOW64."
+    },
+    {
+      "prediction": "`@exception_presenter.fully_formatted_lines(failure_number, colorizer)` should be `@exception_presenter.fully_formatted_lines(failure_number, colorizer).join(\"\\n\")`",
+      "reference": "Is it really worth expanding our public API for this? After all, isn't calling this the same as calling `notification.fully_formatted(...).lines`? If so, I'd rather not widen our API (and thus, increase our maintenance burden) when it's so simple to get all the lines already."
+    },
+    {
+      "prediction": "I think we can remove the `public` modifier from this interface.",
+      "reference": "Is it necessary to keep this class around just to define this constant? Or can we place it somewhere else?"
+    },
+    {
+      "prediction": "I think we can remove these two fields from the rule struct, since they are no longer used by the CLI or the API.",
+      "reference": "Question - do we cache NetworkPolicy itself? If so, here we can point to NetworkPolicy?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "For my view, `Register` means the process of **insert node resource to etcd through api-server**, which is called by upstream rather than here, how about changing the func name to `OnConnected`?"
+    },
+    {
+      "prediction": "I think this should be `protected boolean telemetryInitialized = false;` instead of `protected boolean telemetryInitialized = null;`",
+      "reference": "I would prefer an active verb for this property, e.g. `initializeTelemetry`. The reason is that this property refers to something the engine does once on startup. Other properties that use passive voice (e.g. `authorizationEnabled`) refer to a state of the engine during its lifetime."
+    },
+    {
+      "prediction": "I don't think we need this property. We can just use the `Contains` method.",
+      "reference": "please use new line to maintain reasonable line width"
+    },
+    {
+      "prediction": "I'm not sure if this is the right approach. I think we should check if the tag exists before adding it. If it does not exist, we should add it. Otherwise, we should return an error.",
+      "reference": "I would skip this check or add an internal server error response."
+    },
+    {
+      "prediction": "I think this should be `config.notarizeUser` and `config.notarizePassword` instead of `config.notary_user` and `config.notary_password`",
+      "reference": "missing `{` here (and then `}` after `notary_password = config.notary_password`); it's only going to do the first one"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it should be moved to `test/testutil.go` since it's not related to the `ddev import-db` command.",
+      "reference": "This is odd because it's actually in our generated SiteSettingsPath, not in the SiteLocalSettingsPath (wp-config-ddev.php). Are these two files swapped somehow? I'd expect SiteSettingsPath to be wp-config.php and SiteLocalSettingsPath to be wp-config-ddev.php. BTW, I'm *way* ok with renaming that to SiteDdevSettingsPath."
+    },
+    {
+      "prediction": "I think this should be `kythe/kythe-extraction-config` instead of `kythe/go/extractors/config/default`.",
+      "reference": "It seems like this must necessarily be a stopgap. Can you please add a TODO(#xyz) to point to the appropriate issue?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think we should move this to `agent.go` and then call `startKubelet` and `startKubeProxy` from there.",
+      "reference": "Is the return signature necessary since we never actually return if there is a problem (I'm referring to the call to `logrus.Fatalf`). Let's pick a pattern and stick with it."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think it's better to use `this.get('model.name') + ' - ' + this.get('model.id')`",
+      "reference": "@btecu why is this override needed here? I'm not seeing a `new` being used here."
+    },
+    {
+      "prediction": "I think it would be better to move this into `initLeafNodeSmapAndSendSubs` so that we don't have to worry about closing the connection in `initLeafNodeSmapAndSendSubs`.",
+      "reference": "Should it not be more something like: `MaxAccountConnectionsExceeded` here?"
+    },
+    {
+      "prediction": "I think this should be `assert.Equal(t, \"default\", parseAndValidate(t, \"-compute_service_account\", \" default\t\").ComputeServiceAccount)`, and then the next line should be `assert.Equal(t, \"\", parseAndValidate(t, \"-compute_service_account\", \" \t\").ComputeServiceAccount)`.",
+      "reference": "minor: use a non-default CE service account for testing (since default is handled in a specific way in the code)"
+    },
+    {
+      "prediction": "Why do we need to change this?",
+      "reference": "This is how I ran the 512MiB test. It's probably too much data for a fixture, and makes for longer functional tests (although still less than a minute on my laptop)."
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "reference": "So we weren't able to hide the fact that the file wasn't pre-generated and had to introduce this \"option\" to detect that?"
+    },
+    {
+      "prediction": "I think it's better to make this a `map[string]int` so that it can be used in other places as well.",
+      "reference": "Agree that exporting these maps is the best tradeoff, just registering my ongoing discontent with the lack of `const` collections."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `WebKitCaret`'s `onSelectionChanged` method.",
+      "reference": "This could probably use `const` as well?"
+    },
+    {
+      "prediction": "I think this should be `tmp`, not `orig`.",
+      "reference": "it doesn't look like these need to be exported?"
+    },
+    {
+      "prediction": "I think this should be `fmt.Errorf`",
+      "reference": "I think you should remove \"fmt\" here. And change `fmt.Errorf` to `errors.Errorf`."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Here's the `resources` attribute @jferris. I'm not against renaming `Product` to `Resource`, and this one... something else."
+    },
+    {
+      "prediction": "I think this should be `defer wg.Drain()`",
+      "reference": "If the context is unused, this is equivalent to `var wg errgroup.Group`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Highlighter` class, not the `HighlightParams` class.",
+      "reference": "very minor: I'd prefer these two added rows are switched so that fragsizeIsMinimum directly follows fragsize"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for a loopback address. I think we should be looking for an IPv4 address that doesn't start with 127.0.0.1 or 255.255.255.255. We could also look for an IPv6 address that doesn't start with 00:00:00:00:00:00 or 2001:0db8:85a3:0001:0000:0000:0000:0000. We could also look for an IPv6 address that doesn't start with 00:00:00:00:00:00 or 2001:0db8:85a3:0001:0000:0000:0000:0000. We could also look for an IPv6 address that doesn't start with 00:00:00:00:00:00 or 2001:0db8:85a3:0",
+      "reference": "Given that 115 returns `True`, why would this be `False` here?"
+    },
+    {
+      "prediction": "I think this test should be moved to `org.mule.module.oauth2.server.oauth2.server.OAuth2LoginConfigurerTests#oauth2LoginWithOneClientConfiguredAndRequestFaviconNotAuthenticatedThenRedirectDefaultLoginPage`",
+      "reference": "Please move this test method just below `oauth2LoginWithOneClientConfiguredThenRedirectForAuthorization()`"
+    },
+    {
+      "prediction": "I think this should be set to 1000 seconds.",
+      "reference": "Why this new configuration?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the partition value is not a byte array, then it's not a valid binary type, and we should throw an exception.",
+      "reference": "I don't think that we need to check `hasArray` here. I think the reason why this didn't previously check `hasArray` is that the array passed to `DataByteArray` must start at offset 0 and be valid through the array length, so a copy was needed in almost every case. It may be simpler to change this to use `ByteBuffers.toByteArray` and pass the result to create `DataByteArray`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this. I think `SpanContext` and `SpanReference` could be more specific, e.g. `BaggageContext` and `BaggageReference`?",
+      "reference": "`spanReference` -> `baggage`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put it. I think it should be in `web` folder, but I'm not sure.",
+      "reference": "Why the `/-/verdaccio/` ?"
+    },
+    {
+      "prediction": "I think this is the only place where we need to change the `AppName` and `GlobalOpts` here.",
+      "reference": "Has been waiting for a million years"
+    },
+    {
+      "prediction": "Shouldn't these be `min` and `max`?",
+      "reference": "Minor grammar detail: the 'than' is unnecessary"
+    },
+    {
+      "prediction": "I think this should be `self._internal._sdf.select(scol).apply(func, axis=0, skipna=skipna)` instead of `self._internal._sdf.select(scol).head(2).apply(func, axis=0, skipna=skipna)`",
+      "reference": "I think you can just simply fix this line to `self[:2].to_pandas()`"
+    },
+    {
+      "prediction": "I think this should be `if (debuggerAddress.match(/\\/se\\/cdp\\/ws\\//))` instead of `if (debuggerAddress.match(/\\/se\\/cdp\\/ws\\//))`",
+      "reference": "if we are returning the `ws` here when passing in `se:cdp` we can just return it straight or do we have to make a request to get the `ws` address?"
+    },
+    {
+      "prediction": "I think we should move this file to `src/store.js` and rename it to `registerSiteData`",
+      "reference": "I could add a resolver for this (like above) but it seems more sensible to just change the one reference!"
+    },
+    {
+      "prediction": "I think this should be `minimumSecondsSinceParent + actualTimeSinceParent` instead of `actualTimeSinceParent`.",
+      "reference": "not sure \"actual\" adds value - secondsSinceParent is probably closer."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Line is too long. [94/80]"
+    },
+    {
+      "prediction": "I think it would be better to add this to the `repeater` section.",
+      "reference": "The error should be `At least :number items are required`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `HandleBlockPropose` function.",
+      "reference": "singleCaseSwitch: should rewrite switch statement to if statement (from `gocritic`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `KeyManager` interface.",
+      "reference": "Could combine this with `encryptionKeyGetter` to have a single `keyGetter` interface. I'm not sure which way is better."
+    },
+    {
+      "prediction": "I think we can remove the deleteInstance() call here.",
+      "reference": "Is there any security issue here? If we integrate with Hive security (and I don't know if we have or not) is there some notion of re-authentication at connection time?"
+    },
+    {
+      "prediction": "I think we should keep `writeConcern` as it was before.",
+      "reference": "Does this mean we do not support writeConcern on `createCollection`?"
+    },
+    {
+      "prediction": "I think this should be `id` instead of `sid.ID`",
+      "reference": "I guess `sid` was for `sessionID`, but now it looks strange: `sid.ID`."
+    },
+    {
+      "prediction": "I think we need to add the source security group ID to the `SourceSecurityGroupIDs` slice above.",
+      "reference": "We'll need `infrav1.SecurityGroupBastion` as well"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/fuse/fuse.go`",
+      "reference": "This `fileName` should probably be `remote` and be relative to the Fs root as per normal rclone usage."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move it to `org.mule.runtime.module.extension.runtime.test`",
+      "reference": "That way we are back to having just one index on k_0 So we can really compare the execution times with and without re-indexing"
+    },
+    {
+      "prediction": "I think we should use `StringEscapeUtils` instead of `String.format` to escape the characters.",
+      "reference": "Expression that uses a negation pattern to exclude all characters that aren't in the expression."
+    },
+    {
+      "prediction": "I think it would be better to call it `EchoServiceHeader` to be consistent with the name of the header.",
+      "reference": "Similar to my suggestion for HTTP, let's remove this and use `ServiceHeader` instead."
+    },
+    {
+      "prediction": "I think this should be `if commandSet == targetGlobalCommandPath {` instead of `if serviceDirOnHost[0:1] == \".\" {`",
+      "reference": "The serviceDirOnHost[0:1] was completely wrong. It wasn't just Windows, glad you got this fixed!"
+    },
+    {
+      "prediction": "I'm not sure why we need to change the timeout here?",
+      "reference": "Would membership take 10+ seconds to detect the failed node?"
+    },
+    {
+      "prediction": "I think this should be a `RuntimeError` instead of a `ValueError`",
+      "reference": "hmm, in this case perhaps the whole try-except could be removed altogether? because if the file does not exist, it's already returning `None`. In any other case it should actually probably fail with the exception, no?"
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `filepath.Join(dir, \"..\", filepath.Join(paths...))`",
+      "reference": "This function does not need to be public"
+    },
+    {
+      "prediction": "I'm not sure if we want to add a margin-bottom to this, or not. I think it's fine to not add a margin-bottom if we don't want to add it.",
+      "reference": "Any chance we can get a CSS class for this? I know it's just in one place but feels like a dangerous precedent to set."
+    },
+    {
+      "prediction": "I don't think we need to check for the next link method here, since we are already checking for the next link method in line 115.",
+      "reference": "> && this.Extensions.ContainsKey(\"nextMethodName\") [](start = 96, length = 48) From line 124, looks like we don't need `&& this.Extensions.ContainsKey(\"nextMethodName\")` condition or we don't need line 124 #Closed"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `componentWillMount` hook, since that's where the ref is set.",
+      "reference": "Moving this code from compat to core shaves 47 bytes out of compat and only adds 6 bytes to core so I thought it was worth it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the base dir of the current process.",
+      "reference": "Please use \".NET Core 3\" - i'm trying hard to use one form, and this is the one MS advices"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `Telemetry` package.",
+      "reference": "Unrelated to this change, but this is a fix for when TestDoStartCgroupInitHappyPath has a failure after the test goroutine has already exited."
+    },
+    {
+      "prediction": "I think this should be `check_not_scrolled` instead of `check_scrolled`",
+      "reference": "Hmm, I'd really expect this to work, and yet it doesn't. I'll investigate later, though it might get Monday until I get the time."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think you should be able to use `std::make_int64_t` instead of `std::int64_t` here.",
+      "reference": "Does this changes affect process building dynamic libraries anyhow?"
+    },
+    {
+      "prediction": "I don't think we need to import `addPPIType` from `scapy.contrib.ppi`.",
+      "reference": "Could you limit what is imported here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `lambda_handler.py` file, since that's where the lambda handler is defined.",
+      "reference": "just to clarify - on line 1074 we update `zip_file_content` for non-local lambdas, but never store it, which means lambda never picks it up"
+    },
+    {
+      "prediction": "I think this should be `packetWithPorts(1, src: \":0\", dst: \":0\")` instead of `packet{protocol: 1, srcAddr: src, srcPort: 0, dstAddr: dst, dstPort: (icmpType << 8) | (icmpCode)}`",
+      "reference": "Golang naming convention is to use camel case `icmpPktWithTypeCode` Often the linter will complain"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `catch` block below, but I'm not sure if that's the right place to do it.",
+      "reference": "We can do away with assigning the response here too."
+    },
+    {
+      "prediction": "I think this should be `newViewportHeight = naturalHeight + 4;`",
+      "reference": "Bumping the newViewportHeight"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `js-hll` package.",
+      "reference": "Is this change correct? Looks like a typo and not sure this should be changed?"
+    },
+    {
+      "prediction": "I think this should be `qname.split('.')[-1]` instead of `astroid.modutils.get_module_part(qname)`.",
+      "reference": "I checked the tests coverage and strangely it look like those three lines are not covered (?!). Do you have the same result on your side ?"
+    },
+    {
+      "prediction": "I think it would be better to use `#pragma warning disable CS0618` instead of `#pragma warning restore CS0618` here.",
+      "reference": "nit: please try to avoid changing end of the file (it adds a noise to the code review, but does not provide too much value)"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could come up with that worked.",
+      "reference": "I don't see the \"only\" part reflected in the code but maybe I'm missing it. Also, the indentation is misleading: the `.SelectMany` calls are perfectly aligned but are *not* operating on the same \"level\". I'd expect the second `SelectMany` to be on the same height as the inner `Where`, just break `pathObj => pathObj<HERE>.Where` and lines won't be that long."
+    },
+    {
+      "prediction": "I think this should be `resources => { type => Optional[Hash[String[1], ResourceInstance]], kind => given_or_derived },` and not `plugin_hooks => { type => Optional[Hash[String[1], Data]], kind => given_or_derived },`",
+      "reference": "It seems like most of these will never be nil, are they optional just in case?"
+    },
+    {
+      "prediction": "I don't think we need to pass -1 here.",
+      "reference": "`.addColumn(\"c1\", IntegerType.INT, true)` means `PkHandle=true`, maybe should add `.setPkHandle(true)` in line 31"
+    },
+    {
+      "prediction": "I think this should be `throw new TiExpressionException(\"Cannot encode row key with non-long or non-integer type\");`",
+      "reference": "how about `Cannot encode row key with non-integer type` directly?"
+    },
+    {
+      "prediction": "I think this method should be renamed to `fetchUnfinishedExecutionsByExecId`",
+      "reference": "The method names `fetchUnfinishedExecutions` and `fetchUnfinishedFlows` are too similar to each other. Would it be better to use the name `fetchUnfinishedFlowsMetadata` since you are only fetching metadata info about the flow?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it's the only way I could get Travis to pass.",
+      "reference": "Note that `sudo` is not available by default on BSDs; there are a few places in the codebase here where that's explicitly worked around by using `su`."
+    },
+    {
+      "prediction": "This should be `package importpath` instead of `package importpath is used to implement a test on Go import paths.`",
+      "reference": "ultra nit: missing Oxford comma :P"
+    },
+    {
+      "prediction": "I'm not sure if we want to do this in this PR, but I think it would be good to do it in a follow-up PR.",
+      "reference": "This is not supported in Spark 3.0.0-rc2 yet. I'd skip this for now."
+    },
+    {
+      "prediction": "I think this should be `Query(ctx context.Context, from, to types.Address, method string, params ...interface{}) ([]byte, *exec.FunctionSignature, error)`",
+      "reference": "BLOCKING: Why does `Query` return an `*exec.FunctionSignature`?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to do this with a single call to `resetCachesWithShutdown`",
+      "reference": "please move this down to immediately above the assignment to `c.dirtyBcache`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `examples/variable`.",
+      "reference": "Don't need the `, err` part since you've already handled it."
+    },
+    {
+      "prediction": "nit: you can use `std::string` instead of `std::wstring` here.",
+      "reference": "Please add two additional SECTIONs that show that this also works if the dative bond is in the reactant (reaction SMARTS `[O:1]->[H+]>>[O:1].[H+]`) or in the agents (reaction SMARTS `[O:1][H]>N->[Cu]>[O:1].[H]`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to `pkg/configurator/server/configurator.go` since it's only used by that file.",
+      "reference": "What happens if the certificate rotation fails and we are prematurely removing this file?"
+    },
+    {
+      "prediction": "I think this should be `s.once.Do(func() { s.introspect() })`",
+      "reference": "does this change belong here?"
+    },
+    {
+      "prediction": "I think this should be `workflowDir` and `cwd` should be `workingDir`.",
+      "reference": "This returns an error"
+    },
+    {
+      "prediction": "I'm not sure if this is a good place to add this. I think it would be better to move it to `libcalico/common/config.go` since it's only used by `libcalico/common` package.",
+      "reference": "If we think we might someday write a native nftables backend, do you think it would make sense to just use generic dataplane configuration? e.g, `dataplane = iptables | ebpf | nftables`, but for now selecting `nftables` uses iptables in nft compat mode?"
+    },
+    {
+      "prediction": "I think we can remove the `require.NoError` as it's already done in the `require.NotNil` above.",
+      "reference": "Looks like that helper function is general enough that it should reside in testhelpers"
+    },
+    {
+      "prediction": "I think this should be `return nil, fmt.Errorf(\"drushrc file already exists\")` instead of `return nil`",
+      "reference": "This isn't incorrect IMO, but I think it would be better to fix this in drupal7PostStartAction and also in drupal6PostStartAction. It seems to me like those were both somehow neglected on this. Use drupal8PostStartAction as example. Congrats on your first golang PR! Please make sure to test it manually."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `beaker_test.rb` under the `provision` section.",
+      "reference": "I'd like to ensure that the error message at least has some reference to the flag that is not allowed. Something like `/ERROR(.+)--hosts/` would work."
+    },
+    {
+      "prediction": "I think we can remove the `if err == selector.ErrSubnetsNotFound {` part here.",
+      "reference": "Should we say \"specifying two public subnets\"?"
+    },
+    {
+      "prediction": "I'm not sure what this TODO is referring to, but it seems like it should be addressed in a separate PR.",
+      "reference": "Issue number? When? Why?"
+    },
+    {
+      "prediction": "I don't think you need to import `speech` here. You can just use `import nvwave`.",
+      "reference": "You should not rely on `PitchCommand` being imported into speech. Please import it from `speech.commands`."
+    },
+    {
+      "prediction": "nit: I think this should be `authType` instead of `auth_type`",
+      "reference": "Is this a constant used in the Nessie project itself? If so, perhaps you might consider a follow up for adding `NessieCatalogProperties` class at some point, to help make them more clear to users looking to adopt Nessie coming from the Iceberg repo itself"
+    },
+    {
+      "prediction": "I think we can remove `ReadLevel` and `MaxReadLevel` fields from the `GetTasks` response.",
+      "reference": "maybe we should rename: MinTaskID -> MinTaskIDExclusive, MaxTaskID -> MaxTaskIDInclusive,"
+    },
+    {
+      "prediction": "nit: I think this should be `ecs-nvidia` instead of `nvidia`",
+      "reference": "You may need to make this configurable if we expect people to be able to use the normal Nvidia runtime on other Linux distributions like Ubuntu or Debian."
+    },
+    {
+      "prediction": "I think this should be `&SubjectPermission{Deny: []string{remote.DenyImports, remote.DenyExports}}`",
+      "reference": "Originally the DenyExport was a subscribe permission because it meant that if on this LN connection, if we deny export of \"foo\" it means that it would reject a subscription (hence subscribe permission) on \"foo\" from the other side. Now you are changing to simply not allowing this server to publish on \"foo\". I am not saying this is wrong, but we completely flip the meaning."
+    },
+    {
+      "prediction": "I think this should be `TESTANY(DR_DISASM_ATT | DR_DISASM_INTEL, DYNAMO_OPTION(disasm_mask));`",
+      "reference": "> k1 src0 src2 -> dst s/src2/src1/ nit: k1 is actually src0 according to instr_get_src(instr, 0) which makes this a little confusing"
+    },
+    {
+      "prediction": "I think this should be `logging.Base().Errorf(\"%s: kmd data dir exists but is too permissive (%o), please change to (%o)\", kc.kmdDataDir, dataDirStat.Mode()&0777, dataDirStat.Mode()&0077)`.",
+      "reference": "I believe that you meant to have the following on the second one, right ? `dataDirStat.Mode()^0077`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to make this change. I think it should be in `protocol`, not in `stream`.",
+      "reference": "The reason a `ByteCount` is used here, is that the H2 mapping in gQUIC requires this layer violation, which is why `CloseRemote` is not part of the public API. This layer violation will be resolved in IETF QUIC"
+    },
+    {
+      "prediction": "This should be `return !IsEffectInSpell(spell_id, SE_DivineAura);`",
+      "reference": "I dig the helper function!"
+    },
+    {
+      "prediction": "This file should not be included in this PR.",
+      "reference": "Please add the Copyright header."
+    },
+    {
+      "prediction": "I don't think we need to import FileSystemTarget here, since it's not used in this module.",
+      "reference": "Is this the best way to import `FileSystemTarget`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem here is that we're using `thread_id_t` as a pointer, which is 64 bits on x86 and 32 bits on x64. So when we try to cast it to `int32_t`, it doesn't work.",
+      "reference": "Wait -- os_local_state_t.tid is thread_id_t though, so we need to read a pointer-sized value via READ_TLS_SLOT_IMM, rather than changing these locals to ints. Maybe have a READ_TLS_TIDSZ_SLOT_IMM or sthg."
+    },
+    {
+      "prediction": ">!clrstack -i is unsupported on this target. The ICorDebug interface cannot be constructed. [](start = 22, length = 34) I think this is the right thing to say.",
+      "reference": "Don't use ! in the messages because it isn't used on xplat. There is a SOSPrefix define that can be used (blank on xplat and ! on Windows). Do we really need 3 newlines?"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(dagster_home_path, candidate_run.__class__.__name__)` instead of `os.path.expanduser(dagster_home_path)`",
+      "reference": "print out what we got here in the error"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Does this need to be global?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `values()` returns a `Seq<V>`, but `map()` returns a `Seq<Map.Entry<K, V>>`. So you can't just call `map` on the result of `iterator().map(Tuple2::_2)` because it's not a `Seq<Map.Entry<K, V>>`. You need to call `map` on the result of `iterator().map(Tuple2::_2).collect(Collectors.toMap(Tuple2::_2, null, Comparator.comparing(Comparator.comparingInt(o -> o.getKey().compareTo(o.getValue().getKey()))))`",
+      "reference": "Yep, could be simplified. Now looks like other *Map.values() impls"
+    },
+    {
+      "prediction": "I think it would be better to use `echo json_encode($this->response->getContent());` instead of `echo $this->response->getBody();`",
+      "reference": "This method will be for debug?"
+    },
+    {
+      "prediction": "I think it would be better to have this as a separate interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to be a part of the `Key` interface. It's not clear to me why we need this to",
+      "reference": "// Key returns the private key for a specified name that was encrypted with the // provided password. If the private key does not exists it creates a new one // with a name and the password, and returns with `created` set to true."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this class. It's not a class in the usual sense, it's a `PackageEntity` that represents a class. I think `Package Class Entity` or `Package Class` would be better names.",
+      "reference": "Adding the type hints revealed that it was necessary to distinguish between a ``PackageEntity`` and a ``ClassEntity``, because the ``ClassEntity`` has additional attributes that were dynamically added in the previous code, which confused ``mypy``."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pandas/_frame_base.py`.",
+      "reference": "These two functions should be available now."
+    },
+    {
+      "prediction": "I think this should be `if file.o.size != file.fd.Size() || file.o.modTime != file.fd.ModTime()`",
+      "reference": "`fi` is what the result of Stat is called elsewhere in this file not `finfo`"
+    },
+    {
+      "prediction": "I'm not sure if this is a good or a bad thing, but I think it's a good thing.",
+      "reference": "I'm confused; are we just dropping these methods without deprecation?"
+    },
+    {
+      "prediction": "I think it would be better to use `std::vector<std::pair<kvstore::KV, std::string>>` instead of `std::vector<std::pair<kvstore::KV, std::string>`.",
+      "reference": "can we use PC's time to version in Distributed Systems?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check if the user is linked to the current user or not, but it works for now.",
+      "reference": "is this \"unless\" actually needed?"
+    },
+    {
+      "prediction": "I think this should be `if (length == 0) { return list; }`",
+      "reference": "We don't accept empty words for lookup anymore, again"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to exit the program, but it's the only way I could get to work.",
+      "reference": "We'll have to account for CI installations being faster than local ones. Do y'all think we should leave it at < 30 green / < 50 orange | >= 50 red or lower our thresholds?"
+    },
+    {
+      "prediction": "I don't think we should have a separate command for this. I think we should have a single command that can be used to turn on/off w3c features.",
+      "reference": "No need for this to be `W3C_` since there is no JWP equivalent"
+    },
+    {
+      "prediction": "I think this should be `::RSpec::Core::Backports::Random`",
+      "reference": "As far as I can tell, there's nothing that creates an instance of `Random` or that calls `rand` or `seed`. Am I missing it? If not, let's remove the `Random` class since we don't really need it and we can move the definitions of `shuffle` into `RSpec::Core::Ordering`. One less type :)."
+    },
+    {
+      "prediction": "I don't think we need the lock here.",
+      "reference": "Is this ever actually used?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `consensus.ProcessTipSet` since it's the main place where we process the tip set.",
+      "reference": "this section of the codebase should be noted as a candidate for caching, and as a place where multiple tipsets is making things extra tricky"
+    },
+    {
+      "prediction": "I'm not sure why this is ARM specific. It seems like it should work on all architectures.",
+      "reference": "Do we need to save the existing value of the stolen reg somehow?"
+    },
+    {
+      "prediction": "I think we should be able to set the maxResults parameter to 1000 or 10000 and it should work just fine.",
+      "reference": "maxResults should come from FLAGS once #244 is submitted."
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "is this wanted?"
+    },
+    {
+      "prediction": "I think this should be `linode-cli` instead of `linode`",
+      "reference": "What exception is actually happening here? AFAIK `pytest.importorskip` with just return `None` if there's nothing to import. Which means that this check'd always return `True`. `pytest.importorskip` is specifically designed to trigger skipping the current test anyway so I don't know why you would wrap it like this."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to do something like: DEFAULT_HEADERS = {'Accept' => 'application/x-www-form-urlencoded', 'Content-Type' => 'application/x-www-form-urlencoded', 'User-Agent' => 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/14.0 Safari/605.1.15', 'Accept-Language' => 'en-US,en-US;q=0.8,en;q=0.5', 'Accept-Encoding' => 'gzip, deflate', 'Connection' => 'keep-alive', 'Accept-Charset' => 'ISO-8859-1,utf-8', 'Keep-Alive' => 100, 'Strict-Transport-Security' => 'max-age=3111111111111111111111111111111111111111111111111111111111111",
+      "reference": "Does it send requests with urlencoded bodies anywhere? I thought it sends only json. Maybe content-type should be `application/json` by default?"
+    },
+    {
+      "prediction": "I think this should be renamed to `--ignore-chromium` or `--ignore-chromium` to be consistent with the other flags.",
+      "reference": "I'm not sure about adding more flags here when we're trying to simplify things, I thought we were going to check for patches changes to decide if we needed to update or not?"
+    },
+    {
+      "prediction": "I think this should be `testutil.SignedTransfer(ta.Addrinfo[\"charlie\"].String(), ta.Keyinfo[\"producer\"].PriKey, 1, big.NewInt(10), []byte{}, testutil.TestGasLimit, big.NewInt(testutil.TestGasPriceInt64))`",
+      "reference": "`testExecutionPb` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "~Method is public and removing it would be a BC Break, you can deprecate it if you want.~"
+    },
+    {
+      "prediction": "I think this should be `%s`, not `%v`. The former makes it clear it's a debug statement. The latter hides the fact that it's a panic.",
+      "reference": "this got me a little confused when reading `build.log`. The err is printed with a log.Error later on but that ends up after the stack trace."
+    },
+    {
+      "prediction": "I think it would be better to add this option to `--flatbuffers-include-prefix` instead of adding it to `--flatbuffers-usage`.",
+      "reference": "Can we make these more explicitly for Rust only? It seems like JS and Go use flags with their name in them."
+    },
+    {
+      "prediction": "I think this should be `InvalidQueryParameter` instead of `InvalidQueryParameter %q`",
+      "reference": "Should unsupported query parameters just be ignored? I think that's more common than error out."
+    },
+    {
+      "prediction": "I think it would be better to say \"reset to :py:attr:`Register._UNSET_NAMESPACE`.\"",
+      "reference": "Actually can we avoid mentioning `Register._UNSET_NAMESPACE`. I thought of it as an implementation detail of the simpler concept of \"being unset\". Perhaps the docs become more natural if we remove the \"If called without arguments or with ...\" part and just say \"you have to call this function without arguments at the end of any file it has been used. That is to ensure blah blah\". Something like that. Just a suggestion."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You need to check if the index is in the range before trying to get the item.",
+      "reference": "I assume it raises COMError? If so perhaps best to just catch that specifically, so as to not hide other more critical errors."
+    },
+    {
+      "prediction": "I think it's better to use `github.com/hacdias/caddy-minify/minify` instead of `github.com/abiosoft/caddy-git`",
+      "reference": "Since minify can generate errors, it should at least go after the errors middleware. And you definitely don't want to be minifying after the gzip writer has closed."
+    },
+    {
+      "prediction": "I think this should be `provider *TracerProvider` and not `*TracerProvider`",
+      "reference": "nit: we could remove the `config` field as it is a duplicate of the `provider.config` field."
+    },
+    {
+      "prediction": "I think this should be `forseti-server-allow-grpc-internal` instead of `forseti-server-allow-grpc`",
+      "reference": "Nit: from the Internet."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems to be the only place where this is used, so I figured I'd put it here.",
+      "reference": "this is internal package, move to bottom and run 'make fmt'"
+    },
+    {
+      "prediction": "I think we can remove this feature completely.",
+      "reference": "Why is this better than ApplicationStopping?"
+    },
+    {
+      "prediction": "Shouldn't this be \"dns01\" and \"http01\" be \"http01\" and \"dns01\"?",
+      "reference": "Maybe worth expanding that these 2 are supported by cert-manager but other values exist"
+    },
+    {
+      "prediction": "I think this should be `$options['src'] = $provider->getHelperProperties($media, $format, $options['src'], $options);`",
+      "reference": "Why was this merged? It should have raised some questions IMO @core23 @OskarStark . It's already in 3 releases now, so we can't revert it can we? How can we fix this? Please have a look at #1065"
+    },
+    {
+      "prediction": "I think it would be better to use `node.jsonRpcConfiguration().getAuthenticationPublicKeyFile().isPresent()` instead of `node.jsonRpcConfiguration().getAuthenticationPublicKeyFile() != null`",
+      "reference": "What if `node.jsonRpcConfiguration().getAuthenticationPublicKeyFile()` is empty string, would that cause a problem here?"
+    },
+    {
+      "prediction": "I think we can remove the `Duration` and `RenewBefore` fields from the `CertificateSpec` struct, since they are not used in the `Certificate` struct anymore.",
+      "reference": "The `omitempty` struct tag does not do anything for non-pointer structs. I think we may need to consider making both of these fields pointers, so that they are excluded from output when not set, and also to make it easier to compare to the zero value. That said, I'm happy to merge this now and open an issue to verify the behaviour is as we want ahead of cutting v0.6.0."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it would be better to have a method that does the comparison and returns null if they are the same, and then use that in the iterator.",
+      "reference": "I think this is should absolutely be `==`, as the interface cannot control the implementation of equals (and it's part of the contract of the enclosing function). Can we add this interface to the exceptions of the rule?"
+    },
+    {
+      "prediction": "I think we can remove `namespaceUID` and `namespaceLabels` fields from `ns` struct.",
+      "reference": "It looks like the UID is the only other thing that we use from `namespace`, so I suggest to copy that out and lose `namespace`, so we don't have to worry about setting it to nil."
+    },
+    {
+      "prediction": "I think this should be `DefaultImage = \"aws/codebuild/amazonlinux2-x86_64-standard:3.0\"`, not `DefaultImage = \"aws/codebuild/amazonlinux2-x86_64-standard:3.0\"`",
+      "reference": "Can we define this constant in the`deploy` pkg instead? this would allow us to keep it private"
+    },
+    {
+      "prediction": "I think this should be `toto/mydir/toto`, not `mydir/toto`.",
+      "reference": "I don't like this fixture name, should be a noun like `initialized_linter` ? But it's outside of the scope of this MR."
+    },
+    {
+      "prediction": "I think it would be better to move this check to the constructor of `JsonPrinter`, so that it can be called only once.",
+      "reference": "Could we keep this as a static reference (JsonPrinter.jsonAsDatasetDto(releasedVersion); and perhaps change the constructor line to just be a static set method and use that here instead? (I suspect it was the warning that 'static methods should be accessed in a static manner' from my IDE that caused me to make this change in the first place...) (I can make a PR to your branch if you want.)"
+    },
+    {
+      "prediction": "I think it would be better to say something like \"If the provided key is none of those, the modifier key cannot be released.\"",
+      "reference": "Keys.COMMAND seems to be an alias to Keys.META. That isn't mentioned?"
+    },
+    {
+      "prediction": "Why do we need both of these? I don't think we need either of them.",
+      "reference": "This should go in `ConfigLocal` (and then re-generate the mocks)."
+    },
+    {
+      "prediction": "I don't think this is correct. If `release_mbid` is `None` then `release_msid` should be `None` as well.",
+      "reference": "Nitty but entry.get for the same thing"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in `libcontainer/nsenter` since that is where the nsenter process is started.",
+      "reference": "Might not hurt to mention the function never returns (since this all ends in `execve`) so `main` never actually runs."
+    },
+    {
+      "prediction": "I think this should be a warning, not an error.",
+      "reference": "nit: add new line after this block"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure it's the right thing to do. I'm not sure it's the right thing to do in general, but I'm not sure it's the right thing to do in this case.",
+      "reference": "I believe there is some javascript that also uses this 60px number for calculations. I think I'd rather keep the number in px instead of rem to make that correspondence easier to see. If 3.25rem != 60px, can you also change the other place where `60` is used to in the JS?"
+    },
+    {
+      "prediction": "I don't think we should be using `retrying` here. I think we should be using the built-in `retry` from the `retry` module.",
+      "reference": "retrying is an installed 3rd party library just like jinja and sendgrid; perhaps try grouping all of them together?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `ArrayList` and `Vector` don't have a `prepend` method.",
+      "reference": "you could always use List.listIterator with previous() and hasPrevious() to traverse all kinds of j.u.List backwards. No need for special cases for ArrayList and Vector."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to do this in a more efficient way, like storing only the difference of bitmaps or steps followed to edit.",
+      "reference": "remove the bitmap of index 1 from the list not the 0th one. because when we keep on undoing, it would be better if we end up with the original image rather than some randomly edited image. I am not sure whether only just removing bitmap from the list would clear memory. I think you should call bitmap.recycle before removing that bitmap from the list."
+    },
+    {
+      "prediction": "I think this file should be renamed to reflect the copyright year.",
+      "reference": "this is a existing file so let's keep 2019"
+    },
+    {
+      "prediction": "I think this should be a `Map<Integer, ConnectionType>`.",
+      "reference": "can you change this instead to just `type.toString()` and then you wouldn't have to expose the getBitMask in the enum. (Alternatively you could have used `type.hashCode()` but that doesn't feel as nice)"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it, but it's the only way I could get it to work.",
+      "reference": "AsReadOnly() would be better here"
+    },
+    {
+      "prediction": "I don't think we need to add `partial_sync` to the `changes` dictionary. We can just use `newModifications` and `deletions` here.",
+      "reference": "I suggest that we use `query_based_sync` instead of `partial_sync`."
+    },
+    {
+      "prediction": "I think we should add a note here that these signed URLs will not work unless the PUT request includes a \"x-ms-blob-type\" header set to \"BlockBlob\".",
+      "reference": "How does the user use this? By converting the type using `As` function and add the header? Maybe add an example on how."
+    },
+    {
+      "prediction": "I think we can remove the `ApiMethodParamTransformer` from this class as it is not used anymore.",
+      "reference": "I'm trying to understand the effects of this change. Is setting this to something other than the old default (that is, `initCodeTransformer = new InitCodeTransformer()`) generally necessary, or are we doing this only to support Python?"
+    },
+    {
+      "prediction": "I don't think this is accurate. The docstore doesn't take a zero-value key, it just doesn't care if it's `nil` or not.",
+      "reference": "\"Constructor\" isn't a standard term in Go or this project, although we use it informally amongst ourselves. And I think it will confuse people coming from languages like Java. So can we leave this as it was?"
+    },
+    {
+      "prediction": "I think this should be `ctu_failure` instead of `ctu_on_demand_failure`",
+      "reference": "Why do we have to change the test dir?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `TEXTAREA` and `SELECT` cases since they are the only elements that have a `summary` attribute, but I'm not sure if there is a better place.",
+      "reference": "This should test the flattened tree instead. details > summary works across shadow tree boundaries."
+    },
+    {
+      "prediction": "I think this should be `MvvmCross.Logging.MvxAndroidLog` instead of `MvvmCross.Logging.MvxAndroid`",
+      "reference": "Not sure I am a big fan of these duplicated Log classes."
+    },
+    {
+      "prediction": "I think we should use a more descriptive name for this error. It's not a cleanup error, it's not a clean up error, it's not a clean up error. It's an error in the domain verification process.",
+      "reference": "It's long-winded, but the convention is to give constants a common prefix which, see `reasonDomainVerified` ."
+    },
+    {
+      "prediction": "I think this should be `20.0` instead of `20.0`",
+      "reference": "Using lanes later down, this threshold could be reduced for similar effects. Otherwise we look a bit to far."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this struct. It's a bit confusing because it looks like it's a container, but it's not really a container because it doesn't have any container-specific fields. Maybe something like `conda_env` or `conda_env_info` would be better?",
+      "reference": "Let's get more verbose about names. struct `wq_conda_environment`: if it can really only be used for conda. struct `wq_software_environment`: if it has potential use outside of conda."
+    },
+    {
+      "prediction": "I think this is the only place where we're using the `lastPageInPaginatedResp` variable. I think it should be removed.",
+      "reference": "Should we set this back to `false` inside each `t.Run`? so that we can have more than one testcase that can have paginated responses"
+    },
+    {
+      "prediction": "I think this should be `ImmutableList.of()` instead of `ImmutableList.copyOf()`.",
+      "reference": "What about using `dataManifests == null || deleteManifests == null`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `IndexSearcher.java`.",
+      "reference": "The `HitsThresholdChecker` should be created once and shared within the collectors ? We also don't need to use the `GlobalHitsThresholdChecker` if the executor is null or if there is a single slice."
+    },
+    {
+      "prediction": "I think this is the only place where we need to use the cookie package. I'm not sure if we need to use it in other places.",
+      "reference": "move this to below `antrea.io` import section"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to express it. I think it's better to say something like: > This extension method should only be used on <see cref=\"Activity\"/> instances that were created via older means that predate <see cref=\"ActivitySource\"/>.",
+      "reference": "Seems likely to confuse people. What if we moved it into ActivitySourceAdapter and made it private?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this error message. I think it should go in `eth/transaction.go` since that's the place where the error message is actually generated.",
+      "reference": "gas price don't need to be a multiple cuz we support allll gas prices now"
+    },
+    {
+      "prediction": "I don't think we need this method. We can just remove it from the interface.",
+      "reference": "We are exposing implementation details in the interface. What if there is no processes involved in an implementation of `ITestRequestSender`?"
+    },
+    {
+      "prediction": "I think it would be better to use `traceback.format_exc()` instead of `sys.exc_info()[1]`.",
+      "reference": "It would be awesome not to handle the broad exception here. Instead, raise a custom exception, something like `ModelNotSetException`, from the `require_model()`, and then handle it here with a nicer message to tell the user to set the model first."
+    },
+    {
+      "prediction": "Shouldn't this be `customize_theme_page`?",
+      "reference": "This already exists under the `theme` lang key, please remove this."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `GoogleCloudPlatform` component, since that's where the flag is set.",
+      "reference": "Let's add a simple inline HOC around the default export below to provide the value as a prop (no need to introduce a reusable function for this yet)."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "I believe that you end up printing out the result of the assignment here instead of the netmask."
+    },
+    {
+      "prediction": "Do we need this? I don't see it being used anywhere.",
+      "reference": "We have to update the filename here."
+    },
+    {
+      "prediction": "I think this should be `schema snake_case emits snake_case` or `schema snake_case emits snake_case` to be grammatically correct.",
+      "reference": "The `--cpp-field-case` looks like `--cpp-field-style` instead of `-case`. - 'unchanged' - leave unchanged (default); - 'upper_camel' -upper camel case; - 'lower_camel' - lower camel case."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `origin_changed` variable above.",
+      "reference": "I haven't thought about this a lot yet, but if we were to try this change, we'd still need to keep the code to delete it from the `.origin` file for backcompat."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this. I think it's better to put it in `org.hyperledger.besu.crypto`",
+      "reference": "In my opinion I think it is possible to add `private final` here"
+    },
+    {
+      "prediction": "I'm not sure this is correct. The `exit` instruction should jump to the `exit` label, not the `deny` label. The `exit` label should jump to the `exit` instruction, not the `deny` instruction. The `deny` label should jump to the `exit` label, not the `exit` instruction.",
+      "reference": "Feel like \"exit\" doesn't convey that the packet will be dropped. \"drop-and-exit\" or \"error-exit\" maybe?"
+    },
+    {
+      "prediction": "This should be `a Main actor must have a create constructor which takes an Env parameter`",
+      "reference": "How do you feel about \"The Main actor\" instead of \"A Main actor\", while we're already here changing the message?"
+    },
+    {
+      "prediction": "I think this should be `ZK_READ_PERM(\"zk-read\", unmodifiableSet(new HashSet<>(asList(\"*\", null)))),` and not `ZK_READ_PERM(\"zk-read\", null)`.",
+      "reference": "I cannot see that this new permission is used anywhere? And if the new zk handler is covered by `zk-read`, should not also existing `ZookeeperInfoHandler` handler implement PermissionNameProvider and declare the same permission, for consistency?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to have a function like `isinstance(x, np.ndarray)` that returns `True` if `x` is an instance of `np.ndarray` and `False otherwise`. Then you can call `isinstance(x, np.ndarray) if isinstance(np.ndarray, type(x)) else ...`",
+      "reference": "Is this reasonable?"
+    },
+    {
+      "prediction": "We need to add these transforms to the `TRANSFORMS` attribute of the `AutoAugment` class.",
+      "reference": "We may move this TODO to Line15."
+    },
+    {
+      "prediction": "Shouldn't this be `ConfigQuery::checkAvailableStock() && return $this->checkStockNotEmpty();` ?",
+      "reference": "You have to verify the return type. If it's a reponse, return it. Otherwise do nothing."
+    },
+    {
+      "prediction": "I think this should be `3==int32_t, 4==int64_t` and not `1==float, 2==double, 3==int32_t, 4==int64_t`",
+      "reference": "No. Keep the header include and do `extern template ...` to skip the instantiation from that compile unit. (I pushed a fix directly to your branch)"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it's fine to leave this as it is for now, but I'm not sure what the best way is to handle this.",
+      "reference": "these tests are currently somewhat nonsensical. This code expands to `a, b = {}.keys` when it really should be expanding to `a, b = {}.keys()` -- though fixing this causes the test to fail so I suspect something worse is going on here that I don't quite understand?"
+    },
+    {
+      "prediction": "I think this test is failing because of a bug in the AvroNameMappingTest class. It should be `assertNull`, not `assertNull(\"location.value.long, should not be read\", projectedL1.get(\"long_r2\"));`",
+      "reference": "Why did this need to change?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface. I think it should be moved to `action/branch` package.",
+      "reference": "/go/pkg/mod/golang.org/x/xerrors@v0.0.0-20190410155217-1f06c39b4373/adaptor_go1_13.go:16:21: Frame not declared by package errors (from `typecheck`)"
+    },
+    {
+      "prediction": "I think this should be called \"configName\" and not \"collection.configName\"",
+      "reference": "Probably doesn't go here because I think it's specific to the HTTP API layer. This class is too internal to declare such a name."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `zipkin.yml` file.",
+      "reference": "in preparation of directly writing the /health endpoint one day.. we should have content checks."
+    },
+    {
+      "prediction": "I think this should be `string` and not `[]string`",
+      "reference": "question (non-blocking): Is there a benefit from using `%q` instead of `%s`?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "I would suggest to keep two empty lines between functions in cpp files"
+    },
+    {
+      "prediction": "I'm not sure about this test. It seems to me that if the other stream synchronizes on the marker, then the other stream should be able to send the marker to the other stream, and then the other stream should be able to receive the marker from the other stream. So I'm not sure why this test is failing.",
+      "reference": "This change causes the test to fail on both rocm-1.9.x and rocm-head."
+    },
+    {
+      "prediction": "I think it would be better to have a `Fixer` interface, and have a `RuleViolationFixer` implement that interface. This way, we don't need to add a new interface for each rule violation, and we don't need to change the `RuleViolationFixer` in order to fix a new rule violation.",
+      "reference": "I'd consider moving all fix related stuff to a distinct package to avoid contaminating the base package"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The original include directory is not guaranteed to be in the same directory as the compiled files.",
+      "reference": "Use `os.path.join` for path concatenation."
+    },
+    {
+      "prediction": "I don't think we need to add `YolactProtonet` and `YolactSegmHead` to the `__all__` list.",
+      "reference": "Use upper case: YOLACTHead, YOLACTProtonet, YOLACTSegmHead"
+    },
+    {
+      "prediction": "I think this should be a `*crypter` instead of a `crypter`",
+      "reference": "The driver tests should be updated to use the concrete type instead of the driver directly; right now your test coverage of the concrete type is 0%."
+    },
+    {
+      "prediction": "I think this should be `throw new DocblockParseException('Badly-formatted @method parameters for ' . $method_entry;`",
+      "reference": "I would prefer the message to be more actionable. Like 'There should be no space between & and the variable name' or something similar."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. The `jjtGetChild` method is only called from within the parser, so it should be called from within the parser, not here.",
+      "reference": "The image attribute is not supposed to be the text of the node. I'd rather keep it separate (the previous `@Text` attribute was fine). A practical reason for that is in the future, other languages may have a way to get the text of their node, in which case that wouldn't be fetched with `getImage`, for compatibility, but probably with a `getText` method, or `getSourceCode` like in #1728. Also we *may* deprecate or otherwise put `getImage` out of use as soon as 7.0, given its contract is super loose."
+    },
+    {
+      "prediction": "I think this should be `@system_exclusion_patterns = [] + @system_exclusion_patterns` instead of `@exclusion_patterns = [] + @system_exclusion_patterns`",
+      "reference": "The `[] +` seems weird to me. Why is it there?"
+    },
+    {
+      "prediction": "Layout/AlignHash: Align the elements of a hash literal if they span more than one line.",
+      "reference": "Not sure if this constant is defined at the right place style-wise. Maybe move it up to before 'belongs_to' like DISCOUNT_TYPES in coupon.rb?"
+    },
+    {
+      "prediction": "I think this should be `return isEmpty || isEmptyYaml([]byte{string(k)})`",
+      "reference": "What about a line with just spaces/tabs? Or a line with a couple spaces followed by a `#`?"
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "if funder_is is not blank (L375) there is not need to check if org_id is not blank (L379) since you will never enter in the if (L374)."
+    },
+    {
+      "prediction": "I think this should be `auto_fire = !GetTarget()->IsWithinDistance(GetPlayer()->GetMapId(), GetTarget()));`",
+      "reference": "Don't need this-> here."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `google/cloud/compute/engine/metadata.py`.",
+      "reference": "I'm a little concerned about relying on a private module, as they could change the implementation at some point, I'd like to have some test to validate this is working as intended."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for a dict, but it's the only way I could come up with.",
+      "reference": "There might other values that we don't support here (eg. `[]`). So it might be safer to check for the supported types instead (string or number)."
+    },
+    {
+      "prediction": "I think this should be `const char*` instead of `std::string`.",
+      "reference": "FYI, It won't have any benefit to move from trivial types."
+    },
+    {
+      "prediction": "I think this should be `secrets-ssm-environment-variables` and not `secrets-ssm-environment-variables`",
+      "reference": "i missed these discussions- but ` \"secrets-ssm-environment-variables\"` is what was agreed upon with cp?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "whats the difference between this package and `github.com/golang/protobuf` ? or did vscode just decided this was the package it wanted to used?"
+    },
+    {
+      "prediction": "I think this should be done in the constructor of `AddRepositoryDialog` instead of here.",
+      "reference": "Just asking: What does this line do? Does it add the stylesheet from the parent to the dialog?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be in `requirements-dev.txt` because it's not a requirement for everyone to use scikit-learn, but only for people who want to run tests on it.",
+      "reference": "Including the scikit-learn dependency here would pull in scikit-learn for everyone who depends on Dagster. If you put it in the setup.py under docs_snippets, we'd avoid that problem (although I think it's already there)."
+    },
+    {
+      "prediction": "I think it would be better to use `std::set` instead of `std::vector<std::string>`.",
+      "reference": "Why remove this?"
+    },
+    {
+      "prediction": "I think you can remove this test now.",
+      "reference": "Why remove this? It seems like we should keep this around as a regression test, unless we can guarantee this will never be called without a model set (does your new code guarantee that?)"
+    },
+    {
+      "prediction": "I think this should be `undefined` or `[]` instead of `global._googlesitekitUserData?.permissions || {}`.",
+      "reference": "Because this defaults to `{}` if `global._googlesitekitUserData?.permissions` is `false`-y, the checks below around `state.capabilities` always result in the `!! capabilities === true` path. This shouldn't have a default value of `{}` if there's the possibility that `global._googlesitekitUserData?.permissions` can be `undefined`/`false`/`null` legitimately."
+    },
+    {
+      "prediction": "I think this should be `log_handler.change_log_capacity(capacity)`",
+      "reference": "Please add a space after the comma here (generally, with arguments there's always a space after commas)."
+    },
+    {
+      "prediction": "I think it would be better to say \"Flags defined in those class must be used with cautiousness, and strictly reserved to experimental EIPs.\"",
+      "reference": "From the context, should it say \"Flags defined in this class must be used with caution...\" ?"
+    },
+    {
+      "prediction": "I think it would be good to have a test for this file.",
+      "reference": "We don't need this file, we can use the one that is in the grid-ui directory"
+    },
+    {
+      "prediction": "I think this should be `infoImportedKey` instead of `infoImportedKey`",
+      "reference": "\"Exported key for account\"?"
+    },
+    {
+      "prediction": "I think this should be `\\n` instead of `\\r\\n`",
+      "reference": "Normally we use Environment.NewLine unless you really want it to always be \\r\\n"
+    },
+    {
+      "prediction": "I don't think we need this function, we can just call `c.VerifyHeader(header, uncles)`.",
+      "reference": "why did you remove `verifyHeaders`?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Thanks for cleaning up these deprecated calls"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this validation. I think it should be moved to the `User` model's `__init__` method.",
+      "reference": "@akarve this breaks error catching logic on the front end, I'll fix this. That's why I prefer to use some machine-friendly error codes that won't change frequently."
+    },
+    {
+      "prediction": "I think we can remove this function now, it's no longer used.",
+      "reference": "We support process remoteQuery for Service. Why are we removing it ?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better idea. I'm not sure what the best thing to do here is.",
+      "reference": "This is specifically for testing?"
+    },
+    {
+      "prediction": "Shouldn't this be `resourcetags` instead of `resourcetags->getStatistics()`?",
+      "reference": "I think this is related to the social stats and probably does not need to be removed."
+    },
+    {
+      "prediction": "I think this should be `jsdoc/require-description-completed-sentence`",
+      "reference": "Yep, jsdoc again .. It seems that when the plugin description is wrapped within `eslint-disable/enable` expression it's not generated at all. After adding the `@class AutoColumnSize` tag right after the `@plugin` tag the plugin appears in the docs. Please review the other plugins."
+    },
+    {
+      "prediction": "I think we can remove the `len(values) == 0` check here.",
+      "reference": "This usage of the result doesn't look quite right. I would not expect the first page of values to be empty if there were multiple pages of results. This also obscures the error case when no values are returned (handled in the next `if` block) by returning a more generic error from the Azure SDK from `result.NextWithContext()`."
+    },
+    {
+      "prediction": "I think we can remove the `err == nil` check here.",
+      "reference": "unknown field AddressDetails in struct literal (from `typecheck`)"
+    },
+    {
+      "prediction": "I think it would be better to use `aws.StringValue(region)` instead of `aws.String(region)` here.",
+      "reference": "Just noticed, are we specifically missing the addition of `SharedConfigState: session.SharedConfigEnable` ?"
+    },
+    {
+      "prediction": "I think this should be `oldTimeIntensities != null && oldTimeIntensities.NumPoints > 0` instead of `oldTimeIntensities != null`",
+      "reference": "Not sure what this is. Bad merge?"
+    },
+    {
+      "prediction": "I think we can remove the `if acc.claimJWT != \"\"` check here.",
+      "reference": "I think this may be a RW lock."
+    },
+    {
+      "prediction": "I think this should be `package helpers` as well.",
+      "reference": "Perhaps we should tuck this away into a dedicated subdir and name it `config` or something similar? Or maybe it would be happy living in `common`?"
+    },
+    {
+      "prediction": "This should be `spec.must_learn & 0xf` not `(spec.must_learn & 0xf) && !spec.has_learnt`",
+      "reference": "Doing a select query inside of a loop like this is not ideal. We should fetch recipes once and then loop through it in memory to perform this check"
+    },
+    {
+      "prediction": "I think this should be `array([1, 2, 3], dtype=object)`.",
+      "reference": "too long here. (104 > 100)"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the `MessageHandlers` section of the `Handle` method.",
+      "reference": "Use TryGetValue instead of catching exception"
+    },
+    {
+      "prediction": "I think we can remove the `as: user` param from the `visit` and `click_link` calls below since it's no longer being used.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `add_submessage_to_cache()` function.",
+      "reference": "I think this method should be either: - a static method of `RTPSWriter` to avoid a StatelessWriter redefinition of the function. - a setter in the `CacheChange_t` struct."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to call this. I think it should be in `Http2ConnectionAbortedException`, but I'm not sure if that's the right place to call it.",
+      "reference": "You add a stream back reference for trailers, and all of the sudden we have tight coupling!"
+    },
+    {
+      "prediction": "I think this should be `using System.Linq.Expressions;`",
+      "reference": "I don't think we use `AspNetAmbientContext` anymore since . We can probably delete this file."
+    },
+    {
+      "prediction": "I think this should be `__webpack_require__(1)`, not `__webpack_require__(0)`.",
+      "reference": "Please avoid committing build artifacts, as they produce unnecessary conflicts and noise."
+    },
+    {
+      "prediction": "`callable` is not a good name for this function. `Callable` is a type, not a function.",
+      "reference": "this is vendored code so i would hesitate to make random changes to it"
+    },
+    {
+      "prediction": "I'm not sure if we should expose the EthAddress here. It's not part of the protocol, and we don't want to make it publicly available.",
+      "reference": "We are using a different style for JSON tags. `ethAddress` - camalCase `referral_code` - snake_case"
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. I think it should be something like `DefaultTraceContextTypeName` or `DefaultRequestContextTypeName`.",
+      "reference": "Should this be `\"Microsoft.AspNetCore.Http.DefaultHttpContext\"`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `Object#copyRange` and `Object#lengthOf`.",
+      "reference": "removed asserts from `Vector` as it's stable enough and it may hinder inlining, even if turned off :/"
+    },
+    {
+      "prediction": "I think it would be better to have a `get_message_definitions_by_id` method that takes an id and returns the corresponding message definition. This way you don't have to disable the message and you don't have to worry about the size of the class.",
+      "reference": "Do we want to raise on the function or on the decorator?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix, but it's the only one I could come up with.",
+      "reference": "How complex is to to replace `album_*` with `release_*` everywhere in the `Listen` class and places that use it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to do `n.add(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t, path)`, and then `n.remove(t,",
+      "reference": "You haven't changed any of the callers of `.Add` to check the return value, so we're now ignoring duplicates. Please fix that."
+    },
+    {
+      "prediction": "I think this should be `const vNode = getNodeFromTree(el);` instead of `const vNode = axe.utils.getNodeFromTree(el);`",
+      "reference": "@straker should this not push to `vnode._cache.isHiddenWithCSS`?"
+    },
+    {
+      "prediction": "I think this is the wrong package for this interface.",
+      "reference": "\"an interface\" in conceptual sense vs \"interfaces\" or \"set of interfaces\", referring to the Go interfaces - database/sql uses the latter, should we?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could come up with at the time of writing this.",
+      "reference": "This now checks nearly all errors that can occur. How about explicitly listing cases where we want it to retry? I think there are not many cases where we want that, as the many added conditions in the last years show :)"
+    },
+    {
+      "prediction": "I think this should be `PrefetchStatus, err := brq.checkCaches(ctx, kmd, ptr, block)`",
+      "reference": "Can you make this accept an action instead?"
+    },
+    {
+      "prediction": "I think we can remove this `Given` statement now, since we already have it in the `Given` statement above.",
+      "reference": "Tab inconsistency here (3 spaces instead of 2 spaces)"
+    },
+    {
+      "prediction": "why 12500 and not 12000?",
+      "reference": "these change means we are changing epoch to 1 hour?"
+    },
+    {
+      "prediction": "I think this should be `PANEL_DESCRIPTION_WIDTH = 544` instead of `PANEL_DESCRIPTION_WIDTH = 545`",
+      "reference": "Why this is defined in the middle of imports?"
+    },
+    {
+      "prediction": "I don't think this should be in this PR.",
+      "reference": "Could you move this to the original position?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to move this. I think it should be in `pkg/deployment/stage.go` instead of `pkg/deployment/source.go`",
+      "reference": "Is there a test that'd have caught this?"
+    },
+    {
+      "prediction": "I think this should be `&visibility.ListWorkflowExecutionsRequest{`",
+      "reference": "ListWorkflowExecutionsRequest <- this can be nil?"
+    },
+    {
+      "prediction": "I think this should be `forge api v1 canonical source` and `forge api v3 canonical source`",
+      "reference": "Do we need to continue to support the old link, or is it dead dead dead?"
+    },
+    {
+      "prediction": "I think this should be `GlobalTags` instead of `GlobalTags` and `HeaderTags` instead of `HeaderTags`.",
+      "reference": "Do we need `ConcurrentDictionary`? `Dictionary` can be safely read from multiple threads. See `GlobalTags`."
+    },
+    {
+      "prediction": "I think this should be `seelog.Infof(\"Server OSFamily: %s\", config.OSType)`",
+      "reference": "this seems like more of a Debugf statement"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be in the `ValidateConcurrentBarrierAtSubmit` function.",
+      "reference": "We're capturing cb_state non-const, but the function is const so, safe enough, but after going to reader/writer locks we're all going to have to key a careful eye on anything in this pattern."
+    },
+    {
+      "prediction": "I think this should be `if req.Config.Endpoint != nil || req.Config.EndpointID != \"\" { return }`",
+      "reference": "Probably want to wrap the endpoint discovery work that follows in this `if` statement instead of returning early. This will make it easier to add additional request code generation logic in the future. e.g. if we add any code generation after the endpoint discovery block this check will cause it to be skipped."
+    },
+    {
+      "prediction": "I think this should be `Error` instead of `JSX twice`.",
+      "reference": "This test was giving a false positive because my change caused it to throw a different error that contained `createElement` when it should've been throwing this error. Caught this by looking at the code coverage and noticing that the line under the condition I changed was no longer covered lol."
+    },
+    {
+      "prediction": "I don't think we need this function at all.",
+      "reference": "I think this is added to wrong file. You wanted to add to cmd/agent/, right?"
+    },
+    {
+      "prediction": "I think this should be `String elementText = element.getAttribute(\"text\");`",
+      "reference": "Each `ExpectedCondition` implements `java.util.Function` These are expected to be stateless. This condition will leak previous `elementText` on the second usage, which doesn't seem ideal."
+    },
+    {
+      "prediction": "I think this should be `DeviceID string` and not `DeviceID   string`",
+      "reference": "Could we consider a more explicit name, like `devicePciAddress`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "Why change this from a foreach? I can't see it gaining anything here and code styles shouldn't change just for the sake of it."
+    },
+    {
+      "prediction": "Why not use `simplexml_load_string($response->getRawResponse(), $parser, true, false, false);` ?",
+      "reference": "Is this supposed to be part of this PR?"
+    },
+    {
+      "prediction": "I think we can remove the `actualNode` parameter from this function, since it's not used anywhere else.",
+      "reference": "DRY, worth extracting this method to axe.utils"
+    },
+    {
+      "prediction": "I think this should be called `applicationName` and not `applicationName`",
+      "reference": "I think there's another spot for this in DefaultCapabilityMatcher"
+    },
+    {
+      "prediction": "This should be `except Exception as e:` instead of `except OSError as e:`",
+      "reference": "`QtOSError` inherits `OSError`, so I don't think it's needed to list them both here."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the subscriber.",
+      "reference": "Instead of patching vendor, let's update the version of sdk-go to a more recent one that doesn't have this line."
+    },
+    {
+      "prediction": "I think this should be `[]string{\"BUILD\", \"PULUMI_BUILD_FILE_NAME\"}`.",
+      "reference": "Don't think this should be here. The default is set somewhere else."
+    },
+    {
+      "prediction": "I think this should be `if self._last_status[\"progress\"] >= 100 or self._last_status[\"status\"] == \"STOPPED\"`",
+      "reference": "Not really important in this PR, but does BlazeMeter describe \"progress\" values anywhere? It might be nice to have a set of constants like `PROGRESS_DOWNLOADING_IMAGE`, `PROGRESS_BOOTING`, `PROGRESS_RUNNING_TEST`, etc in our BZA client."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this method. I think it would be better to put it in the `ContentFile` interface.",
+      "reference": "nit: `<p>` after the line"
+    },
+    {
+      "prediction": "I think this should be `(\"fonts\", glob(\"fonts/*.ttf\"))` and not `(\"fonts\", glob(\"fonts/*.ttf\", recursive=True))`",
+      "reference": "Should we include the files with otf extension here, too? If not, I wonder why we do allow them in the source but we don't include them as per the setup"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `dequeueAll` function, since that's where the blocks are dequeued from the queue.",
+      "reference": "i think it's OK to leave this?"
+    },
+    {
+      "prediction": "I think it would be better to move this to the `org.apache.flink.table.types.logical` package.",
+      "reference": "Does this need to be public? The only reference to `FlinkTypeVisitor` that I see in this PR is here, so I'm not sure why this is needed."
+    },
+    {
+      "prediction": "This should be indented 4 spaces, not 2.",
+      "reference": "No line at end of file warning"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the blockchain processor, but I'm not sure if there is a better place.",
+      "reference": "This looks unused, shouldn't it point to inner processor in any way?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the scan itself.",
+      "reference": "There may be a few isolated cases where the FirstN node is added during preCodeGen. Please see GenPreCode.cpp RelRoot::preCodeGen(). The example given there about Order by where sort is added in optimizer, or a FirstN where the N value is to be specified with a param seem to be cases where we would add the FirstN later. Will current change cause such views to marked as updateable? This is minor and could be resolved later, if any change is necessary."
+    },
+    {
+      "prediction": "I think this should be `public static String getStaticLangStreamingReturnTypeName(Method method, MethodConfig methodConfig)`",
+      "reference": "I think you should compose the %sClient part using the Name class, and then do the remaining composition with plain concatenation. Side question: why does the return type name look like it is a client type name?"
+    },
+    {
+      "prediction": "I think this should be `padding: 0, height: 0, width: 0` and not `padding: 0`",
+      "reference": "All stories in this file also need the default padding."
+    },
+    {
+      "prediction": "I think this should be `MeterName = \"PrometheusMiddleware.Meter\"`.",
+      "reference": "Curious - what would be the actual value?"
+    },
+    {
+      "prediction": "I think it would be better to use `datasets.__version__` here.",
+      "reference": "Is it necessary to record the palette in the checkpoint? Any reason?"
+    },
+    {
+      "prediction": "I don't think we need to import `socketio` in this file.",
+      "reference": "Not sure how to best do this without creating a new connection to the server everytime. Do this in another thread?"
+    },
+    {
+      "prediction": "I think this test is a bit too specific to this PR. Can we move it to a different PR?",
+      "reference": "should we have a similar test for gsa18f procurements?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be.",
+      "reference": "change `v.recipient` to address.Address, so can use `AccountState(v,recipient)`"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "Open to doing away with this altogether and maybe injecting this logic somehow into `requires_approval_notice?` in LinearDispatcher."
+    },
+    {
+      "prediction": "I think this should be `int` instead of `int?`",
+      "reference": "remove this and always enable tracer, just set the default timeout to something higher (20 seconds)"
+    },
+    {
+      "prediction": "Shouldn't this be a `Optional<URI>`?",
+      "reference": "Can we give this field a more descriptive name? When I use a variable named `icon` I normally expect an `Image` object."
+    },
+    {
+      "prediction": "I think this should be `Include` and `Exclude` instead of `in` and `ex`.",
+      "reference": "can't these just use the global `include` and `exclude` flags?"
+    },
+    {
+      "prediction": "I think this should be `rois = rois.to(dtype=torch.float32)`",
+      "reference": "Is this still necessary?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm not sure if it works for everyone. I'm not sure if it's the best way to do this. I'm",
+      "reference": "Interesting! Happen to catch an exception message or anything I can look into?"
+    },
+    {
+      "prediction": "Shouldn't this be \"when the promise is fully resolved\"?",
+      "reference": "did this get moved to mpromise? I don't recall if it's in that lib or not."
+    },
+    {
+      "prediction": "I don't think we need to add `Ports` field here.",
+      "reference": "json objects should be `camelCase`"
+    },
+    {
+      "prediction": "I think this should be an `int64` instead of a `uint64`.",
+      "reference": "QuarterGiBSectorSize? SectorSize265MiB? Spelling out 256 seems overly verbose."
+    },
+    {
+      "prediction": "I think we can remove the `using OpenTelemetry.Tests;` and `using OpenTelemetry.Trace;`",
+      "reference": "Maybe with this change we can remove `using OpenTelemetry.Logs`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to check if the value is either \"on\" or \"off\" instead of \"on\" or \"off\" and \"on\" or \"off\" as the default.",
+      "reference": "Fix this name too."
+    },
+    {
+      "prediction": "I think this should be `int(record_get_field_value(self.record, \"001\"))`",
+      "reference": "why this is an issue?"
+    },
+    {
+      "prediction": "I think this should be `bool $was_static = false`",
+      "reference": "The condition should be inverse?"
+    },
+    {
+      "prediction": "I think this should be `public static final String AZKABAN_FLOW_NODE_TYPE = \"flow\";`",
+      "reference": "this seems to be a property key name, so should it be a inside configurationkey.java?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we need to make sure that the volume is fully destroyed before attempting to destroy it. If the volume is not fully destroyed, then we should not be attempting to destroy it.",
+      "reference": "G204: Subprocess launching should be audited (from `gosec`)"
+    },
+    {
+      "prediction": "I think this should be `True` by default.",
+      "reference": "Please add docs here. And also warn users that this mode should be avoided whenever possible, because any blocking IO will make the keep-alive-thread not run."
+    },
+    {
+      "prediction": "I think this should be `Object.keys( dimensionFilters ).every( (dimension) => typeof dimensionFilters[dimension] === 'string' );` instead of `Object.keys( dimensionFilters ).every( (dimension) => typeof dimensionFilters[dimension] === 'string' );`",
+      "reference": "We shouldn't require dimension values to be a string. They just need to be scalar values (probably we could check whether it's either a string or a number). Something more important to cover in the validation here though is to ensure that a map of `dimensionName => dimensionValue` is passed. The keys here actually need to be strings. For example someone shouldn't be able to pass an array of values."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "let's do a pointer cast of nil, similar to the `MapRouter` above"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this action. I think it should be moved to the `settings.php` file.",
+      "reference": "action should use a prefix of `hfg` rather than `neve` as we plan to bootstrap this as a standalone library. Moreover, previously filter/actions used in this library was using the same pattern."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `Http2Options` class.",
+      "reference": "I'm halfway tempted to disable this for all https connections. Do you know of any clients that actually renegotiate for any reason?"
+    },
+    {
+      "prediction": "I don't think this is a good name for this class. It's not just checking that an element is visible, it's checking that an element is visible and interactable. I think something like `expect_any_element_is_interactable` would be better.",
+      "reference": "shouldn't **call** return a boolean?"
+    },
+    {
+      "prediction": "I think it would be better to move this to the `awsdynamodb` package.",
+      "reference": "I don't think you need the blank import here. This one uses awsdynamodb directly."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be moved to `DeleteFilter.java`.",
+      "reference": "I think this should be initialized to `null` instead of a predicate. There is no need to run an extra predicate (with an extra method dispatch for each row in a data file. That's a tight loop so we should do more work here to avoid it. Instead of using `isDeleted.or`, this should test whether `isDeleted` is `null` and either initialize `isDeleted` or call `isDeleted.or`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `build_clangsa.py` under the `--ctu` option.",
+      "reference": "What is this and the next entry in config_handler used for?"
+    },
+    {
+      "prediction": "I think this should be `memory.Plugin{}` instead of `memory.Empty`",
+      "reference": "will your editor integrate `gofmt`, `goimports`, etc...?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think you should check if the intent is not null and if it is not null, you should extract the state from the intent and assign it to `mState`",
+      "reference": "looks like the code is not properly formatted. for example here are some missing spaces in **if** command. please take care and reformat the code using default android formatting."
+    },
+    {
+      "prediction": "I think this should be `type=str` instead of `type=argparse.SUPPRESS`",
+      "reference": "Please add some valid examples to help message like `\"4:10\"`, `\"4:\"`, `\":10\"`"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Is it only Symbols we're worried about converting? Is it not safe to just call `to_s` anyway?"
+    },
+    {
+      "prediction": "I think this should be `const BlockTimeTest = time.Millisecond * 500`",
+      "reference": "Note for those who come across this later: It was moved to `testhelpers.mining.go` so that `testhelpers.NewDaemon` and the `mining/worker_test.go` can share it."
+    },
+    {
+      "prediction": "I think this should be `select { |trail| trail.incomplete? && !trail.in_progress? }`",
+      "reference": "What about `sort_by(&:in_progress?)`? Maybe with a `.reverse` thrown in?"
+    },
+    {
+      "prediction": "I think it would be better to have a `FlagGRPC` and a `FlagGRPCEnabledHostVerification`",
+      "reference": "Flag is removed because it is only gRPC now."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this option. It's a little confusing because it's not just a template, but it's also not just a cookiecutter template. What about something like `--template-path path` or `--template-dir path`?",
+      "reference": "Might be better to rephrase to: Please refer to the ``init scenario`` command in order to generate a custom ``molecule`` scenario. Since you aren't customizing the default scenario since it already exists, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be moved to `pathlib.py` since it's only used in that module.",
+      "reference": "Shall we merge these `if` for `path lib` and then do the `attrname` one. I'm trying to count `if`-calls and its getting late but I think we can reduce the number checks needed to get into L648 from 3 to 2 if you understand what I mean"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/recovery` since it's related to recovery.",
+      "reference": "why depend on recovery???"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "the done channel only aims to be closed, so `chan struct{}` is better then `chan interface{}`"
+    },
+    {
+      "prediction": "This should be wrapped in `if err != nil { return nil, wrapErr(err, \"error creating contract binding\") }`",
+      "reference": "Do you mean to have gas_oracle changes in this PR?"
+    },
+    {
+      "prediction": "I think we can remove the partial here as well, since it's not used anymore.",
+      "reference": "Is this effectively requiring browsers support javascript?"
+    },
+    {
+      "prediction": "I think this should be `setStyle(R.style.SalesforceSDK_Dialog, (isDarkTheme ? R.style.SalesforceSDK_Dialog_Dark : R.style.SalesforceSDK_Dialog));`",
+      "reference": "Lowercase `boolean` - use the primitive type."
+    },
+    {
+      "prediction": "I think this should be `blobExists` instead of `blobExists == true`",
+      "reference": "This isn't the correct logic. if allowOverride == true and !blobExists then upload. Just don't do the exist check if allowOverride is set to true."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `wlr_net_wm_state_read()`.",
+      "reference": "Can you replace these by `else if` please?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `componentDidMount` method of the `Calendar` component, since it's only called when the component is mounted.",
+      "reference": "Would be nice if the field had a unique class name that could be used to target it, instead of `fromRange`   But doesn't seem to be a common thing in the code base."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it. I think it would be better to have a function that returns true if a header is refed or not, and then use that in the `reduce` function.",
+      "reference": "Does this code fit on one line under 80 characters? It would be more consistent with our existing style."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the same place as the 0.7.0 `RegisteredArtifactsFor070` function, since that is where the 0.7.0 `OpenEBSCRDArtifactsFor070` and `JivaVolumeArtifactsFor070` functions are defined.",
+      "reference": "Not a requirement here. But i see a change in naming convention. We can rename the function to `CstorSparsePoolArtifactsFor070`."
+    },
+    {
+      "prediction": "I'm not sure if this pin is used by Pico, but it's better to have it.",
+      "reference": "Why do you configure the UART like this? It's already configured by default."
+    },
+    {
+      "prediction": "Please remove this header file. It is not used anywhere in this file.",
+      "reference": "Wow. Good catch. \"This `#include` was brought to you by the department of redundancy department.\" ;)"
+    },
+    {
+      "prediction": "I think we can remove this logf import now.",
+      "reference": "Well it's actually calling `AuthorizeOrder` and this seems like a DebugLevel message to me, maybe it should be dropped entirely."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the device code, but I don't know how to make it work.",
+      "reference": "Should this be using the `CUDA_SAFE_CALL` macro here?"
+    },
+    {
+      "prediction": "I'm not sure why we're changing the color here. I think it should stay the same.",
+      "reference": "What cases do you see as being valid for not having any tests to run? If a test project doesn't have any tests to run we should avoid running it at the project level."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we need to do this in `processMirrorMsgs()` on the `mset.mirror` object.",
+      "reference": "and is it possible that mset.mirror.msgs be nil? because if so, then you would get a panic also because of that."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `surface_update` function, since it's the only place where we update the opaque region.",
+      "reference": "`next->opaque` should not be cleared."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `processClientAuthentication`.",
+      "reference": "already what? I know what, but seems like smth. is missing in this sentence"
+    },
+    {
+      "prediction": "I think we can remove the `details` class from the `textarea` as it's not used for anything related to the proposal.",
+      "reference": "Can you put spaces after the `{`and before the `}`"
+    },
+    {
+      "prediction": "I think this should be `__version__` and not `__version__str__`",
+      "reference": "We said we would put `[ ]` around the version. `[forseti-security] [v2.3.0]`"
+    },
+    {
+      "prediction": "I think this should be `if len(cnb) > 0 && len(cnb[0]) > 0 && n >= int64(len(cnb[0]))`",
+      "reference": "Seems len(cnb[0]) will be the length after the write since of the *v thing we saw no?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should go in `queue.go`",
+      "reference": "can we make use of storagepoolclaim type rather than the generic interface{} type?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I'm not sure we want to wait for all the peers to connect before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I'm not sure we want to wait for all the peers to connect before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I'm not sure we want to wait for all the peers to connect before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I think we want to wait until all the peers are connected before we start accepting connections. I think we",
+      "reference": "this should happen before the `go k.manage()`, otherwise the node will always try the bootnodes first"
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix. I think we need to keep the `%n` and `%b` parts, but remove the `%s` part. The `%s` part should be removed from the `procCode` string, not from the `var procComponents` array.",
+      "reference": "Is this actually a lint related change?"
+    },
+    {
+      "prediction": "Do we need this autoload? I don't see it being used anywhere.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `opt_nonempty_str_param` instead of `opt_nonempty_str_param_param_name`",
+      "reference": "what does this do that str_param does not?"
+    },
+    {
+      "prediction": "I think this should be `fmt.Sprintf(\"%s/%s-%s\", prefix, api, version)`",
+      "reference": "For GCP we included the API name (see below) so you can distinguish between uses other than `blob` once they exist, WDYT?"
+    },
+    {
+      "prediction": "I'm not sure this is necessary, but I'm also not sure we need to check for boto3 and botocore here.",
+      "reference": "what would you think about doing this include on-demand within S3ObjectStore methods and then failing in a loud way?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this check. I think it would be better to do this check in the `node` mutex lock above.",
+      "reference": "May be better make `stopped` atomic, then it can be just `for (auto i (node.store.latest_begin (transaction_a, next_frontier_account)), n (node.store.latest_end ()); i != n && && !stopped && elections_count < max_elections; ++i)`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it's the only way I could get it to work.",
+      "reference": "In `src/storage/InternalStorageServiceHandler.h ` file, the client in `client->future_chainAddEdges ` should be InterStorageServiceAsyncClient?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/apis/monitoring/v1`",
+      "reference": "`go.sum` needs to be updated for this. Please run `go mod tidy`"
+    },
+    {
+      "prediction": "nit: I think this should be `SecretBackendSSMParameterStore` and `SecretBackendSecretsManager` should be `SecretBackendAWSSecretsManager`",
+      "reference": "Not sure if this was discussed previously, but is there a specific use case in mind for having this as an API field rather than a configuration option on the controller manager? Is there a specific use case in mind where one would want to choose different backends for individual Clusters/Machines vs having it a global configuration for an instance of the infrastructure provider?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to add the `PeriodicExportingMetricReader` to the `builder`, and then add the `OtlpMetricsExporter` to the `builder.AddMetricReader(...)` call.",
+      "reference": "Same thing with the OtlpExporter"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should go in `jx_print_subexpr.c`",
+      "reference": "Should not be removed."
+    },
+    {
+      "prediction": "I think we can remove the `include DSL::InstallUtils::PuppetUtils` line as well.",
+      "reference": "Not a blocker for anything, but should this be moved to beaker-pe? Is that ticketed anywhere?"
+    },
+    {
+      "prediction": "Shouldn't this be `afterClickOn(element, driver)`?",
+      "reference": "this shouldn't be beforeClickOn, but rather beforeSubmit? and added to WebDriverEventListener. Since submit does not synthesize the 'click' events, this isn't accurate."
+    },
+    {
+      "prediction": "I think this should be `self.approving_official && self.approving_official.email_address || self.system_approver_emails.first` instead of `self.approving_official ? self.approving_official.email_address : self.system_approver_emails.first`",
+      "reference": "this ternary operator is hiding an `if/else` within an `if/else` - any chance we could move the logic ELSEwhere? (see what I did there -- ?? :100: )"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this annotation. I think it should be in `org.pmd.lang.java.typeresolution.type`",
+      "reference": "Is this really internal? Or will the API be different in PMD 7.0.0? Well, the interface doesn't offer much functionality anyway..."
+    },
+    {
+      "prediction": "I think this should be `ragged` instead of `raggeded`",
+      "reference": "Does it have to be arrays? Isn't it ragged 'data' (i.e multiple elements of different lengths)?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to set this to `false` by default.",
+      "reference": "Per my understanding, this flag only impacts \"NoCleanup\" disk. If so, we may name it more clear. The reason of this ask is because I plan to add some other flag to force cleanup for other non-NoCleanup disks."
+    },
+    {
+      "prediction": "I think this should be `StreamingWrite` instead of `SparkStreamingWrite`",
+      "reference": "In my changes for `RequiresDistributionAndOrdering`, this class went away and is replaced by an inner class. I think that pattern worked well. Maybe we could do that before this one to reduce the number of changes here."
+    },
+    {
+      "prediction": "These should probably be `ITEM_PARSE_HITPOINTSLEECHANCE`, `ITEM_PARSE_HITPOINTSLEECHAMOUNT`, `ITEM_PARSE_MANAPOINTSLEECHANCE`, and `ITEM_PARSE_MANAPOINTSLEECHAMOUNT`.",
+      "reference": "Shouldn't this be called `hp` instead of `life`?"
+    },
+    {
+      "prediction": "I think it would be better to add this as a constant in `DispatcherConstants.java` and then use `DispatcherConstants.ALLOWED_METHODS` and `DispatcherConstants.ALLOWED_ORIGINS` instead of `DispatcherConstants.CORS_ALLOWED_METHODS` and `DispatcherConstants.CORS_ALLOWED_ORIGINS`",
+      "reference": "Because the default value of allowedOrigins is * (all origins), so it isn't necessary to set again at all."
+    },
+    {
+      "prediction": "I think this should be reverted.",
+      "reference": "Undo this modification"
+    },
+    {
+      "prediction": ",\"\"boolKey\"\":true,\"\"library.name\"\":\"\"CreateTestActivity\"\"}}}}]\",\n                 content);\n         }\n \n```\nReview comment?",
+      "reference": "Please check my thoughts here... I changed the test to not `ToString()` attribute values. This was important to test things when `net.peer.port` was both an int or a string, but I was unsure if Zipkin supported non-string attributes."
+    },
+    {
+      "prediction": "I think this should be `returnBase64NoMetadata(call, exif, bitmapOutputStream);`",
+      "reference": "Looks like you made changes on `CameraResultType` class, but didn't commit them. And you also have to do the changes on the types in @capacitor/core"
+    },
+    {
+      "prediction": "I think we can remove the `BootDiskKmsKeyring` and `BootDiskKmsLocation` fields from the `ImportParams` struct as well.",
+      "reference": "Can you explain the history of the bug a bit, and why this fixes it?"
+    },
+    {
+      "prediction": "Layout/AlignHash: Align the elements of a hash literal if they span more than one line.",
+      "reference": "What do you think about `mentor_mail_to` or `mentor_mail_to_link` in order to match Rails' `mail_to` method, which is what this calls? I'm not sure about this suggestion..."
+    },
+    {
+      "prediction": "`metricPoint.Keys.Length` is `int` and `((metricPoint.Keys == null) ? 0 : metricPoint.Keys.Length)` is `null` when `metricPoint.Keys` is null.",
+      "reference": "Looks like this will make the check in every loop. Consider extracting the null check."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `SharedExampleGroup` module, since that's where the `clear` method is defined. The `SharedExampleGroup` module should contain all of the logic related to shared examples, including resetting the registry.",
+      "reference": "Maybe `it \"can safely be reset when there are not yet any shared example groups\"`? That's the edge case that wasn't working, right?"
+    },
+    {
+      "prediction": "I think this file should be renamed to `ssmfactory.go`",
+      "reference": "naming: please use `asmfactory`"
+    },
+    {
+      "prediction": "I think we can remove this now that we have the API v2.1 release.",
+      "reference": "FYI you could use a raw string for this which removes the need for escaping the inner quotes."
+    },
+    {
+      "prediction": "I think this test should be `@pytest.mark.parametrize('keystr, is_special_key', ...)`",
+      "reference": "2 blank lines here (between functions)"
+    },
+    {
+      "prediction": "nit: can we rename this to `BaseRewriteFilesWithPartitionSpec` to make it more clear what this class is?",
+      "reference": "Nit: we try to avoid whitespace changes because they can easily cause unnecessary commit conflicts."
+    },
+    {
+      "prediction": "I think this should be `substr($file_contents, $end_pos, 2) === '::'`",
+      "reference": "As mentioned below, I'm not sure why this is only allowing refs that are tagged with `*` at the start!"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it would be better to have a method in `CombinedAppsFilter` that checks if this is equal to another `CombinedAppsFilter` object.",
+      "reference": "Don't forget the {} symbols, and please use EqualsBuilder and HashcodeBuilder as possible"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `testutil/testutil.go`",
+      "reference": "Allocate buffer once outside loop."
+    },
+    {
+      "prediction": "This is a bit of a hack, but I don't see a better way to do this.",
+      "reference": "It would be better to check that system.nodes[1]->network.port is somewhere in the target without specifying its exact position. But it is a very minor point and I have no string opinion on it just thought I'd mention it because our tests in general have too implementation detail."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this exception message. I think it should be moved to `org.eclipse.jetty.websocket.WebSocketClientHandler` and `org.eclipse.jetty.websocket.WebSocketServerHandler` instead of `org.eclipse.jetty.websocket.WebSocketClientHandler` and `org.eclipse.jetty.websocket.WebSocketServerHandler`. This is because `org.eclipse.jetty.websocket.WebSocketClientHandler` and `org.eclipse.jetty.websocket.WebSocketServerHandler` are the only two classes that implement the `javax.ws.rs.core.Response` interface. The `org.eclipse.jetty.websocket.WebSocketClientHandler` and `org.eclipse.jetty.websocket.WebSocketServerHandler` classes are the only two classes that implement the `javax.ws.rs.core.Response` interface. The `org.eclipse.jetty.websocket.WebSocketClientHandler` and `org.eclipse.jetty.websocket.WebSocketServerHandler` classes are the only two classes that implement the `javax.ws.rs.core.Response` interface. The `org.eclipse.jetty.websocket.WebSocketClientHandler` and `org.eclipse.jetty.websocket.WebSocketServerHandler` classes are the only two classes that implement the `javax.ws.rs.core.Response` interface. The `org.eclipse.jetty.websocket.WebSocketClientHandler` and `org.eclipse.jetty",
+      "reference": "Should this be in a bundle?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `ProcessRequestAsync` since it's a synchronous call.",
+      "reference": "When Activity is lost (more precisely, ExecutionContext is lost) in the HttpModule we restore the root (HttpIn) Activity. That makes this assert invalid. I tried to fix the HttpModule so that it restores the Activity that was last running, but it is impossible to retrieve do to the way ExecutionContext works. It isn't an issue to remove the assert, but it is unnerving. Any instrumentation running in IIS reliant on Activity.Current could run into trouble."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `RegisterHandlers` function.",
+      "reference": "what was the point in moving handlers registration into a separate method?"
+    },
+    {
+      "prediction": "I think this should be `if state.Config.Cache.RetrieveCommand != \"\" && len(mplex.caches) == 0 {`",
+      "reference": "At this point we probably want to ensure there's a store command set."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to make sure that the indexes are unique, and that we don't have two indexes for the same thing.",
+      "reference": "This code is way too complicated for what it does. I don't want to be a PITA, but what can you consider rewriting it in a simpler way? I'm talking about the whole `_getDerivationIndexes` function"
+    },
+    {
+      "prediction": "Shouldn't this be `string result = await _vaultService.Sign(Guid.Parse(vaultId), Guid.Parse(keyId), message.Replace('0x', string.Empty));` ?",
+      "reference": "Don't throw base exception, specify more precise type"
+    },
+    {
+      "prediction": "I don't think this is the right place for this function. This function is only used in `tests/unit_tests/functional_tests.py`. I think it should be moved to `tests/unit_tests/functional_tests.py`.",
+      "reference": "Recursive -> Recursively converted -> convert"
+    },
+    {
+      "prediction": "I think this should be `num_fields > 0 && num_fields < 127`",
+      "reference": "why is this object API only? I think the reason why we didn't do this before was because this code was shared with Java which doesn't have value structs."
+    },
+    {
+      "prediction": "I think we can remove the `DockerVersion: ` portion of the query.",
+      "reference": "just to confirm, so backend will deal with both old format and new format?"
+    },
+    {
+      "prediction": "I think this should be `where T : struct, Enum`",
+      "reference": "huh, look at that, I wasn't aware you could constrain a type by Enum"
+    },
+    {
+      "prediction": "I don't think this is the right place for this function. It should be in the `adios2_mpi` module, not the `adios2` module. The reason is that the `adios2_mpi` module is for MPI applications, and the `adios2` module is for non-mpi applications. The `adios2_mpi` module should have the functions that are common to both MPI and non-mpi applications, and the `adios2` module should have the functions that are specific to MPI applications.",
+      "reference": "Even though the debug parameter is effectively ignored, should this default to `false` instead of `true`?"
+    },
+    {
+      "prediction": "I don't think we need this interface. We can just use `View` directly.",
+      "reference": "What is the purpose of this interface? Unlike ParamDocView, it doesn't seem to be used anywhere to enable polymorphism"
+    },
+    {
+      "prediction": "I think this is the only place where we're using caddy. I think we should remove this file and all of its code from this PR.",
+      "reference": "We added a library to generate a UUID for the request_id directive \"github.com/nu7hatch/gouuid\" can we uses the same one for what you want rather than add another dependency? Or if the google one is better can we remove the nu7hatch library?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to just do `p.pingConfig.Timeout / p.pingConfig.Interval` and avoid the `time.Duration(i)*p.pingConfig.Interval` part.",
+      "reference": "Shouldn't this be `i * Interval`? You're multiplying nanos with a Duration, I'm not sure what's the result :)"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. The problem is that `make_unwritable()` is called for every allocation, which means that if the allocation fails, we'll end up allocating again and again. I think we need to find a way to do this in a way that avoids allocating and unprotecting the same page over and over again.",
+      "reference": "I guess it's not no-access to match Windows where guard pages are readable."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this logic. I think it should be in the `before_save` callback of the `Proposal` model, since it's related to the `Proposal` object itself and not to the `NcrJob`.",
+      "reference": "Not a blocker, but would probably be useful to have a `scope :non_pending_approvers` on the `Proposal` model."
+    },
+    {
+      "prediction": "I think this should be `TaskState`, not `worker_state`. It's the state of the worker, not the state of the task itself.",
+      "reference": "Add trailing comma"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in `init.go` and then `GracefulShutdown` should be called when `init` is done.",
+      "reference": "I see someone say SIGKILL can not be caught by process. The original code here about `SIGKILL` is useless?"
+    },
+    {
+      "prediction": "I think this should be `output[Setting.THEME_OLED_DARK] = _('OLED dark');` and not `output[Setting.THEME_OLED_DARK] = _('OLED dark');`",
+      "reference": "As it is a mobile only theme, please make sure the option appears only on mobile"
+    },
+    {
+      "prediction": "`strip()` is not needed here.",
+      "reference": "Would like a docstring and type annotations here"
+    },
+    {
+      "prediction": "I think this should be `self.shard_key_fields = {}`",
+      "reference": "This is an API change. Why was it made?"
+    },
+    {
+      "prediction": "I think this should be `using Microsoft.VisualStudio.TestPlatform.Core.Utilities.Tracing;`",
+      "reference": "Please move using inside namespace."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could get it to work.",
+      "reference": "shall we create some enum flags for setasmain shouldprocess and pos?"
+    },
+    {
+      "prediction": "I think this should be `msi_opts['PUPPET_AGENT_STARTUP_MODE'] == \"Automatic\" ? \"Auto\" : msi_opts['PUPPET_AGENT_STARTUP_MODE']` instead of `msi_opts['PUPPET_AGENT_STARTUP_MODE'] == \"Automatic\" ? \"Auto\" : msi_opts['PUPPET_AGENT_STARTUP_MODE']`",
+      "reference": "Should this `PUPPET_AGENT_STARTUP_MODE` have a corresponding yardoc change?"
+    },
+    {
+      "prediction": "I think we can remove this file now, it's not used anymore.",
+      "reference": "Why not simply remove the `pytest` import now that it's not needed anymore? :wink:"
+    },
+    {
+      "prediction": "This file should be deleted.",
+      "reference": "Yeah, not filled in yet."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to add a `:checked` attribute to the `input` tag and then use `f.check_box :active, checked: true` in the view.",
+      "reference": "don't use style. Use a class instead. reducing `col-xs-12` down to 8 or 6 or whatever should do the trick"
+    },
+    {
+      "prediction": "nit: can we use `Objects.requireNonNull(table, \"table is null\");` instead of `Objects.requireNonNull(database, \"database is null\");`?",
+      "reference": "should we also do `reloadCache` when database is null?"
+    },
+    {
+      "prediction": "I think this function should be called `getQueryMol` instead of `getQueryMol2`.",
+      "reference": "Why not just return QueryMol directly? Why require the copy?"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(os.path.abspath(__file__), '../../', fname)`",
+      "reference": "Could you add a docstring?"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "looks like you added newlines after blocks in a few files - I generally like newlines before/after multi-line blocks _except_ when the end the block is directly nested inside another block (eg: two `end`s next to each other) what do you think?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. It hides the fact that this visitor is not thread-safe.",
+      "reference": "@jlerbsc 's fix in #2918 in action - thanks! :smiling_face_with_three_hearts:"
+    },
+    {
+      "prediction": "I think this should be `self.zero_count = zero_count or not count`",
+      "reference": "nitpick: Please lower-case `Zero_count` (as it's a literal argument name) and `Argument` here."
+    },
+    {
+      "prediction": "I think this should be `a.Name != issuer`",
+      "reference": "change the name of the function to match its functionality? account IsIssuing?"
+    },
+    {
+      "prediction": "I think this should be `from dask.dataframe import GroupBy`",
+      "reference": "for later: add docstring (can just copy paste Pandas')"
+    },
+    {
+      "prediction": "I think we should rename this file to `mocks.go`",
+      "reference": "nit: @sonofachamp pointed out to me that the idiomatic way is \"archermocks\" (lowercase for package names)"
+    },
+    {
+      "prediction": "I think we can remove this error now, since it's no longer used.",
+      "reference": "Ugh I'm stupid, can you change this to `yarpc.InternalErrorf`?"
+    },
+    {
+      "prediction": "I think this should be `private final String name;` instead of `private final String name = table.name() + \".all_data_files\";`",
+      "reference": "We instantiate some metadata tables in tests so I kept the old constructor too. Won't harm if someone is using it directly too."
+    },
+    {
+      "prediction": "I think this should be `throw new Error('Can not use `limit` and `perDocumentLimit` at the same time. Model `' + mod.model.modelName + '`; limit: ' + queryOptions.limit + ' perDocumentLimit: ' + queryOptions.perDocumentLimit)`",
+      "reference": "I think the better place to put this might be `lib/options/PopulateOptions.js`. That should make it easier - checking options in `populate()` can get confusing."
+    },
+    {
+      "prediction": "nit: `params` -> `Config`?",
+      "reference": "Lets have simple function arguments here, because now struct `traversal.Params` started to have 2 purposes: 1. used as contract in DTO between consumer-provider 2. as function parameters for internal code calls"
+    },
+    {
+      "prediction": "This file should be renamed to `RemoteDistributorTest.java`",
+      "reference": "Probably best not to have an empty test...."
+    },
+    {
+      "prediction": "I think it would be better to have a link to the website where you can learn more about upcase.",
+      "reference": "`image_tag` in the `testimonials` files?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. This is only used for debugging, right? I think we should move it to the `FlowController` or `FlowManager` class.",
+      "reference": "I believe flow.toObject() method is throwing NPE if SLA option list (i.e. this.executionOptions.getSlaOptions()) is null or one of the value in the list (i.e. this.executionOptions.getSlaOptions()) is null. If that is the case we could fix root cause of NPE in the ExecutableFlow.toObject() method. The corresponding code to populate SLA options can be modified as follows. final List<Map<String, Object>> slaOptions = Optional.ofNullable(this.executionOptions.getSlaOptions()).orElse(Collections.emptyList()).stream() .filter(Objects::nonNull) .map(slaOption -> slaOption.toObject()) .collect(Collectors.toList());"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but it's the only thing I can come up with that doesn't break anything.",
+      "reference": "So I haven't tested this - but since the icon location is just directly dumped into the img _src_ attribute, won't this cause issues with Ghost blogs in a subdirectory? If I'm misunderstanding what the purpose of the default is then let me know"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think it would be better to use `::testing::AssertionResult` instead of `int result = RUN_ALL_TESTS();` and then use `::testing::AssertionStatus` instead of `int result = RUN_ALL_TESTS();`",
+      "reference": "Why swallow the exception here rather than propagate it?"
+    },
+    {
+      "prediction": "I think this should be `Optional[List[Tuple[str, str]]]` and not `Optional[List[str]]`",
+      "reference": "How about `column_index_names`? `column_names` sounds ambiguous."
+    },
+    {
+      "prediction": "I think we should rename this to `WRITE_METRICS_MODE_DEFAULT` to be consistent with the others.",
+      "reference": "+1 on this. Do we want to have it as `WRITE_METRICS_MODE_COLUMN_CONF_PREFIX` to be consistent with defaults? Is there a possibility we will have `READ_METRICS_MODE_COLUMN_CONF_PREFIX`? Not sure."
+    },
+    {
+      "prediction": "I think this test should be moved to `test_cumprod.py`",
+      "reference": "Shall we use different inputs for `PySpark < 2.4` where `transpose` won't work with different data types."
+    },
+    {
+      "prediction": "I don't think we need to add this here, since it's only used in `pyspark.sql`, which is only used in `dask`",
+      "reference": "we should sort the headers like in spark: public packages, then pyspark, then internal"
+    },
+    {
+      "prediction": "Why do we need to check for clientTLSConfig and not just clientTLS?",
+      "reference": "looks like we should drop this var from the transport options struct"
+    },
+    {
+      "prediction": "I think this is the only test that needs to be changed. The rest of the tests pass.",
+      "reference": "Ah, yeah, this is the test for the validation that I mentioned in a remark above about relaxing the validation."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this function. I think it should be in `transformations` or `transforms.py`",
+      "reference": "For tensors, it is better to illustrate the shape."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move this test to `test/unit/admin/components/site-kit/index.js` since it's only used in that file.",
+      "reference": "This was removed in a recent PR for the admin bar, but should have been kept. It's been restored in the other admin bar PR but I've added it here to for completeness."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it would be better to move it to `InstrumentationTests` or `InstrumentationCoreTests`.",
+      "reference": "nit: probably swap the if condition and `if/else` and check `NETCOREAPP2_1` which looks more natural."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `account.go` since it's only used there.",
+      "reference": "for the tag, should it be `requestor_rtt` since the other is `responder_rtt`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `MutateInventory` instead of in `Monster` itself.",
+      "reference": "The one sad part of this is that is will generate a lot of overhead if you loop through a vector, since it obtains the vector every time. But with the current API there is no alternative I guess, and it is better to have the option than not."
+    },
+    {
+      "prediction": "I think this should be `github.com/ipfs/go-ipfs-files` instead of `github.com/filecoin-project/go-filecoin`",
+      "reference": "nit: there should be a newline between `types` and `go-ipfs-files`."
+    },
+    {
+      "prediction": "I think we can remove this file and just use the `Duration` from the `retryAfter` struct.",
+      "reference": "would be nice to have this as a parameter (Options), re: technical debt discussions"
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "complete ignorance here  but is it ok to import this from `com.sun`? can it break under non-oracle JREs?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move it to `span_processor.go` since it's only used there.",
+      "reference": "Is this some way of saying that the span is going to be sampled? I'm basing this question on the condition I saw for the `OnEnd` to be called - `mustExport := s.spanContext.IsSampled() &&  )`. Shouldn't the condition be `s.spanContext.IsSampled()` to match the condition for calling the `OnEnd`?"
+    },
+    {
+      "prediction": "I think we can remove the `volumereplica.GetAndUpdateSnapshotInfo(c.clientset, cvr)` call here.",
+      "reference": "just calling UpdateSnapshotInfo() should be fine"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in `GlobalConfiguration` since it's a global setting that affects all traces.",
+      "reference": "Was `DefaultLogMessageRateLimit` not configurable anywhere?"
+    },
+    {
+      "prediction": "I think this test should be moved to `docker_test.go` since it is not related to the pull image functionality.",
+      "reference": "Please use the `assert` instead. Same below."
+    },
+    {
+      "prediction": "Please add a space after the comma.",
+      "reference": "Owww, I think I found a typo. `suche`"
+    },
+    {
+      "prediction": "I don't understand why this is removed.",
+      "reference": "This can happen because we could get `PhantomNodes` that are not admissible as source/target right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be.",
+      "reference": "Should we add a note here to file a bug if this ever happens?"
+    },
+    {
+      "prediction": "I think this should be `return 0, errors.Wrap(err, \"benchmarks quantity overflows int64\")`",
+      "reference": "overflow is still possible, isn't it?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "This seems more like a personal preference, so maybe this should be in your `~/.rspec`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `scapy.error` since it's a superclass of all exceptions raised by `scapy`.",
+      "reference": "To avoid a circular import between pcapdnet and automaton"
+    },
+    {
+      "prediction": "I think this should be indented 4 spaces instead of 2.",
+      "reference": "Please remove this blank line."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `Reconcile` method of the `CloudAuditLogger`.",
+      "reference": "it feels that the WorkloadIdentityCondition should be added to this Status... And if you have workload identity enabled, but not working, then you can mark that condition failed, and also Mark the ready false... not entirely sure though"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `if (pPipeline->graphicsPipelineCI.pRasterizationState->pNext)` check above.",
+      "reference": "I think these got refactored to `LvlFindInChain` to better conform with the coding guidelines. I think MarkL left these in for backwards compatibility."
+    },
+    {
+      "prediction": "I don't think we need this extra line.",
+      "reference": "It's fine for this PR, but in the future try to minimize unrelated changes."
+    },
+    {
+      "prediction": "I think this should be `:$HOME/.ansible/roles:$HOME/.ansible/plugins/roles:$project_directory/roles`",
+      "reference": "The library -> modules and filters -> filter path changes are breaking changes or?"
+    },
+    {
+      "prediction": "I don't think we should be importing `boto3` and `botocore.config` at all. We should just use `google-cloud-storage` and `google-auth`.",
+      "reference": "Can we encapsulate this check (`os.environ.get(\"TEST_TARGET\") == \"AWS_CLOUD\"`) into a small config/util function? (and also use it in `integration/conftest.py`) (I'd generally try to avoid accessing `os.environ` directly across the codebase. I'm aware that we're already doing this in a few places, but maybe we can start pulling things out - this will also help us create more systematic config/environment abstractions as we go...)"
+    },
+    {
+      "prediction": "`[].concat(child).reverse().forEach((item) => {` should be `[].concat(child).reverse().forEach(item => {`",
+      "reference": "An `if` statement would be better here. Functions, concat and reverse are all very expensive."
+    },
+    {
+      "prediction": "I think this should be `goerli` instead of `go`",
+      "reference": "`Log here: connect to default factory for goerli network <address>`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `DeploymentConfig` struct.",
+      "reference": "nits, should add `break` to avoid redundant iterations."
+    },
+    {
+      "prediction": "I think this test should be moved to the `EIP_158` folder.",
+      "reference": "I think we want an exception here, there should never be a transfer to a nonexisting account."
+    },
+    {
+      "prediction": "This should be `LOG.debug`",
+      "reference": "This code is new and not tested well. While we may drop the log level before we ship 4.0, right now this is extremely helpful to users."
+    },
+    {
+      "prediction": "why not 10?",
+      "reference": "Can we make these asserts more strict/precise? I remember this test helped to spot the issue of inconsistent results on different platforms due to using `sort` instead of `stable_sort`."
+    },
+    {
+      "prediction": "I think this should be `select_interpreter(executable, interpreters || {})` instead of `select_interpreter(executable, interpreters)`",
+      "reference": "I thought about adjusting for case here as well? For example `.rb,rb,.RB,.rb` would all map to ruby interpreter specified? Is that too much \"magic\"? Is there a sane workflow that someone would choose to map `.rb` to one interpreter and `.RB` to another?"
+    },
+    {
+      "prediction": "`ostree_sysroot_get_booted_deployment` is called from `ostree_sysroot_get_subbootversion` which is called from `ostree_sysroot_load` which is called from `ostree_main` which is called from `ostree_daemon` which is called from `ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree` which is called from `lib/ostree/ostree",
+      "reference": "In this and the one below, should we assert `self != NULL` first?"
+    },
+    {
+      "prediction": "Why did you change the status code to 200?",
+      "reference": "Why this is needed?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. It should be moved to `Package`.",
+      "reference": "we already have .meta() so I don't think we want get_meta() for PackageEntry"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to get the start timestamp. I think it should be in the `coprocessor-config.properties` file.",
+      "reference": "question: can we always get the startts from dagrequest?"
+    },
+    {
+      "prediction": "I think it would be better to use `accountID` instead of `propertyID` here.",
+      "reference": "Nit-picking, but this can be simplified to reference `properties[ 0 ]` since that is already retrieved 2 lines above."
+    },
+    {
+      "prediction": "I think this should be `delete` instead of `delete!`",
+      "reference": "This technically works because we call `get_targets` before we remove the target, so it's bound to have been resolved. But I'm not sure I would like to rely on that assumption. Should we also remove it from `@unresolved_targets` to be safe?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the contract is null, then the code should be empty, not new Code().",
+      "reference": "So what happens if I try to get code on a EOA?"
+    },
+    {
+      "prediction": "I think this should be `values = [i for i in inferred.values() if isinstance(i, astroid.Assign)]`",
+      "reference": "Can't we return an empty list and then just compare the lengths? Thereby removing this `if`."
+    },
+    {
+      "prediction": "I'm not sure if this is a good or bad thing. I think it's a good thing, but I'm also not sure if it's a good thing to throttle at 500ms. I think it's a good thing to throttle, but I'm also not sure if it's a good thing to throttle at 500ms. I think it's a good thing to throttle, but I'm also not sure if it's a good thing to throttle at 500ms. I think it's a good thing to throttle, but I'm also not sure if it's a good thing to throttle at 500ms. I think it's a good thing to throttle, but I'm also not sure if it's a good thing to throttle at 500ms. I think it's a good thing to throttle, but I'm also not sure if it's a good thing to throttle at 500ms. I think it's a good thing to throttle, but I'm also not sure if it's a good thing to throttle at 500ms. I think it's a good thing to throttle, but I'm also not sure if it's a good thing to throttle at 50",
+      "reference": "Mb change it to 'doesPassRestrictions()'? Doesn't sound enough like a boolean-returning method to me."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should go in `pkg/feeds` since it's a part of the feeds package.",
+      "reference": "TODO: in fact At when used with noncurrent time should not return a next index. only latest should.."
+    },
+    {
+      "prediction": "I think this should be `isTopLevelFragment ? Array.isArray(tmp) ? tmp : [tmp] : tmp`",
+      "reference": "There are scenario's where we skip an update with a placeholder, in this case tmp returns null with render, this is not an array so we `[null]` it"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to move this, but it seems like it should be in `java.util.stream.Stream`?",
+      "reference": "@danieldietrich, this wasn't tested so I presumed it wasn't working before either :p"
+    },
+    {
+      "prediction": "This should be `use Typo3\\Core\\Utility\\GeneralUtility;` instead of `use ApacheSolrForTypo3\\Solr\\Facet\\Facet;`",
+      "reference": "Please leave two empty lines after the `use` statement"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Please move this import in the 3rd group which is for external dependencies"
+    },
+    {
+      "prediction": "I think this should be `not node or isinstance(node.get(RESERVED['file']), (str, bytes))`",
+      "reference": "So what exactly is this fixing? The unit test is passing with or without this change."
+    },
+    {
+      "prediction": "I think this should be `fs::path` instead of `folly::stringPrintf`",
+      "reference": "auto targetFile = fs::FileUtils::joinPath(newPath, folly::stringPrintf(\"%019ld.wal\", it->first));"
+    },
+    {
+      "prediction": "I think we can remove this method and use `snapshotIdsBetween` instead.",
+      "reference": "I don't see any place that checks whether the from snapshot is an ancestor of the to snapshot. That seems like a requirement for this to work correctly to me."
+    },
+    {
+      "prediction": "I think it would be better to use `isset($functionConfig['helpText'])` instead of `isset($functionConfig['helpText'] ?? '')`",
+      "reference": "Other calls to getHelpText() are wrapped in an `isset()` check. Would it make sense to be consistent, and either restore that here or remove it elsewhere?"
+    },
+    {
+      "prediction": "I think this should be `@configuration.output_stream = out || stdout`",
+      "reference": "Why the change? If `output_stream` is set to something non-nil, it seems odd (and potentially wrong) to overwrite it...."
+    },
+    {
+      "prediction": "Align the parameters of a method call if they span more than one line.",
+      "reference": "Thinking this should have an underscore."
+    },
+    {
+      "prediction": "I think this should be `assert MSG_STATE_SCOPE_MODULE == linter._get_message_state_scope(\"W0102\", 3)` instead of `assert MSG_STATE_SCOPE_MODULE == linter._get_message_state_scope(\"W0101\", 3)`",
+      "reference": "To accommodate the change to the method being private."
+    },
+    {
+      "prediction": "I think this should be `CollectionRule` and not `PropertiesNamesCamelCase`",
+      "reference": "Shouldn't we traverse all definitions and apply this rule over all model definitions in the doc"
+    },
+    {
+      "prediction": "I think it would be better to have this as a separate file, so that it's clear that it's not part of this file.",
+      "reference": "Please rename this class that it ends with the suffix \"Test\" - otherwise the tests won't be executed."
+    },
+    {
+      "prediction": "I'm not sure this is the best approach. I think we should be able to tell if a file has a Prettier config by looking at the `prettierignore` file, which is located in the root of the project. I don't think we should be able to tell if a file has a Prettier config by looking at the `prettier-config.js` file, which is located in the root of the project.",
+      "reference": "As a JS file, this should also receive our standard file header."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the same place as `numBMOsInFrag` and `numStreams`.",
+      "reference": "I take it memQuotaRatio is an \"out\" parameter at line 1810? Also, why use MIN_QUOTA in both the cases of memQuota being too low and too high? Contrast this code with RelRoot::codeGen where we set MIN_QUOTA only in the too low case."
+    },
+    {
+      "prediction": "I think we can remove this line now, since we don't need it anymore.",
+      "reference": "It looks like for newpullsubscription, we have two functions. One for pullsubscription with default, one for pullsubscription without default. If you remove setdefault here, then there is no differences between these two functions."
+    },
+    {
+      "prediction": "I think this should be `uint32` instead of `uint64`",
+      "reference": "not dramatic, but why 32bits if we allow a range that is smaller than 8?"
+    },
+    {
+      "prediction": "I don't think we should add a new field to the existing DTOs.",
+      "reference": "- Use type from `service_discovery/dto/price.go` - Rename `Price` -> `Money`"
+    },
+    {
+      "prediction": "I think it would be better to use `System.out` instead of `System.out.println`.",
+      "reference": "one space missing at the beginning"
+    },
+    {
+      "prediction": "Should this be `yaml:\"platform\"` or `yaml:\"platform,omitempty\"`?",
+      "reference": "I worry that having `Platform` as `string` instead of `*string` would result in it getting overriden with `\"\"` if the environment manifest config doesn't specify `platform`   . If it gets overridden as `\"\"`, that could be a problem for users that are not using `linux/amd64` right? If this indeed is a potential impact, could we double check in `ApplyEnv`'s unit test? Same with the `Platform` in `TaskConfig`. Sorry for not seeing this in my previous review!!!"
+    },
+    {
+      "prediction": "I think this should be `'<%= connect.test.options.port %>/test/integration/api/rules/'`",
+      "reference": "I like the new directory `/api`, what is the thinking behind `/external`?"
+    },
+    {
+      "prediction": "Why do we need to sort here?",
+      "reference": "staticcheck flags this: replace with `for k := ..`"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be using `GMLAN_PM` to enable programming mode, not `GMLAN` itself.",
+      "reference": "There's already a sleep right above"
+    },
+    {
+      "prediction": "I think this should be `$var_id = $thisClass->name . '::' . substr($assertion->var_id, 6);`",
+      "reference": "This feels hacky, is there a better way?"
+    },
+    {
+      "prediction": "I don't think this is necessary. `collections.OrderedDict` should be available by default.",
+      "reference": "Can we move this logic into the `boto.compat` module? Then it's just `from boto.compat import OrderedDict` instead. Also, this is introducing a new dependency. What about users on 2.6.x that don't have the OrderedDict module installed? We may need to fall back to an ordinary `dict` so that existing code in the wild does not break."
+    },
+    {
+      "prediction": "I think this should be `SerializationSettings` and `DeserializationSettings` should be `JsonSerializerSettings`",
+      "reference": "\"Gets the JSON serialization settings.\" would be better?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `server/file`.",
+      "reference": "The endpoint should be in plural `/files`."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "??? Why do we need to have a different API in Python? Oh, wait, I see what you did. It's to fake the namespaces. Given that I don't think the namespaces are necessary, and that less of these should be exposed anyway, I think these should go."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/context.go`",
+      "reference": "I think this line is gofmt issue? @daixiang0"
+    },
+    {
+      "prediction": "I think this test should be renamed to `TestParseSubrepoLabelParts`.",
+      "reference": "probably better to assert the values directly; technically you could pass this test with an implementation that always returned \"\" for the subrepo for example."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `addTestingTsfBlocks` function.",
+      "reference": "line is 161 characters (from `lll`)"
+    },
+    {
+      "prediction": "I think this should be a `@Immutable` interface.",
+      "reference": "Are these changes needed?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think this should be done in `org.apache.gobblin.util.FileSystemUtils` instead of `UserUtils`",
+      "reference": "It seems like reducing the sensitivity would make the test failure less likely to occur, but not fix the issue?"
+    },
+    {
+      "prediction": "This should be `use Thelia\\Model\\Tools\\I18nTimestampableTrait as I18nTimestampableTrait`.",
+      "reference": "missing use statement for importing the full namespace"
+    },
+    {
+      "prediction": "I think this should be `ErrIncorrectAddressFormatForService`",
+      "reference": "Generally error messages in Go shouldn't start with capital letter. I see this rule is broken in many places here. Why?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be in `testSelection.js` under the `selection` section.",
+      "reference": "These test checks if selection coordinates don't return negative values in that edge cases. Please revert the changes and set `autoWrapCol` and `autoWrapRow` to `false` to the Handsontable instance. This change applies to the entire Core_selection.spec.js file."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could come up with right now.",
+      "reference": "I would prefer flag SkipGasPricingValidation"
+    },
+    {
+      "prediction": "This should be `auto exists = blocks.find (root_a);`",
+      "reference": "could be `const &` ?"
+    },
+    {
+      "prediction": "I think we can remove `hideProgressAfterFinish` as it is not used anywhere.",
+      "reference": "The docs should also be updated with the new option."
+    },
+    {
+      "prediction": "I don't think we should be importing boto. We should be using `sqs` or `smtplib` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or `jose` or `josepy` or `sentry` or `sentry-python` or",
+      "reference": "Name is _way_ to generic for this patch to be accepted I think. Can you rename it to maybe `send_sns_notification` or something else you find suitable?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to pass in the config directly, but I'm not sure how to do that.",
+      "reference": "Out of the scope."
+    },
+    {
+      "prediction": "`callback` can be `undefined` in this case.",
+      "reference": "I think we need an if block to check that if this API is really missing. We wouldn't want to override the browser's implementation."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think we should check if `editorClass` is an instance of `Editor` and then use `editorClass.EDITOR_TYPE` if it is, or `Editor.EDITOR_TYPE` if it isn't.",
+      "reference": "I was wondering, things like EDITOR_TYPE, VALIDATOR_TYPE, RENDERER_TYPE are not obligatory. Therefore if the custom editor will be registered with this method, `name` will be set to `undefined`. Should we handle this scenario?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "This line should not be removed, as it is not the description of the package, but a copyright header."
+    },
+    {
+      "prediction": "I think this should be called `LastPrefetchTag` or `LastPrefetchTime`.",
+      "reference": "Wouldn't it be more space efficient to make this a `byte` enum or bitfield?"
+    },
+    {
+      "prediction": "I think this should be `\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"",
+      "reference": "Flake8 doesn't like the assignment, but non-use of `e` here. I think we can just remove the try/except here for `cursor.execute(sql)`."
+    },
+    {
+      "prediction": "I think this interface should be removed. It is not used anywhere.",
+      "reference": "Maybe add short example list: .... type body declarations, such as AnnotationMembers, Methods, Fields"
+    },
+    {
+      "prediction": "I don't think we need to add any new capabilities here. We already have them in the existing capabilities.",
+      "reference": "These are really meant to be the keys in the capabilities, not the keys of values within the capabilities"
+    },
+    {
+      "prediction": "I think this should be `@return the full name for this table`",
+      "reference": "It is debatable whether we have to default this. I did this to avoid breaking custom implementations."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "# `attr_value['string'] is not None` probably not required as `attr_value['string'] == value` check is already being done"
+    },
+    {
+      "prediction": "I think this file should be renamed to reflect the fact that it is an addon, not a package.",
+      "reference": "we shouldn't need the \"addon\" rename"
+    },
+    {
+      "prediction": "I think this should be `before_filter :must_be_admin_user`",
+      "reference": "These are covered by routing constraints."
+    },
+    {
+      "prediction": "I think this should be `aws.String([]string{*instanceID})`",
+      "reference": "Could we use just a normal string here?"
+    },
+    {
+      "prediction": "nit: can we move this to the top of the file?",
+      "reference": "`isFunction(Symbol)` triggers a `ReferenceError` if Symbol is undefined. Using `typeof Symbol === 'function'` is OK however."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that this will cause the election thread to run in a race condition. The election thread may transition to the active state before the lock is unlocked. In that case, the election thread will try to acquire the lock, which will fail because the lock is held by some other thread. The election thread will then try to transition back to the active state, which will fail because the lock is still held by some other thread. The election thread will then try to transition back to the active state, which will fail because the lock is still held by some other thread. The election thread will then try to transition back to the active state, which will fail because the lock is still held by some other thread. The election thread will then try to transition back to the active state, which will fail because the lock is still held by some other thread. The election thread will then try to transition back to the active state, which will fail because the lock is still held by some other thread. The election thread will then try to transition back to the active state, which will fail because the lock is still held by some other thread. The election thread will then try to transition back to the active state, which will",
+      "reference": "Is there an else clause (with a lock.unlock ()) missing here as well? If we think we do not need an else clause then lets add an else clause containing assert zero."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to make sure we have enough balance to pay for the transaction.",
+      "reference": "do we foresee `32` ever changing? (as well as its use in error checking in `clerk.go`) if \"no\" this is fine, if \"yes\" it's probably still fine"
+    },
+    {
+      "prediction": "nit: remove trailing whitespaces",
+      "reference": "you don't need to create a enum for this, java allows you to switch on a string ;) (since java 7?)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could get it to work.",
+      "reference": "This should be `modelPromises.roles = this.get('store').query('role', ...` and the promise body should just `return roles;`. This is because the `model()` hook will pause for the returned promises to be resolved before continuing so when returning `RSVP.hash` you end up with a hash containing the values returned from the promises so the controller and everything else relying on the `model` object doesn't need to worry about things being async."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "This is basically a repeat of `main-is-top-level.js`. I suggest having these 3 checks point to the same evaluate file. You can change what element they test through the rule selector as well as the check.json options property. So you would have `main-is-top-level.json`, `banner-is-top-level.json` and `contentinfo-is-top-level.json` each point to `landmark-is-top-level.js`, and pass in a different `options` object which tells it what element you're trying to test."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to do this in a more efficient way, but I don't know how to do that.",
+      "reference": "Why do we need this check? setCore already handles nulls."
+    },
+    {
+      "prediction": "I think this should be `__dir__` instead of `__getitem__`.",
+      "reference": "good catch, thanks. This is going to make the user experience much better."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a great solution either.",
+      "reference": "TODO: remove this unused import"
+    },
+    {
+      "prediction": "I think this should be \"compact JWS format must have two parts\"",
+      "reference": "Hmm leaking the library name out in the Workload API doesn't seem great. Do we want to catch these errors and return a canned \"could not parse token\" or something instead?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get Travis to pass locally.",
+      "reference": "I think `-f gradlew` is better for code unity"
+    },
+    {
+      "prediction": "I think it should be 1024 here.",
+      "reference": "Is there precedent for this config key? What other keys are similar? The others in this file start with `table.exec.iceberg`. Is there a reason for not continuing with that convention?"
+    },
+    {
+      "prediction": "This file should be reverted.",
+      "reference": "Can we avoid no-change changes being committed - they make it harder to distinguish actual changes from code editor changes"
+    },
+    {
+      "prediction": "I'm not sure if we want to remove the `UpdateServerCertificate.java` sample, as it's still useful to demonstrate how to update the name of an AWS Identity and Access Management (AWS IAM) server certificate. I think we should keep the `UpdateServerCertificate.java` sample, as it's still useful to demonstrate how to update the name of an AWS Identity and Access Management (AWS IAM) server certificate.",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `tequilapi` command.",
+      "reference": "You dont use this dependency anymore"
+    },
+    {
+      "prediction": "I think this should be `assertNotContains` instead of `assertNotContains`.",
+      "reference": "Here is the test that checks that the sqlcipher in use was compiled with the right flags"
+    },
+    {
+      "prediction": "I think this should be `if (mdmLoginServers == null || mdmLoginServers.isEmpty()) {`",
+      "reference": "Turns out that if the hosts are not in an `array`, an `Exception` is not thrown anymore :-( It simply returns `null`. So, we need to attempt to parse a `string` outside the `catch` block for it to work."
+    },
+    {
+      "prediction": "I don't think you need to import `URI` here.",
+      "reference": "this test only covers the happy path. what about no acceptable rest template found?"
+    },
+    {
+      "prediction": "nit: `LOG.info(String.format(\"Firefox Session %d assigned to %s\", i, chromeSession.getUri()));` `LOG.finer(String.format(\"Firefox Session %d assigned to %s\", i, chromeSession.getUri()));`",
+      "reference": "Since this is in a test, I imagine that the choice of `info` level was deliberate."
+    },
+    {
+      "prediction": "This file should be renamed to `langserver_test.go` and this file should be renamed to `langserver_test_test.go`",
+      "reference": "this should probably be named somewhere if you want to reuse it."
+    },
+    {
+      "prediction": "I think this should be `BoundServiceAccountSigningKeyFileEnvVar`.",
+      "reference": "needs a rename to not include AWS"
+    },
+    {
+      "prediction": "I think it would be better to move this header to the top of the file, so that it is only included once.",
+      "reference": "Isn't sigma have `double` type?"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "May be better use nano::keypair that generates ramdom value?"
+    },
+    {
+      "prediction": "I think we can remove this check now, since we don't know what the user selected for BSON options here.",
+      "reference": "I think the more complete solution will save off a `ignoreUndefined` in the base class for the bulk operation (with a default of `false`), and use that for calculations. Was there a reason you didn't want to support the option from the operation level?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the logout code path.",
+      "reference": "The catch (Exception e) block that follows does an updateSync, which will get a SmartStore instance (while logging out is taking place) - and then terrible things will happen - a database gets created for the outgoing user that won't be openable by the returning user causing the app to crash at logout. This is a somewhat superficial fix - SmartStore should not return an instance for an outgoing user."
+    },
+    {
+      "prediction": "I think this test should be moved to `integration/test/integration_test.go` since it's not related to the PR.",
+      "reference": "huh does this work without a `Return`?"
+    },
+    {
+      "prediction": "I think this is the only place where `set_exception` is used, so I think it's safe to remove it.",
+      "reference": "Don't think we want this line, right? (Plus \"failing\" is spelled wrong)."
+    },
+    {
+      "prediction": "I think this should be `setItemVisibility(menu, R.id.share_item, !selectedItem.getFeed().isLocalFeed());` and not `setItemVisibility(menu, R.id.share_item, selectedItem.getFeed().isLocalFeed());`",
+      "reference": "I think it would be more clear to have local-feed-hiding all in one place (bottom of this method?). Further up the method, there already is some code that hides the website icon, for example."
+    },
+    {
+      "prediction": "This should be removed as well.",
+      "reference": "Why remove this one? It doesn't work yet, but we'll add it sooner or later."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `models`.",
+      "reference": "Why did you remove the struct tags?"
+    },
+    {
+      "prediction": "I don't think this is the best way to do it. I think it would be better to have a function that takes the data points and returns a dict with the key being the name of the report and the value being a list of tuples containing the link and the name of the report.",
+      "reference": "Let's be neutral with tag names. Let's just have \"ReportURL\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/integration/test_span_data.go`",
+      "reference": "Use an explicit time to ensure conversion is not copy-paste and wrong."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I don't have a better suggestion.",
+      "reference": "The logic now is to look for `<account_out` if that is present return the string starting from there. If it is not present look for `<error` and return the string from there (which can produce malformed XML because you do not check if the closing tag matches). If this is also not found return an empty `<account_out/>`. This `<error>` output is also not consistent with the other two functions you mentioned. Do the other two RPCs not report an error?"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundMethodBody: Extra empty line detected at method body end.",
+      "reference": "Use nested module/class definitions instead of compact style."
+    },
+    {
+      "prediction": "Why is this here? I don't see it being used anywhere in this file.",
+      "reference": "Do you think we need to keep this in, or is it just leftover personal debugging? Seems like it wouldn't be too useful most of the time, but I don't care too much if you want to keep it."
+    },
+    {
+      "prediction": "I think this should be `config.showPlayers() | config.showNpcs() | config.FreezeTimers() | config.Veng() | config.TB()`.",
+      "reference": "showNPCs is not needed"
+    },
+    {
+      "prediction": "I think this should be `ErrParameterConflict` instead of `ErrParameterAlreadyExists`",
+      "reference": "Docstring for public function pls"
+    },
+    {
+      "prediction": "I think this test should be more specific about the error message it's expecting to be trimmed. I think it should be something like `expected error message to be unchanged` instead of `expected temporary file to be trimmed from error message`",
+      "reference": "This should use `t.Errorf` instead. You want the second part of the test to run even when the first part fails."
+    },
+    {
+      "prediction": "I don't think we need to check for `ratio.http3 > 0` here, since it's already checked in `if (ctx.protocol_selector.ratio.http3 > 0) {`",
+      "reference": "Should this if block better be changed to `#if H2O_USE_LIBUV \\n #else ... #endif`? The reason I wonder is because that's the way the QUIC context is being initilaized at the beginning of this function. Though I would not expect practical difference between the two approaches, because ATM the only case where we create QUIC connections is when the protocol selector is set to non-zero, and because when it is set to non-zero `h2o_quiy_close_all_connections` becomes a no-op."
+    },
+    {
+      "prediction": "I think it would be better to use `Object.assign` instead of `Object.is` here.",
+      "reference": "`Object.is` is an ES6 feature of JS so I don't think we can use it here (or we have to change our browser support matrix or specify that an Object.is polyfill is pre-req of `preact/hooks`). Should we just do an `===` check in `preact/hooks` and provide a `Object.is` polyfill and version of `useReducer` in `preact/compat`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to handle this. I think we should check if `options.roles` is an array, and if it is, we should just return it. If it's not, we should throw an error telling the user to provide an array of roles.",
+      "reference": "This was a bug I picked up by using the TS interface as a guide, this seems like it was / is the intention, also is a bug in master (needs port)"
+    },
+    {
+      "prediction": "I think this should be `COLLECTION_WITH` instead of `COLLECTION_WITH_COLLECTION`.",
+      "reference": "It would be helpful to explicit here what this really means and assumes."
+    },
+    {
+      "prediction": "I think this should be `map[string]*big.Int` instead of `map[string]map[string]*big.Int`",
+      "reference": "`ExpectedBalances` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `spec/formatters`.",
+      "reference": "This is another place where I'd feel more comfortable with declaring the class private. We can always make it public in the future if someone makes a case for that, but I prefer to err on the side of privateness for things like this that 99% of RSpec users won't have a reason to use."
+    },
+    {
+      "prediction": "I think this should be `nuk.IssuerAccount = optIssuerAccount`",
+      "reference": "Feels like this should be left as is and add a new createClientWithIssuers or something like that. Avoid all the \"\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `SPRequestHandler` class, since it's only responsible for handling the success case.",
+      "reference": "Minor unrelated bug in the IDP flow where the error wasn't getting displayed as a `Toast`."
+    },
+    {
+      "prediction": "This should be `const mongoErrorContextSymbol = require('../../lib/core').mongoErrorContextSymbol;`",
+      "reference": "I thought we got rid of this thing?"
+    },
+    {
+      "prediction": "I think this should be `@param string $file_name` and `@param string $file_path` instead of `@param string $file_name` and `@param string $file_path`",
+      "reference": "Please convert int params as well."
+    },
+    {
+      "prediction": "I think this should be `{0}{1} {2} {3}: {4}`",
+      "reference": "Fixed bug in implementation. Now I can run the validator from the VS Tools menu, double-click on an output line, and navigate to the site of the issue."
+    },
+    {
+      "prediction": "I don't think we need to add this to the docs.",
+      "reference": "Can we rename this field to `HttpServerErrorCodes` or `HttpServerErrorStatuses`? It will contain a list of status _codes_, not a list of _errors_. (Personally I prefer \"codes\" over \"statuses\", but we can't change `DD_HTTP_SERVER_ERROR_CODES`.)"
+    },
+    {
+      "prediction": "I think this should be `strncasecmp(tmp, \"false\", 3) == 0` instead of `strcasecmp(tmp, \"true\") == 0`",
+      "reference": "would you please use: flb_utils_bool(...) here ?, that function wraps the on/off/true/false stuff."
+    },
+    {
+      "prediction": "Should this be nullable?",
+      "reference": "`invocationInfoTokensToRedact: null` (I really like the convention of using a named parameter whenever the value doesn't communicate the meaning, such as for literal values.)"
+    },
+    {
+      "prediction": "I think this should be renamed to `async_test` to be consistent with other modules.",
+      "reference": "maybe renamed to `async_simple_test` if we consider supporting aug test later on?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `Document`.",
+      "reference": "So where is default_attributes now? Is it used anywhere else still? If not, can it be removed along with any tests of it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that the `StartTimestamp` and `Timestamp` are always in the same timezone.",
+      "reference": "We could move this to the base class and remove all these changes. What do you think?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do here, but I couldn't come up with a better idea.",
+      "reference": "While we're doing some translation work, could you extract this button text and the other static text from the view?"
+    },
+    {
+      "prediction": "I think you can remove the `if(headersValue != null)`.",
+      "reference": "Nit: Please fix indentation (note: Azkaban uses 2 spaces and not tabs)."
+    },
+    {
+      "prediction": "I think this should be `schema: []` instead of `schemaVersion: 1, schema: []`",
+      "reference": "This change is necessary because the schema version is ignored unless a schema is specified. I think that has always been the intent."
+    },
+    {
+      "prediction": "This should be `case WINDOWS_VERSION_10_1803: info->version = DR_WINDOWS_VERSION_10_1803; break;`",
+      "reference": "The api/docs/release.dox changelog message is missing: maybe you planned to add it once NtAllocateVirtualMemoryEx and NtMapViewOfSectionEx support is in? I would say, add it here in the same diff that raises max_supported_os_version."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "Just do: `return node.hasAttribute('aria-hidden')`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it should go in `spec/views/catalog/search_controller_spec.rb` since it's a controller spec.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `private static int _currentProcessId;`",
+      "reference": "Do we need `_processDataPoisoned`? Can the name or the id of the current process ever change? Or the machine name?"
+    },
+    {
+      "prediction": "I'm not sure why we need to set the environment variable here.",
+      "reference": "Did it not work out to wait until after containerd and kubelet are started?"
+    },
+    {
+      "prediction": "This should be `ITEM_PARSE_COOLDOWNREDUCTION`, not `ITEM_PARSE_COOLDOWN`.",
+      "reference": "variable names suggestions from me: - damageboost, ITEM_PARSE_DAMAGEBOOST, \"damage boost +x%\" - healingboost, ITEM_PARSE_HEALINGBOOST, \"healing power +y%\" - managainboost, ITEM_PARSE_MANAGAINBOOST, \"mana restoration +z%\" alternatively the other names can stay, because \"increase\" convention isn't that bad, just change mana values to: - \"increasemanarestored\", ITEM_PARSE_INCREASEMANARESTORED, \"mana restoration +z%\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `data` table, since that's where the data is stored.",
+      "reference": "Why can't this be determined by 'PARTIAL_SUCCESS' in the `status` field?"
+    },
+    {
+      "prediction": "Shouldn't this be `sprintf('%s/%04s/%02s', $media->getContext() ?? '', $repFirstLevel + 1, $repSecondLevel + 1)`?",
+      "reference": "Does it make sense to generate a Exception here? Maybe yes, because it wouldn't make sense to generate a path without the id of the media right?"
+    },
+    {
+      "prediction": "I think you can remove the `INSTANCE` variable and just call `getInstance()` in the caller.",
+      "reference": "This should be made private."
+    },
+    {
+      "prediction": "I think this should be `BaseSVIDPath: a.getBaseSVIDPath(),` and not `BaseSVIDPath: a.getBaseSVIDPath(),`",
+      "reference": "perhaps this is better modeled as a pkg-level var?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the same place as the other constants.",
+      "reference": "I don't think there is a maximum scale, neither for Hive nor for Trafodion. The only condition right now is that the scale can't exceed the precision. Example of a valid scale: DECIMAL(18,18). The maximum of 6 digits applies only to TIMESTAMP columns, where we don't support resolution below microseconds."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `FakeOAuthServer` since it's related to OAuth.",
+      "reference": "`ctx` is unused in UpdateProjectStaticUser"
+    },
+    {
+      "prediction": "I think this should be `String.format(\"https://content.googleapis.com/discovery/v1/apis/%s/%s/rest\", apiName, apiVersion);`",
+      "reference": "Why is this not the default, and why only for JS?"
+    },
+    {
+      "prediction": "Align the parameters of a method call if they span more than one line.",
+      "reference": "A repository has many badges so we should also add type column in scope."
+    },
+    {
+      "prediction": "I think this should be `result.CodeFlows[0].Should().Equal(new[] { new AnnotatedCodeLocation {`",
+      "reference": "Now a hash set, so can't index into it."
+    },
+    {
+      "prediction": "I think this should be `!!ENV['PSModulePath']` instead of `!!ENV['PSYTHON_MODULE_PATH']`",
+      "reference": "@jpogran Does this seem like a reasonable way to know if we're in powershell vs. CMD or *sh?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `contentType` can be `null` in this case.",
+      "reference": "1.toLowerCase(Locale.US)? 2.where is accept bug fix?"
+    },
+    {
+      "prediction": "I think this should be `KeyError` instead of `SystemExit`",
+      "reference": "We should still be providing something to the user in the case of a `KeyError` instead of a stack trace!?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `MarketplaceContractResolver` class.",
+      "reference": "check Lukasz's fix with NonceReserving - is that not better?"
+    },
+    {
+      "prediction": "I think this should be `this.colour_ = colour;` instead of `this.colour_ = colour || this.colour_;`",
+      "reference": "Would you explain why here please? E.g. from the PR description > the renderer expects to be able to use the secondary color as the fill for a shadow."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `hd.lock.RLock()` and `hd.lock.RUnlock()` calls below.",
+      "reference": "Lock needs to go to `RecoverFromDb` instead. Here it may cause deadlocks. My convention was that un-exported functions do not lock, only exported ones (with name starting with a capital letter)"
+    },
+    {
+      "prediction": "I think this should be `public String[] getDomains(String... domains)` and not `public String[] getDomains(String domain)`",
+      "reference": "German keyboard would also be used in Austria and Switzerland, so this should have included `.at` and `.ch`."
+    },
+    {
+      "prediction": "I'm not sure why we're removing these. I think they're still useful. I'm not sure why we're removing them. I think they're still useful. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm not sure why we're removing them. I'm",
+      "reference": "since this makes the indexer thing really only used for a single table, now, you can put in a TODO for me to cleanup and simplify this"
+    },
+    {
+      "prediction": "I think this should be named `NSCD_SOCKET_VOLUME_MOUNT_PATH` instead of `NSCD_SOCKET_VOLUME_NAME`",
+      "reference": "Should we have this path passed from the Azkaban properties? If this is the standard path for nscd even outside LinkedIn, then we can keep it."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think we should be able to just do `cfg.StreamConfig.AllowNoSubject = mqttStreamName != \"\"`",
+      "reference": "Could be a non-public field. `allowNoSubject`"
+    },
+    {
+      "prediction": "I think it would be better to have a `ForceRebuild` flag here, so that we don't have to keep track of the `ForceRerun` flag and the `ForceRebuild` flag at the same time. I think it would be better to have a `ForceRebuild` flag and a `ForceRerun` flag, so that we don't have to keep track of the `ForceRerun` flag and the `ForceRebuild` flag at the same time. I think it would be better to have a `ForceRebuild` flag and a `ForceRerun` flag, so that we don't have to keep track of the `ForceRerun` flag and the `ForceRebuild` flag at the same time. I think it would be better to have a `ForceRebuild` flag and a `ForceRerun` flag, so that we don't have to keep track of the `ForceRerun` flag and the `ForceRebuild` flag at the same time. I think it would be better to have a `ForceRebuild` flag and a `ForceRerun` flag, so that we don't have to keep track of the `ForceRer",
+      "reference": "Don't think this is quite right - think the build one also needs an `IsOriginalTarget` (c.f. code in `src/build`)"
+    },
+    {
+      "prediction": "I think this should be `array('results' => $this->processResults($paginator->getCurrentPageResults(), $targetEntityConfig));` and not `return array('results' => $this->processResults($paginator->getCurrentPageResults(), $targetEntityConfig));`",
+      "reference": "what about adding typehints instead? I don't think we need to add docblocks for every private methods."
+    },
+    {
+      "prediction": "I think we need to add `-o` for `cStorVolumeReplica.Spec.VolumeType == \"pd-shard\"` and `cStorVolumeReplica.Spec.VolumeType == \"pd-shard-shard\"`",
+      "reference": "Is it fine to pass the \"-o compression=on\" in the middle instead of at the end ?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I'm not sure we want to mix in `load_config` into the `Bolt::Analytics::NoopClient` class. I'm not sure we want to mix in `load_config` into the `Bolt::Analytics::BundledContent` class. I'm not sure we want to mix in `load_config` into the `Bolt::BundledContent` class. I'm not sure we want to mix in `load_config` into the `Bolt::BundledContent` class. I'm not sure we want to mix in `load_config` into the `Bolt::BundledContent` class. I'm not sure we want to mix in `load_config` into the `Bolt::BundledContent` class. I'm not sure we want to mix in `load_config` into the `Bolt::BundledContent` class. I'm not sure we want to mix in `load_config` into the `Bolt::BundledContent` class. I'm not sure we want to mix in `load_config` into the `Bolt::BundledContent` class. I'm not sure we want",
+      "reference": "Probably make them all keyword arguments."
+    },
+    {
+      "prediction": "I think this is the only place where `backdrop.setBackdrops([item])` is used. I think it should be removed.",
+      "reference": "Why not just check the backdrop setting here as well for the time being? We can combine the code later but at least then it would be configurable if people like the backgrounds."
+    },
+    {
+      "prediction": "I think this should be `Warning` instead of `Info`",
+      "reference": "why are we turning this one into a Warning?"
+    },
+    {
+      "prediction": "I think this should be `String.valueOf(sourceIDs.size() + 1)`",
+      "reference": "Wouldn't it be better to use double quotes for the opening parenthesis ?"
+    },
+    {
+      "prediction": "`blessings` is a `uint32_t` so this should be `static_cast<uint32_t>(player->blessings);`",
+      "reference": "Remove the cast"
+    },
+    {
+      "prediction": "I'm not sure if we want to include this file in the release?",
+      "reference": "What criteria do we use to bump the copyright years? An update to the module?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `transforms.py` file, under the `Resize` class. The reason is that the `Resize` class is the base class for all image transforms, so it should have access to all the transforms defined in the `transforms.py` file. For example, it should have access to the `ResizeWithDilations` class defined in the `transforms.py` file.",
+      "reference": "should we rm line 269?"
+    },
+    {
+      "prediction": "I think this should be `setattr(self, \"_isNvdaStartupComplete\", True)` instead of `setattr(self, \"_isNvdaStartupComplete\", lambda: setattr(self, \"_isNvdaStartupComplete\", True))`",
+      "reference": "Perhaps the code at `source/core.py:564: postNvdaStartup.notify()` should be queued instead? If we are saying the loop must have started before NVDA's startup is complete, then the `postNvdaStartup` action is incorrect."
+    },
+    {
+      "prediction": "Layout/AlignHash: Align the elements of a hash literal if they span more than one line.",
+      "reference": "Needed this for testing locally with mail view but not used for any tests. Might be nice to have some seed scripts at some point."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle the case where we don't have an `AcceptResponseError` in the response header.",
+      "reference": "wait, if we only do this for successful status codes won't we miss a whole slew of responses? We should be checking this for all codes right?"
+    },
+    {
+      "prediction": "I think this should be `new String[] { \"library.yaml\" }` instead of `new String[] {\"library.yaml\"}`",
+      "reference": "Adding in the service yaml `library.yaml` here."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The problem is that the `RemoteUserCtxKey` is a key, not a value.",
+      "reference": "Why is this dereference needed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/repo`.",
+      "reference": "Might want to add that this only happens if you have write permissions to the folder."
+    },
+    {
+      "prediction": "I think it would be better to make this a `bool` and set it to `true` by default.",
+      "reference": "Adding a small blurb about why someone would want to enable this flag would help clarify what it is for. Such as something about when providing a custom retry handler and how`ShouldRetry` will be handled with and without the flag enabled."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the old format for backwards compatibility.",
+      "reference": "assume the user will be able to register for a custom service name in future?"
+    },
+    {
+      "prediction": "This file should be renamed to `GraphUtilities.cs`",
+      "reference": "The extra two empty lines seem unnecessary. Please review your diffs more carefully."
+    },
+    {
+      "prediction": "I think we should rename this to `EthAPI` instead of `OracleAPI` to be consistent with the naming of the other API interfaces.",
+      "reference": "What the reason to embed the Oracle?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this route. I think it should be under `workshops`, since it's related to the `Workshop` controller. But I don't know if it's better to put it under `pages`, since it's related to the `Pages` controller. Can you help me decide?",
+      "reference": "Shouldn't this happen automatically with High Voltage?"
+    },
+    {
+      "prediction": "I think this should be `entityValue = context.getNamer().varReference(Name.from(initValue.getValue()));`",
+      "reference": "This is changed because we need `projectId` to be printed as `$projectId` in PHP."
+    },
+    {
+      "prediction": "I think this should be `customization_of` instead of `customization_of_template`",
+      "reference": "This title change seems more accurate to me however not sure if this will confuse users."
+    },
+    {
+      "prediction": "I think this should be `^(?:.*:)*wikipedia$` and not `if key =~ /^(?:.*:)*wikipedia$```",
+      "reference": "Here (and on line 158 as well), would it be more efficient to use the String end_with() method? Not tested"
+    },
+    {
+      "prediction": "I think this should be `ram_handler = RAMHandler(capacity=capacity)` instead of `ram_handler = RAMHandler(capacity=self.capacity)`",
+      "reference": "Why are you creating a new (second) `RAMHandler` here? Is this just an oversight from deleting the previous code?"
+    },
+    {
+      "prediction": "I think this should be `resp_.set_code(req.get_if_not_exists() ? cpp2::ErrorCode::SUCCEEDED : cpp2::ErrorCode::E_EXISTED);`",
+      "reference": "Move the line 27 to line 31 is better?"
+    },
+    {
+      "prediction": "I think this should be `fbo.rekeyFSM.Event(NewRekeyEvent(rmd.IsRekeySet()))` instead of `fbo.rekeyFSM.Event(NewRekeyNotNeededEvent())`",
+      "reference": "In slack I mentioned we should only do this if the update is from some other device, to make sure our own updates don't cause issues. Maybe the revision check above is good enough to prevent this, but I just want to make sure you thought about it."
+    },
+    {
+      "prediction": "I think this should be `--seed 1234` and not `--seed 123456789`",
+      "reference": "I would instead explain that random is the default, but you can switch back to defined, or another ordering scheme."
+    },
+    {
+      "prediction": "I think this should be `<= 1` instead of `<= 0`",
+      "reference": "Is there a reason you left this at 0?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `AbstractRememberMeServices` constructor.",
+      "reference": "The value of `key` will never be `null` at this point, because `getKey()` will generate a value if it is `null`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with at the time of writing.",
+      "reference": "NIT: You should first check if AdoNetExcludedTypes.Count is greater than 0 before calling Contains (fast-path)"
+    },
+    {
+      "prediction": "Can you remove this empty line?",
+      "reference": "should rearrange so as to follow import ordering."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "what is the use case for create workflow with completed state?"
+    },
+    {
+      "prediction": "This file should be renamed to `RemoteDistributorTest.java`",
+      "reference": "Probably best not to have an empty test...."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to push the region.",
+      "reference": "Missing the license"
+    },
+    {
+      "prediction": "I think it would be better to use `p.logger.Errorw` instead of `p.logger.Error` to be consistent with the rest of the code.",
+      "reference": "Looks like this was previously guarded by `EnableCleanupReplicationTask` flag. Now looks like this is always needed. Just want to confirm the intention is if `GlobalNamespace` is enabled then we want run cleanupReplicationTasks in all cases?"
+    },
+    {
+      "prediction": "This is not needed anymore.",
+      "reference": "Potentially it'll be good to remove all of them excluding \"tbb/tbb.h\" But let's do it next time"
+    },
+    {
+      "prediction": "I don't think this is the right place for this function. It's not about finding the range, it's about determining if the range is large enough to be considered bounded by the data.",
+      "reference": "I would just mention which tolerance - namely as reported by ``sys.float_info``"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": ":+1: Ideally this'll become something you could pass in to the config"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to pass the executor to `RaftIterator` and then call `RaftIterator::replicateLogs` instead of `RaftPart::replicateLogs`.",
+      "reference": "Should using `thenValue` to replace `then` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to get the context from the `TraceContext` in the `OnError` event handler.",
+      "reference": "This follows the convention of each callback starting with the `IsIntegrationEnabled` check."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for these properties. I think they should be in the `Meter` class.",
+      "reference": "this is not required to be part of this PR right? (With UpDownCounter being absent in .NET, we can make this hardcoded for now, i think)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface. I think it would be better to move it to `KestrelTransportHandler` or `TransportHandler` interface.",
+      "reference": "Add the PipeFactory here"
+    },
+    {
+      "prediction": "I think this is the wrong place to put this. It should be in `common.ChaosImpl`, not in `Delegate`.",
+      "reference": "please add `var _ impltypes.ChaosImpl = (*Delegate)(nil)`"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it should go in the `Snapshot` interface.",
+      "reference": "nit: is this added deliberately?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Given the number of tests which need this (I think I counted seven) should we maybe just install this one globally? Is there even a place to do that? Something in `test_helper` maybe?"
+    },
+    {
+      "prediction": "I think we can remove this function and use `typeof maybePromise.then === 'function'` directly.",
+      "reference": "we don't use this anymore, so we can delete it."
+    },
+    {
+      "prediction": "I think this should be `fmt.Fprintln(os.Stdout, \"algoh telemetry configured from file:\", telemetryConfig.FilePath)`",
+      "reference": "nit: change to \"Telemetry configuration loaded from '%s'\""
+    },
+    {
+      "prediction": "896da21682d925964ece3a0601024704010402040304\")\n \tif err := assert.Equal(t, expected, expected); err != nil {\n \t\tt.Errorf(\"Expected: %v, Got: %v\", expected, err)\n \t}\n```\nReview comment?",
+      "reference": "Why did this value change?"
+    },
+    {
+      "prediction": "I think we can remove `gcpduckv1alpha1` now that we have `gcp.pubsub.v1alpha1`",
+      "reference": "maybe `duckpubsubv1alpha1` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. It seems like it should be in `extensions/manifest.json`, but I don't see it being used there.",
+      "reference": "The id is not mandatory for web extensions. Could you update this to support a web extension without an id?"
+    },
+    {
+      "prediction": "I think this should be `serviceName = \"pubsub.googleapis.com\"`",
+      "reference": "I actually prefer them to be defined here. Client who calls `lib` should define the args, not `lib` itself."
+    },
+    {
+      "prediction": "I think this should be done in `epollConsole.Close()`",
+      "reference": "I think this should've been done with the `defer`-named-return trick. I'll submit a PR for it."
+    },
+    {
+      "prediction": "I think it would be better to use `boost::asio::ip::tcp::socket` instead of `nlohmann::tcp::Endpoint`.",
+      "reference": "The bind address should be configurable."
+    },
+    {
+      "prediction": "This method is not used anywhere, so I think it can be removed.",
+      "reference": "Maybe the empty `delete()` should be implemented as default in the interface."
+    },
+    {
+      "prediction": "Please use `CMAKE_CURRENT_LIST_DIR` instead of `CMAKE_SOURCE_DIR`.",
+      "reference": "How to ensure CMAKE_SOURCE_DIR is set correctly?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to make sure that the `Balance` and `VotingWeight` are both set to `0` before we create the account state.",
+      "reference": "safer to make a copy of incoming *big.Int"
+    },
+    {
+      "prediction": "I think we should use `pullImageTimeout` instead of `ListImagesTimeout`",
+      "reference": "how did we choose this value? does this hold good for Windows too?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `libcontainer/storage` or `libcontainer/network` or `libcontainer/image` or `libcontainer/registry` or `libcontainer/network` or `libcontainer/storage` or `libcontainer/image` or `libcontainer/registry` or `libcontainer/network` or `libcontainer/storage` or `libcontainer/image` or `libcontainer/registry` or `libcontainer/network` or `libcontainer/storage` or `libcontainer/image` or `libcontainer/registry` or `libcontainer/network` or `libcontainer/storage` or `libcontainer/image` or `libcontainer/registry` or `libcontainer/network` or `libcontainer/storage` or `libcontainer/image` or `libcontainer/registry` or `libcontainer/network` or `libcontainer/storage` or `libcontainer/image` or `libcontainer/registry` or `libcontainer/network` or `libcontainer/storage` or `libcontainer/image` or `libcontainer/registry` or `libcontainer/network` or `libcontainer/storage` or `libcontainer/image` or `libcontainer/registry` or `libcontainer/network` or `libcontainer/storage` or `",
+      "reference": "To follow the other build tags, maybe we can call it `criu`?"
+    },
+    {
+      "prediction": "I think this should be `timeout` instead of `connect_timeout` to be consistent with the rest of the gem.",
+      "reference": "Should we load this from the config file too?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it.",
+      "reference": "Can you file an issue for this?"
+    },
+    {
+      "prediction": "I think we can remove the `this[kIsMaster] == null` check here.",
+      "reference": "@nbbeeken what were the cases where `!!this[kIsMaster]` was yielding an incorrect value? we should try to cover them in the tests"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. It seems like it should be in the `Bolt::Inventory` class, not the `Bolt::Catalog` class. The `setup_inventory` method is only called in the `Bolt::Catalog` class, and it seems like it should be moved to the `Bolt::Inventory` class as well. Can you move this method to the `Bolt::Inventory` class?",
+      "reference": "This whole function feels messy. I don't have a better idea at the moment though."
+    },
+    {
+      "prediction": "I think you can remove the `typeMap.get(\"airplaneMode\").toString()` line as well.",
+      "reference": "should use Number instead of Long"
+    },
+    {
+      "prediction": "I think this should be `func (m MockProvider) WriterForFn(ctx context.Context, conf influxdb.Config) (influxdb.Writer, error)`",
+      "reference": "Can you rename this and the file `InfluxDBProvider` and `influxdb_provider.go` respectively?"
+    },
+    {
+      "prediction": "I think this should be `!== 0`",
+      "reference": "This is the real culprit, as now vlen will always be a Number (and known to the compiler as such). You could probably go one step further and avoid the ToBoolean on `vchildren` as well by writing something like `vlen = (vchildren !== undefined) ? vchildren.length : 0` if that matches the contract."
+    },
+    {
+      "prediction": "I think you can remove the `txtNoDataString` view here.",
+      "reference": "missing space before `TextView`"
+    },
+    {
+      "prediction": "This should be `break;` instead of `return`.",
+      "reference": "so this attribute is used as u8 in code, but saves as int32? Why?"
+    },
+    {
+      "prediction": "I don't think we need to change this, but I think it would be good to keep the old name for backwards compatibility.",
+      "reference": "can RDSE be added to the list? It has its capnp files."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `client` since it's the client that's going to be making the API calls.",
+      "reference": "These functions only ever read from the API and should always be reading from a cache, so I have made that explicit by expecting a Reader here and supplying a `cache.Cache` as the client."
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to accomplish, but it seems like it should be removed.",
+      "reference": "No, we can't skip any of the mangling after the suspend point: this should follow the other mangling and undo its push."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "this is generated code.. these changes will need to be made in the code generator to have them stick."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to test shadow support, but I don't have a better idea.",
+      "reference": "Should we abstract this into a reusable utility so it doesn't have to get repeated in every test file needing Shadow DOM support?"
+    },
+    {
+      "prediction": "nit: can you move this to the top of the file?",
+      "reference": "Maybe we rename this to `command_monitoring.test.js` to match the directory name change? Or would you rather do that as part of the greater test cleanup? I'm fine either way."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to fix it, but it's the only way I could get it to work.",
+      "reference": "Why is this change necessary? `handleSingle()` will cast it to a string, no?"
+    },
+    {
+      "prediction": "I think this should be `default String name() { return \"table(\" + hashCode() + \")\"; }`",
+      "reference": "This newly added method should be well defined, should it return `TableIdentifier` or just `String`?"
+    },
+    {
+      "prediction": "Shouldn't this be \"Multimedia has %count% relations with %relations%\"?",
+      "reference": "have a relation with a product"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think we need to make sure that the pod is actually gone before we try to delete it. If the pod is still there, then we should not be able to delete it. If the pod is not there, then we should be able to delete it.",
+      "reference": "I guess that there is still a window here, because the GC could happen between the `PodExist` and `Delete` calls. Would it be better instead to check `err` and allow it if it says \"pod has already been deleted\"?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be using `transaction.commit()` and `transaction.rollback()` instead of `session.commit()` and `session.rollback()`. This is because `transaction.commit()` and `transaction.rollback()` should be called in a transaction, and `transaction.commit()` and `transaction.rollback()` should only be called if the transaction is in a consistent state (i.e. all operations have completed and no errors have occurred). If we call `transaction.commit()` and `transaction.rollback()` here, it means that if there was an error while committing or rolling back the transaction, then the entire transaction will be rolled back, which may not be what we want.",
+      "reference": "SQLAlchemy's session management does magic to make sure that if the rollback fails you still get the original exception that caused the rollback. Also it looks nicer."
+    },
+    {
+      "prediction": "Why do we need to convert from actionPb to transferPb?",
+      "reference": "This switch statement can be removed. Just return q.items[act.Nonce] != nil"
+    },
+    {
+      "prediction": "nit: I think you can simplify this to `n.equals(n.getParent().getChild(1))`",
+      "reference": "I think `==` for nodes is more readable than equals. An equals calls looks like it could be recursing, because intuitively two nodes are equal if their subtree are the equal. But everywhere you replaced, we don't want to test whether the subtrees are structurally equal, we want to know whether they're the same. Only `==` captures this intent, using equals obscures this. Not to mention that equals may NPE and is in that sense less ergonomic. I think we should enhance the rule with a list of types for which `==` is correct. Edit: in the meantime i think we could suppress those new violations..."
+    },
+    {
+      "prediction": "This should be renamed to `BenchmarkFilters` or something similar.",
+      "reference": "What's the scenario for passing multiple filters?"
+    },
+    {
+      "prediction": "I think this should be `iter->logEntry()` instead of `std::get<3>(logEntry)`",
+      "reference": "move to after check log.empty()"
+    },
+    {
+      "prediction": "I don't think we should be using `docutils` for this. We should be using `jinja2` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils` or `jinja2_docutils`",
+      "reference": "This line needs to be removed"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this. I think it should be something like `cloud.build.source.buildId` or `cloud.build.source.buildStatus`?",
+      "reference": "Can we delete such const like `CloudBuildSourceEvent` and `CloudBuildSourceBuildId` since all those are contained under schemas/v1?"
+    },
+    {
+      "prediction": "Is this package used anywhere? I don't see it referenced anywhere in this PR.",
+      "reference": "This is unused."
+    },
+    {
+      "prediction": "I think this should be `ITextFormatBaggage` instead of `ITextFormat`",
+      "reference": "Any reason not to put `using System.Collections.Generic` (guess there shouldn't be naming conflicts)?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the easiest I could come up with at the time of writing.",
+      "reference": "Could the name of this controller just be shortened to `TwitterCardsController`? Is the `Player` there adding information?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check if the host is healthy. I'm not sure if we should be checking if the host is healthy or if we should be checking if the upstream is healthy. If we should be checking if the upstream is healthy, I'm not sure if we should be checking if the host is healthy or if we should be checking if the upstream is healthy. I'm not sure if we should be checking if the host is healthy or if we should be checking if the upstream is healthy. I'm not sure if we should be checking if the host is healthy or if we should be checking if the upstream is healthy. I'm not sure if we should be checking if the host is healthy or if we should be checking if the upstream is healthy. I'm not sure if we should be checking if the host is healthy or if we should be checking if the upstream is healthy. I'm not sure if we should be checking if the host is healthy or if we should be checking if the upstream is healthy. I'm not sure if we should be checking if the host is healthy or if we should be checking if the upstream is healthy. I'm not sure if we should be checking if the host is healthy or if we should be",
+      "reference": "Backends with hostnames that depend on the request are always up?"
+    },
+    {
+      "prediction": "I don't think we should be using `pyramid_exceptions` here. I think we should be using `jsonpatch` and `jsonpointer` from `pytz`.",
+      "reference": "I'd rather catch those to `utils.py` and raise a simple ValueError from them. From the resource point of view, these are details of implementation"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "make this 1.0 ? I don't think all that many people do `language: julia` without any `julia:` specifiers, but may as well keep that possible?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "nit: `AZURE_KEYVAULT_AUTH_VIA_CLI` is defined by us, so it might be helpful to differentiate it from azure's env names. Maybe starts with `GOCDK`."
+    },
+    {
+      "prediction": "I think we should use `storage.SessionStorage` instead of `storage.Storage` here.",
+      "reference": "Depend on interfaces not on structures"
+    },
+    {
+      "prediction": "I'm not sure about this change. The `stop()` function is called by the `StorageServerThread` and the `notifyStop()` function is called by the `StorageServerThread` and the `stop()` function. I think it would be better to call `stop()` in `StorageServerThread::stop()` and `notifyStop()` in `StorageServerThread::notifyStop()`.",
+      "reference": "This fix looks good. My concern, don't forget to fix metad too!"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I'm not sure we should be relying on the fact that the latest branch is a squash branch to know that we should use the earliest branch as the journal branch ID. I'm not sure we should be relying on the fact that the earliest branch is a local branch to know that we should use the latest branch as the journal branch ID. I'm not sure we should be relying on the fact that the latest branch is a squash branch to know that we should use the earliest branch as the journal branch ID. I'm not sure we should be relying on the fact that the earliest branch is a local branch to know that we should use the latest branch as the journal branch ID. I'm not sure we should be relying on the fact that the latest branch is a squash branch to know that we should use the earliest branch as the journal branch ID. I'm not sure we should be relying on the fact that the earliest branch is a local branch to know that we should use the latest branch as the journal branch ID. I'm not sure we should be relying on the fact that the latest branch is a squash branch to know that we should use the earliest branch as the journal branch ID. I'm not sure we",
+      "reference": "Just noticed this nil context. Perhaps plumb through ctx too, or change to `Debug`?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to do something like: DEFAULT_HEADERS = {'Accept' => 'application/x-www-form-urlencoded', 'Content-Type' => 'application/x-www-form-urlencoded', 'User-Agent' => 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/14.0 Safari/605.1.15', 'Accept-Language' => 'en-US,en-US;q=0.8,en;q=0.5', 'Accept-Encoding' => 'gzip, deflate', 'Connection' => 'keep-alive', 'Accept-Charset' => 'ISO-8859-1,utf-8', 'Keep-Alive' => 100, 'Strict-Transport-Security' => 'max-age=3111111111111111111111111111111111111111111111111111111111111",
+      "reference": "Does it send requests with urlencoded bodies anywhere? I thought it sends only json. Maybe content-type should be `application/json` by default?"
+    },
+    {
+      "prediction": "I think this should be 11810.",
+      "reference": "Again use of `constexpr` is recomended."
+    },
+    {
+      "prediction": "I don't think it's necessary to import LayoutAlgorithm as ds_layout = None here, since it's not used in this function.",
+      "reference": "Minor point but I would call this ``redim_graph`` instead."
+    },
+    {
+      "prediction": "I think we can remove `apis/cluster/v1alpha2` and `apis/infrastructure/v1alpha2` from `apis/cluster/v1` and `apis/infrastructure/v1`",
+      "reference": "As I was trying to figure out how the error messages are written out, since they are returned from `getOrCreate` and `reconcile`, it looks like we are using plain `klog` to write them out instead of using the logger from the scope above"
+    },
+    {
+      "prediction": "I think this should be `log.Warnf` instead of `log.Warnn`",
+      "reference": "Now it's double logging, is not it? Because `service.ipForward.Enable()` logs extra `warning`"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Whatever the decision is about the request type to use, this should only match what is needed and not the other one."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be.",
+      "reference": "...and in turn, this should probably be var updatePreviews = true; if (!(localConnection && localConnection.type == Blockly.OUTPUT_VALUE) && (Blockly.localConnection_ && Blockly.highlightedConnection_)) { since the first clause is a no-op. If you want to leave it this way for clarity, that's fine too."
+    },
+    {
+      "prediction": "I think we should rename this method to `find_violations` as well.",
+      "reference": "Please remove this TODO, since they will not apply anymore after you are done. :) Can you please remove this everywhere else in this PR?"
+    },
+    {
+      "prediction": "I don't think this is a valid test. `enum` and `enum clashes with datatype`. `enum` and `enum` and `enum clashes with datatype`. `enum` and `enum` and `enum` and `enum clashes with datatype`. `enum` and `enum` and `enum` and `enum` and `enum clashes with datatype`. `enum` and `enum` and `enum` and `enum` and `enum` and `enum clashes with datatype`. `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum clashes with datatype`. `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum clashes with datatype`. `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum clashes with datatype`. `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum` and `enum clashes with datatype`. `enum` and `enum` and `enum` and",
+      "reference": "This seems like an incomplete schema definition. will that have any effect on the tests?"
+    },
+    {
+      "prediction": "why do you need this check?",
+      "reference": "The testing cases are failing because `SetHeader` does not only handle cases where input are from files. It also reads categorical feature indices from the config parameters (see the part outside the `if (filename != nullptr) { ... }`). Skipping `SetHeader` directly here will cause errors when we load data from numpy or pandas arrays (where `filename == nullptr`) and use categorical features. So I think we should move the the check `filename != nullptr && CheckCanLoadFromBin(filename) == \"\"` into `SetHeader`. That is, we change `if (filename != nullptr) { ... }` into `if (filename != nullptr && CheckCanLoadFromBin(filename) == \"\") { ... }`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `cmd.Wait()` method, since that is the place where the process is started.",
+      "reference": "I think we should change the variable name of `err` in `if err, ok := err.(*exec.ExitError); ok `, this re-assignment of `err` still affects the next `else` branch"
+    },
+    {
+      "prediction": "nit: I think it would be nice to have a `folly::Future<StatusOr<bool>> MetaClient::dropSpace(std::string const& name)` method that does the same thing as the `folly::Future<StatusOr<bool>> MetaClient::dropSpace(std::string name)` method above.",
+      "reference": "`Status` has identical semantics with `StatusOr<bool>`."
+    },
+    {
+      "prediction": "I think this should be `\"/readyz/addlabels\"`.",
+      "reference": "why is this named in a different style from other paths?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this check. I think it should be done in the `Cluster` controller, but I'm not sure if that's the right place to do it.",
+      "reference": "Swapped these to ensure that we aren't passing a nil cluster in to util.IsPaused"
+    },
+    {
+      "prediction": "I think it would be better to use a map[string]platformPkgManagerTuple instead of a list of tuples.",
+      "reference": "I don't see pkgManager used anywhere, is there a reason we need this?"
+    },
+    {
+      "prediction": "Shouldn't this be `instanceDir.toPath()`?",
+      "reference": "A bit late, but I don't think this is necessary, as all callers will send absolute paths. And if you ever get a relative path, resolving it with `toAbsolutePath()` leads to it being relative to whatever CWD the app is started with, while the typical resolving of relative `instanceDir` is to resolve it relative to CoreContainer#coreRootDirectory."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `modelSchema.paths[localField]`",
+      "reference": "There's an awful lot of test failures here because `localFieldPathType.schema` may not contain a `getters` array."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Mysql allows for `||` concatenation (e.g. `firstname||' '||surname`) if you enable it: `set sql_mode=PIPES_AS_CONCAT;`. I think this check is safer though"
+    },
+    {
+      "prediction": "I think this file should be renamed to `ens.go`",
+      "reference": "NOTE: this is the default ENS registry address. If the ENS suite is deployed from the current builds to any chain, it will always be found at this address."
+    },
+    {
+      "prediction": "Layout/AlignParameters: Align the parameters of a method call if they span more than one line.",
+      "reference": "I'm curious why you are using `draw` in this route file?"
+    },
+    {
+      "prediction": "I don't think we need to panic here.",
+      "reference": "Those variables are only been used once, I think we don't need to create them at all, just pass values to the `GetRawDiff` directly."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "I like the idea of replacing all the `#if NET461` with `#if NETFRAMEWORK`... is that worth doing now? Means fewer changes if we go to 4.7.2 at some point"
+    },
+    {
+      "prediction": "I think this should be `clutchProviderName` instead of `clutchProvider`",
+      "reference": "to fix the docs build failure, make this const var named without the suffix `Name`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to move the fallthrough. I think it should be moved to the top of the function where it's first used.",
+      "reference": "This bugfix also included, in case a subjectAlternativeName was present in the cert, then URIs (e.g SVID SPIFFE auth) would not have been attempted."
+    },
+    {
+      "prediction": "I don't think we need this interface. We can just use `Uri.fromFile(file)`",
+      "reference": "Same here. Feels like the wrong location."
+    },
+    {
+      "prediction": "I don't think this should be in this PR.",
+      "reference": "Please make `_path_suggestion` public (i.e. remove the `_`) in `downloads.py`."
+    },
+    {
+      "prediction": "I think this should be `external-port-443-is-local-port` instead of `external-port-80-is-local-port`",
+      "reference": "Woah, these flag names gotta get shorter. (Edit: I'll try to suggest some if needed, will think on it.)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Is this the code that's supposed to use the Rails app's only when in Rails4? What's the point of the first `if app.respond_to?(:secrets)`, both the `if` and the `elsif` have the same body, is only the second one needed? If `app.config` has a `secret_key_base`, then use it, the end. Is there a need for first checking if `app.respond_to?(:secrets)`?"
+    },
+    {
+      "prediction": "I think this should be `enum_decl->isObjC() || (enum_decl->isObjCProxy() && !m_is_objc_enum);`",
+      "reference": "Does this not apply to C enums on Linux?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `if (isRetryableWriteError(err)) {` block.",
+      "reference": "Should this check if the error is `RetryableWriteError` before adding the label?"
+    },
+    {
+      "prediction": "I'm not sure why we need to import `get-form-data` from `rollup-plugin-commonjs` here. Can't we just import it from `get-form-data` in `rollup-plugin-commonjs`?",
+      "reference": "Oh wow, that's one complicated require :)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this snippet. It's in the `aws-sdk-go` package, which is the Go SDK for AWS, but the snippet is about using `context.Context` with SDK requests, which is the Go SDK for Golang. Should it be in `aws-sdk-golang` or `aws-sdk-go`?",
+      "reference": "do you mean how to \"use\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this test. I think it should go in `test/utils.js` since it's only testing a specific case.",
+      "reference": "We can't really test this anymore since ref and props.ref are at the backing node level now"
+    },
+    {
+      "prediction": "I don't think this is necessary. The iterator is initialized in the constructor.",
+      "reference": "Did you intend to set this in the constructor?"
+    },
+    {
+      "prediction": "I think this should be `err = s.defaultAMILookup(\"ubuntu\", \"18.04\", machine.Machine.Spec.Versions.Kubelet)`",
+      "reference": "This looks fine as a first approach, and can we put a TODO here about ubuntu 18.04 being hardcoded?"
+    },
+    {
+      "prediction": "I think this should be `#if defined(USE_OPENCV) && !defined(OPENCV_NO_DEPRECATED_6_0)`",
+      "reference": "This isn't strictly needed since lmdb / leveldb data that is unencoded does not require OpenCV IO. I think `DataLayer` should only require lmdb / leveldb. If one tries to use encoded data without OpenCV the `DataTransformer` guards will report the dependency failure."
+    },
+    {
+      "prediction": "I think this should be `hLog.WithError(err).Errorf(\"unable to apply hiveapi cluster role binding: %s\", err.Error())`",
+      "reference": "Should we take out this change? This was only in there to try to brute force the start of the apiserver yesterday."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I don't have a strong opinion. I just wanted to make sure I'm doing the right thing.",
+      "reference": "I worry that this association name is a bit confusing. Yoz was asking me about the concept of completion yesterday, which is why I think of it. It wasn't clear to him exactly what it was. what about calling this a `step_completer` ? we know that a proposal has many steps, so I think that might be clearer. in the future, we also might want the rename the `approver` relation because a step is not always an approval these days."
+    },
+    {
+      "prediction": "I think we can remove the `ApiVersion` field from the `OperationParametersValidation` class.",
+      "reference": "Just add a to-do saying we need to add api-version in the check some time in the future when we enable the single-swagger spec mode for validation."
+    },
+    {
+      "prediction": "I think this should be `Propel\\ConnectionManager::DEFAULT_DATOSRC`",
+      "reference": "@lunika Please see this. It allows us to do `Propel::getConnection()` easily, as we have only one database."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `AspNetWebApi2Integration.UpdateSpan` since that's the only place where we have access to the `AspNetTags` instance.",
+      "reference": "Not related to this particular cast but I wonder if we shouldn't have a cast helper that would log if null to gain visibility."
+    },
+    {
+      "prediction": "I think this should be `nextPageToken` instead of `incompleteSearch`.",
+      "reference": "why do we need to explicitly declare these fields?"
+    },
+    {
+      "prediction": "I think this should be done in the `AbstractOaiController` and not in this `OaiController`. The `AbstractOaiController` is the base class for all oai controllers.",
+      "reference": "Note that there are multiple OAI servers -- that's why `$serverClass` is a variable here. You'll want to fetch `$serverClass` from the service manager rather than a hard-coded value, and also set up a module.config.php for the authority record version, `VuFind\\OAI\\Server\\Auth`. Should be easy enough since it can share the same factory."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "This is no longer needed and should be removed."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "could use named return params here to avoid declaring the opts, and using naked returns instead."
+    },
+    {
+      "prediction": " <svg viewBox=\"0 0 58 58\" opacity=\"0.6\">\n       <path d=\"M36.537 28.156l-11-7a1.005 1.005 0 0 0-1.02-.033C24.2 21.3 24 21.635 24 22v14a1 1 0 0 0 1.537.844l11-7a1.002 1.002 0 0 0 0-1.688zM26 34.18V23.82L34.137 29 26 34.18z\" /><path d=\"M57 6H1a1 1 0 0 0-1 1v44a1 1 0 0 0 1 1h56a1 1 0 0 0 1-1V7a1 1 0 ",
+      "reference": "this is an unrelated fix. I noticed when an instagram carousel post is mixed with images and videos, the videos don't come with thumbnails, so I am adding a fallback thumbnail for this case."
+    },
+    {
+      "prediction": "I think this should be `RunStateTransitionWithReceipts` or `RunStateTransitionWithReceipts` to be consistent with `RunStateTransition`",
+      "reference": "\"... after protocol version 1\"?"
+    },
+    {
+      "prediction": "`[^:\\{\\}]+` should be `[^:\\{\\}]+` to match both `:` and `{`",
+      "reference": "Is there a unit test that covers this regexp? I am afraid that changing it might break things =/"
+    },
+    {
+      "prediction": "I think this should be `name.toCamelCase()`",
+      "reference": "This change is okay IMO because an `@` variable in Ruby is not a reference -- `@` variables are instance variables. Also, it's never used by the Ruby transformers."
+    },
+    {
+      "prediction": "I think it would be better to use `sprintf` instead of concatenation.",
+      "reference": "Why this change was needed? Because in `po` files it's with curly braces?"
+    },
+    {
+      "prediction": " | (0x00ffffe0 & off >> 2 << 5);\n+        ASSERT(off + 0x40000 < 0x80000);\n+        *pc_writable = (enc & 0xfff8001f) | (0x00ffffe0 & (off >> 2 << 5));\n     } else if ((enc & 0x7e000000) == 0x36000000) { /* TBNZ, TBZ */\n-        ASSERT(off + 0x2000 < 0x4000);\n-        *pc_writable = (enc & 0xfff8001f) | (0x00ffffe0 & off >> 2 << 5);\n+        ASSERT(off + 0x40000 < 0x80000);\n+        *pc_writable = (enc & 0xfff8001f) | (0x00ffffe0 & (off >> 2 << 5));\n     } else if ((enc & 0x7e00000",
+      "reference": "Wait -- where did this change come from? This seems unrelated to trace building and seems like it could affect basic execution, unlike the rest of this PR which is all under the off-by-default trace option. Please separate this into its own PR and probably ask @AssadHashmi to review."
+    },
+    {
+      "prediction": "I think this file should be removed.",
+      "reference": "seems to be redundant (mention `docusaurus`   )"
+    },
+    {
+      "prediction": "I think we should rename this method to `cherryPick` instead of `cherrypick`.",
+      "reference": "Can we combine this with the Rollback API? We could still support the `rollback` method here, but combine `Rollback` and `CherryPick` into something like `ManageSnapshots`. Then we could reuse logic for enforcing checks of the current snapshot. What do you think?"
+    },
+    {
+      "prediction": "I think this should be `const char*`",
+      "reference": "Maybe just `env_prefix`, I don't think this is referring to the WQ manager."
+    },
+    {
+      "prediction": "I think this file should be removed as it is not used anymore.",
+      "reference": "Unrelated line change :)"
+    },
+    {
+      "prediction": "I think we can remove this file.",
+      "reference": "I raised an eyebrow at making `net` depend on `k8s.io`, but it seems we're already doing that."
+    },
+    {
+      "prediction": "I think this should be `if (countlyGlobal && countlyGlobal.company) {`",
+      "reference": "I think ticket meant, not applying replacement in this method, but rather applying encodeSomeHtml to each and every localization string"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this check. I think it should be in the `catch` block of the `try/catch` below.",
+      "reference": "ticket for the 4.0 epic?"
+    },
+    {
+      "prediction": "I think you can use `cfg.model.neck.pretrained = None` instead of `if cfg.model.neck.rfp_backbone.get('pretrained', False):`",
+      "reference": "`cfg.model.get('neck')` will return None if neck does not exist, thus we can omit the default value."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Trailing whitespace detected."
+    },
+    {
+      "prediction": "I think this should be `CXXFLAGS`, not `CXXFLAGS`.",
+      "reference": "I don't see why a `cxxflags` key is necessary? C flags are important in the target file because they define things like the float ABI. But these flags are also used for C++. I can't think of a reason why you would want to configure C++ flags in a target file."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the same place as the other instrumentation code.",
+      "reference": "Update year range in Copyright notice, and elsewhere too."
+    },
+    {
+      "prediction": "I don't think we need to expose the `SecretKeySelector` here.",
+      "reference": "Can we update this PR to not be a breaking change? i.e. if a user specifies `accessKeyID`, it is still used. But if a user specifies `accessKeyIDSecretRef`, it takes precedence?"
+    },
+    {
+      "prediction": "I think this should be `nil` or `\"\"` instead of `1` or `0`.",
+      "reference": "You don't actually need both tests here as \"truthiness\" means that `\"0\"` is false and hence the first test will be true..."
+    },
+    {
+      "prediction": "Do we need to wait for the page to load before clicking this link?",
+      "reference": "Everything else in here is using `I18n`. Should we do that here to be consistent?"
+    },
+    {
+      "prediction": "/0:tcp:2379,0.0.0.0/0:tcp:2380,0.0.0.0/0:tcp:5473,0.0.0.0/0:tcp:6443,0.0.0.0/0:tcp:6666,0.0.0.0/0:tcp:6667;die-on-fail\"`\n }\n \n // Validate validates the config\n```\nReview comment?",
+      "reference": "What do you think about changing the ordering to be `<protocol>:<cidr>:<port>`? I think it matches the most with what we previously had so that it doesn't change too much for our users."
+    },
+    {
+      "prediction": "I think this should be `static_cast<unsigned int>(rootedAtAtom) < mol.getNumAtoms()`",
+      "reference": "I'm fairly sure this precondition is always true due to the mod (%)"
+    },
+    {
+      "prediction": "I don't think we should be changing the semantics of the `noDups()` method here. If we do, we should do it in a way that doesn't change the semantics of the `noDups()` method. For example, we could change the semantics of the `noDups()` method so that it returns `false` if there are duplicates.",
+      "reference": "I love seeing diffs like this one, adding a `String` message to an otherwise cryptic `assert`! It makes me realize you must have had a hellacious debugging session!"
+    },
+    {
+      "prediction": "I think this should be `if c.version == 0` instead of `if a config file does not have version, it is assumed to be zero. All fields listed in migrate() might be changed if an actual value matches to default value from a previous version.`",
+      "reference": "1. No logger exists to produce a message 2. Can't assume the latest version because missed version means 0."
+    },
+    {
+      "prediction": "Shouldn't this be `sha1($this->hash . ':' . $this->level) . ':' . $this->hash` ?",
+      "reference": "I saw a wrong reuse of the cache between a partial analysis of a single file and a full run where errors due to lack of context on the first partial run was reported on the full run. Shouldn't we use a hash that is composer.lock + psalm.xml + command line to be safe?"
+    },
+    {
+      "prediction": "I think this should be `if not isinstance(self.index, (ks.DatetimeIndex, pd.DatetimeIndex)):`",
+      "reference": "Could you also update it in `at_time`?"
+    },
+    {
+      "prediction": "I think it would be better to use `Boolean.parseBoolean(options.get(\"locality\").orElse(false))` instead of `Boolean.parseBoolean(localityFallback)`",
+      "reference": "Because schema is non-final we can't just switch this to a lambda"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to handle this. I think it would be better to have a separate method for this, something like `ThrowExceptionIfMustBeRethrown(string message, Exception exception)` or something like that.",
+      "reference": "proposal: **always** add the exception to the end of the message. So no need to to pass exception twice (implicit and explicit). Need a lot of changes."
+    },
+    {
+      "prediction": "I think it would be better to move this logic to the `Producer` and `Consumer` classes themselves, so that we don't have to duplicate the logic here.",
+      "reference": "why? event.getInvocationType is not you want?"
+    },
+    {
+      "prediction": "I think we can remove the `ID` and `Version` fields from the `Info` struct. They are no longer used.",
+      "reference": "nit: how about `GitSHA`?"
+    },
+    {
+      "prediction": "I think this should be `FAIL_DIR_NAME` and not `makeflow.failed.%d`",
+      "reference": "If you move the other functions, this should also be moved to `makeflow_gc.h`."
+    },
+    {
+      "prediction": "I think this should be `listen['track_metadata']['additional_info'].get('duration_ms', current_app.config['PLAYING_NOW_MAX_DURATION'])` instead of `listen['track_metadata']['additional_info']['duration'] // 1000`",
+      "reference": "This whole block bugs me. Expire_time to me suggest that an absolute time of when something happens and duration is an interval of time. While the code looks correct, it feels awkward to read."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `coerceToVNode` since it's the only place where `possibleVNode._dom` is set.",
+      "reference": "Can we add `_dom` as an argument to createVNode here? I think it might be shorter (could totally be wrong!)"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now. It's too late to do it now.",
+      "reference": "I think that the iret handling is not yet good."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to pass in the `tabIndex` to the `Tab` component and then set the `tabIndex` of the `TabBar` component to the value of the `tabIndex` of the `Tab` component. This way the `Tab` component will always be the last child of the `TabBar` component and the `TabBar` component will always be the first child of the screen.",
+      "reference": "Huh, that's kinda weird. I get it, but it's unexpected to me..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `Tag` struct itself. The reason is that we don't want this to happen in the `Tag` struct itself. It should only happen in the `Splitter` struct.",
+      "reference": "why is this removed? it is important when you upload from stream of unknown size"
+    },
+    {
+      "prediction": "Style/ClassAndModuleChildren: Use parentheses to call a method defined in a module instead of parentheses to call a method defined in the class.",
+      "reference": "Out of curiosity, why is this needed? Does it enforce what class can be assigned?"
+    },
+    {
+      "prediction": "I think this should be `os.Setgid(os.Getgid())` and `os.Setuid(os.Getuid())` instead of `os.Chmod` and `os.WchMod`.",
+      "reference": "non-blocking: Now that we have a handful of statements which deal with creating a listener, it may make sense to introduce a `createListener` method or something similar"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `LastExecutor` field of the `MevBlockProductionTransactionsExecutor` class.",
+      "reference": "factory should be stateless if possible, looks like much complexity added"
+    },
+    {
+      "prediction": "I think this should be `project_id` instead of `project_id`.",
+      "reference": "Could you please add short description for the default value as well?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to log the error. I think it's better to log the error in the `CommitStateChanges` function.",
+      "reference": "Leave a TODO message to say we will fix the log level in the future, once committing the block and the state become a transaction"
+    },
+    {
+      "prediction": "I think this should be `write.parquet.row-group-check.max-record-count`",
+      "reference": "this can be integer"
+    },
+    {
+      "prediction": "I think this should be `font: {{ font['completion.category'] }};` and not `font: {{ font['completion.category.border.top'] }};`",
+      "reference": "nitpick: Please remove the blank line here"
+    },
+    {
+      "prediction": "Style/InverseOf: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "What is this for? The only methods I can see by that name in the rails doc are generating a foreign key name from a model class name but you seem to be giving a key name as argument?"
+    },
+    {
+      "prediction": "I think it would be better to move this logic into the `AuthenticationPlugin` and `PkiAuthenticationPlugin` classes. The `AuthenticationPlugin` and `PkiAuthenticationPlugin` classes should be responsible for initializing their respective metrics. The `AuthenticationPlugin` and `PkiAuthenticationPlugin` classes should also be responsible for registering their respective metrics with the `SolrMetrics` class.",
+      "reference": "Is this second check necessary? we know that just after the plugin was created its metricRegistry is null, it's set only after `initializeMetrics` has been called."
+    },
+    {
+      "prediction": "This should be `return TALKACTION_CONTINUE;` instead of `return TALKACTION_BREAK;`",
+      "reference": "shouldn't this be `return TALKACTION_BREAK;` as the player does not meet the required group access?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to check the block is created by the proposer.",
+      "reference": "does this still work if minerNode1 has already proposed a block before 2 & 3 are stopped?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Reconcile` method of the `Reconciler` interface. The `Reconcile` method is where the actual work is done.",
+      "reference": "I am wondering if we can either make this a feature or have a way to turn it off in production. My concern is that we might be adding more of these kind of code in the future."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the same place as `kdf = kdf.stack(level=1)`, but with the `reset_index` and `reset_index(level=1)` in between.",
+      "reference": "We can just use `.stack()` here? Then I guess we can reuse when supporting multi-index columns."
+    },
+    {
+      "prediction": "I think we can remove the `#if NETSTANDARD2_0` from this file.",
+      "reference": "What are the possibilities here? Since we plan on supporting .net standard 2.0 from day 1, we might as well tackle that now."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. I think it should be something like `Add<T>(in SpanContext context, T value)` or `Add<T>(in SpanReference context, T value)`.",
+      "reference": "should the parameters be called reference or context?"
+    },
+    {
+      "prediction": "I think this should be `in_buffer.byteslice(index, in_buffer.length)` instead of `in_buffer.slice(index, in_buffer.length)`.",
+      "reference": "Do we need to also change `length` here (and below) to `bytesize`? Perhaps we ought to make a copy of `in_buffer` encoded as binary and then the existing algorithm should work."
+    },
+    {
+      "prediction": "I think this should be `url.toString()`, not `url.toDisplayString()`.",
+      "reference": "This should be `toDisplayString()` to not contain e.g. passwords."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Can we just say that having a non-nil `autocomplete_path` implies that autocomplete is enabled?"
+    },
+    {
+      "prediction": "I think this is the only place where `proto` is used in this PR.",
+      "reference": "Note that the package github.com/golang/protobuf/proto is deprecated. We're instructed to use the \"google.golang.org/protobuf/proto\" package instead. I didn't want to change it now to avoid intrusive changes whose consequences are, currently, unknown."
+    },
+    {
+      "prediction": "I don't think we should be removing these",
+      "reference": "Are these gone routes? Do we need to 301 them?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/apis/networking/v1beta1` since it's only for v1beta1.",
+      "reference": "move them to internal group of imports"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "can we rewrite this as: `ignored event '%s' for cstor pool '%s'`"
+    },
+    {
+      "prediction": "I think this belongs to the `ApprovingOfficial` model, not the `ClientRequest` model.",
+      "reference": "not all client data types have an approving official (eg: 18F does not). do we still want to include the relation here?"
+    },
+    {
+      "prediction": "I think this method should be renamed to `findElementsByXPath` or `findElementsByXPathInWebDriver` or something like that.",
+      "reference": "This change should also probably go into the corresponding method of the abstract By class?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to use `req.protocol` and `req.hostname` instead of `req.get('host')` and `req.get('host')`",
+      "reference": "why this remove?"
+    },
+    {
+      "prediction": "I think this should be `FileReaderReader` or `FileReaderReader` to be consistent with the rest of the package.",
+      "reference": "As the test failures indicate, making this change is going to mean changing everything which currently implements this interface."
+    },
+    {
+      "prediction": "I think this should be `http` instead of `http://`",
+      "reference": "Do we want to redirect this to the forum or something in case people have it linked/bookmarked?"
+    },
+    {
+      "prediction": "I think it would be better to use a ternary operator instead of `||` here.",
+      "reference": "@peterzimon came up with this 'NM' piece as a placeholder for New Member initials, without this the screen looks broken. It changes to normal initials calculation once email or name is entered. Lmk if you'd like to put something different here ;)"
+    },
+    {
+      "prediction": "I'm not sure if this is a good or a bad thing, but I think it's a good thing. The `member-subscription` is a bit of a mouthful, and I don't think we should be using it in the UI.",
+      "reference": "@kevinansfield Would be cool if you can take a   if this is the best way to add `subscriptions` info on member model. This uses the transform + separate model definition way which seemed to be the right way from other references in Admin"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this. I'm not sure if we want to suppress the warning in all cases, but also don't want to make it impossible to plot without Matplotlib.",
+      "reference": "This is the common way of dealing with optional dependencies"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `testhelpers.go`",
+      "reference": "nit: we can replace all `kr` in this file with like `src`."
+    },
+    {
+      "prediction": "I think this should be `const mode = readPreference.mode || 0;` instead of `const mode = readPreference.mode;` since `0` is the default value for `typeof readPreference.mode === 'object' && readPreference.mode !== null && typeof readPreference.mode === 'object'`",
+      "reference": "same concern here as above"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `Mongoid::Search::Base` since it's a base class for all search engines, not just Mongoid.",
+      "reference": "Thank you for this, I gather this repairs failures that I've seen in another PR."
+    },
+    {
+      "prediction": "I think it would be better to use `distinct().withColumn(\"file_path\", from_json(\"file_path\", new StructType().add(\"file_path\", new StructType().add(\"file_path\", jsonType(table.getSchema())))))` instead of `distinct().flatMap(...).toDF(\"file_path\").distinct()`",
+      "reference": "This place probably makes sense to me. We can consider exposing an argument to make the dedup step optional (I am not sure it is a good idea but I want to think this through together). The dedup step we are adding is going to trigger a shuffle. Technically, we are fine in the existing expire snapshots action as it does the dedup in `except`. The question is what kind of performance impact deduplicating here will have. We only have duplicates if multiple manifests reference the same files. In `rewrite_manifests` procedure, we rewrite all manifests, meaning we produce a snapshot with new manifests where entries are old data files. Also, there are updates and deletes that may rewrite manifests. I think deduplicating here makes sense to me in all cases. Thoughts, @rdblue @flyrain @RussellSpitzer @karuppayya?"
+    },
+    {
+      "prediction": "I think we need to log the error here as well.",
+      "reference": "can we move the code before deleteNodeFromDB ? this way even if the node crashes after remove config we can still re-run decommission again ?"
+    },
+    {
+      "prediction": "`var_name` is not used in this function. I think it should be removed.",
+      "reference": "Seems the default value of `melt`'s `var_name` at namespace.py should be changed as well."
+    },
+    {
+      "prediction": "I think this is the only test that needs to be updated. The rest of the tests pass locally and I don't see any regressions.",
+      "reference": "I think botocore's serializer should already create the correct headers. However, they are currently not used in `_botocore_parser_integration_test`(line #217). Maybe we could remove the headers here and just use a fallback in the `_botocore_parser_integration_test` (i.e. use the given headers if they are set, otherwise use the headers generated by botocore's serializer)?"
+    },
+    {
+      "prediction": "I think it would be better to use the singular form of the name here instead of the plural one.",
+      "reference": "This code is passing a non localized string when it should be passing in a localized string or it should use the name of the item being deleted."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `AuthProvider` class, since that's where the `authorize` method is.",
+      "reference": "in case you want to rename one of the plugins when instantiating, from options?"
+    },
+    {
+      "prediction": "I think this should be `ctu_func_mapping`",
+      "reference": "This default variable should be moved into a config variable, created by `package_context`, and read from `config\\package_layout.json`."
+    },
+    {
+      "prediction": "I think this test should be more specific and test the case where `visibleCalc.startColumn` is not equal to `visibleCalc.endColumn`.",
+      "reference": "Can I ask you to add a new line after `const` assignment? I believe that this increases the code readability by encapsulating assignment and logic (expecting) blocks."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. This is not the place where we initialize the host space. This is the place where we initialize the device space. The device space should be initialized before the host space is initialized. I think this should be done in the device space initialization function.",
+      "reference": "So the intention is you configure with `-CMAKE_CXX_FLAGS=\"-D KOKKOS_IMPL_TURN_OFF_CUDA_HOST_INIT_CHECK\"`?"
+    },
+    {
+      "prediction": "why is this 1.0 and 0.0?",
+      "reference": "This was a bug."
+    },
+    {
+      "prediction": "I think we can remove the `class` attribute from the `image.html.twig` template as well.",
+      "reference": "Is OK that we don't have some placeholder?"
+    },
+    {
+      "prediction": "I think we should set it to 2.",
+      "reference": "Sectorbuilder behaves differently depending on whether it's given 1 or more threads. It won't seal if only given 1."
+    },
+    {
+      "prediction": "nit: you can use `import com.google.inject.Inject;` instead of `import static azkaban.ServiceProvider.SERVICE_PROVIDER;`",
+      "reference": "Sort of believe We should put MetricManager in constructor parameter given this case."
+    },
+    {
+      "prediction": "I think it would be better to make `sourceId` and `fieldId` fields of type `Integer` instead of `int`. This will allow us to avoid boxing and unboxing when serializing and deserializing the object.",
+      "reference": "Do we need to check `fieldId` is larger than 1000?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to check for acked status. I think it should be in the listener's `onWriterChangeReceivedByAll` method.",
+      "reference": "I think this should be called after the if below (the one for disable positive acks)"
+    },
+    {
+      "prediction": "I think it is better to use `URI` instead of `URI`",
+      "reference": "I'd like to avoid changing this method since it is public and using a URI will probably change behavior for users passing strings (String -> URI -> Path instead of String -> Path)."
+    },
+    {
+      "prediction": "I think this should be done in a separate PR.",
+      "reference": "Should these restore the stashed env after the test runs?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The reason is that `kw_defaults` can contain `None` values, which means that we should not consider them as defaults. Instead, we should check whether a given default node is not `None` before trying to access its `name` attribute. This can be done by checking if the node is not `None` before trying to access its `name` attribute. For example: `for default_node in scope.args.defaults: ... if default_node is not None: ... for default_name_node in default_node.nodes_of_class(nodes.Name): ... if default_name_node is not None: ... if default_name_node.name is not None: ... return True`",
+      "reference": "We're calculating the full list of kwargs here (even if the first element of the list would return True) so we could improve performance by using a generator line 408 in ``for default_node in all_defaults:``."
+    },
+    {
+      "prediction": "I think this should be `accountDeltaCount` instead of `accountDelta`",
+      "reference": "Can ndelta differ depending on when the deltas are compacted? That is, when intermediate updates are dropped?"
+    },
+    {
+      "prediction": "I think it would be better to add a `@since TODO` to the `XPathRule` class.",
+      "reference": "Our checkstyle config likes it better when the `<p>` is before the first word of the next paragraph, and not on a blank line"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be.",
+      "reference": "This looks to be the only place where we return early before the close below, so if the \"double close\" is bothering people, closing manually here (and removing the `defer`) could be an option"
+    },
+    {
+      "prediction": "I think this should be `execSync( 'npm run qunit', { stdio: [ 0, 1, 2 ] } );`",
+      "reference": "This task can probably be deleted since we don't need gulp to run Jest."
+    },
+    {
+      "prediction": "I don't think we need to check for `Errno::ECONNREFUSED` since it's the most commonly used error code.",
+      "reference": "Doesn't this need to include `Errno::ECONNRESET` to fix the issue?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "never heard of a qirector... what's their role again? (also should be 'for a movie')"
+    },
+    {
+      "prediction": "nit: I think it would be better to move this into the `else if` block above.",
+      "reference": "Implement the \"collect all props for one vertex\" logic inside QueryVertexPropsProcessor"
+    },
+    {
+      "prediction": "This should be `break` instead of `result = true;`",
+      "reference": "This is to prevent parsing things like \"1.1\" as \"1\" until we implement it properly."
+    },
+    {
+      "prediction": "I think this should be `runExperiment(..., _getCheckpointParentDir(...))` instead of `runExperiment(..., _getCheckpointParentDir(...))`",
+      "reference": "Can we change `_getCheckpointParentDir` to be a public function explicitly? It seems like something that clients of experiment runner may wish to use."
+    },
+    {
+      "prediction": "I think this should be `self.ConsoleUIACheckBox.SetValue(config.conf[\"UIA\"][\"consoleUIA\"])` and not `self.ConsoleUIACheckBox.defaultValue = self._getDefaultValue([\"UIA\", \"consoleUIA\"])`",
+      "reference": "Why is a restart necessary?"
+    },
+    {
+      "prediction": "I think this should be `if !cfg.External.Enabled() && !cfg.Internal.Enabled() {`",
+      "reference": "Should setting the task runtime also go under the above if condition?"
+    },
+    {
+      "prediction": "I think this should be 100, not 50.",
+      "reference": "SPIRE uses the leadingLowerCase convention for unexported constants"
+    },
+    {
+      "prediction": "I don't think this is needed. The `Sarif` namespace is only used by `SarifReader` and `SarifWriter`, which are only used by `Microsoft.CodeAnalysis.Sarif.Readers`.",
+      "reference": "The newest JSchema code generator orders `System` usings first."
+    },
+    {
+      "prediction": "I think it would be better to move this logic into `IcebergObjectInspector.create` and `IcebergObjectInspector.createCaseInsensitive`",
+      "reference": "I understand that we want to have a possibility to configure the IcebergInputFormat to be case sensitive or case insensitive since it can be used by other MR jobs as well. Do we want to allow the users of Hive to shot themselves on the foot and enable case sensitivity? My first guess would be that we should not use the configuration here, just go with `false`, but if you have some specific use-case in your mind I can be easily convinced"
+    },
+    {
+      "prediction": "I think we can remove the `pool.PoolAddEventHandled = true` line here.",
+      "reference": "Will the above call to cStorPoolAddEventHandler() get executed in the same thread as this line?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it makes sense to me.",
+      "reference": "I don't think alloca instructions are always at the start of the entry block. I think it would be safer to check the entire entry block for alloca instructions, just in case some are not the first instruction. (Technically they can be anywhere in the function but we check in other places that this isn't possible)."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `ifaces` package, but I'm not sure if that's the best place to put it.",
+      "reference": "Why isn't this 4789 (which I think is the Felix default)?"
+    },
+    {
+      "prediction": "I think we should use `FailureImport` instead of `FailureImported` here.",
+      "reference": "Can this be `Import operations failure`? In addition, is this the right time to change the variable names? e.g. `ReasonImportOperationsFailure`, `ReasonImportFailure`, `ReasonImportSuccessful`, `ReasonCreateResourceFailure`, `ReasonImportResourceSuccessful` Naming convention that i have assumed here is as follows: `Reason<Action><Entity/Adjective of the Entity><Sucess/Failure>`"
+    },
+    {
+      "prediction": "I think this should be `rb_define_singleton_method` instead of `rb_define_global_variable`.",
+      "reference": "I suspect this isn't needed because I think `rb_intern` gets marked already."
+    },
+    {
+      "prediction": "I think this should be `log.finest(\"In development mode. Copying required files in case we're using a WebDriver-backed Selenium\");`",
+      "reference": "This change is incorrect: the current log level is correct."
+    },
+    {
+      "prediction": "I don't think this is correct. I think we want VERSION_SELECTIVE_INDEXING to be the highest.",
+      "reference": "Can you add a space before the equal sign? There are a couple other places where spaces are missing in this PR."
+    },
+    {
+      "prediction": "I think it is better to remove the `flow.getUpdateTime() >= updateTime` condition and just use `flow.getUpdateTime()` in `updateList.add(flow.toUpdateObject(updateTime))`",
+      "reference": "Please follow the todo naming convention."
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "These are no longer used and changes should instead be made directly to the files in `deploy/chart/crds`   /assign @JoshVanL /unassign"
+    },
+    {
+      "prediction": "`xy` -> `x`, `xywh` -> `xwh`, `pred_map[..., 2:4]` -> `wh`, `multi_lvl_anchors[i]` -> `anchor`, `self.num_attrib` -> `num_attrib`",
+      "reference": "Is this really necessary to work around this way? The old version works fine with me when converting to ONNX."
+    },
+    {
+      "prediction": "I think this should be `return 0, fmt.Errorf(\"Blob already closed\")`",
+      "reference": "Same as above: don't start error strings with a capital letter, and don't fix unneeded things in an unrelated commit ;)"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Is this a testing debug statement?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to pass the `options` to the `parse` function and have the `parse` function return the connectionStringObject. This way we don't need to pass `options` to the `createUnifiedOptions` function, and we don't need to pass `options` to the `mergeOptions` function.",
+      "reference": "you'll need to wrap the _whole thing_ within this async call (or refactor it so everything below is in a new temporary function). The code below this section depends on a value for `object` after the parsing occurs, but what you're doing here is kicking off a race condition where `object` is null, but _might_ parse quickly enough to assign it the parsed parts. Instead, you need to do everything that depends on the parsed object in the callback of `parse`."
+    },
+    {
+      "prediction": "I think this should be `std::unique_lock<std::mutex, std::logic_error> lk (mutex);`",
+      "reference": "This is now going to be protecting `stopped` when it doesn't need to. If you are going to do this then can you make `stopped` non-atomic and protect it in `stop`. It's cheap to make a `unique_lock` fyi, it's just a pointer to a mutex and a `bool`. The `lk.lock ()` in the loop is superfluous so probably shouldn't have been there to begin with."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Lets not give packages names if we dont have to. If a package requires a rename, we should change the actual name while/if we can."
+    },
+    {
+      "prediction": "I think this should be `TEST(1U << 31, enc) && TEST(1U << 15, enc)` instead of `TEST(1U << 31, enc) && TEST(1U << 15, enc)`",
+      "reference": "Not sure I see where this is coming from. This is a bug fix? Should be mentioned in the commit message? Same below."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this cleanup. I think it should be in the `afterEach` callback of the `cleanupAfterSuite` function.",
+      "reference": "I reverted this change in the last commit, I think we should revert it here. I know this leads to confusing test output for skipped tests, but I would prefer to tackle that in a separate PR specifically focused on that problem. In fact, I think the issue might really be with the leak checker, and not how this test runner is defined. Please make a ticket for this (if it doesn't already exist) and revert this change"
+    },
+    {
+      "prediction": "I don't think we need to add normed_bbox to the public API.",
+      "reference": "Can this part be done in `dataset.get_ann_info`? Just like CocoDataset converts xywh to xyxy in `_parse_ann_info`"
+    },
+    {
+      "prediction": "I think this is the only change that should be in this file. The rest of the changes are unrelated to this PR.",
+      "reference": "Should we change the name of this test struct to be `testTracerProvider` as well?"
+    },
+    {
+      "prediction": "I think we can remove the `iotexapi` package and use `github.com/iotexproject/iotex-api` instead.",
+      "reference": "\"google.golang.org/grpc\" should be together with \"go.uber.org/zap\" all \"\"github.com/iotexproject/iotex-core/*\" stay together"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in `WebEngineTab.__init__`, since that's the place where the WebEngine is initialized.",
+      "reference": "No need for a separate variable here - just call `utils.read_file` directly below."
+    },
+    {
+      "prediction": "I think this should be a warning, not an error. I don't think we want to log every message that gets sent to the anon channel.",
+      "reference": "Thanks for the fixing, small question, Here the message just be written to the channel, will it be blocked?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right name for this. I think it's more accurate to call it `ReconstructedBlock` or `ReconstructedBlockHash` or something like that.",
+      "reference": "maybe we will have something like `Bodies []byte(\"bodies\")` instead of them being a number? with numbers we are making it harder to insert stages mid-process."
+    },
+    {
+      "prediction": "I think we can remove the `Pod` field from the `GroupMember` struct and the `PodReference` field from the `ExternalEntityReference` struct. The `Pod` field is only used in the `PodReference` field, and the `PodReference` field is only used in the `ExternalEntityReference` field, so we can remove the `Pod` field from the `GroupMember` struct and the `PodReference` field from the `ExternalEntityReference` struct. The `PodReference` and `ExternalEntityReference` structs are only used in the `GroupMember` struct, so we can remove the `PodReference` and `ExternalEntityReference` structs from the `GroupMember` struct. The `PodReference` and `ExternalEntityReference` structs are only used in the `GroupMember` struct, so we can remove the `PodReference` and `ExternalEntityReference` structs from the `GroupMember` struct. The `PodReference` and `ExternalEntityReference` structs are only used in the `GroupMember` struct, so we can remove the `PodReference` and `ExternalEntityReference` structs from the `GroupMember` struct. The `PodReference` and `ExternalEntityReference` structs are only used",
+      "reference": "out of curiosity, why making this change? If no real reason, could you restore to avoid unnecessary changes of several generated files?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `manager.Connect()` function.",
+      "reference": "Could we public events if `manager.serStatus()`?"
+    },
+    {
+      "prediction": "I think this should be `Please make sure that the element exists on the page, or that the plugin you are targeting has not been installed. Check that the <script> tag initializing Uppy comes at the bottom of the page, before the </body> (see https://github.com/transloadit/uppy/issues/1042).`",
+      "reference": "could be misread as `<body>`, maybe something redundant but very clear like \"before the closing `</body>` tag\"?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that the response body is closed after it's been written to the response stream. Otherwise, the response body will remain open and leak memory.",
+      "reference": "This test is specifically checking behavior on sync I/O (note there are tests with Write in the name, and other tests with WriteAsync). Enable sync I/O here."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to go about it, but it seemed like a good way to get the preview working.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "Can we rename this to `CstVolumeClaimSpec` or `CSTVolumeClaimSpec`?",
+      "reference": "nit: Should we have something like `volumeName@snapShotname` and this will be filled only if it is a clone creation"
+    },
+    {
+      "prediction": "This should be `if err = os.MkdirAll(dir, 0755); err != nil {`",
+      "reference": "This seems like an improved pattern :)"
+    },
+    {
+      "prediction": "I think this should be `configuration.get_config().get(\"hadoop\", \"version\", default=\"cdh4\").lower()`",
+      "reference": "I think it would be better to have a separate configuration option for specifying which HDFS interaction layer to use instead of using the existing hadoop version option. I think there are still cases where things might have to rely on Hadoop version for other things, and overwriting that information with \"snakebite\" would make that difficult. The existing hadoop version option could still be used to distinguish the different HDFS CLI flavours."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Why are we filtering content twice? It seems like just filtering it in PAL should work."
+    },
+    {
+      "prediction": "I think we should rename this to `ACCOUNT_STATUS_APPROVED_MULTIPLE` or something similar.",
+      "reference": "These should probably move to `constants.js`"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Again, you don't want to depend on glassfish."
+    },
+    {
+      "prediction": "This should be `Faker::Space.launch_vehicule` not `Faker::Space.launch_vehicule, 2019, 06`",
+      "reference": "it should be `Faker::Space.launch_vehicle`, otherwise it will crash"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to clean up old archives. I think it would be better to check if the file is older than the archive period and then delete the old archives.",
+      "reference": "Isn't it quite expensive to check whether the file exists on every write?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for CORS. I think it would be better to check for `url[0]` being a `/` or `window.location.protocol + '//' + window.location.host'` being the same as `origin + url`.",
+      "reference": "Please hide this behind a flag and add a setter function to the builder."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the backup API, since it's related to the backup functionality.",
+      "reference": "old api should be deprecated if we don't intend on using it anymore. Again this should happen through swagger api and then removed from next release. We should also release not this as a breaking change. (same with all the other cloudsnap api renames."
+    },
+    {
+      "prediction": "I don't think this is the right place to log this. I think it should be in the Vert.x logs.",
+      "reference": "so strange...... canTcpListen is to detect, if this transport can init when customers deployed vertx and servlet rest transport, we must detect which one will be used only both of them refused to init is a problem. we have already print that......"
+    },
+    {
+      "prediction": "I think this should be `SELECT COUNT(*) FROM %s`",
+      "reference": "Did you mean for this to be `FORM`, instead of `FROM`?"
+    },
+    {
+      "prediction": "I think this should be `Optional` as well.",
+      "reference": "What are your thoughts on the long term solution for the naming of this variable? Do you think we should revert the logic when we are ready to make the next set of breaking API changes?"
+    },
+    {
+      "prediction": "I think this should be `$stderr.puts(\"#{command} failed\")` and not `$stderr.puts failure_message`",
+      "reference": "There's a syntax error on this line."
+    },
+    {
+      "prediction": "This should be a `goto` out; it's not the end of the function.",
+      "reference": "You lose the value of `result` when you `goto out`"
+    },
+    {
+      "prediction": "I think this should be `this.ratio_` instead of `ratio`.",
+      "reference": "Just a style thing, why not `this.ratio_ || 0`?"
+    },
+    {
+      "prediction": "I think this should be `CIDRFromIPNet` instead of `CIDRFromIPNet`",
+      "reference": "I think the stdlib's `net.IP` has a `Mask()` method, does that do what we need?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should go in `runtime/structs.go`",
+      "reference": "Why does substitute need to be public? Are we allowing/encouraging variable usage when using Daisy as a library?"
+    },
+    {
+      "prediction": "I think this should be `if js_elem == \"focused\" or \"selected\"`",
+      "reference": "We should probably use ctrl-enter if `tab` is given, just like in `follow_selected`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not sure if it's the right dispatcher to use, but it's the only one I could find. I'm not",
+      "reference": "This file looks like anti pattern `ServiceLocator` and lucky it's not used anywhere. Can I remove it? (in separate PR)."
+    },
+    {
+      "prediction": "I think this should be `bytesToHumanReadable` instead of `getDownloadSizeNumeric`",
+      "reference": "we can just have this method call the next, passing in the file metadatas (deduplication of logic)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `tabbed_browser` module.",
+      "reference": "What happens here if you send the `wheelEvent` directly to the `tabbed_browser` and let it take care of propagating it to its children (i.e. the focused tab)? If that works, I guess that'd be cleaner."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. It should be done in the native code.",
+      "reference": "it seems python 2 don't support `end=''`. @StrikerRUS @henry0312 any better solution? or we drop the support of python 2?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to check this, but it seems to be the only place I could think of.",
+      "reference": "This test already exists in the object_tracker layer."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `EthGetCode` class itself.",
+      "reference": "Same here, can we presume if `privacyParameters` is set we need goQuorum handling? Or is there a goQuorum sub option in `privacyParametes`?"
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "The addition of delete files in `newManifests` far above the addition of data files threw me a bit. Is it intentional to ensure the delete files are at the front of the list?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this validation. I think it should be in the `resolve_top_level_references` method, since that's where the actual validation happens.",
+      "reference": "Interesting. The call to `Array(groups)` below this would seem to indicate this doesn't _have_ to be an array. I think the problem is actually that passing a hash to `Array()` actually converts the hash to an array of key/value pairs rather than wrapping it in an array. So maybe `Array(groups)` should be `groups = [groups] unless groups.is_a?(Array)`."
+    },
+    {
+      "prediction": "I think this should be `history.expire.retain-min-num-days` to be consistent with other properties.",
+      "reference": "This is not a particularly good default but we need it to keep the old behavior."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/testutil.go` since it's only testing the `applySignatureMessage` function and not the whole `PaymentBroker` class.",
+      "reference": "We have other tests to assert all properties of a condition are valid. It's enough to assert the condition is not nil here."
+    },
+    {
+      "prediction": "I think we can remove the `opsv1alpha1` package now.",
+      "reference": "shouldn't the import be renamed `crdv1alpha1` by convention? It seems that you took care of renaming everywhere else..."
+    },
+    {
+      "prediction": "I think this should be `PauseContainerImageName` and not `PauseContainerTag`",
+      "reference": "nit: Can Linux-specific paths be moved to a conditionally-compiled file for Linux?"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it would be better to put it in `SessionMapImpl` since that is where the implementation of this method is.",
+      "reference": "This is not the right approach. The `Distributor` maintains a model of the current state of the Grid. That model already contains the information about every active session. We don't need to modify `SessionMap` to expose it further."
+    },
+    {
+      "prediction": "I think this should be `errors.New(\"Lists for non-fetches unsupported for now\")`",
+      "reference": "Git doc seems to suggest `for-push` is a optimization. Do we have a plan/ticket for treating them differently?"
+    },
+    {
+      "prediction": "I don't think we need to add `ltp` to the `contribs` list.",
+      "reference": "This not necessary since you moved the layers to `contrib/`."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "I think this also needs to be removed from `resources/bolt_bash_completion.sh`, `rakelib/pwsh.rake`, `lib/bolt_spec/plans.rb`, and `lib/bolt/config/options.rb`."
+    },
+    {
+      "prediction": "I think this test is still a bit out of date, as it doesn't test that the email address in the session matches the one in the request. I think we want to make sure that the email in the session matches the one in the request.",
+      "reference": "I fixed this today with a direct push to master. My build was failing otherwise."
+    },
+    {
+      "prediction": "I think it would be better to have a `@Rule` annotation instead of this implementation.",
+      "reference": "not sure why this is in a \"rule\" child package from the interface..."
+    },
+    {
+      "prediction": "I think this should be `icon` and not `icon:`.",
+      "reference": "Not needed, please remove."
+    },
+    {
+      "prediction": "I think this should be `if (!h2o_lcstris(loc_parsed.host.base, loc_parsed.host.len, match->host.base, match->host.len, match->host_is_unix_path))`",
+      "reference": "Shouldn't the logic be: * the result is a mismatch if either is a unix path but the other is not * if both are unix paths, then perform case-sensitive comparison * if both are not unix paths, then perform case-insensitive comparison"
+    },
+    {
+      "prediction": "This should be removed.",
+      "reference": "This is just rearranging it alphabetically"
+    },
+    {
+      "prediction": "I think this should be `too-many-arguments` instead of `too-many-locals` to be consistent with the rest of the tests.",
+      "reference": "I think this disables it for the rest of the file - you should be able to use it inside of the function instead."
+    },
+    {
+      "prediction": "Shouldn't this be `MediaInterface|mixed $media`?",
+      "reference": "this could receive the MediaInterface directly or an identifier, it is correct to type it with int|string? or just use mixed?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. It should be moved to the `MultiLeafFieldComparator` class. The `MultiLeafFieldComparator` class is designed to compare fields within a multi-leaf field, not individual leaf fields. The `MultiLeafFieldComparator` compares the values of the fields within the multi-leaf field, not the values of the individual leaf fields.",
+      "reference": "Why not return `Object[]`?"
+    },
+    {
+      "prediction": "I don't think we should have a `sequenceNumber` field here. It's not a sequence number in the sense that it's not a sequence of rows, it's a sequence of manifest files.",
+      "reference": "Looks like we are trying to match the new ordering of fields in `ManifestFile`. Earlier, we co-located `...FilesCount` with `...RowsCount` to match the ordering of methods in `ManifestFile` and args in constructors. Is this change intentional?"
+    },
+    {
+      "prediction": "This should be `summon->setLossSkill(true);` instead of `summon->setLossSkill(false);`",
+      "reference": "You're already setting drop loot to false just above."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure if it's the right place for it. I think it should be in the middleware, but I'm not sure",
+      "reference": "`once` is very generic and doesn't specify it's purpose - `closeOnce`?"
+    },
+    {
+      "prediction": "I think we can remove this method and use `HiveMetastoreTest`'s `getTableLocationPath` instead.",
+      "reference": "Why did this need to change from `private`? Oh for the test below. And this is only a test class too."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Should test the `has?` method here"
+    },
+    {
+      "prediction": "`$this->clickCss($page, '.fa.fa-save');` should be `$this->clickCss($page, 'a.fa.fa-save');` and `$saveLink = $link;` should be `$saveLink = $links[0];`",
+      "reference": "I think you want to be asserting/clicking $saveLink, not $link."
+    },
+    {
+      "prediction": "I think this should be `Optional[str|List[str]]`",
+      "reference": "We don't have `target_name` on `@sensor` - why introduce it for `@asset_sensor`? If we do want to introduce it everywhere, can we just call it `job_name` to avoid introducing a new term?"
+    },
+    {
+      "prediction": "I think this should be `FeedSettings` instead of `SleetSettings`",
+      "reference": "Missing copyright header (+other new files)."
+    },
+    {
+      "prediction": "I think we can remove the `TimestampType` class altogether. It's not used anywhere.",
+      "reference": "Could we do the flag check here as well to check whether or not we have enabled the \"Handle without timezone\" flag here as well? We may be using this not on the read path (like in the migrate/snapshot code) and it would be good to catch it here as well and make sure users know what is happening."
+    },
+    {
+      "prediction": "I think this should be `from dagster.core.definitions.pipeline_failure_sensor import PipelineFailureSensorContext`",
+      "reference": "why is the file named `pipeline_definition_definition_sensor`?"
+    },
+    {
+      "prediction": "I think this should be `if not conf[\"SQLALCHEMY_TIMESCALE_URI\"]`",
+      "reference": "None not needed, can just be `return`. not important for this release"
+    },
+    {
+      "prediction": "I think we can remove the `target` input now, since it's no longer used.",
+      "reference": "I think it's missing closing `</p>` tags in line 108 and 109 unless they don't matter..."
+    },
+    {
+      "prediction": "I think this should be `enableScriptEngineLoadExternalResources = false;`",
+      "reference": "So it seems that loading a script file from the classpath needs to be enabled explicitly now since this is `false` by default. Is it possible to enable this via the YAML config file in a Camunda Platform Run distribution?"
+    },
+    {
+      "prediction": "I think we can remove the `instagram` and `instagram/graph` from the package.json as they are not used anymore.",
+      "reference": "can we delete any code in the provider implementation? i imagine the entire thing is unused now?"
+    },
+    {
+      "prediction": "This function is not used anymore, can you remove it?",
+      "reference": "I suggest we remove all these unused functions. I don't expect any external projects to be importing packages from `cert-manager/test/...`"
+    },
+    {
+      "prediction": "I think you can remove the `continue` here as well.",
+      "reference": "Looks like the template argument doesn't resolve by default so you need `std::lock_guard<std::mutex> lck`"
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "reference": "Are we removing this public method? Could it be useful to someone? Maybe we could implement it once in Node using the metamodel"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `PlansController`, but I'm not sure.",
+      "reference": "I think this method is unused now?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check if we need to start the network policy status controller.",
+      "reference": "I know we discussed this before and decided to reuse the AntreaPolicy gate. But do you think realization status will introduce much overhead or not?"
+    },
+    {
+      "prediction": "I think we can remove the `RunController` and `RunEdgecore` constants, since they are no longer used.",
+      "reference": "please add a blank line at the end of this file. Applicable to similar cases below"
+    },
+    {
+      "prediction": "Shouldn't this be `if (pendingConfiguration.remoteHost != null && pendingConfiguration.remoteHost.length() > 0)`?",
+      "reference": "This is not really needed, the line added in `GridNodeConfiguration.java` is what really fixes the `remoteHost` regression."
+    },
+    {
+      "prediction": "I think we should add a similar message to the synchronous filter cache case as well.",
+      "reference": "filterCache can be null"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be moved to `core` package.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be `return SET_PARAM_CMD_FAIL`",
+      "reference": "If strstr returns NULL, the check on 378 won't fire, because param_name will be at least SET_PARAM_CMD_LEN."
+    },
+    {
+      "prediction": "I think this should be `sum_singlesubgroup` instead of `sum_singlepass`.",
+      "reference": "Does this change affect the performance of other algorithms, except KMeans?"
+    },
+    {
+      "prediction": "I'm not sure why this is necessary, but it seems like it should be.",
+      "reference": "What do these changes have to do with the request ID?"
+    },
+    {
+      "prediction": "I think this should be Kokkos::DefaultExecutionSpace, not Kokkos::HostSpace.",
+      "reference": "I'm kind of surprised this doesn't require `typename`?"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "nit: move this one down."
+    },
+    {
+      "prediction": "I don't think this is the right name for this. I think it should be `GroupedBoxWhisker` or `GroupedBoxWhiskerChart`.",
+      "reference": "maybe 'standard Tukey boxplot definition' if it is standard? Otherwise sounds like it is just *a* definition for boxplots..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `gitlab-ci.yml` under the `env` section.",
+      "reference": "I feel we should not fix this."
+    },
+    {
+      "prediction": "Can you move these headers to the top of the file, so they are sorted from newest to oldest?",
+      "reference": "I think these are not necessary"
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "Please update the user guide too."
+    },
+    {
+      "prediction": "I think this should be `$name = $location->discoveryDisplayName ?? $location->name;` instead of `$name = $location->discoveryDisplayName ?? $location->name; $code = $location->code; $locationMap[$location->id] = [$name, $code];`",
+      "reference": "If you use `compact('name', 'code')` here, you'll get an associative array, which might make the rest of the code more readable (instead of using hard-coded 0/1 indexes)."
+    },
+    {
+      "prediction": "I don't think this is a good idea.",
+      "reference": "Should it be 2011-14 instead ;-)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to set the max idle connections.",
+      "reference": "I want to write some type of test for this but there are not Getter methods to assert this value. I tried to extract the value via the stats that are exposed without luck."
+    },
+    {
+      "prediction": "I think this should be `gsuite_id (str): Gsuite id.`",
+      "reference": "This needs to take the group id, not the gsuite id."
+    },
+    {
+      "prediction": "I think we can remove the `beta.5` part from the version string as well.",
+      "reference": "These should use the constants defined in JsonTests.cs"
+    },
+    {
+      "prediction": "I'm not sure if we want to remove the `error` variable from the `getDataErrorComponent` call above.",
+      "reference": "Kind of unrelated to this PR, but let's update this to `getDataErrorComponent( 'search-console', error.message, false, false, false, error )` so that everything is passed as expected."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this function. It's not related to loading an image from a file. I think it should be moved to `mmdet/dataset/__init__.py`.",
+      "reference": ":obj:\\`dataset\\` cannot be correctly rendered"
+    },
+    {
+      "prediction": "I think this should be `user_id` instead of `document_id`",
+      "reference": "I think `email` is something we get from devise, and I think the goal of putting it in the generator was not to tie others to that particular implementation?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to make sure that we don't process more than `MaxExportBatchSize-1` items at a time. Otherwise, we could run out of memory if we have too many items in the queue.",
+      "reference": "If the span producers pushing to the queue do so faster than the this can drain, it will cause this to hang. I'm guessing we can update the `enqueue` method to check if the `stopCh` is closed and not send any more spans while this flushes what has already been pushed."
+    },
+    {
+      "prediction": "This should be `iterableCollection` instead of `Collection`",
+      "reference": "Not sure about that, an array is iterable right? But if I pass array, that clear method wont work"
+    },
+    {
+      "prediction": "Why is this file not part of this PR?",
+      "reference": "make sure the config changes stuck. are there any invalid config values? maybe write tests around empty `trust_domain`, negative/missing `ttl`, invalid `key_size`, etc..."
+    },
+    {
+      "prediction": "I think this should be `window.__mochaResult__` instead of `window.__mochaResult__`",
+      "reference": "Why are you changing this?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Layout/SpaceInsideStringLiteralBraces: Space between { and | missing.<br>Layout/SpaceInsideStringLiteralBraces: Space missing inside }.",
+      "reference": "@phirefly Can we look at a list of RWAs, or ask someone to double-check that this format is correct? Otherwise :shipit:"
+    },
+    {
+      "prediction": "I think it is better to use `SparkWriteOptions.DISTRIBUTION_MODE` here.",
+      "reference": "Still request a local sort for bin-packing based on the defined table sort order."
+    },
+    {
+      "prediction": "I don't think we need to pass users to this function, since it's only used in `request_model` and `request_candidate_sets`, which don't need to know about users.",
+      "reference": "As with the other PR, user name is better."
+    },
+    {
+      "prediction": "I think it would be better to use `s.chequebook.AvailableBalance` instead of `s.chequebook.AvailableBalance.Set(float64(balance.Int64()))`",
+      "reference": "a peer's accounting lock is held during `Pay`. we should avoid adding additional blockchain calls here if possible."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "Maybe wording could be improved. Is this \"not enabled\" or it is \"not available\". Not sure about that."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `TaskReconciler` since it's the only one that has access to the `node` object.",
+      "reference": "I looked at the new `conditionalBranchesEvaluated` function and it looks like the part added is a duplicate of the line above?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this information. I think it should be in the store, but I'm not sure how to get the information from the store.",
+      "reference": "can you put this flag in `blockRefInfo` instead? It would be a shame to add one more file per block, especially since we've run into inode limits. I guess `blockRefInfo` should maybe then be renamed to `blockInfo` or something. But we're stuck with the filename."
+    },
+    {
+      "prediction": "I don't think this is the right way to fix the issue. We should not be setting the IP and port if the IP is nil.",
+      "reference": "Shouldn't we keep other parameters if only IP is empty? Or maybe return an error if it's a mandatory argument?"
+    },
+    {
+      "prediction": "I think this should be `self.assertEqual(redimmed.kdims, [Dimension('Integer', type=int), Dimension('floatdim', type=float)])`",
+      "reference": "Was this just wrong before? The names indicated types but type wasn't specified. I guess the tests passed as comparison worked with ``type=None``?"
+    },
+    {
+      "prediction": "I think we can remove `icebergColOpt.get()` and `effectiveMetricsConfig.columnMode(icebergCol.name())` as they are not used anymore.",
+      "reference": "icebergCol.name() is the unqualified column name"
+    },
+    {
+      "prediction": "I think this should be `dimensions, alarm_actions, insufficient_data_actions, ok_actions`",
+      "reference": "It's generally a bad idea to use mutable types like lists as default values for parameters. Lots of strange, difficult to debug side effects can occur. I see that there was already one example of this prior to this commit which probably explains why it seemed innocuous to add more but I'm going to rework this before committing to master."
+    },
+    {
+      "prediction": "I think this should be `eslint-config-prettier@types/eslint-plugin-prettier`",
+      "reference": "We run prettier as an eslint plugin, so this harmful"
+    },
+    {
+      "prediction": "Why is this `DynamicType_ptr` instead of just `DynamicType`?",
+      "reference": "Check if the TypeDescriptor and MemberDescriptor includes are necessary"
+    },
+    {
+      "prediction": "Can you remove this empty line?",
+      "reference": "delete the empty line here."
+    },
+    {
+      "prediction": "Layout/AlignParameters: Align the parameters of a method call if they span more than one line.",
+      "reference": "Line is too long. [89/80]"
+    },
+    {
+      "prediction": "I think we can remove the `protected` modifier from `getUrlMapper` and `setUrlMapper` methods as well.",
+      "reference": "You don't need to expose this method to do what you want. There are already public addNewGetMapping, addNewPostMapping, and addNewDeleteMapping methods."
+    },
+    {
+      "prediction": "Nit: I think this should be renamed to `HeadResponseCanNotContainContentLengthHeaderButBodyNotWritten()`.",
+      "reference": "This test passes in `dev`. Why wouldn't this work?"
+    },
+    {
+      "prediction": "Nit: You don't need to start with a capital letter. Also, you don't need to add a `#` at the end of the snippet.",
+      "reference": "ARN -> Amazon Resource Number (ARN)"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "Why not use the overload that takes an instance?"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "`Blockchian` is a misspelling of `Blockchain` (from `misspell`)"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "`agent/gpu/nvidia_gpu_manager_unix.go:47: struct field lock has json tag but is not exported`"
+    },
+    {
+      "prediction": "I think this should be `TEST_ARN = 'arn:aws:sqs:eu-west-1:000000000000:testq'` instead of `TEST_ARN = 'arn:aws:sqs:eu-west-1:000000000000:testq'`",
+      "reference": "nit: better rename to `TEST_QUEUE_ARN` or `TEST_EVENT_SOURCE_ARN`"
+    },
+    {
+      "prediction": "Why do we need both `p.Spec.ServiceMonitorSelector != nil` and `p.Spec.AdditionalScrapeConfigs != nil`?",
+      "reference": "Would this mean that podMonitorSelectors suffer from the same issue if they are the only configuration set?"
+    },
+    {
+      "prediction": "I think this should be `dev` instead of `fake`",
+      "reference": "`ctx` is unused in GetEnvironment"
+    },
+    {
+      "prediction": "This should be called `OK` instead of `OKDialog`",
+      "reference": "We usually use OkDialog()"
+    },
+    {
+      "prediction": "I think this should be `--enable-unrestricted-privacy` and `--disable-unrestricted-privacy` instead of `--privacy-unrestricted-enabled` and `--unrestricted-privacy-enabled`",
+      "reference": "could we please change that to something that indicates that this feature is not \"production\" ready!"
+    },
+    {
+      "prediction": "I don't think you need to add `ms-crop-3x_coco.py` here. You can use the `mask_rcnn_swim-t-p4-w7_fpn_fp16_ms-crop-3x_coco.py` file instead.",
+      "reference": "swim -> swin. Other configs and file names also should be modified."
+    },
+    {
+      "prediction": "I think we should rename this method to `get_project_ids` or `get_project_ids_from_config` since it's not just for bigquery.",
+      "reference": "nit: arg description for \"key\"?"
+    },
+    {
+      "prediction": "I think we can remove the `Fsync` method from the `TLF` interface, since it's now implemented in the `TLF` struct.",
+      "reference": "If we don't have a `dir` yet, we wouldn't need to do a sync right? If so, perhaps we can just `getStoredDir()` like `Attr()`?"
+    },
+    {
+      "prediction": "I think this should be `IF_HOST_X86_ELSE(..., ...)` instead of `IF_X86_ELSE(..., ...)`.",
+      "reference": "I still don't seem to fully understand this. Why are we testing the host if DR_HOST_NOT_TARGET is not set?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move this test to `client_test.go`",
+      "reference": "in the current implementation, both schemes are supported (you can call either `/bzz` or `/v1/bzz`). i would suggest to keep it this way"
+    },
+    {
+      "prediction": "I think we can remove the `reason` field from the `assembly` object since we don't care about it anymore.",
+      "reference": "hmm, I think we can just do `new Error(assembly.error)` and that should set `error.message` correctly too. I don't know why it was done this way with a separate `.message` assignment before :sweat_smile: Should we do `error.assembly = assembly` so the template editor can access it that way, rather than parsing `error.details`?"
+    },
+    {
+      "prediction": "I think this should be `if (trim($this->pageAccessRootline) !== \"\") {` and not `if (trim($access) !== \"\") {`",
+      "reference": "This should never be empty. The access field always needs a value of at least `c:0` or `r:0`"
+    },
+    {
+      "prediction": "I think this should be `return parser.ParseFile(f, src), nil`",
+      "reference": "Another question is, do we want to call `os.Getenv()` every time we parse a file? That seems a lot."
+    },
+    {
+      "prediction": "`append` and `prepend` seem to be the only methods on `Tuple` that don't have a `@Override` annotation. Is there a reason for this?",
+      "reference": "minor: please rename all occurrences of `A` to `T`. please also rename `v` to `value`."
+    },
+    {
+      "prediction": "I think this should be `public int FastSyncLag => (Current & SyncMode.Beam) == SyncMode.BeamSyncFastSyncLag ? SyncModeSelectorConstants.BeamSyncFastSyncLag : SyncModeSelectorConstants.NotBeamSyncFastSyncLag;`",
+      "reference": "Do we want it to be so dynamic? My first idea was to base it on SyncConfig.BeamSync ."
+    },
+    {
+      "prediction": "I think this should be a `Map<Integer, ConnectionType>`.",
+      "reference": "can you change this instead to just `type.toString()` and then you wouldn't have to expose the getBitMask in the enum. (Alternatively you could have used `type.hashCode()` but that doesn't feel as nice)"
+    },
+    {
+      "prediction": "why do we need to check the number of patches?",
+      "reference": "I'm confused by these tests. Shouldn't the template string `@importance_type@` have been replaced with the actual value of `importance_type`?"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think it would be better to have a `NewHibernatingKeeper` function that takes the secret key as an argument and returns a hibernating keeper that can be used to communicate with the outside world.",
+      "reference": "Let's keep using NewKeeper so that it got covered by tests."
+    },
+    {
+      "prediction": "I think we should move this to the `org.apache.solr` package. It's not part of the core solr API.",
+      "reference": "I only see new imports. Is there any functional change in this file?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `createMemorialTransfer` function.",
+      "reference": "ineffectual assignment to `err` (from `ineffassign`)"
+    },
+    {
+      "prediction": "I think this should be `service = getDefaultService() || service.start();`",
+      "reference": "Would you mind removing the default parameter above? (I doubt I'll ever use defaults again since you still have to protect against callers explicitly passing `null` or `undefined`)"
+    },
+    {
+      "prediction": "I don't think we need to warn about this, since it's only used in one place.",
+      "reference": "Make sure the old logging functions still work, but add deprecation warnings (unfortunately Python suppresses deprecation warnings by default, so I'm not sure if anyone'll notice)."
+    },
+    {
+      "prediction": "I don't think this is the right place to enable telemetry. I think it should be in the `ManagementServiceTest` class.",
+      "reference": "Not directly related to the topic of this ticket: I think this API design is a bit confusing. To disable delemetry, I would write `managementService.enableTelemetry(false)` which is not intuitive to read. Maybe `toggleTelemetry` instead of `enableTelemetry` is more clear."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to capture the annotations. I think it should be in the `CaptureAnnotatedCodeLocationCollections` method above.",
+      "reference": "Again I suggest returning `AnnotatedCodeLocation[][]`."
+    },
+    {
+      "prediction": "I think we should be able to get rid of the `v1alpha1` package here.",
+      "reference": "We should probably require an increased verbosity to output for anything below the cluster itself to avoid spamming the logs. It would also be good to give additional context as to what we are attempting to reconcile since the controller has multiple workers."
+    },
+    {
+      "prediction": "I don't think we should have this in the OptionsNetwork struct.",
+      "reference": "Not needed anymore"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `TraceListener` class, since it's related to tracing.",
+      "reference": "Am I right in thinking this completely avoids the situation where `enable_calltarget_state_by_ref` is `true`, but the managed integrations don't expect a by ref argument? For example, if there's an exception here, that seems like a fatal problem, as we would have a mismatch for jit rewriting? We should disable instrumentation entirely in that case shouldn't we? i.e. if this (or the existing `EnableByRefInstrumentation`) calls fail, should we bail out of this method entirely as we can't safely instrument?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/oc/resource/types.go`",
+      "reference": "nit: spaces seem off"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I'm not sure how to fix it either.",
+      "reference": "shouldn't the em parameter be called ec?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix, but it's the only thing I can come up with that doesn't break anything.",
+      "reference": "Please change this to use o.DidSign(o) as implements this logic without performing any of these checks that duplicate something that JWT can do correctly. If the JWT deserialized properly, and DidSign returns true, the JWT was signed by one of the listed operator keys."
+    },
+    {
+      "prediction": "I think this should be `return ((FeedItem) o2.getComponent()).getPubDate().compareTo(o1.getComponent().getPubDate());`",
+      "reference": "Talking about style, IMO extracting either `getComponent()` or `getPubDate()` to variables would make this line easier to read."
+    },
+    {
+      "prediction": "I think it would be better to call this something other than `Option` since it's not an option, but rather a function that can be used to customize the task. Something like `Fn` or `FnFunc` or something like that.",
+      "reference": "package header missing"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `componentDidMount` callback, since that's where the component is created.",
+      "reference": "Just a thought (doesn't change the output or anything) - do you think we'll ever want to access the other properties of `oldVNode` from within `_childDidSuspend`? If so we could pass `oldVNode` here and then check these properties in the _childDidSuspend implementation. I don't have strong feelings either way, just occurred to me."
+    },
+    {
+      "prediction": "Why not using the parent generateId() method?",
+      "reference": "We should be configuring the base paths somewhere.."
+    },
+    {
+      "prediction": "Why do you need to set `min_sizes` and `max_sizes` at the same time?",
+      "reference": "May also need to test the normal functionality with min/max_sizes"
+    },
+    {
+      "prediction": "I think this should be `https://mystanet.mysterium.network/`",
+      "reference": "We can remove this one. It's just a random one I got for free for testing."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `AddFile` component itself.",
+      "reference": "(maybe this should be MyDevice or just removed entirely?)"
+    },
+    {
+      "prediction": "I think it would be better to have this as a separate field in the `UnrecognizedTaskError` struct. This way the `Task` field can be omitted from the `apitaskstatus.TaskStopped` message.",
+      "reference": "Can you do a nil check here, as the task isn't always nil here?"
+    },
+    {
+      "prediction": "I think we can remove the `map` import here.",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "4004000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "Does order matter? Should we sort by enode key to make duplication checking easier?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could come up with that worked.",
+      "reference": "Can you explain this too? This seems to go beyond linting."
+    },
+    {
+      "prediction": "I think this test should be moved to `src/test/integration/test-component.js`",
+      "reference": "I just moved this into the render.js UT file since this test doesn't actually use an components"
+    },
+    {
+      "prediction": "I think this should be `app.CreateSettingsFile()`, not `app.CreateSettingsFile()`",
+      "reference": "We do have to capture and check the err here right? I don't think we need to capture the location of the file though for this context."
+    },
+    {
+      "prediction": "I don't think we need to import `os` here.",
+      "reference": "an unused import"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could come up with that worked.",
+      "reference": "If this value doesn't exist (do an `, ok := ...` check), we should initialize and store it"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to use `Number.MAX_SAFE_INTEGER` here.",
+      "reference": "Would use parenthesis to explicitly state the associations of operands ```js const MAX_SAFE_INTEGER = Number.MAX_SAFE_INTEGER || (-(2 ** 53) - 1)"
+    },
+    {
+      "prediction": "I think this should be `self.inWaiting = self._ser.inWaiting or self._ser._inWaiting` instead of `super(Serial, self).__init__(self._ser.hComPort, onReceive)`",
+      "reference": "This handle is now private to pyserial, but if there is no public function to retrieve it I guess this is the best we can do."
+    },
+    {
+      "prediction": "Shouldn't this be `wlr_xdg_surface_schedule_configure_fullscreen`?",
+      "reference": "This function is what **the compositor** calls when it wants to fullscreen a client, not what the client calls."
+    },
+    {
+      "prediction": "I think this is the only change that needs to be made to this test case to make it pass.",
+      "reference": "s/not existed Pod/non-existing Pod"
+    },
+    {
+      "prediction": "I think we can remove the `ecsSvcGetter` and `aasSvcGetter` now. They are no longer used.",
+      "reference": "Why did we not leave this as just `svcDescriber`?"
+    },
+    {
+      "prediction": "I don't think we need to import `Queue` from `google.cloud.security.common.data_access` at all. We can just use the built-in Python `Queue` module.",
+      "reference": "nit: I think we've been (in)consistent in leaving a blank line here."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this column. I think it should be called `full_name`, since it's the full name of the rule.",
+      "reference": "Maybe we can change it to a more generic name like source_id."
+    },
+    {
+      "prediction": "I think you can just do `errCh := make(chan error, 1)` and then do `errCh <- err` in the `defer` above.",
+      "reference": "What's the theory on why this could fix the hang? It seems like the call to `GetRootNodeForTest` will always need to call into the crypto object before returning an error, and so it should block on that `c` receive, after which the test immediately drains `errCh`. So I don't quite see how buffering would help..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. It seems like it should be in the same place as the `TEST(PREFIX_EVEX_VV, di->prefixes)` check.",
+      "reference": "Maybe add an assert in arch_init or somewhere?"
+    },
+    {
+      "prediction": "Why do we need to nolint here?",
+      "reference": "I wonder what does this `// nolint` ignore? And what are the standards that you choose which error to fix and which to ignore?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think it's better to check if `cores` is not `null` before calling `getZkController()`.",
+      "reference": "what about the line below, forceUpdateCollection?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Doesn't this go in the opposite direction of #1051 / #1062 by putting platform-specific logic back into this module?"
+    },
+    {
+      "prediction": "I think this should be `authenticated`, not `authenticated`.",
+      "reference": "I think this is the only line in JS legacy code that should be updated based on the renaming. `global.googlesitekit.setup.isAuthenticated = response.authenticated;` would be sufficient to keep the old name for the legacy JS data (which will be phased out in the long term) while supporting the new name in the API route. All updates to the JS components above wouldn't be necessary."
+    },
+    {
+      "prediction": "I'm not sure if this should be 500ms or 1000ms. I think 1000ms would be a bit too long, but 500ms might be too short.",
+      "reference": "Technically, this would allow `pingWaitTime` to be `false`. I don't know if it matters here."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in the `kinesis` test suite.",
+      "reference": "I only want the proxy request for this to run for kinesalite. So this just returns true if the KINESIS_PROVIDER is kinesis-mock"
+    },
+    {
+      "prediction": "I think this should be `Copyright (C) 2006-2019 NV Access Limited, Peter V\u00e1gner, Aleksey Sadovoy, Joseph Lee`",
+      "reference": "Please add your name to the list of copyright holders."
+    },
+    {
+      "prediction": "I don't think we need to make it public.",
+      "reference": "Maybe maxRetryTimes should be bigger than 2. Return Max(maxRestryTime, 2)?"
+    },
+    {
+      "prediction": "I don't think we need to pass the signedUserName here.",
+      "reference": "Call `userName` is good enough, `signedUserName` is narrowing the actual usage of this method."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `TraceContext` class, but I'm not sure if that's the best place to put it.",
+      "reference": "How would you feel about a static method to express intent?"
+    },
+    {
+      "prediction": "I'm not sure if we want to remove this, but I think it's fine for now.",
+      "reference": "Do we know for sure that no other code changed its value? (Maybe some weird testing use case or something) Why not just use the actual value instead of risk lying?"
+    },
+    {
+      "prediction": "This file should be renamed to `root.sql` and `part.sql` instead of `root.sql` and `part.sql`.",
+      "reference": "the partInstall.go isn't an auto-generated file, and the part.sql doesn't exists either."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should go in `cache/ttl.go`.",
+      "reference": "`k` is unused in PutHash"
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. It is not related to the task of copying a stream to another stream.",
+      "reference": "Use IOUtils instead"
+    },
+    {
+      "prediction": "I think this should be `// TODO: each caller should use it's own mocked manager part instead of global one`",
+      "reference": "It's not related to the line of code, but I don't know how to write it in a more proper place. Looks like `e2e/myst-provider/db/myst.db` should be in `.gitignore` file and should not be commited."
+    },
+    {
+      "prediction": "I don't think this is a good name for a property. It's not a property of an option, it's a property of the ChromeOptions object itself.",
+      "reference": "nice touch since Google likes to change things every now and then."
+    },
+    {
+      "prediction": "I think it would be better to move this into the constructor of `CircuitBreakerManager`",
+      "reference": "Make `CircuitBreakerManager` implement `PluginInfoInitialized`"
+    },
+    {
+      "prediction": "This should be `variant.convert(QMetaType(QMetaType.UInt.value))`",
+      "reference": "This is extra awkward, QVariant.convert() takes a QMetaType which takes the value of a QMetaType.Type."
+    },
+    {
+      "prediction": "I think we can remove the `ctx` field from the `Writer` struct and just pass it to `NewTypedWriter` and `NewRawWriter`.",
+      "reference": "Nit: finished -> finish, \"done passing it\" -> \"passed it\"."
+    },
+    {
+      "prediction": "Do we need to do this if we're already setting `S3_BUCKET_NAME` and `S3_ACCESS_KEY_ID`, `S3_SECRET_ACCESS_KEY` and `S3_PERMISSIONS`?",
+      "reference": "Maybe mention that it will be saved to the filesystem otherwise?"
+    },
+    {
+      "prediction": "I think this should be `string.Format(Resources.ImportTransitionListColumnSelectDlg_DisplayData_Column_The_input_text_did_not_appear_to_contain_column_headers__Use_the_dropdown_control_to_assign_column_meanings_for_import_, (i+1).ToString());`",
+      "reference": "Probably cleaner and more general to do this: string,Format(\"Column {0}\", i+1) That way if the local language has a convention like \"3 column, this is\" it still works."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it works for now.",
+      "reference": "This can be uniformed too"
+    },
+    {
+      "prediction": "I think we can remove the `AddPeersErr` variable and just use `d.addPeerErr`.",
+      "reference": "It looks to me that this does not have to be exported."
+    },
+    {
+      "prediction": "I think this should be `ApiConfig.createDummyApiConfig()`",
+      "reference": "It looks like this is only used in order to extract the snippet file names from GapicProviderFactory. I wonder if it would be better expose the snippet file names directly, without exposing a test-only function in non-test code."
+    },
+    {
+      "prediction": "This method is not used anywhere in the codebase, so it can be removed.",
+      "reference": "Another question: this is being removed, but I don't see a corresponding removal if it being called? Was it never used (or am I just missing it)? (is it related to what was in the onclick that got removed, i.e. was that supposed to call this and not something in the backing bean?)"
+    },
+    {
+      "prediction": "I think this should be `apiutil.CertificateRequest.HasApproved()`",
+      "reference": "Suggestion: `.. does not have an Approved condition set to true`"
+    },
+    {
+      "prediction": "I think this should be `self.client.delete_files_before_test = self.parameters.get(\"delete_files_before_test\", False)`",
+      "reference": "I would say that client class should have default of False and Cloud prov should enable deleting."
+    },
+    {
+      "prediction": "I think we should remove this file from the PR.",
+      "reference": "Did this remove the BOM or something?"
+    },
+    {
+      "prediction": "I think it would be better to have this method as a static method in the `Rule` class, and call it from the `addRule` method in the `BlockHeaderValidator` class. This way, you don't need to create a new instance of `Rule` every time you add a new rule.",
+      "reference": "This makes the BlockHeaderValidator mutable, and IMHO should not be done. Note we have a builder that has the exact same method, so instead of mutating a produced BlockHeaderValidator we should hook into wherever the builder is being created."
+    },
+    {
+      "prediction": "I don't think we should change the default value of `async` to `True` here. It might be useful in the future, but it's not necessary right now.",
+      "reference": "May be add information about parameters to the doc string while at it?"
+    },
+    {
+      "prediction": "I'm not sure if this is a legacy RPC or a new one, but I think it should be renamed to `eth_getTransactionCount`.",
+      "reference": "Naming of this sounds a bit awkward. Perhaps priv_legacy_getTransactionCount or priv_legacyGetTransactionCount?"
+    },
+    {
+      "prediction": "I think this should be `log_msg(report_data, VK_DEBUG_REPORT_WARNING_BIT_EXT, get_debug_report_enum[obj.type], obj.handle, kVUID_Core_MemTrack_FreedMemRef, \"VK Object %s still has a reference to mem obj %s.\", report_data->FormatHandle(obj.handle).c_str(), report_data->FormatHandle(mem_info->mem).c_str());`",
+      "reference": "This one should probably be added to the list in #24 . It's possibly indicative of an issue, but there isn't a VUID for it."
+    },
+    {
+      "prediction": "I think we can remove the `api_name` arg from the `build_network_url` docstring, since it's no longer used.",
+      "reference": "Why this indentation change? Seems the original indentation would be fine, and leaves more room."
+    },
+    {
+      "prediction": "`update` is not used in this function, it should be removed.",
+      "reference": "Would be nice to know what the `update` argument does in the docstring."
+    },
+    {
+      "prediction": "I'm not sure if this is a good or a bad thing. I think it's a good thing, but I'm not sure if it's a good thing to do in general.",
+      "reference": "I think we can just use `io.BytesIO` here."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `azure.go` file, but I'm not sure if that's the right place to do it.",
+      "reference": "This ensures that the User-Agent header matches during replay."
+    },
+    {
+      "prediction": "I think this should be `1.3` instead of `0.05`.",
+      "reference": "What units are used for this default?"
+    },
+    {
+      "prediction": "I think it would be better to use `JsonProperty` instead of `[JsonProperty]`",
+      "reference": "I am wondering if it makes sense to make this a flag to `flatc` rather than a preprocessor flag, since this is generated code, it would make more sense to simply omit this code completely if the flag is not specified"
+    },
+    {
+      "prediction": "I think this should be `tab_count` instead of `tab_count == 1`",
+      "reference": "Doesn't this change the behavior (edit: oh, I see, never mind)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `else` branch of the `if` statement above, since that's where we're checking for `xenial` or `bionic`.",
+      "reference": "I think you need a `do` at the end of this line."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Not passing in a `:params` key here."
+    },
+    {
+      "prediction": "Why do we need to set the power state here?",
+      "reference": "Feels like this should be an option and not the default. Leaving them running would give you immediate response to claims, which might be what some people would want. Hibernation is AWS only right now and probably other clouds will just disregard the setting but arguably validation should be rejecting attempts to create / update powerstate on gcp/azure/etc. I'd pitch we put this onto the pool spec and make you opt in."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this. I think it would be better to put it in `IntegrationTestHelpers` since it's only used in integration tests.",
+      "reference": "I think I saw somewhere that it has been fixed and that you can use #NETCOREAPP3_0_OR_GREATER but I may be wrong."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to activate the consensus. I think it should be moved to the Consensus module.",
+      "reference": "what's the second part for?"
+    },
+    {
+      "prediction": "nit: remove trailing whitespaces",
+      "reference": "It's not clear to me what the \"super\" test above is for. As far as I can see, the test runs a spellcheck for \"super\" but then uses \"fob\" as the index into suggestions, which will never find an entry."
+    },
+    {
+      "prediction": "I think we should rename `ioctlUtil` to `ioctlutil` to be consistent with the rest of the package.",
+      "reference": "revert this file change"
+    },
+    {
+      "prediction": "I think this should be `pylint: disable=no-member, useless-suppression` instead of `pylint: disable=not-callable, useless-suppression`",
+      "reference": "I think this is the most elegant way to allow us to still fail on `I`."
+    },
+    {
+      "prediction": "I think this should be `addcslashes` not `addcslashes`",
+      "reference": "Wouldn't it be better to use a sql `IN ( :exact_value )` ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I've been able to get it to work.",
+      "reference": "So far this is not working as expected... working to further investigate WinRM protocol and the gems behavior."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it's better to have a `get_requests` method that returns a list of requests, and then call `self.scenario.get_requests()` in the `__init__` method of the `ApacheBenchmarkExecutor` class. This way, you don't need to create a new instance of the `ApacheBenchmarkExecutor` class every time you run the benchmark.",
+      "reference": "So now there're two ways to get requests: `ScenarioExecutor.get_requests()` and `Scenario.get_requests()`. To avoid confusion we should probably make `Scenario.get_requests()` method private, or hide it in some other way."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `if (connection->type == nano::bootstrap_server_type::realtime) {` part of the `if (message_a.response->second) {` block.",
+      "reference": "This existed already, but seeing as you are changing the variable to have an explicit type it could be made a const reference to remove a copy."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `base/fake` or `base/fake_words` or `base/fake_noun` or `base/fake_noun_phrase` depending on what you want people to be able to use this for.",
+      "reference": "I don't think this code is being tested; this method can be removed without the test failing"
+    },
+    {
+      "prediction": "I don't think this is needed. The `onPreExecute` method is called before the `onCreate` method, so the `ArrayList` is not initialized yet.",
+      "reference": "@codepoet2017390 You have already initialized this ArrayList in the onPreExecute method then why do it here too?"
+    },
+    {
+      "prediction": "Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "This covers the `RESTRICT_ACCESS=true` x non-gsa pair, but we probably want to test the three other permutations: (`RA=true`, gsa), (`RA=false`, non-gsa), (`RA=false`, gsa)"
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "Please don't do this - if you want to disable code permanently delete the code, if you need to disable it to test don't commit. Such sections are very poorly highlighted during review and can cause unspotted issues slipping through the process..."
+    },
+    {
+      "prediction": "I think we should rename this interface to `ConnTrackInterfacer` to be consistent with other interfaces.",
+      "reference": "I'm starting too feel we have too much abstraction and too many interfaces here. But this method is definitely the most confusing IMO. For one, it's called \"Get*\" but it does not return anything... I don't have a good understanding of what's going on anymore, so it's hard for me to suggest an alternative, but it's pretty clear to me that this interface needs to be rethought. An interface where all the parameters are themselves arbitrary objects (`interface{}`), whose actual type depends on the specific interface implementation, is not a helpful abstraction"
+    },
+    {
+      "prediction": "I think this import should be removed.",
+      "reference": "It fills me with endless sadness that we can't use Selenium's own `SessionId` here."
+    },
+    {
+      "prediction": "I think this should be `return stat.getVersion() > aliases.getZNodeVersion();` instead of `return stat.getVersion() <= aliases.getZNodeVersion();`",
+      "reference": "Interesting. So you've found that it's faster to request only the \"Stat\" without the data so long as this is the typical path?"
+    },
+    {
+      "prediction": "Can you add a note to the changelog that this is a bug in the querier?",
+      "reference": "Is this a future fix? I thought the local identifier scope and the package issue were the same one."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for an application object. I think it would be better to check if the object is an instance of `IApplication` or `IApplicationDialog`.",
+      "reference": "Not super important, but this line could be split up."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `h2o_storage_close()`. Also, I think it would be better to do this in `h2o_storage_free()`.",
+      "reference": "Maybe we should better call the dispose function without checking the value of `item.data` (since the dispose function is setup before `data` is set up)."
+    },
+    {
+      "prediction": "Why do we need to set the mode back to CPU here?",
+      "reference": "Why set to GPU, is it CPU?"
+    },
+    {
+      "prediction": "I think this should be `return t('appointments.todayTitle');` instead of `computed('i18n', () => t('appointments.todayTitle'));`",
+      "reference": "This should be `computed('i18n.locale'....`"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for the channel. It doesn't seem to have anything to do with the state of the session.",
+      "reference": "can we not expose the channel directly? or at least expose it as a read only channel?"
+    },
+    {
+      "prediction": "This should be `AC_MSG_ERROR([bison not found but required])`",
+      "reference": "Can we add a `--with-bashcompdir` here to override this? I'd like to be able to do unprivileged builds."
+    },
+    {
+      "prediction": "I think we should rename this to `CreateSnapshot` and `DestroySnapshot` to be consistent with the other commands in this package.",
+      "reference": "We will be removing this maya/volume package completely. Do not use this."
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix. The problem is that `host_bad_alloc` is a function that is only defined in the `host.cpp` file, so it won't be available in the `client.cpp` file.",
+      "reference": "Why you think that `model_ptr` will be `nullptr` when memory allocation is failed? I think `new daal_knn::Model(column_count)` would throw `std::bad_alloc` if `operator new` is not overloaded. Is it overloaded by DAAL in the way it simply returns `nullptr` in case of bad allocation?"
+    },
+    {
+      "prediction": "I think it would be better to say \"The miner will not be able to commit additional sectors if its collateral drops below 0.001 FIL per sector.\"",
+      "reference": "Where does this 0.001 come from? Is this from the network collateral cost function, which happens to be hardcoded right now? I think it's going to rot very quickly: it depends on sector size and (probably) block height."
+    },
+    {
+      "prediction": "I think this should be `Copyright (c) 2018 by the CodeChecker infrastructure developers.`",
+      "reference": "Please update the license header in each source file. For more information see: #2697"
+    },
+    {
+      "prediction": "I think this should be `@param {number=} retries The current number of retries.`",
+      "reference": "!numbers -> numbers"
+    },
+    {
+      "prediction": "why is this needed?",
+      "reference": "so we don't have universal rules afterwards? In that case I'd remove this entirely (and maybe some places where it's used get cleaner) since we shouldn't ever add a universal rule. Not only because it's way simpler in TS, I'm also just super skeptical about \"universal\". In case of this rule, even control characters may have their place in Swaggers, e.g. in `examples` sections where you can have raw response bodies where control characters are perfectly valid."
+    },
+    {
+      "prediction": "I think this should be `options.stdio || 'ignore'` instead of `options.stdio || 'ignore'`",
+      "reference": "You're welcome to surface this option through the API, but I'm not going to make this the default behavior."
+    },
+    {
+      "prediction": "I think this should be `return buildBinaryFor(source, target, targetOS, targetArch, buildStatic)`",
+      "reference": "Just a thought that any further extensions of `buildBinaryFor` signature could be done as `buildBinaryFor(..., opts... BuildOptions)` for better readability."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I could come up with.",
+      "reference": "I'll fix the InMemoryExporter right after this PR lands"
+    },
+    {
+      "prediction": "I'm not sure if this should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not sure if it should be a warning or an error. I think it should be a warning, but I'm not",
+      "reference": "Good call, none of these are technically invalid"
+    },
+    {
+      "prediction": "I think this should be `msbuild /property:Configuration=${{matrix.configuration}} /property:Platform=${{matrix.platform}} /target:platform=x86 /target:platform=x64 dd-continuous-profiler-dotnet\\src\\ProfilerEngine\\Datadog.AutoInstrumentation.Profiler.Native.Windows\\Datadog.AutoInstrumentation.Profiler.Native.Windows.WithTests.proj`",
+      "reference": "this one should be updated too"
+    },
+    {
+      "prediction": "I think it would be better to use `getInstanceId()` instead of `getMicroserviceName()`.",
+      "reference": "maybe need to add microserviceId too."
+    },
+    {
+      "prediction": "I think this should be `missing-param-doc-and-type-doc-with-no-accept-no-param-doc`",
+      "reference": "I'm curious, do you create this commit at the very end by rebasing and taking the value you had after moving everything ?"
+    },
+    {
+      "prediction": "I think this test should be renamed to `should not add more source rows than defined in maxRows when trimming rows using the modifyRow hook`",
+      "reference": "It's no longer possible to use `modifyRow` to trim rows and that's ok to use TrimRows instead. But I think we should move this test to TrimRows plugin tests as `trimRows` setting is not a part of the `core`."
+    },
+    {
+      "prediction": "I think this should be `return NULL;` instead of `return node->data.scalar + 1;`",
+      "reference": "The function seems to either return \"default\" or return the extension stripping the preceding dot. Does the fact mean that we can no longer register `.default` as an extension?"
+    },
+    {
+      "prediction": "I think this should be called `ImageHash` instead of `ImageDigest` to be consistent with the other fields.",
+      "reference": "this may require a statefile version update"
+    },
+    {
+      "prediction": "This class is deprecated and will be removed in the next major. Use SplFileObject::fgetcsv() instead.",
+      "reference": "I like when `@deprecated` contains how to resolve it - eg. \"use `SplFileObject::fgetcsv()` instead\". The information in upgrade notes is usefull only during upgrade."
+    },
+    {
+      "prediction": "I'm not sure if we want to keep the v1 and v10 in the same package. I think it would make more sense to have the v1 and v10 be in the same package as compute.",
+      "reference": "Nice to see that your local environment is up-to-date now for these changes."
+    },
+    {
+      "prediction": "`countlyVue.sideBarComponent.$store.getters[\"countlySidebar/getSelectedMenuItem\"]` should be `countlyVue.sideBarComponent.$store.getters[\"countlySidebar\"][\"getSelectedMenuItem\"]`",
+      "reference": "Instead of setting it to null, i think lets set it to {}"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for the group. I think it should be something like `monitoring.monitoring.openshift.io` or `monitoring.monitoring.openshift.io/v1alpha1`.",
+      "reference": "We should bump this before releasing and remove all legacy fields that are safe to do so. Since we cannot have multiple versions anyway there's little value in walking around with the \"alpha\" stamp."
+    },
+    {
+      "prediction": "I think this should be `AC_DEFINE([DEEPBIND], [no], [deepbind is unsupported with asan, musl and so-forth])`",
+      "reference": "I might suggest a different name for the `DEEPBIND` config.h macro. Perhaps `FLUX_DEEPBIND` to give a hint in the code that we're using a locally provided define."
+    },
+    {
+      "prediction": "Can you add a license to this file?",
+      "reference": "Code generated by protoc-gen-go. **DO NOT EDIT.**. The correct way to fix `deprecated` in this file may be updating the protoc / protobuf / grpc and regenerate this file."
+    },
+    {
+      "prediction": "Why is this deprecated?",
+      "reference": "We should make sure that we cover the testcases in a new unit test class."
+    },
+    {
+      "prediction": "I think this should be `SectorSize(ssType Mode) uint64`",
+      "reference": "Is this only used in test code? Is there future utility in it for production code? If not, it would be nice if this could be moved into a test package. I'm not sure how hard that would be. CC @laser"
+    },
+    {
+      "prediction": "Why do we need both `config.WebhookLabel` and `config.APIServiceLabel`?",
+      "reference": "Super curious how you noticed this, as SPIRE itself currently only configures once...?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `NotFoundExceptionMapper` class.",
+      "reference": "This `startsWith` seems a bit brittle because in the code base we'll occasionally spell it \"DataFile\"."
+    },
+    {
+      "prediction": "I think this should be `99999` instead of `999999`.",
+      "reference": "if it keeps reconnecting for too long, how should we get alerted?"
+    },
+    {
+      "prediction": "I don't think we need to set the Keepalive here, since it is already set in TCPIPSession()",
+      "reference": "it's better to use single name style in one name."
+    },
+    {
+      "prediction": "I think this should be 10043",
+      "reference": "As defaultSendPort is being removed, and I don't like TODOs on examples, please remove the whole line"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to update the DNS delegation accounts. I think it should be done in the `cloud-init` package, since that's where the `cloud-init` cloud provider is used.",
+      "reference": "Should we return `nil` when `err == errChangeSetEmpty` instead of calling l.103 `WaitUntilStackUpdateCompleteWithContext`?"
+    },
+    {
+      "prediction": "I think this should be `private int[] ordMap = new int[0];` instead of `private int[] ordMap = new int[0];`",
+      "reference": "Hmmm I thought spotless would wrap this line, but it doesn't seem to complain about it"
+    },
+    {
+      "prediction": "I think we can remove the `ParentId` and `SpiffeId` fields from the `SpiffeIDSpec` struct.",
+      "reference": "I think this will require updating the api version. we are planning to do this anyways, so perhaps we can lump this in so we don't have to update too many times?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the tile layer's constructor.",
+      "reference": "revert this for parity"
+    },
+    {
+      "prediction": "(style) Macros with complex values should be enclosed in parentheses",
+      "reference": "INCR_BATCH and DECR_BATCH could be exposed as tuneable command line options but that might be too low level to expose. I think keeping private for now where \"in the know\" folks could use to try and tune makes sense. These values were choosen mostly \"at random\" and have worked out for us thus far."
+    },
+    {
+      "prediction": "I don't think this is needed. The `setCustomTagWriter` function is only called from `setTagWriter`, so it doesn't need to know about the global `writerTable`.",
+      "reference": "Currently this isn't very clean because we have to re-use some of the predefined parser types - the demo uses WRITER_U_CTAGS but there should be some support for custom writer type."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "I think this was left it accidentally :)"
+    },
+    {
+      "prediction": "I think we can remove `NEWUTS` and `NEWPID` now.",
+      "reference": "Can you finish re-ordering this list to match the one that was deleted?"
+    },
+    {
+      "prediction": "Is it possible to use `spec.ContainerName` instead of `spec.Spec.ContainerName` here?",
+      "reference": "Do we need to check if `exp.Target.StressChaos` is `nil`?"
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think we can remove this task as it is no longer used.",
+      "reference": "not calling this is going to impact current functionality which depends on PVC.. let us still call this and set the values to default as 'none' (or empty) by handling the error from 'get' operation"
+    },
+    {
+      "prediction": "Shouldn't this be `FPGA_ERR(\"IRQ write %zu bytes\")`?",
+      "reference": "`!=` is no the same as `<`"
+    },
+    {
+      "prediction": "This should be `__all__ = []`",
+      "reference": "No need to create a dir for transformer. Simply move transformer.py into mmdet/models/utils/"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the SelectSubscriptionActivity.",
+      "reference": "Can't you just add `SelectSubscriptionActivity` directly to the manifest instead of opening `MainActivity` that then starts it?"
+    },
+    {
+      "prediction": "`run-check` -> `check`?",
+      "reference": "Maybe say \"the spec's %check section\" so it's clearer what we mean."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to check if sorting is enabled. I think it would be better to check if `this.sortingEnabled` is true.",
+      "reference": "I think after change this check `rowInMapper === null` is always false, so it's useless."
+    },
+    {
+      "prediction": "Why do we need to change this?",
+      "reference": "ClockIpLpi2c4.Enable() is required. The same change is required for enablePeripheralClocks()."
+    },
+    {
+      "prediction": "I think we can remove the `Sha` field from the `PrometheusConfig` struct as well.",
+      "reference": "nit: this should be `SHA` since it is an acronym for `Secure Hash Algorithm`, no?"
+    },
+    {
+      "prediction": "I think this should be `const determinedTimezone = window.Intl && window.Intl.DateTimeFormat && window.Intl.DateTimeFormatresolvedOptions().timeZone ? window.Intl.DateTimeFormat.timeZone : 'UTC'```",
+      "reference": "Not sure if Intl can be defined but not `Intl.DateTimeFormat`. Both seem to have the same percentage on CanIUse."
+    },
+    {
+      "prediction": "I think this should be `pvc.Name` instead of `pvc.Name`",
+      "reference": "PTAL, looks like the order of PVC & PV is reversed for different storage-engine types, for Jiva/CStor the args sent to sendEventOrIgnore are `PV, PVCName`."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Remove the trailing white space"
+    },
+    {
+      "prediction": "I don't think this is the right place to add this check. I think it should be in `setup.py`, not `setup.py.test`.",
+      "reference": "No need to modify this."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it. I think we should be able to get the hostname from the IP address without having to know the hostname of the DNS server that resolves to the IP address.",
+      "reference": "would be nice at some point to clean this signature up... not relevant for this PR"
+    },
+    {
+      "prediction": "I think this should be `default: 'text'`, not `default: 'text'`.",
+      "reference": "Here, is it `null` because it is `null` in `$fieldConfiguration`, or because it has been set to `null` above in the `array_replace` ? If it doesn't exists at all in `$fieldConfiguration`, I think you should add `type => 'text'` above in the `array_replace`."
+    },
+    {
+      "prediction": "I think this should be `node.getName().toString().endsWith(className)`",
+      "reference": "I was thinking about this for #2928 too, which also uses `endsWith`... Presumably `SomeOtherObject` would match `endsWith(\"Object\")`, meaning that something more sophisticated like splitting it then iterating right to left would be needed in order to do this robustly?"
+    },
+    {
+      "prediction": "Should this be `BatchExportProcessorOptions` or `BatchExportOptions`?",
+      "reference": "I think this name is a bit confusing because if you asked someone what type of exporter they were using they would probably say Zipkin or Jaeger, etc. How about `ProcessorType` with `Simple` and `Batch` definitions?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `FakeWebServer`?",
+      "reference": "`req` is unused in RecreatePipedKey"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add this message. I think it would be better to add it in the `else` case of the `if ( existingTag !== '' ) {` block above.",
+      "reference": "The \"Adsense\" term needs to be capitalized - I just noticed this was wrong through ACs and IB, so not a problem of the PR itself really. I'll quickly fix it."
+    },
+    {
+      "prediction": "I think it would be better to move these functions to `pkg/beam/functional` since they are only used in `pkg/beam/functional`.",
+      "reference": "do which methods need to be registered? is it basically just the ones that are directly referenced in ParDo calls? (The Beam GoDoc doesn't say anything about it. :/ )"
+    },
+    {
+      "prediction": "I don't think we need to add this as part of the download instance. I think it should be part of the file metadata.",
+      "reference": "@bmckinney does this mean that the bibtex citation will be available via API?"
+    },
+    {
+      "prediction": "I don't think we need to bump this anymore, right?",
+      "reference": "We should probably tag the cliquet-fxa version as well."
+    },
+    {
+      "prediction": "I think this should be `memset(&sin, 0, sizeof(sockaddr_in));` instead of `memset(&sin, 0, sizeof(sockaddr_in) + sizeof(sockaddr_in->sin_family) + sizeof(sockaddr_in->sin_port) + sizeof(sockaddr_in->sin_addr.s_addr));`",
+      "reference": "IIRC we need to use memset, because an empty brace is not C99 comformant, and because we cannot use `{0}` because how the struct is organized is not defined in POSIX (the initializer cannot be `{0}` if the first property of the struct is a struct)."
+    },
+    {
+      "prediction": "I think we can remove this field now.",
+      "reference": "Will we be adding this to `AppliedTo` as well? If not, any reason why not?"
+    },
+    {
+      "prediction": "Shouldn't this be `@return: Boolean`?",
+      "reference": "Just a note that this feels weird being on the base NVDAObject rather than IAccessible, but right now, I understand that's how it has to be because we fire MSAA caret events on the focus object regardless of whether it's IAccessible. I think we should consider restricting these caret events to focus objects that are IAccessible subclasses in future, but that change is probably too risky for this PR. CC @MichaelDCurran for his thoughts."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "This change and other similar ones should not be part of this commit. This commit is about adding a ImageUploadType not fixing controllers and stuff."
+    },
+    {
+      "prediction": "I think this should be `AccountPingReqSubj`",
+      "reference": "I am a bit confused about PING here. Ping is implied and it overlaps with \"$SYS.REQ.ACCOUNT.%s.%s\" where the last token is what is requested (CONNZ) and the one before is the requested account id."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "nit: use the OptSubFolder constants. Also in the client you are using \"maxdepth\" but in the server the constant for \"depth\" is being used."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/unit/config.js`",
+      "reference": "I also moved this to `/verdaccio/storage` since it's written in runtime, and the approach here is to make the configuration by default read-only."
+    },
+    {
+      "prediction": "I think this should be `gax.EventEmitter` instead of `Stream|gax.EventEmitter`",
+      "reference": "nit: s/has cancel method/has a cancel method"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundBlockBody: Extra empty line detected at block body end.",
+      "reference": "Useless assignment to variable - `second_quiz`. Use `_` or `_second_quiz` as a variable name to indicate that it won't be used."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `encoding/gzip.go`.",
+      "reference": "Is this still needed since there's the `GzipPrecompressed` type?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "this is spelled wrong and also not used so remove it"
+    },
+    {
+      "prediction": "I think this should be `@SuppressWarnings(\"unused\")`",
+      "reference": "It might be right here that's the issue ^"
+    },
+    {
+      "prediction": "I don't think this is a good idea. It will break backwards compatibility. I think it would be better to just remove this method and let the user override `comparator()` in `SortedMap<K, V>`.",
+      "reference": "@danieldietrich let's leave existing extension. Here we just mark `keyComparator()` with `@Deprecated` annotation"
+    },
+    {
+      "prediction": "Please don't include `io.h` in this file.",
+      "reference": "This breaks CI. I am not sure why you need to include io.h in this file especially on linux."
+    },
+    {
+      "prediction": "I think this should be `parentDom.childNodes.length - 1`",
+      "reference": "this seems equivalent"
+    },
+    {
+      "prediction": "I'm not sure if we should change this to 20*1024 here. I think it's fine to leave it at 200*1024 for now.",
+      "reference": "change size to 50MB criticial, 10MB warn... change count to 50k critical, 5k warn..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `mergeResponseFromShard` method, since it's the responsibility of the `mergeResponseFromShard` method to determine which pivot facet to use.",
+      "reference": "This has nothing to do with master/slave replication. Is this something we want to change too? If so, I guess this is an aggregation?"
+    },
+    {
+      "prediction": "I'm not sure why we need to detach the process here, but I think it's safe to do so. The `Process.waitpid(pid, 0)` will return the pid of the process that was just forked, so we can use that pid to detach the process from the kernel process table.",
+      "reference": "Maybe it makes sense to mention that those zombies are not forever, but only up to the moment when the parent process exits? Is that correct according to your observations, @benoittgt ? Please disregard this note if zombies remain after."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be moved to `moto/moto.py` where the mocking is done.",
+      "reference": "moto mock setup needs to happen before boto usage."
+    },
+    {
+      "prediction": "feval is not None means you have a function that computes a metric at each fold. If it is not None and `feval` is not None, it means you have both `first_metric_only` and `feval` at the same time. In that case, it means you are using `feval` but not `first_metric_only`, which is not possible.",
+      "reference": "`hasattr(cb, 'first_metric_only') and cb.first_metric_only` -> `getattr(cb, 'first_metric_only', False)`"
+    },
+    {
+      "prediction": "This is a bit of a hack, but I think it's the only way to make this work. The `ServerStream` interface doesn't have a `Close()` method, so we need to implement a `Close()` method in the `StreamHandler` interface. That's where this `streamHandler` type comes in. The `ServerStream` interface has a `Close()` method, so we don't need to implement that in the `StreamHandler` interface. The `ServerStream` and `StreamHandler` interfaces have a `Close()` method, so we don't need to implement that in the `ServerStream` and `StreamHandler` interfaces. The `ServerStream` and `StreamHandler` interfaces have a `Close()` method, so we don't need to implement that in the `ServerStream` and `StreamHandler` interfaces. The `ServerStream` and `StreamHandler` interfaces have a `Close()` method, so we don't need to implement that in the `ServerStream` and `StreamHandler` interfaces. The `ServerStream` and `StreamHandler` interfaces have a `Close()` method, so we don't need to implement that in the `ServerStream` and `StreamHandler` interfaces. The `ServerStream`",
+      "reference": "no newline for function"
+    },
+    {
+      "prediction": "I think this should be a warning, not an error.",
+      "reference": "This makes me feel icky. We could also go the route of making `wlr_output.scale` a double instead of a float, but that would be a more widely reaching change."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this check. I think it would be better to do it in the `if (vp->atp) {` block above.",
+      "reference": "Again, would be awesome if you could remove the tabs in these 3-4 lines as well."
+    },
+    {
+      "prediction": "`step.ref` is a `std::string`, so `std::move` won't work with it. You need to convert it to a `std::string` first.",
+      "reference": "You're moving multiple times from `step.name`; you're not allowed to do that. Instead you have to copy the `step.name` string and (if you want to) move once at the very last."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to show this message. I think it should be in the `CalendarService` instead of the `CalendarController`.",
+      "reference": "What about `may not end`? @jancborchardt"
+    },
+    {
+      "prediction": "I think it would be better to say something like \"You can't approve this deployment because you are not in the approver list.\"",
+      "reference": "If you think about it, the status should be PermissionDenied?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this metric. I think it should be in `metrics.py` under the `exact_match` metric.",
+      "reference": "Calculation looks good but don't know what format etc grafana expects this to be in."
+    },
+    {
+      "prediction": "I think this test should be moved to `tests/test_docker.py` since it's not related to the `--add-host` feature.",
+      "reference": "nit: We could use the `to_str(..)` util here. (let's not bother changing for this PR..   )"
+    },
+    {
+      "prediction": "I think this should be `github.com/temporalio/temporal/.gen/proto/execution` not `github.com/temporalio/temporal/.gen/proto/executiongenproto`",
+      "reference": "Can you rename alias to `executiongenpb`. The rule is package+\"pb\" for `temporal-proto` and package+\"genpb\" for server protos."
+    },
+    {
+      "prediction": "I don't think you need to copy here.",
+      "reference": "Any noticeable performance hit? Any other way to avoid the allocation and copy even of the array?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The idea here is that if the bus is ready, then we should be able to read the data. If the bus is not ready, then we shouldn't be able to read the data. I'm not sure what the right way to do this is, but I'm not sure this is the right way to do it.",
+      "reference": "What if you remove this line entirely from here?"
+    },
+    {
+      "prediction": "`self.add_extra_convs == 'on_output'` -> `self.add_extra_convs in {'on_input', 'on_output'}`",
+      "reference": "`on_lateral` can also be implemented"
+    },
+    {
+      "prediction": "I don't think this is the best way to check for an author. I think it would be better to do something like: `if (podcast.feedUrl != null && !podcast.feedUrl.contains(\"itunes.apple.com\") && podcast.author != null && !podcast.author.trim().isEmpty()) {`",
+      "reference": "Please add a space between `if` and `(`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `zpool_status.go` under the `ZpoolStatus` struct.",
+      "reference": "This statuses might have been better off with some custom status type. However, good for now."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "Thanks for catching the bug. Here your change lets `v4IPNet` and `v6IPNet` take the first IPv4/IPv6 address in `addrList`. Maybe choose the last one also works and makes code simpler? What's your idea? @tnqn"
+    },
+    {
+      "prediction": "I think this is the only place where we use the `detail_dict` variable. I think it would be better to move it to the top of the function so it's clear what it's used for.",
+      "reference": "nitpick: you can name this variable just `details`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. It seems like it should be in `general.h` instead of `parse.h`.",
+      "reference": "The character class in the regex should probably be \"[a-zA-Z0-9_]\"."
+    },
+    {
+      "prediction": "I think we should keep the old settings for now, and add a new setting for `openpgpProvider` and `openpgpSupportSignOnly` in the future.",
+      "reference": "Please also increment `Settings.VERSION` and update `AccountSettings`."
+    },
+    {
+      "prediction": "I think this should be `builddate` instead of `versionString`.",
+      "reference": "I think this should just say \"build <date>\". And when builddate wasn't set, e.g. somebody built weave with 'go build' or some other way, this should return \"unknown\"."
+    },
+    {
+      "prediction": "I think it would be better to have a `JobGroupInfo.Builder` instead of a `JobGroupInfo` class.",
+      "reference": "Can we move this into a separate var and call it `desc` like below?"
+    },
+    {
+      "prediction": "I don't think you need to change this line.",
+      "reference": "why RpcSchema need this annotations?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to clean up replies. I think we need to do this in the `if s.sys != nil && s.sys.replies != nil` block, and then we can remove the `if s.sys.replies == nil` check below.",
+      "reference": "Not sure it makes sense to send the reply if we did not delete from the map, but I guess it does not hurt.."
+    },
+    {
+      "prediction": "`continueProcess` is not used in this function. Can we remove it?",
+      "reference": "Most of these variables can be moved after `return` or to the body of `if` statement."
+    },
+    {
+      "prediction": "I think we can remove this error code and use the one from `PrivacyMultiTenancyError` instead.",
+      "reference": "nit: Is this actually just a http \"unauthorised\" response? Or is that too vague? If this is the right way to go ... is \"token\" the right word\" Or do we want to say something else?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think we should move this to `Bolt::PuppetDB::Config` since it's used by both `Bolt::PuppetDB` and `Bolt::PuppetDB::Config`.",
+      "reference": "We check for `$future` in PuppetDB::Config, so this won't change behavior if you don't have `future` set."
+    },
+    {
+      "prediction": "I think this should be `issubclass(Series, anchor).__name__` instead of `issubclass(Series, Series).__name__`",
+      "reference": "I think you should remove `if isinstance(by, str):`,. We also should fix the error message `raise ValueError('Not a valid index: TODO')` to match with pandas'"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `ClientInstrumentation` class.",
+      "reference": "Curious - do we want this to be `OTel.GrpcHandler.Request`? For example, if someone is debugging a crash dump, they would get some hint that it was added by OpenTelemetry."
+    },
+    {
+      "prediction": "I think this should be \"the state of the commit is unknown\" instead of \"the state is unknown\"",
+      "reference": "nit: unnecessary `</p>`"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this check. I would put it in the `else` branch of the `else if` above.",
+      "reference": "Should just be `#ifdef ADIOS2_HAVE_HDF5`"
+    },
+    {
+      "prediction": "I think we can remove the `stack_name[: 63 - 2 - (len(random_id_part) + len(resource_id_part)]` part of the conditional, since it's the same as `short_uid()`.",
+      "reference": "FYI: I deployed a stack with multiple lengths for the stack name and resource id to cloudformation, to see how it truncates the parts for the final FunctionName"
+    },
+    {
+      "prediction": "I think we can remove this now, since we don't need it anymore.",
+      "reference": "Did we not use this field @_@"
+    },
+    {
+      "prediction": "I think it would be better to use `getNPMConfig(['android_keystore_path', 'android_key_alias', 'android_keystore_password', 'android_key_password'])`",
+      "reference": "shouldn't this be `android_keystore_name`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for the token secret. I think it would be better to check for `iss.Spec.TokenSecretRef.Name == secret.Name`",
+      "reference": "Double check the issuers package too :)"
+    },
+    {
+      "prediction": "I think this should be `Continue -> continue or break loop to next consumer`",
+      "reference": "This is out of date right? Now that `_is_first_level_self_reference` uses it as well? Perhaps we should make the docstring more general and not refer to method names that could change."
+    },
+    {
+      "prediction": "I think this should be `testPromRemoteWriteWithTLSParallel`",
+      "reference": "This test alone takes around 10-12 minutes to run. It has 20 subtests each of which takes around one minute. We should see if we there are redundant subtests that we can remove, or look for a way to optimize them."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to just do `Object.keys(newProps).filter(k => oldProps.hasOwnProperty(k) && newProps[k] !== oldProps[k]).forEach(k => setProperty(dom, k, newProps[k], oldProps[k], isSvg));`",
+      "reference": "Good call, we need to keep that check :+1:"
+    },
+    {
+      "prediction": "I think we can remove the `BufferReader` ctor and use `ReadOnlySequence<byte>` directly.",
+      "reference": "Should the interface have 2 `ParseHeaders` methods?"
+    },
+    {
+      "prediction": "I think this should be a constant in Iceberg.",
+      "reference": "Nit: Does this make more sense as `iceberg.hive.metadata...`? The rest of the configs seem to start with `iceberg.hive`."
+    },
+    {
+      "prediction": "I think this should be `numProviders > 0` instead of `len(p.DNSMadeEasy.APIKey) > 0`",
+      "reference": "Is `baseURL` not required?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it.",
+      "reference": "Just delete it, `embedding_rpn_head` is not a `Detector` which should have `show_result`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this method. I think it should be in `Datadog.Trace.ClrProfiler.ExtensionMethods` since it's only used in `ExtensionMethods.cs`.",
+      "reference": "Where do we _get_ tag values? edit: I mean, where are these `GetFoo()` methods called from? It just seems weird to me that we are getting tag values from spans. I thought we only ever _set_ tags on spans."
+    },
+    {
+      "prediction": "I think this should be `log.Printf(\"Reading kzip with %s encoding\", pref)`",
+      "reference": "This seems like log spam. Maybe expose the encoding as a method and log at the point of interest?"
+    },
+    {
+      "prediction": "I don't think we should add a new protocol type here. I think we should either remove the protocol type altogether or add a new protocol type for the challenge handshaking authentication protocol.",
+      "reference": "I looked up that one, and it seems correct: Authentification is French, Authentication seems to be correct"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. This is a mixin, not a standalone module.",
+      "reference": "Can we print a more human-friendly warning here, saying that plotting won't work or something along those lines?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "@mhils didn't we try to get rid of these (now empty) wrapper classes?"
+    },
+    {
+      "prediction": "I think this is the only place where we're using the ipset package.",
+      "reference": "I would avoid importing `felix/ipsets` package because this package is the windows equivalent and should be at the same level of `felix/ipsets`. We could add linux specific dependencies into `felix/ipsets` later and it will break Windows build."
+    },
+    {
+      "prediction": "I think this should be `registerTable(String identifier, String metadataFileLocation)`.",
+      "reference": "Maybe, `import` would be a better name."
+    },
+    {
+      "prediction": "I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I think it would be better to have a constant for the name of the parameter. I",
+      "reference": "Wouldn't we want `PARAMETER_MAX_RESULTS` in here as well? (I might be miunderstanding how this is used)"
+    },
+    {
+      "prediction": "I think this should be done in `config.before(:each) do` instead of `config.before(:suite) do`",
+      "reference": "We can re-use the config in our tests."
+    },
+    {
+      "prediction": "I think this should be `test \"$pmix_hwloc_source\" != \"cobuild\"`",
+      "reference": "Wouldn't we have the same issue with the internal libevent? IIRC, the integration is pretty close to an exact duplicate. Or does the libevent m4 code already take that into account when setting this variable? PMIx _requires_ libevent, so it isn't like pmix_libevent_support can be anything other than 1. I'm okay with leaving this question for separate resolution - just wondering why we don't have the same issue here."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for Chrome on iOS. I think we need to check for Chrome on iOS and see if it's an actual Chrome browser or if it's Chrome on iOS. If it's Chrome on iOS, then we need to see if it's an actual Chrome browser or if it's Chrome on iOS. If it's Chrome on iOS, then we need to see if it's an actual Chrome browser or if it's Chrome on iOS. If it's Chrome on iOS, then we need to see if it's an actual Chrome browser or if it's Chrome on iOS. If it's Chrome on iOS, then we need to see if it's an actual Chrome browser or if it's Chrome on iOS. If it's Chrome on iOS, then we need to see if it's an actual Chrome browser or if it's Chrome on iOS. If it's Chrome on iOS, then we need to see if it's an actual Chrome browser or if it's Chrome on iOS. If it's Chrome on iOS, then we need to see if it's an actual Chrome browser or if it's Chrome on iOS. If it's Chrome on iOS, then we need to see if it's an actual Chrome browser or if it's Chrome",
+      "reference": "The actual User-Agent string uses \"CriOS\" not \"CRiOS\""
+    },
+    {
+      "prediction": "I think we can remove the `StartBatch` and `CommitBatch` methods from the `IKeyValueStoreWithBatching` interface as well.",
+      "reference": "Unused code deletion. Is that a desired thing? No implementation of this actually used this indexer, so removing it seemed reasonable (and was suggested by the compiler). Maybe this had an indexer so plugins could use it? This change is definitely pushing the limits of what is appropriate in a \"cleanup\" as it is changing public API surface area."
+    },
+    {
+      "prediction": "I think this should be `logger.error(\"Raised during teardown with_puppet_running_on: #{dumping_exception}\")`",
+      "reference": "We'd lose the data here from the original teardown_exception and only get the dump_exception - I fear that that could make the actual error difficult to track if it cascades."
+    },
+    {
+      "prediction": "I think this should be `dom.multiple = newVNode.props.multiple;` instead of `dom.multiple = newVNode.props.multiple && newVNode.type==='select';`",
+      "reference": "At this point in `diff()` we should know if we're dealing with a `component` or a native `html` element. We can reuse that information in this if-statement here :+1:"
+    },
+    {
+      "prediction": "I think it would be better to use `if epochNum >= a.foundationBonusP2StartEpoch && epochNum <= a.foundationBonusP2EndEpoch {` instead of `(epochNum >= a.foundationBonusP2StartEpoch && epochNum <= a.foundationBonusP2EndEpoch) `",
+      "reference": "somewhere in reward protocol, we need to do if epoch is Fairbank { a.foundationBonusP2StartEpoch = genesis.xxx a.foundationBonusP2EndEpoch = genesis.yyy p.state(sm, adminKey, &a) } that is to write the 2 new values at Fairbank height then this logic can follow"
+    },
+    {
+      "prediction": "I think it would be better to drop the table in a transaction, so that it gets rolled back if anything goes wrong.",
+      "reference": "Dropping and recreating would delete all t he previous records, are they any important?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `cgroup_test.go`",
+      "reference": "While it is technically OK to use `m.GetPaths()` here, and it's a mock code so it doesn't really matter, I'd still like to have `m.Path(\"devices\")` used here, because since commit 714c91e9f73a1512808476eb532b4aa36bbb7530 we're not supposed to use GetPaths() for anything other than state save/restore."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `system-config` package.",
+      "reference": "Let's be consistent everywhere: either \"Disable\"/\"Enable\" or \"Disabled\"/\"Enabled\". I'm voting for the latter.:)"
+    },
+    {
+      "prediction": "I think it would be better to have a separate exception for this case. I think it would make more sense to have it as a separate exception from the `ParameterException` that is thrown when `isPruningEnabled` is true. The `ParameterException` that is thrown when `isPruningEnabled` is true and `isGoQuorumCompatibilityMode` is true should be a separate exception.",
+      "reference": "this works and is the smallest possible change but we could definitely fail earlier"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Should not have snuck in here right? This is a golang 1.17 feature, wii definitely want to update these"
+    },
+    {
+      "prediction": "I think we can remove this, since we don't need it anymore.",
+      "reference": "Could you explain the logic with prevFolderId?"
+    },
+    {
+      "prediction": "I think we can remove this function and use `time.Now()` directly.",
+      "reference": "Please add license header"
+    },
+    {
+      "prediction": "`delete` is not needed here.",
+      "reference": "It bothers me that the user calls `initialize` without also calling `finalize`. It seems to me that we're essentially making `lbann_comm` a singleton object. Going further down this path, we would put `initialize` inside `lbann_comm`'s constructor and `finalize` in the destructor. This has it's own weirdness - the user must make sure it's lifespan matches the beginning and end of the program. We may also run into the case where someone might want multiple instances of `lbann_comm`, but this is not too important."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "We should probably make a unique_ptr<ROMol> typedef in ROMol.h and start using that liberally."
+    },
+    {
+      "prediction": "I think this should be `fillHandle: {autoInsertRow: false}`",
+      "reference": "Can you add missing asterisk?"
+    },
+    {
+      "prediction": "I think this should be `public static String capitalizeDigitsAfterNumbers(String toUpperCamel)`",
+      "reference": "late to the party, but I bet it meant to be `capitalizeLettersAfterNumbers` :)"
+    },
+    {
+      "prediction": "I think this should be `private boolean firstTimeEmptyObject = true;` instead of `private boolean firstTimeEmptyObject = false;`",
+      "reference": "The Activity is recreated on configuration changes and the value of the field is lost. So, e.g. pressing 'send' once will display the error message. If you rotate the device and press 'send' again, the message will show another time. Use `onSaveInstanceState()` and `onRetainInstanceState()` to save and restore the value of the field."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "Looks like a \"find and replace\" unintended change here"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `compute` method of the `sfdr.SpatialPooler`. The `stripUnlearnedColumns` method should only be called when the `sfdr` is in learn mode and the `spatialPooler` is being used.",
+      "reference": "haven't we got rid off `randomSP` recently? (I think `not learn` implied that)"
+    },
+    {
+      "prediction": "I think we can remove the `Status` field from the `{{.Type}}` struct as well.",
+      "reference": "Why we need use `in.GetStatus()` function here? Can we use `in.Status.xxxx` directly?"
+    },
+    {
+      "prediction": "I'm not sure if this is intentional or not, but I think it's a good idea to force invalidate when renewing a managed cert.",
+      "reference": "@cretz Just a thought: what if another renewal process updates the certificate between the beginning of this for loop (above on line 67) and actually calling RenewCert? Even though we have a read lock on the certCache, something else could have renewed it by now, and finished, which would cause this certificate to be renewed twice. Could you double-check my thinking here and see if that's a plausible 'race' condition? I wonder if the lock needs to be put over this whole function. (It's still early here; I could be wrong...)"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a function that takes a pointer to a service object and returns a pointer to the service object itself. This way you don't have to pass around a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer to a pointer",
+      "reference": "We don't need this for the PInvoke?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `scheduler.go` since it's only related to slack.",
+      "reference": "nits, I think add a log (using s.logger) to show why does this return with no error is better."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this check. I think it should be done in the `pipedProjectCache` cache.",
+      "reference": "Btw, It would be nice if we have some tests for those validation functions."
+    },
+    {
+      "prediction": "Should this be `public abstract String methodName();`?",
+      "reference": "Should this be called something that indicates its function, rather than its content? validationMessagePrefix, or similar?"
+    },
+    {
+      "prediction": "I don't think you need to change this line.",
+      "reference": "Please remove this line"
+    },
+    {
+      "prediction": "I think this should be \"Run the client\" and not \"Don't run the client\".",
+      "reference": "You could change the name of the command line switch too. Short options are typically one character after hyphen. wxWidgets' command line parser seems to handle `-nd` without confusing it with `-n` or `-d` but I'm not sure if that's by design or by accident. I'd remove the short option. `--no-daemon` is with hyphen but the rest of Manager and client uses underscore. Please change that."
+    },
+    {
+      "prediction": "This should be `flags.BoolVarP(flagSet, &fs.Config.CompareDest, \"compare-dest\", \"\", fs.Config.CompareDest, \"Include additional server-side path during comparison.\")`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be `tdnf error` instead of `tdnf error (will continue if the only errors are toybox conflicts)`",
+      "reference": "Why are we mentioning toybox in this message? AND it still says \"tdnf error\". What's the actual error? Should it be resolved instead of flagged?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to fix this. I think it should be fixed in the `client.js` file.",
+      "reference": "You can remove these if they've been resolved in NODE-3405 and this isn't depending on it"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I've been able to get it to work.",
+      "reference": "don't need the class here if its empty"
+    },
+    {
+      "prediction": "I think this should be `t('appointments.searchTitle')`",
+      "reference": "This should be computed('i18n.locale'...."
+    },
+    {
+      "prediction": "I think this should be `job_id` instead of `job_job_id`",
+      "reference": "Please add return and return type description to docstring of this method."
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "reference": "What does 1 << 32 do? In irb typing this yields 4294967296. I didn't find the append operator in Numeric or Integer."
+    },
+    {
+      "prediction": "I think this should be `succeeded` instead of `succeeded?`",
+      "reference": "We generally use metadata for this sort of thing, indeed there is a `metadata[:execution_result]`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the base class, since it's not specific to a particular plot type.",
+      "reference": "This was very confusing until I realized this might refer to the *bokeh* use of the word 'streaming'."
+    },
+    {
+      "prediction": "I think we should rename this to `containerEvents` or `containerEventsMap` to be consistent with `tasksToEvents` and `tasksToEventsLock`.",
+      "reference": "Can you rename this as `tasksToContainerStates` ? If you do that, you can rename the lock as well"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in `UserAccountManagerImpl` since it's only used by that class.",
+      "reference": "Fixing `lint` warnings that have existed for a while."
+    },
+    {
+      "prediction": "I think this should be `io.vavr.collections.Collections.fillObject(int start, int end, T element)`",
+      "reference": "An iterator **of {\\@code n} sequential elements,** where each element ~are~ **is the** given {\\@code element}."
+    },
+    {
+      "prediction": "This should be `nameof(IntegrationIds.AdoNet)`",
+      "reference": "If all of the `SqlCommandIntegrationName` values are the same, can we just remove it from `IAdoNetClientData` entirely and put the constant there?"
+    },
+    {
+      "prediction": "I'm not sure we should be using `cert-manager` as the prefix here. I think we should be using `venafi` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager` or `cert-manager",
+      "reference": "Maybe regroup this import with the other cert-manager packages."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Theme` class.",
+      "reference": "There are spaces missing ;) Just have a look at the code style of the `if` statement above"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `worker.go` file. The `worker.go` file is where the worker logic is.",
+      "reference": "We should pass the size limit as a parameter to Drain so that we don't take messages out of the pool and then drop them. It's ok to merge as is, I will file an issue."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. It seems like it should be in `emit_special_ibl_xfer()`, not `instr_encode_to_copy()`.",
+      "reference": "Patch lists are used in many places so this is going to break other updates. This should just be a local change in emit_special_ibl_xfer() to pass the second-to-last instruction instead of the last instruction."
+    },
+    {
+      "prediction": "I'm not sure if this is correct. It looks like it should be `7b42a00ee91d5810` but I'm not sure.",
+      "reference": "@argakiig previous work was above live threshold, causing the high difficulty to bleed into other blocks when genesis open was processed in tests."
+    },
+    {
+      "prediction": "Prefer `4.3` over `4.2`.",
+      "reference": "Metrics/LineLength: Line is too long. [89/80]"
+    },
+    {
+      "prediction": "I think this should be `Hidepid` instead of `HidepidDisable`",
+      "reference": "Please also extend the config tests now to include the new field."
+    },
+    {
+      "prediction": "I think this should be `this.RenderEmptyLiteral = false;` instead of `this.RenderEmptyLiteral = true;`",
+      "reference": "I think it's wise to keep the behavior backwards-compatible (so `false` as default)"
+    },
+    {
+      "prediction": "Do we need this here? I don't see it being used.",
+      "reference": "We don't need this line, do we?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "I don't know if we've actually had this problem on Upcase. I'd lean towards having a blank file unless (like on robots) we see multiple errors per week from this crawler."
+    },
+    {
+      "prediction": "I think this should be `0.29804` instead of `0.09804`.",
+      "reference": "I'm not good at picking colors, if anyone else has a preference."
+    },
+    {
+      "prediction": "I think this should be `cancelRefreshFoldersTimers` to be consistent with the name of the function.",
+      "reference": "Note that even if you cancel the timer, the refreshFolders function might still be running since it's async. Could that be a problem for the test units? One big issue I had with tests is they sometimes would work and sometimes fail randomly, and that's because there are still code running in the background. One example was the ItemChange class, which saves data asynchronously. So I've implemented a waitForAllSaved(), which can be used for test units, to make sure nothing is running once the test ends. So just wondering if we might end up with this issue here."
+    },
+    {
+      "prediction": "I think this test should be moved to `tsd-test` package.",
+      "reference": "can we make this something more descriptive? like... types should compile? I'm still not sure exactly what this is testing... this says \"Exported types\" - but what does tsd actually look at?"
+    },
+    {
+      "prediction": "I think we can remove the `onHideCustomView` method as well.",
+      "reference": "this method is need for full screen video button to appear in players"
+    },
+    {
+      "prediction": "I don't think we need a pointer here.",
+      "reference": "Why is this a pointer?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in `org.apache.iceberg.relocated.com.google.common.base.Preconditions`",
+      "reference": "nit: Space after comma"
+    },
+    {
+      "prediction": "I don't think this is needed. The `i2h` and `h2i` functions are already in `LongField` and `XLongField`.",
+      "reference": "A similar field is already define in `scapy/layers/bluetooth.py`. Can you merge both definitions ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the same file as the `#include` statements, since that's where the code is being generated.",
+      "reference": "`std::strtok` isn't necessary. The `std::string::find_first_of` is better C++ alternative to `std::strtok`. For example, see `CheckedError Parser::ParseEnumFromString` method."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "This property is already declared in a parent with the same visibility/type/value. This one is redundant."
+    },
+    {
+      "prediction": "I think this should be `def flush(self):`",
+      "reference": "This is required, otherwise multiprocessing will (for some reason) crash"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "Please put in a separate line-separated block"
+    },
+    {
+      "prediction": "I think this should be `fmt.Printf(\"Disconnect error: %s\\n\", err)` instead of `fmt.Printf(\"Disconnect error: %s\\n\", err)`",
+      "reference": "Should error be eaten? If so, not clear why"
+    },
+    {
+      "prediction": "I think this should be `self.force_empty_list = self.resource_name == 'bucket'` instead of `self.on_collection and self.resource_name == 'bucket'`",
+      "reference": "This attribute is not defined if the condition is not met. Instead, you could define another RouteFactory (e.g. `BucketRouteFactory` with a class attribute like `allow_empty_list`)"
+    },
+    {
+      "prediction": "I think we should use `WithInstalledTimestamp` to add the `InstalledTimestamp` field to `ClusterDeployment.Status` and not `ClusterDeployment.Condition`.",
+      "reference": "Can we use the existing `InstalledTimestamp` function?"
+    },
+    {
+      "prediction": "I think you can remove the `else if (isNutritionDataAvailable() && addProductNutritionFactsFragment.containsInvalidValue()) {` part of this if statement.",
+      "reference": "weird naming: `is` but `fields` (plural or singular ?) . If the method checks if the fragment has an invalid value, then \"hasInvalidValue\" or \"containsInvalidValue\" is fine no ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `user_bp` is a blueprint, which means it's a collection of `Blueprint`s, not a single `Blueprint`. So when you call `user_bp.route`, you're actually calling `timescale_connection.route` twice, which is not what you want. You need to call `timescale_connection.route` once, and then call `timescale_connection.route` again with the `user_bp` as the first argument.",
+      "reference": "Does this mean that `listenbrainz.org/user/iliekcomputers` will start getting 404s?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Line is too long. [82/80]"
+    },
+    {
+      "prediction": "I think this should be `async_select` instead of `async_select_unrequired`",
+      "reference": "That's a strange name =)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Thoughts on this? I'm not crazy about the brittleness of this but haven't found a good way to test more generically that a specific (custom) template has been rendered."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `objectTypes` since it's an object type, not a result type.",
+      "reference": "Have we reached binding-level agreement on the `subscribe` terminology? My only concern is that it doesn't seem descriptive enough and may be confused with subscribing for notifications."
+    },
+    {
+      "prediction": "I think this should be `if l != nil` instead of `if l != nil && l.SpanContext.IsValid()`",
+      "reference": "would it not be safer to move this condition to the `addLink` method?"
+    },
+    {
+      "prediction": "I think we should add a check for `req.ProtoMajor == 2` in `copyResponse`",
+      "reference": "I think this might read better: > some apps need the response headers before starting to stream content with http2, so it's important to explicitly flush the headers to the client before streaming the data."
+    },
+    {
+      "prediction": "I think this should be `PartitionSpec` instead of `StructLike`",
+      "reference": "What does this return if there is no partition? I think that we should consider adding a struct type that describes this tuple. That way, we can use an empty struct for unpartitioned and a non-empty struct for tasks that are combined by partition. We could also support more combinations, like combining across day partitions by not across bucket partitions or similar."
+    },
+    {
+      "prediction": "I think this should be `ErrRepoDoesntExist`",
+      "reference": "Also, it might be worth returning this for the above lookup on `kbfsRepoDir` as well, which could happen if someone tries to delete a repo in a TLF that has never had any repos at all."
+    },
+    {
+      "prediction": "I think this should be `Comma separated list of allowed domains. Prepend value with dot for wildcard mask`",
+      "reference": "Can this be `Hidden: true` as we already have a bunch of flags that are printed on `--help` and this seems like an average person should not care about it."
+    },
+    {
+      "prediction": "I think this should be `||` instead of `||:`",
+      "reference": "Wouldn't this set $strictVariables = true when Config::get returns false?"
+    },
+    {
+      "prediction": "Why is this being ignored?",
+      "reference": "was this meant to be checked in?"
+    },
+    {
+      "prediction": "I think this should be `time.Second` instead of `time.Hour` to be consistent with the other RTT measurements.",
+      "reference": "check spaces after const name..."
+    },
+    {
+      "prediction": "I think this should be `if driverVersion != \"\" && driverVersion != agent.resourceFields.NvidiaGPUManager.GetDriverVersion()`",
+      "reference": "If the GPU setting is enabled, resourceFields.NvidiaGPUManager will be nil, right?"
+    },
+    {
+      "prediction": "I think this should be `unless file_exist?(script)` instead of `unless file_exist?(script)`",
+      "reference": "We should probably verify that it's readable too"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "See, here's an example where you removed the variable but kept the call, which makes sense."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add the `request.isInitiatedByUser()`. I think it would be better to move this logic to the `FeedParserTask` itself, since it's not specific to the `FeedParserTask` and could be used by other tasks as well.",
+      "reference": "Please use `request.getTitle()` instead of `feed.getHumanReadableIdentifier()`: In this case, `feed` does not have a human readable title yet"
+    },
+    {
+      "prediction": "I think this should be `WARNING: cannot connect to clef signer: %v`",
+      "reference": "I think the wording on the left is better"
+    },
+    {
+      "prediction": "I think it would be better to use `append` instead of `i.interceptors = append(i.interceptors, interceptor)`",
+      "reference": "Please update the method described to point out that it maintains a chain of interceptors and they will be called in the same order passed in the options."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to check if `args.Name` starts with `Datadog.Trace, Version=` and if it doesn't start with `Datadog.Trace, Version=`, then it's not Datadog Trace and we should just return null.",
+      "reference": "Is this case purely hypothetical or did it solve an issue that occurred in your testing? I'm trying to understand this change a little better"
+    },
+    {
+      "prediction": "Why do we need to cast to an int here?",
+      "reference": "Use instead `convert.Int32ToString` that was added in #762 for this purpose."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `HashSet` is not serializable, so we can't use `tree.put(element, element)` here. Instead, we need to check if `contains(element)` first, and then call `tree.put(element, element)` if it's true.",
+      "reference": "If key is present, tree.put(k,v) needs to return a new instance for Maps and the same instance for Sets. Therefore we currently perform an additional 'contains' check for Sets. A future optimization may add an additional flag `replace` to the backing HAMT.put() / RedBlackTree.insert() methods. Sets set it to replace=false, Maps set it to replace=true. But for now this solution is working for us."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "This is a regression test for code I found while working on this."
+    },
+    {
+      "prediction": "nit: I think it would be better to use `new[]` instead of `char *[]` here.",
+      "reference": "I'm wondering if we need to delete jsonStr and jsonAttr after the json_extract_path_text call to avoid unnecessary heap pressure. Though if json_extract_path_text itself does new's on the same heap, we'd get heap fragmentation. Another approach would be to allocate these on the stack instead, avoiding both concerns: char jsonStr[len1+1]; char jsonAttr[len2+1]; ..."
+    },
+    {
+      "prediction": "I think this should be `int(request.args.get('from', 0))` instead of `int(request.args.get('from', 0))`",
+      "reference": "Does it come as `str`?"
+    },
+    {
+      "prediction": "I think this should be `for (int i = 0; i < allergens.size() - 1; i++) {`",
+      "reference": "This should actually read as the following `for (int i = 0; i < allergens.size(); i++)`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `AppendValue` function, not here.",
+      "reference": "This is so useful and safety, I want to propose making AppendBool, AppendInt, etc. all private functions, and forcing us to use AppendValue(j, values.New(false)), etc."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think it's better to check if ref is defined before calling createHandle() on it.",
+      "reference": "Really, really small nit   I believe there's 3x tabs in here? And should it be just 2x?"
+    },
+    {
+      "prediction": "nit: I think it would be better to have this as an option to the `aws s3v4` command itself, rather than to the `aws s3v4` provider.",
+      "reference": "Perhaps note that rclone can't do this yet?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/quic/transport.go` since that's where the quic stuff lives.",
+      "reference": "You probably don't need to set an empty config here."
+    },
+    {
+      "prediction": "I think this should be `util.datetime_types = util.datetime_types + cftime_types`",
+      "reference": "I think you are missing `cftime.DatetimeAllLeap` here. That said, all of these are subclasses of `cftime.datetime`, so I think you could get away with just using `cftime.datetime` here, rather than enumerating all of the different subclasses (since `cftime_types` is only used for instance checks)."
+    },
+    {
+      "prediction": "I think this should be `openebs.io/bd-predecessor` and not `openebs.io/bd-successor`",
+      "reference": "don't we need to set this label on pool pods? I don't see them being set"
+    },
+    {
+      "prediction": "I think this should be `throw std::runtime_error(\"k-NN brute force method is not implemented for CPU!\");`",
+      "reference": "can remove, it isn't reachable."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [132/120]",
+      "reference": "`expect(cancellation).to be_can_downgrade_instead` would be preferred :rainbow: :rainbow:"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `TransportSendLoop` since it's the loop that is responsible for sending packets.",
+      "reference": "Gross. The OS should timeout writes for completely unresponsive clients to begin with. Long term, the better solution is to enforce a minimum minimum data rate for responses. This might require a way to cancel LibuvAwaitables, but it definitely not OK to immediately kill the socket and any ongoing writes just because the request processing loop completed. You could consider doing that after some kind of timeout, but at that point you might as well just enforce a minimum rate."
+    },
+    {
+      "prediction": "I think this should be `std::this_thread::get_id()` instead of `io_thread_pool.get_local_thread_id()`.",
+      "reference": "If this is not used, we should just delete the line."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `exec` package.",
+      "reference": "Are there other callers to this function. How will they be impacted."
+    },
+    {
+      "prediction": "This should be `from plugin import NotReadyToAuthenticate`",
+      "reference": "Let's be consistent. The majority of boto code does not use relative imports. Let's just stick to the existing standard of \"from boto.package.subpackage import Thing\"."
+    },
+    {
+      "prediction": "I think it would be better to have this as a command line parameter instead of a CLI option.",
+      "reference": "Suggestion: specify `arity` for this option."
+    },
+    {
+      "prediction": "`show_legend = param.Boolean(False, default=True, doc=\"\"\"\r\n        Whether to show legend for the plot.\"\"\")`",
+      "reference": "This is a parameter available for the matplotlib backend IIRC. In which case, it is good to see this support added to the Bokeh backend."
+    },
+    {
+      "prediction": "`trim()` is not needed here.",
+      "reference": "trim has performance impact and this is the lowest level library... seems the trim if occurs should happen in the UI or Query controller instead.."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I can't think of a better way to do this.",
+      "reference": "Probably also needed for the `http.client.FOUND` below?"
+    },
+    {
+      "prediction": "I think it would be better to add a note that this option is experimental.",
+      "reference": "Can you also add this to Compiler.md ?"
+    },
+    {
+      "prediction": "I think this should be `@param {number=} retries The current number of retries.`",
+      "reference": "!numbers -> numbers"
+    },
+    {
+      "prediction": "I think it would be better to keep the format as \"com.google.orgname.artifact.name\"",
+      "reference": "This description doesn't seem quite right for the longrunning case - `longrunning` isn't an orgname."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `DiskFileCollector` since it's only used by that type of collector.",
+      "reference": "maybe give it a more descriptive name, e.g. `competitiveFilter`"
+    },
+    {
+      "prediction": "I think it would be better to add the keys into a master list only if the issuerConfig.usesHttpsJwk() is true. Otherwise, it would be better to add the keys into a master list only if the issuerConfig.usesHttpsJwk() is false.",
+      "reference": "This occurrence of \"master\" is unrelated to master/slave replication. Maybe simply remove the word \"master\" or replace it with \"reference\" but \"primary\" doesn't really make sense."
+    },
+    {
+      "prediction": "I think this is not needed anymore.",
+      "reference": "BLOCKING: This is a problem. We shouldn't be introducing new dependencies on proofs into actors. The miner has some dependencies that should be removed (#2555). This could be accomplished either by moving `proofs.Mode` to `types.ProofsMode` or by forgoing a new type altogether and replacing it with a boolean. The later would have the additional advantage of reducing the number of types we have to export through ABI. I get why `proofs.Mode` has value as a type over a boolean, but I'm not sure it's worth it here."
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix. I think we need to make sure that we don't have multiple messages with the same message id.",
+      "reference": "We probably want to log the fact that a `MessageId` was not present in the response. In addition the `RequestID` from `request.Request` should be included in the message so that the user is aware of the issue."
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "It feels a bit strange to re-assign that here, as that line itself has no context on why it's reassigned. I'd prefer if we could pass `{ propertyID: webPropertyId }` below instead (that's how we do that elsewhere too). This also applies to other similar re-assignments below - let's rather pass the values within the objects as applicable, since then there is more clear context on why this \"rename\" / \"reassignment\" is necessary (because of an object expecting specific properties)."
+    },
+    {
+      "prediction": "Can you move this to `frontend/express/public/core/events/javascripts/countly.models.js`?",
+      "reference": "remove this in the new pr aswell when moving compare to plugins as discussed."
+    },
+    {
+      "prediction": "nit: you can use `std::move(std::unique_ptr<...>)` instead of `return Status::Error(...);`",
+      "reference": "check varHolder is nullptr is redundant, ExecutionContext ensure"
+    },
+    {
+      "prediction": "I'm not sure if we want this to live here or in `app/services/unsubscribe_helper.rb`. I think it would make sense to move this to `app/services` since it's a helper.",
+      "reference": "Make these frozen constants?"
+    },
+    {
+      "prediction": "I don't think you need the `replace` here.",
+      "reference": "Please use something like html.escape() to ensure that the text is totally safe to include within html. &lt; is not enough."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `Peers` struct, but I'm not sure if that's the right place for it.",
+      "reference": "why is a buffered channel larger than 1 needed here? the separate goroutines can just try to write to the channel with a select-default block, and then it is not needed. i find the current implementation a bit convoluted, maybe we could simplify it a bit? not sure why it is needed for example to communicate an empty result. these patterns can be simplified using waitgroups"
+    },
+    {
+      "prediction": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "There seem to be more changes in here than I'd expect (mostly `limit: 255` additions)."
+    },
+    {
+      "prediction": "I think we can remove `TraceStateKeyMaxLength`, `TraceStateKeyTenentMaxLength`, `TraceStateKeyVendorMaxLength`, and `TraceStateValueMaxLength` from here.",
+      "reference": "I'm not sure, but I think it's `Tenant`"
+    },
+    {
+      "prediction": "I think this should be `ParseCustomKeyValues(string data, bool allowOptionalMappings = true)`",
+      "reference": "Not sure if it's an issue, but this is a breaking change in a public API. Maybe we should add as an overload without optional parameter instead and delegate the existing call to this one?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Outbound` struct, not the `OutboundConn` struct.",
+      "reference": "as opposed to exposing the chooser function and increasing our API exposure for the purpose of tests, can we move the HTTP transport config test into this package?"
+    },
+    {
+      "prediction": "I think this should be `wlr_drm.connector_set_dpms` and `wlr_drm.connector_set_mode` should be `wlr_drm.connector_set_dpms` and `wlr_drm.connector_set_mode` should be `wlr_drm.connector_set_mode`.",
+      "reference": "This is using the legacy DRM interface. DPMS levels have been removed from the atomic interface IIRC. Should we handle DPMS levels at all?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to do something like `watchs := make([]*transactionWatch, 0, len(tm.watches)) for _, watch := range tm.watches { if watch.nonce < nonce { watches = append(watches, watch) } } return watches`",
+      "reference": "I think this was correct before. The nonce passed in here is the nonce at a block (which is the next nonce not yet used, e.g. 0 if the account was never used, 1 if only the tx with nonce 0 has been sent). So if the in-block nonce is 12, then a transaction with nonce 12 cannot have been included yet."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do this. I'm not sure if it's possible to get rid of the `data` and `close` functions.",
+      "reference": "Please name the file ShareExtension.js, as you import it under this name (also could you convert it to TypeScript please?)"
+    },
+    {
+      "prediction": "I don't think we should have a default value for this.",
+      "reference": "Not sure if we need it, but if we set it, be sure to add `,omitempty`."
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this. It's not a header, it's a payload.",
+      "reference": "It's the whole `PublicHeader` that will change soon, and we should state that here, not in the member functions. If Jana's proposal for a new header is accepted, we might also want to rename the fields here. For example, there won't be a dedicated version flag anymore, so a more appropriate name might be `ContainsVersion` instead of `VersionFlag`."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `indexOf` can be called on a `CharSequence` without any bounds checking, which can lead to unexpected results. For example, if you have a `CharSequence` that is a substring of another `CharSequence`, calling `indexOf` on the first `CharSequence` will return the correct index, but calling `indexOf` on the second `CharSequence` will throw an `ArrayIndexOutOfBoundsException`. Instead, you should use the `indexOf` method with a custom implementation that checks the length of the `CharSequence` before calling `charAt` and throwing an `ArrayIndexOutOfBoundException` if the length is less than the required index.",
+      "reference": "I think `element` can stay of type `Character` because `Character` is a final class. The we do not need the cast in the line below."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I would move it to `ASTAnnotationTypeDeclaration` since it's only used there.",
+      "reference": "Not sure, if we should do it the other way round. If you ask be in the middle of the night, whether Interface should return true as \"isInterface\" and what Annotation would return. I'd tell you, Annotations should return false.... I want to say, on first glance, I'd say, it's easier to have \"isInterface\" and \"isAnnotation\" rather than \"isInterface\", \"isAnnotation\" and \"isRegularInterface\"... But I see, we also have \"isRegularClass\".... We still have the possibility to change the semantics of those APIs in PMD 7 *Update:* Or in other words: What is less surprising? That isInterface returns true for annotations or that it returns false?"
+    },
+    {
+      "prediction": "I think this should be `terminate_after = request.args.get('terminate_after', None)`, otherwise it will be interpreted as `terminate_after = request.args.get('terminate_after', None)`.",
+      "reference": "This will potentially skip package results. Do we really want that?"
+    },
+    {
+      "prediction": "I think this should be 720.",
+      "reference": "Please remove `HIP_UNSUPPORTED`"
+    },
+    {
+      "prediction": "I think this should be `MutableStateStats` instead of `MutableStateStatus`",
+      "reference": "if this is a \"Status\" type then the variable name \"Stats\" (which implies \"statistics\") seems wrong (or at least confusing to me)"
+    },
+    {
+      "prediction": "I think this should be `leakedInsecureKeystorePath`",
+      "reference": "maybe will be better to put it into the default worker's config rather than hardcoding the address?"
+    },
+    {
+      "prediction": "I think we can remove the `done()` call here since it's not being used.",
+      "reference": "It seems like this is not meant to test the option failure, but rather that `raw` can be set. This may be a superfluous test now, and we might want to delete it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be in `roots_view_set_fullscreen()`.",
+      "reference": "This is already done by `view_set_fullscreen`"
+    },
+    {
+      "prediction": "I think we can remove this function now.",
+      "reference": "mmm I followed the convention of this collection. It has Add, Remove, RemoveAll, so All sounds reasonable. Else It would likely be AddPeer, RemovePeer, RemovePeers and Peers."
+    },
+    {
+      "prediction": "Shouldn't this be `@return \\Shopsys\\FrameworkBundle\\Model\\Product\\Flag\\Flag|null`?",
+      "reference": "_nitpick_ I'm thinking about whether this method should be in the previous commit or not. I know it's not yet used there, but in theory, neither do UUID itself."
+    },
+    {
+      "prediction": "I think you can remove the `if (term_ != currTerm) {` check here.",
+      "reference": "\"break\" is right? You skip the line 768."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "Can this use the route helper instead?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Server` struct, since that's where it's used.",
+      "reference": "Why is this field exported and others aren't?"
+    },
+    {
+      "prediction": "I think this file should be under `/test/functional/`",
+      "reference": "should we rename this like `mongodb_srv_tests.js`?"
+    },
+    {
+      "prediction": "I think this should be `'' === googlesitekit.pageTitle ? '' :` instead of `const pageTitle = '' === googlesitekit.pageTitle ? '' : sprintf( __( 'Search Traffic Summary for %s', 'google-site-kit' ), decodeHtmlEntity( googlesitekit.pageTitle ) );`",
+      "reference": "This should still have a value when there is no page title, e.g. just `__( 'Search Traffic Summary', 'google-site-kit' )`."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Oops.. meant to link to SplitOp here. I'll clean up in my next commit"
+    },
+    {
+      "prediction": "I think this import should be removed as it is not being used.",
+      "reference": "Remove these unnecessary imports that you've added."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I'm not sure if it's a bad idea either.",
+      "reference": "That space shouldn't be here :wink: This lead to an exception whenever a page with an error was loaded, e.g. an inexistent host - I just fixed that :smile:"
+    },
+    {
+      "prediction": "I think this should be `self.remote_port = addr[1]`",
+      "reference": "It seems this is not the way to fix issue #505."
+    },
+    {
+      "prediction": "`pos_weight` -> `label_weight`?",
+      "reference": "Please take a look at the docstring of `F.binary_cross_entropy_with_logits`. `weight` should be a tensor that matches the input tensor shape. It is Not the class-aware weight. `pos_weight` should be a vector with a length equal to the number of classes."
+    },
+    {
+      "prediction": "I think we can remove the `testSuiteName` variable from the `testsMap` and just use `testSuiteName` from the `testconfig.Project`",
+      "reference": "Same here, merge with an existing test."
+    },
+    {
+      "prediction": "I think this should be `SLOT_INIT_TEMPLATE` instead of `SLOT_INIT_TEMPLATE  gSdMmcPciHcDriverBinding = {0, UnknownSlot, 0, 0, 0, 0}`",
+      "reference": "@aimanrosli23 Could you help to confirm if you do not revert the change in below commit: SHA-1: 643623147a1feaddd734ddd84604e1d8e9dcebee * MdeModulePkg/SdMmcPciHcDxe: Send SEND_STATUS at lower frequency"
+    },
+    {
+      "prediction": "I think it would be better to have a separate method for converting Avro schemas to Iceberg schemas. It would make it easier to test and ensure that the conversion is working as expected.",
+      "reference": "Can we do this without a breaking API change? What about adding a different name to convert directly to a Schema?"
+    },
+    {
+      "prediction": "I think we can remove the `callback` param from `connectToDb` now that we're using `expect` instead of `chai#expect`.",
+      "reference": "Why do we need to pull bluebird in here?"
+    },
+    {
+      "prediction": "I think it would be better to move the `FileIOUtils.deleteDirectorySilently(tempDir)` call inside the finally block, so that the directory is deleted even if an exception is thrown.",
+      "reference": "If *two* threads successfully download the same project into two different temp dirs, and one of them successfully renames one to its final destination, the `if`on line 102 will be false _in the other thread_, and it would leave its temp dir behind."
+    },
+    {
+      "prediction": "I think this should be a `struct pool_entry` instead of a `struct pool_entry_t`.",
+      "reference": "I would appreciate it if you could rename `link` to `all_link` so that the roles of the two links (the other is `target_link`) become clearer."
+    },
+    {
+      "prediction": "I think this should be `smi.create_ConfigProfileTriggerCommand(t1, True, expectedToBecomeIndex=1)` and `smi.create_ConfigProfileTriggerCommand(t2, True, expectedToBecomeIndex=2)` instead of `smi.create_ConfigProfileTriggerCommand(t1, False, expectedToBecomeIndex=1)` and `smi.create_ConfigProfileTriggerCommand(t2, False, expectedToBecomeIndex=2)`",
+      "reference": "Is there a reason not to use `smi.create_expectedIndex` here (and in other places)? The `smi.create_expectedIndex` function will also check to make sure that you don't accidentally skip or duplicate any indexes. With the goal of avoiding errors in the test. It does mean that you have to be diligent with how the expected indexes are created, which can be annoying too. My logic was that it is better to be annoyed by constraints during development time than to find out some time later that the test didn't do what I thought."
+    },
+    {
+      "prediction": "I think we can remove the `description` field from the `StripeSubscription` model as well.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I think this should be `bool`, not `*bool`.",
+      "reference": "I don't know if we even need this parameter."
+    },
+    {
+      "prediction": "I think this should be `missing-docstring` instead of `arguments-renamed`.",
+      "reference": "What is the new result of this file if we keep it the same than before? It's just to picture easily what changed in this MR :) (thinking is hard)"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure we're closing the stream properly.",
+      "reference": "This needs to be synchronous for the timeout to work."
+    },
+    {
+      "prediction": "I think we can remove the `should you have any questions or concerns` part of this sentence.",
+      "reference": "Can you break this onto multiple lines?"
+    },
+    {
+      "prediction": "I think this should be `0xf3800000` instead of `0xf3800010`.",
+      "reference": "OK, it looks like all children have bit 4 set, but it doesn't really matter at this split point: just informative, nothing reads it."
+    },
+    {
+      "prediction": "I think this should be `LinkToParentDomain string` instead of `LinkToParentDomain bool`",
+      "reference": "Sorry for the very basic question but why do we still have a DNSZone CRD in light of the DNSEndpoint, I don't fully understand the differences."
+    },
+    {
+      "prediction": "I think it would be better to use `assert.Equal(t, length, len(manager.Sessions))` instead of `assert.Len(t, length, len(manager.Sessions))`",
+      "reference": "Looks like loop with overhead. You can: - create SID - test private function `sessionIsUnique()`"
+    },
+    {
+      "prediction": "I think this should be prefixed with `prefQueueSortOrder_` instead of `prefQueueSortOrder`.",
+      "reference": "In my opinion, it would be better to have two separate settings. One that stores if sorting is enabled and one that stores the order. That allows to toggle automatic sort without having to re-select the order."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "/me can't wait to rename this repo :)"
+    },
+    {
+      "prediction": "I think this should be `const collLoader = makeLazyLoader(`${__dirname}/../collection`).`",
+      "reference": "nit: maybe call these `loadCollection()/loadDb()`"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it should go in the `Task` class itself, since that's where it's used.",
+      "reference": "Did you consider just do `return Task is an external data dependency and data does not exist (yet?).'`? Less `None`'s and `ifs`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I've been able to get it to work.",
+      "reference": "Some day, we really need to make these toolbars pluggable.."
+    },
+    {
+      "prediction": "I think this should be `self._show_source_html()`, since the `pygment` argument is only relevant for the HTML case.",
+      "reference": "This should be `pygment=False` like in `browsertab.py`"
+    },
+    {
+      "prediction": "I think this should be `c.msubs = opts.MaxSubs`",
+      "reference": "Should this be moved down (or up) to under the client lock instead of server lock?"
+    },
+    {
+      "prediction": "Why remove this?",
+      "reference": "Why are we removing this?"
+    },
+    {
+      "prediction": "I think this should be `\"#{self.cart.id}\"` and not `\"##{self.cart.id}\"`",
+      "reference": "What do you think about using `self.id` ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "Please log once here at Error level \"Failed to connect to Typha, will retry...\""
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Can you revert this change? The `VERSION` value will be updated after maintainer determines the version at release."
+    },
+    {
+      "prediction": "I don't think this is a good name for this key. It's not a key, it's a qualifier. It should be called something like \"qualifiedServiceName\", \"qualifiedServiceVersion\", \"qualifiedServiceRole\" or something like that.",
+      "reference": "It could be better if we use the up case letter to specify the constant string name. Such as QUALIFIED_SERVICE_NAME_KEY can let the user know it's a constant string."
+    },
+    {
+      "prediction": "I think this should be `return @configuration.failure_exit_code if examples_count == 0` instead of `return @configuration.failure_exit_code if examples_count && @configuration.fail_if_no_examples`",
+      "reference": "This should be `@configuration.failure_exit_code`, we don't want to hard code 1."
+    },
+    {
+      "prediction": "This should be `@protected`.",
+      "reference": "Changed per review in Blockly."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the L3 routing table.",
+      "reference": "Hi @wenyingd . so what's the original dst MAC of the reply packet from kube-proxy?"
+    },
+    {
+      "prediction": "I think it would be better to add a `help` argument to `BaseChecker` instead of adding a new option here.",
+      "reference": "Note: I wasn't sure what the right wording would be for this, so I used existing language from elsewhere in the codebase."
+    },
+    {
+      "prediction": "I think this should be `instance['vm_name'] = utilities.format_instance_name(instance['name'], platform, self.config['vagrant']['instances'])` instead of `self.config['vagrant']['instances'] = {utilities.format_instance_name(instance['name'], platform, self.config['vagrant']['instances']) for instance in self.config['vagrant']['instances']}`",
+      "reference": "Docstring for `platform`."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to do this. I think it's better to do it in the `cleanup()` function.",
+      "reference": "This should be called before any `os.Exit` otherwise it could still mess up things."
+    },
+    {
+      "prediction": "This should be `public static final int RED = 1;` and `public static final int GREEN = 2;`",
+      "reference": "shouldn't this be `short` ?"
+    },
+    {
+      "prediction": "I think we can remove the `Capacity` from the `out` array.",
+      "reference": "Can you please paste the output of `volume list` command which will show capacity of a openebs volume."
+    },
+    {
+      "prediction": "I think it would be better to throw `IllegalArgumentException` instead of `IllegalStateException`",
+      "reference": "I'd make this a `throw new NullPointerException(\"metric == null\");` instead"
+    },
+    {
+      "prediction": "This should be moved to the top of the file.",
+      "reference": "Include should be order lexicographically."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to store the pin length. I think it would be better to store it in the `OAuthWebviewHelper`'s `setPinLength` method.",
+      "reference": "We need to pass in the Activity context here, so that LoginActivity is used, and so that LoginActivity can get onActivityResult from the PasscodeActivity. Without this, the application context is used, and we don't get a callback when the passcode is done."
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.",
+      "reference": "Can you move these to the top please, for consistency with other models - normally we put scopes immediately after the associations at the top of the model. As to names I agree with @gravitystorm that these names may be confusing but I'm not sure the ones I suggested are perfect either so I'm not really sure what's best. The goal of my names was just to separate point which could be ordered (as `ordered`) from those which can't (as `unordered`) but I realise that may not be very clear from the names."
+    },
+    {
+      "prediction": "Shouldn't this be `if (shouldRepeat && !item.getFeed().getPreferences().getCurrentAutoDelete())`?",
+      "reference": "Looks like the episode should also not be deleted when repeating"
+    },
+    {
+      "prediction": "I think this should be `github.com/elastic/beats/v7/x-pack/elastic-agent/caddy/caddyhttp/httpserver` instead of `github.com/mholt/caddy/caddyhttp/httpserver`",
+      "reference": "I feel like goimports would move this import line by the other non-std packages... can you double-check that? Run goimports or set up your editor to run it on save. :+1:"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I don't have a better idea. It seems like we should be able to do something like `@capabilities[:chrome_options] = { user_data_dir: \"#{profile_json[:directory]}\", extensions: profile_json[:extensions] }`",
+      "reference": "`options['binary'] = WebDriver::Chrome.path` if set?"
+    },
+    {
+      "prediction": "I think this should be `SQLITE3:SQLITE3{SQLite3File: \"./explorer.db\"}` and not `SQLITE3:SQLITE3{SQLite3File: \"./explorer.db\", SQLite3File: \"./explorer2.db\"}`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if there is a better way to do this.",
+      "reference": "Nit: Supported -> Enabled."
+    },
+    {
+      "prediction": "I think this should be `csv.at[0, 'Int0'] = 42` instead of `csv.set_value(0, 'Int0', 42)`",
+      "reference": "Yay! I was too lazy to fix this."
+    },
+    {
+      "prediction": "I'm not sure if this is the right package to use here. I think we should be using `errors` from `pkg/errors` instead of `errors` from `github.com/pkg/errors`.",
+      "reference": "other declaration of errors (from `typecheck`)"
+    },
+    {
+      "prediction": "I think this should be `githubRepoExp.MatchString(val)` instead of `githubRepoExp.ReplaceAllString(val, \"https://github.com/$1\")`",
+      "reference": "nit: error starts with capital letter"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `Flag` class has a `valueOfByRealName` method that should be used here.",
+      "reference": "the original code ignores \\Draft and \\Recent, whereas your change will not. what this matters, i'm not sure."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's the only way I could come up with.",
+      "reference": "Why was this necessary?"
+    },
+    {
+      "prediction": "I think this should be `rerun` rather than `rerun` because it's a verb and not a noun.",
+      "reference": "Can you also do this for Cover?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add this check. I think it should go in `ValidateImageSubresourceRange`.",
+      "reference": "So I tried adding this VU a long time ago, thought it would be this simple of a check, but turns out there were other validation in the way for getting here in `ValidateImageAspectMask` I assume that is what is failing CI here, realized it would require some more refactoring"
+    },
+    {
+      "prediction": "I think this should be `coalesce_and_sum` instead of `coalesce_and_sum_and_sum`",
+      "reference": "This method is not needed for postgresql, by default sum function omits null values so we can remove this file"
+    },
+    {
+      "prediction": "This should be `systemdsystemgeneratordir=` instead of `systemdsystemgeneratordir=$($PKG_CONFIG --variable=systemdsystemgeneratordir systemd)`",
+      "reference": "Can you provide a `--with-systemdsystemgeneratordir` option here? So that I can do unprivileged installs without completely turning off systemd. I can add it as a follow-up PR too."
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to do.",
+      "reference": "`classification` is misspelled"
+    },
+    {
+      "prediction": "I think this should be `methodViewBuilder.initCode(methodSampleViews.get(0).methodInitCode());`",
+      "reference": "Can you change this to `sampleInitCode`?"
+    },
+    {
+      "prediction": "I think we can remove the `startTs` field from the `TIDAGRequest` as well.",
+      "reference": "I added a startTs information in `TiDagRequest` output. Not really sure if it is useful? @zhexuany"
+    },
+    {
+      "prediction": "I think this should be `width = parseInt(frame.node.getAttribute('width'), 10) || 0` and `height = parseInt(frame.node.getAttribute('height'), 10) || 0` instead of `width = parseInt(frame.node.getAttribute('width'), 10);` and `height = parseInt(frame.node.getAttribute('height'), 10);`",
+      "reference": "This prevents margin/border/padding CSS from affecting the width/height (otherwise in Chrome a width and height of 1 would result in a rect size of 5x5)."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "prefer to just add this to the `verify_authorized` exception list above since we're already making exceptions."
+    },
+    {
+      "prediction": "I think it would be better to use `array_walk` here.",
+      "reference": "was there a reason to not loop over parent classes here?"
+    },
+    {
+      "prediction": "`computed` is not used in this file.",
+      "reference": "@billybonks Why move `import Ember from 'ember';` to the top?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now. I'm not sure if there is a better way to do this.",
+      "reference": "Everything's so simple (a `ThreeStateWorkflow`) until it isn't :smirk:"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. The problem is that `op_contours` is a list of `OperationContour` objects, not a single `Contour` object. So you can't compare `op_contours[0]` to `contour[0]` as you did here. You need to compare `op_contours[0].pts` to `contour[0].pts` and `op_contours[0].colors` to `contour[0].colors` and `op_contours[0].labels` to `contour[0].labels` and `op_contours[0].labels` to `contour[0].labels` and `op_contours[0].labels` to `contour[0].labels` and `op_contours[0].labels` to `contour[0].labels` and `op_contours[0].labels` to `contour[0].labels` and `op_contours[0].labels` to `contour[0].labels` and `op_contours[0].labels` to `contour[0].labels` and `op_contours[0].labels` to `contour",
+      "reference": "I don't quite understand where the NaNs come from..."
+    },
+    {
+      "prediction": "I think we can remove the `MsgLogsCleaningInterval` and `MsgLogRetention` as they are no longer used.",
+      "reference": "File is not `goimports`-ed"
+    },
+    {
+      "prediction": "Shouldn't this be `//_please:wheel_resolver`?",
+      "reference": "I wonder if we should look at moving the Python stuff to a plugin and doing this there rather than adding to core plz at this point?"
+    },
+    {
+      "prediction": "I think this should be `inventory_file` instead of `inventory-file`",
+      "reference": "This is the actual bug fix. I prob should have broken this out into two PRs."
+    },
+    {
+      "prediction": "Shouldn't this be `IsEip3238Enabled`?",
+      "reference": "I am not sure if we need this, we already have DifficultyBombDelays"
+    },
+    {
+      "prediction": "I think this should be disposed in the `Dispose` method of the blockchain contract.",
+      "reference": "move to finally - I can see at least one return that would omit it, also any exception and so on"
+    },
+    {
+      "prediction": "why do we need this check?",
+      "reference": "remove `genesisConfig.GravityChainStartHeight == 0`"
+    },
+    {
+      "prediction": "This should be `__DIR__ . '/../vendor/autoload.php'` instead of `__DIR__ . '/../' . $symfonyDumpFunctionPath`",
+      "reference": "this might kill whole application if `var-dumper` will not be installed. What about two separate file_exists conditions (monorepo/project)?"
+    },
+    {
+      "prediction": "I think this should be `CreateSettingsFile()` instead of `app.CreateSettingsFile()`",
+      "reference": "Hrm. Maybe this isn't the correct spot to run this? Because it errors if the app type doesn't support settings. But we want to check that before running it."
+    },
+    {
+      "prediction": "`sortBy` should be an int, not a string.",
+      "reference": "Why are you changing this name ?"
+    },
+    {
+      "prediction": "I'm not sure if this is enough. I think we need to check that the user is not `localstack` and that the current user is not `localstack` as well.",
+      "reference": "as we discussed, maybe it would be better to - [ ] set OVERRIDE_IN_DOCKER=1 in the Dockerfile - [ ] replace the if statement with `os.path.exists(\"/run/.containerenv\")` as @dfangl suggested"
+    },
+    {
+      "prediction": "I think this should be `is_being_called` instead of `is_being`",
+      "reference": "I like this style of doing things better than having super big conditional without explanation. Even though I know this won't be re-used it's nice to have a function if only for clarity."
+    },
+    {
+      "prediction": "I think this should be `unless @group_level > 0` instead of `if @group_level > 0`",
+      "reference": "Our preference here is to either have inline statements e.g. `@group_level -= 1 if @group_level > 0` but stylistically here I think `return @group_level if @group_level == 0` might be nicer. As it stands however our rubocop is picking this up as a fail :("
+    },
+    {
+      "prediction": "I think we can ignore this test for now, as Spark3 is not supported, so we need to fix it in the parent class `TestRemoveOrphanFilesAction`.",
+      "reference": "Spark3 Is supported in OSS Iceberg, Do you mean this particular function is not supported in Spark3? If so I think we should have a more clear reason why"
+    },
+    {
+      "prediction": "I don't think we need this function at all. We can just call `io.read_text()` on the `os.path.abspath(notebook_path)`.",
+      "reference": "something about the arbitrary file access part here makes me nervous. Can we keep the `if not path.endswith(\".ipynb\")` check from before here as well? Just imagining a weird attack scenario where somebody somehow uses this to get access to confidental files or something"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Blacklight::Solr::SearchBuilder` class.",
+      "reference": "I wonder if we should push this and the search_params_logic into the blacklight config."
+    },
+    {
+      "prediction": "I think this should be `1000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "don't need this line, since it is not tested here?"
+    },
+    {
+      "prediction": "I think this should be `ami_version` instead of `ami_version=\"1.0\"`.",
+      "reference": "I wonder if adding this new param in the middle of the param list might screw up some existing code that is depending on the old parameter order. Might be better to add it to the end of the parameter list?"
+    },
+    {
+      "prediction": "I think this should be `String.format(\"%s/build/(build|src/generated.*)/.*\", project.getProjectDir().getAbsolutePath())`",
+      "reference": "I wonder if this absolute path is gonna affect the build-cache relocatability of these compile tasks??"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `resolveWriteConcern` method, since that's where the `WriteConcern` is being set.",
+      "reference": "possible bug if `fullResponse` is false"
+    },
+    {
+      "prediction": "I think this should be `ascii_lowercase + ascii_uppercase`",
+      "reference": "nitpick: Please only `import string` here and use `string.ascii_lowercase` - I really like to see what module stuff is coming from without looking at the imports."
+    },
+    {
+      "prediction": "I'm not sure if this is what we want to see, but I think it's what we want to see.",
+      "reference": "nitpick: the only reason we have a catch for every one of these promises is because the test isn't properly chaining promises (returning them within each `then`). We could reduce the `catch` with `console.dir` to a single instance if that was done"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "why do this?"
+    },
+    {
+      "prediction": "This should be `System\\MediaLibrary::class`",
+      "reference": "For consistency, would it not be best to also replace the first class name? `traceLog('Class ' . static::class . ' has been deprecated, use ' . SystemMediaLibrary::class . ' instead.')`"
+    },
+    {
+      "prediction": "I think this should be `EnableTSDBAdminApi` to be consistent with the rest of the Prometheus API.",
+      "reference": "this should be `EnabledAdminAPI` and the json string should then be `enableAdminAPI`"
+    },
+    {
+      "prediction": "I think this should be `from sqlalchemy import engine as engine` instead of `from time import sleep`.",
+      "reference": "Noticed some missing imports in manage.py so thought I'd add them"
+    },
+    {
+      "prediction": "I think this is the only change that needs to be made to the code. The rest of the code looks good.",
+      "reference": "as a future improvement, maybe we should just change `ForAllFlowRecordsDo` so that `updateOrSendFlowRecord` uses a flow record pointer instead of a copy of stored flow record. This whole code is executed with the lock any way."
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to represent. It's not a literal string, it's a Unicode character.",
+      "reference": "I am not sure what GitHub wants to tell me with this icon here :)"
+    },
+    {
+      "prediction": "I think this should be `reference_1{1, 2, 12}; auto result_1 = parseParameters<TileParameters>(\"tile(1,2,12).mvt\");`",
+      "reference": "can we also check for smaller zoom levels returning valid error codes?"
+    },
+    {
+      "prediction": "I don't think we need to add these methods to the base class. We can just add them to the derived class.",
+      "reference": "1) These names are not great, doesn't help explain what they do. 2) Instead of methods like this that hard code some number of = and *, why not a method that takes the string, defaults to a char to surround with, and a width? 3) Do we really even need headers like this in the output? Maybe we ditch them."
+    },
+    {
+      "prediction": "Why did you change this test?",
+      "reference": "The autocomplete editor shouldn't change its own behavior, I assume that `autoWrapCol` and `autoWrapRow` options should be set to `false` in handsontable editor."
+    },
+    {
+      "prediction": "I don't think we should be setting this to `true` here. I think we should be setting it to `false`.",
+      "reference": "out of curiosity, with IPv6 introduced, why there's no change to the expected output of openflow in the test?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. It seems like it would be better to do it in the decode routine, but I don't have a good idea of what that would look like.",
+      "reference": "nit: grammar: instruction"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "We can revert this to reduce the diff of the PR."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `ReconcileHiveConfig` itself.",
+      "reference": "i think this missed the reconciler to callsite change"
+    },
+    {
+      "prediction": "I think this should be `getNotImplementedString(\"SurfaceNamer.getClientConfigImportModule\");`",
+      "reference": "Please reuse `getClientConfigPath`."
+    },
+    {
+      "prediction": "I think we can remove the `matchers` folder entirely. It's no longer used.",
+      "reference": "Should we use the shared chai-addon.js file here? Actually we should consolidate any `chai.use` calls, I see 20 instances, quick win?"
+    },
+    {
+      "prediction": "I think this should be `c.OrderByDescending(c => c.ProtocolCode).Last())`",
+      "reference": "Maybe move _agreedCapabilities and _availableCapabilities from List to SortedSet ? We could order them by code and then by version already."
+    },
+    {
+      "prediction": "I'm not sure if this is correct, but I think it is correct enough for now.",
+      "reference": "I see the `+1` issue, but we used to check if `o.lss != nil` and we don't anymore, so we are sure that it won't be nil here, right?"
+    },
+    {
+      "prediction": "I think this should be `<%= document.send(:export_as_mla_citation_txt).html_safe %>`",
+      "reference": "why is the `send` needed?"
+    },
+    {
+      "prediction": "`if isinstance(bboxes, torch.Tensor):` -> `if torch.is_tensor(bboxes):`",
+      "reference": "Is this necessary? If so, we need to update the docstring."
+    },
+    {
+      "prediction": "I'm not sure if this is the right approach, but it's the only one I could come up with.",
+      "reference": "Checking for actionable here instead of grabbing the first pending approval. I tested this with proposal ID 265 on staging."
+    },
+    {
+      "prediction": "I think it would be better to use `assertEqual` instead of `self._assert_result`",
+      "reference": "I made this change because I wasn't getting printouts of the 'result' value when the assert was inlined for some reason. I think it's related to creating new CallbackCheckers maybe?"
+    },
+    {
+      "prediction": "I don't think we need this interface at all. The only callers of `HttpParser` are `HttpParser.ParseStartLine` and `HttpParser.ParseHeaders`, which are both implemented by `HttpParser`.",
+      "reference": "nit: the correct name is \"request line\", so let's use the right one here :)"
+    },
+    {
+      "prediction": "I think this should be `ALLOW_INSECURE_CATALOG_ACCESS`",
+      "reference": "This should now use `[]`, so it crashes if the var is missing"
+    },
+    {
+      "prediction": "This should probably be `return -DER_NONEXIST;` instead of `return -1;`",
+      "reference": "you could use `bpf_map_update_elem` with `BPF_EXIST`"
+    },
+    {
+      "prediction": "I think this should be `throw new UncheckedIOException(\"Failed to create output stream for location: \" + uri(), e);`",
+      "reference": "thanks for noticing this! But I think we should not have it in this PR."
+    },
+    {
+      "prediction": "This should be `Input.DEFAULT_LOCKTIME_SEQNUMBER + 1`",
+      "reference": "A few things: - This may not serialize because I think the property for the sequence number on an input is `.sequenceNumber`. - There should be tests to cover these statements - Indentation is incorrect (should be two spaces, as similar to the statements above)"
+    },
+    {
+      "prediction": "I think this should be `data_format not in cls.supported_data_formats`",
+      "reference": "If this is a base method that could be used by others, should you hardcode 'GCS uploader' here?"
+    },
+    {
+      "prediction": "I think this method can be simplified to `return this.padTo(length, element);`",
+      "reference": "just a little thing: let's introduce a `final int bound = length - back.length();` that saves us calculating the bound each time checking the loop condition."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put these endpoints. I think they should be in the `balance` router, since they are related to peer balances. But I'm also not sure if they should be in the `balance` router, since they are related to settlement balances. I think they should be in the `settlements` router, since they are related to settlement balances, but I'm also not sure if they should be in the `balance` router, since they are related to peer balances. I think they should be in the `balance` router, since they are related to peer balances, but I'm also not sure if they should be in the `settlements` router, since they are related to settlement balances. I think they should be in the `balance` router, since they are related to peer balances, but I'm also not sure if they should be in the `settlements` router, since they are related to settlement balances. I think they should be in the `balance` router, since they are related to peer balances, but I'm also not sure if they should be in the `settlements` router, since they are related to settlement balances. I think they should be in the `",
+      "reference": "I think that this should be a POST method, as it changes the state. Get requests should not change the state. With different method, I would suggest to have the handler under `\"/settlements/{peer}\"` path."
+    },
+    {
+      "prediction": "This should be `0x4200000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "This is the default value?"
+    },
+    {
+      "prediction": "I think this should be `from datetime import strftime`",
+      "reference": "The spaces are actually legitimate here: in python, you: - first import standard library modules (hint: time is one) - then import 3rd party library modules - then import the modules from you project (here kinto). Between each block you need an empty line."
+    },
+    {
+      "prediction": "I think this should be `if name := obj.Name(); name != \"\" && name != \"\" {`",
+      "reference": "Haha, whoa, I'm not sure how I let that one go by. :)"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think it would be better to have a list of supported file extensions in the `MimeTypeDetectionByFileExtension.properties` file, and then just check if the file extension is in that list.",
+      "reference": "FWIW: We have FileUtil.generateOriginalExtension for tabular files but it uses a hardcoded list rather than the properties file."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Kubernetes` section, since it is specific to Kubernetes.",
+      "reference": "I would consider extracting this logic in to a helper function to make it a bit more readable, but that's just my opinion."
+    },
+    {
+      "prediction": "I'm not sure if this should be configurable or not. If it should be configurable, then I think we should have a default of `false`. If it shouldn't be configurable, then I think it should be `true` by default.",
+      "reference": "Are all fields beginning with \"Debug\" automatically `;local` ? (I guess so, but just checking.)"
+    },
+    {
+      "prediction": "`DeprecationWarning` is not a real warning, it's just a deprecation message.",
+      "reference": "We should let this module inherits from the EvalHook supported in MMCV, and only override the `__init__` function by adding the warnings. The other functions can use those in MMCV."
+    },
+    {
+      "prediction": "I don't think this is the right place to set the role. This should be done in the `setContainerRole` method of `PostgreExecutionContext` which is called from `setSessionRole` method of `DBRProgressMonitor`. The reason is that the `setSessionRole` method is called from the `executeQuery` method which is called from the `setContainerRole` method which is called from the `openSession` method which is called from the `executeQuery` method which is called from the `setSessionRole` method which is called from the `openSession` method which is called from the `executeQuery` method which is called from the `setContainerRole` method which is called from the `openSession` method which is called from the `executeQuery` method which is called from the `setSessionRole` method which is called from the `openSession` method which is called from the `executeQuery` method which is called from the `setContainerRole` method which is called from the `openSession` method which is called from the `executeQuery` method which is called from the `setContainerRole` method which is called from the `openSession` method which is called from the `executeQuery` method which",
+      "reference": "Let's use prepared statements here instead"
+    },
+    {
+      "prediction": "I think it's better to use `import React from 'react';` and `import { Button } from 'yoast-components';` instead of `import { Button } from '../assets/js/components';`",
+      "reference": "Just leaving this in one place but thanks so much for re-organising all these imports."
+    },
+    {
+      "prediction": "I think this should be `#if defined(__x86_64) && __x86_64__ != 0` instead of `#if defined(__x86_64) && __x86_64__ != 0`",
+      "reference": "Ditto, here and below"
+    },
+    {
+      "prediction": "`typeof vnode.type!=='function' && vnode.props[key]!=null` should be `typeof vnode.type!=='function' && vnode.props[key]!==undefined`",
+      "reference": "Let's move the check outside the `for`-loop so that we don't have to do it for every prop. If we change it to check for `vnode.type === 'string'` we can automatically skip checking text nodes which don't have props"
+    },
+    {
+      "prediction": "I don't think this is a good name for the hook list. I think it should be something like `finalize_column_hooks` or `finalize_column_hooks_list` to indicate that it's a list of hooks, not a list of hooks.",
+      "reference": "Seems like the docstring here should simply be \"Deprecated; use `hooks` instead.\"."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could get it to work. I'm not sure if there is a better way to do this.",
+      "reference": "It would be nice to have an `attr` in `styles.xml` for this, so we don't need the same `if` statement in multiple places"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Line is too long. [81/80]",
+      "reference": "How about `null: false` on these two lines? Also `t.timestamps null: false`?"
+    },
+    {
+      "prediction": "I think it would be better to log the full exception here, so that the user can see the full stack trace.",
+      "reference": "This recommendation isn't very helpful because it isn't clear what \"the metastore\" is. How about this instead: \"To fix this, use an alternative metastore\"."
+    },
+    {
+      "prediction": "I think this should be `contains` instead of `containsKey`",
+      "reference": "The `SymbolTable` deals only with symbols, I think we can just call this `isUsed` here."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for CORS. I think it would be better to check for `url[0]` being a `/` or `window.location.protocol + '//' + window.location.host'` being the same as `origin + url`.",
+      "reference": "Please hide this behind a flag and add a setter function to the builder."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to get the route table from the RouteTableAssociation and then check if it's Main or not.",
+      "reference": "Just a side-node, no need to change it here: The term \"recent\" might be confusing at a later point in time."
+    },
+    {
+      "prediction": "nit: `Object.keys(val1).map(key => `${errorMessage ? `${errorMessage}: ${key}` : key})`",
+      "reference": "When can it be undefined? Wouldn't that be a bug?"
+    },
+    {
+      "prediction": "I don't think we need to change anything in this file.",
+      "reference": "JMX offers normal APIs for this. I don't think you want the glassfish one."
+    },
+    {
+      "prediction": "I think this should be `F.percentile(q).alias(\"{}_percentile_{}\".format(colname, int(q * 100)))` instead of `alias(\"{}_percentile_{}\".format(colname, int(q * 100)))`",
+      "reference": "Could you remove `,' at the end?"
+    },
+    {
+      "prediction": "I think this should be `assert(typeof opt_port !== 'function', `",
+      "reference": "Maybe it would simpler to ignore opt_port if type !== 'number'?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Account` struct, since that's where the account data is stored.",
+      "reference": "do you want to rename the codec tag here too?"
+    },
+    {
+      "prediction": "I think we should rename this to `START_CAST_TAB_MIRRORING` as it is not related to media-routers.",
+      "reference": "These command names are specific to Chromium-based browsers. Please move to `ChromiumDriverCommand`"
+    },
+    {
+      "prediction": "I think it would be better to use `getattr` instead of `if not hasattr(..., ...)`: `logger = getattr(F, name, getattr(functions, name))`",
+      "reference": "Hm.. should we auto-patch? I am less sure on this yet as `percentile_approx` seems a Spark function and auto-patching it in Koalas seems a bit counterintuitive."
+    },
+    {
+      "prediction": "Why not use `ServiceProviderTestCase`?",
+      "reference": "I found domain use in `AllFeedsTest::getAllFeedExportCreationData` and class AllFeedsTest is descendant of `FunctionalTestCase`. I think so we can rewrite this too.."
+    },
+    {
+      "prediction": "I think it would be better to use `ImmutableList` instead of `BasicTypeInfo.STRING_TYPE_INFO` here.",
+      "reference": "I am wondering if it is better to consolidate all committer checkpoint states into a single structure, e.g. Pojo class or Avro record which are Flink supported state types for schema evolution."
+    },
+    {
+      "prediction": "I think this should be `log_signal` instead of `debug`.",
+      "reference": "I assume this is a performance optimization to only check this once when creating the signal filter? While you're at it, maybe make `BLACKLIST` a set instead?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it would be better to do this in the `installExtension` method of the extension node.",
+      "reference": "Please use async/await and try/catch"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the same place as the nested flatbuffer check.",
+      "reference": "Like I said, this code looks incorrect. Why is it still looking up the attribute? Why is it getting the type from `nested->type.struct_def` instead of `field.nested_flatbuffer`? Are you running `generate_code.sh`, because I think `nested->type.struct_def` would be null?"
+    },
+    {
+      "prediction": "nit: you can remove the `return` statement here.",
+      "reference": "Actually, I'm not sure why lines 1663-1668 are here. Looks like Push descriptors aren't used in the test at all. @ziga-lunarg?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `wrapIfKeywordOrBuiltIn` is used by `wrapIfKeyword` and `wrapIfBuiltIn`, both of which are used by `convertToCommentedBlock`. I think we need to fix `wrapIfKeywordOrBuiltIn` instead of `convertToCommentedBlock`.",
+      "reference": "Don't we still want to check it it's a keyword?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "this can just me `link_text` :tomato:"
+    },
+    {
+      "prediction": "I think this test should be moved to `test_series_argsort.py`",
+      "reference": "What if the Series contains null values?"
+    },
+    {
+      "prediction": "I think we can remove the `Uri = TestAnalysisTarget.CreateUriForJsonSerialization()` call here as well.",
+      "reference": "Every place you see a change like this, it's because `PhysicalLocation` is now its own object, not an array of `PhysicalLocationComponent` objects. The `Uri` is the key into the `FileInfo` dictionary."
+    },
+    {
+      "prediction": "Why do we need this change?",
+      "reference": "Like in atss_head"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the pool, but I'm not sure how to do it.",
+      "reference": "nit : I think it would be better to have the multiplier being calculated along with `pool.numPendingWholeBlocks`, so that we could always multiply the feePerByte with that multiplier. Not only that it would be faster, but it would also have constant time. btw - Correct me if I'm wrong, but with this change, you're practically rolling back your previous change while making it `pool.expFeeFactor` adjustable, right ?"
+    },
+    {
+      "prediction": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Is there a reason we _wouldn't_ want to do this? A bit of a security hole :grimacing:"
+    },
+    {
+      "prediction": "I think this should be `public static ProjectSpec load(File projectSpecFile)` instead of `public ProjectSpec load(File projectSpecFile)`",
+      "reference": "Please add a copyright notice to all new files. Is there a tool to check and auto update copyright notices?"
+    },
+    {
+      "prediction": "I think it should be `e-mail` instead of `E-mail`",
+      "reference": "A new translation dump should be made after this change (removing the \"Email\" message)"
+    },
+    {
+      "prediction": "I think this should be `map(implicitSelector => implicitSelector + ':not([role])')` and not `map(implicitSelector => implicitSelector + ':not([role])')`",
+      "reference": "CI is failing saying `expected [TypeError: null is not an object (evaluating 'implicitNodes(role).map')] to equal null`."
+    },
+    {
+      "prediction": "I think this should be `thing = this.props` instead of `thing = this`.",
+      "reference": "@JoviDeCroock I replaced the refs here with simple assignments."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "nit: Could we use VisibleForTesting annotation here?"
+    },
+    {
+      "prediction": "Layout/AlignParameters: Align the parameters of a method call if they span more than one line.",
+      "reference": "Can we create some post records to assert sort_by"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do here. I think we should reset the stream before we attempt to connect to it.",
+      "reference": "is this necessary?"
+    },
+    {
+      "prediction": "I think this should be `public List<Func<Span, bool>> Filters { get; private set; } = new List<Func<Span, bool>>();`",
+      "reference": "What is the purpose of this? It seems like it is used to indicate whether there is a filter on a span. But what filter? Could there eventually be multiple filters per span? Just not understanding."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `Condition` struct, since it's specific to the condition itself. But I'm not sure if there's a better place to put it.",
+      "reference": "using a certain type instead of using string directly."
+    },
+    {
+      "prediction": "I think this test should be renamed to `aliceAndBothRenameSameFile` to be consistent with the other tests.",
+      "reference": "fix trailing ,"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "I've fixed it for this change but for future reference there's no point doing string substitution for a single value like that - it will just slow things down. Just use `:mlat => trace.latitude` etc instead."
+    },
+    {
+      "prediction": "Why is this class not inheriting from `ProductAssert\\UpdateAttributeFormModel`?",
+      "reference": "The form no logger needs validation?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that we don't export metrics that are not present in `metricsToExport.Metrics` and that we don't export metrics that are present in `metricsToExport.Metrics` but not present in `this.getMetrics`",
+      "reference": "there'd be more changes in the area, when we implement \"clean-up\" of aggregators if no new updates. (i think we currently send metricitem to exporters with zero as value)"
+    },
+    {
+      "prediction": "`self.assertIn('id', resp.json['details']['resource_name'])`",
+      "reference": "Usually we only keep the assertions that are relevant to the specification (ie. the test title). For example, the test on the values of id and details is not properly relevant for the spec _collections are isolated by bucket_. I suggest that you only keep the assertions regarding `details` in dedicated tests similar to `test_unknown_collection_raises_404` from `test_views_records.py`, but for each of bucket, group, collection and record in the other `test_views_` files."
+    },
+    {
+      "prediction": "I don't think this is correct. It should be the other way around, i.e. `access type dependant on constructor type arguments`",
+      "reference": "The test sources are ok though, you don't need to remove those changes"
+    },
+    {
+      "prediction": "I think this should be `width * 2`",
+      "reference": "`maxWidth: width * 2,` Potentially, this will be the same as `getImageUrl`: 80px or 500px. I can't verify this - need for normal LiveTV tuner."
+    },
+    {
+      "prediction": "Shouldn't this be `toLowerCase()` on line 165 as well?",
+      "reference": "Not really related to this issue but should we use fuzzy search here as well (like for apps)?"
+    },
+    {
+      "prediction": "I think this should be `forceQuotaReclamation` instead of `forceQR`.",
+      "reference": "isn't the usual capitalization for subcommands force-qr? (don't care too much about it)"
+    },
+    {
+      "prediction": "I'm not sure about this one. I think it's fine, but I don't want to make a big deal out of it.",
+      "reference": "Do you know how many repos have hit this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the filter results. I think it would be better if the filter results were displayed in the same way as they are displayed on the usage dashboard, which is to say that the filter results are displayed in descending order of the number of records.",
+      "reference": "I'll be interested to see if our gettext rake tasks will pick up on this text to translate"
+    },
+    {
+      "prediction": "This should be `setField(L, -1, \"params\", spell.getHasParam());`",
+      "reference": "I think you need to change line 928 value from 6 to 7 too, since now table will have 7 values"
+    },
+    {
+      "prediction": "I think you can just do `assert set(f1.names) == {\"cat\", \"dog\", \"mouse\", \"dog, cat\"}`",
+      "reference": "Does `f0[::-1, :]` mean all the rows taken in a reverse order?"
+    },
+    {
+      "prediction": "I don't think you need to change this line.",
+      "reference": "Looks like there are strange characters here."
+    },
+    {
+      "prediction": "Shouldn't this be `!chaosy.GetPause()`?",
+      "reference": "It is better to have a check to ensure the format is valid."
+    },
+    {
+      "prediction": "I think it would be better to have this as a constructor parameter instead of a public property.",
+      "reference": "How about `nextAlreadyDefined` variable name in method signature?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it should be moved to `integration_test.go`",
+      "reference": "In older SDK versions the data converter was called 3 times but now it is 2. This is because logic was changed deep in the interceptor chains on when values are serialized and encoded and such. Previously we invoked the data converter on results _even if there was an error_ (and most of the time they were nil). Now we don't do such a thing, resulting in fewer calls."
+    },
+    {
+      "prediction": "Can you add a license header to this file?",
+      "reference": "formatting... can this import go next to the other sources ones?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this option. I think it should go in the `--attach` subcommand of the `podman attach` command.",
+      "reference": "I don't understand why anyone would pass 0? If you don't want to attach to an existing process you just wouldn't pass -attach. I would remove that sentence and make 0 an invalid argument."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "This change is responsible for #2775"
+    },
+    {
+      "prediction": "I don't think we need to do this. We can just use the `exception_type:CustomException1` string in the test.",
+      "reference": "Should we add this to the other \"never\" verification in lines 71-73 above, too?"
+    },
+    {
+      "prediction": "I think this should be `list(self.dao.iter_cai_assets(ContentTypes.resource, 'google.cloud.bigquery.Dataset', '//cloudresourcemanager.googleapis.com/projects/{}'.format(project_number), self.session))`",
+      "reference": "Fixed this since it no longer needed to be switched from a generator to a list, this was missed when I cleaned up the dataset code in an earlier PR."
+    },
+    {
+      "prediction": "I think we should use `LOG.info(String.format(\"Adding %s %d times\", caps, info.getMaximumSimultaneousSessions()));` instead of `LOG.finest(String.format(\"Adding %s %d times\", caps, info.getMaximumSimultaneousSessions()));`",
+      "reference": "This is an informational message that allows someone to read the console output and understand how the grid node is configured. Please leave."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Can we convert this to an arel as below ContributorFact .joins(:project).where(projects: { id: project_id}) .where('name_id = ? or name_id in (?)', name_id, AnalysisAlias.select(:preferred_name_id) .joins(:project) .where(commit_name_id: name_id) )"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `libpod/cgroup`",
+      "reference": "Does this cause issues if you have lots of load on the machine? Having a global variable for this seems like a bad idea IMO."
+    },
+    {
+      "prediction": "Shouldn't this be `PrivacyParametersBuilder.setEnclavePublicKeyUsingFile(privacyPublicKeyFile());` instead of `PrivacyParametersBuilder.setEnclavePublicKeyUsingFile(privacyPublicKeyFile());`?",
+      "reference": "We probably don't need to print the msg with e.getMessage(). It will be printed by the logger since we are passing the exception as a parameter."
+    },
+    {
+      "prediction": "I don't think you need this fixture at all. You can just do `mocker.patch('molecule.verifier.testinfra.Testinfra._get_tests')`",
+      "reference": "Can we get a test with multiple directories as per the initial example?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You need to check if the object is a status bar, and if it is, you need to check if it is the STATUSBAR object, and if it is, you need to check if its role is STATUSBAR_ROLE.",
+      "reference": "I assume this is because you were seeing cases where left was a negative number? When?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "this is undoing recent changes for increased consistency of include file statements. please restore to the original."
+    },
+    {
+      "prediction": "I think this should be `OptionalOrElseMethodInvocation` to be consistent with the rest of the codebase.",
+      "reference": "If these are applied in order, should `LambdaMethoReference` come after `OptionalOrElseMethodInvocation`"
+    },
+    {
+      "prediction": "I think this should be `IEnumerable<FileData>`, not `IDictionary<string, FileData>`.",
+      "reference": "Why this rename?"
+    },
+    {
+      "prediction": "I think this should be `if mock_main and sys.version_info < (3, 3) and sys.modules.pop(\"__main__\")`",
+      "reference": "Why do we do this check here? `sys.version_info` cannot be less than 3.3 as pylint does not support older versions. I think we should avoid popping `__main__` altogether."
+    },
+    {
+      "prediction": "I think this should be `--verbose` and `--debug` instead of `debug` and `--debug`",
+      "reference": "This can be removed too."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `Put` method of the `RootMetadataSigned` struct. The `Put` method is called by the `Put` method of the `MDServerMemory` struct, which is called by the `Put` method of the `RootMetadataSigned` struct, which is called by the `Put` method of the `MDServerMemory` struct, which is called by the `Put` method of the `RootMetadataSigned` struct, which is called by the `Put` method of the `MDServerMemory` struct, which is called by the `Put` method of the `RootMetadataSigned` struct, which is called by the `Put` method of the `MDServerMemory` struct, which is called by the `Put` method of the `RootMetadataSigned` struct, which is called by the `Put` method of the `MDServerMemory` struct, which is called by the `Put` method of the `RootMetadataSigned` struct, which is called by the `Put` method of the `MDServerMemory` struct, which is called by the `Put` method of the `RootMetadataSigned` struct, which is",
+      "reference": "I'm a little worried about the new extra calls to `MakeMdId`, since in most local-server KBFS profiles, hashing is the biggest CPU user, and not re-using the value calculated by the client will result in even more hashing. That said, I understand why you did it this way, and I don't have a concrete suggestion on how to make it better. Just something to keep in mind; plus it might be worth comparing the test speed before and after this change to make sure it didn't slow things down too much."
+    },
+    {
+      "prediction": "I think we can remove the `__` and `wp.i18n` imports from this file.",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "I think this file should be named `miranda32.py`",
+      "reference": "Just curious, but how did you work out that Aleksey Sadovoy, Peter Vgner and Joseph Lee contributed to this file? Did you look at the repository history?"
+    },
+    {
+      "prediction": "I think this is the only place where `ReadStop` is called.",
+      "reference": "You could call OnStopRude() directly here to do the same thing."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `app-service` package.",
+      "reference": "This TODO and this `sort` depend on how the UI looks like. So I think we should fix the UI before continuing on updating this web API. For example, in case the UI flow is as below, this API will receive a Piped ID and then return just ones belonging to that Piped. - User picks where they want to deploy first (select one Piped and one cloud provider) - Then API will be called to show the list of suggested apps for that Piped and cloud provider - User picks app they want to create"
+    },
+    {
+      "prediction": "I don't think this is a good name for a help text. It's not a help text, it's an explanation of what a command does.",
+      "reference": "This command rolls back the volume data to the specified snapshot. Once the roll back to snapshot is successful, all the data changes made after the snapshot was taken will be post. This command should be used cautiously and only when there is an issue with the current state of the data."
+    },
+    {
+      "prediction": "I think this should be `fmt.Sprintf(\"ecs-functional-tests/awslogs/%s/%s\", containerId, taskId)`",
+      "reference": "should this be a separate/new functional test?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "Can these also assert at least the length of arrays returned and ideally contents?"
+    },
+    {
+      "prediction": "I think this should be `getattr(task, 'param_visibilities', None) or _get_default(param_visibilities, {})`",
+      "reference": "`hasattr(task, 'param_visibilities')` is more concise with the same effect, but either one works."
+    },
+    {
+      "prediction": "I think this should be \"Positional arguments are out of order, the arguments are being passed in a different order\"",
+      "reference": "I would hedge this a little, maybe `Positional arguments appear to be out of order`. After all, it's just a guess."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `AccountActivity` class, since it's specific to that screen.",
+      "reference": "you can directly call show() method from **SnackBarHandler.show(coordinatorLayout, getString(no_account_signed_in)).show();** no need to create an extra object and then call the show method."
+    },
+    {
+      "prediction": "I think this should be `folly::optional` instead of `std::string`.",
+      "reference": "concatHosts may be a better func name to explain this?"
+    },
+    {
+      "prediction": "This should be `const stubString = () => '';`",
+      "reference": "A function that returns empty string."
+    },
+    {
+      "prediction": "I think this should be `else if filename != \"\"`",
+      "reference": "If it's not the root package, or if it is the root package but we don't add the psudo :pleasings rule there, print an error."
+    },
+    {
+      "prediction": "nit: can you move this line to the top of the file?",
+      "reference": "can delete this line"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it would be better to have a function like `get_model_dir()` that returns the model directory, and then call that from `get_local_file_dir()`. That way you don't have to pass the model directory to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't have to pass the model to `get_local_file_dir()`, and you don't",
+      "reference": "Will it still be an issue if all the models use the same data set?"
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "reference": "How does this controller_name method work? Does this method grab the context of the url request? Wouldn't this always result in 'widget'?"
+    },
+    {
+      "prediction": "I think this should be moved to `flb_fs_chunk.h`",
+      "reference": "Let's get rid of these for the final version and if you get rid of the macro I left last time I'd be more than grateful."
+    },
+    {
+      "prediction": "I think it would be better to use `Integer.MAX_VALUE - (101 - 200)` instead of `Integer.MAX_VALUE - (200 - 101)`",
+      "reference": "Instead of `DELETE_MARK`, how about `IS_DELETED`? I don't think that \"mark\" is clear enough to describe what this is. Similarly, I think the docs should be \"Whether the row has been deleted\". There's no need to include \"delete mark\" because that's identifying something that is not defined (this column is _deleted and \"mark\" is not introduced), and \"or not\" is unnecessary because it is implied by \"whether\"."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. It should be done in the build system, not in the compiler. I think it would be better to move this logic to the build system and have the build system call the compiler to generate the headers.",
+      "reference": "Does it make sense to have this very thin wrapper that only calls `create_dependencies`? And maybe having `create_dependencies` return a set would be better than convert here?"
+    },
+    {
+      "prediction": "I think it would be better to make this a `size_t` instead of an `int` so that it can be used with `sizeof` instead of `sizeof(int)`.",
+      "reference": "How about renaming `escape` to `use_path_normalized` or something to better indicate that it is a selection between `path` and `path_normalized`?"
+    },
+    {
+      "prediction": "I think it would be better to have a `@Nullable Accountable` instead of a `@Deprecated Accountable`",
+      "reference": "Interface members are always `static final` right?"
+    },
+    {
+      "prediction": "This is not exported. I think it should be.",
+      "reference": "It seems to me that this could be in the runtime_nintendoswitch.go file, or is there a reason to keep it separate?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "I was actually looking at this code over the weekend and wondering whether we were trying to link here. I assume it's been cargo culted from something else in the compiler."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this, but I don't have a better suggestion.",
+      "reference": "nice extraction of the form for being used at new/edit. We should have more of these"
+    },
+    {
+      "prediction": "I don't think you need to change this line.",
+      "reference": "Are the coefficients the actual bug fix?"
+    },
+    {
+      "prediction": "I think this should be `psycopg2.ProgrammingError` instead of `IntegrityError`",
+      "reference": "Why are the error code options in a list and not a tuple?"
+    },
+    {
+      "prediction": "I think this should be `TEST_BAD_ANCESTORS = [Project('123'), Folder('456'), Organization('ABC')]`",
+      "reference": "Need alignment with the preceding line."
+    },
+    {
+      "prediction": "I think this should be `fromConfig(GenesisConfigFile.getMainnetConfigOptions(), new HashMap<>(), false)`",
+      "reference": "Changing from a static method to a constant is inessential to Type Transactions and has a long reach, adding many unneeded files and diff lines."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem here is that `comparator` can be `null` if `values` is `null` as well. I think we need to add a check for `comparator != null` before calling `RedBlackTree.ofAll`.",
+      "reference": "There is one caveat: javaslang.collection.Iterator is Iterable. With Iterator it does not work. Let's revert this line. Does this happen elsewhere, too? I will double-check the code..."
+    },
+    {
+      "prediction": "I think it would be better to have `format_amount` and `format_satoshis` in the same file. It would make it easier to see what's going on under the hood.",
+      "reference": "Why change the default value of `is_diff`?"
+    },
+    {
+      "prediction": "I'm not sure that this is the best way to do this. I think it would be better to use `>=2.7.5` and `<3` instead of `>=2.2.0` and `<3`.",
+      "reference": "I think tacking on `,<7` would be a wise precaution in case of future incompatible changes in tenacity."
+    },
+    {
+      "prediction": "Why did you remove the `?array $orderBy = null`?",
+      "reference": "I don't think is necessary to remove the return type"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to decrypt the client secret. I think it should be in the `OAuth2.refreshAuthToken` method.",
+      "reference": "Also added this check, since client secret is not used except in the IP bypass scenario."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `getProfiles` resolver, since that's the place where the profile ID is being fetched from the server.",
+      "reference": "That's unrelated, but shouldn't there also be a call to `setInternalWebPropertyID`, making it empty, for this case? Otherwise, when selecting to create a new property, any previous internal web property ID will still be in state, potentially causing problems."
+    },
+    {
+      "prediction": "I'm not sure if this is the right test to have. I think we need to test that we can run `ddev login` and `ddev logout` and that we can run `ddev logs` after logging out.",
+      "reference": "I wonder if we should trigger a PHP error and ensure it ends up in the log?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "The `move_topic` action be a `post` request since it changes the state of the object."
+    },
+    {
+      "prediction": "I think this should be `metadata = new(core.BuildMetadata)`",
+      "reference": "Should we just return a non-nil metadata from `Test`?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Noticing this is really just the same as cart_notification_email. Would be nice to be able to view custom templates. I may update this with some set up that will make viewing specific custom emails easily."
+    },
+    {
+      "prediction": "I think it would be better to have this in a separate PR.",
+      "reference": "please don't use wildcard imports"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think it would be better to have a `sourceDataReference` that is set to `null` before the test starts, and then set it to the actual data when the test is done.",
+      "reference": "Should we even fix this test? IMO it should be refactored to check that values are recalculated after `setSourceDataAtCell` and `setSourceDataAtRowProp` calls. This will be the correct way to alter the data since now and we're not testing this use case."
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "I'm not sure if this is standard practice, but I would prefer to leave this initialization here, if only to show that such a variable exists in this file (instead of inferring it from the globals). It can also cause weird issues if, for example, `init` isn't run before other methods."
+    },
+    {
+      "prediction": "This should be `throw std::logic_error(\"Value of nbits is Negative\");`",
+      "reference": "This isn't how you throw an exception in C++. I'd suggest you take a look elsewhere in the RDKit C++ codebase and see how ValueError exceptions are thrown. Hint: there's a function called `throw_value_error()` that you may find useful."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test_transactions.cpp`.",
+      "reference": "Was changing this to a write transaction required?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure it's necessary. If we want to keep it, I think it would be better to add a `must` field to `PaymentVoucher` and set it to `map[string]interface{}{}` or `nil` in the `Create` function.",
+      "reference": "nit: I think \"They\"?"
+    },
+    {
+      "prediction": "I think this should be 1_099_511_627_777",
+      "reference": "is it a result of some multiplication or some other meaningful number? would be nice to display as a power of something or the mult"
+    },
+    {
+      "prediction": "I think we can remove `FULFILLABLE_FEATURES` now, since it's no longer used.",
+      "reference": "I think this class is a good example of where Payload could significantly improve things. I'm okay with this class as it stands. If you merge this in, I'd like to have a go at refactoring it after introducing Payload."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Just a small tip: I think you don't need to include pytest here. You can run tests using: - `pytest tests/path/to/my/test` under the virtualenv to run a single test. - `pytest tests/` under the virtualenv to run all the tests. - `make tests` to run all the tests for all python versions, like travis do. Hope this helps! :)"
+    },
+    {
+      "prediction": "I think this should be `sanitizer` instead of `sanitizer_`",
+      "reference": "Might be worth sorting the replacement items so this is deterministic."
+    },
+    {
+      "prediction": "I don't think we need this envconfig, we can just set the default value in the constructor.",
+      "reference": "Should we set the default `TargetNamespace ` to `chaos-testing`? I noticed that the installation script defaults to chaos-testing."
+    },
+    {
+      "prediction": "Why was this removed?",
+      "reference": "Why was this `Task.Yield` removed? It was added in purpose."
+    },
+    {
+      "prediction": "I think this should be `getRequiredFieldConfigs` instead of `getRequiredFields`",
+      "reference": "You'll also want to do this below, for the `optionalFieldConfigs`. See `GapicMethodConfig.java:180`."
+    },
+    {
+      "prediction": "`schemaManager->toEdgeType` can return `nullptr` if the edge is not a valid one.",
+      "reference": "We'd better use \"spaceId\" as variable name here. Because i can't figure out the really type for \"space\" at the first glimpse due to \"auto\" used."
+    },
+    {
+      "prediction": "I think this should be `RootDirectory / \"packages\" / \"libsqreen.1.1.2.2\" / \"1.1.2\"`",
+      "reference": "I assume we have a package reference somewhere that pulls down the nuget package?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to set up the event tracking. I think it should be in the `remoteUpdates` goroutine, but I'm not sure if that's the right place to set up the event tracking.",
+      "reference": "But why introduce publish on `accConnsEventSubjNew` subject if we never subscribe on that subject. What is the plan then? Replace subscription on \"old\" with \"new\" at one point? Not sure about all that.."
+    },
+    {
+      "prediction": "I think this should be `fpga_token_init()` instead of `fpga_token_fini()`.",
+      "reference": "can we use `nullptr` for consistency?"
+    },
+    {
+      "prediction": "Why do you need this `Shopsys` prefix?",
+      "reference": "strange, for templating\\EngineBundle it was `\\Symfony\\Bundle\\FrameworkBundle\\`"
+    },
+    {
+      "prediction": "I think this should be `||` instead of `(!added_new) && is_merge_or_booted`.",
+      "reference": "But if we're on the merge deployment, and the next one is the booted deployment, we'll still be inserting between the merge and booted deployment, right? It seems like we would need e.g. `met_merge` and `met_booted` vars to keep track."
+    },
+    {
+      "prediction": "I think this should be `return (axe.utils.tokenList(virtualNode.attr('role')).filter(role => !axe.commons.aria.isValidRole(role, { allowAbstract: true }).length > 0).length > 0`",
+      "reference": "Can you put the output of this into data, and list which roles are invalid? Same thing with abstract role. I know that's not strictly part of this PR, but it'd make the messaging about this a bunch better."
+    },
+    {
+      "prediction": "I think we should keep the name `ibftTwo` as it was in the previous version.",
+      "reference": "rename ibftTwo arg to bft"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for `skip_session_tracking`",
+      "reference": "Should we go ahead and push `agent_is_crawler?` into the `skip_session_tracking?` method?"
+    },
+    {
+      "prediction": "`AttributeError`: Please upgrade mmcv to >= 1.3.16 `AttributeError`: You are trying to access a attribute of an object that has been deleted.",
+      "reference": "How about we raise a warning here and simply change to use self.data_infos = self.load_annotations(self.ann_file)"
+    },
+    {
+      "prediction": "This file is covered by the GNU General Public License. See the file `COPYING` for more details.",
+      "reference": "Please remove this line to comply with the standard for copyright headers."
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. It should be part of `AuRaValidator` class, not `AuRaValidatorsCollectionExtensions`.",
+      "reference": "seems incorrect - for 5 validators it will tell that 3 is enough to seal and you need 4 2/3 * 5 is 3.3"
+    },
+    {
+      "prediction": "I don't think we need to import `Executors`, `ScheduledExecutorService` and `ScheduledFuture` here. We can just use `java.util.concurrent.Executors` and `java.util.concurrent.ScheduledExecutorService` directly.",
+      "reference": "Nit: please remove unnecessary whitespace changes."
+    },
+    {
+      "prediction": "I think it would be better to rename `resource_` to `resource` in the Node.js client library generator as well.",
+      "reference": "Is this captured in a test?"
+    },
+    {
+      "prediction": "I think this should be `protected`.",
+      "reference": "This is used in o.a.l.sandbox.search.LargeNumHitsTopDocsCollector."
+    },
+    {
+      "prediction": "why grad_clip is None?",
+      "reference": "We may also do the same for `schedule_2x.py` and `schedule_20e.py`."
+    },
+    {
+      "prediction": "Align the elements of a hash literal if they span more than one line.",
+      "reference": "Should foreign key here be something like `diary_entry_id`? Or above, in `diary_entry.rb`, it should be `diary_entry_id`?"
+    },
+    {
+      "prediction": "I think this should be `map[string]struct{}` instead of `map[string]struct{}`",
+      "reference": "session is a new term. Should we call more like currentCommandBatch or something else?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea.",
+      "reference": "Will go fmt before squash."
+    },
+    {
+      "prediction": "I think this should be set to `false` by default. If we want to allow privileged containers, we need to set this to `true`.",
+      "reference": "Setting this to `true` means that it will always get merged in. The way merging is done here is that if a value is its zero value (`false` for `bool`), the value is considered unchanged. In order for this to work, you'll need to change this to be a `*bool` type instead."
+    },
+    {
+      "prediction": "why remove this?",
+      "reference": "It was decided that blockSignerAddress is not only redundant (use the miner owner public key instead which is already stored), but does not belong in config."
+    },
+    {
+      "prediction": "I think we can remove `jsonschema` from the `pip install` requirements.",
+      "reference": "are we not asking for trouble here by not pinning this? or does 3.* imply all of the draft versions we'd try to validate?"
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to do.",
+      "reference": "No, we want to compare `tm_py` and `tm_cpp` in this test."
+    },
+    {
+      "prediction": "I think this should be `100Mb` instead of `1000Mb`",
+      "reference": "Let me know if we'd rather have this as an env variable."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could think of.",
+      "reference": "question: what is happening here?"
+    },
+    {
+      "prediction": "I think this should be `@Override` and not `abstract protected`.",
+      "reference": "\\[Checkstyle\\] ERROR: 'protected' modifier out of order with the JLS suggestions\\."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `client_encryption_test.js` file, under the `client_encryption` module.",
+      "reference": "Can we use EJSON from bson here? and in doing so avoid bringing in the deprecated `mongodb-extjson` lib `const { EJSON } = require('bson')`"
+    },
+    {
+      "prediction": "I don't think this is a good idea. It's not uncommon for people to want to display tick labels on the x-axis. I think it would be better to have a `show_xaxis` parameter that defaults to `True` or `False` depending on whether tick labels should be shown on the x-axis.",
+      "reference": "So this parameter is now deprecated?"
+    },
+    {
+      "prediction": "(style) Macros with complex values should be enclosed in parentheses",
+      "reference": "Actually, it looks like HAVE_RSEQ is only used to determine whether the rseq.h header is around, which then only determines whether the regression test is built: it does not affect whether the core handles an application using rseq. The core rseq mangling code and all the code in rseq_linux.c is always enabled. I believe the idea is that we want DR built on one machine to work on as many others as possible, rather than requiring a custom build for every variant of kernel feature or whatnot. What is the top-level problem being solved here? If it requires disabling or enabling the core handling rseq, probably a separate option should be put in place and maybe HAVE_RSEQ renamed to make it clearer what its consequences are."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "I don't see a corresponding call to freeSelectorSet... am I missing something?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good change, but I don't have a better idea at the moment.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure if we want to keep the `DatabaseTypeKey` and `DatabaseNameKey` here, but I don't think we need to worry about it right now.",
+      "reference": "Is this part of spec?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. It's too late to wait for the app to be started. We need to wait until dynamo_thread_init is called.",
+      "reference": "Hmm, do we need to do this for Windows too? It won't have this private loader TLS issue but it will run DR code in a separate thread before DR init is fully done which I think violates some assumptions. For Windows we would move this wait from win32/os.c to win32/callback.s intercept_new_thread where it checks whether it's a client thread right before calling dynamo_thread_init. If you do not want to deal with Windows you could bail on my suggestion and leave the issue open (or maybe file a new issue) for Windows, or if you want to put the suggested core/win32/ change in place and your new test passes on Appveyor we could then claim to have fully solved this related issue too for all platforms."
+    },
+    {
+      "prediction": "nit: I think it would be better to use `lambda_name = attributes[-1]` instead of `lambda_name = attributes[-1] if len(attributes) > 1 else None`",
+      "reference": "We can remove this function and use `lambda_function_name(..)` in this file instead."
+    },
+    {
+      "prediction": "I think it would be better to have a `NewSenderWithApplicationVersion` instead of a `NewSender`",
+      "reference": "nitpick: `ApplicationVersion` could be `AppVersion`. It's smaller but gives the same understanding for purposes of the field."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pollWorkflowTaskQueue.go`",
+      "reference": "nit: maybe prefer `resp = &workflowservice.PollWorkflowTaskQueueResponse{}` over adding another function exit points especially for non-error cases?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could come up with at the time of writing this.",
+      "reference": "Do we specifically need to keep the read preference as a symbol for use elsewhere? If not, I think it would be cleaner to just modify the options before caching them in the `client_options` method so that we don't do this check every time. If we do need it a a symbol elsewhere, I'd suggest either putting `return @client if @client` at the top of the method (and changing the assignment below from `||=` to `=`) or moving this within the parenthetical block in the last assignment so that we don't repeat this work."
+    },
+    {
+      "prediction": "`preferredMimeType` is not supported in `Webcam` extension. You can use `mimeType` instead.",
+      "reference": "Since this is for video only, should it be called `preferredVideoMimeType`? If we add it for pictures later, it will likely need to be a different option."
+    },
+    {
+      "prediction": "I think this should be `LayoutEntry` and not `ImageSubresourceLayoutMap::LayoutEntry`",
+      "reference": "That doesn't make sense. LayoutEntry doesn't match the generator concept."
+    },
+    {
+      "prediction": "I think we need to make sure that this is not used in the app deploy, right?",
+      "reference": "I think the customer visible strings should be \"Service\" instead of \"Svc\" to make it obvious to them"
+    },
+    {
+      "prediction": "I think this should be `true` by default.",
+      "reference": "Explain why this property is necessary?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `ExecuteContract` function.",
+      "reference": "It should accept executions as the input"
+    },
+    {
+      "prediction": "`batch_size` is not a good name for a variable. Maybe `batch_size` should be changed to `batch_num` or `num_buttomms`?",
+      "reference": "Is this limitation necessary?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be under `ACMEIssuer` since it's a part of the ACME stack.",
+      "reference": ":question: Should there be validation to ensure that only one of these attributes is used?"
+    },
+    {
+      "prediction": "This should be renamed to `VufindSearch\\Backend\\SolrBackend`.",
+      "reference": "I would recommend using Psr\\Container\\ContainerInterface here instead of Interop\\Container\\ContainerInterface; both currently work, but the Interop version is deprecated and will eventually be removed, so using the Psr version will save us time in the future."
+    },
+    {
+      "prediction": "I think this should be `assert.Contains(out, \"/var/www/html/docroot\")`",
+      "reference": "This would also change back to /var/www/html/docroot if we go that way."
+    },
+    {
+      "prediction": "Why is this public? It's not used anywhere else in the codebase.",
+      "reference": "Does this need to be public? Can't one of the helper functions below pass the data? If it really needs to be public, could it be write-protected?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Thanks. Now that you've improved this we should just init and expose the StorageAPI on the node, like the other.s"
+    },
+    {
+      "prediction": "I don't think this is a good name. It's not the next prefix key but the next key with the next available rid. I think it should be something like `nextAvailableRidKey` or `nextRidKey` or something like that.",
+      "reference": "it will be better if you can provide some examples"
+    },
+    {
+      "prediction": "I think this should be called `operationResponseType` to be consistent with the other fields in this class.",
+      "reference": "\"return type\" and \"response type\" have never really been contrasted before, and it's unclear what their meaning is here. Could you clarify?"
+    },
+    {
+      "prediction": "I think this should be `import json`, not `import json`, because `libcodechecker.log` is a submodule of `libcodechecker`.",
+      "reference": "Import order has been violated here."
+    },
+    {
+      "prediction": "I think this should be `actorState.Query` instead of `actorState.Snapshot`",
+      "reference": "Are you suggesting deleting the concept of message querying from plumbing (sounds like more trouble than its worth) or suggesting deleting the snapshot based implementation?"
+    },
+    {
+      "prediction": "I think this should be `use this key to set the customized labels for nodes. you can input customized labels like {\"key1\": \"value1\", \"key2\": \"value2\"}`",
+      "reference": "I recommend using StringSliceVarP to resolve the label flag, like `-l key1=value1,key2=value2`. What do you think?"
+    },
+    {
+      "prediction": "I think this should be done in `addTabIcon` method.",
+      "reference": "@Samuell1 Will assigning the same icons array to all the tabs result in, for example, a primary tab called \"Colours\" and secondary tab called \"Colours\" having the same icon?"
+    },
+    {
+      "prediction": "I think this should be `return _(translatedMessage)` and not `_translatedManifestPaths`",
+      "reference": "A docstring for this function please."
+    },
+    {
+      "prediction": "I don't think this needs to be in this PR.",
+      "reference": "heads up @JunTaoLuo"
+    },
+    {
+      "prediction": "I think this should be `stackFrame.GetMethod()?.DeclaringType?.IsAssignableFrom(typeof(System.Runtime.CompilerServices.AsyncTaskMethodBuilder))`",
+      "reference": ".GetMethod() cannot return null now, correct?"
+    },
+    {
+      "prediction": "I think this should be `material-icons iconOsdVolumeUp`",
+      "reference": "Same thing here, I think the proper fix is elsewhere."
+    },
+    {
+      "prediction": "I think this should be a separate commit.",
+      "reference": "errno is a global variable that might be set by any system library call. It is dangerous to rely on it except right after the system call where it is set. But in this code, it looks like the library call is buried inside ExpLOBInterfaceSelectCursor. It would be safer if the latter function saved errno after whatever system library call it calls, and returns that as an \"out\" parameter, which this code could then check."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it's better to have the `__repr__` method in the `PackageNode` class itself. This way, we don't have to worry about the `__repr__` method being overridden by a subclass, and we don't have to worry about the `__repr__` method being called with an instance of `self._keys` being passed to `__repr__`. Instead, we can just call `repr()` on the instance of `self._keys` to get a string representation of the `PackageNode` object.",
+      "reference": "`hasattr` is kinda terrible; just append extra info in the subclass."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `datetime` module, since it's specific to datetime objects.",
+      "reference": "Fair warning, the way that exception block is written won't work in python3 Python 3.2.3 (default, Jun 8 2012, 05:36:09) [GCC 4.7.0 20120507 (Red Hat 4.7.0-5)] on linux2 Type \"help\", \"copyright\", \"credits\" or \"license\" for more information. > > > try: > > > ... raise ValueError(\"foo\") > > > ... except ValueError,e: > > > File \"<stdin>\", line 3 > > > except ValueError, e: > > > ^ > > > SyntaxError: invalid syntax"
+    },
+    {
+      "prediction": "Line is too long. [87/80]",
+      "reference": "This only updates the metadata and not the actual thing you are trying to solve for the docker hypervisor."
+    },
+    {
+      "prediction": "I think we can remove this package. It's not used anywhere.",
+      "reference": "should this package be internal? don't we want to use it e.g. in go-contrib?"
+    },
+    {
+      "prediction": "I think we can remove `interface{}` and just use `Document` everywhere.",
+      "reference": "Just curious, why did you make this change?"
+    },
+    {
+      "prediction": "I think this should be `prefix = {column_label: kdf._internal.spark_column_for(label) for column_label in column_labels}`",
+      "reference": "Can you handle error cases such as `pd.get_dummies(pdf, prefix={\"A\": \"foo\"})`?"
+    },
+    {
+      "prediction": "I think this should be `host` or `bridge` instead of `host`",
+      "reference": "Where is this environment variable being set?"
+    },
+    {
+      "prediction": "Shouldn't this be `TypeError(f'module must be a child of nn.Module, but got {module_class.__name__}')`?",
+      "reference": "`module_class ` is already a class and `type(module_class)` always returns `<class 'type'>`"
+    },
+    {
+      "prediction": "I think you can remove the `UTF8String.fromString(path.toString())` part here.",
+      "reference": "Nit: The lambda can be replaced with a method reference: `SparkOrcWriter::new` I think."
+    },
+    {
+      "prediction": "I think this should be `isinstance(obj, (int, np.int64, np.int32, np.int16, np.uint64, np.uint32, np.uint16))` instead of `getattr(getattr(obj, 'dtype', None'), 'kind', 'o') in 'ui'`",
+      "reference": "It would be nice to have a docstring that says what range of integer-like objects will be accepted, as well as what the int_like argument is meant for."
+    },
+    {
+      "prediction": "Use 2 (not 1) spaces for indenting an expression spanning multiple lines.",
+      "reference": "How do we manage updates to this view down the road?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "We can revert this to reduce the diff of the PR."
+    },
+    {
+      "prediction": "I think it would be better to use `expectThrows(IllegalArgumentException.class, () -> w.addDocument(doc));` instead of `expectThrows(IllegalArgumentException.class, () -> w.addDocument(doc));`",
+      "reference": "I think we should refactor or drop this test, as it is not testing the points format now, but IndexingChain/FieldsInfos' logic. Maybe we could rename the test `testMergeMissing` and configure the first segment to not have the `id` field at all."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this formatter. I think it should go in `spec/formatters/base_text_formatter.rb`.",
+      "reference": "not sure I follow why all this goes away?"
+    },
+    {
+      "prediction": "I think this should be `quteproc.send_cmd(':insert-text {}')` and not `quteproc.send_cmd(':paste-primary')`",
+      "reference": "Is this `{clipboard}` or `{primary}`, as the deprecation message for `:paste-primary` says?"
+    },
+    {
+      "prediction": "I think this should be `cli.Float64Flag` instead of `cli.Uint64Flag`",
+      "reference": "Human unreadable. IMHO user should input MYST value: 0.0006 @chompomonim, opinions?"
+    },
+    {
+      "prediction": "I'm not sure about this one, but I think it should be removed as well.",
+      "reference": "May be you could elaborate on why you removed this? I see some of this logic has been moved to ia2Web, but couldn't there be cases outside ia2Web (i.e. UIA) in which case ROLE_TABLEROW and ROLE_TABLECOLUMN are mapped to objects, in which case these objects will be of presentation type content erroneously?"
+    },
+    {
+      "prediction": "`player->tradeItem` and `tradePartner->tradeItem` should be `player->getTradeItem()` and `tradePartner->getTradeItem()`.",
+      "reference": "This is not consistent. Either rename the first to `tradePlayerItem` (which is ugly) or the second to `partnerTradeItem` (:+1:)"
+    },
+    {
+      "prediction": "I think it would be better to use `go get -u github.com/xeipuuv/gojsonschema@vX.X.X` instead of `go get -u github.com/xeipuuv/gojsonschema`",
+      "reference": "is there a reason we don't want to gx this dependency?"
+    },
+    {
+      "prediction": "I think this should be `strip_vendor($model, 'NVIDIA')`",
+      "reference": "I'd suggest to add \"ATI \" and \"Intel(R) \" to this list"
+    },
+    {
+      "prediction": "I think this should be `$warnings[] = Lang::get('backend::lang.warnings.plugin_missing', ['name' => '<strong>'.$plugin['name'].'</strong>']);`",
+      "reference": "This is using a different lang key, we should switch it to using the new key and remove the old key if it is no longer used."
+    },
+    {
+      "prediction": "I think we can remove the `nodesRegex` and just use `nodesRegex + \"|\" + api.SpecNodes + \"=([0-9A-Za-z_-]+),*\"`",
+      "reference": "shouldn't this work? nodesRegex = regexp.MustCompile(api.SpecNodes + \"=(('[0-9A-Za-z,_-]+')|([0-9A-Za-z_-]+)),?\")"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `Account` since it's a model and not a controller action. What do you think?",
+      "reference": "You mentioned this change was prompted because the `before_create` action was actually a defect. This will be done only for a new record; why is `before_validation`, which will be called repeatedly as accounts get updated and saved, correct whereas `before_create` is not? It looks like one would want to assign and activation code, etc., only once."
+    },
+    {
+      "prediction": "I think this should be `globalRate >= 0.0f && globalRate <= 1.0f`",
+      "reference": "What's the rationale for a default setting to not use a GlobalSamplingRate of 1? I don't know much about the sampling rate stuff"
+    },
+    {
+      "prediction": "I think this should be `echo \"$( [ -d \"/cygdrive/c/Program Files (x86)\" ] && echo \"/cygdrive/c/Program Files (x86)\" || echo \"/cygdrive/c/Program Files\" )/Puppet Labs/Puppet Enterprise/bin)`",
+      "reference": "Don't we know if it's 64 by this point? Do we have to test for the directory on every call, or can't we just split the value based on the platform?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this menu item. I think it should be moved to `note-list.js` under `menuItems`.",
+      "reference": "It's not async"
+    },
+    {
+      "prediction": "I think we can remove the `periodic` option from the `AddOtlpExporter` method.",
+      "reference": "Maybe we could add some use of `exporterOptions` in this example, otherwise the proper code would replace this variable name with an underscore?"
+    },
+    {
+      "prediction": "This should be `public $everykey = true;`",
+      "reference": "Maybe would be better to rename it to \"onEveryKey\" or \"fireOnEveryKey\"."
+    },
+    {
+      "prediction": "I think we can remove the `privateUrlUser` check from `FileDownloadHelper.isPreviewAllowed` as well.",
+      "reference": "if we're switching to calling FileDownloadHelper, we can just call that directly from the xhtml (see line 357 for example) and then remove this method completely. This is because the other thing it does is check PrivateURLUser, but the FileDownloadHelper method already does that. (and while we're at it, we can remove the method from FileUtil as this is the only place that method is called)"
+    },
+    {
+      "prediction": "I think this should be `getMethodConfigMap().get(methodSimpleName)`",
+      "reference": "is `fullName` still required as a parameter?"
+    },
+    {
+      "prediction": "This should be `LOG.info(\"Bound legacy RC support\", e);`",
+      "reference": "This informational message is important to users. Please leave."
+    },
+    {
+      "prediction": "I'm not sure why we need to set this to `true` in the `initEnvOpts` struct. I think we can just always set it to `true` in the `deployEnv` function, since it's always going to be `true` anyway.",
+      "reference": "Why do we want to remove this one? Are we planning to substitute it?"
+    },
+    {
+      "prediction": "I don't think we need this, since we don't have a `WithDetails` implementation yet.",
+      "reference": "We may create an issue to track this so we don't forget?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "This file has the executable bit set."
+    },
+    {
+      "prediction": "I think this should be a debug statement.",
+      "reference": "Improve the formatting of the message `.Infof(\"setting db capacity to: %v\", db.capacity)` There is a space before `:` and this way it is easier to see the formatting."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "This lock cannot be released here, as it is protecting m_RTPSParticipants and the participant reference."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `cast` function. The `cast` function is called before `find` is called, so the `castError` is always going to be `null`.",
+      "reference": "we'll need this in `_findAndModify` too"
+    },
+    {
+      "prediction": "Shouldn't this be `bundle.getString(DB_NAME, DBOpenHelper.DEFAULT_DB_NAME);`?",
+      "reference": "This is the fix for the crash. The function to create the intent to launch this activity requires `dbName`, so it has to be set to `null`. In such cases, the value for `dbName` will be set to `null` and cause issues throughout this activity. This adds a default value if the explicitly assigned value in `null`."
+    },
+    {
+      "prediction": "I don't think we need to fire the handler here. The handler is only called when the text changes, not when the text changes and the caret moves.",
+      "reference": "Shouldn't we also trigger vision update here, so that if someone has caret highlighting enabled the correct character is highlighted?"
+    },
+    {
+      "prediction": "I think this should be `context.getNamer().getLocalPackageName() + \";` instead of `context.getProductConfig().getPackageName() + \";`",
+      "reference": "Woah, how does this change, which includes semicolons, result in the baseline change? Are the semicolon chars just part of the internal representation of the import type?"
+    },
+    {
+      "prediction": "I think this should be `pyw -3.7-32` and not `pyw {self._runNVDAFilePath}`",
+      "reference": "Can't you just use runnvda.bat here?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to have a macro like `__unused_1` or `__attribute__((unused))` instead of defining `#define` for every unused parameter.",
+      "reference": "these definitions of UNUSED aren't the same as the one you imported, so callers should be fixed (if any)"
+    },
+    {
+      "prediction": "I think this should be `ECSIPAMPluginName`, not `ECSIPAMPluginName`.",
+      "reference": "is type a free string too?"
+    },
+    {
+      "prediction": "I think it would be better to pass `model` and `model_func` as arguments to `ort_validate` instead of `feats`.",
+      "reference": "here input could be a normal function or an instance of torch.nn.Module."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get Travis to pass.",
+      "reference": "Example output: \"19.03.git.a7f4a860d0c\" At some point `nixpkgsVersion` was renamed to `version` since someone may use an old channel we fallback to `nixpkgsVersion` for backwards compatibility."
+    },
+    {
+      "prediction": "I don't think this is correct.",
+      "reference": "Class import is therefore redundant I guess :)"
+    },
+    {
+      "prediction": "I think this should be `errorNodeCreation` instead of `errorNodeCreationIPFailure`",
+      "reference": "If parsing fails, don't act as no IP specified. Error out."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "... here the semicolon wanted to run away from the enum ;)"
+    },
+    {
+      "prediction": "I don't think we need to add switchTransactionLog here.",
+      "reference": "Note that this is in the context of SolrCloud, so \"secondary\" doesn't apply and should be instead follower"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.",
+      "reference": "This one is a query parameter and probably really should be ID and in fact I'm mystified as to how this is working because the controller code definitely wants a number for that parameter so the test framework must be converting it back to an ID as best I can tell."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the node map.",
+      "reference": "Need to ensure this is set to TRU for LOB datatype access too not just for hive ."
+    },
+    {
+      "prediction": "I think this should be `sfun` and `numeric_only` should be `None`.",
+      "reference": "You're using the Sphinx docstring style here (maybe because it's the default in PyCharm?). Instead, you should use the NumPy style (that you've also used in other places) to be more consistent with the rest of the project."
+    },
+    {
+      "prediction": "Why do we need to check for `!hasServerAddresses`? I think it's enough to check for `!hasListenOptions && !hasServerAddresses`",
+      "reference": "Don't really need this since the rest of the method will no-op but it's also safe to return early."
+    },
+    {
+      "prediction": "Is this needed? I don't see where it is used.",
+      "reference": "Don't use relative paths. src directory is already on the include search path"
+    },
+    {
+      "prediction": "I think we should remove this file. We don't need it for .NET Framework anymore.",
+      "reference": "this file can be removed"
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think it's better to pass the required parameter to the constructor of `RequestAttribute`.",
+      "reference": "Do we have test cases that name is null?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in `SumIntAgg` itself, not in `DoInt`.",
+      "reference": "in a separate PR, I'll check for a table that is empty/all-null and skip this function call completely."
+    },
+    {
+      "prediction": "I think it would be better to move the `catch` to the `try` block, so that the `catch` is always executed, even if the azkaban event reporter is not configured.",
+      "reference": "We really, really want to catch _everything_ here. It is safe to leave `azkabanEventReporter` set to null."
+    },
+    {
+      "prediction": "I think this should be `if (parentDom.lastChild!==newDom) {`",
+      "reference": "Can't we add the check here?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The problem is that we don't know what `body` is until we get the response, so we can't tell if it's `None` or something else.",
+      "reference": "body is being used here for the first time without every being populated. This will error. To run tests do `./go clean test_py` and that will run the Firefox tests"
+    },
+    {
+      "prediction": "I think it would be better to have this as a field in the `Deployment` struct itself. This way we don't have to pass it as a parameter to the `Deployment` struct and we also don't have to pass it as a parameter to the `GenericDeploymentSpec` struct. I think it would be better to have this as a field in the `Deployment` struct itself. This way we don't have to pass it as a parameter to the `Deployment` struct and we also don't have to pass it as a parameter to the `GenericDeploymentSpec` struct. I think it would be better to have this as a field in the `Deployment` struct itself. This way we don't have to pass it as a parameter to the `Deployment` struct and we also don't have to pass it as a parameter to the `GenericDeploymentSpec` struct. I think it would be better to have this as a field in the `Deployment` struct itself. This way we don't have to pass it as a parameter to the `Deployment` struct and we also don't have to pass it as a parameter to the `GenericDeploymentSpec` struct. I think it would be better to have this as a field in the `Deployment` struct itself.",
+      "reference": "Tell me if you have a better field name for this."
+    },
+    {
+      "prediction": "I think we can remove the `|| Framework == TargetFramework.NETCOREAPP3_1 || Framework == TargetFramework.NETCOREAPP3_0` part here as it's already covered by the `|| Framework == TargetFramework.NET5_0` part above.",
+      "reference": "FYI, I have a branch I started to improve this. Nuke can read the target frameworks from the project files, so we don't have to do this mess. Ran into some other issues so it didn't take priority"
+    },
+    {
+      "prediction": "I think you can remove the `unsigned int ai, bi;` line above as well.",
+      "reference": "This is just `mol->atomNeightors()`, is it?"
+    },
+    {
+      "prediction": "I think this should be `infer_kernel_gpu<Float, method::by_default, task::clustering>`, not `infer_kernel_gpu<Float, method::by_default, task::clustering>`.",
+      "reference": "This alias is just for a single occurrence. Maybe it ins't necessary?"
+    },
+    {
+      "prediction": "I think this should be `return false;` instead of `return true;`",
+      "reference": "You can delete this `NONE` case."
+    },
+    {
+      "prediction": "I think we can remove the `Mismatching` part of the description since it's implied by the `Mismatching Layouts` part.",
+      "reference": "@cnorthrop -- here is a test with a name change. Is this going to affect your internal CI?"
+    },
+    {
+      "prediction": "I don't think we need to add a new command here, since it's the same as `SEND_COMMANDS_FOR_DOWNLOAD_CHROME`.",
+      "reference": "Nit: `Headless` is one word, not two, and so doesn't need camelcasing in this way."
+    },
+    {
+      "prediction": "I think this should be `The default is the number of CPU cores detected by the system.`",
+      "reference": "I think this will cause a warning, can you set the default to zero (`0`)?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `service_discovery` since it's related to the service discovery part of the client.",
+      "reference": "confusing naming. Identity from \"id\" package :/ why we need alias here? (and in other imports)"
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to do, but I think it should be removed.",
+      "reference": "It'd be nice to fix up existing configs with the change - that'd mean adding the option to `RENAMED_OPTIONS` in `config.py` and adding something like `_get_value_transformer({'true': '', 'false': '*'})` to `CHANGED_OPTIONS`. I think I never tried adding an option to both, but it should work."
+    },
+    {
+      "prediction": "`Object.getPrototypeOf` is not available in IE11.",
+      "reference": "Perhaps use the new `.keys()` method here instead?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think this should be done in `provider.DiffList`",
+      "reference": "We don't need this option because plan-preview is comparing between 2 commits, not using the live state."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I'm not sure about putting this in the options. I would see it living in an object and then being queried on through a class - a more object oriented, ruby solution."
+    },
+    {
+      "prediction": "I think we can remove this function and just call `card.getMetrics()`",
+      "reference": "So `Status` returns metrics?"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "Can we make a generic method say, `set_infographic_view` and in that method we can check the 4 values (outside_committers, portfolio_projects, outside_projects, affiliated_committers) and check against params[:action] and params[:view] and then trigger the respective method dynamically as it takes the standard argument(s) for all the four methods. What do you think?"
+    },
+    {
+      "prediction": "I think we can remove the `else` part of the `if` as it is the same as the above `if`",
+      "reference": "Why would this be called to copy a non-Date object?"
+    },
+    {
+      "prediction": "I think this should be `native.PutUint32(buf[4:8], uint32(0))`",
+      "reference": "This alignment issue had not appear because no multiple `Boolmsg`s had been serialized at once."
+    },
+    {
+      "prediction": "This should be `BooleanDefaultFalse{Value: NotSet}`",
+      "reference": "Why are we not implementing a BooleanDefaultFalse struct and replacing all of these config vars that default to false with that? Are \"default false\" config vars not affected by this bug?"
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "Replaced this method with a ruby idiomatic setter."
+    },
+    {
+      "prediction": "I think this should be `This program is free software; you can use it, modify it and/or distribute it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.```",
+      "reference": "wrong place hehe"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "Can this be `test -e /var/run/mysqld/mysqld.sock` instead of `ls`? Using `ls` for noninteractive stuff is a recipe for sadness, IMHO. /cc @tianon"
+    },
+    {
+      "prediction": "I think this belongs to `ClientData` instead of `User`",
+      "reference": "Are these `dependent: destroy` attributes intended to implement cascading deletes? I had assumed they were added because of the `paranoia` gem, but since that is no longer part of this PR, are they here as a best practice?"
+    },
+    {
+      "prediction": "I think this should be `if (stream->tunnel == NULL || tunnel == NULL)`",
+      "reference": "This approach would work only if there's a guarantee that the stream does not get freed before `establish_tunnel` is called, even though it _may_ enter CLOSE_WAIT state (at which point `stream->tunnel` is set to NULL). I do not think that assumption holds. Hence the alternate approach."
+    },
+    {
+      "prediction": "This should be `nano_pow_server_path`, not `pow_server_path`.",
+      "reference": "Did you mean to remove the periods here while newly adding to other locations?"
+    },
+    {
+      "prediction": "I think this should be `newLen = 0`",
+      "reference": "the check can now be `if newLen < 0`"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "If we're going to clean up this package, can we also remove those two `_` imports that appear to serve no purpose? (FWIW, they're also in the manager package too...)"
+    },
+    {
+      "prediction": "I think this should be `static std::unique_ptr<MMFFAromCollection> &ds_instance(new MMFFAromCollection(nullptr));` instead of `static std::unique_ptr<MMFFAromCollection> ds_instance(new MMFFAromCollection(nullptr));`",
+      "reference": "Won't this leak like a sieve? The caller can't delete this as one is a unique_ptr and one is not."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the `package.json` file, not in the `verdaccio` package.",
+      "reference": "form => from"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `org.sonar.java.regex`.",
+      "reference": "Ok, we can't switch it now, because the properties are a field of the interface Rule - which makes it public API. Maybe we should remove it here in the (Java) API in 7.0.0? The only API left would be, when using a rule and setting the properties in the ruleset xml. There the type doesn't matter - since the String is then automatically converted into a Pattern."
+    },
+    {
+      "prediction": "I think this should be `GapicProductConfig` instead of `GapicProductConfig.create`",
+      "reference": "Pass in null here, instead of a value? Maaaaybe we should add an overload for `GapicProductConfig.create` that matches the original signature and passes through null? I don't feel strongly about that though, so happy to leave asis if you prefer."
+    },
+    {
+      "prediction": "I think this should be `new String(replacement, 0, breakPos + beginPos)` instead of `new String(word.chars, 0, word.offset + breakPos)`",
+      "reference": "Include all compounds parts from the very beginning for the case check to work later"
+    },
+    {
+      "prediction": "Why do we need to change this?",
+      "reference": "if the user specifies a url scheme on the cli this is going to cause problems, no?"
+    },
+    {
+      "prediction": "I think this should be `width: dom.getWindowSize().innerWidth * 0.75`",
+      "reference": "How about that `scaleFactor`?"
+    },
+    {
+      "prediction": "Why do you need this method? You can just return `Type::getFloat($result);` in line 275.",
+      "reference": "`assert(is_int($result));` would be preferable I think, even if we know it can't ever be false"
+    },
+    {
+      "prediction": "I think it would be better to rename this method to `getQuantifiedProductsIndexedByItemId` or `getQuantifiedProductsByItemId`.",
+      "reference": "I'm unfortunately unable to review whether you've changed everything that used to use cartIds"
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "Please don't touch unrelated files to keep the git history clean"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [121/120]",
+      "reference": "Per our style guide, I don't think we indent `private` keyword. Would you mind fixing that?"
+    },
+    {
+      "prediction": "I think this should be `out_heatmap = torch.nn.functional.interpolate(heatmap, mode='bilinear', align_corners=False, recompute_scale_factor=False)`",
+      "reference": "Will this change the input `heatmap`? Is this behavior expected or not?"
+    },
+    {
+      "prediction": "I think this should be `generateDualArgStringFunctionReturnInt(\"LastIndexAny\", []string{stringArgV, chars}, strings.LastIndexAny)`",
+      "reference": "If I'm reading this right, this changes the parameter name so it'd constitute a breaking change :cold_sweat: Are we missing a test that should have been failing up until now?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Need copyright/license info."
+    },
+    {
+      "prediction": "I think we can remove this file now, it's not used anymore.",
+      "reference": "Is this `info log` necessary? If it is not useful, can you consider deleting it?"
+    },
+    {
+      "prediction": "I think this should be `RCLONE_ENCRYPT_V0` instead of `V0`.",
+      "reference": "Unrelated change, but seems to be the correct purpose of the test."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `componentDidMount` or `componentDidUpdate`",
+      "reference": "Should it be `'uppy-input-`?"
+    },
+    {
+      "prediction": "I think this should be `if (false) {`",
+      "reference": "what's the reasoning to keep these always true or dead-code blocks?"
+    },
+    {
+      "prediction": "Please use `toggleClass` instead of `removeClass` and `addClass`.",
+      "reference": "It would be a good practice to do `.off('click', '.tab-buttons > div').on('click', '.tab-buttons > div', function() {`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the way I've been doing it for a while.",
+      "reference": "I suggest that we rely on `--controllers='*,gateway-shim'` for now, and we can then move from `--controllers='*,gateway-shim'` to automatically enabling the Gateway API support on startup using the discovery API in 1.6 or 1.7, what do you think? Note that the logic I wrote in e5436df521015057e77de3fe02c174ea8a863b93 should also use this `GatewaySolverEnabled` flag I think."
+    },
+    {
+      "prediction": "I think this should be done before `configureNativeLibs`",
+      "reference": "I moved the instantiation of the signature algorithm here, to execute it before the native libraries are configured. Otherwise the wrong signature algorithm could be configured in a future release when more than one is supported in Besu."
+    },
+    {
+      "prediction": "I think this should be `from typing import TYPE_CHECKING, Callable`",
+      "reference": "Why remove this?"
+    },
+    {
+      "prediction": "nit: I think it would be better to have this method return a new instance of the `ServiceBuilder` and then call `this.ieService_ = service;` directly in the `setIeService` method. This way you don't have to pass in a new instance of the `ServiceBuilder` in the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `set",
+      "reference": "`this.ieService_` should be initialized to null in the constructor."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure it's necessary. The `ebpf::BPF` class has a `~ebpf::BPF()` destructor that cleans up the `bpf` object. I'm not sure what happens to the `ebpf::BPF` object when the `ebpf::BPF` class goes out of scope, but I'm not sure what happens to the `bpf` object when the `ebpf::BPF` object goes out of scope. I'm not sure what happens to the `unique_ptr` when the `unique_ptr` goes out of scope, but I'm not sure what happens to the `ebpf::BPF` object when the `unique_ptr` goes out of scope. I'm not sure what happens to the `unique_ptr` when the `unique_ptr` goes out of scope, but I'm not sure what happens to the `ebpf::BPF` object when the `unique_ptr` goes out of scope. I'm not sure what happens to the `unique_ptr` when the `unique_ptr` goes out of scope, but I'm not sure what happens to the `ebpf::BPF` object when",
+      "reference": "Why not `ebpf::BPF bpf;`? Assuming that the instance is not huge, I do not think there is a good reason to use a pointer when it can be retained as a value."
+    },
+    {
+      "prediction": "nit: should be `github.com/jetstack/cert-manager/pkg/apis/core/v1` instead of `github.com/jetstack/cert-manager/pkg/apis/certmanager/v1`",
+      "reference": "nit: move to last block"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/agent` since it's specific to that package.",
+      "reference": "nit: I'm under the impression that `log` is the convention for golang, and that the `er` suffix is usually reserved for interfaces"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the `DbProviderFactories` class.",
+      "reference": "Why is IsolationLevel fully qualified? System.Data is already in scope? What do I miss?"
+    },
+    {
+      "prediction": "I think we can remove `SetEnvelopeContext` and `SanityCheck` from `types.go` and move them to `actions.go`",
+      "reference": "do we still need SanityCheck()? seems they all return nil now"
+    },
+    {
+      "prediction": "I'm not sure about this one. I think we should keep it for now, but let's discuss it with @jridgewell.",
+      "reference": "What I am missing in this PR, and I think we discussed that on the weekly meeting, is that all the code snippets that advise using `loadData` should be changed to one of the two new methods. Otherwise we send confusing mixed signals by promoting `loadData` everywhere in the guides. `loadData` is not deprecated, but is legacy."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this, but I don't have a better suggestion.",
+      "reference": "Line is too long. [100/80]"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to the top of the function so that it is the first thing that is checked.",
+      "reference": "I suggest \"no parent snapshot found, will read all data\\n\"."
+    },
+    {
+      "prediction": "Why is this needed? I don't see anything in this PR that would change the behavior of the test.",
+      "reference": "`LogManager.ThrowExceptions = true` has very special meaning when unit-testing. Why the change to `LogFactory.ThrowExceptions = true` and enabling after the config-load?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `@` and `\\\\` are not escaped by default. You need to add `@` and `\\\\` to the list of special characters that are escaped by default.",
+      "reference": "It should be return \"''\"; I will fix it"
+    },
+    {
+      "prediction": "I don't think you need to check the type image here. You can just check the name of the type.",
+      "reference": "This is wrong, should be `entry.getValue()`"
+    },
+    {
+      "prediction": "I think this should be `{ \"op\": \"replace\", \"path\": \"/spec/servingCerts/namedCertificates\", \"value\": [ %s ] }`",
+      "reference": "nit: why can't we combine the two add operations into one so that we add `{\"nameCertificates\": []}` to `/spec/servingCerts` ?"
+    },
+    {
+      "prediction": "This file should not be included in this PR.",
+      "reference": "These whitespace only changes make it very difficult to review this change. Is there any way you could separate just the code changes into one PR and the whitespace only fixes into another? Or are they sufficiently merged together at this point? Chrome is having a real tough time rendering these large diffs."
+    },
+    {
+      "prediction": "I think this should be `final`",
+      "reference": "why does this suddenly need a config ref"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it would be better to put it in `src/test/javascript/` or `src/test/javascripts/` or `src/test/javascripts/angularX/` or something like that.",
+      "reference": "Unfortunately, I couldn't find a way to get karma to conditionally load stuff from a CDN, so I had to include it."
+    },
+    {
+      "prediction": "I think it would be better to put them in alphabetical order. Also, I think it would be better to put them in the same order as they are defined in the code.",
+      "reference": "Those functions are not currently wrapped, so I don't think it is necessary to exclude them (at least for now)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should go in `pkg/time/timer.go`.",
+      "reference": "Nit: since this is now a private function rather than a variable, would it make sense to move it after the public functions in this file for readability?"
+    },
+    {
+      "prediction": "Why did you change this to `int(255.0 * col.r), int(255.0 * col.g), int(255.0 * col.b)` instead of `col.r, col.g, col.b`?",
+      "reference": "I'm assuming that MolDraw2Qt drops the alpha channel?"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in the tests.",
+      "reference": "the code looks good to me, but I just wonder if it is a real use case: (de)serializing a single integer."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the same place as `re.split(r\"([A-Z][a-z]*)\\s*([-*\\.\\d]*)\", f)`.",
+      "reference": "Pls remove the print statements. Also, you do not need the if statement."
+    },
+    {
+      "prediction": "I think this should be `replace(r'<StringValue>', '<StringValue><![CDATA[')` and not `replace(r'<StringValue>', '<StringValue><![CDATA[')`",
+      "reference": "Thanks for this fix @eltone !   Wondering if we should make this a bit more resilient to upstream changes, e.g., via a regex replacement. Not critical, though - we can pick that up in a follow-up PR.. Thanks"
+    },
+    {
+      "prediction": "Why do we need to set `Sniff=false`?",
+      "reference": "Should this be instead a configurable option?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `ComputeClient.CreateDisk` function.",
+      "reference": "unit test needs an update?"
+    },
+    {
+      "prediction": "I think we can remove the `opts.authProvider` argument from the constructor, since we don't need to pass in an auth provider when creating a new `Provider` object.",
+      "reference": "I realised that this is not used anywhere"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to make sure that the `watcher` is only started if we actually need it.",
+      "reference": "why add the default branch? this may cause there exist multiple configWatchers that notify the eventsCh at the same time?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. The USB driver should not be responsible for sending the ZLP. It should be the driver for the USB device.",
+      "reference": "This line appears to have caused the regression. What is it supposed to be doing?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "# `attr_value['string'] is not None` probably not required as `attr_value['string'] == value` check is already being done"
+    },
+    {
+      "prediction": "I think we can remove the `BaseExportingMetricReader` class altogether, and just use `new BaseExportingMetricReader(exporter)`",
+      "reference": "@alanwest I noticed this while changing the code. I think we _might_ run into some race condition - if a scraper happens to hit the HTTP server before we could add the reader, what would happen (I guess we will hit exception, which turns into HTTP 500)? I haven't looked into the HTTP server logic. I think it _might_ be OKAY. A better version could be - we only start the HTTP server once the exporter/reader are fully ready and both are hooked up to the provider."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with that works.",
+      "reference": "As for Owin, I think we should exclude this URL in `IsNotServerLifeCheck` too?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should go in `HttpClientTelemetryExporterOptions`.",
+      "reference": "Does it make sense to just fold these options back into the exporter options? My thinking is that this option is unlike the processor/metric reader options in that it actually is about the exporter itself."
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "here is `:seconds` missing"
+    },
+    {
+      "prediction": "I think this should be `lambda arg2=arg2: arg2 * arg2` instead of `lambda arg2=arg2: arg2 * arg2`",
+      "reference": "change of name is not needed"
+    },
+    {
+      "prediction": "I think this should be `self.assertEqual(kser.name, None)`, not `self.assert_eq(kser.name, None)`.",
+      "reference": "Hm .. so this case doesn't work anymore?"
+    },
+    {
+      "prediction": "This should be `from PyQt5.QtCore import (pyqtSignal, pyqtSlot, QObject, QProcess, QFileSystemWatcher)`",
+      "reference": "Please indent this so it lines up with the `(`"
+    },
+    {
+      "prediction": "Shouldn't this be `wp_enqueue_script('roots_plugins', get_template_directory_uri() . '/js/plugins.js', false, null, false);`?",
+      "reference": "We should still register the script. Just not enqueue it."
+    },
+    {
+      "prediction": "I think this should be `change = attributes[name]` instead of `change = [changed_attributes[name], attributes[name]]`",
+      "reference": "I think this can stay as a method. And we can write our own attribute_change(name) as required."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `MvvmCross.DroidX.Adapter` since that's where the `ItemTemplateSelector` is.",
+      "reference": "@alexshikov @Cheesebaron I might be missing something but isn't the viewType parameter here the actual layout resource id? Why would the GetItemLayoutId method need to be called again? On line 127, the exact same method is called except this time passing in the index of the object in the backing data source, which makes sense - the template selector can return the layout id based on the index of the item in the list."
+    },
+    {
+      "prediction": "I think this should be `self.project_id + \"/\"` instead of `self.project_id + \"/\" +`",
+      "reference": "Can you change this to use `.format()`?"
+    },
+    {
+      "prediction": "I think we can drop the `&& r.Config.DropVXLANPacketsFromWorkloads` check here as it's already covered by the `&& r.Config.DropIPIPPacketsFromWorkloads` check above.",
+      "reference": "Maybe the parameter should be consulted higher up the stack so that we only set dropEncap if we're rendering a workload egress chain and the flag is set?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in the `SectorBuilder` interface itself, since it's only used by that interface.",
+      "reference": "new casing is less ELiTE"
+    },
+    {
+      "prediction": "I think this should be `return r->impl->renderbuffer_from_image(image);`",
+      "reference": "This leaks EGL/GL implementation details into the generic renderer interface."
+    },
+    {
+      "prediction": "I'm not sure if this is the best error to raise here, but it's the only one I could come up with.",
+      "reference": "Maybe add *key* i.e. `\"The \\\\`deploy.on\\\\` key should be a hash (dictionary).`"
+    },
+    {
+      "prediction": "This should be `new Color($status['color'])` and not `new Color($status['color'])`",
+      "reference": "its be good also change $code na $id, actual name is misleading and that was probably actual error generator"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `ListenStore` class.",
+      "reference": "I changed the name of the key, to make deployment easier."
+    },
+    {
+      "prediction": "This should be `spotInstanceDrainingEnabled` instead of `spotInstanceDRAINING_ENABLED`",
+      "reference": "isn't this going to be true by default? if yes, then let's have the env var named `ECS_DISABLE_SPOT_INSTANCE_DRAINING` and have default as false."
+    },
+    {
+      "prediction": "I don't think unicode literals are allowed in this file.",
+      "reference": "Do you think we might need this anywhere else in HoloViews?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for audio direct, but it's the only way I could find that seemed to work.",
+      "reference": "@MrTimscampi don't we want to remove this term entirely? Might as well do it now if that's the case."
+    },
+    {
+      "prediction": "Line is too long. [108/80]",
+      "reference": "I have followed the convention of `cp`"
+    },
+    {
+      "prediction": "This should be `string|null` instead of `?string`.",
+      "reference": "The docs say it's `false|null`"
+    },
+    {
+      "prediction": "I think it would be better to have a separate type for the status ratio. It would make it clear that this is a percentage value.",
+      "reference": "Thinking on it more, I actually really like your idea to rename `type` to `factor`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `MessageWebView` class, since it's specific to the `MessageWebView`.",
+      "reference": "This error message is redundant. All of this information is included in the stack trace. In general it's a good idea to avoid using method names in error messages. Chances are the method will be renamed at some point, but the string won't be updated accordingly. Then you'll end up with a very confusing error message. I suggest changing the message to: \"Failed to unblock network loads. Missing INTERNET permission?\""
+    },
+    {
+      "prediction": "I don't think you need to add this line, it's already in the top of the file.",
+      "reference": "Feel free to add your own name while at it."
+    },
+    {
+      "prediction": "I think it's better to have a separate file for this. It's not part of this PR, but I think it's better to have it in a separate file.",
+      "reference": "`signer` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "shouldn't this be `&& opts[:localize]` (i.e. both nil and false skip localization?)"
+    },
+    {
+      "prediction": "I don't think this should be in this PR.",
+      "reference": "This can also be removed."
+    },
+    {
+      "prediction": "I don't think we should be changing the semantics of this function.",
+      "reference": "Not critical but \"common\" is not that helpful in the commit title. Maybe go with \"libkvs/txn:\" for this one?"
+    },
+    {
+      "prediction": "I think this should be `self._migrate_bool(settings, 'scrolling', 'when_searching', 'never')`",
+      "reference": "It's been a while, but I just noticed this was wrong: It migrated `True` to `when-searching` (so people with `scrolling.bar = True` suddenly didn't have scrollbars anymore) and `False` to never. Instead, it should migrate `True` to `always` (no behavior change) and `False` to `when-searching` (so people notice the new feature). I fixed this in cc0f5fc6d400e12833ba729049e31d16cf836d53."
+    },
+    {
+      "prediction": "I think this should be `os.path.normcase(fs_path).decode(\"utf-8\")`",
+      "reference": ".decode(sys.getfilesystemencoding()) will be better. Although I think rrdtool should accept bytes.."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should go in `Microsoft.AspNetCore.Server.Kestrel.Internal.RequestHandler`.",
+      "reference": "An interface call per header might be more expensive than we want to pay. Do we really need this to be abstracted? I think a parser abstraction makes sense, but it seems to me like a separate abstraction for handling headers (one by one) might be an overkill."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I don't have a better idea. The problem is that the `GetStackTraceUsage` method returns `StackTraceUsage.None` when the `targets` parameter is `null`, which means that the `logEvent.HasStackTrace` check will always fail. I'm not sure what the best way to handle this is, so I'm hoping someone else will have a better idea.",
+      "reference": "Resharper annotations, always +1"
+    },
+    {
+      "prediction": "I'm not sure if this should be hidden or not. I think it should be hidden because we don't want people to be able to use it.",
+      "reference": "same here, CryptoSm2 won't be changed once command is compiled"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `LoginActivity`'s `runOnUIThread` method.",
+      "reference": "Loading the login page after the certificate has been loaded. Views can only be loaded from UI thread."
+    },
+    {
+      "prediction": "I think this should be `warn` instead of `warnf`",
+      "reference": "Technically it doesn't belong here; let me remove it."
+    },
+    {
+      "prediction": "I think this should be `a4adc8f4da9c7ec0d93b50ca5b73dd14df791c0e`",
+      "reference": "remember to update the hash once the upstream PR is merged"
+    },
+    {
+      "prediction": "I think we can remove the `SFDCFcmListenerService` and `SFDCmcpListenerService` as they are not used anymore.",
+      "reference": "Sends the incoming message to the decryptor, which will then forward it to the interface once processing is complete."
+    },
+    {
+      "prediction": "I'm not sure we want to do this yet. I think we want to do this in a follow-up PR.",
+      "reference": "FYI This will go the other way, with nil being the correct value for \"no params\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the non-peptide keys. I think they should be added after the PeptideLibraryKeys are added to nonPeptideKeySet.",
+      "reference": "Is this necessary? Are there duplicates in your LibKeyIndex?"
+    },
+    {
+      "prediction": "I think this should be guarded by `m.ctrl.AssertCalled(m, \"Next\")`",
+      "reference": "Why will the gomock files be regenerated? It seems to be irrelevant"
+    },
+    {
+      "prediction": "I think this should be `make(chan *reportable_command, 1)`.",
+      "reference": "I'm very curious about why using buffered-channel. Is there something wrong to use an unbuffered channel with zero capacity?"
+    },
+    {
+      "prediction": "I don't think this change is needed.",
+      "reference": "It seems like you accidentally pulled some unrelated changes to util_test into this."
+    },
+    {
+      "prediction": "This should be `using Nethermind.Blockchain.Bloom;`",
+      "reference": "do not toucm NDM please - there will be lots of conflicts"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `toInstant()` returns an `Instant` object, which is not compatible with `Date` objects. We need to convert the `Instant` object back to a `Date` object before passing it to the `IndexableField`'s `stringValue()` method.",
+      "reference": "This looks legitimate - Would cause a classCastException. But have we ever seen it in the wild?"
+    },
+    {
+      "prediction": "I think this should be `True` and not `True` and `False`",
+      "reference": "This looks out of scope for this particular PR."
+    },
+    {
+      "prediction": "I think this should be `time.Duration` instead of `int64`",
+      "reference": "Don't need () if only one."
+    },
+    {
+      "prediction": "Shouldn't this be `private final String mailPort`?",
+      "reference": "Why not have mailPort as int since your are already parsing while calling t.connect ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to keep the `>=3.0` part of `python-daemon` in order to support Python 3.0 and above.",
+      "reference": "In the future, do not remove these trailing commas as they have the purpose of making append-diffs easier to read. :)"
+    },
+    {
+      "prediction": "I think this should be `isset($this->fields['lccn'])` instead of `isset($this->fields['lccn']) && is_array($raw)`",
+      "reference": "Was this added by accident? It doesn't seem related to Syndetics, and I don't think it should be necessary in core VuFind (but maybe it's related to one of your local customizations)."
+    },
+    {
+      "prediction": "I think this should be `AllowedForeignJWTClaims` and `TrustDomain` should be `spiffeid.TrustDomainName`",
+      "reference": "i know the ergonomics are a little nicer with map[string]bool, but I prefer map[string]struct{} for sets for a few reasons: 1 - optimized storage (not very relevant here) 2 - don't have to think about the conditions where the key exists in the map or if the value could somehow be false"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should go in `pkg/server` since it's a server-side thing.",
+      "reference": "This is very unclear what Server means. We need a better name, maybe ServerExtension?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right test to have. I think we need to test that we can run `ddev login` and `ddev logout` and that we can run `ddev logs` after logging out.",
+      "reference": "I wonder if we should trigger a PHP error and ensure it ends up in the log?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "Hmmm not sure about this. Is it fine @ar2rsawseen?"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this interface. I think it should be `InstallerRunner` or something like that.",
+      "reference": "Why do you prefer `go` over `run`?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Unintentional, please remove"
+    },
+    {
+      "prediction": "I think we can remove this method now since we're using `Subscription.active.where(deactivated_on: nil)` directly.",
+      "reference": "What was the reason behind moving this?"
+    },
+    {
+      "prediction": "Do we need to escape `<` and `>` here?",
+      "reference": "Did you mean to put that `join` after the `map` rather than inside it? Also should it be `safe_join` or is interpolating it into an `_html` resource going to have much the same effect>"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it should be in the `Realm` class itself.",
+      "reference": "`Indicates if this Realm has been closed.`?"
+    },
+    {
+      "prediction": "I'm not sure we need to remove these, since they're only used in `pandas/_utils.py` and `pandas/_utils.py` doesn't import them?",
+      "reference": "how is this change adding corr to Series? Do all the methods that are added to Frame automatically get added to Series?"
+    },
+    {
+      "prediction": "I think we can remove the `builtin.Actors` argument from the `GetTipSetStateRoot` call.",
+      "reference": "Just a heads up for anyone else reviewing this, this logic and all the repetitions of it should be greatly simplified by subsequent work relating to issue #2552."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be in `System.Collections.Generic` since it's only used there.",
+      "reference": "@billwert I have kept the `1000` InvocationsPerIteration, but do note this is very small for the small types."
+    },
+    {
+      "prediction": "I don't think you need to add this empty line.",
+      "reference": "Unrelated whitespace change."
+    },
+    {
+      "prediction": "I think this should be `transaction_p = std::make_unique<rai::transaction> (environment, nullptr, true);` instead of `writeTransaction = std::make_unique<rai::transaction> (environment, nullptr, true);`",
+      "reference": "Slight formatting issue here. There should be a space after the asterisk."
+    },
+    {
+      "prediction": "`parent` is not used in this function. Can you remove it?",
+      "reference": "Slight preference for declaring variables inline rather than at the top of the function. That way you can use const for both of these."
+    },
+    {
+      "prediction": "I think it is better to use `org.junit.Assume.assumeTrue` instead of `static final Logger LOGGER = LoggerFactory.getLogger(TestMgr.class);`",
+      "reference": "this is a duplicate file of the one in demo-schema. please remove this file."
+    },
+    {
+      "prediction": "I think we should also add `utils.Eth1CrossDomainMessengerAddressFlag` and `utils.Eth1L1CrossDomainMessengerAddressFlag`",
+      "reference": "Nit: unified names between geth and contracts"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `server_test.go`.",
+      "reference": "Necessary? Could you just use DefaultBuildState here?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "linked hashmap retains order"
+    },
+    {
+      "prediction": "I think this should be `extern const struct wlr_drm_interface *iface_legacy;` and `extern const struct wlr_drm_interface *iface_atomic;`",
+      "reference": "And the old names for these variables made more sense imo."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to use `fmt.Sprintf` to construct the string.",
+      "reference": "Are you sure this change is correct?"
+    },
+    {
+      "prediction": "I'm not sure this is a good test. I don't think we should be relying on the existence of a file on disk to determine that the file is an `ico` file. I think we should be checking that the file exists and is a valid `ico` file, e.g. by checking that the file has the correct file extension and that it is a valid `ico` file, e.g. by checking that the file has the correct file extension and that it is a valid `ico` file, e.g. by checking that the file has the correct file extension and that it is a valid `ico` file, e.g. by checking that the file has the correct file extension and that it is a valid `ico` file, e.g. by checking that the file has the correct file extension and that it is a valid `ico` file, e.g. by checking that the file has the correct file extension and that it is a valid `ico` file, e.g. by checking that the file has the correct file extension and that it is a valid `ico` file, e.g. by checking that the file has the correct file extension and that it is a valid `ico` file, e.g. by checking that the",
+      "reference": "The previously linked bug does not apply anymore? If so, this is LGTM!"
+    },
+    {
+      "prediction": "I think this should be `TestFailingLB(t *testing.T)`",
+      "reference": "Why are we renaming all of the tests?"
+    },
+    {
+      "prediction": "I think this should be `nameserver = strings.TrimPrefix(nameserver, \":\")`",
+      "reference": "Instead of having this logic, would it make sense to require *users* to encompass the specified IPv6 address within `[` and `]`? Why the magic handling here?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The problem is that this function is called from many places, and it is possible that the returned value is not the same type as the input. In that case, the exception will be thrown with the wrong type. Instead, you should check the type of `res` before casting it to the right type.",
+      "reference": "Why not also replace this one with calls to `GetProp<ROMol,T>`?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Still need to remove trailing spaces on this line"
+    },
+    {
+      "prediction": "I think it would be better to have this as an interface rather than an interface with a concrete implementation.",
+      "reference": "Please use the `@Experimental` annotation on this"
+    },
+    {
+      "prediction": "I think this is the right thing to do, but I don't know if it's the right thing to do in general. I'm not sure what the right thing to do is in general. I'm not sure what the right thing to do is in this case. I'm not sure what the right thing to do is in general. I'm not sure what the right thing to do is here. I'm not sure what the right thing to do is in general. I'm not sure what the right thing to do is here. I'm not sure what the right thing to do is in general. I'm not sure what the right thing to do is here. I'm not sure what the right thing to do is in general. I'm not sure what the right thing to do is here. I'm not sure what the right thing to do is in general. I'm not sure what the right thing to do is here. I'm not sure what the right thing to do is in general. I'm not sure what the right thing to do is here. I'm not sure what the right thing to do is in general. I'm not sure what the right thing to do is here. I'm not sure what the right thing to do is in",
+      "reference": "I found ~10 more instances of this typo. we can fix those in subsequent PRs."
+    },
+    {
+      "prediction": "I think this should be `metricNamespace` instead of `metricNamespace string`",
+      "reference": "How about making metricNamespace an option, and by default it's the same as `component`? e.g.m Init(component, WithMetricNamespace(\"trigger\"))"
+    },
+    {
+      "prediction": "I think we can remove the `methodInfo.responsePageTokenName()` and `methodInfo.requestPageTokenName()` as they are the same as `methodInfo.nextPageTokenName()` and `methodInfo.nextPageTokenName()`.",
+      "reference": "Just checking: no need for case manipulation?"
+    },
+    {
+      "prediction": "nit: I think it would be better to have a separate method for this. It would make it easier to see what is going on in one place.",
+      "reference": "Same as above, not that descriptive"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. It seems like it should be in the `special_ibl_xfer` function, but I'm not sure what the right place is.",
+      "reference": "> i#4670 signl-interrupted special ibl xfer frag: Unlink if signal pending nit: This looks like one of many commits with the text before the colon. nit: On the long side: 56 or 60 chars is what some views show Spelling error."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. The `pthread_cancel()` is called by the `flb_pipe_r()` and `flb_pipe_w()` and the `pthread_join()` is called by the `flb_pipe_w()` and `flb_pipe_r()`. I think it would be better to call `pthread_cancel()` in `flb_pipe_r()` and `flb_pipe_w()` and `pthread_join()` in `flb_pipe_w()` and `flb_pipe_r()`. The `pthread_cancel()` is called by the `flb_pipe_r()` and `flb_pipe_w()` and the `pthread_join()` is called by the `flb_pipe_w()` and `flb_pipe_r()`. I think it would be better to call `pthread_cancel()` in `flb_pipe_r()` and `flb_pipe_w()` and `pthread_join()` in `flb_pipe_w()` and `flb_pipe_r()`. The `pthread_cancel()` is called by the `flb_pipe_r()` and `flb_pipe_w()` and the `pthread_join()` is called by the `flb_pipe_w()` and `flb_pipe_r()`. I think it would be better to call `pthread_cancel()` in",
+      "reference": "Should we need to invoke pthread_cancel ?"
+    },
+    {
+      "prediction": "I think this should be `userDomainPrefix + \".\" + name + \":domain\"` instead of `userDomainPrefix + \".\" + name + \":\" + userDomainPrefix + \".domain\"`",
+      "reference": "this is auto generated file so no changes are allowed here"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "All buttons are submit buttons, except if they are `type=reset` or `type=button`. I suggest you do an exclude of those, rather than only include `button[type=submit]`."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but it's the only way I could get Travis to pass locally.",
+      "reference": "can you please make this multi line, this version is hard to read."
+    },
+    {
+      "prediction": "I think this should be `if (typeof xyRatio.x === 'number' && isFinite(xyRatio.x))`",
+      "reference": "For consistency, it's best to either always use strict equality (`===`) or loose equality (`==`) in `typeof` checks, and to not mix them."
+    },
+    {
+      "prediction": "Style/SymbolArray: Use %i or %I for an array of symbols.",
+      "reference": "Work Orders would never have only one approver, right?"
+    },
+    {
+      "prediction": "I don't think this enum is used anywhere. Can we remove it?",
+      "reference": "`code` is written to the message as a `byte`, so I think it's better to declare it in the enum as `byte` as well to avoid a narrowing primitive conversion from `int` to `byte` in runtime."
+    },
+    {
+      "prediction": "This should be `numFmt( row.metrics[ 0 ].values[ 0 ], currencyFormat )`",
+      "reference": "This is the only \"extra\" change here."
+    },
+    {
+      "prediction": "I think this should be `allowBLSSignatureCheck` instead of `bypassBLSSignatureCheck`.",
+      "reference": "Please TODO and link to an issue for changing this."
+    },
+    {
+      "prediction": "I think you can remove the comma here.",
+      "reference": "Can we make the script output trailing commas?"
+    },
+    {
+      "prediction": "I think we can remove this feature flag altogether. We don't need it for development.",
+      "reference": "This file should get a file header"
+    },
+    {
+      "prediction": "This file should be renamed to `FileBasedWalIterator.h`.",
+      "reference": "maybe we need process the os error more safely and friendly when open file failed, because this error is very common when cpu has a high pressure, crash directly is danger."
+    },
+    {
+      "prediction": "I think this should be `return big.NewInt(0).Mul(big.NewInt(2), big.NewInt(uint64(gasPrice))).Uint64()` instead of `big.NewInt(0).Mul(big.NewInt(2), big.NewInt(uint64(gasPrice))).Uint64()`",
+      "reference": "we need to pay attention not to use this flag everywhere."
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "curious about this change, since it is not mentioned in the commit message and now we have `uint8` all over the place"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. The problem is that we don't know what the tag ID of the next chunk will be until it's sent to the channel. If we don't know the tag ID of the next chunk, then we can't reliably know when to stop sending chunks to the channel. I think we need to rethink how we're going to send chunks to the channel.",
+      "reference": "wait, are we not mising `WithStamp` here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to get the current block. I think it's better to use `stages.GetStageProgress` to get the progress of the stage and then use `stages.Finish` to finish the stage and get the current block.",
+      "reference": "oh. didn't know we store this stage progress."
+    },
+    {
+      "prediction": "I think this should be `\\U0000-\\U001F` instead of `\\E000-\\E00F`",
+      "reference": "Going with definition from here: > Does Unicode have private-use characters? > A: Yes. There are three ranges of private-use characters in the standard. The main range in the BMP is U+E000..U+F8FF, containing 6,400 private-use characters. That range is often referred to as the Private Use Area (PUA). But there are also two large ranges of supplementary private-use characters, consisting of most of the code points on Planes 15 and 16: U+F0000..U+FFFFD and U+100000..U+10FFFD. Together those ranges allocate another 131,068 private-use characters. Altogether, then, there are 137,468 private-use characters in Unicode. It looks like we are only ignoring ones in BMP and not the supplementary private-use characters. Shouldn't we consider the supplementary ones too?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "This change is un-tested and requires testing"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in `extensions.ipynb`",
+      "reference": "Should the docstring mention the vscode option?"
+    },
+    {
+      "prediction": "I think this should be `packageConfig.packageType() + \"_\" + packageConfig.packageType() + \"_gapic.yaml\"`",
+      "reference": "ISTM that since `artman` know the \"real\" value of the GAPIC config name, it should pass that value to toolkit, rather than toolkit guessing the name based on a heuristic. Then again, I don't know what this value actually does for Java codegen..."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "Please back out the import changes in files you haven't touched. I think these happen if you run goimports without having the vendor directory populated"
+    },
+    {
+      "prediction": "I think this should be `T` instead of `T?`.",
+      "reference": "Should this be `ISocketOutput<T> where T : struct`?"
+    },
+    {
+      "prediction": "I think it would be better to return `Optional<Long> sequenceNumber()`",
+      "reference": "In which case will this actually return 0? If there is no snapshot, then there is no `Snapshot` object, right?"
+    },
+    {
+      "prediction": "I think this should be `dataset.CLASSES`, not `dataset.CLASSES`.",
+      "reference": "This part of code is valuable and necessary. Can we move it to another place for a more clear logic and only check it once? For example, move it to line 133 before delving into each module."
+    },
+    {
+      "prediction": "I think this should be `== \"psycopg2-binary<2.9\"`, right?",
+      "reference": "Are we at all worried that changing pins will cause release hiccups? I think we've decided that relaxing pins should be safe but adding pins has caused us build issues in the past - so I think we're fine?"
+    },
+    {
+      "prediction": "I'm not sure if we want this to be configurable or not, but if we do want to make it configurable, I think we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we do for the other section sizes. That is, we should make it configurable in the same way we",
+      "reference": "could you please explain a bit where these numbers come from? for example: why is there no `swarm.EncryptedSectionSize`?"
+    },
+    {
+      "prediction": "I think we can remove this method and use `r.r.Attrs().ContentType` instead.",
+      "reference": "s/object content/blob object/ (for consistency with the `Size` docs)"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this check. I think it would be better to put it in `org.jvnet.hudson.test.AbstractJenkinsRule` instead of `org.jvnet.hudson.test.CheckJUnitDependencies`.",
+      "reference": "Rather than deleting this entirely, could we just emit it as a `warn` or `info` log line?"
+    },
+    {
+      "prediction": "I'm not sure why we need a domain name here. I think we can just use the domain name from the op.",
+      "reference": "Is there a chance of this being a BC break of somekind? I.e. what if an old instance of Caddy is running in a cluster with a newer one?"
+    },
+    {
+      "prediction": "I think this should be rolled back in the `finally` block.",
+      "reference": "Thanks for this... is this the only place where the rollback is needed? Are there others?"
+    },
+    {
+      "prediction": "I think this is the only change that needs to be made to make this test pass.",
+      "reference": "Should this also be a pointer? What happens if there are no BLS messages? I guess that's what this test is exercising, and Lotus also uses a non-pointer here."
+    },
+    {
+      "prediction": "I think this should be `perr != nil` instead of `pc == nil`.",
+      "reference": "This was definitely a subtle bug right here."
+    },
+    {
+      "prediction": "I think this should be `secondary_emails: fetch_email_responses.select { |hsh| !hsh['primary'] && hsh['verified'] }.compact`",
+      "reference": "If this method gets all the emails, it should be named appropriately. **all_emails** or just **emails**."
+    },
+    {
+      "prediction": "Why is this port used for the bootnode? I don't see it being used for anything else.",
+      "reference": "Can we still random, but if we randomly get a port is used before, we randomly pick again?"
+    },
+    {
+      "prediction": "I think this should be `create_no_window=True`",
+      "reference": "I would rather no have this as a `kwarg` as it encourages \"growth\" which lead to an unweildy constructor in other classes. Let's add a method or property to take care of this instead as I think it's usage is going to be quite low."
+    },
+    {
+      "prediction": "I think this should be `!approver? || !delegate? || !requester? || !purchaser?` instead of `!purchaser? || !approver? || !delegate? || !requester? || !purchaser?`",
+      "reference": "what if a purchaser is also an approver? or would that not happen?"
+    },
+    {
+      "prediction": "I think this should be `if not network.config.has_fee_estimates() or not network.config.has_dynamic_fee_estimates()`",
+      "reference": "Not sure why this change. You forgot to handle the case where the user has dynamic fees disabled and there are no fee estimates available. He should be able to sweep none-the-less (as he is using static fees anyway). The line with `config.fee_per_kb()` below, that you have deleted, handled that."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `auditLogsSourceCondSet`.",
+      "reference": "sorry, I wasn't familiar with this TopLevelCondition... What it would be in this case? The AuditsLogReady condition?"
+    },
+    {
+      "prediction": "I think this should be `*memory.Allocator` instead of `*memory.Allocator*`",
+      "reference": "Where does the `alloc` field get set?"
+    },
+    {
+      "prediction": "I think this should be `int` instead of `int64`",
+      "reference": "looks expensive to do it on each query"
+    },
+    {
+      "prediction": "I think this should be `nupic.datafiles` instead of `nupic.datafiles`",
+      "reference": "@rhyolight -- but weren't we supposed to be standardizing on using underscores in file names (versus camelcase)?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for emptyness, but it's the only way I could come up with at the time of this write.",
+      "reference": "Can this simply call `in_memory.empty?` ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could find that worked.",
+      "reference": "The styling/wording here probably needs some thought. It's more of a placeholder."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be under `users/passwords` or `users/confirmations` or `users/reset_tokens` depending on how you want to view them.",
+      "reference": "Using a plural route helps in detecting the path automatically for `= form_for @password_reset`."
+    },
+    {
+      "prediction": "nit: `inspect` -> `inspect`",
+      "reference": "Is `metadata` required on all tests? I actually just removed the metadata field entirely from a few tests in my PR."
+    },
+    {
+      "prediction": "I think this should be a `FileAlreadyExistsException` instead of an `IOException`.",
+      "reference": "seems reasonable to also want to include the 'to' location?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just remove the `Runtime.getRuntime().removeShutdownHook(shutdownHook);` line from here. The `Runtime.getRuntime().shutdownNow();` call is supposed to be the last thing that is done before the JVM is shut down, so it should be the last thing that is done even if the `Runtime.getRuntime().removeShutdownHook(shutdownHook);` line is removed.",
+      "reference": "I don't think we need to necessarily check if that returned true or not, we should just remove the shutdown hook. Since nothing would check or do anything with this flag anyways."
+    },
+    {
+      "prediction": "I think we can remove the `klog.V(2).Infof(\"uses cstorvolume policy for volume configuration\")` line above.",
+      "reference": "Better to push getting policy details into a func we can reuse later."
+    },
+    {
+      "prediction": "nit: I think we can remove this `Stop` method and just call `o.Outbound().Start()` and `o.Outbound().Stop()`",
+      "reference": "optional: while there's nothing to do, a good way to catch bugs (where we use an outbound without calling `Start`) might be to have this outbound verify that `Start` is called before `Call` or `Stop`"
+    },
+    {
+      "prediction": "I think we can remove `aws/actuators/machine` and `aws/actuators/cluster` now.",
+      "reference": "Not a blocker by any means, but I think there is value in keeping the local imports in a separate group."
+    },
+    {
+      "prediction": "I think this test is failing because it's using `NewDispatcher(...)` instead of `NewOutboundsDispatcher(...)`",
+      "reference": "nit - I would test the error message as well. For panic, you might need to see if the stack contains the error message, instead of equaling."
+    },
+    {
+      "prediction": "I think this should be `var newSlice WaitForInstancesSignal` instead of `newSlice = append(newSlice, replacer.Replace(v))`",
+      "reference": "Didn't we have specific logic for handling slices vs structs?"
+    },
+    {
+      "prediction": "I think this should be `buf.WriteString(message.c_str());` instead of `buf.WriteString(new_message.c_str());`",
+      "reference": "`c_str()` is not needed."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `services/egress-service.go` since that is where the egress controller is.",
+      "reference": "The handler will be installed twice if you don't remove the below one."
+    },
+    {
+      "prediction": "I think this should be `OptionalLong` instead of `Optional<OptionalLong>`.",
+      "reference": "How about putting `epochLength` (or something like that) in place of `a` just to make it easier to read"
+    },
+    {
+      "prediction": "I don't think we should be removing this. It's a warning, and users will see it.",
+      "reference": "Wouldn't this still fire afterwards if you created one called `pleasings`? or am I missing something?"
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to do, but I think it should be removed.",
+      "reference": "I don't love this syntax; it doesn't match anything we're currently doing and it's not obvious. Why isn't this `/keybase/private/jzila,kb_bot/.kbfs_autogit/public/jzila/kbp.git`?"
+    },
+    {
+      "prediction": "`Tuple::of` is a no-op, so we can remove it from the `@SuppressWarnings(\"unchecked\")` list.",
+      "reference": "Does a type-hint work instead of casting? `return ofAll(toList().<U> zipWith(that, mapper));` (Probably not, just a question.)"
+    },
+    {
+      "prediction": "I think this should be `new GenericManifestFile(localInput(\"file:/tmp/manfiest.3.avro\"), SPEC_5.specId()))`",
+      "reference": "Are these required?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `client.go` since it's only used by `initClient` which is in `client.go`.",
+      "reference": "This is consistent for all clients that we will check yes? Seems like we should just use the singleton, maybe pass it to the parse function or since clients have a server pointer set it at server start and just access that way without locks?"
+    },
+    {
+      "prediction": "I think this should be `DECIMAL_LITERAL` instead of `DECIMAL`.",
+      "reference": "Note that this is a bug, that should be fixed on master. `DECIMAL_LITERAL` cannot match any token, because it's declared with a `#`."
+    },
+    {
+      "prediction": "nit: I think we can remove this empty line.",
+      "reference": "Don't need extra line here."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "For the first look, this change is strange. Were we passing these tests before?"
+    },
+    {
+      "prediction": "I think this should be `matchProcess(appname string, processname string) (bool, error)`",
+      "reference": "On Windows, doesn't `os.Args[0]` include the `.exe`? What if you run the program like `caddy.exe`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should go in the `negotiator.go` file, under the `negotiateOpenvpnConnection` function.",
+      "reference": "I don't see anywhere where `ServiceConfiguration` interface implementation would return any kind of error. Not sure if its needed, but we could ALWAYS return a valid 'pseudo' configuration without possibility of error."
+    },
+    {
+      "prediction": "I think this should be `containsExactlyInAnyOrder` instead of `containsExactly`",
+      "reference": "Do we actually want to split on `\\r`? I thought that the \"newline sequence\" was only `\\r\\n` on windows."
+    },
+    {
+      "prediction": "I think this should be `else if (ttype->is_object())`",
+      "reference": "Nit: these lines are duplicated from above. They can be refactored by setting etype in a conditional and putting 282 and 283 below that."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `Shutdown` function.",
+      "reference": "I would do an s.Notice(\"Initiating Shutdown...\") then after Shutdown() do the original Server Exiting."
+    },
+    {
+      "prediction": "Shouldn't this be `if (timeLeftToClaimRefund < 0)`?",
+      "reference": "Not sure with how many seconds on average are we dealing with but maybe it's better to have hh:mm:ss format in logs - you can make it with `TimeSpan.FromSeconds(seconds).ToString()`"
+    },
+    {
+      "prediction": "I think it would be better to have this validation in `validateUpdate` instead of `validateCreate`.",
+      "reference": "Why is this required? The user should be able to omit it and use the defaults."
+    },
+    {
+      "prediction": "I think this should be `self.handleInjectedMouseControlCheckBox.SetValue(config.conf[\"mouse\"][\"handleInjectedMouseControl\"])` instead of `self.handleInjectedMouseControlCheckBox.SetValue(config.conf[\"mouse\"][\"handleInjectedMouseControl\"])`",
+      "reference": "I don't think the naming of this setting conveys what it does. Maybe something like \"Ignore mouse movement triggered by other applications\""
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `Uri` property on the `OpenTelemetryProtocolExporterEventSource` and have the `OpenTelemetryProtocolExporterEventSource` do the `Uri.Parse` itself. This way you don't have to pass in the `collectorUri` to the `OpenTelemetryProtocolExporterEventSource` and you don't have to pass in the `ex` to the `OpenTelemetryProtocolExporterEventSource`.",
+      "reference": "Didn't went with backwards compatibility since it's still in beta"
+    },
+    {
+      "prediction": "I think this should be `throw_value_error(\"invalid value for `bondsToCut`\");`",
+      "reference": "Perhaps bondsToCut must be None or non empty."
+    },
+    {
+      "prediction": "I think it would be better to have a `chartType` prop that is either `line`, `bar`, or `scatter`. The `line` and `bar` are the most commonly used, so those should be the default. The `scatter` is a little less commonly used, so I think it would be better to have the `chartType` be a string, and the `bar` and `scatter` be props.",
+      "reference": "See above, we could avoid adding that (same applies below)."
+    },
+    {
+      "prediction": "I think you can remove the `typeMap.get(\"airplaneMode\").toString()` line as well.",
+      "reference": "should use Number instead of Long"
+    },
+    {
+      "prediction": "I think this should be `chai-as-promised`",
+      "reference": "Should the rest of the file be updated to use `expect` or should this test use the same format as the rest of the tests?"
+    },
+    {
+      "prediction": "I think this should be `return a.ModifyStatus()` instead of calling `a.ModifyStatus()` again.",
+      "reference": "How would you feel about a different approach where the `DeleteAWSRecordSets` gets passed the zone ID and zone name rather than the `DNSZone`?"
+    },
+    {
+      "prediction": "`prob` is not used in `Expand`, please remove it.",
+      "reference": "it seems `random.uniform(0, 1)` similar to `random.randint(2)`, all have 1/2 probabilities."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to iterate over the items. I think it would be better to iterate over `items[i]` instead of `items[i-1]` and `items[i-2]` etc.",
+      "reference": "Added to support EBS volume creation, used like this: blockDeviceMap = [] blockDeviceMap.append( {'DeviceName':'/dev/sdc', 'VirtualName' : 'ephemeral0'}) blockDeviceMap.append( {'DeviceName':'/dev/sdh', 'Ebs': {'VolumeSize' : '100'} })"
+    },
+    {
+      "prediction": "I think this should be `testMysteriumApiUrl = \"http://testUrl\"`.",
+      "reference": "We shouldn't depend that url is `mysterium` - it can be just `testRequestURL`. Also, `URL` has to be upper-case."
+    },
+    {
+      "prediction": "Is this lock needed? I don't see it being used anywhere in this function.",
+      "reference": "We could delay this look-up now and only do it in the non-skip case. I suspect this is true for (almost) all of the top-level refactored functions."
+    },
+    {
+      "prediction": "I think we can remove the `reinterpret_cast` here, since the size of `reinterpret_cast<GraphSpaceID>(spaceId)` and `reinterpret_cast<TagID>(tagId)` are both `sizeof(GraphSpaceID) + sizeof(TagID) + sizeof(GraphSpaceID) + sizeof(TagID) = sizeof(GraphSpaceID + TagID)`",
+      "reference": "Typically, to be simple, we could reserve a const length space."
+    },
+    {
+      "prediction": "Please remove this file.",
+      "reference": "Need to add a license and copyright header to each file."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I've been able to get it to work.",
+      "reference": "If I'm not mistaken, this line is now outside of the `if signed_out?` block, right? Any concerns about that? Seems odd that we wouldn't accept coupons for signed in users, but I want to make sure we understand the ramifications of this change."
+    },
+    {
+      "prediction": "I think it would be better to make this `@Restricted(NoExternalUse.class)`",
+      "reference": "Please remove the unused import of DBPDataSource."
+    },
+    {
+      "prediction": "I think this should be renamed to `LogWarning` instead of `LogError` to be consistent with the rest of the `Log` methods.",
+      "reference": "I think you should add some extra message here saying it was originally sent as an error, for diagnosability."
+    },
+    {
+      "prediction": "I think this should be changed to `TEST_ASSERT(bv.getNumOnBits() == 48);`",
+      "reference": "I believe that all the changes in this file are not valid for v1.2 of the Avalon toolkit."
+    },
+    {
+      "prediction": "I think this should be done in the constructor of the DBService.",
+      "reference": "we cannot change the behavior of the processRole - that method is used in lots of places. this call must be done only in the method where templates are being handled."
+    },
+    {
+      "prediction": "I think this should be `ExecAgentLogFileName` instead of `configuration/`",
+      "reference": "I think we probably want to follow the same naming convention that we do with `execAgentConfigFileNameTemplate` (using the SHA in the name of the file). This file might be confgurable in the future and when that happens we will be ready. Plus it's pretty much the same code that we already use for the config file."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to send the time left. I think it would be better to send it in the `work_queue_resources_received()` function. The reason is that the `work_queue_resources_received()` function is called from the `send_resource_update()` function, and the `send_resource_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the `link_update()` function, and the `link_update()` function is called from the",
+      "reference": "If following above, this would be: end_time = time(0) + manual_wall_time_option, which is simpler. Also, make the check manual_wall_time_option > 0, otherwise negative times would terminate the worker right away."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "If we aren't using `errors.Is` we don't need the `Unwrap` as we have `Cause` which is what pkg/errors uses."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "redundant empty line"
+    },
+    {
+      "prediction": "This should be renamed to `accountOwnedBy`.",
+      "reference": "The config.lua.dist still shows `houseAccountOwner`"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "I think we can leave this import and use non FQNs in the code."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `switchTo` command, since that is the command that switches to a different window.",
+      "reference": "I'd rather just send the parameter twice than break encapsulation here. There's already precedence with webelement IDs"
+    },
+    {
+      "prediction": "Why was this moved out of the `PushItemsToFeedAsync` method and into the `PushItemsToFeed` method?",
+      "reference": "We should probably exit early if there are no items in the list."
+    },
+    {
+      "prediction": "I don't think this is a good example of what we should be doing here. I think it would be better to have an example of a public function that takes a param with a missing type doc.",
+      "reference": "As we can't use old names we should warn in whats new for 2.12 that this can happen."
+    },
+    {
+      "prediction": "I think this should be `// +build !testbincover`",
+      "reference": "good job finding this package, I hope it's actively maintained"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The `CUDA_VISIBLE_DEVICES` environment variable is used by the CUDA runtime to determine which devices are visible to the user. It's not used by the HIP runtime at all. I think the right fix is to remove the `CUDA_VISIBLE_DEVICES` environment variable from the `getDevicePCIBusNum` function so that it doesn't interfere with the `CUDA_VISIBLE_DEVICES` environment variable used by the HIP runtime.",
+      "reference": "same changes as in hipEnvVar.cpp"
+    },
+    {
+      "prediction": "This should be `#include <wlr/wlr.h>`.",
+      "reference": "NULL check should be before this line"
+    },
+    {
+      "prediction": "This should be `if (client.getLocalPlayer() != null && client.getLocalPlayer().getWorldLocation() != worldPoint)`",
+      "reference": "Seems like client.getlocalplayer is being called a lot, would it be worth it to do final Player player = client.getlocalplayer?"
+    },
+    {
+      "prediction": "I think this should be `(Lng32 *)op_data[2 * MAX_OPERANDS]` instead of `(Lng32 *)op_data[2 * MAX_OPERANDS] = 0;`",
+      "reference": "this change is not correct. 3-valued boolean logic requires that comparison of null to another value during intermediate operations should make the result null and not false. The place where null becomes false is at the very end of that expression. As an example, with the changed logic, this expression will return incorrect result. select * from t where not (a = 1); if a is null, the where predicate will become true which is incorrect. With the original 3-valued result, the where pred will become false and not return any result."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to get the locale from the request. I think it would be better to get the locale from the `X-Frame-Options` header and then from the `X-Content-Type-Override` header if the `X-Frame-Options` header is not set.",
+      "reference": "This can already be done with `{>Detected-Locale}`, why make another way to do it?"
+    },
+    {
+      "prediction": "I think this should be `if (fmd.isRestricted() == false) {`",
+      "reference": "These changes just match the logic update in the next method that was done previously - to assure the whole set of files is valid rather than that there is at least one that would be changed."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `CreateTag` function, right?",
+      "reference": "Why setting the status to Continue as at the end it is written OK? Status should be written only once."
+    },
+    {
+      "prediction": "I think this should be `./proofs/bin/paramcache` and not `./scripts/copy-groth-params.sh`",
+      "reference": "Howdy! You'll want to do this same thing (replace `hydrateParamCache` with `proofs/bin/paramcache` and then do the copy) in `smartdeps`, too."
+    },
+    {
+      "prediction": "I think this should be `note: the volume must exit before itself`",
+      "reference": "nit : Remove the extra lines"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `server.go` under the `ServerConfig` struct.",
+      "reference": "Instead of adding a new option, I wonder if we could decide on a port that would disable listening. For instance, port set to 0 means that we use default port 4222. Setting to -1 means that we let OS pick a random free port. We could say anything negative lower than -1 (say -2) means disabled? You don't have to update the PR for that, just wondering if that would be better or not and let's see what others think."
+    },
+    {
+      "prediction": "I think this should be `report.check_name` instead of `clang-diagnostic-`",
+      "reference": "Should we add some default case if non of the above return some analyzer name? Something like `unknown analyzer`?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Nvm. The original version is best... *hides from the angry rubocop*"
+    },
+    {
+      "prediction": "I think this should be `defer cleanup()` and not `defer body, err := ...`",
+      "reference": "we can drop this and use the returned body above, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I don't have a good suggestion either.",
+      "reference": "Is echoing enabled on purpose here or it's a remainder of your tests?"
+    },
+    {
+      "prediction": "I think this should be `@logger.info(\"Starting command run '#{command}' on #{targets.map(&:uri)}. Description: #{options['_description']}.\") if options.key?('_description')`",
+      "reference": "I think this should just be worked into the next message and follow the verbosity of it."
+    },
+    {
+      "prediction": "I think this should be `@param {string|undefined} urlParams URL parameters to be passed to the query.`",
+      "reference": "This also needs to be reverted."
+    },
+    {
+      "prediction": "I think this should be `TlfJournalPauseType` instead of `TlfJournalPauseType_tlfJournalPausedFromConflict`",
+      "reference": "`FromCommand` seems to fit better, since with my suggestion below, all pauses will raise a signal on `needPause`. Also maybe the format `journalPauseConflict` and `journalPauseCommand` is better, since the journal isn't necessarily paused yet once we raise a pause signal."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "This is fix from another PR #120."
+    },
+    {
+      "prediction": "I think this should be `multi = False` instead of `multi = True`",
+      "reference": "I *think* it makes sense to try the more general dictionary (i.e standard python literals) format first. Might be other implications I haven't figured out yet. Then again, ``MultiInterface`` is pretty new so it probably doesn't matter wrt backwards compatibility."
+    },
+    {
+      "prediction": "I think this should be `sonata.media.manager.classification` and not `sonata.classification.manager.category`",
+      "reference": "lol that variable name"
+    },
+    {
+      "prediction": "I'm not sure if this is the right name for this. I think it should be something like `ChainFailsafeIncoming` or `ChainFailsafeOutgoing` or something like that.",
+      "reference": "Upper case is reserved for our versions of top-level chains i.e. the `FORWARD` chain jumps to `cali-FORWARD`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do it. The idea is to have a function that generates a representative for a new account, and then call that function when opening a new account.",
+      "reference": "json_handler::receive () has already started a read tx that can be used here right?"
+    },
+    {
+      "prediction": "I think this should be `global._googlesitekitBaseData = { ...global._googlesitekitBaseData, isWP5.0+ }` instead of `global._googlesitekitBaseData[ 'isWP5.0+' ] = true;`",
+      "reference": "We should add the definition to `.storybook/config.js` as well."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it should be `if ( *(Lng32 *)op_data[1] == 0 || *(Lng32 *)op_data[1] == -1) {`",
+      "reference": "I don't believe this is correct. Consider the query, \"select a from t1x where not(b = 0 and c = 0)\". When B and C are both null, both equal predicates evaluate to null, and the AND evaluates to null. The NOT then also evaluates to null. The WHERE clause should treat the result of the NOT as false. But with this fix, the result of the AND will be false, making the NOT true. There needs to be three cases here for ITM_AND: If the first operand is false, then the AND is false. If the first operand is true, then the result is the second operand. If the first operand is null, then if the second operand is false, the result is false otherwise the result is null. Similar logic needs to be added to the ITM_OR case."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The idea is that if the target is less than the current element in the array, then we know that the current element is the largest element that is less than the target, so we can skip ahead to the next element in the array. If the target is greater than the current element in the array, then we know that the current element is the smallest element that is greater than the target, so we can skip back to the previous element in the array. If the target is equal to the current element in the array, then we can skip ahead to the next element in the array. If the target is greater than or equal to the current element in the array, then we don't know what the current element is, so we need to check the next element in the array, and so on.",
+      "reference": "`bound/2` is generally the previous bound that we tested, except when `bound` is equal to 1. It won't break in that case since callers are not supposed to call advance on a target that is lte the current doc ID, but this might still make room for bugs?"
+    },
+    {
+      "prediction": "I think this should be `ReadPreference|string|undefined` and `options.fullResult` should be `boolean|undefined`",
+      "reference": "Can we remove `raw`, `fullResult`, and `serializeFunctions`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be under `auth` or `authorization` or something similar.",
+      "reference": "[Nit] I don't have a proposed suggestion, but security seems too generic of a name here."
+    },
+    {
+      "prediction": "I think this should be `config.buildConfig = 'Release'` instead of `config.buildConfig = 'Release' && config.defaultOptions === 'Release'`",
+      "reference": "do we need to force a buildConfig here? I know it's in muon, but that might actually be making things harder for people"
+    },
+    {
+      "prediction": "This should be `auto_fire = false;` instead of `auto_fire = true;`",
+      "reference": "Shouldn't need this-> here."
+    },
+    {
+      "prediction": "why remove this line?",
+      "reference": "Why delete this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `amp-form.js` or `amp-input.js`",
+      "reference": "This line is here to be compatible with the current way `refs` work, right?"
+    },
+    {
+      "prediction": "I think this should be `number_to_currency(value.to_s)` instead of `number_to_currency(value)`",
+      "reference": "Hmm, I wonder if we can safely assume all decimals should be displayed as $$...ok for now I suppose."
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundBlockBody: Extra empty line detected at block body end.",
+      "reference": "Good spec :). Very clear and easy to see what it's doing."
+    },
+    {
+      "prediction": "`params[:filter]` is a string, not an integer, so you can't use `params[:filter].to_i` to convert it to an integer. You'll need to pass in an integer here.",
+      "reference": "Shouldn't we need to sanitize the `params[:filter]` from a defined expected values?"
+    },
+    {
+      "prediction": "I think this should be 100, not 51.1. I don't think we want to back off at 51.1 seconds.",
+      "reference": "Should we bite the bullet and poll for the full 5 minutes that AWS recommends? If we do that, we'll probably want to leave some kind of breadcrumb in the event log to let users know why it's taking so long to launch."
+    },
+    {
+      "prediction": "nit: remove this import.",
+      "reference": "Guava changed `Objects` -> `MoreObjects`"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "All these hand generated mocks can be deleted. I think we already moved everything to gomocks."
+    },
+    {
+      "prediction": "I think this should be `===` instead of `equals`",
+      "reference": "what about string/number etc _ids? this will fail."
+    },
+    {
+      "prediction": "I think this should be `eMS.Labels` and `rMS.Labels` instead of `eMS.Spec.Template.Labels` and `rMS.Spec.Template.Spec.Labels`",
+      "reference": "Looks like the test had a bug prior to this."
+    },
+    {
+      "prediction": "I think this should be `process.platform === 'win32'` instead of `env === 'dev'`",
+      "reference": "Should start with joplin://x-callback-url/. Also maybe create a helper function to check if a url is valid?"
+    },
+    {
+      "prediction": "This should be `id: 1`, not `id: 7`.",
+      "reference": "PR works great. Just out of curiosity: why all the id changes?"
+    },
+    {
+      "prediction": "Shouldn't this be `FLB_CONF_TYPE_INT64` instead of `FLB_CONF_TYPE_INT`?",
+      "reference": "config map first value is wrong, it aims to define the \"value type\", that value will lead to an undefined behavior."
+    },
+    {
+      "prediction": "`*` is a reserved keyword in Go. I think it would be better to use `import \"github.com/pulumi/pulumi/sdk/v2/go/bucket\"` so that we don't accidentally import anything else that might conflict with this package.",
+      "reference": "This doesn't make sense. The thing being constructed is a blob.Bucket. \"Constructing\" the pointer is trivial."
+    },
+    {
+      "prediction": "I think this should be `import optparse as optparse`.",
+      "reference": "Outch. A configuration/parsing refactor is on the radar but this seems more pressing."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with.",
+      "reference": "Unfortunately this won't work when the \"All time\" range is chosen because `memberStats.days` will be `\"all-time\"` and not a number. What you could do instead is to use `memberStats.stats` as the dependent key and use `Object.keys(memberStats.stats.total_on_date)[0]` to grab the first date in the range and build the label from there."
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this scenario. It's not about the user clicking the call to action, it's about the user signing in, clicking the call to action, and then clicking the \"view video trail\" button.",
+      "reference": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this is the only place where `configureOptions` is used, so I think it's fine to remove it.",
+      "reference": "We need this left in as `axe.configure` is for the top-level page `axe` and we still need to configure the axe script inside the iframe as well (so both have to be configured),"
+    },
+    {
+      "prediction": "I think this should be ValidationCategory.RPCMethod.",
+      "reference": "is this rule serviceimpacting? I'm thinking it's mostly around the modeling of pageable in the swagger, thoughts?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `TriggerStatus` struct itself.",
+      "reference": "same here, replace `ks` and `kc` to something more related to Source."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this converter. It's not specific to MVVM, so I don't think it should be in `MvvmCross.Platform.Converters`. I think it should be in `MvvmCore.Platform.Converters`.",
+      "reference": "We should probably ask here if `dict` contains a key with the appropiate value, and leave a trace in case it doesn't, to make it easier to debug for developers."
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.",
+      "reference": "this should only be indented 2 spaces"
+    },
+    {
+      "prediction": "I think this should be `if err != nil { t.Errorf(\"failed to load env config, %v\", err) }` instead of `if err != nil { t.Errorf(\"failed to load env config, %v\", err) }`",
+      "reference": "Nit, these Err not nils should be `t.Fatalf` not error. We want to the test to stop if this error occurs."
+    },
+    {
+      "prediction": "I think we should rename this to `hasDownloadOriginalTracker` to be consistent with `hasTabular` and `hasRsyncScript`.",
+      "reference": "shouldn't every `Boolean` in this pr be `boolean`?"
+    },
+    {
+      "prediction": "I don't think we need to check for w3c here. We can just do `self.execute(Command.MAXIMIZE_WINDOW)`.",
+      "reference": "This `if` is not necessary"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "Are you using this anywhere? (I couldn't find it.)"
+    },
+    {
+      "prediction": "I think it would be better to have this method on the `Step` interface. Then, when implementing `Step`, you just call `s.recordStepTime` instead of the expensive call to `time.Now()`.",
+      "reference": "Can be inlined"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "`json` is not used in this PR, can you remove it?",
+      "reference": "Please use `kinto.core.utils.json`"
+    },
+    {
+      "prediction": "I'm not sure this is the best error message. I think we should be more specific about what port ranges are being used.",
+      "reference": "Since we're returning the error with the same content, we don't need to log it here; it will get bubbled up and emitted that way."
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think we should be able to use `@final` everywhere.",
+      "reference": "What happens if someone import `typing` and thus uses `@typing.final` instead?"
+    },
+    {
+      "prediction": "I think this should be `pandas==0.20.0`, not `pandas==0.19.1`",
+      "reference": "Better remove this dependency altogether"
+    },
+    {
+      "prediction": "I think this should be `extern const h2o_buffer_prototype_t h2o__http2client_wbuf_buffer_prototype;`",
+      "reference": "Could you please move the declaration to httpclient.h? I think we have a consistent pattern in declaring global variables in header files."
+    },
+    {
+      "prediction": "I think this should be 142.",
+      "reference": "We will also need to add this to the spec."
+    },
+    {
+      "prediction": "I think you can remove the `sdeltas` field from the `makeRoundCowState` function.",
+      "reference": "In startEvaluator, we have the paysetHint; we should use it to preinitialize the state delta allocation. The number of transactions and the deltas are typically growing in a linear fashion, so preallocating twice as much in the array initializer would be a good idea."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. This is something that should be handled by the browser, not the driver.",
+      "reference": "we shouldn't add a bunch of static initializers 'everywhere' in the code. Probably just one would be good, during the construction of the 'RemoteWebDriver' class."
+    },
+    {
+      "prediction": "I think this should be `public string SerializeObjectProperty() => JsonSerializer.Serialize((object)_value);`",
+      "reference": "Would it be possible to move the creation of the `new { Prop = (object)_value }` out of the benchmark? In the current version, the benchmark includes the cost of creating and serializing the object, while if possible it should be the cost of serialization only."
+    },
+    {
+      "prediction": "This should be `Translation\\PlatformAdapter\\Loco\\Bridge\\Symfony\\TranslationAdapterLocoBundle::class => ['all' => true],` and not `Http\\HttplugBundle\\HttplugBundle::class => ['all' => true],`",
+      "reference": "only dev and local"
+    },
+    {
+      "prediction": "I think this should be `null` instead of `nullable`.",
+      "reference": "This should not be nullable, instead it should be `->default(0);`"
+    },
+    {
+      "prediction": "I think it would be better to use `ImmutableList.copyOf` with a `Set` instead of `ImmutableList.copyOf` with `Collections.emptySet()`.",
+      "reference": "What about `selectedColumns`? I believe it is frequently used throughout the project."
+    },
+    {
+      "prediction": "I think this should be called `synonymBoostFactor` instead of `synonymBoostByPayload`.",
+      "reference": "I thought we switched the approach from a payload to boost attribute? Besides; it's not clear we need this toggle at all since the user could arrange for this behavior simply by having the new DelimitedBoost filter thing in the chain."
+    },
+    {
+      "prediction": "I don't think we should be changing the behavior of `# environment_string` in this PR. It seems like we should only change the behavior of `# environment_variable_string_pair_array`, but I don't think we should be changing the behavior of `# environment_string` in this PR.",
+      "reference": "the previous code used 'export' when platform is cisco-5, presumably because 'env' cannot be used to load variables? I have no idea if that's the case but this env vs export logic is not here"
+    },
+    {
+      "prediction": "I think this should be `+optional`",
+      "reference": "I'm wondering if we need to add some type of validation here that the SecretCount != 0... Or do we think it would be valid to have a SecretCount of 0?"
+    },
+    {
+      "prediction": "I think this should be `{ limit: 0, skip: 0, slaveOk: false, secondaryOk: false, readPreference: 42 }`",
+      "reference": "This test is being skipped. I unskipped it, and it failed in the same way for both `slaveOk: false` and `secondaryOk: false`. I'm not sure how else to test this broken test."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `OutboundCall` struct, since that's where it's used.",
+      "reference": "I don't believe that we need a pointer to an interface for the `to` side of this since we're just assigning a value to the to field."
+    },
+    {
+      "prediction": "I think this should be `times` * `count` instead of `times` * `count *`",
+      "reference": "No blank line after the docstring."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "nit: move this to the group below."
+    },
+    {
+      "prediction": "I'm not sure this is correct. The `CostScalar` type is a `CostScalarUInt32` and `CostScalarUInt32` is a `CostScalar` with a `UInt32` value. So the `CostScalar` type is a `CostScalarUInt32` and `CostScalarUInt32` is a `CostScalar` with a `UInt32` value. So the `CostScalar` type is a `CostScalarUInt32` and `CostScalarUInt32` is a `CostScalar` with a `UInt32` value. So the `CostScalar` type is a `CostScalarUInt32` and `CostScalarUInt32` is a `CostScalar` with a `UInt32` value. So the `CostScalar` type is a `CostScalarUInt32` and `CostScalarUInt32` is a `CostScalar` with a `UInt32` value. So the `CostScalar` type is a `CostScalarUInt32` and `CostScalarUInt32` is a `CostScalar` with a `UInt32` value. So the `CostScalar` type is a `CostScalarUInt32",
+      "reference": "This is fine, but maybe we should change the variable name to \"bufferSize_as_uint\"? Or \"...uint32\"?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Use snake_case for variable names."
+    },
+    {
+      "prediction": "This is a test for the `RawData` class, not the `Console` class.",
+      "reference": "Obviously, don't merge this to master."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `spec/spec_helper.rb` under `context 'spec reports' do`",
+      "reference": "Nearly there! This should assert on the output, take the result of \"start\" and \"finish\" as an example :)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "This loop is going to burn CPU if len(s.q) == 0. Maybe a small sleep?"
+    },
+    {
+      "prediction": "I think it is better to use `ImmutableList` and `ImmutableMap` instead of `ArrayList` and `HashMap`.",
+      "reference": "can you give a reason why session should be contained in RegionManager?"
+    },
+    {
+      "prediction": "I think this test can be removed now.",
+      "reference": "You are comparing two signerIDs here, error message should be \"Signers should match\""
+    },
+    {
+      "prediction": "nit: I think we can remove the `lock.RWMutex` from the `TaskIAMRoleCredentials` struct and the `GetIAMRoleCredentials` method.",
+      "reference": "Who's acquiring this lock for writes?"
+    },
+    {
+      "prediction": "I think we should keep the `compressed` parameter. It's used in the `PublicKey.fromData` method.",
+      "reference": "This throws a linter error, don't use object creation for side effects. I feel like it's kinda messed up that the errors are thrown by the constructor and not that the constructor will throw them"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it should go in `spec/models/trail.rb`, but I don't know for sure.",
+      "reference": "Should there be a `trail.topic_name` method so we don't annoy Demeter?"
+    },
+    {
+      "prediction": "I think this should be `team` instead of `username`.",
+      "reference": "instead of passing \"None\" it would be more readable to create a constant e.g. TEAM_PUBLIC = None and then call build_package(TEAM_PUBLIC, ...)"
+    },
+    {
+      "prediction": "This should be `__DIR__ . '/internal/exception_handler.php'` instead of `__DIR__ . '/Psalm/ExceptionHandler.php'`.",
+      "reference": "Would it make sense to include this file via composer files-autoloading instead?"
+    },
+    {
+      "prediction": "I think this should be `String elementText = element.getAttribute(\"text\");`",
+      "reference": "Each `ExpectedCondition` implements `java.util.Function` These are expected to be stateless. This condition will leak previous `elementText` on the second usage, which doesn't seem ideal."
+    },
+    {
+      "prediction": "I think this file should be renamed to reflect its new purpose.",
+      "reference": "Extra newline here. Thanks for moving it under the `testing` import though. This just needs to be grouped with the other imports below it."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The problem is that we don't know what encoding is used by the body. So we can't know if we should use quoted-printable or 8bit here.",
+      "reference": "Should we ignore case when doing the comparison?"
+    },
+    {
+      "prediction": "I think this should be `#if defined(WLR_HAS_X11_BACKEND) && WLR_HAS_X11_BACKEND` instead of `#ifdef WLR_HAS_X11_BACKEND`",
+      "reference": "The same `ifdef` should apply to `#include <wlr/backend/x11.h>` line 14"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to log this error. I think it should be in the `BeforeAction(...)` method itself, since it's the first thing that gets called.",
+      "reference": "We're instrumenting a static method, so there's no instance in this case. `diagnosticSource` is the first argument."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to move it to the `MessageCompose` class itself.",
+      "reference": "I feel like this belongs in `RecipientPresenter` (which should really be `MetadataPresenter`), but I suppose having it next to requestReadReceipt makes sense as well. feel free to leave as is."
+    },
+    {
+      "prediction": "Can you update the version to 1.5.0?",
+      "reference": "Let's go back to 1.4.0 until the release is being prepared."
+    },
+    {
+      "prediction": "(style) 'errno' may be misspelled - perhaps 'errors'? (style) 'errno' may be misspelled - perhaps 'errors'?",
+      "reference": "I would avoid the term \"errno\" as it implies the glibc variable. Use error_code or status or something."
+    },
+    {
+      "prediction": "I think it would be better to have this as an argument to `Start` rather than an argument to `inbound.Start` and `inbound.StartWithDependencies`",
+      "reference": "@abhinav @kriskowal what do you think about putting the handler on `d`?"
+    },
+    {
+      "prediction": "I think this should be `restic.TagList` instead of `[]restic.Tag`",
+      "reference": "Why `restic.TagLists` and no `restic.TagList` ? Because I'm using `restic.TagList` here and using `restic.TagLists` is a lot of changes"
+    },
+    {
+      "prediction": "I think this is a bit of a hack, but I think it should work.",
+      "reference": "An hour seems a significant amount of time to wait for within a test, does it need to be that long?"
+    },
+    {
+      "prediction": "I'm not sure why we need to pass `false` here. I think we can just call `dial` with `tlsClientConfig` as the first argument.",
+      "reference": "Nit: (newish to golang, don't hesitate to tell me I'm wrong here, but ...) I'd have written this code, just for readability: keepAlive := false return d.dial(hostName, tlsClientConfig, keepAlive) ... or even return d.dial(hostName, tlsClientConfig, false) // keepAlive=false Personally I think Golang should have included named args; I think it's one of the things python got right."
+    },
+    {
+      "prediction": "I think this should be `Msmq` instead of `Mql`",
+      "reference": "I don't think it would cause any version conflict, but I'd feel safer if you declared it last, so that the numerical id of the existing integrations don't change"
+    },
+    {
+      "prediction": "I think this should be `CertificateRequest %q is an invalid request`, and the message should be `The CertificateRequest %q is an invalid request`",
+      "reference": "Can we provide some more useful actionable info here? `fmt.Sprintf(\"The certificate request could not be completed due to invalid request options: %s\", req.Status.Conditions[InvalidRequestCondition].Message)`"
+    },
+    {
+      "prediction": "I think this should be 15015.",
+      "reference": "yeah, can change to 8545"
+    },
+    {
+      "prediction": "I think this should be `CachedUser` instead of `UserRepositoryInterface`",
+      "reference": "As far as I can see you're not rollbacking transaction anywhere. This could potentially cause further issues(and very hard to debug ones) with not synchronized transactions between message handling. Is the transaction required in this place anyway? You seem only to be fetching data here. ack method is performing some updates so it makes sense"
+    },
+    {
+      "prediction": "I don't think we need to import `org.springframework.stereotype.Component` as it is already imported in `io.servicecomb.foundation.common.net.NetUtils`",
+      "reference": "What's the meaning of null URIEndpointObject?"
+    },
+    {
+      "prediction": "I think this should be `DEFAULT_LOG_LEVEL = logging.DEBUG` and `DEFAULT_LOG_FORMAT = \"%(levelname)-5.5s  %(name)-12s: %(message)s\"`",
+      "reference": "Why did you remove the time? It can be useful if some actions are ran in jobs?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it. I'm not sure if it's possible to have a `mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `go-mega` package. I'm not sure if it's possible to have a `",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "What the reason to change imports?"
+    },
+    {
+      "prediction": "I think it would be better to move this method to `HiveClientPool` instead of `HiveClientPool` and `HiveClient`",
+      "reference": "Nit: continuation indents are 2 indents / 4 spaces."
+    },
+    {
+      "prediction": "I don't think we need to check for `None` here, since `parsed_url.hostname` will never be `None`.",
+      "reference": "You can use `netloc.rsplit(\":\")` for that."
+    },
+    {
+      "prediction": "I think we can remove this line too.",
+      "reference": "nits, I don't get the point of this change"
+    },
+    {
+      "prediction": "I think we can remove this file and just use `errorPrefix` and `warningPrefix` in the `log` package.",
+      "reference": "`warningPrefix` looks the same on both platforms, we could move it up to `log.go`."
+    },
+    {
+      "prediction": "I think we can remove the `Filter` and `Interceptor` fields from the `Config` struct, since they are no longer used.",
+      "reference": "What's the motivation behind switching to plural here? I think we want singular cc @abhinav"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think we should move this to the `Status` struct itself.",
+      "reference": "What do you think about renaming this method `FromErrorAndCode(code Code, err error) *Status`? One of the benefits is that it will be consistent with the naming of `func FromError(err error) *Status {` which does something almost similar."
+    },
+    {
+      "prediction": "I don't think we should change the default port for websockets.",
+      "reference": "we cannot state that by default the same value is used if we stopped using null - it would suggest that when you change JSON RPC and leave WS port default then they would be same."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do here, but it seems like we should be able to just pass in `GinkgoWriter` and `GinkgoWriter` to `gexec.Start` without having to pass them in as arguments to `gexec.Start`.",
+      "reference": "I'm still not convinced dropping Chrome logs is the right thing to do, but if you insist"
+    },
+    {
+      "prediction": "I don't think this is right. We should not throw an exception here, but return an empty string.",
+      "reference": "It would be helpful to include some reasoning here. With the reflection model, it is often the case that names are not available."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to the `log` package.",
+      "reference": "Instead of setting this here, why not have the flag call in the init function set `caddy.LogDestination` directly, and switch on that?"
+    },
+    {
+      "prediction": "I don't think we should be using `__` here. It's not used anywhere else in the codebase, so it's probably best to just remove it.",
+      "reference": "This belongs under \"WordPress dependencies\"."
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "Was there an issue on `net452`?"
+    },
+    {
+      "prediction": "I think it would be better to use `String#StartsWith(String, StringComparison)` instead of `String#StartsWith(String, StringComparison)`",
+      "reference": "Another option is to always write it without quotes. I always have to use SR. And why not?"
+    },
+    {
+      "prediction": "I don't think we need this file at all.",
+      "reference": "This should use the `merge` function added in #473 instead of introducing a new dependency"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `descriptorSetLayout` check above.",
+      "reference": "Shouldn't this say \"a binding with descriptor type VK_DESCRIPTOR_TYPE_MUTABLE_VALVE\""
+    },
+    {
+      "prediction": "This should be `static const struct wlr_tablet_tool_v2_grab_interface default_tool_interface = { .name = \"wp_tablet_tool-cursor\" };`",
+      "reference": "This variable name should probably contain `grab`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `config/initializers/openaire.rb`",
+      "reference": "Which initializer should we be using for open_aire? The one in `external_apis` or this one in `initializers`?"
+    },
+    {
+      "prediction": "I think we need to add the routing package to the libp2p peerstore package.",
+      "reference": "remove todo on line 28?"
+    },
+    {
+      "prediction": "Why do you need to add `mstrain_3x_coco.py` and `retinanet_r50_fpn.py`?",
+      "reference": "clean unnecessary arguments."
+    },
+    {
+      "prediction": "Shouldn't this be 64 instead of 65?",
+      "reference": "Could the name of the status be confusing to users? It says the execution stopped but is it a temporary stop? Will it continue to run at some point? To me the name does not suggest a final status but a transition one. WDYT @sshardool @jakhani @djaiswal83 @aditya1105"
+    },
+    {
+      "prediction": "I don't think this is necessary. You can just import `DockerProvisioner` in `molecule.utilities.provisioners`.",
+      "reference": "Since we created a Provisioners directory, can we do the same with the tests, and follow the same structure as the code?"
+    },
+    {
+      "prediction": "I think we can remove `mapSizeStr` now, since we don't use it anywhere.",
+      "reference": "We don't need that one anymore?"
+    },
+    {
+      "prediction": "I don't think we need to import `STORE` here, since it's imported at the top of the file.",
+      "reference": "Since we're determining the local node's host info in the `Store()` init, we need to import the `STORE` object here. Is this OK, or should I be determining the host info earlier. (When parsing settings?)"
+    },
+    {
+      "prediction": "I think this should be \"type using #getKind() as a cue.\"",
+      "reference": "I think, there is no `#getKind()` method anymore, is it? The approach would now be testing with instanceof. Or could we return a more specific type than JavaNode now?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be using the `done` button instead of the `done` link. The `done` link is meant to be used in the context of the `done` button, and the `done` button is meant to be used in the context of a button on the page.",
+      "reference": "can we localise these? not sure if we have easy access to a translator instance from `generic-provider-views`"
+    },
+    {
+      "prediction": "This should be `const uint16 = 0x239A` and `const uint16 = 0x8029`",
+      "reference": "This is the VID from Adafruit. Doesn't Particle have a different VID? (CC @suda)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `resetScope` function, since it's only called when `delim` is true.",
+      "reference": "The function is only used once, and this body is now only 1 call with the same args (e.g. basically an alias for `vStringTruncate`), it could probably be simply removed. But if you wanna keep the naming for clarity, it also sounds reasonable."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I'm not sure if it's the right thing to do in general. I'm not sure if it's the right thing to do in this case. I'm not sure if it's the right thing to do in general. I'm not sure if it's the right thing to do in this case. I'm not sure if it's the right thing to do in general. I'm not sure if it's the right thing to do in this case. I'm not sure if it's the right thing to do in general. I'm not sure if it's the right thing to do in this case. I'm not sure if it's the right thing to do in general. I'm not sure if it's the right thing to do in this case. I'm not sure if it's the right thing to do in general. I'm not sure if it's the right thing to do in this case. I'm not sure if it's the right thing to do in general. I'm not sure if it's the right thing to do in this case. I'm not sure if it's the right thing to do in general. I'm not sure if it's the right thing to",
+      "reference": "please change to `could not connect to peer: %s`. it is not relevant to say \"from kademlia\" in a warning"
+    },
+    {
+      "prediction": "I think this should be `time::TimeUtils::rmDateTimeMs(date.toString());`",
+      "reference": "Why need to remove this? Datetime has microseconds in fact."
+    },
+    {
+      "prediction": "I don't think we need to check for `nil` here, since `ecdsa.PrivateKey` is a pointer.",
+      "reference": "What's up with this, isn't it nil? Why not use the real T value from the test function? @elcore @wmark"
+    },
+    {
+      "prediction": "I think this should be `@param bool $exact = false`",
+      "reference": "this feels weird"
+    },
+    {
+      "prediction": "I think this should be `PREMERGE` instead of `MERGE`",
+      "reference": "I don't think this is the type of network we should be putting in our named networks."
+    },
+    {
+      "prediction": "I think this should be `debug.open(std::string(\"debug.\") + std::to_string(m_comm->get_rank_in_world()));`",
+      "reference": "debug? I suspect this will be removed before merge?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it would be better to move it to `test/unit/helpers/mongodb/test.js` or `test/unit/helpers/postgresql/test.js`",
+      "reference": "should this be here twice?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the same place as the other tests, but I don't know if that's the right place to do it.",
+      "reference": "`generate_special` - a meaningless name for me. Can we provide more detailed naming?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "This isn't something you introduced in your changes, but the change makes more obvious to me that this test would be better as a unit test of some kind (probably a controller test). Testing a 404 is an edge case that probably doesn't need to be tested with all components in integration."
+    },
+    {
+      "prediction": "I think this should be `logger.info(\"Using delegate supplier: \" + clazz.getName());`",
+      "reference": "We chose `info` in the test code for obvious reasons. Changing to `finest` makes debugging harder and noisier."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The reason why we need to do this is so that we can differentiate between a local host and a remote host. If the remote host is a local host, then the id should be something like `http://localhost:8080`. If it is a remote host, then the id should be something like `http://127.0.0.1:8080`.",
+      "reference": "any reason you're assigning to the local variable 'id' too?"
+    },
+    {
+      "prediction": "I think this should be `molecule.validators` instead of `validators`",
+      "reference": "Need to fix this import :)"
+    },
+    {
+      "prediction": "I think this should be `GetActor` instead of `GetAccountActor`",
+      "reference": "sorry these tests suck so much we are working to kill them"
+    },
+    {
+      "prediction": "I'm not sure this is correct. The `updateTimeDisplay` function takes a number of milliseconds, not ticks. I think you meant to multiply by `1000` instead of `10000`.",
+      "reference": "Why not adjust `currentTime` to be in ms directly?"
+    },
+    {
+      "prediction": "I don't think this is correct. The instruction could have a different offset than the one we're looking for.",
+      "reference": "> restore, we find the matching spill for that restore which uses the same slot nit: Two separate sentences: separate with `.` or something besides `,`."
+    },
+    {
+      "prediction": "I think this should be `UseBadgerDB: true`",
+      "reference": "this flag not used"
+    },
+    {
+      "prediction": "I think this should be `date_default_timezone_get() . ' GMT'` instead of `Y-m-d H:i`",
+      "reference": "I don't think `?? null` works here... that's to ensure PHP doesn't throw an \"undefined\" error about `$profile->expirationDate`, but now the code is assuming that `$profile->expirationDate` will be set. Might be cleaner to do: <pre> $expiration = isset($profile->expirationDate) ? $this->dateConverter->convertToDisplayDate(\"Y-m-d H:i\", $profile->expirationDate) : null; </pre> And then assigning the `$expiration` value in the return array..."
+    },
+    {
+      "prediction": "I think we can remove the `--production` env var now, since we don't need it anymore.",
+      "reference": "An ENV should be used? Are you sure that the Vue examples are built in the production mode?"
+    },
+    {
+      "prediction": "I think this should be `Union[float, int]` instead of `Union[float, int]`",
+      "reference": "Nit: this should actually always be float since we convert it before we construct the namedtuple."
+    },
+    {
+      "prediction": "This should be `LOG(WARNING)` instead of `LOG(FATAL)`.",
+      "reference": "The result of `FileUtils::fileType` could be saved, instead of calling twice."
+    },
+    {
+      "prediction": "Shouldn't this be a `try catch`?",
+      "reference": "Can you log the exception ?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "since this debug message is inside a plugin code, it should use flb_plg_debug(ctx->ins, \"...\"), on this case don't need the component prefix since the API will put it there automatically"
+    },
+    {
+      "prediction": "I don't think we should be changing this import statement.",
+      "reference": "I also took advantage of the changes to change the Timestamps used in this notification from \"java.sql.Timestamp\" to \"com.yahoo.rdl.Timestamp\"."
+    },
+    {
+      "prediction": "Why did you change this?",
+      "reference": "It would be nice all these tweaks to be configurable from cmd line, with sensible defaults"
+    },
+    {
+      "prediction": "I think this should be `600.000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "Why do we need this `check_same_thead` to be false? I feel a bit uncomfortable about this."
+    },
+    {
+      "prediction": "I think this should be `object = realm.create(CustomObject, {intCol: 2});` and not `object = realm.create(CustomObject, {intCol: 2}); object = object.intCol = 200;`",
+      "reference": "We should probably test with constructors which aren't in the schema, and functions which aren't constructors."
+    },
+    {
+      "prediction": "I don't think this is the right place for this function. I think it should be in the `CudaHostPinnedSpace` class.",
+      "reference": "this call doesn't work doesn't? I mean there doesn't seem to be an 2 argument allocate overload. Maybe arg_logical_size should just be defaulted to the arg_alloc_size thing. Or we should just report out physical allocation size instead of logical."
+    },
+    {
+      "prediction": "I think this should be `/var/lib/puppetlabs/puppet` and not `/opt/puppet/bin`",
+      "reference": "This winds up flipping the desired values. Looks like I gave you a bum steer @kevpl"
+    },
+    {
+      "prediction": "I think this is the only place where we need to know the iotex address.",
+      "reference": "nit: let's call our address ioAddr from now on"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this check. I think it should be done in the `gpu_validation_reserve_binding_slot` check above.",
+      "reference": "You're killing this \"else\" case here which currently flags an error when maxBoundDescriptorSets == 0."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. It seems like it should be in `pkginfo.c` or `pkginfo.h`, but I don't know for sure.",
+      "reference": "Nitpick: I'd append `_cb` to the function name here to mark it as a callback. Otherwise it looks a bit like this will do the entire job of removing null checksum refs from a hash table."
+    },
+    {
+      "prediction": "I think we can remove `isView` from `isNotView`",
+      "reference": "or you can use `filter(x => !x.isView)`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `ReconcileBootstrapData` function.",
+      "reference": "Instead of NewBuffer on the line above, do NewReader to avoid having to wrap later."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it. I think it would be better to have a separate function for building the website, and then call that function from here. That way, you don't have to do `pylint: enable=too-many-locals,too-many-statements`, and you don't have to do `# pylint: disable=too-many-locals,too-many-statements`.",
+      "reference": "No need for this, as pylint already only turns things off for this function and it's needed for the entire function."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `transactionGroup` instead of here.",
+      "reference": "extra brownie points ( overall solution ): we've currently triple-encoding the payset - 1. we encode it to calculate the block size. 2. we encode it to calculate the commit hash ( either via flat, or as Merkle tree ). 3. we encode it as a whole for the purpose of preparing the proposal ( I know that this isn't always the case, but this case happens to be on the critical path ). Caching the encoded data of the *first* block could help us repeating the process."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "This assertion should use the actual expanded string, so that it is clear what the message produced looks like. Right now one has to run the code to determine what the message is."
+    },
+    {
+      "prediction": "nit: unnecessary whitespace change",
+      "reference": "similar here for imports, these should be consolidated (and also consolidated with the grouping below)"
+    },
+    {
+      "prediction": "`self.file_client = mmcv.FileClient(**self.file_client_args)` -> `self.file_client = FileClient(**self.file_client_args)`",
+      "reference": "Maybe there is no need to modify it. Because if you don't use the mask, it won't be initialized."
+    },
+    {
+      "prediction": "I think this should be `input = new AutoValue_SelectTraceIdsFromSpan_Input(annotationKey, timestampRange.startUUID, timestampRange.endUUID, limit);`",
+      "reference": "Is there a reason we wouldn't want the trailing `%`? I'm guessing that without the trailing `%` it will just do a strict match vs a partial prefix right?"
+    },
+    {
+      "prediction": "I think this should be `#include \"safe_string/safe_string.h\"` instead of `#include <safe_string/safe_string.h>`.",
+      "reference": "Please update the DMA test app in AFU repo once this gets approved."
+    },
+    {
+      "prediction": "I think this should be `Shopware Framework - the best solution for your shopware.`",
+      "reference": "There should be a migration for that as well to reflect the change on in-production instances"
+    },
+    {
+      "prediction": "I think we can remove this line now.",
+      "reference": "Why does it have to be `\"\"` instead of an empty string"
+    },
+    {
+      "prediction": "I'm not sure if this should be `unless pending` or `unless RSpec.configuration.dry_run`?",
+      "reference": "This is very strongly opinionated so feel free to disagree with me better, but as conditionals get more complex I like turn them into ifs instead of unlesses. What do you think?"
+    },
+    {
+      "prediction": "I think this should be `sort_by { |l| list.map { |x| rand(x) }.sort }` instead of `sort_by { |l| list.sort_by { |x| rand(x) } }`",
+      "reference": "Will this obey the seed set on Kernel?"
+    },
+    {
+      "prediction": "I think this is the only place where `Puppet::Functions.create_function` is used, so I'm not sure we need to deprecate it.",
+      "reference": "We shouldn't use the `&.` syntax here, since we expect that `executor` will never be `nil`. For the functions that _can_ be called from apply / without an executor, `&.` is still appropriate."
+    },
+    {
+      "prediction": "I think this should be `throw new RuntimeException(\"SurfaceNamer::getSourceFilePath(\" + path + \", \" + className + \") not implemented\");`",
+      "reference": "Why the switch from period to colon?"
+    },
+    {
+      "prediction": "I think this should be `ECS_NUM_IMAGES_DELETE_PER_CYCLE_IN_SECONDS`",
+      "reference": "Can you just fix the warning instead? It's actually important for this to have a default of `\"\"` as the subsequent merges with `DefaultConfig()` and `fileConfig()` need to work. If you make this not `\"\"`, you break the assumptions of `Merge()`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it seems like it should be in the same namespace as the other config items.",
+      "reference": "let us not add this"
+    },
+    {
+      "prediction": "Should this be `nil` if the cgroupfs is mounted read-only?",
+      "reference": "Is there a need for group as well? crun sets both."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's the only way I could come up with that works.",
+      "reference": "@cijothomas This build-up pattern was really confusing. It looks like internally ActivityProcessor is intended to be chained but there is nothing in the abstract class that enforces it. We never set a \"Next\" or make sure that the chain is called. Probably need to do some more work in this area?"
+    },
+    {
+      "prediction": "I don't think we need to log the filter here.",
+      "reference": "`this.getFilter()` won't work on `pre('save')`. Perhaps make this `toySchema.pre(['updateOne', 'findOneAndUpdate'])`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to get the timestamp from the `Span` object itself.",
+      "reference": "Break out tests for events into their own tests rather than placing them in other ones. That makes it easier for us to figure out where problems lie and to do a TDD-driven implementation over new APIs."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be.",
+      "reference": "I think this is the only place where we use `UvShutdownReq`. Can we remove the type altogether? Or do you prefer to keep it around?"
+    },
+    {
+      "prediction": "I think this should be `InsecureSkipVerify: true` and not `InsecureSkipVerify: true,`",
+      "reference": "I think we can get this to work by passing in the root ca cert fixture that upstream ca is using"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `mktime` doesn't take the current time as an argument. It takes a `struct tm` as an argument. So you can't use `mktime` with a `tm` that has a different year than the current year.",
+      "reference": "Just occurred to me to ask: is the condor log date in localtime or UTC?"
+    },
+    {
+      "prediction": "This should be `GAUNTLET_ENTERED` instead of `SULLIUSCEP_LOCATION`.",
+      "reference": "Can you remove the sculliscep varbits from here too?"
+    },
+    {
+      "prediction": "I think this should be `do_query(\"alter table host add index host_cpid (host_cpid)\", $retval);`",
+      "reference": "I understand that it is a common practice in this script to add indexes this way but why don't we check index existence before adding it? I'm not very familiar with MySql but it's can be done easily in MSSQL and I'm pretty sure that there is a way to do the same here."
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "we should probably deprecate (and display a warning) the misspelled keyword arg here rather than removing it... and then add the new one. This changes a public API and will break any code that is currently using the misspelled version."
+    },
+    {
+      "prediction": "I think this should be `h2o_fatal` instead of `h2o_fatal`.",
+      "reference": "Should we do something like `h2o__fatal(file, line, \"fatal error: %s, %s\\n\", mess, RSTRING_PTR(...))` here?"
+    },
+    {
+      "prediction": "I think this test is flaky. I'm not sure why it's flaky, but I think it's due to the fact that we're using `request.method` and `request.url` twice in this test. I think we should be able to get away with only using `request.url` and `request.method` in this test.",
+      "reference": "is this one all platforms?"
+    },
+    {
+      "prediction": "I think this test is failing because it doesn't have the `message` key in the JSON response.",
+      "reference": "why capitals grr?"
+    },
+    {
+      "prediction": "I think this should be moved to `nameserver.go`",
+      "reference": "Surely all the above should go into `common`."
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "This is an unrelated change, but was failing CI... probably introduced in master."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to do something like: `const bool useQueryQueryMatches = false;` `if (SubstructMatch(refMol, prbMol, match, recursionPossible, useQueryQueryMatches)) {`",
+      "reference": "This piece isn't backwards compatible, but it's enough of an edge case that I think it's unlikely to be a problem."
+    },
+    {
+      "prediction": "I think this should be `10 * 1024 * 1024` instead of `10 * 1000 * 1000`",
+      "reference": "MB is presumably 1024 * 1024."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `wpcom-components` since it's a component, not a plugin.",
+      "reference": "Didn't we extract this logic to a `react-features` helper?"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "But But, overlapping async writes will get corrupted"
+    },
+    {
+      "prediction": "I think this should be `github.com/filecoin-project/go-filecoin/config` instead of `github.com/filecoin-project/go-filecoin/config`",
+      "reference": "Not your immediate problem, but having everything depend on a package that has the config for everything else is ick. Can you move the MessagePoolConfig here and reverse the dependency?"
+    },
+    {
+      "prediction": "I'm not sure about this change. The enter key will send ACTION_DOWN and ACTION_UP both events out.",
+      "reference": "recently, IME_ACTION_GO action can trigger key_down and key_up event successfully, so we don't need to send enter key separately as before, otherwise will trigger it twice and cause to enter empty passcode, which cause test failed."
+    },
+    {
+      "prediction": "I'm not sure why this is needed. The `checkExprDepth` check is done in `ExpressionUtils::RewriteLabelAttr2EdgeProp`, which is called by `rewriteLabelAttr2EdgeProp` in `ExpressionUtils`, which is called by `Expression::Rewrite`, which is called by `Expression::RewriteLabelAttr`, which is called by `Expression::RewriteEdgeProp`, which is called by `Expression::RewriteExpr`, which is called by `Expression::RewriteExpr2`, which is called by `Expression::RewriteExpr3`, which is called by `Expression::RewriteExpr4`, which is called by `Expression::RewriteExpr5`, which is called by `Expression::RewriteExpr6`, which is called by `Expression::RewriteExpr7`, which is called by `Expression::RewriteExpr8`, which is called by `Expression::RewriteExpr9`, which is called by `Expression::RewriteExpr10`, which is called by `Expression::RewriteExpr11`, which is called by `Expression::RewriteExpr12`, which is called by `Expression::RewriteExpr13`, which is called by `Expression::RewriteExpr14`, which is",
+      "reference": "It seems that you only need to do this `checkExprDepth()` inside `deduceExprType()`. So you don't have to add this check everywhere."
+    },
+    {
+      "prediction": "I think we can remove this line as it is not used.",
+      "reference": "why split into two lines?"
+    },
+    {
+      "prediction": "I think this should be `private static final int PARTITION_DATA_ID_START = 1000;`",
+      "reference": "Does this need to be public or can it be package-private?"
+    },
+    {
+      "prediction": "I think this should be `this.props.getPlugin(this.props.getPlugin(this.props.id).\u0441anEditFile(file))` and not `this.props.getPlugin(target.id).canEditFile(file)`",
+      "reference": "Questionable way of looping through editors and calling `canEditFile` to show the edit button. Is there a better way?"
+    },
+    {
+      "prediction": "`Tuple::of` is a no-op, so this can be simplified to `Tuple::of`.",
+      "reference": "`that is null` doesn't sound very useful to me. Could we rename `that` to `target` or something less context dependent :)?"
+    },
+    {
+      "prediction": "I think it would be better to have a `self.molecule._print_valid_platforms()` and a `self.molecule._print_valid_providers()` method.",
+      "reference": "Just to be consistent, can we use the print function `print()` instead of the keyword. Same goes for line 437."
+    },
+    {
+      "prediction": "I think we should use `wallet[0].String()` instead of `wallet[0]`",
+      "reference": "Having to pass in `price` and `limit` is pretty common. Do we want to have this be another argument, maybe a combined structure that can be used for every action that requires it?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it's better to do it in the `stop` method.",
+      "reference": "Could you move it after `stop`?"
+    },
+    {
+      "prediction": "I think it would be better to have a `getBodyParamName` method in the `Operation` class and use that instead of hard-coding it here.",
+      "reference": "The variable `method` seems not used. Maybe we can remove the parameter `bodyParamName` and generate it by invoking `ParamUtils.generateBodyParameterName(method)` ?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "I don't think this should be here?"
+    },
+    {
+      "prediction": "I think we can remove the `Winning` part of the name here.",
+      "reference": "Either I'm missing something or specs actors should remove this call cc @anorth"
+    },
+    {
+      "prediction": "I think this should be `Print(counter, counterWidth, resultWidth);` instead of `Print(defaultCounter, counterWidth, resultWidth);`",
+      "reference": "Please undo all these changes as they are white-space only."
+    },
+    {
+      "prediction": "I think this should be `array_search($countryCodeAlpha2, $countriesAlpha2, true);` and not `countryCodeAlpha2[] = array_search($countryName, $countriesAlpha2, true);`",
+      "reference": "You forgot an optimization? `$countries = CountryField::FORMAT_ISO_3166_ALPHA3 === $countryCodeFormat ? Countries::getAlpha3Names() : $countriesAlpha2;`"
+    },
+    {
+      "prediction": "I think this should be `current Python interpreter version (e.g., '3.8')`",
+      "reference": "Shouldn't this also return the current value specified by the user? At least that's what the docstring says and what the previous version did."
+    },
+    {
+      "prediction": "I think it would be better to add a `time_range` parameter to the `get_listens` method, instead of adding a new parameter to the `get_listens` function.",
+      "reference": "> the time range the listen search the time range of the listen search?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. Preserving the ZMM caller saved state is a bit of a hack, and I don't think we want to do it for every single call to `read_evex()`. Instead, I think it would be better to do this in the `read_evex()` function itself.",
+      "reference": "This snprintf, etc. needs to all be inside the DO_ONCE: all this overhead happening on every single decode is likely a huge performance hit."
+    },
+    {
+      "prediction": "This should be `--connect-retries` or `--conn_retries` depending on whether you're talking about explicit routes or implicit routes.",
+      "reference": "would call it connect_retries"
+    },
+    {
+      "prediction": "I think this test should be moved to `sql/test/test_sql.py` instead of `sql/test/test_dataframe.py`",
+      "reference": "Why are we using `repr`?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "I use the same zms instance to check authentication in swagger endpoints. Same thing in ZTS."
+    },
+    {
+      "prediction": "I think it would be better to set this to `true` instead of `false` so that we don't have to rely on the binary file to change user in the test case.",
+      "reference": "Consider consolidating the common code in a common setup method in tests?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this test. It should be moved to `org.hyperledger.besu.ethereum.test` folder.",
+      "reference": "q: do you need to run the annotation processor over EthProtocolConfiguration prior to writing this file? (i.e.to ensure ImmutableEthProtcolConfiguration exists)? Does Auto-import etc. still work in the IDE?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be moved to `SchemaTypeFormatter`.",
+      "reference": "What is the motivation for switching from `Schema` to `DiscoveryField` everywhere?"
+    },
+    {
+      "prediction": "I think this should be `ECS_CONTAINER_METADATA_ENABLED`",
+      "reference": "What happens when ECS Init/whatever's starting the Agent mounts some other directory as Agent's data directory (`-v /tmp:/data`) sets `ECS_HOST_DATA_DIR` to `\"/var/lib/ecs\"` It doesn't seem like a strong enough abstraction to be dependent on Agent configuration options to expect `ECS_HOST_DATA_DIR` to be the same as whatever's being mounted as `/data` on the host. My main concern is the late-binding/asynchronous failures this could lead to if the wrong host mount was specified. Wondering if there are better alternatives here. Can we at the very least inspect the Agent container and check if `ECS_HOST_DATA_DIR` is mounted? If the answer is that we'll revisit the validation here at some latter stage, that's fine too. But, I want us to be aware of potential failures here."
+    },
+    {
+      "prediction": "I think it would be better to use `WITH_XCB` instead of `WITH_XCB_ERRORS` for consistency.",
+      "reference": "btw that's obviously out of scope of this PR but since i noticed this line here it reminded me a thing: on some systems i've noticed the font didn't had those characters, so it was just a square or empty space and sometimes they're just a bit shifted in position (if font doesn't have them and using from fallback font which have incompatible dimensions of characters) how do you think is it worth creating an issue to discuss/address it or it could just waste time on bikeshedding?"
+    },
+    {
+      "prediction": "I think it would be better to move this to `pkg/recover` and rename it to `recovery.go`",
+      "reference": "alias is not needed"
+    },
+    {
+      "prediction": "I think this should be `pid` instead of `pid_file`.",
+      "reference": "Do we need to include the path?"
+    },
+    {
+      "prediction": "I'm not sure we need to change this, but I'd like to see a test for this.",
+      "reference": "During our zoom and working on this, I said that I believe it was set to 64MB, not 32MB, but the diff shows that I was wrong. So we could change it back to 32MB."
+    },
+    {
+      "prediction": "I think this should be `log.Logger` instead of `log` as `log.Logger` is a package while `log` is not.",
+      "reference": "Here I thought you always insisted on ` log \"github.com/Sirupsen/logrus\" ` :) I might do a PR that just globally switches that out wherever we have log."
+    },
+    {
+      "prediction": "I think this should be `bool` instead of `bool, error`",
+      "reference": "We maintain that each (logical) key generation has the same set of device keys, so no need to plumb through `keyGen`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I've been able to get it to work.",
+      "reference": "Please use `from unittest import mock` instead of this package."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `cursor` pipeline, since that's where the cursor value is coming from.",
+      "reference": "Should be `if (this.options && this.options.cursor) {`. Options may be undefined. Also, going forward, mongoose will always use curly braces around if blocks."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "This should remain as mailrelay.blackducksoftware.com for staging. Not sure this will be valid, but it shouldn't go through the production mail server."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in `MemoryPoolBlock2.cs`.",
+      "reference": "Maybe we should have a Debug.Assert for when `Slab != null` to ensure that we are always returning the block (in our tests at least)."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to check if `props.value` is defined before deleting it.",
+      "reference": "Okay sorry about this I was really confused on how to support uncontrolled in this case, so what we want to do is apply `defaultValue` when `props.value` is falsy but not undefined, right? In the case of us seeing it is controlled (value & onChange present) we apply defaultValue to `props.value` (and delete props.defaultValue) In the case of uncontrolled we shouldn't enter this branch at all and go for full native dom behavior"
+    },
+    {
+      "prediction": "`map` is not used in this file.",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure if we want to make this change. I think it's fine for now, but I'm not sure",
+      "reference": "The latest yaml in openebs/master has changed to name=admission-webhook. cc: @prateekpandey14"
+    },
+    {
+      "prediction": "I think this should be `Amazon-Simple Notification Service` instead of `Amazon Simple Notification Service`",
+      "reference": "using **an** Amazon Simple... Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "I think this should be `||` instead of `1`.",
+      "reference": "Probably don't intend for this `1 ||` to still be here"
+    },
+    {
+      "prediction": "I think this should be `skip |= val_obj->LogError(val_obj->device, \"VUID-VkDescriptorSetLayoutCreateInfo-flags-04590\", \"pCreateInfo->flags cannot contain both \" \"VK_DESCRIPTOR_SET_LAYOUT_CREATE_PUSH_DESCRIPTOR_BIT_KHR and \" \"VK_DESCRIPTOR_SET_LAYOUT_CREATE_HOST_ONLY_POOL_BIT_VALVE.\");`",
+      "reference": "Looks like this is failing to compile on Windows (VS 2015 I think)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `BeforeQuery` function of the `Query` struct.",
+      "reference": "nit: call the arg something simple like `f` to avoid repeating \"beforeQuery\"."
+    },
+    {
+      "prediction": "I think we should move this to the `auth.go` file.",
+      "reference": "do you think this should be `osd/auth` as a pkg to make it more unique and void conflicts with other `auth` packages?"
+    },
+    {
+      "prediction": "Should this be `Message = string.Empty`?",
+      "reference": "Why are you passing out the Message?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `RegisterSpanProcessor` function.",
+      "reference": "Yikes, this was a bug :grimacing:"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we need to make sure that the subscriber is not closed before we call `subscriber.Finish()`.",
+      "reference": "I am not sure which one of these is considered idiomatic go... maybe @azdagron has an opinion here? /me prefers the verb since it's shorter. I guess I don't care _too_ much, but it would be good to get some general agreement so we are all following the same conventions"
+    },
+    {
+      "prediction": "I think it would be better to have this in a separate `if` statement so that we don't need to check for `scaffold` every time we run the command.",
+      "reference": "Please check this for error and emit appropriate debugging information. I know this was pre-existing code, but we just have to fix things like this as we go."
+    },
+    {
+      "prediction": "I think this should be `cond == nil || cond.Status == corev1.ConditionUnknown`",
+      "reference": "This seems unrelated, should be a separate PR."
+    },
+    {
+      "prediction": "This should not be included in this pull request. It is not related to the content of this pull request.",
+      "reference": "Never use relative includes, prefer full path"
+    },
+    {
+      "prediction": "I think this is a good test, but I don't think it covers all cases. For example, `@register` with `count` and `zero_count` both set to `True` works. I think it would be good to have a test where `count` is `None` and `zero_count` is `True` and vice-versa.",
+      "reference": "This should fit on the line above without going over the 79-char limit, no?"
+    },
+    {
+      "prediction": "I'm not sure why this is here.",
+      "reference": "Is this related to the other contents of this pull request?"
+    },
+    {
+      "prediction": "I think this should be `string` instead of `[]string`",
+      "reference": "Should this be `*string`?"
+    },
+    {
+      "prediction": "I think we can remove this line now.",
+      "reference": "In the old code here there was no Source Selector, but CreateDefaultProfile will specify a Source Selector of `\"\"`. Is that equivalent?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "We should pick this across for all our deprecation specs"
+    },
+    {
+      "prediction": "I think this should be `kubectlImageTar` for consistency with the rest of the CLI.",
+      "reference": "Longer term, I think it could be valuable to move this to cluster-api, and then other repos could take advantage of this as well."
+    },
+    {
+      "prediction": "I think this should be `externalClient.EnsureBinary(\"test-connection\", \"test-connection\")`",
+      "reference": "Is this a related change?"
+    },
+    {
+      "prediction": "I think this should be done in the reconciler, not in the actuator.",
+      "reference": "Given that this is repeated, can we do it outside here and the Create call, what do you think about moving it outside?"
+    },
+    {
+      "prediction": "`instance.propToCol(changes[0][1])` should be `instance.prop(changes[0][1])`",
+      "reference": "Changing `prop`=>`col` by reference?   I think you can prepare map `prop`=>`col` in L51."
+    },
+    {
+      "prediction": "I think this should be `--bfbs-builtin-attributes` or `--bfbs-builtin-attrs` to be consistent with the rest of the CLI.",
+      "reference": "I hate to ask, but can we make this flag shorter so it lines up? :("
+    },
+    {
+      "prediction": "I'm not sure if we want to remove the `apiv1` package from the `pkg/apis` directory. I think it's fine to keep it, but I'm not sure if we want to remove the `extensions` package from the `pkg/apis` directory. I think it's fine to keep it, but I'm not sure if we want to remove the `extensions` package from the `pkg/apis` directory. I think it's fine to keep it, but I'm not sure if we want to remove the `extensions` package from the `pkg/apis` directory. I think it's fine to keep it, but I'm not sure if we want to remove the `extensions` package from the `pkg/apis` directory. I think it's fine to keep it, but I'm not sure if we want to remove the `extensions` package from the `pkg/apis` directory. I think it's fine to keep it, but I'm not sure if we want to remove the `extensions` package from the `pkg/apis` directory. I think it's fine to keep it, but I'm not sure if we want to remove the `extensions` package from the `pkg/apis` directory. I think it's fine to keep it",
+      "reference": "Note to reviewers: latent, dedup import"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "Seems this func is no more needed after your refactoring."
+    },
+    {
+      "prediction": "I think this should be `io\\\\.usethesource` instead of `io\\\\.usethesource`",
+      "reference": "already added (see several rows below)"
+    },
+    {
+      "prediction": "I think this should be `ifaceName == \"tunl0\" || ifaceName == \"wireguard.cali\"` instead of `ifaceName == \"tunl0\" || ifaceName == \"wireguard.cali\"`",
+      "reference": "nit: a switch perhaps?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that we don't add a retry status to a workflow that is already in that state.",
+      "reference": "Great you caught this. Super critical to not retry these errors."
+    },
+    {
+      "prediction": "I think this should be `instancePath = coreContainer.getCoreRootDirectory().resolve(coreName);`",
+      "reference": "Just a little change to make the var effectively final, which is clearer."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to format the address. I think it would be better to use `human.format_address` instead.",
+      "reference": "Using `\"{}\".format(...)` is a bit beside the point..."
+    },
+    {
+      "prediction": "I think it would be better to use `json.dumps` instead of `json.load` and `json.dump` to be consistent with the rest of the tests.",
+      "reference": "I think these should remain as they are."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Let's just call this PERF_PACKAGES and have them in the same format as WINDOWS/SLES/DEBIAN_PACKAGES constants - since it is only a single package and, as written, you'd have to update the code to add more package names anyway. Might as well be consistent with the rest of the code."
+    },
+    {
+      "prediction": "Should this be `validTaskIndex(TaskIndex)` instead of `const deviceIndex_t DeviceIndex = getDeviceIndex_from_TaskIndex(TaskIndex);`?",
+      "reference": "We don't have a `getPluginIDfromTaskIndex` function? I think it should be implemented in Globals/Plugins.h / .cpp"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to check if the field is present in the schema before loading it. If it is not present, then we don't need to load it. If it is present, then we need to load it.",
+      "reference": "@shalinmangar If lazy field loading isn't enabled, I don't think this actually changes the behavior of `SolrIndexSearcher`, since it was previously ignoring the fields list anyway. What it _should_ do is allow certain distributed queries, like the ones in `DistribJoinFromCollectionTest`, to co-exist with `SolrIndexSearcher#doc()` respecting the `fields` set instead of just discarding it."
+    },
+    {
+      "prediction": "I think this should be `{{ .plugin }}-${DDEV_SITENAME}-db` instead of `{{ .plugin }}-{{.name }}-db`",
+      "reference": "Should we be using version 3 now?"
+    },
+    {
+      "prediction": "I think it would be better to move this to the top of the file, so that it's clear that this is the only place it's used.",
+      "reference": "Would be better to throw a more specific exception, but for now it will do."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "We should not replace Wei with a type that is shared across multiple units. We need this as `Wei` for type and unit safety."
+    },
+    {
+      "prediction": "I think this should be `if err != nil && !os.IsNotExist(err)`",
+      "reference": "I'm pretty sure this should emit a util.Warning*(), since we're skipping by here and never hitting anything that will give them a warning about what's happened."
+    },
+    {
+      "prediction": "I think this should be `return GetBuiltins(state)`",
+      "reference": "This function isn't useful? it's just a clone of GetParserWithBuiltins?"
+    },
+    {
+      "prediction": "I think it would be better to move this to the `circuitBreaker` section.",
+      "reference": "I'm not sure if the boolean flags should always contain `is`, also I generally hate too long names... ;) we already know this is a section for circuit breakers, so the name doesn't have to repeat all of it. How about `cpuBreakerEnabled`, `memoryBreakerEnabled` etc?"
+    },
+    {
+      "prediction": "I don't think we should add a new method to the rewriteFiles interface. Instead, I think we should add a new method to the SnapshotUpdate interface and have it return a new instance of the RewriteFiles type. This would allow us to keep the interface agnostic of the underlying implementation.",
+      "reference": "Before we start the replacing equality deletes with position deletes, I think we need to refactor the RewriteFiles API to adjust more cases: 1. Rewrite data files and remove all the delete rows. The files to delete will be a set of data files and a set of delete files, and the files to add will be a set of data files. 2. Replace equality deletes with position deletes, the files to delete will be a set of equality delete files (we will need to ensure that all delete files are equality delete files ? ) , the files to add will be a set of position delete files. 3. Merging small delete files into a bigger delete files. The files to delete will be a set of equality/position delete files, the files to add will be a set of equality/position delete files."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "Layout/TrailingWhitespace: Trailing whitespace detected."
+    },
+    {
+      "prediction": "I think it would be better to use `entry.getValue().get(0)` instead of `entry.getValue().size() > 0 && entry.getValue().get(0) != null`",
+      "reference": "Is it possible that the shortcuts list is empty?"
+    },
+    {
+      "prediction": "I think this should be `errors.Cause(err) == context.Canceled || errors.Cause(err) == context.DeadlineExceeded || errors.Cause(err) == context.Unknown`",
+      "reference": "This just changes the error message from `unable to save snapshot: [...] context canceled` to `canceled`."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that the `Content-Length` header is not part of the response headers, it is part of the request headers. So if the request has a `Content-Length` header, then the response must also have a `Content-Length` header. This means that the `Content-Length` header must be added to the response before it is sent to the client.",
+      "reference": "maybe it's better to remove HttpHeaders.CONTENT_LENGTH after the loop"
+    },
+    {
+      "prediction": "I think this should be `if (n < 0 || n > length())`",
+      "reference": "Very cool. We are now collecting the fruits after your length implementation!"
+    },
+    {
+      "prediction": "I think this should be `xpn-master` instead of `xpn-network`",
+      "reference": "Is this the same copy of the rules file from above? If so, can just use the copy from above, and this can be dropped."
+    },
+    {
+      "prediction": "I think we can remove the `enigid` field from the `eniAttachHandler` struct.",
+      "reference": "Please break this into multiple lines."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add the header tags. I think it should be done in the `AddHeaderTagsFromHttpResponse` method, since it's the responsibility of the `AddHeaderTagsFromHttpResponse` method to add the header tags.",
+      "reference": "Shouldn't this new line replace the few lines above?"
+    },
+    {
+      "prediction": "I think this should be `universe.SystemTime` instead of `transformations.SystemTime`",
+      "reference": "Perhaps this belongs in universe?"
+    },
+    {
+      "prediction": "I think this should be `EPS` instead of `1e-12`",
+      "reference": "there is a `kEpsilon` you can use directly."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "This should be grouped with the below imports, and the two separate groups of imports below should likely also be grouped together"
+    },
+    {
+      "prediction": "I think this should be `std::string *error_code, std::string *error_msg`",
+      "reference": "It looks like this error_code field is used to return a string to the caller, not just a const input string to be consumed as in the rest of this PR? If so, my personal preference would be to leave it as a std::string* instead of a reference; I don't know offhand what the VL coding style dictates."
+    },
+    {
+      "prediction": "This should be `static const std::string MetaUtils::partPrefix = kPartsTable;`",
+      "reference": "If return by value, then why bother to make it `static`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `utils.py` since it's only used there.",
+      "reference": "You'll need to add docs for the argument, see e.g. the `debug_log_filter` docstring above."
+    },
+    {
+      "prediction": "I think we can remove this line now.",
+      "reference": "`sdktrace` or `tracesdk`, pick one."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "can we return null instead?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit()`, which is called by `dr_process_child_exit",
+      "reference": "You mean XXX and \"we should\" or \"NYI\" or sthg. Maybe put it on the ifdef line."
+    },
+    {
+      "prediction": "I think this should be called `SingleImageImportLoggable` and not `SingleImageImportFileLoggable`",
+      "reference": "minor: can be in-lined in return"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "why the additional \"\\n\"?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should just return `err` here, since we don't care about the result of `s.first.Do(f)` at this point.",
+      "reference": "This seems to be a big omission. If the first buffer reports an error, stream seems to just discard it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the block listener. I think it would be better to move it to the `node` package.",
+      "reference": "Maybe move this one down too?"
+    },
+    {
+      "prediction": "I think this is the only place we need to use this.",
+      "reference": "We do this in lots of places I'm starting to notice...typically you'd avoid taking concrete dependencies, but we also avoid package dependencies as well which makes sense (i.e. not using an IoC, etc.). Any reason we don't implement some form of poor-man's DI and/or a simple factory instead of new-ing things up everywhere. This may be a simple example (as I'm guessing this is likely the only place this is ever or ever will be used), but separating them doesn't take any more time/effort either really."
+    },
+    {
+      "prediction": "I don't think this is correct. The error message should not be included in the error itself, it should be a part of the error object.",
+      "reference": "What if formatPrefix doesn't include a space? Will the resulting error have e.Error() stuck at the end without any space?"
+    },
+    {
+      "prediction": "I think this should be `// Package pushsync provides the pushsync protocol implementation`",
+      "reference": "that doesn't say a lot to me. Can you please be more specific on the function of pushsync?"
+    },
+    {
+      "prediction": "I think this should be `subpass_desc->pColorBlendAttachment->layout == VK_IMAGE_LAYOUT_COLOR_ATTACHMENT_READ_ONLY_OPTIMAL` instead of `subpass_desc->pDepthStencilAttachment->layout == VK_IMAGE_LAYOUT_DEPTH_STENCIL_READ_ONLY_OPTIMAL`",
+      "reference": "> subpass uses a depth/stencil attachment in renderPass with a read-only layout for the depth aspect Assume - VK_IMAGE_LAYOUT_DEPTH_READ_ONLY_STENCIL_ATTACHMENT_OPTIMAL - VK_IMAGE_LAYOUT_DEPTH_READ_ONLY_OPTIMAL count as well seems there is a `NormalizeDepthImageLayout` for this"
+    },
+    {
+      "prediction": "I think this widget should be under `WIDGET_AREA_STYLES.BOXES` instead of `WIDGET_AREA_STYLES.COMPOSITE`.",
+      "reference": "This shouldn't be altered, you probably meant to update `AREA_PAGE_DASHBOARD_ALL_TRAFFIC` further below :)"
+    },
+    {
+      "prediction": "`dict(mount_volumes)` is a dict, not a dict-like object, so you can't call `.items()` on it. You'll have to do `list(mount_volumes)` and then call `.items()` on the result.",
+      "reference": "what does this change do exactly?"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this. It's not a background context, it's the context in which the gossip node runs.",
+      "reference": "What's the purpose of this `backgroundCtx` field?"
+    },
+    {
+      "prediction": "I think this should be `uint32(1024 * 1024)`",
+      "reference": "non contract incarnation"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to handle the case where `message.get('message', None)` returns `None` instead of an empty string.",
+      "reference": "I would recommend to leave out `None` because None is already the default."
+    },
+    {
+      "prediction": "I think this should be `log.Debugf(\"genesis cid: %s does not match: %s, disconnecting from peer: %s\", &hello.GenesisHash, h.genesis, from)`",
+      "reference": "Should the \"bad hello message\" case above also disconnect?"
+    },
+    {
+      "prediction": "I think this should be `public static boolean fragmentEnabled = true;` and `public static boolean fragmentVisible = true;`",
+      "reference": "Shouldn't this be in the child and not in the base plugin? Wouldn't having it here enable all derived plugins at once?"
+    },
+    {
+      "prediction": "I think this should be `Run examples whose full nested name matches REGEX (may be used more than once)`",
+      "reference": "Should probably say REGEX or PATTERN rather than string."
+    },
+    {
+      "prediction": "I'm not sure about this one. I think it's fine, but just wanted to make sure.",
+      "reference": "I felt we don't need to set it that strict, like if somehow pubsub happens to allow extra 20 bytes for metadata, this test won't work. We would suggest to use 11MB instead."
+    },
+    {
+      "prediction": "I think this should be `isset($metadata['associationType'])`",
+      "reference": "In theory `isset($metadata['associationType'])` is `false` if the linked field is not an association field, isn't?"
+    },
+    {
+      "prediction": "This should be `grant_flows: %w(authorization_code implicit password client_credentials)`",
+      "reference": "@tute the only thing I needed to enable this like as `password` is not in default flow anymore I think."
+    },
+    {
+      "prediction": "I'm not sure about this one. I think it would be better to have a test for `allUsers` and a test for `user:anything` and a test for `user:someone@somewhere.tld` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `user:anything@mozilla.com` and a test for `",
+      "reference": "nit: to be consistent with `self.members`, please move this up one line, so that it's closer to `allUsers`? You will need to update your test reference."
+    },
+    {
+      "prediction": "I think this should be `--log` instead of `--verbose`",
+      "reference": "we might as well just access `config.gClientVerbose` directly here. If any other options are passed to `gclientSync` or `gclientRunhooks` then the verbose option will be lost the way it's used now"
+    },
+    {
+      "prediction": "I think this should be renamed to `system_logs` or something like that.",
+      "reference": "Change the key to `default_tab` instead and I'll merge this"
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "we still can end up passing a blank org_id to the scope method if for any reason the params is not present..."
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. The problem is that `value` can be a `String` or an `Enum`, so we need to check the type of `value` before casting it to a string.",
+      "reference": "I think yes, this makes sense to expose this. +1"
+    },
+    {
+      "prediction": "I think this should be `{{env \"OPENEBS_IO_LOCALPV_HOSTPATH_DIR\" | default \"/var/openebs/local\"}}` and not `{{env \"OPENEBS_IO_LOCALPV_HOSTPATH_DIR\" | default \"/var/openebs/local\"}}`",
+      "reference": "let us put the sample yaml snippet i.e. maya api server deployment that makes use of this env & value"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it's fine to leave this as it is for now, but I'm not sure if it's something we want to do in the future.",
+      "reference": "What changed to cause this?"
+    },
+    {
+      "prediction": "I think this is the wrong place to put this.",
+      "reference": "Repeating big interface in many places shows smtg is wrong with design"
+    },
+    {
+      "prediction": "I'm not sure if we want to expose the `scope` here, but I think it's fine.",
+      "reference": "Inheriting from `AccountsController` to get access to the `redirect_if_disabled` filter."
+    },
+    {
+      "prediction": "I think you can remove the `# snippet-start` and `# snippet-end` tags here.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Enter at the end of the file! :-D (and at the end of lines ;-) )"
+    },
+    {
+      "prediction": "I think this should be `:scheduled_for_deactivation_on`",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that the `RemoveSuccessor` is called on `root`, but the `AddSuccessors` is called on `newRoot`. I think we need to call `AddSuccessors` on `newRoot` and remove `RemoveSuccessor` on `newRoot` before calling `AddSuccessors` on `root`.",
+      "reference": "Don't you want to place this check before you add `succ` to `newSucc`?"
+    },
+    {
+      "prediction": "These imports are no longer used.",
+      "reference": "Commit 676eeeab10884456a5f70ce708a8aa5116ecbaf7 contains a lot of unrelated code style changes. Please get rid of these."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we should be using `bokeh.io._get_screenshot_as_png` instead of `bokeh.io._get_screenshot_as_png(plot.state, None)`.",
+      "reference": "Well this sucks, I wish you'd mentioned you were using a private API, perhaps we could have made a public one with better guarantees before `0.12.6` was released."
+    },
+    {
+      "prediction": "I think we can remove the `capacity` field from the `peerlist.List` struct.",
+      "reference": "Consider instead moving this into an _test file so it's public but only usable in tests."
+    },
+    {
+      "prediction": "I think this should be `fmt.Printf` instead of `\tlogrus.Infof`",
+      "reference": "Any reason why we want to remove this code ? It helps debugging AWS ebs attach issues. Ideally yes a library should have minimal logging, but there is no way we can track long running AWS attach calls."
+    },
+    {
+      "prediction": "Why do you need the `tstzrange` function here?",
+      "reference": "nit: indentation doesn't match (unfortunately i don't know of a way to automatically lint these sql statements with how we're using them currently)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `before` block of the `setup_profiler` method, but I'm not sure if that's the right place to do it.",
+      "reference": "Should we remove the other `setup_profiler` call site and just let this be the one call site for it, since, AFAIK, the other one is no longer necessary with this? Also, can we make `setup_profiler` private?"
+    },
+    {
+      "prediction": "I think this should be `Hidden: true`, since it's a command that can be run from the CLI.",
+      "reference": "Are there any other templates where this command needs to be added, or were those issues specific to storage requiring us to enable a new command group?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "perhaps file can be removed altogether (trying to avoid an empty package for these build tags)"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `get_module_handle_64` can return `NULL` if it can't find a module with the given name.",
+      "reference": "NULL change, could you move that out too?"
+    },
+    {
+      "prediction": "I think this should be renamed to `FormatKV` since it's only used for KV.",
+      "reference": "A brief explanation about this func is required."
+    },
+    {
+      "prediction": "I think it would be better to have this as a method on the `MergedCell` class.",
+      "reference": "Could you add description for below static function?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to do this in a single call to `cs.ctx.Err()`.",
+      "reference": "Since this is only counting the failed attempts, can you call it failedAttemptsCount or retryCount instead of attemptsCount?"
+    },
+    {
+      "prediction": "I don't think we need to import `OOOOOOOOOOos` here.",
+      "reference": "I don't think this will work."
+    },
+    {
+      "prediction": "I don't think we need this interface at all. We can just implement `IHttpParser` in `KestrelHttpHandler` and call `HttpParser.ParseRequestLine` and `HttpParser.ParseHeaders` in `ParseRequestLine` and `ParseHeaders`.",
+      "reference": "Since http2 is coming do we want to call this something more specific?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it's better to keep the `navbar-search` tag around and just add `aria-label` and `aria-describedby` to the `<div>` tag inside the `<nav>` tag.",
+      "reference": "I don't think we want to revert the accessibility issues the tool caught, do we?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that didn't break anything.",
+      "reference": "I prefer naming this rateLimitedRemoteHost This can be local host but reads remote Host Is Non Local."
+    },
+    {
+      "prediction": "I think this should be `type._lastDomChild !== undefined && type._lastDomChildSibling !== undefined`",
+      "reference": "Hmm I think I might've messed up this condition here... Need to look more at it"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for `package_stats` being a dict, but it's the only way I could come up with.",
+      "reference": "This also makes the exception to be raised on non-dict falsy values which I think was expected."
+    },
+    {
+      "prediction": "I think this should be `_auto_indexable_1d = False`",
+      "reference": "Is this correct? The auto indexing behavior caused problems fixed in a recently merged PR..."
+    },
+    {
+      "prediction": "I think we should bump this to `1.10`",
+      "reference": "Hmm, didn't we just find out that 1.6 *does NOT* work? We should bump the minimum as well."
+    },
+    {
+      "prediction": "I think this should be `return` instead of `return 0;`",
+      "reference": "Will calling `abort()` cause problems for our test suite?"
+    },
+    {
+      "prediction": "I think this should be `break;` instead of `return AST_FATAL;`",
+      "reference": "for my own edificaton, what's this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I could come up with.",
+      "reference": "make it harmless to commit ws == nil"
+    },
+    {
+      "prediction": "I think this should be `\\$\\(abs_out_exe_location ([^\\)]+)\\)`",
+      "reference": "Hmmmm, I'm not sure I like these becoming a replacement any build rule has access to. It seems like if it were used in an actual build rule it would only be doing bad things. Can we just use `filepath.Abs` on the returned path in run_step.go?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `yaraServer/echo` since that's where the yara engine is.",
+      "reference": "That's actually not as bad as I thought it was going to be."
+    },
+    {
+      "prediction": "I think this should be `em.createQuery(\"SELECT object(o) FROM Dataset o, DvObject d WHERE d.id=o.id and d.publicationDate IS NOT NULL ORDER BY o.id ASC\", Dataset.class).getResultList();`",
+      "reference": "Hmm, I'm surprised if we don't already have a \"find all unpublished datasets\" method but I don't really know. Maybe @scolapasta knows."
+    },
+    {
+      "prediction": "I think this should be `old-style-super` instead of `C1601`.",
+      "reference": "The checks in this file are disabled by default since they are meant for Python 3 porting, and this one does not have to do with Python 3 porting at all."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this feature. I think it should be in the `HttpServerTracing` feature.",
+      "reference": "I would change \"should be sent\" to \"must be received\"."
+    },
+    {
+      "prediction": "Can we move this function to storagepool.go?",
+      "reference": "Please put it in following way : `// DeleteStoragePool receives StoragePoolClaim delete event and calls the required handlers to delete other pool related resources.`"
+    },
+    {
+      "prediction": "I think we can remove these constants, as they are not used anywhere.",
+      "reference": "It's better to use servicecomb now."
+    },
+    {
+      "prediction": "I think this should be `defaultRenewBeforeExpiryDuration time.Duration`",
+      "reference": "Make the flag derived default available in the readiness controller"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this.",
+      "reference": "Confirmed this test fails when the fix is reverted"
+    },
+    {
+      "prediction": "I think this should be an error from the collector, not the caller of the collector. The caller can decide if it wants to continue or not.",
+      "reference": "This method is still possible insecure as it may archive not only logs files. I suggest to filter file names to ensure that they start with `.mysterium-log*`."
+    },
+    {
+      "prediction": "Why do we need to zero out `conv_offset`?",
+      "reference": "DCN should not be able to use `init_cfg`."
+    },
+    {
+      "prediction": "I think this should be `throw new ArgumentException($\"{nameof(spanContext)} must not be null or empty.\");`",
+      "reference": "It is a bit weird to see a message like \"something must be valid\"."
+    },
+    {
+      "prediction": "I think we can remove the `LOG.info(\"Logging configured.\");` and `LOG.finest(\"Logging configured.\");` as they are redundant now.",
+      "reference": "Again. This is part of the start up of the server, and the diagnostics are helpful."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `spec/travis/build/travis.yml`.",
+      "reference": "I think you need a `,` at the end of the line?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `org.camunda` package, since it's not related to the `bpm` package.",
+      "reference": "Please avoid wildcard imports."
+    },
+    {
+      "prediction": "I think we should be able to get away with just 100 seconds. I'm not sure how long it takes to upload a 4MB file, but I'm pretty sure it's less than 100 seconds.",
+      "reference": "Did we consider making the timeout an argument to the build task? Would amke it easer to configure in the future: change a build property vs. rebuilding the DLL."
+    },
+    {
+      "prediction": "I think this should be `nodeConfig` instead of `nodeConfig.NodeIPv4Addr`",
+      "reference": "dumb question: the commit message talks about dual-stack support but it seems that in a dual-stack cluster, Egress will only support the IPv4 address family?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The `set_tlsext_host_name` is only called if `self.tls[server].set_tlsext_host_name(server.sni)` is `True`. If `self.tls[server].set_tlsext_host_name(server.sni)` is `False`, then the `set_tlsext_host_name` will never be called. I think we need to make sure that `self.tls[server].set_tlsext_host_name(server.sni)` is only called if `self.tls[server].set_tlsext_host_name(server.sni)` is `True`.",
+      "reference": "Do you want to cherry-pick these separately and get them merged already or keep it in here? Both is fine with me!"
+    },
+    {
+      "prediction": "I think this should be 1_000_000 instead of 1_200_000",
+      "reference": "Hmm what are these magical constant numbers? And why does this change mean they should increase?"
+    },
+    {
+      "prediction": "I don't think this should be changed. If you want to change it, you need to do it in the `setup.py` file of `boto` package, not in the `__init__.py` file of `boto` module.",
+      "reference": "This change shouldn't be included with this push. Shouldn't push version changes with a pull-request."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add this. I think it should be in the `callHandler` method, since that's where the `handle` method is.",
+      "reference": "nit: let's move this right under where we create the `responseWriter`, before `h.callHandler`"
+    },
+    {
+      "prediction": "This should be `release_assert_success<MDB_val, MDB_val> ((*this), s);`",
+      "reference": "The way it was implemented, needed to pass the template values. Please check whether we want this."
+    },
+    {
+      "prediction": "I think this should be `propagationDelay` instead of `propagationDelay`",
+      "reference": "Just FYI we need this to align with `builtin.EpochDurationSeconds` from specs-actors or the state machine computations will be off. We'll need to turn this into a config option and follow the same monkey patching paths as the min miner size, seal proof types etc. The EpochDurationSeconds is currently a `const` so can't do it yet tho."
+    },
+    {
+      "prediction": "I think this should be `from past.builtins import print_function` instead of `from __future__ import print_function`.",
+      "reference": "Is this import needed?"
+    },
+    {
+      "prediction": "I think this should be `padding: 0, margin-bottom: 2rem`",
+      "reference": "Dashboard stories also need to have the default padding."
+    },
+    {
+      "prediction": "I think this should be renamed to `REWRITE_BINPACK` or `REWRITE_BINPacks`.",
+      "reference": "Is there a reason why we wouldn't use this as the default?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `CreateEdgeProcessor` class itself, since it's only used by that class.",
+      "reference": "if there is an tag of the same name?"
+    },
+    {
+      "prediction": "I think this file should be renamed to `util` or `libutil` or something like that. It's not really a part of the net package, and it's not really a part of the netlib package either.",
+      "reference": "I'd add `// +build go1.10` to prevent from accidentally compiling with older Go."
+    },
+    {
+      "prediction": "I don't think we need to change this line.",
+      "reference": "Not sure how this import relates to the PR but seems like a good idea anyway..."
+    },
+    {
+      "prediction": "I think this should be moved to the header file where it is used.",
+      "reference": "Hmm this is strange, why should the SettingsStruct import something from `WebServer` ? If that's really needed, then we must move stuff as it makes no sense to have some code related to viewing things included in a settings struct. It should be the other way around."
+    },
+    {
+      "prediction": "I think this should be `loggerFactory = new LoggingFactory(loggerFactory);` and not `loggerFactory.AddConsole(LogLevel.Debug);`",
+      "reference": "nit: Space between these lines."
+    },
+    {
+      "prediction": "I think it would be better to pass the translator as an argument to the constructor. This way, you don't have to pass it as a parameter every time you create a new instance of the `ResultsProvider` class.",
+      "reference": "Why aggregator should have this translator? It's Selenium specifics for now..."
+    },
+    {
+      "prediction": "I think this method can be removed.",
+      "reference": "I don't understand why this function is needed ? Can't you just pass the information when the `DocComparator` is created in the SortField ?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [152/120]",
+      "reference": "This rescue is repeated exactly in `app/jobs/mailchimp_removal_job.rb`. Could we extract a method and pull it up into `MailChimpJob` for reuse?"
+    },
+    {
+      "prediction": "I think this should be `options.headers = {content-type: xhr.getResponseHeader('content-type')}`",
+      "reference": "It's not directly related to this PR, but it would be nice to flow the original error as we're going to print that if the promise is rejected and it'll be more informative than the vague `Network request failed`. Also, I'm not a js dev, so maybe I'm wrong, but throwing a TypeError feels a bit odd."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to do this, but I don't have a better suggestion.",
+      "reference": "Can we print a debug message when this is used? Since it's a debugging variable, I wouldn't want users to have issues because of a left-over env."
+    },
+    {
+      "prediction": "I think it would be better to have this in a separate file, so that we don't have to update this file whenever we add a new component.",
+      "reference": "Should we have an avatar, truncate, etc examples, too?"
+    },
+    {
+      "prediction": "I'm not sure if this should be removed or not. I think it should be removed, but I don't have a strong opinion.",
+      "reference": "could you add this to `docs/source/reference/indexing.rst` ?"
+    },
+    {
+      "prediction": "I think this should be `from rdkit import six` instead of `from rdkit.VLib.Node import six`",
+      "reference": "Why the change from rdkit.six to six?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think it would be better to just always prefetch remote data, even if it's not in the local cache.",
+      "reference": "Looking good, let's just move this inside the `if` to avoid doing the extra work if we don't need it."
+    },
+    {
+      "prediction": "I think this should be `\"Batch cannot be re-execed, this batch has already been executed, create new batch to execute\"`",
+      "reference": "I think we should change this to an instanceOf check for MongoBatchReExecutionError"
+    },
+    {
+      "prediction": "I think this should be `conversion.Converter` and not `conversion.NewConverter`",
+      "reference": "Required by the libcalico-go changes"
+    },
+    {
+      "prediction": "I don't think we should have a special error type for this. I think we should just return an error when we try to get a block that doesn't exist.",
+      "reference": "Explain why we don't reuse the corresponding server error locally?"
+    },
+    {
+      "prediction": "I think this should be `DateUtil.formatDate(this.releaseOrCreateDate, \"yyyy-MM-dd'T'HH:mm:ssZ\")`",
+      "reference": "@JayanthyChengan This change is still missing (and should be last one!): change this line to call getDateToDisplayOnCard(), that way the logic is in only one place (that method)."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "What was wrong with more concrete error?"
+    },
+    {
+      "prediction": "Shouldn't this be `$classes[] = basename(get_permalink());`?",
+      "reference": "can you replace the tabs with 2 spaces please to match the rest of the project?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `flux` since that's where we store the integrals.",
+      "reference": "Why are we adding it to the integral function?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just set the ID to be the same as the src address, since we don't know the proto or mode of the src address.",
+      "reference": "Same for this file"
+    },
+    {
+      "prediction": "Shouldn't this be `FileCharacteristics` instead of `FileAppenderFileCharacteristics`?",
+      "reference": "Looks good! But why it's still a out parameter? We can refactor the whole tree because it's internal?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "probably better to use `wlr_seat_keyboard_end_grab()`."
+    },
+    {
+      "prediction": "I think this should be `UNPROTECTED_TRANSACTIONS_ALLOWED` instead of `UNPROTECTED_TRANSACTION`",
+      "reference": "Such a large test impact for such a little 3 line change ;)"
+    },
+    {
+      "prediction": "Why do we need this change?",
+      "reference": "Is the `cascade` necessary? It should work also without the casdade, because Payments are always already persisted when use cases with them. If it doesn't work, please tell me the reason. It is as same in Transport."
+    },
+    {
+      "prediction": "I think this should be `235160`",
+      "reference": "This is not very short. I'm not sure sure if there's a system for assigning shortcodes to issues."
+    },
+    {
+      "prediction": "I think this annotation should be called `edit-in-place-annotation` instead of `edit-in-place`",
+      "reference": "We should probably prefix this with `acme-http01` to bring it in-line with other annotations, e.g. `certmanager.k8s.io/acme-http01-edit-in-place`. It's quite wordy, but it *is* more specific (and I'd hope more obvious to the user what it is for?)"
+    },
+    {
+      "prediction": "I think it would be better to make this `public static final` and use it in both places where it is used.",
+      "reference": "No need to add this file change for this PR. Please remove it."
+    },
+    {
+      "prediction": "I think it would be better to put this constant in `java.lang` instead of `net.sourceforge.pmd.lang.java.rule`",
+      "reference": "this extra dot at the end is causing a couple tests to fail on Travis"
+    },
+    {
+      "prediction": "I think this should be `print(\"Capturing on '%s'\" % kargs.get('iface', conf.iface))`",
+      "reference": "Did you try using `global` with a regular integer ?"
+    },
+    {
+      "prediction": "I think this should be `sizeof(extractor::NodeBasedEdge) == 20` and not `sizeof(extractor::NodeBasedEdge) == 24`",
+      "reference": "windows size might be off"
+    },
+    {
+      "prediction": "I think this should be `#if defined(ADIOS_HAVE_DATAMAN) || defined(ADIOS_HAVE_ADIOS1)` instead of `#if defined(ADIOS_HAVE_DATAMAN) || defined(ADIOS_HAVE_ADIOS1)`",
+      "reference": "This should still be guarded with an `#ifdef` since the DataMan code will only be supported for shared lib builds."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the attribution, but it's the only way I've been able to get it to work.",
+      "reference": "thanks for cleaning up these old length checks"
+    },
+    {
+      "prediction": "I think this should be `e.config.Input.Namespace` instead of `e.config.Input.Name`",
+      "reference": "What will be logged when the `Input.Namespace` was not specified?"
+    },
+    {
+      "prediction": "I think this should be `true` by default.",
+      "reference": "I wouldn't expose this until the feature is ready. We are most likely to cut releases while still in development"
+    },
+    {
+      "prediction": "`ariaQuery` is not a real polyfill for the `aria-describedby`/`aria-for`/`aria-labelledby`/`aria-indices`/`aria-labelledby`/`aria-for`/`aria-index`/`aria-count`/`aria-hidden`/`aria-invalid`/`aria-invalid-msg`/`aria-labelledby`/`aria-for`/`aria-count`/`aria-hidden`/`aria-invalid-msg` polyfills.",
+      "reference": "I don't think we should pull aria-query into axe-core at this point. There's a much simpler fix for this. Please don't overthink it."
+    },
+    {
+      "prediction": "I think we can remove the `Slot.Set()` call here, since it's not doing anything.",
+      "reference": "@ejsmith What if instead of making these two methods public we made it a partial class and moved them to a separate file so that file could be part of the include list? I bring it up just because I don't think it is a very nice API that was ever intended to be public? I think they are only called by a couple of other very internal-y spots. /cc @reyang"
+    },
+    {
+      "prediction": "I think this should be `triggerCond` instead of `expireCond`",
+      "reference": "would be good to turn these into constants."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should leave this as it was, since we don't have a good way to tell if it's a wire or go-wire at this point in the code.",
+      "reference": "Seems like it should be `generated by Wire` here and in the SUT?"
+    },
+    {
+      "prediction": "I think this test should be moved to `storage_test.go` since it is not related to the rest of the tests.",
+      "reference": "I know the main cases show up in the daemon test already but a little bit more unit test coverage would be nice and probably not too annoying to set up. Ideas: 1. Verify that the string \"no duplicates\" is in the LastDuplicate field when we propose a new deal. 2. Verify that proposing a duplicate deal fails when the duplicates boolean is set false. 3. Start out in a state with more than 1 duplicates. Verify that the LastDuplicate field is set to the last duplicate when we propose with a true duplicates bool."
+    },
+    {
+      "prediction": "I think this should be `Timeout.InfiniteMilliseconds` instead of `Timeout.Infinite`",
+      "reference": "`Timeout.Infinite` is actually `-1`. Here the correct message can be taken from Line:67."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "I think we need `#if !SILVERLIGHT` here and at the test."
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "Can you change this import to the apache dependency? I didn't even know that jgit contains such a method..."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think we should use the service account name as the name of the service account. If we want to use the k8s service account name as the name of the service account, then I think we should use the `Name` field of the `k8s.ServiceAccount` resource.",
+      "reference": "Four strings in a row makes me nervous, can we use an arguments struct instead? Or reuse the struct I recommend in the reconciler."
+    },
+    {
+      "prediction": "I don't think we should be using `-A` here. I think we should be using `-a` instead.",
+      "reference": "should work with no options, if my suggestion is accepted. JSON is the one that gets weird since a JSON object appended to a JSON object isn't valid JSON."
+    },
+    {
+      "prediction": "I think this should be renamed to `check-streaming` to match the rest of the commands.",
+      "reference": "I presume this isn't some special term..."
+    },
+    {
+      "prediction": "I think this should be `return (headingLength > paragraphLength) || (headingLength < paragraphLength / 2) || (headingLength === nextSibling.textContent.trim().length);`",
+      "reference": "Did you mean to return here? ;) You'll also need to add an `incomplete` message to the rule metadata."
+    },
+    {
+      "prediction": "I am not sure if this is the right place for this annotation. I think it should be in the `PersistentVolumeClaim` and `PersistentVolume` annotations.",
+      "reference": "Why are we calling these as annotations? I guess this is an old type that we are re-using. But lets make it a point to avoid once we feel that this design/code is not good."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. The rest client is already sending the request, so we shouldn't wait for it to finish.",
+      "reference": "`invocation.getInvocationStageTrace().startSend()` is inside `RestClientSender.send`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `actpool.go` since it's only relevant to that pool.",
+      "reference": "use \"actionHash\" to be consistent with other places"
+    },
+    {
+      "prediction": "I think this class should be renamed to reflect that it is a property descriptor rather than a property descriptor factory.",
+      "reference": "Can we derive `Class<E>` by looking at one of the label choices, if at least one is always provided? Given these are immutable, a real-world case for an empty enumerations eludes me."
+    },
+    {
+      "prediction": "I think it would be better to use `vk::Format` instead of `VkFormat` to be consistent with the rest of the code.",
+      "reference": "I am now confused here, the spec says > If format is VK_FORMAT_UNDEFINED, all members of samplerYcbcrConversionComponents must be the identity swizzle. The spec also says > samplerYcbcrConversionComponents is the component swizzle that **should** be used in VkSamplerYcbcrConversionCreateInfo. so you are allowed to set the swizzle to anything I want, so I guess this is valid usage here... but now I am more curious if I have an external format does it not have to follow any component swizzle rules and can be anything regardless of the `ycbcrModel`? (I guess really a question for @critsec and is not a blocking concern for the scope of this PR)"
+    },
+    {
+      "prediction": "I think this should be `app.SiteStatus() != platform.AppNotRunning`",
+      "reference": "I'd say the error should be an error, not instructions to the user. So error would be something like \"site should be running and is not\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be moved to the `SNSTests` class, since this test is related to the `SNS` module.",
+      "reference": "Thanks for this PR @pinzon !   Rather than resetting the `sms_messages` list, can we fix the assertion in line 902 - rather than asserting `len` equality, we should assert that all expected messages are contained in the list. (In this case the phone numbers should be reasonably random, and ideally we should not have duplicates across different tests). We're planning to introduce test parallelization in the near future, and this will help us avoiding undesired side-effects.   Thanks"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "Some weird blocks formation again."
+    },
+    {
+      "prediction": "I think this should be `fmt.Sprintf(\"%[1]s-worker-0\", testInfraID)`",
+      "reference": "This is worrisome."
+    },
+    {
+      "prediction": "I think we can remove the `AutoFSTrimStatus` and `AutoFSTrimUsage` methods as well.",
+      "reference": "nit try rephrasing to something like \" AutoFilesystemTrimUsage returns the volume usage and trimmable space of locally mounted pxd volumes\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be under `ENV['PULP_SERVER_BIN_DIR']`",
+      "reference": "currently the puppet-agent package doesn't symlink files into /opt/puppetlabs/bin. Also many puppet acceptance tests that execute on Unix assume puppetbindir is a single directory (windows doesn't have this issue). So I think if we want to unblock testing, this should just be changed to `h['puppetbindir'] = '/opt/puppetlabs/puppet/bin`"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be using the `configDB` property of the `PouchDBService` instead of creating a new `PouchDBService` instance every time we need to query the config DB.",
+      "reference": "any point in keeping this here ?"
+    },
+    {
+      "prediction": "I think it would be better to use `checkAndPanic(CaseSensitive, g.Entries)` instead of `checkAndPanic(CaseInsensitive, g.Entries)`",
+      "reference": "I don't think we need this check. Instead `Entries.merge` should check that its argument is correctly sorted (in addition to the existing pre/post check on the method receiver)."
+    },
+    {
+      "prediction": "Is there a reason for this ifdef? I don't see it being used anywhere.",
+      "reference": "This is not really good to mix different code bases. @TheMarex I understand the point about placing code into cpp file, but we still need to move templated version into an impl header, remove references to offline code here and keep template instantiations in a cpp file to avoid rebuild. So offline code could instantiate own implementations using private impl headers. Otherwise we again make wrong dependencies into OSRM code and block changes in separate projects."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just remove the entry from the history.",
+      "reference": "Then, where is it removed?"
+    },
+    {
+      "prediction": "I think this should be `CaseSensitivePath = !CaseSensitivePath` instead of `CaseSensitivePath = true`",
+      "reference": "The tests are not run concurrently? (I don't know. Better check.)"
+    },
+    {
+      "prediction": "This should be `const`.",
+      "reference": "Regardless of scale?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should go in `pkg/algo`, but I'm not sure if that's the right place to put it. I think it should go in `pkg/telemetry`, but I'm not sure if that's the right place to put it. I think it should go in `pkg/version`, but I'm not sure if that's the right place to put it. I think it should go in `pkg/telemetry`, but I'm not sure if that's the right place to put it. I think it should go in `pkg/version`, but I'm not sure if that's the right place to put it. I think it should go in `pkg/version`, but I'm not sure if that's the right place to put it. I think it should go in `pkg/version`, but I'm not sure if that's the right place to put it. I think it should go in `pkg/version`, but I'm not sure if that's the right place to put it. I think it should go in `pkg/version`, but I'm not sure if that's the right place to put it. I think it should go in `pkg/version`, but I",
+      "reference": "Nit: should not we hide them from printing out?"
+    },
+    {
+      "prediction": "Why do we need to change this?",
+      "reference": "What is the reason to modify those things? I noticed it breaks the examples using `asyncio`"
+    },
+    {
+      "prediction": "I think this should be `transport.Protocol` instead of `transport.Encoding`",
+      "reference": "WHOOOOOOOPS! so this is a wire protocol breaking change. I deployed a dev version of yarpc 1.9 to demo-yarpc-go and it caused bad request errors because it couldn't determine the procedure anymore. (in coordination with yarpc-prober)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `core/uri_parser.js` file.",
+      "reference": "if we keep this, can we stick to our convention of using UPPER_CASE for constants? Otherwise this looks like a floating variable to me."
+    },
+    {
+      "prediction": "I think this should be `getattr(self.underlayer, 'seq', None)` instead of `getattr(self.underlayer, \"seq\")`",
+      "reference": "You probably don't want to assume that `self.underlayer` has a `seq` attribute."
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "reference": "This apparently breaks the language selection."
+    },
+    {
+      "prediction": "I think this should be `ALTER TABLE tags CHANGE name name VARCHAR(255) CHARACTER SET utf8 COLLATE utf8_bin;` and not `ALTER TABLE tags MODIFY name varchar(255) CHARACTER SET utf8 COLLATE utf8_bin;`",
+      "reference": "since we're not using mysql, do we need this migration?"
+    },
+    {
+      "prediction": "I think this should be `ENABLE_PARTITIONS_ADJUSTMENTS`",
+      "reference": "i'm not sure about htis naming here. both the word partition and adjustment are very ambiguous and it's not clear what it does. i would maybe go for something less generic and more concrete like `GOV_CLOUD_SUPPORT`"
+    },
+    {
+      "prediction": "I think we can remove this method and use `this.activeFlows` directly.",
+      "reference": "unfinishedFlows might not be the same as activeFlows. Will it cause confusion here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could find that worked.",
+      "reference": "Same, remove the error handling."
+    },
+    {
+      "prediction": "Should this be `Task<IDisposable>` instead of `Task`?",
+      "reference": "We should not change this at this time."
+    },
+    {
+      "prediction": "`stride` should be a tuple of length 2, and the first element should be `(w, h)`.",
+      "reference": "We may also update the docstring in the AnchorGenerator's arguments at around line 15."
+    },
+    {
+      "prediction": "I think this should be `inserted.first->commandBuffer`, not `inserted.first->second->commandBuffer`.",
+      "reference": "FormatHandle isn't current typesafe w.r.t. requiring *VULKAN* handles... so this slipped through."
+    },
+    {
+      "prediction": "I think this file should be renamed to `eventbus_test.go`",
+      "reference": "What kind of dependency is here, I think we have own implementation of event wo external dependencies"
+    },
+    {
+      "prediction": "I think this should be Kokkos::abort(error_message.str().c_str()); instead of `std::stringstream error_message; error_message << \"Requested GPU with id \" << id << \" but only \" << gpu_devices.size() << \" GPU(s) available!\\n\"; Kokkos::abort(error_message.str().c_str());`",
+      "reference": "Throw an exception instead"
+    },
+    {
+      "prediction": "I think you can remove `org2` now, it's no longer used.",
+      "reference": "we shouldn't need the `FactoryBot.` in these. Ok to leave for now since all of these specs have them. Can clean up later"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that the uploads path is always relative to the storage path. If the uploads path is relative to the storage path, then it will be relative to the root of the filesystem, which is not what we want. I think we need to make sure that the uploads path is always relative to the storage path, and then we can remove the `Storage::url()` call.",
+      "reference": "Isn't the better option to actually do `$this->getDisk()->url($uploadsPath)` without changing the filesystems config? Not sure why you think that change is required but it's definitely not getting merged."
+    },
+    {
+      "prediction": "I think this should be `or itemmethod.parent.frame().name in SEQUENCE_TYPES` instead of `or itemmethod.parent.name not in SEQUENCE_TYPES`",
+      "reference": "Not sure if we need this. It's not tested because it only guards against a new conditional definition of a dunder method in a built-in package. I am fine with leaving it out."
+    },
+    {
+      "prediction": "I think this should be `internal` so that it can be used by other projects that don't have `Microsoft.DotNet.Build.Common.Desktop` package.",
+      "reference": "Is this the only task in this library? If so this is good, if not then we should do it for each task."
+    },
+    {
+      "prediction": "I think we should use 60, 10 here.",
+      "reference": "I am not sure what is 120 and what is 10. Hence unable to review."
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. The firewall is used to configure the firewall rules, and firewall rules are not the same thing as OpenVPN rules.",
+      "reference": "Not needed anymore"
+    },
+    {
+      "prediction": "I think this should be `// Package pingpong exposes the simple ping-pong protocol which measures RTT with other peers.`",
+      "reference": "Please not abbreviations"
+    },
+    {
+      "prediction": "Nit: I think we can remove the `bool(false)` return value here.",
+      "reference": "This shouldn't get stomped over on line 207"
+    },
+    {
+      "prediction": "I think this should be `int` instead of `zero-indexed`",
+      "reference": "are they not 1-indexed?"
+    },
+    {
+      "prediction": "I think we can remove the lastTime variable now.",
+      "reference": "I know that previously this check was here but @ralph-pichler replaced it with the `CheckTimestamp` field stating that checking against `Timestamp` was incorrect. @ralph-pichler can you elaborate?"
+    },
+    {
+      "prediction": "I think it would be better to use `from typing import Tuple, Union` instead of `from typing import ....`",
+      "reference": "`short_version = __version__`"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this snippet. It should go in the `iam` subpackage of the `aws-iam` module.",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think we can remove `Serializable` from this interface.",
+      "reference": "Required for the SparkDataFile SerDe"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Do you want to make this an assertion instead?"
+    },
+    {
+      "prediction": "I think this should be `Copyright (C) 2015-2016 NV Access Limited`",
+      "reference": "When updating the copyright, we tend to cover previous years, not just the current. So, if we started the file in 2015, we'd update to say 2015-2016. I'm not sure if there's a legal requirement surrounding this, but it's the convention we and many others have adopted for code."
+    },
+    {
+      "prediction": "I think this is the only place where we need to check the merged status.",
+      "reference": "Wait, so it looks like this function is called _every_ time we fetch the root node, i.e. we do an MD head fetch every time we fetch the root node? Not for this PR, but it seems like we should make that unnecessary, since this function only ever does anything when head is `nil`, i.e. the first time. Maybe add a TODO?"
+    },
+    {
+      "prediction": "I don't think we should change the signature of `__init__` in this PR.",
+      "reference": "Why are you changing this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way I could get it to work.",
+      "reference": "`json.RawMessage` is a `[]byte`, which is already a pointer. Why do we need to have pointer here if we just reading this value?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it's the only way I could get it to work.",
+      "reference": "should we actually switch to `All` or just add `0.4.0` explicitly to the list of supported CNI versions? What is a new CNI version is released?"
+    },
+    {
+      "prediction": "I think this should be `gcp_type` instead of `resource_util`",
+      "reference": "pylint might complain that this isn't alphasorted?"
+    },
+    {
+      "prediction": "I think this should be `--ignore-certificate-errors --enable-features=NetworkService --enable-logging`",
+      "reference": "It seems that NetworkService is behind a feature flag because it's still experimental. If we see new issues, we should look at flipping this back off."
+    },
+    {
+      "prediction": "I think we can remove this function and just use the `BackEndStatus` function in `backend.go`",
+      "reference": "I think it makes more sense to have this be a method on TaskStatus. Also, nit, Backend, not BackEnd."
+    },
+    {
+      "prediction": "I think this should be `Kokkos::Tools::Experimental::Impl::get_instance_id` or `Kokkos::Tools::Experimental::Impl::instance_id` instead of `idForInstance`.",
+      "reference": "Could you please elaborate on why you need all this magic here? In which cases do you want the same id to be returned? Does this work the same way for all backends?"
+    },
+    {
+      "prediction": "I think we can remove the `PluginsConfigs` field from the `Config` struct. We don't need to keep track of all the config files.",
+      "reference": "nit: `PluginConfigs` feels a little better"
+    },
+    {
+      "prediction": "Why do we need to stop the probe server here?",
+      "reference": "think we need to find out the root-cause and fix, instead of just removing `NoError`? it erred b/c `context canceled`, so call this before context cancelled is the right fix"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we need to make sure we don't have two goroutines writing to the cache at the same time.",
+      "reference": "Please expand \"later behavior\". Presumably something to do with the prefetch logic?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do here. I think we should just return `response` and let the caller decide what to do with it. If we do this, we need to make sure that `swarm.ZeroAddress` is returned in all cases where `response.Message == \"upload error\"` and `response.Message == \"upload cancelled\"`.",
+      "reference": "The `response` variable is not used, it can be removed, also the fmt.Errorf."
+    },
+    {
+      "prediction": "I think this should be `return value?.Length > 0 ? string.IsNullOrEmpty(value.Trim()) : true;`",
+      "reference": "Do you find this really easier to read?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the `addOneRowAggregates` function.",
+      "reference": "A small nit: This variable should start with a lower case letter."
+    },
+    {
+      "prediction": "nit: I think we can remove `context.getNamer().getPageSizeFieldName(pageStreaming)` as it is not used anywhere else.",
+      "reference": "Where is this used for Node?"
+    },
+    {
+      "prediction": "I think this should be \"Press tab to select identity or provider identity to connect <your-identity> <provider-id>\"",
+      "reference": "Is `<your-id>` really easier to understand than `<your-identity>` for CLI user? For me, `identity` seems like a concept we use publicly, and `id` is just an internal shortcut for it."
+    },
+    {
+      "prediction": "I think this should be `if newGroup.ServiceReference != nil || newGroup.Selector != nil`",
+      "reference": "nit.. `if` .. `else` for logging probably can be avoided.. maybe only log that an internal group was created for cluster group"
+    },
+    {
+      "prediction": "I think this file should be renamed to `aws_aws_aws_aws_aws_aws_test.go`",
+      "reference": "There's a lot of changed tests -- but no new tests. You need unit and integration tests for this."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `TxPoolFilter` class, since that's where the `Accept` method is.",
+      "reference": "WithMessage what gas limit is?"
+    },
+    {
+      "prediction": "I think this should be `skip: true` instead of `skip: 'cursor stream errors'`",
+      "reference": "Why are these skipped?"
+    },
+    {
+      "prediction": "I think we can remove the MetaServer struct and use MetaManager instead.",
+      "reference": "We'd better add the `json:\"metaServer,omitempty\"` here"
+    },
+    {
+      "prediction": "This should be `pip_install = [\"install\", \"--no-deps\", \"-e\", \"../..\"]`",
+      "reference": "So... when we want to install mitmproxy with pinned dependencies, we first install all pinned dependencies, and then in a second step install mitmproxy as editable, making sure that no additional dependencies are sneaking in."
+    },
+    {
+      "prediction": "I think this should be `return fmt.Errorf(\"Failed to exec %s: %s\", cmd, err)` and not `return fmt.Errorf(\"Failed to exec %s: %s\", cmd, err)`",
+      "reference": "not sure how this got in master, but the linter is unhappy about it so I fixed it here"
+    },
+    {
+      "prediction": "I think this should be 10 min.",
+      "reference": "No need for this TODO anymore, right?"
+    },
+    {
+      "prediction": "I think this should be `if 'kinto.plugins.default_bucket' not in settings['excludes']` instead of `if 'kinto.plugins.default_bucket' not in settings['excludes'] and 'kinto.plugins.default_bucket' not in settings['includes']`",
+      "reference": "I would assign the key you're looking for into a variable first, and then use it in the comparisons."
+    },
+    {
+      "prediction": "I think this should be done in the `PluginManager::load` method.",
+      "reference": "Would be better to normalize it at the start of the method to pass through to all the other calls"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "This can't exist here either because the code editor can be used on other models than CMS templates. You could probably implement this as a `hint` property (so `hint: cms::lang.cms_object.safe_mode_enabled`) on the codeeditor formwidget instead though and I'd be fine with that."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. The scrollIntoView should be done in the `WordProcessDocument` class, not in the `WordDocumentTextInfo` class. The `WordProcessDocument` class is the base class for all process documents, including `WordDocumentTextInfo`. The `WordDocumentTextInfo` class is a subclass of `WordProcessDocument`, so it should be able to call the methods defined in the `WordProcessDocument` class. The `WordProcessDocument` class is also the base class for all process documents, so it should be able to call the methods defined in the `WordDocumentTextInfo` class. The `WordDocumentTextInfo` class is a subclass of `WordProcessDocument`, so it should be able to call the methods defined in the `WordDocumentTextInfo` class. The `WordDocumentTextInfo` class is a subclass of `WordProcessDocument`, so it should be able to call the methods defined in the `WordDocumentTextInfo` class. The `WordDocumentTextInfo` class is a subclass of `WordProcessDocument`, so it should be able to call the methods defined in the `WordDocumentTextInfo` class. The `WordDocumentTextInfo` class is",
+      "reference": "This line is unnecessary."
+    },
+    {
+      "prediction": "I'm not sure why this was removed.",
+      "reference": "How did this compile in the first place? Is there a matching declaration in a header that should be removed too?"
+    },
+    {
+      "prediction": "I think this should be `h2o_mruby_assert(mrb, \"h2o_mruby_new_str:failed to create string:\", file, line);`",
+      "reference": "Please remove the colon at the end, and merge the PR once the CI succeeds. Thank you for the PR."
+    },
+    {
+      "prediction": "I'm not sure why this is removed.",
+      "reference": "whoops. this should be protoc-gen-doc, since that is the only bit left that is built out-of-tree (everything else is either downloaded or built via modules already depended on by spire's go.mod)"
+    },
+    {
+      "prediction": "I think it would be better to have a separate interface for reporting events and a separate implementation for reporting events. It would make it easier to see what's going on.",
+      "reference": "The convention is to use <p> see the google style guide and use checkstyle to check. You can change it later."
+    },
+    {
+      "prediction": "I don't think this is the right place to add the list markers. I think they should be added in the same way as the item markers.",
+      "reference": "Using non-keyword args after the keyword arg (parent-self) will cause a SyntaxError."
+    },
+    {
+      "prediction": "Style/BlockArguments: Use %w or %W for an array of words.",
+      "reference": "I think you'll want to add some YARD docs to this method to get the PR approved"
+    },
+    {
+      "prediction": "(style) 'nullable' may be misspelled - perhaps 'nullable'? (style) 'ostree' may be misspelled - perhaps 'ostree'?",
+      "reference": "Hmm, AFAICT `*out_variant` isn't explicitly set to `NULL` in the `ENOENT` case."
+    },
+    {
+      "prediction": "I think this should be `log.NewZapAdapter(b.logger.With(\"error\", err))` instead of `log.NewZapAdapter(b.logger).Warn(\"Failed to create SDK client\", zap.Error(err))`",
+      "reference": "Nit: \"Failed to configure TLS for SDK client\"?"
+    },
+    {
+      "prediction": "I don't think we need to change this line, but I'm not sure why it's here.",
+      "reference": "please update your branch to the latest `master`. This change was already made in #4359"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "This line should be formatted"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `boost/program_options/program_options.hpp` under `boost::program_options::value`.",
+      "reference": "\"default 65536\" Not sure if it's important to state, but block_processor_full_size has a different default for fast_bootstrap."
+    },
+    {
+      "prediction": "I think this should be 512MB, not 512KB.",
+      "reference": "remove unnecessary whitespace changes"
+    },
+    {
+      "prediction": "I'm not sure why this is removed. It seems to me that if `Sig` is non-nil, then `Msig` must be non-nil as well, and vice versa.",
+      "reference": "Are these conditions covered elsewhere ?"
+    },
+    {
+      "prediction": "I think this should be `const auto same_name = !util::guidance::requiresNameAnnounced(name_table.GetNameForID(in_data.name_id), name_table.GetRefForID(in_data.name_id), name_table.GetNameForID(out_data.name_id), name_table.GetRefForID(out_data.name_id), street_name_suffix_table);`",
+      "reference": "the above check (by transitivity) checked for `in_data.name_id == out_data.name_id && in_data.name_id != EMPTY_NAME_ID && out_data.name_id != EMPTY_NAME_ID`. The final check of `out_data.name_id != EMPTY_NAME_ID` is missing now."
+    },
+    {
+      "prediction": "I think this should be `return -1;` instead of `return FPGA_INVALID_PARAM;`",
+      "reference": "The event handle is being passed by reference, so the dereference on 591 is actually correct. To follow the convention that variables be declared before executable statements, please change the code to the following: struct _fpga_event_handle *_eh; fpga_result result = FPGA_OK; int err = 0; if (!event_handle) { return FPGA_INVALID_PARAM; } _eh = (struct _fpga_event_handle *) *event_handle;"
+    },
+    {
+      "prediction": "This should be moved to the top of the file.",
+      "reference": "reorder includes to be alphabetically ordered"
+    },
+    {
+      "prediction": "I think it would be better to put this argument before sampling, so that the user can choose whether or not to map the target values when need_unmap is true.",
+      "reference": "Rename `need_unmap` to `unmap`."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "In line 28, it has been declared twice, so removing `cm` should not impact, can you explain me little bit? :)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this interface. I think it should be something like `TopologyCache` or `TopologyCacheLock`.",
+      "reference": "i don't think these unexported functions belong on the public interface, they can exist solely on the struct."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `helper.go` since it's only used by `NewBindAddresses`.",
+      "reference": "I'm not entirely clear on why this is necessary?"
+    },
+    {
+      "prediction": "I'm not sure that this is the best way to do it, but I think it's the only way I could come up with.",
+      "reference": "Why do we need to cast to `str` here?"
+    },
+    {
+      "prediction": "Why is this test being ignored?",
+      "reference": "The ignore import above should be unused now and can be removed."
+    },
+    {
+      "prediction": "I think it would be better to move this method to `QueryVisitor` and have `visit` method call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort` and have `visit` call `visitFilter` and `visitSort`",
+      "reference": "redundant null check?"
+    },
+    {
+      "prediction": "I think this should be `Available: m.Total` instead of `Total: m.Total`",
+      "reference": "initialization of available ram is here"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `text` is a string, and `'\\u202B'` and `\\u202E` are both strings, so the concatenation doesn't work. I think we need to do something like `text = text.replace(/\\s/g, '') + '\\u202B' + text + '\\u202C';`",
+      "reference": "Can have negative numbers with the '-' on the right (`10-`) by just using line 610."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this snippet. I think it should go in the `aws-sdk-java-samples` repository under the `samples_iam` folder.",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think this should be `if len(check_commands) > 0:`",
+      "reference": "Shouldn't we use store a message which indicates that there was no available checker command? By the way how is it possible that metadata.json doesn't contain the checker command? Maybe this was the case in the earlier versions?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. It seems like it should be in `spec_helper.rb`, but I can't find where it is used.",
+      "reference": "Given that this is a boolean predicate, maybe it should be `full_backtrace?`."
+    },
+    {
+      "prediction": "I don't think we should be importing ldap here. It's not used in this file.",
+      "reference": "Just curious why the blank line here?"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "I like the idea, but we don't want a dependency on mitmproxy in pathod."
+    },
+    {
+      "prediction": "I'm not sure if we want to keep this as the string representation to avoid having to parse. For now, it's relatively easy to append new values when needed.",
+      "reference": "Why not adding this to the interface and keep passing the interface (as the interface is internal I don't get why you're not adding it there)"
+    },
+    {
+      "prediction": "I don't think we need to import `PY310_PLUS` from `astroid.const` here.",
+      "reference": "`PY310` should probably be imported from `pylint.const`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "Can we propagate this error up the stack?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I would expect it to be in the `compute_reference` function, not in the `compute_reference` function.",
+      "reference": "Why reference is converted to table?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `roots_server.c` where the `roots_server` is initialized.",
+      "reference": "You shouldn't use assert for runtime errors."
+    },
+    {
+      "prediction": "I think this is the only thing that needs to be changed here.",
+      "reference": "Does this really belong in the vendor directory if you wrote it? At very least, should it get its own repo somewhere in addition to being dropped here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that the `@Rule` annotation on the `SampleXPathRule` matches the `@Rule` annotation on the `TestRule` in the `sample.dummy.java` file.",
+      "reference": "I think it would be better to place the rule name before the message. Messages are of varying lengths and the rule names will end up misaligned"
+    },
+    {
+      "prediction": "I think this should be `R0125` instead of `R0124`.",
+      "reference": "Can we remove the argument from this message? A message will include a line number already that points to the place in the code that has the problem. I think we can remove it."
+    },
+    {
+      "prediction": "Why is this virtual method needed? I don't see it being called anywhere.",
+      "reference": "This whole class is now a mix of virtual and non-virtual. I tried to see what would happen if I made everything non-virtual in dev, and the build still passes. It turns out that none of the other methods actually need to be virtual. Can you make *all* the logging methods either virtual or not for consistency? Thanks."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this flag. I think it should be something like `--mmn-address` or `--mysterium-api-address`.",
+      "reference": "This looks a bit cryptic to me, and also it contains a dash. How about `mymysterium.url`?"
+    },
+    {
+      "prediction": "`&var` is not guaranteed to be the same value as `var`. It is guaranteed to be the same value as `var` if and only if `var` is a pointer to an integer. If `var` is not a pointer to an integer, then `&var` will refer to a different location in memory than `var`. In that case, you will need to use a pointer to an integer to compare `&var` and `var`.",
+      "reference": "You did notice the no newline at the end of this file, right?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the `SarifResult` class.",
+      "reference": "Here's the object model change. `CodeFlows` is a set of objects of type `CodeFlow`, rather than a list of list of `AnnotatedCodeLocation`. Exactly parallel with `Stacks`."
+    },
+    {
+      "prediction": "I think we should move this to the top of the file, so that it's visible to linting tools like `flake8` and `flake8-bugbear`.",
+      "reference": "Sorry, one more thing... You should use `app.config` instead of importing it directly. See the code below."
+    },
+    {
+      "prediction": "I think this method should be `void close()`.",
+      "reference": "This was correct before; \"setup\" is a noun and \"set up\" is the verb form."
+    },
+    {
+      "prediction": "I think this should be `maybeSetAccount: false` instead of `maybeSetAccount: true`",
+      "reference": "What's the reason for this change here? I didn't see it mentioned in the PR and it's a bit of a confusing param name"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I'm not sure if there is a better way to do this.",
+      "reference": "I wonder why ~ and not a wildcard instead?"
+    },
+    {
+      "prediction": "I think it's better to add a check for `null` here.",
+      "reference": "when will this happened? if happened, just ignore it Silently?"
+    },
+    {
+      "prediction": "I think this is not the right way to test this. We need to test that we can merge 2 manifest files with different sizes. We can do this by writing two files of different sizes, and then merging them.",
+      "reference": "3x the smaller size would be around 17k, and we need it to be at least about 13k, which is 2x the larger size. I'd probably set this to 15k to split the difference and hopefully avoid needing to update this again as tests change. This is minor, though."
+    },
+    {
+      "prediction": "I think this should be `assert isinstance(input, (datetime.datetime, Timestamp))`",
+      "reference": "@scottpurdy, strictly speaking, should this scenario raise a ValueError exception instead of AssertionError?"
+    },
+    {
+      "prediction": "I think it would be better to have this enum as a separate file from the rest of the rest of the package. This way, you can easily add a new type of executor in the future without having to change the rest of the rest of the rest of the package. For example, if you wanted to add a new type of executor, you could just create a new class in the `executor` package, and have that class extend this `ExecutorType` class. Then, you could just add a new entry to the enum, and no changes will be needed to the rest of the package.",
+      "reference": "Can you please add open source disclaimer?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to keep the wireError and set it to the new error.",
+      "reference": "I dont want to do composite keys on purpose to verify at compile time if the message is completely filled out appropriately @sectioneight"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "don't need this line, since it is not tested in api_test?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check the status of the CStorVolumeReplica CRD. I think it's better to use `clientset.Client().V1().CStorVolumeReplicas()`.",
+      "reference": "Does this mean the list operation is done for all the namespaces? How to list the volume replicas per namespace?"
+    },
+    {
+      "prediction": "I think it would be better to use `np.argmax` instead of `np.max` to get the max coordinate.",
+      "reference": "I suggest adding `class_agnostic` as an argument of `batched_nms()`, with the default value False."
+    },
+    {
+      "prediction": "I think we should rename this to `ROW_ID_FIELD_NAME` to be consistent with the other `ROW_ID_FIELD_NAME`s.",
+      "reference": "@rdblue, did we not add the name on purpose?"
+    },
+    {
+      "prediction": "I think we can remove the `BinaryType` enum class altogether. It's not used anywhere else in the codebase.",
+      "reference": "Why is this now public?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to do this is. I'm not sure what the best way to",
+      "reference": "I guess we should backport some deprecation for this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Selenium::WebDriver::Chrome::Selenium::Remote::Chromium` module.",
+      "reference": "Moving this code inside the last case statement doesn't seem right. Why are we defining `name` and `is_relative` variables there if we aren't using them anywhere?"
+    },
+    {
+      "prediction": "I think this should be `DR_REGARG(reg)`, not `VSIBY(reg)`.",
+      "reference": "I would say just change REGARG: no need for a separate thing."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `mailer.css` file, since it's specific to the `mailer` module.",
+      "reference": "we can remove this template, it is no longer in use"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "This doesn't look like it has as much coverage of the recursive code as it once did."
+    },
+    {
+      "prediction": "Nit: I think it would be better to use `aws.StringValue(securityGroup)` instead of `securityGroup`.",
+      "reference": "Do we actually expect users to provide more information here than possibly a SecurityGroup ID? If not, then this should probably use a new type, otherwise users might be confused by what options would be respected/ignored when used in this context."
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "This seems like a new dependency on rometools. Do we need it?"
+    },
+    {
+      "prediction": "nit: `fine` is a verb, `log.fine` is a noun.",
+      "reference": "Waiting for the pull takes a long time. This message informs the user that at least one of the images being pulled is available. Please leave."
+    },
+    {
+      "prediction": "I think we can remove this line now.",
+      "reference": "I can not think in a good use for it, but may we allow a way to setup a fake with a hook for logs? this fake is special, and we may need access to some of thoe logs, at the same time it may be an overkill because we dont want to test \"sql\" implementation but results.. but we can create some tests with end to end logs in some scenaries, What do you think about it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the same way as the other variables.",
+      "reference": "I don't think you should be changing the dag unilaterally at parse time. If cores isn't specified, then it isn't specified."
+    },
+    {
+      "prediction": "I think this is a bit of a misnomer, as we don't have at-least-once semantics in the default mode, we have at-least-once semantics in the delete mode, and we have at-least-once semantics in the receive-delete mode.",
+      "reference": "Instead of just \"See SubscriberOptions....\", say something like \"Use ... to choose between the two.\""
+    },
+    {
+      "prediction": "I think this should be `sdf = Chem.SDMolSupplier(\"cdk2.sdf\")` instead of `sdf = Chem.SDMolSupplier(\"cdk2.sdf\")`",
+      "reference": "Use with statement for `f`"
+    },
+    {
+      "prediction": "I think this should be `trait T\\n` and `class C is T`",
+      "reference": "It's a small style point, but could you remove the \"extra\" space before the colon that precedes the return type? This would make it more closely match the prevailing style in these tests and in the standard libraries."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `MDOpsStandard`.",
+      "reference": "seems clunky to me to have an `MDOps` interface method that some implementations don't implement. Perhaps define a separate interface, like, `BranchResolver`, and then callers that have an `MDOps` object can check via type assertion?"
+    },
+    {
+      "prediction": "I think this method should be renamed to `findElementsByXPath` or `findElementsByXPathInWebDriver` or something like that.",
+      "reference": "This change should also probably go into the corresponding method of the abstract By class?"
+    },
+    {
+      "prediction": "nit: I think it would be better to use `byteutil` here and in other places where it's used.",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I think this should be `return err, nil`",
+      "reference": "This is likely to return an error value if the client fails to read the response we write, which is why I chose to ignore this error."
+    },
+    {
+      "prediction": "I think this should be `bool(self.storage.get('upfront_shutdown_script_enabled', False))`",
+      "reference": "this method is not used"
+    },
+    {
+      "prediction": "I think it would be better to move this to the `reload` method, as it's used in many other places as well.",
+      "reference": "Right now it's a bit unclear that this needs to be called explicitly by the implementing class. If, for example, there's another backend, it won't get this update unless we add the same function as webkit. Could you either add a note to this docstring explaining that this needs to be called, or find some way to automatically trigger this so the webkit override isn't needed (and some override passed on webengine)."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Why this change (test name still says \"must pass with metadata\")?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get Travis to pass locally.",
+      "reference": "Another nitpick: parens around the arg pretty please"
+    },
+    {
+      "prediction": "I'm not sure why this test is here, but it seems to be testing something I don't understand.",
+      "reference": "This spec can be modified, giving you extra strength (Check this fetch key doesn't work and therefore returns `nil`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this. I think it should be `caddy_rewrite_uri` or something like that.",
+      "reference": "Oh, I guess I mentioned/pressed this point in the other issue, that this should probably go into the httpserver package. In fact, so should the const above this (URLPathCtxKey). These are specific to the HTTP server."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in `validation.go` since it's only used there.",
+      "reference": "Not added by you, but we don't really need the '5' here.."
+    },
+    {
+      "prediction": "I think this should be `goog:driver`",
+      "reference": "The vendor prefix is still being used on Chromium based browsers like Edge Chromium and Chrome. Did you mean to remove this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `kb.py`.",
+      "reference": "I don't think this is needed anymore?"
+    },
+    {
+      "prediction": "I'm not sure why this is a helper function. I think it should be a separate function.",
+      "reference": "This is the only occurrence I see in the code where we now need to convert from a boolean protobuf wrapper to a boolean pointer. This felt a little cumbersome here; should we consider moving it somewhere else as a helper function?"
+    },
+    {
+      "prediction": "I think we can remove the `info.version_ = *req.version_ref();` line as well.",
+      "reference": "Need you to delete the `version_` from `HostInfo` struct ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for the primary language. I think it would be better to check for `lang` and `xml:lang` attributes and use `getBaseLang` to determine the primary language. This way, we don't have to assume that the language is the same as the XML language.",
+      "reference": "This could be the cleanest looking check we've got. Great job Jey."
+    },
+    {
+      "prediction": "I think this should be `preg_match('/^\\/[a-zA-Z0-9\\-]/', $url, $matches);`",
+      "reference": "Who invited yoda?   In all seriousness though, wouldn't an `if (empty())` be better here?"
+    },
+    {
+      "prediction": "I think this should be `appRouter.showDirect('/');`",
+      "reference": "You can use `appRouter.goHome` to do this (It's defined in site.js), it's less hacky than overriding the href."
+    },
+    {
+      "prediction": "I think this should be `LOG(WARNING)` instead of `LOG(INFO)`.",
+      "reference": "When could this happen?"
+    },
+    {
+      "prediction": "I think we can remove the `Content-Type` header from the `PostSegmentAsync` call, since it's already added by the `FlushAsync` call.",
+      "reference": "I assume this was removed because it was redundant? and conflicted with the now dynamic contentType? Edit: Oh I see it being set was moved to the `PostSegmentAsync` call."
+    },
+    {
+      "prediction": "I don't think this is a good idea.",
+      "reference": "Here is a newline missing"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Did you mean to make this http rather than https?"
+    },
+    {
+      "prediction": "I don't think you need to change this line.",
+      "reference": "The updated copyright header should be: # Copyright (C) 2006-2018 NV Access Limited, yourname"
+    },
+    {
+      "prediction": "I think it would be better to use a `try/catch` block here instead of assigning `$performanceTestSample` to a variable.",
+      "reference": "wow :+1: , i do not even know how this test works."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `MvvmCross.UI.Platform.Uap`.",
+      "reference": "File should be renamed `PlugIn` -> `Plugin`"
+    },
+    {
+      "prediction": "I think this is the only thing that needs to change in this file. The rest of the code should remain the same.",
+      "reference": "I scanned the new test cases quickly. Do we have one for an egressIPPool with multiple different IP ranges?"
+    },
+    {
+      "prediction": "Shouldn't this be `public`?",
+      "reference": "you can use a different name, like `maybe_activate` or `is_active` without `component` in the method name as this is used in the class name. E.g: `$component->should_component_be_active` is using twice the `component` word"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this check. I think it should be done in the `execute` method of the `OrderBulkOperation` class.",
+      "reference": "I think this might not be a complete enough check: what if `options` is `null`/`undefined`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this scenario. It's not really about masquerading, it's about an admin being able to click masquerade as a user. Maybe something like \"Admin masquerages as a user\" or \"Admin clicks masquerade as a user\"?",
+      "reference": "Useless assignment to variable - `user`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should go in the `build` section of the `.travis.yml` file, under the `deploy` section. The `build` section is where we do the build steps, and the `deploy` section is where we do the deploy steps. In this case, we want to download the archive, unpack it, and put it in the `otp` directory under the user's `~/.kerl/otp` path.",
+      "reference": "I think we would want to keep `-O` instead. I wonder how this worked before, though."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `AggregatorStore` itself, since it's only used by that class.",
+      "reference": "another alternate option which doesn't have the risk of too many entries (when user keeps providing keys in different order) Have the dictionary as before If tagKeys lookup fail, sort and lookup again. If fails, insert both original tagKeys and its sorted one to the dictionary. So that we only store atmost 2 entries per key set. And we only do a single lookup in hotpath, as opposed to 2 look ups."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to update the symbols. I think it should be done in the `PlanNode` constructor.",
+      "reference": "if (out != nullptr)"
+    },
+    {
+      "prediction": "I don't think we should be using `request.args` for anything other than `GET` requests. We should be using `request.form` or `request.files` for `POST` requests and `request.args` for `PUT` and `PATCH` requests.",
+      "reference": "In the listenstore you catch the exception and then pass it on via Raise. However, here you do not catch the exception. You should catch the exception, but since this is a minor aspect of this page, perhaps show an error message when the count cannot be loaded in time. Then the rest of the page can still be rendered, rather than failing the whole page."
+    },
+    {
+      "prediction": "I think this should be `transfer: bool` instead of `transfer: bool|str`",
+      "reference": "wouldn't it be simpler to just add `or transfer` here? That way the more complicated set of conditionals down below don't have to get more clauses."
+    },
+    {
+      "prediction": "I think this should be `from googleapiclient import errors as googleapiclient_errors`",
+      "reference": "ditto; please fix everywhere"
+    },
+    {
+      "prediction": "I think this should be `Specifies the path or command to run a sftp server on the remote host. The subsystem option is ignored when server_command is defined.`",
+      "reference": "Can you break this line into two parts - the first line makes the option help text for `--sftp-server` and its too long! `Specifies the path or command to run a sftp server on the remote host. The subsystem option is ignored when sftp_server is defined.`"
+    },
+    {
+      "prediction": "I think you can remove the `: ?bool` here.",
+      "reference": "I reverted that one in a previous PR because of a CI failure but it was actually unrelated"
+    },
+    {
+      "prediction": "I think we can remove the `latestVersion` variable and just do `kubeVer = version[1:]`",
+      "reference": "The for loop is useless here, since any error will make the function return."
+    },
+    {
+      "prediction": "I think you can remove this line.",
+      "reference": "Fix file endings here and elsewhere."
+    },
+    {
+      "prediction": "I think this should be a `const std::string&` instead of a `const char*`.",
+      "reference": "I'd better increase `buf` length to MAXPATHLEN and not cut p->get_project_name() output twice"
+    },
+    {
+      "prediction": "I think this should be `fmt.Printf(\"here it is: %v\\n\", err)`",
+      "reference": "Please remove the prints, even though this code will be trashed."
+    },
+    {
+      "prediction": "I think we should rename this to `dataplane` instead of `linux` and `unix`?",
+      "reference": "Not sure if you added deliberately but I've seen these popping up; is goimports adding them?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this conversion. I think it should be in `aws-cluster` since it's a specific type of `aws` resource.",
+      "reference": "I think we might need some special handling for `Spec.ControlPlaneLoadBalancer.SecurityGroups`, it looks like we are currently only handling the case that `Spec.ControlPlaneLoadBalancer` is nil."
+    },
+    {
+      "prediction": "I don't think you need to add `requires_grad=True` and `norm_eval=True` here.",
+      "reference": "Is this from a third-party library?"
+    },
+    {
+      "prediction": "I think this should be `dom.isFocusable(el) && !dom.isNativelyFocusable(el)` instead of `!dom.isNativelyFocusable(el)`.",
+      "reference": "Please put the `el.tabIndex > -1` in the first part of the expression. The matcher will run on every node on the page, so we should put the fastest part first."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `services` package, but I'm not sure if that's the right place to put it.",
+      "reference": "`identity` is unused now, is it for future usage?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could think of that works.",
+      "reference": "This would have assisted me a lot! I am on the \"functions\" page more than any other page in the docs"
+    },
+    {
+      "prediction": "I think this should be `d = d.replace(b'c__main__', b'c' + module_name.encode('ascii'))`",
+      "reference": "I think it's better we keep `(` and encode with default encoding (utf-8)."
+    },
+    {
+      "prediction": "I think you can remove this line.",
+      "reference": "this is gonna be different in an IDE vs from gradlew??"
+    },
+    {
+      "prediction": "I think this should be `RSpec::Matchers.configure do |matcher|` and `RSpec::Mocks.configure do |mock|` instead of `RSpec::Matchers.configuration.syntax` and `RSpec::Mocks.configuration.syntax`",
+      "reference": "Do we not already have something for isolating syntax?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add the copyright.",
+      "reference": "1: D100 Docstring missing 4: I102 copyright year is outdated, expected 2014 but got 2012 158: D103 Docstring missing 168: D101 Docstring missing 170: D102 Docstring missing 180: D102 Docstring missing 187: D102 Docstring missing"
+    },
+    {
+      "prediction": "nit: remove the period at the end of the sentence.",
+      "reference": "how **to** deny"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "The only beef I have with this is that it's app domain global."
+    },
+    {
+      "prediction": "I think we can remove the `runtime.GOOS != \"openbsd\"` check now, since we don't care about `openbsd` at all.",
+      "reference": "Can you link to the GitHub issue and/or PR so that it is easy for future readers to find out more about this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `manual_PostCallValidateAllocateMemory` function.",
+      "reference": "Typical to report the value found as well as the valid bounds."
+    },
+    {
+      "prediction": "I think this should be `transaction.LastValidRound` instead of `transactionLastValidRound`",
+      "reference": "maybe we want 0 as a default value?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `GetAccount`.",
+      "reference": "can we get height from sf at very beginning of the function and use that as height of the block?"
+    },
+    {
+      "prediction": "Is there a reason we don't want to have this in the `Kestrel` assembly?",
+      "reference": "Does this have to be public and in the primary namespace?"
+    },
+    {
+      "prediction": "I think this should be done in `initMDLocked`, not here.",
+      "reference": "I'm a bit confused. Doesn't `mdWriterLock` already synchronizes calls to this function?"
+    },
+    {
+      "prediction": "I think this should be `null === $media->getProviderReference() ? '' : sprintf('http://www.dailymotion.com/video/%s', $media->getProviderReference());`",
+      "reference": "Does it make sense to generate a dailyMotion url without the video reference? It does not to me, but can we throw an exception here?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Why the separation of imports here?"
+    },
+    {
+      "prediction": "I think this is the wrong place for this function. It's not a function that returns a side effect. It's a function that evaluates an expression.",
+      "reference": "Random extra newline."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could come up with.",
+      "reference": "No need for this variable."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. It seems like it should be in the `auth` module, since it's only used by the `authenticator.py` file, but I can't find where it's used there.",
+      "reference": "Why not use `msg` directly below?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think it would be better to use `@SuppressWarnings(\"checkstyle:IllegalType\")` instead.",
+      "reference": "I think other places changed to use `Deque` instead of suppressing the warning."
+    },
+    {
+      "prediction": "I think this should be `#include <boost/range/algorithm_ext/erase.hpp>`.",
+      "reference": "missing: utility, cstddef also: vector is already included in the header (interface requires vectors as in/out types)"
+    },
+    {
+      "prediction": "Why is this needed? `fetchMock` is not a global variable.",
+      "reference": "I'm not sure this is necessary in the config here, as Jest would be setting the global `fetchMock = ''`. See below."
+    },
+    {
+      "prediction": "I think this should be `mFolder.getName() != null ? mFolder.getName().hashCode() : 0`",
+      "reference": "Can `mFolder == null` happen during regular operations? Or is it only so we can be lazy in tests?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "I'd suggest using the actual MIME types here - so `text/plain` and `multipart/alternative` or does that cause problems with what postgres allows for enumeration names? If so then maybe just replace the slash with an underscore?"
+    },
+    {
+      "prediction": "I think this should be `%(team_name)s:%(user)s/%(pkg)s is private. To share it with the team, run `quilt access add %(team_name)s:%(user)s/%(pkg)s team`. `",
+      "reference": "Ohh. `TeamName` is actually a \"friendly\" name displayed in the Catalog - not the name used in the CLI. So I guess we'll need a new variable here. (That is, this is going to be mainly a `quilt.yaml` change. You won't need the `.lower()`, though.)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the same place as `flow.response is None` check above.",
+      "reference": "Nit: Don't move it between `require_dummy_response` definition and usage, this can live above or below :)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be.",
+      "reference": "With the current code, it not work if `EnableCatchupFromArchiveServers` is disabled. to fix it: in getDNSAddrs, change the predicate to `if wn.config.EnableCatchupFromArchiveServers || wn.config. EnableCatchupFromArchiveServers {` and in the catchup/service.go and catchup/catchpointService.go, use the `PeersPhonebookArchivers` only when `wn.config.EnableCatchupFromArchiveServers` is enabled."
+    },
+    {
+      "prediction": "I think this should be `GetOrCreate` instead of `GetOrCreateAndAddToCache`.",
+      "reference": "Why not make `SearchForNodes` do a `GetOrCreate` for the root node? (Not advocating for it, but just wondering if there's another reason than avoiding having to pass in the path.)"
+    },
+    {
+      "prediction": "I think this should be done in a finally block, to ensure that it is always called even if there is an exception.",
+      "reference": ":+1: amazing that this has been missing since it was originally implemented"
+    },
+    {
+      "prediction": "I think this should be `file.getSize()`",
+      "reference": "Just want to note here that although we do check `length` while writing to choose whether to close and start a new file, that doesn't happen for ORC already so it is fine to use a FS call in this method."
+    },
+    {
+      "prediction": "I think you can remove the `else` clause here.",
+      "reference": "nit: I feel multiple `if (...) { return ... }` is more readable than `if ... else ...`."
+    },
+    {
+      "prediction": "I think this should be `setup_logging()`, not `setup_test_logging()`.",
+      "reference": "This is needed for us to work in IDE UTs"
+    },
+    {
+      "prediction": "I think this should be `Only show files that are starred or not in the trash.`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to keep the `user` label, even if it's empty. This is because the `user` label is used to identify the user who is running the container. Without the `user` label, we can't identify the user who is running the container.",
+      "reference": "this `label=` is docker specific, while here in libcontainer there shouldn't be any mention to docker. `DisableSecOpt` and `DupSecOpt` should just deal with `disable,role,type,level`. Both CRI-O and docker should just pass `disable,role,type.level` stuff and not `label=...`."
+    },
+    {
+      "prediction": "I think this should be `recurse` instead of `recursive`.",
+      "reference": "You need to declare variable `recurse` above, same place as `localTime`."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the same place as the `clone` method.",
+      "reference": "Why does `clone` not already copy `xdensity` and `ydensity` from what it is cloning?"
+    },
+    {
+      "prediction": "I think this should be `throw new ArgumentException($\"No meter with name '{name}' was added to the sdk.\");`",
+      "reference": "Curious - do we do the same for traces (when no ActivitySource / legacy source are added)? (and why we want to do it for metrics?)"
+    },
+    {
+      "prediction": "I think we can remove the `private` and `slot` fields from the `SdMmcHcStartSdClock` function signature as well.",
+      "reference": "@aimanrosli23 For the changes in file SdMmcPciHci.c, please make sure that you do not revert the changes made by the below commits: SHA-1: 49accdedf956f175041040e677163b7cbb746283 * MdeModulePkg/SdMmcPciHcDxe: Hook SwitchClockFreq after SD clock start SHA-1: c67617f3c677c342efde780e229f841f4e0f6c7e * MdeModulePkg/SdMmcPciHcDxe: Add retries for sync commands SHA-1: 6d387610855ee27b49cb720de29ed809ea59dadc * MdeModulePkg/SdMmcPciHcDxe: Read response on command completion SHA-1: 63fd7f389874081ea37452b1f"
+    },
+    {
+      "prediction": "I don't think this is a valid test, since there is no `input` element with `type=\"text\"`.",
+      "reference": "You'll also need to add `webelem.Group.all` everywhere as that matches as well"
+    },
+    {
+      "prediction": "I think this should be `if ($(\"input:radio[name='ncr_proposal[expense_type]']:checked').val() == 'BA80') {` instead of `if ($(\"input:radio[name='ncr_proposal[expense_type]']:checked\").val() == 'BA80') {`",
+      "reference": "Does this need to be an ERB template?"
+    },
+    {
+      "prediction": "I think we can remove this file and just use the `SubmitFeedback` method in the `service/feedback.go` file.",
+      "reference": "do we want to register this in the mock server for testing?"
+    },
+    {
+      "prediction": "I think we need to make sure that we don't generate more than 1000 records here.",
+      "reference": "shouldnt this be `0:996` (+4) so the 1000s below fit?"
+    },
+    {
+      "prediction": "I think it would be better to use `MultiDimensionalMapping._sorted = self.enabled` and `MultiDimensionalMapping.sort = self._enabled` instead of `MultiDimensionalMapping.sort = self.enabled` and `MultiDimensionalMapping._sorted = self.enabled`",
+      "reference": "The docstring of this context_manager should be updated. As now ``sort=False`` is valid, it should just say it disables sorting regardless of whether the NdMapping has ``sort=True`` or ``sort=False``. I also think the line 'Should only be used if values are guaranteed to be sorted before or after the operation is performed.' should just say something else - maybe just that the initial ordering (whatever it is) should be preserved?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `spyAll` calls for `Receiver.prototype` since we don't need them anymore.",
+      "reference": "This tests failed unless I swapped the order here. Perhaps some new class transform broke the old form?"
+    },
+    {
+      "prediction": "This should be `if not cls.debug_java_port:`",
+      "reference": "extract the port and set to `debug_java_port`"
+    },
+    {
+      "prediction": "I think this should be `if not os.path.isabs(sourcefile):`",
+      "reference": "When the argument list is `[\"one\", \"two three\"]` then you concatenate it as you do it here, you won't be able to get the original list back with split. You need to annotate the list items better. (Or if you won't do the split by yourself, the called shell will do it. So you need shell escaping. How portable is that?) The main driver to change this in `intercept-build` was to get rid of these concatenation and splitting errors. The current compilation database parser in Clang understands the `arguments` filed... So the question, why don't you pass the command around as a list instead of a string?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think we should move it to the `webserver` package since it is only used there.",
+      "reference": "`ctx` is unused in EnablePiped"
+    },
+    {
+      "prediction": "I think this should be `self.authenticated = authenticated and user is not None` instead of `self.authenticated = authenticated or user is not None`",
+      "reference": "Why do we have both a list of valid sessions and an instance variable if a session is destroyed?"
+    },
+    {
+      "prediction": "I think this should be `var flushcount map[basics.Address]int = make(map[basics.Address]int)`",
+      "reference": "nit : I think that it would be \"cleaner\" to set the `flushcount` to nil in case of an error nice catch ;-)"
+    },
+    {
+      "prediction": "I think this test should be renamed to `assertNoApplicationsInNullRepository`",
+      "reference": "You may want to use assertNull"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to get the timestamp from the `Span` object itself.",
+      "reference": "Break out tests for events into their own tests rather than placing them in other ones. That makes it easier for us to figure out where problems lie and to do a TDD-driven implementation over new APIs."
+    },
+    {
+      "prediction": "Why is this removed? It seems to be the only place where `wlr_output->scale` is used.",
+      "reference": "Why did you remove this?"
+    },
+    {
+      "prediction": "I think this should be `log.Warnf` instead of `log.Infof`.",
+      "reference": "Why we need another error printouts? We already have it inside addMapping.."
+    },
+    {
+      "prediction": "This should be `location_map[*location]` instead of `location_map[location].output`",
+      "reference": "This LGTM, but I'm curious if this fixed a specific error you were hitting?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we need to make sure that we don't have any empty entries in the cgroup map. If we don't have any empty entries in the cgroup map, then we don't need to do anything. If we do have any empty entries in the cgroup map, then we need to make sure that we don't have any empty entries in the cgroup map.",
+      "reference": "Actually we rely on this functionality in cgroup v2, where the subsystem is empty."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "suggestion: use `core` package name"
+    },
+    {
+      "prediction": "This should be `Optional[int]`, not `Optional[Union[int, float]]`.",
+      "reference": "I think we can entirely remove this `limit` parameter for now to be consistent with other APIs."
+    },
+    {
+      "prediction": "I think this should be `start_moto_server('resource-groups', int(port), name='Resource Groups Tagging API')` and not `start_moto_server('resource-groups', int(port), name='Resource Groups Tagging API', asynchronous=False, update_listener=None)`",
+      "reference": "nit: `Resource Groups Tagging API` -> `Resource Groups API`"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "This type of list + delete always make me a bit uneasy but guess there's no atomic way to do this...... Could you put a todo to retry the \"ClearRepository + delete repo\" flow a few times? Basically, imagine a new image is added right after we call `ListImages`, then `DeleteImages` will not delete that newly added image. Then the delete repo call will fail"
+    },
+    {
+      "prediction": "I think this should be `import * as scrollManager from 'scrollManager'`",
+      "reference": "You can remove this since ES6 modules are strict by default. And thanks for contributing to the Jellyfin Project."
+    },
+    {
+      "prediction": "58578063a02b161e14610066578063dbdf0831461007757005b610060610161565b60006000f35b6100716004356100d4565b60006000f35b61008560043560243561008b565b60006000f35b600054600160a060020a031632600160a060020a031614156100ac57610058578063a02b161e14610066578063dbdf08314610077570",
+      "reference": "Why did eip1559 go from a list (0xf9020280.....) to a wrapped string (0xb902060ff9020280...) and not just concatenation (0x0ff9020280...)? implementation detail or is this how it sits on the wire now?"
+    },
+    {
+      "prediction": "I think this should be `!strings.Contains(err.Error(), \"configs/include_bad_conf_check_b.conf:10:19: unknown field \\\"monitoring_port\\\"\")`",
+      "reference": "@wallyqs Since on Windows it would be `\\` instead of `/` I just look at the suffix past `configs/`. Let me know if that's ok or not."
+    },
+    {
+      "prediction": "I think this should be `model.InsightSampleStream` instead of `*model.InsightSampleStream`",
+      "reference": "nit: Remove this assignment because it is not necessary."
+    },
+    {
+      "prediction": "I think this should be `ot_file_mapat_bytes (repo->repo_dir_fd, \"summary\", error, 0);`",
+      "reference": "This seems like a common enough pattern to offer an equivalent wrapper in `ot-fs-util.c`, no?"
+    },
+    {
+      "prediction": "I think this should be `media.hasFile()` instead of `media.fileExists()`",
+      "reference": "Wouldn't this crash the app if the user is currently listening to a stream?"
+    },
+    {
+      "prediction": "I think this should be `if prefix != \"\" && !strings.HasSuffix(prefix, \"/\") {`",
+      "reference": "This looks like an unrelated change? What is it for?"
+    },
+    {
+      "prediction": "I think this should be called `isDone` instead of `IsDone`",
+      "reference": "Wait, what is this doing? Is this a way to create methods that are only accessible from tests?"
+    },
+    {
+      "prediction": "I think this is not the right place to do this. I think it should be in the `archive` package.",
+      "reference": "Is it more reasonable to return some errors here, such as returning error codes (403 and so on?) ?"
+    },
+    {
+      "prediction": "I think we can remove the `_lastDomChild` property entirely.",
+      "reference": "Thinking out loud... would `_nextDom` be a better name for this?"
+    },
+    {
+      "prediction": "I think this is the only test that needs to be updated. The rest of the tests pass locally.",
+      "reference": "Turns out the unittests also had a false positive."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this. I think it should be `Int64` instead of `long` since it's an `Int64` and not a `long`.",
+      "reference": "I think these method names would be clearer if written as \"Create a {type} counter|gauge|measure with given name\". eg \"Create a int64 counter with given name\" \"Create a double gauge with given name\""
+    },
+    {
+      "prediction": "`getNumber<uint32_t>(L, -1)` should be `getNumber<uint32_t>(L, -2)`.",
+      "reference": "it pushes a boolean, so it has to be explicitly casted to boolean with != nullptr, same with all other similar stuff"
+    },
+    {
+      "prediction": "I think this should be done in the `createPublisher` method, not here.",
+      "reference": "It would be nice to have an equivalent of this call on `DataWriterImpl::enable`"
+    },
+    {
+      "prediction": "I think this should be `Status: model.ChainBlockStatus_DEPLOYMENT_BLOCK_RUNNING`",
+      "reference": "Isn't this PENDING?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the same place as the `aria.labelVirtual` function, since the `aria.label` and `aria.labelVirtual` functions are both called from the same place.",
+      "reference": "minor detail, why aim to sanitize title if tile is empty (in some cases)? worth adding an && to check for that."
+    },
+    {
+      "prediction": "I think you can remove the `QuerySpec` class as well, it's not used anymore.",
+      "reference": "Code shared by ParentChildrenSyncDownTarget and ParentChildrenSyncUpTarget moved to ParentChildrenSyncTargetHelper"
+    },
+    {
+      "prediction": "I think this should be `if not self.available_slots or self.available_slots == sys.maxsize:`",
+      "reference": "The bug is not about the priority , But we use 2 config items CAPACITY and SEQUENTIAL to control 1 action: thoughput. If 'sequential' is given to 'False' as default in configfile and then I use 'capacity' to 10 in my test.yml , the 'capacity' 's priority should be higher than default 'sequential'. So, I suggest we use only one 'capacity', it is enough. '1' is sequential, more than 1 is capacity, None is sys.maxsize."
+    },
+    {
+      "prediction": "I think this test should be in `integration_test.go`",
+      "reference": "I don't know if I'd call any of these FAST tests integration tests. They are unit tests for FAST. They use a mock plugin which doesn't actually start any external processes, etc."
+    },
+    {
+      "prediction": "I think this should be `CurrentVersion` instead of `Version`",
+      "reference": "Can you provide corresponding UT for this."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "nit: this should be below the yarpc group"
+    },
+    {
+      "prediction": "I think this should be `prometheus.ErrAlreadyRegistered`.",
+      "reference": "singleCaseSwitch: should rewrite switch statement to if statement (from `gocritic`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "Please log once here at Error level \"Failed to connect to Typha, will retry...\""
+    },
+    {
+      "prediction": "I think we should be able to remove this `runBpfUnitTest` function altogether and just call `runBpfTest` directly.",
+      "reference": "Do we need the forXDP parameter in runBpfUnitTest? If not, I think better to revert in order to save a few changes."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it would be better to do this in the `__init__` method of the `NameGenerator` class, since that's where the class is instantiated.",
+      "reference": "So we fix it only for kinto and not for all cliquet resources?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "No empty line between"
+    },
+    {
+      "prediction": "I think this should be done in `reviseRootDir` and not in `main`.",
+      "reference": "We have `ResolveRootfs` in `libcontainer/utils` so maybe use that one here?"
+    },
+    {
+      "prediction": "I think we need to add a test for the `long-running-operation-extension-2.json` file as well.",
+      "reference": "Just curious, do we have positive test covered somewhere?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it should be `roots_surface->pending_move_resize_configure_serial = roots_surface->pending_move_resize_configure_serial == 0 ? 0 : roots_surface->pending_move_resize_configure_serial + 1;`",
+      "reference": "Style error, put a space between `if` and `(`"
+    },
+    {
+      "prediction": "I think we should rename this to `EndpointHostPrefix` to be consistent with `EndpointHost` and `EndpointHosts`.",
+      "reference": "It's a bit odd to have a \"disable<x>\" boolean, though it looks like you already do that for a few other flags."
+    },
+    {
+      "prediction": "I'm not sure this is the right timeout. I think it should be the same as the `pullImageTimeout` above.",
+      "reference": "Why did you pick two hours for this?"
+    },
+    {
+      "prediction": "I think it would be better to use `ReadPassphrase` instead of `ReadPassword` here.",
+      "reference": "What about \"enter new password\"? That would be a bit more consistent to the `ReadPasswordTwice` usage in cmd_init."
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "Another redundant import"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just close the stream when we're done with it.",
+      "reference": "What kind of errors might cause us to reach this point? Could any of them be recoverable i.e. might work if we asked the same server again? Similar question for below, but perhaps with a slightly different answer"
+    },
+    {
+      "prediction": "Why not just use `oneapi::dal::preview::undirected_adjacency_vector_graph`?",
+      "reference": "What is the difference between GraphType and graph_type. Naming should be more accurate."
+    },
+    {
+      "prediction": "Why do you need this?",
+      "reference": "hmmm in docblock of contructor it is EntityManagerInterface....maybe it should be that way.... :confused:"
+    },
+    {
+      "prediction": "I think this should be `mol` instead of `self`",
+      "reference": "I'm curious why you went the route of adding a new method name for this. Given that you it would be possible to just have an an overload of `GetPos()`, and that this change breaks existing code, I wonder why you chose to go this way."
+    },
+    {
+      "prediction": "I think we can remove the `statusLabel` variable since we don't use it anywhere else in the component.",
+      "reference": "Does this need to change `'active'` to `'Active'` and that? Or is that done in CSS or something?"
+    },
+    {
+      "prediction": "I think this should be `MinAllowedMaxFrameSize` instead of `MinAllowedMaxFrameSize`",
+      "reference": "Why are these constants public? They should also be listed above members."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to check the v value. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction. I'm not sure if it's a good idea to check the v value of a private transaction.",
+      "reference": "This TODO isn't related to this change. We should remove it."
+    },
+    {
+      "prediction": "I think we can remove the `Stream<FieldMetrics>` method altogether and just use `FieldMetrics` directly in the `ValueWriter` implementation.",
+      "reference": "`FieldMetrics` is parameterized, but this is a bare reference. Could you update it? I think it should be `FieldMetrics<?>` since the metrics are not necessarily for the written value type, `D`."
+    },
+    {
+      "prediction": "I think we can remove the `SetProcessor` and `SetProcessor` methods from the `MeterProviderBuilder` class as well.",
+      "reference": "@cijothomas do we want this to be `Set` or `Add`?"
+    },
+    {
+      "prediction": "I think this should be 100,000,000.",
+      "reference": "Why are we dropping GasCap?"
+    },
+    {
+      "prediction": "I think we can remove the `throw new ArgumentNullException(nameof(span), \"Parameter cannot be null\")` from the `SpanShim` constructor.",
+      "reference": "should we end the text with a .?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I don't think we should be changing the semantics of `dispatch_enter_dynamorio()`. I think we should be changing the semantics of `dispatch_exit_dynamorio()`. I think we should be changing the semantics of `dispatch_enter_dynamorio()`. I think we should be changing the semantics of `dispatch_exit_dynamorio()`. I think we should be changing the semantics of `dispatch_enter_dynamorio()`. I think we should be changing the semantics of `dispatch_exit_dynamorio()`. I think we should be changing the semantics of `dispatch_enter_dynamorio()`. I think we should be changing the semantics of `dispatch_exit_dynamorio()`. I think we should be changing the semantics of `dispatch_enter_dynamorio()`. I think we should be changing the semantics of `dispatch_exit_dynamorio()`. I think we should be changing the semantics of `dispatch_enter_dynamorio()`. I think we should be changing the semantics of `dispatch_exit_dynamorio()`. I think we should be changing the semantics of `dispatch_enter_dynamorio()`. I think we should be changing the semantics of `dispatch_exit_dynamorio()`. I think we should be changing the semantics",
+      "reference": "Could this instead keep the `go_native` and add to it \"or the last exit was the special reset exit\"?"
+    },
+    {
+      "prediction": "This should be `connect(this, &ConfigPanelDialog::reset, mPluginsPage, &ConfigPluginsWidget::reset);`",
+      "reference": "Oh, I missed this one: The other instances of `this` you've added as lambda contexts aren't really needed, although they're harmless. However, in the above connection, the missing context is `panel`, not `this`. Please correct it! Clazy may show warnings about lambda contexts (I don't use Clazy) but, if so, that's a problem in clazy: it isn't smart enough to know about them."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. We should kill the process before trying to wait for it to terminate.",
+      "reference": "can we add a try / except around this to make it more stable? I like the idea of any mistakenly un-quit drivers closing down after the process is quit (the java server does this... also i forget to do driver.quit often when i use the command line repl :) )"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do here. I think we should check if the request body is empty before decoding it. If it is empty, we should return an error. Otherwise, we should proceed with decoding the request body and passing it to the `CloudMigrateStatus` function.",
+      "reference": "when would this happen? backward compatibility?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. We should be using the `tagName` property of the `node` object.",
+      "reference": "I'm not sure what the value of this is. So far we've solved this by always doing `tagName.toUpperCase()` for everything. I think we should stick with this."
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `--volname <vol>`.",
+      "reference": "This command displays available snapshots on a volume."
+    },
+    {
+      "prediction": "I think we should rename this proto to `storage_go_proto` instead of `storage_go_proto`",
+      "reference": "`spb` is the standard package name we use for this proto."
+    },
+    {
+      "prediction": "I think this should be `handleName` instead of `handle: string`.",
+      "reference": "/javascript/firefox-driver is the Selenium implementation of a WebDriver for Firefox. Since it generally isn't W3C compatible, it shouldn't change. We can just drop this change."
+    },
+    {
+      "prediction": "I don't think we need to upgrade to parallel process.",
+      "reference": "is this pr support parallelly process application?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check if the user can enable pull requests for this repo. I think it would be better to check if the user has permission to view the repo's pull request settings.",
+      "reference": "Based on the assumption that when `EnablePulls` is `true`, `CanEnablePulls` must be `true` as well, then this `if` check is redundant. Actually... we have `AllowsPulls` now... why not use that?"
+    },
+    {
+      "prediction": "I think it would be better to use `org.apache.azkaban.model.Flow` instead of `org.apache.azkaban.model.Job`",
+      "reference": "I am not sure why the API requires both project name and project ID as inputs. Wouldn't it introduce more opportunities for clients to make mistakes?"
+    },
+    {
+      "prediction": "I'm not sure if we want this to be a flash message or an alert. If it's a flash message, I think it would be better to use `flash[:notice]` instead of `flash[:warning]`. If it's an alert, I think it would be better to use `flash[:alert]` instead of `flash[:warning]`.",
+      "reference": "I believe this was happening for several different flash messages - should we add `now` to all flash messages? (there might be a downside to doing that, but I am not sure what it would be)"
+    },
+    {
+      "prediction": "Why do you need this?",
+      "reference": "There is no docblock on the parent method. Also, not related, but I forgot to typehint the `$templating` constructor argument above."
+    },
+    {
+      "prediction": "I think this test should be moved to the `xdebug` test suite.",
+      "reference": "Silly me. This shouldn't be done in the tests, but instead in the actual definition of ValidPHPVersions on arm64. Same with Mysql and MariaDB."
+    },
+    {
+      "prediction": "I think this should be `queryRelation == QueryRelation.INTERSECTS || queryRelation == QueryRelation.CONTAINS`.",
+      "reference": "Shouldn't this work as well for Disjoint?"
+    },
+    {
+      "prediction": "I think this should be `o.Status.Reason = fmt.Sprintf(\"Failed to retrieve Order resource: %v\", errUpdate)`",
+      "reference": "You've not updated the `o.Status.Reason = fmt.Sprintf(\"Failed to retrieve Order resource: %v\", err)` line below here when you changed this, so here we are checking `errUpdate` but will print the contents of `err` instead."
+    },
+    {
+      "prediction": "I don't think we need to check for `Errno::ECONNREFUSED` since it's the most commonly used error code.",
+      "reference": "Doesn't this need to include `Errno::ECONNRESET` to fix the issue?"
+    },
+    {
+      "prediction": "Shouldn't this be `azkaban.project_cache_throttle_percentage`?",
+      "reference": "Please consider keeping the config key in one line as it will help in case someone is looking at how this config is used. public static final String PROJECT_CACHE_THROTTLE_PERCENTAGE = \"azkaban.project_cache_throttle_percentage\";"
+    },
+    {
+      "prediction": "I think this should be `ListPowerSelectorEntries` instead of `ListSelectorEntries`",
+      "reference": "This is a hard one, naming-wise... `ListPowerSelectorEntries` is logical given `ListSelectorEntries`. Another option could be `ListMatchingEntries` or even `FindMatchingEntries`. :shrug:"
+    },
+    {
+      "prediction": "I think this should be `types.UVarint` and not `types.UVarintType`",
+      "reference": "This should be called `Uint` or `Uint32` or simply using our `Uint64` type instead. All of our numbers have varint encoding. It's not inherently a varint."
+    },
+    {
+      "prediction": "I don't think we should add a new argument to `view_source` here.",
+      "reference": "No need for this when you don't need to customize anything about the argument."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `app/views/org_admin/sections/_section.html.erb`",
+      "reference": "We're now going to have 3 `#sections_accordion` will need to see how JS behaves. We should probably switch to a class selector instead"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Use 2 (not 4) spaces for indentation.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "did you mean to leave this in here?"
+    },
+    {
+      "prediction": "I think this should be `Whether to generalize image using Windows Sysprep or UEFI.`",
+      "reference": "Mention that it only applies to Windows. This is kind of implied, but better to be explicit."
+    },
+    {
+      "prediction": "I think this should be `uint64(100)`",
+      "reference": "again: it is a bad practice to have some global parameters with such common names for unit test purpose."
+    },
+    {
+      "prediction": "Why do we need this change?",
+      "reference": "Should we use `\\DateTimeInterface` or `DateTimeInterface` and declaration of `DateTimeInterface` in `use`?"
+    },
+    {
+      "prediction": "nit: I think this should be `task-` instead of `task-suffix`",
+      "reference": "What do you think of moving this stack related constant to the `stack` pkg?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this error. I think it should be in the `TxnPool` struct.",
+      "reference": "Hm, I guess that you're doing this here since `WellFormed` is on a `transactions.Transaction` and not a `transactions.SignedTxn`, but quickly grepping through our code, it looks like we always a `SignedTxn` around when calling `WellFormed` (except maybe some tests?)... this doesn't have to happen here, but maybe we should update that in a separate PR since this really feels like a `WellFormed` check."
+    },
+    {
+      "prediction": "I think this should be `padding: 0, margin-bottom: 20px;`",
+      "reference": "This story also needs the default padding."
+    },
+    {
+      "prediction": "I don't think this change is needed.",
+      "reference": "Hmm why did we remove the randomness about which (confusingly) different exception to throw here? This randomness was (is?) useful to help test that Lucene indeed catches `FNFE` and `NSFE` interchangeably."
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "Hm, so before in that case `current_collection_id` looks like it was actually an uninitialized pointer. It seems weird to me that the tests pass both ways. I haven't looked closely at the tests - is there something wrong there that causes this to spuriously pass before?"
+    },
+    {
+      "prediction": "I think it would be better to have a single enum for the kinds of symbols, like `K_SYMBOL`, `K_LITERAL`, `K_CONST`, etc.",
+      "reference": "s it the best way to tag \"foo\" of \"`define foo ...\" with constant kind? Introducing \"d/definition\" kind and tagging the \"foo\" with the definition kind is an alternative way. Using the unified kind \"constant\" in many areas reduces the information that client tools get from a tags file. In my experience, it is better to tag different things with different kinds. If ctags hides the differences, and just reports various language objects as \"constant,\" in some cases, a client tool must parse the raw .sv source file for distinguishing whether a given tag is \"`define\"'ed or not. IMHO, ctags should not be too smart. Instead, ctags should be stupid; it should report what it sees in source input files as-is to client tools that will do something smart."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `config.js`",
+      "reference": "is this a breaking change? If I'm not mistaken, this is translating a user-provided `wtimeout` in the connection string into `wTimeoutMS`"
+    },
+    {
+      "prediction": "I don't think we need this file. We already have a `lambda_function.py` that does the same thing.",
+      "reference": "Can we rename this file to `lambda_echo.py`, to be more descriptive and in line with the other files in this directory? (all files in this folder are lambda functions, after all)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `security` since it's related to security.",
+      "reference": "why not do it like this ? <pre><code> const ( Kill Action = iota Errno Trap Allow Trace ) <pre><code>"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "lock(this) is a bad practise, since you lock the entire instance. cosider: var appenderCache = xxxx. lock(appenderCache) { Thread.Sleep(TimeSpan.FromHours(5)); } while inside AppenderCache instance lock(this) { is blocked 5 hours. } Its always better to make a class private lock objects."
+    },
+    {
+      "prediction": "I think we should use `grpcClientConn` instead of `grpcheader.NewContextWrapper().WithCaller(serviceName + \"-client\").WithService(serviceName)`.",
+      "reference": "should we use the constant above?"
+    },
+    {
+      "prediction": "I'm not sure about this change. It seems to me that this should be a separate commit.",
+      "reference": "I think something like `size_res` would be more descriptive."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Bolt::Event` class, since that's where the event metrics are stored.",
+      "reference": "Oh lol so this was...really more of a bug?"
+    },
+    {
+      "prediction": "I think we can remove the `System.Diagnostics.CodeAnalysis.SuppressMessage` and `System.Diagnostics.CodeAnalysis.JoinMessage` from this file as well.",
+      "reference": "Well, it is providing a platform neutral way of calling Task or TaskEx. We could just call it PlatformTask."
+    },
+    {
+      "prediction": "This should be `xwayland->server = wlr_xwayland_server_create(wl_display, &options);`",
+      "reference": "Typo: this should free `xwayland` instead"
+    },
+    {
+      "prediction": "I think we should use `crypto/rand` instead of `math/rand` here.",
+      "reference": "Is it ok to use \"math/rand\" instead of \"crypto/rand\" here? Do we need to protect this ID somehow?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `github.com/cihub/seelog`",
+      "reference": "Why did we have `logxi` in the first place? Maybe it makes sense to remove it from `glide.lock? It will safe us in the future from using logger which is not working."
+    },
+    {
+      "prediction": "I think it would be better to have a `UNF` field in the `DataCitation` object itself, rather than having a separate `getUNF` method. This would make it easier to change the `getUNF` method in the future without having to change the `DataCitation` object.",
+      "reference": "Is there a reason to use `getUNF()` and `UNF` in the same block?"
+    },
+    {
+      "prediction": "I think this should be `and \"b\" !~ /c`",
+      "reference": "Again this seems odd that megacheck cares"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `Transport` class, since it's only used there.",
+      "reference": "I assume this will resolve to nil if this isn't in the config?"
+    },
+    {
+      "prediction": "I think we can remove the `allowOptionalMappings` parameter from this method.",
+      "reference": "I looks like we can make `allowOptionalMappings` non-optional since we have another overload without it. (I thought this generated a compiler warning, but maybe it's just R#)"
+    },
+    {
+      "prediction": "`conv_cfg` -> `conv_cfgs`",
+      "reference": "` per scale.` -> ` per scale.` need to clean the unnecessary blank space"
+    },
+    {
+      "prediction": "This should be `@Restricted(NoExternalUse.class)`",
+      "reference": "there's no import for Export"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `ListenAndServeTLS` function.",
+      "reference": "i don't think we want/need to make the node API available over UDS"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "How about to skip if and returning null and return a result of `ArticleFacade::findById` directly?"
+    },
+    {
+      "prediction": "`push` is not a method on `Array`, it's a method on `Object.prototype`. You can use `Array.prototype#push` instead.",
+      "reference": "I think this change is currently unnecessary. I like the idea of `map()`, `filter()`, etc. returning vanilla JavaScript arrays rather than mongoose arrays, but I think that's a bit too heavy of a change for a patch release. I think that all we need for this release is the `if (!parent || parent.$__.populated == null) return;` change."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `Travis::Services` since it's only used by that class.",
+      "reference": "how about \"unknown\" instead of \"fafafaf\"?"
+    },
+    {
+      "prediction": "I think we can remove this interface and just use `func (a *Actor) GetSignature(ctx context.Context, actorAddr address.Address, method string) (_ *exec.FunctionSignature, err error)`",
+      "reference": "nit: do we gain anything from naming the output error `err` in this definition?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `inputManager.on(window, onInputCommand)` line above.",
+      "reference": "This can happen with any dialog. Maybe the best place to add this change would be the dialog component instead."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to dump the environment. I think it should be in the `beforeEach` of the test.",
+      "reference": "Should this be `TearDown` since it now calls this method? Something I would like to see is, on test failure don't teardown completely instead leave the FAST repo in place with the stderr and stdout files and direct the user to that location. What do you think?"
+    },
+    {
+      "prediction": "I think negLL is not a real metric, so it should be removed.",
+      "reference": "There's not particular reason to use a short name. Can we use a more descriptive name for the metric? Perhaps \"negativeLogLikelihood\"?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `skipArgumentList` and not in `parseFunction`.",
+      "reference": "not handling in case of `repr` makes the behavior probably a little too unpredictable. Is there a reason not to, apart missing stuff in the `repr`? Does it lead to some problem?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that we don't advance the input when an exception is thrown. I think we need to make sure that we only advance the input if the exception is not due to an unhandled exception.",
+      "reference": "Remove the finally?"
+    },
+    {
+      "prediction": "I think this should be `WCHAR *` instead of `const TCHAR *`.",
+      "reference": "Build on AArch64 fails because WCHAR is not defined."
+    },
+    {
+      "prediction": "nit: you don't need the space around `:=` here.",
+      "reference": "Do we use this const?"
+    },
+    {
+      "prediction": "I think this should be `SetAnnotations` and not `SetCertificateRequestFailureTime`",
+      "reference": "I think this may need to be `SetCertificateRequestAnnotations` as it returns a `CertificateRequestModifier`"
+    },
+    {
+      "prediction": "Style/ClassVars: Replace class var identifier_with_with with the explicit identifier_with variable name.",
+      "reference": "it would likely be better to do this in the Identifier.by_scheme_name method itself."
+    },
+    {
+      "prediction": "I think this should be `created_at`",
+      "reference": "@janette looking at this test code, I'm pretty sure that it's going to give a false positive. Where is it actually checking the contents of the first four datasets against expected values? All the assertion at the end of the test seems to check is that at least four datasets exist."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should go in `app/_ehcache.js` or something like that.",
+      "reference": "Would this break things if Ghost is run in a subdirectory or no?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle the case where the key is `id` but the property name doesn't start with `prop_`. I think it would be better to check if the key starts with `prop_` and then use the `prop_` as the property name, otherwise use `prop_` as the property name. WDYT?",
+      "reference": "maybe switch the order in order to make this a positive assertion instead of a negative one? i.e. `if key =~ /id/` first."
+    },
+    {
+      "prediction": "I don't think we should be using `import` statements in our code. We should be using `from ... import ...` or `from ... import ... as ...` or `from ... import *` or `from ... import * as ...` or `from . import *` or `from . import * as ...` or `from .. import *` or `from .. import * as ...` or `from ... import *` or `from ... import * as ...` or `from .. import *` or `from .. import * as ...` or `from ... import *` or `from ... import * as ...` or `from .. import *` or `from .. import * as ...` or `from ... import *` or `from ... import * as ...` or `from .. import *` or `from .. import * as ...` or `from ... import *` or `from ... import * as ...` or `from .. import *` or `from .. import * as ...` or `from ... import *` or `from ... import * as ...` or `from .. import *` or `from .. import * as ...` or `from ... import *` or `from ... import *",
+      "reference": "Is this an oversight for some code in the body?"
+    },
+    {
+      "prediction": "I think you can use `putExtra(\"state\", mProduct, this);` instead.",
+      "reference": "Reason for adding this? I couldn't find where you are retrieving it back."
+    },
+    {
+      "prediction": "I think this should be `h2o_mruby_eval_expr(mrb, \"require \\\"preloads.rb\\\"\");`",
+      "reference": "Can we expect adequate information emitted to the log in case either requiring preloads.rb or doing the requires in preloads.rb fails? If the answer is yes, I think we can merge this PR right away."
+    },
+    {
+      "prediction": "Layout/ExtraSpacing: Unnecessary spacing detected.<br>Layout/SpaceBeforeFirstArg: Put one space between the method name and the first argument.",
+      "reference": "How about a `app/views/layouts/pages.html.erb` layout that contains the HTML head, body, wrappers and yield's the `new-topics` template into it? I think we might be able to delete the `app/controllers/pages_controller.rb` file at that point."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think you should check if `this.$__.selected` is defined before calling `this.$__.selected._id`.",
+      "reference": "This is a coarse solution. This check is helpful, but you should also add a check in `$__version()` to avoid calling `isSelected()` if `key === false`"
+    },
+    {
+      "prediction": "I'm not sure if this should be `ApplicationGitPath` or `GitPath`. I think it should be `ApplicationGitPath` because that's what it's actually used for.",
+      "reference": "Passing only needed data to make it can be reused by `planpreview` package where there is no deployment data."
+    },
+    {
+      "prediction": "I think we should rename this to `CustomAuth` instead of `Auth`",
+      "reference": "Could we rename those to `CustomClientAuthentication` (same for Router) to remove ambiguity between Authentication and Authorization (permissions)?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should be in `MvvmCross.Core.Views`",
+      "reference": "Is the name \"IMvxFormsViewPresenter\" too close to \"IMvxFormsPagePresenter\"? Perhaps something like \"IMvxFormsNativeViewPresenter\" or \"IMvxFormsPlatformViewPresenter\" or inline with Forms naming \"IMvxFormsOnPlatformViewPresenter\""
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. The problem is that `child` can be `null` or `undefined`, which is going to crash the browser. I think you need to check if `child` is truthy before you assign it to `children`.",
+      "reference": "`typeof child != 'boolean'` maybe?"
+    },
+    {
+      "prediction": "I think this should be `!rawErrorData.containsKey(CLASS) && !rawErrorData.containsKey(STACK_TRACE)`",
+      "reference": "Why && changed to || ?"
+    },
+    {
+      "prediction": "I think this is the only place where `InitAddress` is used.",
+      "reference": "nit: not a big deal right now but spec assigns ID 0 to InitAddress"
+    },
+    {
+      "prediction": "I think this should be `handleName` instead of `handle: string`.",
+      "reference": "/javascript/firefox-driver is the Selenium implementation of a WebDriver for Firefox. Since it generally isn't W3C compatible, it shouldn't change. We can just drop this change."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to set the zoom factor directly in the `WebView` constructor. This way we don't need to call `super().setZoomFactor(fact)` or `self._default_zoom_changed = True` in the `WebView` constructor.",
+      "reference": "There are various places where `setZoomFactor` is used but `fuzzyval` isn't set: - `__init__` and `on_config_changed` (should be okay as `init_neighborlist` gets called which essentially does the same) - `zoom` (calls `zoom_perc` with `fuzzyval=False`), which is used by `:zoom-in` and `:zoom-out`. Are you sure this won't break those calls? (Yes, I know there should be tests for this - I'm working on that.) Other existing places probably should get simplified."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to pass in an `IList<Annotation> annotations` instead of `bool addAnnotations = true`.",
+      "reference": "Is there a benefit to keeping this in RefinementSettings? Or could we move it into its own class in TestUtil?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `query` package.",
+      "reference": "I would have expected the Ready method to need to change to be a `<-chan Result` instead of a `<-chan map[string]Result`. The difference being that now the Ready channel can produce more than one set of results. In fact the name Ready is inaccurate now since its not about the query being ready but just a mechanism to deliver results. If this should be a follow up PR that is fine, lets create an issue for it."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it should be moved to the `ProjectManager` interface.",
+      "reference": "why? what if we want to access those methods outside of azkaban-common?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `MergeTestBlockchain` class.",
+      "reference": "Rename file too"
+    },
+    {
+      "prediction": "I think this should be `awserr.New(r.HTTPResponse.StatusCode, \"\", err)` instead of `awserr.New(r.HTTPResponse.StatusCode, \"\", \"\")`",
+      "reference": "Need to populate the `RequestID` field for these errors. This can be obtained from r.RequestID i think, but need to make sure. The `r.RequestID` should of been populated from the `UnmarshalMeta` handler list."
+    },
+    {
+      "prediction": "I think this should be `private static final boolean TEST_IMPLEMENTS_DELETE = true;` and `private static final boolean TEST_IMPLEMENTS_CANCEL = true;` instead of `private static boolean TEST_IMPLEMENTS_DELETE = false;` and `private static boolean TEST_IMPLEMENTS_CANCEL = false;`",
+      "reference": "Why is the `final` being removed here?"
+    },
+    {
+      "prediction": "@abumusamq can you update the copyright year?",
+      "reference": "Maybe let's take this opportunity to change easylogging to a git submodule?"
+    },
+    {
+      "prediction": "I think this should be `residueWindows` instead of `residueNewline`",
+      "reference": "Should we just be generous and elide all extra whitespace? `strings.TrimSpace`"
+    },
+    {
+      "prediction": "This method is no longer used, can we remove it?",
+      "reference": "This is an inaccurate API that has outlived it's purpose. We can't make the determination of phone (vs) tablet, with a 7\" screen in the mix. Also, with the advent of fragments, this API means very little now. It's not being used anywhere, since we switched to `ActionBar`."
+    },
+    {
+      "prediction": "`log.info(version());` should be `log.finest(version());`",
+      "reference": "This change means that users can't easily see which version of the selenium server they're using. This is `info` level information."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be in `auth/__init__.py`",
+      "reference": "You should just use `with patch(...):`, so it unpatches it automatically. Also, much simpler."
+    },
+    {
+      "prediction": "I think we can remove the `protected IMvxViewsContainer ViewsContainer { get { return _viewsContainer; } }` from the `IMvxViewDispatcher` interface.",
+      "reference": "This didn't use to work for me. Are you sure that returns the correct instance from startup?"
+    },
+    {
+      "prediction": "I don't think we should change the signature of `MethodCallExpr` to allow for a `name` argument. I think we should keep it as `Expression` and add a `@Nullable` annotation to `name` to indicate that it might be `null`.",
+      "reference": "Looks good. Eventually we could remove some of these constructors, but for now adding these two seems the way to go"
+    },
+    {
+      "prediction": "I think this should be `[\"Need action\"]` instead of `[\"Need action\"]`",
+      "reference": "@jalavik should be have somewhere enum/list/registry of possible tags?"
+    },
+    {
+      "prediction": "I think we should be able to remove the logging too-many-args pylint disable.",
+      "reference": "why is this pylint disable needed?"
+    },
+    {
+      "prediction": "This should be `true` by default.",
+      "reference": "The new sync feature is only being supported from within Terraform. Installations using the deprecated method will have this feature disabled."
+    },
+    {
+      "prediction": "I think it would be better to add `onnx_trace_repr` as well.",
+      "reference": "This type of return (f'xxx') is not recommended, it is recommended to return 'xxx'"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it should be done in the same place as the check for `ZMM_ENABLED()`. I think it",
+      "reference": "I would make this level 1 or 2."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Is that path in a variable anyway? Not a big fan of hard coding it here."
+    },
+    {
+      "prediction": "I think this should be `k8serrors` instead of `k8serrors`",
+      "reference": "nit: Move `IdentityTemplateLabel` to be under `IdentityTemplate` to match the struct in config_crd.go"
+    },
+    {
+      "prediction": "Style/MutableConstant: Freeze mutable objects assigned to constants.",
+      "reference": "Not necessarily related to this PR, but don't you think having a centralized place for all interactions with analytics might make this into a very big class? Is that something that you don't worry about until it happens?"
+    },
+    {
+      "prediction": "I think this should be `return Errf(\"step %q error: %s\", s.name, err)`",
+      "reference": "it seems that my terminal is dead when I press ctrl+c during running daisy cli today, is this PR fixing that?"
+    },
+    {
+      "prediction": "I think this should be `chainStore`, not `nc.Repo.Config().Mpool`",
+      "reference": "If we are going to add config to specify the maximum message pool size, we should probably also add a parameter for the maximum nonce gap and pass it into the `IngestionValidator`. This could be done in this PR or added as an issue."
+    },
+    {
+      "prediction": "I think we can remove this import as it is not used.",
+      "reference": "please inline imports"
+    },
+    {
+      "prediction": "I think this should be `0` by default.",
+      "reference": "Should IptablesPostWriteCheckIntervalSecs be set back to its previous smaller value, if use of the iptables lock is disabled?"
+    },
+    {
+      "prediction": "I think this should be `throw new Error('Unknown profile version. Most likely this is an old version of Joplin, while the profile was created by a newer version. Please upgrade Joplin at https://joplinapp.org and try again.' + p.name + ' version: ' + p.version + '\\nProfile version: ' + fromVersion + '\\nExpected version: ' + existingDatabaseVersions[existingDatabaseVersions.length - 1])`",
+      "reference": "Please use packageInfo here instead. As it is used in `app.js`."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the file doesn't exist, we can't use the cache.",
+      "reference": "this seems unrelated to the rest of the PR, although related to the original report"
+    },
+    {
+      "prediction": "I am not sure this is the right fix. I think we need to check if the column has a non-null default value. If it does, then we need to set the predicate to true. Otherwise, then we need to set the predicate to false.",
+      "reference": "Would a check for added columns with non-null default be sufficient for aligned format?"
+    },
+    {
+      "prediction": "I think this flag should be called `compute_service_account_name` to be consistent with other flags.",
+      "reference": "Is this PR implementing compute service account for OVF import as well? I thought it's only image/one-step import? Btw, OVF importer uses \"-\" instead of \"_\" in flags (should be `compute-service-account`)."
+    },
+    {
+      "prediction": "This should be `/collections/{collection}/elements/{product}/`",
+      "reference": "Why change to element ?? in whole application use plural convention ??"
+    },
+    {
+      "prediction": "I think this should be `<a tabindex=\"-1\" href=\"https://uppy.io\" rel=\"noreferrer noopener\" target=\"_blank\" class=\"uppy-Dashboard-poweredBy\">Powered by <svg aria-hidden=\"true\" class=\"UppyIcon uppy-Dashboard-poweredByIcon\" width=\"11\" height=\"11\" viewBox=\"0 0 11 11\" xmlns=\"http://www.w3.org/2000/svg\"><path d=\"M7.365 10.5l-.01-4.045h2.612L5.5.806l-4.467 5.65h2.604l.01 4.044h3.718z\" fill-rule=\"evenodd\" /></svg><span class=\"uppy-Dashboard-poweredByUppy\">Uppy</span></a>`",
+      "reference": "Tiniest nit but this can be `tabindex={-1}`, `width={11}`, `height={11}`"
+    },
+    {
+      "prediction": "I think this should be `assert type(input_image) == itk.Vector[output_pixel_type, 2]` instead of `assert type(output_image) == itk.Vector[output_pixel_type, 2]`",
+      "reference": "Unrelated to this PR, we should remove this exception. We now require `numpy`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the same place as `st.w.NewSubWorkflowFromFile`",
+      "reference": "Is the loop over vars and adding them to the subworkflow below in this file needed, since it's already done by NewSubWorkflowFromFile?"
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `core/pandas_kernel`.",
+      "reference": "Having to write this util makes me think that maybe we should have kept the csv_input stuff. I don't know."
+    },
+    {
+      "prediction": "I think this should be `q: newBlockRetrievalQueue(queueSize, config.Codec(), config.BlockCache())` instead of `q := newBlockRetrievalQueue(queueSize, config.Codec(), config.BlockCache())`",
+      "reference": "Passing in and saving a reference to the `BlockCache` at init time is going to break if something calls `config.ResetCaches()`, because it replaces the `BlockCache` instance completely. This happens on user logout or by a manual write to `.kbfs_reset_caches`. So you probably want to give it the whole `config`, or maybe just the `config.BlockCache` function pointer."
+    },
+    {
+      "prediction": "I think this should be `if err == multistream.ErrNotSupported || err == multistream.ErrIncorrectVersion {`",
+      "reference": "I am not sure this could happen. st should be nil if the error happens. If it does happen, we could also do st.Reset(), since it is an erroureous state so it closes both sides of the stream."
+    },
+    {
+      "prediction": "I think this should be `console.error(e.stack)` instead of `console.log(e.stack)`",
+      "reference": "I'm guessing this was just added for debugging? Should probably remove it so we aren't modifying 3rd party libs."
+    },
+    {
+      "prediction": "I think this should be `print_event(node, event) { print_result(node, event[:result]) }` and not `print_event(node, event) { print_result(node, event[:result]) }`",
+      "reference": "This seems reversed... if `success?` is true, wouldn't we use `success`?"
+    },
+    {
+      "prediction": "I'm not sure we should have this enum here. I think we can just use `TraceLevel` for this.",
+      "reference": "This seemed like the easiest way to allow log level checks, it will require an update to the vendors tool, to ignore that file on update."
+    },
+    {
+      "prediction": "Line is too long. [87/80]<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Line is too long. [87/80]<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Line is too long. [87/80]",
+      "reference": "Should we disallow entry authors to unsubscribe to their entries?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It's not part of NV Access, it's part of the NV Access framework.",
+      "reference": "2016 should be 2017."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check if the schema has changed. I think it would be better to check if the content of the schema file has changed.",
+      "reference": "boolean exists = (existSchema != null) && (existSchema .getSummary() != null);"
+    },
+    {
+      "prediction": "I think we can remove the `Run: func(_ *cobra.Command, args []string) {` portion of the function.",
+      "reference": "It does not look idiomatic."
+    },
+    {
+      "prediction": "This file should be renamed to `sql_test.go`.",
+      "reference": "Generally, we do not use newlines between project imports, we only separate them from the stdlib imports"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the keyboard steps. I think it would be better to move this logic to the `MediaSource` class, since that's where the RunTimeTicks value is set.",
+      "reference": "can this long line be wrapped?"
+    },
+    {
+      "prediction": "I think this should be `W0143: Comparing bare callable, might have skipped parenthesis after callable`",
+      "reference": "I'd use the same message that I suggested in the release changelog."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "Keep semi-colons at end of function defs please"
+    },
+    {
+      "prediction": "I think we can remove `this.readConcern = resolveReadConcern(parent, this.options);` and `this.writeConcern = resolveWriteConcern(parent, this.options);`",
+      "reference": "A little skeptical of `explain` being here. Why is this now a property of all commands?"
+    },
+    {
+      "prediction": "I think we can remove the copyright header from this file as well.",
+      "reference": "License copied two times!!"
+    },
+    {
+      "prediction": "I think it would be better to have this as a class argument instead of a kwarg.",
+      "reference": "... their **superclass**"
+    },
+    {
+      "prediction": "I don't think we should be setting this to 0 in the first place. I think we should be setting it to the value of `options['master-start-curl-retries'] || 0`.",
+      "reference": "I'm not sure if this is the right thing to be doing here. It seems like there should be a more systemic fix instead of this patch."
+    },
+    {
+      "prediction": "I think this should be `null` instead of `undefined`.",
+      "reference": "Not sure why these are up here   can we move these down to be part of the `@return` tag instead? This is what we usually do   (e.g. `getCurrentEntityURL`)"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to handle the case where the field is an array of strings, but it's the only way I could come up with that didn't break something.",
+      "reference": "Layout/EmptyLinesAroundBlockBody: Extra empty line detected at block body beginning."
+    },
+    {
+      "prediction": "I think this should be renamed to `NoSuchInterceptingElement` instead of `InterceptingElementException`.",
+      "reference": "We keep who wrote the code anonymous."
+    },
+    {
+      "prediction": "I think this should be `RegisterIdentity` and not `NodeRegister`",
+      "reference": "Mention something in logs, that fake registration was used e.g. `Fake identity registered`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this action. I think it should go in the `Property` reducer since it's related to that specific property.",
+      "reference": "I think it would make sense to also set `internalWebPropertyID` to `''` here. Otherwise, it might keep its previous value (let's say a property was selected before, but then the user chooses \"Create new property\"), but since it always should depend on `propertyID`, that would be incorrect."
+    },
+    {
+      "prediction": "I think this should be `std::unique_ptr<RWMol> m(std::unique_ptr<RWMol>(static_cast<RWMol *>(SmilesToMol(\"C=C1=CC=CC=C1\"))));`",
+      "reference": "It might be reasonable to add a check that sanitization still fails here."
+    },
+    {
+      "prediction": "I think it would be better to use `if (this.node.getAttempt() > 0)` instead of `if (this.node.getAttempt() > 0 && this.node.getAttempt() <= 0)`",
+      "reference": "Why not use attemptNo instead of another call to getAttempt? Even better, consider using a private method to calculate this string. This logic is in two places."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "Yuck. Rails bug."
+    },
+    {
+      "prediction": "I think this should be `return [l.__class__ for l in self.layers()]`",
+      "reference": "It's useally a common practice to not use `l` as a variable (we have disabled this test in scapy because our code has tons, but it would have PEP8 failed otherwise), but it would be better not to create any. In general, single letter variables are not liked that much. Especially with windows where the fonts make 1=l How about `lay` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `ChunkUploadHandler` in `server/server.go`.",
+      "reference": "DRY these 5 lines into a function call, no?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should log the error and exit gracefully, then close the channel.",
+      "reference": "should we be trying to close here? or just return directly @fabxc ? same goes for `stopc`..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in the `before` block of the `setup` block.",
+      "reference": "just re-ordered for clarity"
+    },
+    {
+      "prediction": "I think we can remove the error message here.",
+      "reference": "Can you fix this also to print what is the IP address used to connect. Rename --> \"M_API server\" to maya apiservice"
+    },
+    {
+      "prediction": "I think this should be `internal/bytealg.IndexByteString` instead of `internal/bytealg.IndexByte`",
+      "reference": "This fails on Go 1.11 because `indexByte` is already defined in strings_go111.go. I would suggest simply renaming one of the two, for example you could rename the other one to `strings_indexByte` (that will be one of the things we'll drop when we drop Go 1.11 support so it's better to make that one look ugly instead of the function that we will likely keep for a longer time)."
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `''`",
+      "reference": "Now `dbType=' MockDbType.int32 '` is correctly interpreted and the value is rendered as an integer instead of a string; hence no more double quotes."
+    },
+    {
+      "prediction": "I think this should be `new \\DateTime('now')` instead of `new \\DateTime()`.",
+      "reference": "typo in `$contend`"
+    },
+    {
+      "prediction": "nit: move imports at the top of the file",
+      "reference": "Nit: empty line"
+    },
+    {
+      "prediction": "I think this should be `if fb == nil {` instead of `if fb == (FolderBranch{}) {`",
+      "reference": "I figured this was more fool-proof than trying to plumb the error up from everywhere."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I've been able to get this to work.",
+      "reference": "We know debug and trace as globals exist and are defaulted to 0. In this instance we need to know that c is non- nil, srv is non-nil and logging is non-nil before we can trust this statement not to panic."
+    },
+    {
+      "prediction": "I think this should be `app.ConfigPath = filepath.Join(app.RootPath, \".ddev\", \"config.yaml\")`",
+      "reference": "Good removal here for two reasons: we're redeclaring `app.ConfigPath` immediately below without having used the initial value, and `app.GetConfigPath()` will build the value using the same process as in this line anyway."
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "Should BuildBase be removed?"
+    },
+    {
+      "prediction": "I think this should be `github.com/stretchr/testify/require` instead of `github.com/stretchr/testify/assert`",
+      "reference": "this was a bad Goland auto-import that I missed in the previous PR. Thanks to @rkowalick and @acruikshank for catching it"
+    },
+    {
+      "prediction": "This should be `new QueryAtom(1);` and then `m2.getAtomWithIdx(1)->setAtomicNum(1);`",
+      "reference": "Good catch here. Reading the code, it actually seems kind of dumb that `replaceAtom` needs to copy the atom it's passed. Ah well, we're more or less stuck with that."
+    },
+    {
+      "prediction": "I think this should be `wlr_signal_emit(wrld, &output->events.present, &event);`",
+      "reference": "What's the reason for adding the output pointer here? I'd expect it to be derivable from the actual handler that's called."
+    },
+    {
+      "prediction": "I think this should be `discounted_annual_payment - 1` instead of `discounted_annual_payment - discounted_annual_payment`",
+      "reference": "I don't see tests for this (or the similar method on `Team`)."
+    },
+    {
+      "prediction": "I think you can remove the `typeMap.get(\"airplaneMode\").toString()` line as well.",
+      "reference": "should use Number instead of Long"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do this in the `parse()` function.",
+      "reference": "We had the checks for those two errors for a reason, why do we not need them anymore?"
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "Why didn't you just switch to `range` here too?"
+    },
+    {
+      "prediction": "Why do we need to pass in the client here?",
+      "reference": "I don't even think `nil` should be allowed, since it uses global state to get the client."
+    },
+    {
+      "prediction": "I think it would be better to say something like \"Remote debugging server started successfully. You can now connect to the remote debugging server by pointing a Chromium-based web browser to this URL.\"",
+      "reference": "You somehow got a change from `master` into your branch again, in e6d2167085688264e5ee6a81cfd2a7a8f10ded13. While it's not something terribly bad (as it will be fine after merging), it's kind of confusing, and I have no idea how it happens :laughing: Are you trying to update changes from `master` while working on your branch somehow? How do you do that?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "Minor, can we drop the leading `!` and just swap the if and the else blocks?"
+    },
+    {
+      "prediction": "`if self.with_semantic and (bbox_feats.shape[0] == 0 or bbox_feats.shape[-2:] != bbox_semantic_feat.shape[-2:])`",
+      "reference": "`bbox_feats.shape[0] > 0` requires the number of proposal is not 0."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `componentDidMount` function.",
+      "reference": "Just FYI, I had to remove this unused import because Travis/`npm run lint` were complaining."
+    },
+    {
+      "prediction": "I think you can use `std::cout` instead of `std::endl` here.",
+      "reference": "format.h is not a system header but bundled with LAMMPS. This should be `#include \"fmt/format.h\"`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be moved to `BKDWriter.java` since it's only used there.",
+      "reference": "could we always call visitDocValuesWithCardinality? It seems to include the version check already?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this exception. I think it should be in the `register` method of the `Registry` class.",
+      "reference": "I think if they _don't_ specify a port, we should assume grid default (which is 4444)"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `TautomerCatalogParams` that has a `getTautomers` function that returns a `std::vector<Tautomer>` and then use that in the `TautomerEnumerator`",
+      "reference": "This looks to be a breaking change. I know it will affect some of my code that uses the Tautomer enumerator."
+    },
+    {
+      "prediction": "I think it would be better to use `if len(m.Bucketlist) > 0 {` instead of `if len(m.Bucketlist) == 0 {`",
+      "reference": "Perhaps remove this line"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `if (!ps.removeHydrides && atom->getAtomicNum() == 1 && atom->getFormalCharge() == -1) {` block.",
+      "reference": "Ah, missed this one. Sorry: you don't need to test the atomic number here. that's taken care of above on line 683."
+    },
+    {
+      "prediction": "I'm not sure if it's a good idea to just mute the panic. I think it's better to fix the root cause of the panic.",
+      "reference": "roundInfo() is called by couple of places in consensus for safety, better test/verify the delay of return err due to this Sleep() would not affect the normal functionality, like consensus can still be reached, full-node can correctly sync"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `GeneratePoSt` function itself, since it's the caller of this function that needs to know about the flattening.",
+      "reference": "This 32 could go away too."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to put it in the `Notifications` reducer itself.",
+      "reference": "These props should be less specifically named since the component is generic, for example `anchorLink`, `anchorLinkLabel`."
+    },
+    {
+      "prediction": "nit: I think this should be `Unable to locate EC2 instance by ID or tags`",
+      "reference": "In the future, it'd be great if we can separate commits or PRs that aren't in the same scope"
+    },
+    {
+      "prediction": "I think this should be `copy.copy(cls._%s__gestures)`",
+      "reference": "can you define the string format on a different line, so it can be named, and so that it does not have to be constructed twice?"
+    },
+    {
+      "prediction": "I think this should be called `bake()` instead of `execute()`, since it's a separate method from `execute()`.",
+      "reference": "Can we write a unit test for this."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to do this. I think it would be better to do it in the `on_load_finished` method of the `AbstractTab` class, since that's where the `loadFinished` signal is emitted.",
+      "reference": "Same as above, use `.delete('_autosave')`, not the command handler."
+    },
+    {
+      "prediction": "I think we can remove the `idToConstant` variable now, since we don't need to keep track of it anymore.",
+      "reference": "It isn't necessary to check whether there are projected ID columns. The code is shorter if the values are available by default, even if they aren't used. This fixes the problem where there are constants to add (like `_file`) but no identity partition values are projected."
+    },
+    {
+      "prediction": "I don't think we need to add this to the `COMMANDS` list, since it's not really a command. It's more of a `POST` request, right?",
+      "reference": "Update after command rename"
+    },
+    {
+      "prediction": "This change is not related to this PR. Can you please revert it?",
+      "reference": "@KroArtem This is the Checkstyle violation. This import should be grouped with other `net.sourceforge.pmd` imports below, in alphabetical order"
+    },
+    {
+      "prediction": "I think this should be `import \"github.com/ncw/rclone/fs\"` instead of `import \"github.com/ncw/rclone/fs\"`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "This change seems unrelated to copying email addresses to the clipboard. Please limit the scope of pull requests as much as possible."
+    },
+    {
+      "prediction": "I think this should be `mmap(resource, keymap_size, PROT_READ | PROT_WRITE, MAP_SHARED, keymap_fd)` instead of `mmap(NULL, keymap_size, PROT_READ | PROT_WRITE, MAP_SHARED, keymap_fd)`",
+      "reference": "I'm assuming that there isn't a way for us to skip sending an fd here (-1?), and that we're allocating this fd just to send a legal but useless value across?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems reasonable to put it here.",
+      "reference": "might be good for the units (bytes?) to be in the variable name."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Metrics/BlockLength: Block has too many lines. [166/25]"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `RecordReader` class, since that's where the file reader is.",
+      "reference": "I'd end the sentence after `special indicator (line 3)`. And then start a new sentence: `The special field can indicate that the field specifies a reset, is a sequence ID, or is a timestamp for the record.`"
+    },
+    {
+      "prediction": "nit: \"a batch of images\" -> \"a batch of images\"",
+      "reference": "These empty lines can be kept."
+    },
+    {
+      "prediction": "I think this should be `h.getCoreContainer().getCoreDescriptor(h.coreName).setConfigSetTrusted(true);`",
+      "reference": "For some reason these tests started failing due to an extra `NEWLINE` returned from `h.query()`. Have not tried to check whether this may be a problem for real templates or not, just fixed the failing tests. Anyone with an opinion here?"
+    },
+    {
+      "prediction": "I think this should be `DefaultPropagator = TraceContext`",
+      "reference": "It is better to have a method than a var."
+    },
+    {
+      "prediction": "Should this be configurable? I think it would make sense to limit the number of concurrent tasks that can be running at the same time. It would also make sense to limit the number of concurrent activity tasks that can be running at the same time. I think it would make sense to limit the number of concurrent tasks that can be running at the same time. It would also make sense to limit the number of concurrent activity tasks that can be running at the same time. I think it would make sense to limit the number of tasks that can be running at the same time. It would also make sense to limit the number of tasks that can be running at the same time. It would also make sense to limit the number of tasks that can be running at the same time. It would also make sense to limit the number of tasks that can be running at the same time. It would also make sense to limit the number of tasks that can be running at the same time. It would also make sense to limit the number of tasks that can be running at the same time. It would also make sense to limit the number of tasks that can be running at the same time. It would also make sense to limit the number of tasks that can be running at the same time. It would",
+      "reference": "concurrent poller won't be larger than concurrent execution. Jump from 2 -> 16 seems aggressive. I suggest to use 8. Ideally, this should be dynamic config. :)"
+    },
+    {
+      "prediction": "I think this should be `['./work_dirs/yolact_r101']` instead of `['./work_dirs/yolact_r101']`",
+      "reference": "List is redundant."
+    },
+    {
+      "prediction": "I think this should be `if (getSupportActionBar() != null)` instead of `if (getSupportActionBar() != null)`.",
+      "reference": "Please use a string resource here rather than the hardcoded string \"user\". This allows the app to be multilingual."
+    },
+    {
+      "prediction": "I think we can remove this file now, it's not used anymore.",
+      "reference": "@aswathkk Why did we remove this import? I guess we are using it."
+    },
+    {
+      "prediction": "Why remove this line?",
+      "reference": "Why? I thought this is for graceful shutdown."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with at the time of writing.",
+      "reference": "Can `result.stdout` or `result.stderr` be nil?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it would be better to just return 1, which would be consistent with the rest of the issue parser.",
+      "reference": "@mprobst Looks like the `isStdLib` function has been removed in the latest Python plugin. Any thoughts on how to handle this? The new version doesn't seem to have any relevant alternatives."
+    },
+    {
+      "prediction": "I think this should be `MaxIPSetSize int` instead of `IptablesRefreshInterval time.Duration`",
+      "reference": "How can there be a configured max ipset size? Surely we need to write ipsets with however many members are implied by the Calico datastore?"
+    },
+    {
+      "prediction": "I think this should be `incident.messages.delete_attachment`",
+      "reference": "This code is passing a non localized string when it should be passing in a localized string or it should use the name of the item being deleted."
+    },
+    {
+      "prediction": "I think this should be `public <T> R isNaN(BoundReference<T> ref, Literal<T> literal)`",
+      "reference": "why not be consistent with `isNull` and `notNull` and return null?"
+    },
+    {
+      "prediction": "I think this should be `map[string][]string`",
+      "reference": "Did we forget to remove this field?"
+    },
+    {
+      "prediction": "I think this should be `this.values().stream().mapToLong(elem -> elem.elementCount()).sum();`",
+      "reference": "What if there is no element?"
+    },
+    {
+      "prediction": "I think we can remove the `d` variable from the `if` statement above.",
+      "reference": "I think you need to be careful to honour isTest and stamp here."
+    },
+    {
+      "prediction": "I think this should be `com.google.api.codegen.viewmodel.SimpleParamDocView`",
+      "reference": "why zero value for `arrayType` for the default value of the optional args?"
+    },
+    {
+      "prediction": "I think this should be `foreach ($this->getScopes() as $scope => $value) {`",
+      "reference": "@robinbonnes ~~Would you mind putting the array key and value on a new line - it makes it more readable as an array.~~ Never mind, see discussion below."
+    },
+    {
+      "prediction": "I think this should be `cniNetwork` and `cniIPAM`",
+      "reference": "I've never seen us do this anywhere else - why here? Why not `address, meshAddress string`?"
+    },
+    {
+      "prediction": "I think this should be `req.getContext().computeIfAbsent(JsonQueryConverter.contextKey, (k)->new JsonQueryConverter());` and not `req.getContext().computeIfAbsent(JsonQueryConverter.contextKey, (k)->new JsonQueryConverter());`",
+      "reference": "I think it's necessary only for the earlier \"dynamic\" approach. I suppose it's not necessary in the eager one."
+    },
+    {
+      "prediction": "This should be `public function getBackendConfig() { return $this->backendConfig; }`",
+      "reference": "Same here than [there](#r26168925)"
+    },
+    {
+      "prediction": "I think this should be `builder.id.matches(\"^[a-zA-Z0-9_.]+$\")`",
+      "reference": "Again we should think about moving this pattern to a static final field"
+    },
+    {
+      "prediction": "I think it would be better to pass the whole `EasyAdminControllerResolverInterface` instead of just the `EasyAdminControllerResolverInterface|null`",
+      "reference": "haven't been implemented the `__constructor` for `NormalizerConfigPass` ?"
+    },
+    {
+      "prediction": "I think it would be better to have this as a class method, since it's not specific to `Collect`.",
+      "reference": "Docstring is missing."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. The problem is that `BaseEvents::loadFromXml` can be called from multiple threads, so we need to make sure that we don't try to register an event that has already been registered. One way to do this would be to use a mutex to protect against concurrent calls to `registerEvent`, but I'm not sure that this is the best way to do it.",
+      "reference": "I don't see the reason of this return as it will stop loading in half because an error, thats bad imo. Or you load it all and report errors or you load nothing and terminate, the first is as it always was so I wouldn't change that."
+    },
+    {
+      "prediction": "I think it would be better to make this `func (m *Mock) AddPeers(ctx context.Context, addresses ...swarm.Address) error`",
+      "reference": "change from AddPeer to AddPeers added a space in some occurrences"
+    },
+    {
+      "prediction": "I don't think this is the right place for this test. It should be moved to `org.sonar.plugins.java.api.tree.AssignStmtTree`.",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "I think we can remove the `NewIPTableRules` function altogether and just use `NewIPTableRules() *iptablesRules`.",
+      "reference": "Do you remove \"s\" from \"iptables\" to IPTable\" on purpose? I feel it should be \"IPTablesRules\"."
+    },
+    {
+      "prediction": "I think this is the wrong place for this method. It should be in the `PrivacyApiMethod` interface.",
+      "reference": "nit: not your problem - but I wonder if this should have been a \"ConstantResponseRpc\" - which takes a name and a value .... maybe a more reusable concept."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The `MeterProvider` is a public API, so we should be able to call it directly.",
+      "reference": "These changes have been resolved upstream. We can revert changes to this file when synced."
+    },
+    {
+      "prediction": "I think this should be `Slice` instead of `T`.",
+      "reference": "The description predates this diff but I think we can come up with something more grounded."
+    },
+    {
+      "prediction": "I think we should rename this to `SnapshotAPISpec` to be consistent with the naming used for the VolumeAPISpec.",
+      "reference": "Let the Kind, APIVersion & Metadata follow the rules followed by Volume. What is the reason for a change for snapshot struct ?"
+    },
+    {
+      "prediction": "Why is this suppression needed? I don't see any usage of `ServoRegistry` in this PR.",
+      "reference": "seems that we did not strong depend servo no need to keep it? change to another one?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this check. I think it should be in the `wait_response` function, since it's called by the `wait_response` function itself.",
+      "reference": "Looks like this used to be UB before with an empty blocks and then calling `.second` on the `.front()`. Did it ever crash though?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I'm not sure we should be modifying the `testSuites` array here. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `testSuites` array at all. I'm not sure we should be modifying the `",
+      "reference": "Sorry about the code churn here, reworking this use normal loops and async await made debugging far more straight forward. For example, if configuring the fail point fails you find out when stepping over that line vs having to look for the ultimate catch clause that handled it."
+    },
+    {
+      "prediction": "I think this should be `util.run('gn', ['gen', '-j', options.test_launcher_jobs], config.defaultOptions)`",
+      "reference": "this is wrong and should be reverted"
+    },
+    {
+      "prediction": "I think this should be `runtime.GOMAXPROCS(runtime.NumCPU())` instead of `runtime.GOMAXPROCS(0)`.",
+      "reference": "I think you'll need to add an import for runtime, before you can merge this PR. (Presumably it was already there for something else when you were working on this, but has since disappeared.)"
+    },
+    {
+      "prediction": "I think we should remove the `SyncState` and `SyncManager` classes from this test class. They are no longer used in this test class.",
+      "reference": "Unrelated typo fix."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the context of the request itself.",
+      "reference": "I don't think we need an empty value here; any middleware that wants a default value for empty placeholders will pass in what that value should be."
+    },
+    {
+      "prediction": "I think this should be `--storybook-host=${ hostname || 'localhost:9001' }`",
+      "reference": "Why was this changed?"
+    },
+    {
+      "prediction": "Why is this needed? If `value` can be null, why not just return `engine->get(key)`?",
+      "reference": "Compared to using traditional enums, I suggest to use `Status` or `StatusOr`, since they are more expressive and informative. Besides, isolating the definitions of error code of individual modules from the ones of the RPC interface is a good practice, isn't it?"
+    },
+    {
+      "prediction": "I think we can remove `@current_user` and `@invitation` as they are not used anymore.",
+      "reference": "Use the return of the conditional for variable assignment and comparison."
+    },
+    {
+      "prediction": "I think this should be `Optional[AnyKey]`",
+      "reference": "just realised... This breaks things since there is not selenium types module, is there a PR for this?"
+    },
+    {
+      "prediction": "I think this file should be renamed to `testutil.go`",
+      "reference": "move to line 23 below"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `PasscodeActivity` class.",
+      "reference": "the first problem i found in the passcode change flow is that these two lines were happening in 'lock()' but not in 'showLockActivity' which is what the passcode change flow calls. I've just moved them to happen in that method."
+    },
+    {
+      "prediction": "I think this should be `if self._solid_config or self.resources`",
+      "reference": "Doing this adds the actual resources themselves to the dictionary. I think for now, there's no way to know what the resource config provided may be, as we don't permit resource config on `build_solid_context`, so this part can be omitted."
+    },
+    {
+      "prediction": "I'm not sure I like the name `resubscriptions` for this controller. I think we should just call it `subscribers` or `users` or something like that.",
+      "reference": "1 trailing blank lines detected."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `./test/unit/test-partition-value.js`",
+      "reference": "I would personally prefer a filename without an abbreviation, a newcomer to the project wouldn't be able to translate \"pv\" to \"partition-value\"."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in `IMvxViewModel` itself.",
+      "reference": "This check looks odd to me. I can't quite grok it. Why is it needed now? What's changed?"
+    },
+    {
+      "prediction": "I think we can remove the `enableJetStream` function and just call `EnableJetStream` in the `if err != nil` block.",
+      "reference": "This change is needed because otherwise the path was being changed from `path/store/jetstream` to `path/store/` and files stored differently after re-enabling."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The problem is that `session` is a closure, so it will always be referencing the same instance of the `session` object. If you want to end the session, you need to pass in the session object as an argument to `session.endSession()`, not the session object itself.",
+      "reference": "is this necessary? doesn't `client.close()` imply all sessions will be ended?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Extra line, it's not needed here."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to move this to `spec/bolt/pal_spec_helper.rb`?",
+      "reference": "Is this OK to add in the spec helper? I need to call `Bolt::PAL.load_puppet` so that I can use the new `Bolt::PAL::Issues` module when verifying that plans forbid functions."
+    },
+    {
+      "prediction": "I think this method should be called `padEnd` instead of `padTo`.",
+      "reference": "or is it `i < length - back.length()`? `\"12345\".padTo(10, 'a')` should be `12345aaaaa`"
+    },
+    {
+      "prediction": "`cfg.get('custom_imports', None)` -> `cfg.get('custom_imports', [])`",
+      "reference": "custom_imports -> custom_hooks"
+    },
+    {
+      "prediction": "I think this should be `if question[:text].present? && question[:text][0].present?` instead of `if question[:text].present? && question[:text][0].present? %></p>`",
+      "reference": "I don't get why are we baking new hash structures to represent phases, sections, questions. We are not only loosing the references defined in models but also we have to figure out keys and values for this newly structures."
+    },
+    {
+      "prediction": "I think this should be `protected` instead of `protected protected`",
+      "reference": "It's somewhat confusing that the `preload` parameter shadows the instance variable; maybe rename the instance variable to `globalPreload` or `preloadDefault` or so to prevent future confusion?"
+    },
+    {
+      "prediction": "I think this should be `name: string` instead of `name: *string`.",
+      "reference": "Are we setting these values anywhere?"
+    },
+    {
+      "prediction": "I think it would be better to move these headers to the top of the file.",
+      "reference": "This has to be in front of all other includes"
+    },
+    {
+      "prediction": "This change is not correct. The implicit valence should be calculated for both ends of the bond.",
+      "reference": "These changes make sense to me. @ricrogz : you wrote (I think ) the original version of this as part of #2553 Do you see any reason to not make the change?"
+    },
+    {
+      "prediction": "I think this should be `self.location.href` instead of `window.self.location.pathname`.",
+      "reference": "`window.self === window`"
+    },
+    {
+      "prediction": "This test is failing because `svcInitPublisherHelpPrompt` is never called.",
+      "reference": "maybe we should keep one test case for ErrFileNotExists and update only one file for ErrWorkspaceNotFound?"
+    },
+    {
+      "prediction": "I'm not sure why we need to do this. I think we can just call `FlushAsync` now that we have the `FlushAsyncInternal` method.",
+      "reference": "Please add `ConsigureAwait(false)` here and on `ShutdownAsync` below."
+    },
+    {
+      "prediction": "I'm not sure this is a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure it's a good example of what you're trying to do, but I'm not sure",
+      "reference": "Isn't this a false positive? Why are we reporting `unused-variable` on a `del` operation with a `undefined-variable`."
+    },
+    {
+      "prediction": "I think this should be `mysteriumAPILogPrefix, \"SessionStats sent: \", sessionId`",
+      "reference": "Is it correct that `SessionDto` should be here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `ovsctl dump-meters`",
+      "reference": "sorry I missed this earlier. We no longer use ovs-ofctl for flow programming AFAIK, why not add this support to ofnet / libOpenflow?"
+    },
+    {
+      "prediction": "I think we can remove the `cache` parameter from the `addJs` method.",
+      "reference": "Make this consistent with the other definition please `'false'` vs `false`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to run cleanup. I think it should be in the `RunCleanupHandlers` function.",
+      "reference": "I think this is a good way to do it. :+1:"
+    },
+    {
+      "prediction": "I think we can remove this line now.",
+      "reference": "Why this is needed?"
+    },
+    {
+      "prediction": "Why do we need to pass the ovsctlClient to the connTrackDumper?",
+      "reference": "Can we define this directly in the function `NewConnTrackOvsAppCtl` instead of passing this as an argument?"
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to be testing, but it seems like it should be `nolint:illuminate` instead of `nolint:unparam`",
+      "reference": "False positives are annoying..."
+    },
+    {
+      "prediction": "Shouldn't this be `public function generateSuffix(string $name, int $iterationIndex, string $suffix, int $iterationIndexSuffix = 0): string` ?",
+      "reference": "Is the infrastructure layer an appropriate one? I don't think so tbh, more like an application, or even a domain one."
+    },
+    {
+      "prediction": "I think this should be `private static readonly byte _powerOfTwoToHighByte = PowerOfTwoToHighByte();`",
+      "reference": "Why is this required to turn the static readonlies into jitted const? Is this a bug in the jitter?"
+    },
+    {
+      "prediction": "I think this should be `traceback.format_exception(exc_type, exc_value, exc_tb)`",
+      "reference": "feel free to make this a \"typed\" named tuple (overriding __new__) if you are feeling frisky"
+    },
+    {
+      "prediction": "I don't think we need to check for `Number.MAX_SAFE_INTEGER` at all, since it's already guaranteed to be positive.",
+      "reference": "Would use parenthesis to explicitly state the associations of operands ```js const MAX_SAFE_INTEGER = Number.MAX_SAFE_INTEGER || ((2 ** 53) - 1)"
+    },
+    {
+      "prediction": "I think this should be `diff.WithEquateEmpty()`.",
+      "reference": "The bug was caused by this order change in the last refactoring."
+    },
+    {
+      "prediction": "nit: this should be `from docker.utils import kwargs_from_env as kwargs`",
+      "reference": "This is already imported as docker. You're safe to simply use `docker.utils.kwargs_from_env()` below."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in `actions.go`",
+      "reference": "No need to add this fun. Using RunnableActionsBuilder#AddActions instead"
+    },
+    {
+      "prediction": "I don't think it's a good idea to have this in `core` since it's not part of the public API. I think it would be better to have it in `dom` or `core` depending on which is more relevant.",
+      "reference": "I don't think it's possible to do this for hooks"
+    },
+    {
+      "prediction": "I think this should be `this.wait(sec * 1000 + 1000);`",
+      "reference": "wait(1ms) when sec=0. wait(0ms) waits forever so that's why this is better."
+    },
+    {
+      "prediction": "I think this should be `H2O_STRING` instead of `H2O_VECTOR`",
+      "reference": "Do we need `domain` and `so_reuseport` now that we open new sockets immediately after calling `add_listener`?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. It's not about closing the connection, it's about closing the connection to the server.",
+      "reference": "The formatting seems different from the rest of the code"
+    },
+    {
+      "prediction": "I think we can simplify it to `return strings.Join(zoneStrs[:len(zoneStrs)-1], \"-\"), nil`",
+      "reference": "What about calling this package \"parameters\"? utils\\parameters are utils related to parameters."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think it would be better to use `System.out` instead of `System.lineSeparator()`.",
+      "reference": "what is the actual difference here?"
+    },
+    {
+      "prediction": "I think it would be better to use `split(\"&\")[0]` instead of `split(\"\\\\?\")[0]`",
+      "reference": "Parsing URLs is hard. Any fixes should avoid manually parsing the URL."
+    },
+    {
+      "prediction": "I think this should be `@group_level = 0` instead of `@group_level = @group_level - 1`",
+      "reference": "This is a private api, it doesn't need to return anything so `@group_level -= 1 if @group_level > 0` is preferred."
+    },
+    {
+      "prediction": "I think this should be `auto value_type_stride = value_type.GetByteStride(m_process);`",
+      "reference": ".GetValueOr(0) ... but really, shouldn't m_value_strife also be optional?"
+    },
+    {
+      "prediction": "I think this should be `Record<string, ?string>`.",
+      "reference": "Working around a possible limitation in the JSDoc plugin's typescript syntax checking. This means the same thing"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `HttpServerTracingHandler` class.",
+      "reference": "Should be readonly"
+    },
+    {
+      "prediction": "I don't think this should be in this PR.",
+      "reference": "I'm not going to stop accepting the PR, but I really hate these artifacts getting dropped in the Python code just to stop things like coverage checkers and linters from complaining."
+    },
+    {
+      "prediction": "I don't think this file should be deleted.",
+      "reference": "FYI modifying an existing file generally doesn't invalidate the copyright date. The date here is when the copy right _begins_, so moving it later is arguably misleading-it still applies. Not a big deal, just something I've seen a few times in passing."
+    },
+    {
+      "prediction": "I think it would be better to add this as a constant in `DispatcherConstants.java` and then use `DispatcherConstants.ALLOWED_METHODS` and `DispatcherConstants.ALLOWED_ORIGINS` instead of `DispatcherConstants.CORS_ALLOWED_METHODS` and `DispatcherConstants.CORS_ALLOWED_ORIGINS`",
+      "reference": "Because the default value of allowedOrigins is * (all origins), so it isn't necessary to set again at all."
+    },
+    {
+      "prediction": "I think this should be `hive.warehouse.dir` instead of `hive.metastore.warehouse.dir`",
+      "reference": "Style: indentation should be 4 spaces (2 indents) from the start of `Preconditions`."
+    },
+    {
+      "prediction": "I think we should add a note here that this is not supported yet and will be fixed in a future version.",
+      "reference": "I don't know why I suggested a html attribute at first (my bad), or maybe you think it makes sense. Otherwise, what about a simple class ?"
+    },
+    {
+      "prediction": "I think this should be `config.tracker_root = \"UoE\"`, not `config.tracker_root = \"UoE\"`",
+      "reference": "Probably want something more generic here like 'DMPRoadmap' so that other installations aren't using UoE by default."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to log the full stack trace. I think it would be better if the full stack trace was logged in the caller of the method that threw the exception. For example, in the case of an unhandled exception, the caller of the method that threw the exception should be able to see the full stack trace.",
+      "reference": "Must this not change to `MustBeRethrownImmediately`?"
+    },
+    {
+      "prediction": "I think we can remove the `lastTf` variable as well.",
+      "reference": "Can it be pointer?"
+    },
+    {
+      "prediction": "I think we can remove `response.requiredScopes.length < response.requiredScopes.length` now that we have `global.googlesitekit.setup.requiredScopes = response.requiredScopes;`",
+      "reference": "Not related to this issue but this condition seems weak."
+    },
+    {
+      "prediction": "I think this should be `array_key_exists($this->mainConfig->Record->preferredMarcField, $this->fields[$marcField])`",
+      "reference": "I think this might be a little more readable in two lines as: <pre> $preferredMarcField = $this->mainConfig->Record->preferredMarcField ?? 'fullrecord'; $marc = trim($this->fields[$preferredMarcField] ?? $this->fields['fullrecord']); </pre>"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "does this still allow this file to be used with Python installations that don't have numpy?"
+    },
+    {
+      "prediction": "I think this should be `goPkg = goPkg.replaceFirst(\"v(.+);\", \"apiv$1;\").replace(\"\\\\/v([a-z1-9]+);\", \"\\\\/apiv$1;\");`",
+      "reference": "what about 0? Version probably can' start with 0, but v10 is theoretically possible."
+    },
+    {
+      "prediction": "Why do we need this method? I don't see it being used anywhere.",
+      "reference": "Usually we consider APIs that have filesystem interactions to be part of IFileHelper, this would allow us to inject a testable implementation easily. `Path.IsRootedPath` doesn't access the file system. It is string comparison I believe."
+    },
+    {
+      "prediction": "I think this should be `status: \"Init\",` instead of `status: \"\"`",
+      "reference": "Do we need to use quote? Why not `status: Init`"
+    },
+    {
+      "prediction": "I think it would be better to use `is not None` instead of `== \"\"`.",
+      "reference": "Nice catch ! I don't think I would have caught that, did you search the whole code base for pre-existing functions ?"
+    },
+    {
+      "prediction": "I think we can remove this test now that we have `testClusterDeploymentWithoutRegionLabel` and `testClusterDeploymentWithRegionLabel`",
+      "reference": "Remove this line."
+    },
+    {
+      "prediction": "I think we can remove this test now, since we don't need to wait for all the blocks to be returned.",
+      "reference": "This is no longer called in any of our unit tests, right?"
+    },
+    {
+      "prediction": "I think this should be `uint32` instead of `*uint32`",
+      "reference": "I think typically K8s APIs will use `int32` and not `uint32`. I think we should keep it consistent with the rule `Priority` above. I believe that the rationale for using `int32` in general is: * some programming languages don't have native support for unsigned integers * it's easier to catch sign errors with signed integers, depending on the programming language (it it's negative, it's clearly an error but if it's a very large number, it could be normal or it could be a negative number overflow)"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion at the moment.",
+      "reference": "If you end up changing grant to an association, this may need to change to a reference to enforce the foreign_key `add_reference :plans, :grant`"
+    },
+    {
+      "prediction": "I'm not sure about this one. I think it's better to have a function that returns the image based on the type of notification. For example: `const getWinImage = notification => { return `${ global._googlesitekitLegacyData.admin.assetsRoot }images/${ notification.type }` };` Then you can use it in the alert template: `winImage={ getWinImage(notification).winImage ? `${ global._googlesitekitLegacyData.admin.assetsRoot }images/${ notification.type }` : '' }`",
+      "reference": "This won't work because `notification` comes from an API response here. I think this would be the same as the change to `Alert` where it would get `SmallSunSVG`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to set these environment variables. I think it would be better to move them to the `ecs-env` package.",
+      "reference": "This is not checked below?"
+    },
+    {
+      "prediction": "I think this should be `map[string]interface{}`",
+      "reference": "Will this ever return a \"Stack does not exist\" error? Should we handle it silently here rather than making higher level packages do error checking?"
+    },
+    {
+      "prediction": "I think this should be `configure_from_url` instead of `configure_from_default`",
+      "reference": "minor: at this scale, one import per line reads better"
+    },
+    {
+      "prediction": "I think this should be `t.text` not `t.string`",
+      "reference": "Should remove the default templates as well."
+    },
+    {
+      "prediction": "I think this should be `config = require('./webpack.dev.js');` instead of `config = require('./webpack.prod.js');`",
+      "reference": "Why not `let`?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `ZipkinServerApplicationInitializer` instead.",
+      "reference": "ns -> DNS My first reading was this is referring to System.nanoTime and thought hrm?"
+    },
+    {
+      "prediction": "I think this should be `start_date` instead of `start_timestamp`.",
+      "reference": "is there a reason not to remove the param now? I think the serdes will still work?"
+    },
+    {
+      "prediction": "`self.assertAlmostEqual(bst.upper_bound(), 3.3200000000000003, places=2)`, `self.assertAlmostEqual(bst.lower_bound(), -3.1200000000000003, places=2)`, `self.assertAlmostEqual(bst.upper_bound(), 3.32, places=2)`, `self.assertAlmostEqual(bst.lower_bound(), -3.12, places=2)`, `self.assertAlmostEqual(bst.upper_bound(), 3.3200000000000003, places=2)`, `self.assertAlmostEqual(bst.lower_bound(), -3.1200000000000003, places=2)`, `self.assertAlmostEqual(bst.upper_bound(), 3.3200000000000003, places=2)`, `self.assertAlmostEqual(bst.lower_bound(), -3.1200000000000003, places=2)`, `self.assertAlmostEqual(bst.upper_bound(), 3.320000000",
+      "reference": "`places=2` seems to be very poor comparison. Do you have any thoughts why is it fail with more strict checks?"
+    },
+    {
+      "prediction": "I think this should be `logger.info(\"Project \" + name + \" doesn't exist in cache, fetching from DB now.\", e);`",
+      "reference": "should we make level debug? Also just FYI, with debug level, we can add more detailed logging for better debuggability without concerning overwhelming logging message"
+    },
+    {
+      "prediction": "I think this should be `from nupic.bindings import GetNTAReal` instead of `from nupic.bindings.math import GetNTAReal`",
+      "reference": "I needed to add this import for Sphinx to inspect C++ runtime objects properly. This should be the only code change in this PR."
+    },
+    {
+      "prediction": "I'm not sure I like the name `RESOURCE_CHANGED` here. What about `NAME_GENERATOR_CHANGED` or `NAME_GENERATOR_ACTIONS_CHANGED`?",
+      "reference": "Changes in this file aren't related to the PR, are they? You took the opportunity to change the group deletion to using the subscriber too?"
+    },
+    {
+      "prediction": "I think it would be better to have a static method `PromoteSortedColumnDefault(MetricsMode defaultMode)` instead of a private method `promoteSortedColumnDefault(MetricsMode defaultMode)`.",
+      "reference": "I'd probably move this into `MetricsConfig` as a private method. Seems like we only use it there."
+    },
+    {
+      "prediction": "I think this should be `Lint` instead of `Linter`",
+      "reference": "Not clear here what you mean with this list"
+    },
+    {
+      "prediction": "I don't think we need to suppress the warning here.",
+      "reference": "Curious, is there a way we can get around this without suppressing?"
+    },
+    {
+      "prediction": "I think this should be called `ProposalID` instead of `ProposalID`.",
+      "reference": "I don't think we this field in session, it's a serial number for proposal but not unique index itself"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's the only way I could come up with that works.",
+      "reference": "Can't have two benchmarks with `Baseline = true`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `src/test/unit/wp-api-endpoints.test.js`",
+      "reference": "Same here: Empty line before and no `Internal dependencies` docblock (mandatory anyway once #217 lands)."
+    },
+    {
+      "prediction": "I think this should be `The path to the configuration file. Default is .pipe.yaml`",
+      "reference": "sorry, the default value for this field is no longer allowed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be using a different approach.",
+      "reference": "Why does this work? I'd've expected that an admin could create a file anywhere."
+    },
+    {
+      "prediction": "I think this should be `\"/p:Configuration=#{config[:configuration] || 'Release'}/p:Platform=\\\"#{config[:platform] || 'x64'\\\" #{config[:solution]}\", timing: true if config[:solution]",
+      "reference": "falling back to `x64` seems like a really bad idea given that the VS templates default to x86 these days."
+    },
+    {
+      "prediction": "I think this should be `mpl_renderer.comms['default'] = self.default_comm` and `bokeh_renderer.comms['default'] = self.default_comm`",
+      "reference": "Bit surprised by this. I would have thought either you want to set it for both mpl and bokeh...or alternatively it was only ever meant for bokeh and was always wrong?"
+    },
+    {
+      "prediction": "I don't think this is correct. The `from .` import is a special kind of import, and it should not be included in the list of imported names.",
+      "reference": "Check modname instead, but only if it's a ImportFrom."
+    },
+    {
+      "prediction": "I think this should be `write.metadata.previous-log-max-count`, not `write.metadata.previous-log-max-counts`.",
+      "reference": "How about `write.metadata.previous-versions-max`? No need to refer to these as a log."
+    },
+    {
+      "prediction": "I think this should be `windows: false` and not `windows: true`",
+      "reference": "We generally do this via tags in the Rakefile, not separate environment variables."
+    },
+    {
+      "prediction": "I don't think we need to add this to the `COMMANDS` list, since it's not really a command. It's more of a `POST` request, right?",
+      "reference": "Update after command rename"
+    },
+    {
+      "prediction": "I think this should be `if not self._predictedFieldName in self._input[0]:`",
+      "reference": "`if X not in Y:`"
+    },
+    {
+      "prediction": "I think this should be 1000, not 10.",
+      "reference": "Do we know why the pool is exhausted? In the past, we had a few leaks in the Spark catalog code which led to this. It can be also a valid use case too if we simply need a larger pool."
+    },
+    {
+      "prediction": "I think this should be `if options[:verbose]` instead of `if options[:verbose] && options[:debug]`",
+      "reference": "Logging in Bolt still seems to be a little messy. I think this is more correct than what I had before, but made sure I wouldn't overwrite an existing level. And if console logging gets more options, both debug and verbose need to be fixed here."
+    },
+    {
+      "prediction": "Why not use `boost::algorithm::to_string` here?",
+      "reference": "This could be simplified (if including <boost/algorithm/string/erase.hpp> is fine): `auto address_string = boost::algorithm::erase_first_copy (address.to_string (), \"::ffff:\");`"
+    },
+    {
+      "prediction": "I think this should be `LOG.info`",
+      "reference": "This is wildly unhelpful to users --- they need to know when a proxy has been registered."
+    },
+    {
+      "prediction": "I think this should be `path: \"topics/:attachment/:id_partition/:style/:filename\"`",
+      "reference": "Why do we need `do_not_validate_attachment_file_type` is we're doing it in the previous line?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the `onCreate` method of the `LoginActivity`.",
+      "reference": "Loading login page right away only for the regular use case."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "IIRC this flag disables re-rendering entirely (`s/recursively //`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do this. I think it's better to add/remove the class on the `window.location.href` change.",
+      "reference": "Did you test this redirect? @thornbill mentioned it might need `web` at the front, but if this works fine I'd rather leave it this way."
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "FYI, this returned `StringValues` which is a value type (aka can never be null)."
+    },
+    {
+      "prediction": "I think this should be `nil` if `expr.Reserve` is `nil` and `sizeHint` is `nil` if `expr.Reserve` is not `nil`.",
+      "reference": "Oh no, that's a bug."
+    },
+    {
+      "prediction": "I think this should be `ApplicationError string` instead of `ApplicationError error`",
+      "reference": "Let's leave a note in CHANGES to replace this with an error metadata struct or interface, unless it would be less effort overall to take a run at it in this change."
+    },
+    {
+      "prediction": "I'm not sure this is a good name for this property. I think it should be something like `nameTo` or `nameFrom`.",
+      "reference": "The changelog entry does a better job of explaining how this is used than this description. In particular, this says that you need to set `mapTo` if the underlying name is different, but not that `mapTo` *is* the underlying name."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure it's necessary.",
+      "reference": "These imports will probably need to be refactored so they aren't here. That applies for all of the database drivers. The reason for this is because we sometimes want a driver to be available and sometimes we don't. When we include this library in our cloud offering, the sqlite3 connector needs to be gone because it's a security vulnerability and there's no use case for it in a cloud environment. But for flux itself, I can see how sqlite3 would be really useful and so I like this change. But this import needs to only exist within the tests and within the `cmd/flux` package. It cannot be here."
+    },
+    {
+      "prediction": "I think this should be `$this->relationModel->table . ':' . $this->relationModel->getKeyName();`",
+      "reference": "We can use `getQualifiedKeyName` here instead. I will update."
+    },
+    {
+      "prediction": "`zip`, `x-7z-compressed`, `x-rar-compressed`, `x-gtar`, `x-apple-diskimage`, `x-diskcopy` -> `zip`, `x-7z-compressed`, `x-rar-compressed`, `x-gtar`, `x-tar`, `x-apple-diskimage`, `x-diskcopy`",
+      "reference": "Could you clarifty why some of those are removed?"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it should be in the `FakeDeploymentConfigTemplates` fixture.",
+      "reference": "`ctx` is unused in ListDeploymentConfigTemplates"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "nice to see the usage of block for a more readable link name"
+    },
+    {
+      "prediction": "I don't think this is correct. The error message should be \"could not read %llu bytes. read only: %llu\", not \"could not read %llu bytes. read only: %llu\".",
+      "reference": "Can these use `static_cast<unsigned long long>` instead? Other than that, it's fine."
+    },
+    {
+      "prediction": "I think it would be better to use `app.config[cfg_name] = cfg_value` instead of `app.logger.debug(\"{0} = {1}\".format(cfg_name, cfg_value))`",
+      "reference": "This output is only present when `DEBUG=True`. Does it really bother so much?"
+    },
+    {
+      "prediction": "I think this should be done in the `background` function, not here.",
+      "reference": "I think it should pass wallet transaction as well, otherwise there will be 2 wallet read transactions in 1 threads (next in scan_receivable)"
+    },
+    {
+      "prediction": "`MMFF::Utils` should be renamed to `MMFF::Utils::Internal`",
+      "reference": "RDKit::Utils is now in the namespace for localeswitcer... We could change it to something else."
+    },
+    {
+      "prediction": "I don't think you need to do this. You can just do `order = [(*levs, method) for _ in range(len(order))]`",
+      "reference": "@itholic, can we fix it in `_normalize_keyword_aggregation`?"
+    },
+    {
+      "prediction": "I think this should be `DEVELOP` instead of `prime-maintain`",
+      "reference": "It seems like we have the main plan in the database but the downgrade plan in the code. Probably okay for now, but as our thinking of how downgrades/plans develops we may want to consolidate."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. This is something that should be handled by the browser, not the driver.",
+      "reference": "we shouldn't add a bunch of static initializers 'everywhere' in the code. Probably just one would be good, during the construction of the 'RemoteWebDriver' class."
+    },
+    {
+      "prediction": "I think it should be `clef-signer-` instead of `clef-signer-enable` and `clef-signer-endpoint` instead of `clef-signer-enable`",
+      "reference": "do we need both flags? maybe infer that `clef-signer-enabled` whenever `clef-signer-endpoint` is defined?"
+    },
+    {
+      "prediction": "I think it would be better to use `node_modules/identity-obj-proxy/scss` instead of `node_modules/identity-obj-proxy` here.",
+      "reference": "Do we need this? I didn't see css files (only scss)."
+    },
+    {
+      "prediction": "I think it would be better to use `settings.REMOTE_EXCLUDE_LOCAL` instead of `not is_local_interface(host)`",
+      "reference": "This feels like a weird `if not ... or not` to me. Can you talk out the intent here?"
+    },
+    {
+      "prediction": "This is a bit of a nitpick, but I think it would be better if this was a flash message.",
+      "reference": "Align the parameters of a method call if they span more than one line."
+    },
+    {
+      "prediction": "This class should be renamed to `OptionsPackage`.",
+      "reference": "`@package *` can be removed as well."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "@ScottGarman is there a reason that `redhat_fips` needs to be a beaker platform? Why can't it be another platform that beaker just understands as `redhat`?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Where is the other place?"
+    },
+    {
+      "prediction": "I think this should be `Sequence` instead of `UserList`, because `list` is a subclass of `Sequence` but `Sequence` is not a subclass of `list`.",
+      "reference": "Are you subclassing from `UserList` instead of `list` to satisfy some constraint of Jython? My main concern here is that this is a subtle breaking change to the API because `UserList` is not a subclass of `list`. This has the unfortunate side effect of breaking any user that is using `isinstance` checks to see if a `ResultSet` is a `list`."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "We can DRY the above three functions"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `HashMap` is a collection, not a list. You should be able to use `toList()` on the `HashMap` and then call `finisher` on the result.",
+      "reference": "Only no good is good code. Nice that you were able to remove all this duplicate stuff!"
+    },
+    {
+      "prediction": "I think this should stay as it is, since it's the only thing that's changed in this PR.",
+      "reference": "Looking at the code, it seems very odd that this method even exists in the SurfaceNamer, since it is only implemented and used in the PhpSurfaceNamer, and in fact the RetrySettings class name is hardcoded elsewhere. WDYT of just removing this method from SurfaceNamer and PhpSurfaceNamer?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface.",
+      "reference": "these interfaces are defined twice. we should use only one and have it in `storage` package maybe"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the `models` folder.",
+      "reference": "This is the wrong context for usage of `global` keyword. It need only be used inside functions."
+    },
+    {
+      "prediction": "I think this should be `if (child.isAssignableFrom(targetType))`",
+      "reference": "this change should be described in the changelog. I'll update it when merging if nothing else arises."
+    },
+    {
+      "prediction": "I think it would be better to have a separate interface for GRPC dispatchers, so that we don't have to duplicate code.",
+      "reference": "Just to stay consistent with naming let's call it CreateGRPCDispatcherForOutbound. Also rename 'CreateDispatcherForOutbound' to 'CreateTChannelDispatcherForOutbound'"
+    },
+    {
+      "prediction": "nit: `LOG.info(String.format(\"Pulling %s:%s\", name, tag));`, `LOG.finest(String.format(\"Pulling %s:%s\", name, tag));`, `LOG.trace(String.format(\"Pulling %s:%s\", name, tag));`, `LOG.warn(String.format(\"Pulling %s:%s\", name, tag));`",
+      "reference": "This will always need to be displayed to users."
+    },
+    {
+      "prediction": "I think this should be `binascii.hexlify(s)`",
+      "reference": "This looks very much like `repr` now (we miss 127 though) - maybe just do `repr(x)` + strip outer parantheses?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put visibility timestamp. I think it's better to put it in `TaskInfo` itself.",
+      "reference": "I don't see this field is set in task_generator.go. worth to double check if we put the timestamp when creating replication tasks."
+    },
+    {
+      "prediction": "I think this should be `return callback(null, false);` and not `return callback(null, false).then(() => callback(null, false));`",
+      "reference": "nit: I prefer the space before and after this check to make it very easy to scan when reading through code"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the `finally` block of the `LogManager.Flush()` call above.",
+      "reference": "should users also do a triple-flush?"
+    },
+    {
+      "prediction": "I think this should be `uint64` and `complex64` should be `float64`",
+      "reference": "What is the implication of removing these from the test? I thought the idea was to have all types supported by `docstore` here, and do things in the drivers to make them support them (e.g., list of size 2 for complex). If you're removing them because firestore doesn't have native support, then we're going to end up only testing the intersection of all types supported by all drivers; we need a better way."
+    },
+    {
+      "prediction": "I think this should be `if not isinstance(vals, list) or isinstance(vals, str)`",
+      "reference": "Let's also adjust the somewhat weird type signature here as well. This probably should be `typing.Any` instead of `str` in the last line if we intend to support ints."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "The body can just be `os.RemoveAll(dir)`"
+    },
+    {
+      "prediction": "I think this should be `throw new UnsupportedOperationException(\"Only support cast to Long, Double and String\");`",
+      "reference": "Is this covered by regression tests? Also you might make a patch onto refactor branch."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in the `SolrRequest` interface.",
+      "reference": "This is maybe more of an admin request? WDYT?"
+    },
+    {
+      "prediction": "`toList()` can return a `List` or `Collection` depending on the `TraversableOnce` implementation, so we need to check the return type in the `map()` call.",
+      "reference": "Here I thought of `toHashMap` and later add `toTreeMap`. But I start to see, what you may have in mind. Alternatively we could provide a `toMap` and `toSortedMap`, which is great, too. On the other hand is always good to be as specific as possible. What do you think?"
+    },
+    {
+      "prediction": "I think this should be `localProjectApplicationNames` since it's an array of strings.",
+      "reference": "maybe localProjectAppNames? Just a thought."
+    },
+    {
+      "prediction": "I think this should be `height += nextHeightWidth.height;` instead of `height -= Blockly.BlockSvg.NOTCH_HEIGHT; // Exclude height of connected notch.`",
+      "reference": "This isn't new, but getHeightWidth is now defined in block_svg.js, block_render_svg_horizontal.js, and block_render_svg_vertical.js. One of these should be unnecessary."
+    },
+    {
+      "prediction": "I think this should be `confirmed := fixture.WaitForAllTxnsToConfirm(curRound+10, txids)`",
+      "reference": "I can't see any reason why this would fix a failure in `a.True(confirmed, \"txgroup\")`. We might have an issue elsewhere, but increasing the wait time wouldn't help."
+    },
+    {
+      "prediction": "I think this should be `aggressive` instead of `aggressive/perf`",
+      "reference": "This section needs to be modified; crontab format differs between OS releases (Debian and CentOS, at least)."
+    },
+    {
+      "prediction": "Should this be `resume_from_failure: bool = True`?",
+      "reference": "if we think this is likely to be augmented with a additional 'run coordination' features or configuration in the future, we could make it an object of some kind instead (or an enum, if we think there may be other resume modes in the future besides just on/off). Just imagining 6 months in the future, it would be unfortunate if there were 7 new args here as the feature gets more complex"
+    },
+    {
+      "prediction": "I think it would be better to move this to the top of the file.",
+      "reference": "`affine_grid` and `grid_sample` are currently unused. We may remove it from `ops/__init__.py` to speedup the loading of mmdet."
+    },
+    {
+      "prediction": "I think this should be `config.AttackTimeout` not `config.IdleTimeout`",
+      "reference": "Is there a reason why you need to support negative `AttackTimeout`? Why not just throw an error?"
+    },
+    {
+      "prediction": "I don't think we need to import `F` here, since it's not used in this file.",
+      "reference": "nit: an unnecessary change?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `http.Handler` interface.",
+      "reference": "Is trailer parsing something we have to do in this PR? It would be really helpful to separate stuff like this into smaller, self-contained PRs."
+    },
+    {
+      "prediction": "I think this should be `wait_for { |v| v.ready? }`",
+      "reference": "Any chance of a wait-forever situation here? Is there a reasonable timeout?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it's better to do it in the worker.",
+      "reference": "line is 165 characters"
+    },
+    {
+      "prediction": "I think this should be done after the namespace cache refresh, so that we don't wait for the refresh to be complete before we check if the namespace is loaded.",
+      "reference": "is this for cross DC case?"
+    },
+    {
+      "prediction": "Why do we need to change this?",
+      "reference": "Remove this line."
+    },
+    {
+      "prediction": "I'm not sure if we want to remove the entry at all. If we do, we should remove it from `Profile.TranscodingProfiles` and add it to `Profile.TranscodingProfiles_vp8`.",
+      "reference": "Maybe it would make sense to use `webmVideoCodecs` here and just append `vpx`. I'm not sure if av1 is currently supported when transcoding to webm though. It looks like it can be included in the mp4 transcoding profile now."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `org.apache.druid.query.aggregation.DocValuesAggregatorStats`",
+      "reference": "Interestingly, this has nothing to do with replication, no clue what this means here"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `CreateRepoIfNecessary` function.",
+      "reference": "Is that because we don't provide a UI for these?"
+    },
+    {
+      "prediction": "I think this should be `github.com/influxdata/flux/internal/errors`",
+      "reference": "Standard is either to use `HappyPath` (CamelCase) or `happy path` (lowercase sentence). I usually prefer the latter."
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "instead of using blkMemDao, we could generate a blockMemDao, and then use it as the second parameter."
+    },
+    {
+      "prediction": "I think it would be better to move this function to the top of the file, so that it's clear what it's doing.",
+      "reference": "nit(2/10): `topology && topology.description != null`;"
+    },
+    {
+      "prediction": "I don't think this is needed. You can just import `pyspark.sql` here.",
+      "reference": "why the space? I may not know all the style conventions."
+    },
+    {
+      "prediction": "I think this should be `it('should fail if the dlitem has a parent <dl> with a changed role', function() { var checkArgs = checkSetup('<dl role=\"menubar\"><dt id=\"target\">My list item</dl>'); assert.isFalse(checks.dlitem.evaluate.apply(null, checkArgs)); });`",
+      "reference": "Should be \"should fail if the **dt element** has a parent <dl> with a changed role\""
+    },
+    {
+      "prediction": "I don't think we should be setting this in the config file. I think we should be setting it in the `Workshops::Application` class.",
+      "reference": "Just to be clear: 1. This is required for compression, right? 2. This won't actually result in our dynos serving the assets, since they'll be served via the CDN, correct?"
+    },
+    {
+      "prediction": "I'm not sure we should remove the warning here. I think we should keep it, but we should add a note that this method is deprecated and will be removed in a future version.",
+      "reference": "not sure about removing this ping @Soullivaneuh"
+    },
+    {
+      "prediction": "I think this should be `allow_nil: true` instead of `allow_nil: false`",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I think this should be `expecting error when instance could not be found: %v`",
+      "reference": "If instance could not be found when the provider id is set, `findInstance()` returns `ErrInstanceNotFoundByID` error. So that during reconcileNormal(), we don't create a new instance. In `reconcileDelete()`, when ErrInstanceNotFoundByID is seen, deletion continues to clean up even if the instance is gone (may be manually deleted)."
+    },
+    {
+      "prediction": "I think this should be `private HashSet<string> _memory = new HashSet<string>();` instead of `private Tuple<string, HashSet<string>> _memory = new Tuple<string, HashSet<string>>((string)null, new HashSet<string>());`",
+      "reference": "I would recommend making this a Tuple&lt;ReplicateValue, IColllection&lt;object>> You only need to convert things to strings if you need to persist them in Settings or something. If they only need to live for the life of dialog, you can keep everything as objects. You can use \"null\" for the ReplicateValue for when they have not chosen a property."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "I think there is no point providing action key here, as it will be called specifically, rather than randomly"
+    },
+    {
+      "prediction": "Nit: I think we can remove `_runtimeDescription` now that we have `RuntimeDescription`.",
+      "reference": "`XUnitIntegration` doesn't need to cache this anymore."
+    },
+    {
+      "prediction": "I think this should be `themes/*` instead of `themes/*/*.lua`",
+      "reference": "Can we have something line `# noqa` there instead? I think it's fine like this though."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "this variable is used in both branches so I've decided to declare it above"
+    },
+    {
+      "prediction": "I think you can remove the `FLAG_ACTIVITY_NEW_TASK` here.",
+      "reference": "`setFlags()` overwrites the previous flag, we should be using `addFlags()` to append flags."
+    },
+    {
+      "prediction": "I think this should be `if repo.IsMirror && isNameChanged {`",
+      "reference": "Need `return` after this."
+    },
+    {
+      "prediction": "I think it would be better to use the CacheableResponseTrait and add the cache headers in the `addCacheHeaders` method.",
+      "reference": "I would say we should not use the `CacheableResponseTrait` within the `JsonResponseTrait`. Traits within traits tend to lead to a bad developer experience as it can be really hard to find the actual method you're seeing in the implementing class, and in this case it looks like we're using _both_ the cacheable and the JSON traits in the same controller, so it's redundant anyway. I'd recommend just removing that use statement from `JsonResponseTrait` and make sure `ChaceableResponseTrait` is included everywhere it's needed specifically."
+    },
+    {
+      "prediction": "I think this should be `LOG.finest(String.format(\"XPUB binding to %s, XSUB binding to %s\", xpubAddr, xsubAddr));`",
+      "reference": "This change is unhelpful: it precludes users from knowing which ports are being used for what purpose within the system."
+    },
+    {
+      "prediction": "Style/SafeNavigation: Use safe navigation (&.) instead of checking if an object exists before calling the method.",
+      "reference": "Whoa, we weren't doing this before?? Derp."
+    },
+    {
+      "prediction": "I think we can remove this TODO. We don't need to copy the actions to a new flow, we can just use the existing `CopyActionsToNewFlow` function.",
+      "reference": "I'm surprised we didn't go with something like `flow := f.Flow.Copy()` to take care of all the fields at once, but as long as it works it's good enough for me"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `if config.BPFEnabled && config.XDPEnabled {` block.",
+      "reference": "We would like this code to run, even when `config.XDPEnabled` is false, so that Felix can clean up its own XDP state after a restart."
+    },
+    {
+      "prediction": "Why is this needed? I don't see any usage of this interface.",
+      "reference": "I'd declare this field `qualifiedName` private to hide it. Unless it really needs to be modified from somewhere else... (e.g. unit tests..), but then, we should find a solution, where this field can stay private."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get this to work.",
+      "reference": "is this exhaustive? or rather, any psudo header started with `:` is un-parsable/invalid in HTTP/1 right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move it to `test/unit/sessions/unit-tests.js`",
+      "reference": "Is this change implying something or just seems fit b/c regardless of outcome the sessions should still be length 1?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this TODO. I think it should be in the `ApiaryConfig` class, since that's where it's used.",
+      "reference": "Maybe throw an exception if there is more than one, so that discovering the lack of support is easier when someone tries to use it down the road."
+    },
+    {
+      "prediction": "I'm not sure if we want to keep the `new_languages` and `new_languages_thanks` routes for the `marketing#new_languages` and `marketing#new_languages_thanks` controllers, or if we want to move them to a separate controller. The reason is that the `new_languages` and `new_languages_thanks` routes are only used in the `marketing#new_languages` and `marketing#new_languages_thanks` controllers, so they don't seem to be needed outside of that controller. However, the `new_language_confirmations#index` route is used in both the `new_languages` and `new_languages_thanks` controllers, so they may still be needed outside of the `marketing#new_languages` and `marketing#new_languages_thanks` controllers. I think it would be better to keep the `new_languages` and `new_languages_thanks` routes for the `marketing#new_languages` and `marketing#new_languages_thanks` controllers, and move the `new_language_confirmations#index` route to a separate controller. This would make it easier to maintain and update the `new_language_confirmations#index` route in the future.",
+      "reference": "Prefer single-quoted strings when you don't need string interpolation or special symbols."
+    },
+    {
+      "prediction": "I think this should be `os.Getenv(\"TRAVIS_OS_NAME\") == \"\" || os.Getenv(\"TRAVIS_OS_NAME\") == \"linux\"`.",
+      "reference": "I feel the name of this function could be better, conveying it's a predicate. Something like `HasDockerTestEnvironment` or `CanRunLocalServerTests`, etc."
+    },
+    {
+      "prediction": "I think this should be `MyGame.Example.MonsterStorage.Retrieve` instead of `Example.MonsterStorage.Retrieve`",
+      "reference": "This seems unrelated to your PR, how did this end up in here?"
+    },
+    {
+      "prediction": "I think this should be `user.getUsername().toLowerCase()`, not `user.getUsername()`.",
+      "reference": "Is `username` expected to be case insensitive?"
+    },
+    {
+      "prediction": "I think this file should be renamed to `util_test.go`",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I don't think we should be deprecating anything.",
+      "reference": "Is there any point to the deprecation? If someone was using these APIs, can they continue to do so in any meaningful fashion now that you've disconnected them from the PMD internals? Normally \"compiles but doesn't work\" is considered a bug."
+    },
+    {
+      "prediction": "I think this should be `str`, not `QItemSelection`.",
+      "reference": "You can remove the `QItemSelection` import in this file now."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the same place as the `resp_.set_code(...)` call.",
+      "reference": "Use `std::shared_lock` and `std::try_to_lock_t`. BTW. Why we fail on contention?"
+    },
+    {
+      "prediction": "I think this should be `[\".js\", \".jsx\", \".js\"]` and not `[\"\", \".jsx\", \".js\"]`",
+      "reference": "Not sure what this actually does, but do you need to add `\"\"` here? Or does `[\".jsx\", \".js\"]` work too?"
+    },
+    {
+      "prediction": "I think this file should be in the same directory as the rest of the code.",
+      "reference": "let's group imports in the order: std third party first party"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in the test.",
+      "reference": "So the tests run only when the db exists? Else everything passes?"
+    },
+    {
+      "prediction": "I think this should be `io.WriterCloser` or `io.WriterCloser`",
+      "reference": "`Remove` is unclear. It seems the point is to allow the `FifoByteSink` to delete the underlying file. There seems to be an implicit contract that consumers of this interface call `Remove` after `Close` without necessarily understanding what `Remove` does. I get that it's a little awkward to have `FifoByteSink` delete the file in `Close`. It would probably be best to just remove this from the interface, since `AddSink` is working with an actual `FifoByteSink`. I would rename it in `FifoByteSync` to `DeleteFile` or `Cleanup` or something that's makes it more clear what's being removed."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `core/ask.go`",
+      "reference": "Mentioned elsewhere but repeating again: would be nice to have units (ideally `Size *Bytes`)."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "I think a test in `spec/integration` might help keep this file a bit more manageable, especially since we're just checking file contents and errors. But basically the same idea!"
+    },
+    {
+      "prediction": "I'm not sure why this test is here, but it seems to be testing something I don't understand.",
+      "reference": "This spec can be modified, giving you extra strength (Check this fetch key doesn't work and therefore returns `nil`)"
+    },
+    {
+      "prediction": "I don't think this is right. If the label callback function is already set, it should not be passed `NULL` for the `cb` argument.",
+      "reference": "So this means that we can remove a callback via instr_set_label_callback(). Maybe consider adding another API function for removal, and add a new assert here such that cb != NULL?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the copyright notice, but it's the only way I could get to pass the linter.",
+      "reference": "why are we using an instance variable instead of accessing the visibility method?"
+    },
+    {
+      "prediction": "I think this should be `//nolint: gosec`",
+      "reference": "Ok. Can we just disable this linter?"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(os.path.dirname(os.path.realpath(__file__)), 'testdata')` instead of `os.path.join(os.path.dirname(os.path.realpath(__file__)), 'test_data')`",
+      "reference": "this is `test_data` and all other instances are `testdata` - does this change affect any test files, or are there actually no directories called `test_data` in the repo? (I just checked, it seems like there aren't...)"
+    },
+    {
+      "prediction": "I think we can remove this method as well, since it's not being used anywhere.",
+      "reference": "Fixing `lint` warnings that have existed for a while."
+    },
+    {
+      "prediction": "I think this should be `CASDefaultCstorPoolEnvKey`",
+      "reference": "Better to rename this to CASDefaultCstorSparsePool."
+    },
+    {
+      "prediction": "I think we should move this to the `S3FileIO` constructor.",
+      "reference": "Since we'e not 100% sure if `close` will be called more than once, should we set `client` to `null` or add an `AtomicBoolean closed` that will then handle the idempotency issue?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this check. I think it would be better to do it in the constructor of the motorway handler.",
+      "reference": "And here - what happens if name id is invalid"
+    },
+    {
+      "prediction": "I think this should be `echo` instead of `echoes`",
+      "reference": "__toString() is not a part of QueryInterface"
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "nit: Could you add an empty line between `import datetime` (built-in library block) and `import numpy as np` (third-party library block)?"
+    },
+    {
+      "prediction": "I think this should be `CONV_NUMERIC_LITERAL` instead of `API_DRIVER`",
+      "reference": "Please remove `HIP_UNSUPPORTED`"
+    },
+    {
+      "prediction": "Shouldn't this be `return parseInt( daysMatch[ 1 ], 10 ) || 0`?",
+      "reference": "Having boolean parameters to change function behavior is not a good practice because it violates the principle of functions being responsible for a single task, so the need for this indicates we need to split something out of here instead. We can introduce a function like `getCurrentDateRangeDayCount` or similar, which does this."
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in the test.",
+      "reference": "Is this not required?"
+    },
+    {
+      "prediction": "I think we can remove the `CachedState` and `PutState` methods.",
+      "reference": "savedStates is for confirmed states, not needed in working set"
+    },
+    {
+      "prediction": "I think this should be `WireGuard` instead of `WireGuard Protocol`",
+      "reference": "Why should we move it into `bootstrap.go` file?"
+    },
+    {
+      "prediction": "I think it would be better to use `@fontsize` instead of `@measures` here.",
+      "reference": "In the next major version, I plan on changing the em-unit breakpoints to be pixel values to keep everything consistent. Would be great to handle this with pixels as well, but still allow em-units to be defined with a string"
+    },
+    {
+      "prediction": "I'm not sure this is correct. It seems to me that `rjust(2)` is equivalent to `str(idx + 1).rjust(2, '0')`.",
+      "reference": "I don't think this will actually align the indexes, if you have > 100 tabs, the tabs over 100 will be misaligned. In addition with less than 10 tabs, there will be a pointless space."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `v4.signing.go`",
+      "reference": "hrm, unsigned headers now won't be included."
+    },
+    {
+      "prediction": "I think this should be `'' !== $value` instead of `strlen($value) < 256`.",
+      "reference": "what if `$value = ' ' `?"
+    },
+    {
+      "prediction": "I think this should be `if (files.size() == 0)`",
+      "reference": "What about using `!files.hasNext` instead? I'm not sure we want to assume that the iterable can be consumed multiple times. Plus there's no need to consume the entire iterable just to check whether it is empty."
+    },
+    {
+      "prediction": "This should be `const ItemType& it = parent->getItemType(); return parent->getNameDescription(it, this, -1, false);`",
+      "reference": "I think we could add bool addArticle here defaulted to false like its done in item class so std::string Container::getName(bool addArticle /* = false*/) const { and pass that variable to getNameDescription call"
+    },
+    {
+      "prediction": "I think it would be better to keep the `axes` and `type` parameters as `int` instead of `int[]` to be consistent with the rest of the code.",
+      "reference": "What if a library function on the outside still calls the old method? It is then no longer blocked properly. Have you tested the change?"
+    },
+    {
+      "prediction": "I'm not sure if it's a good idea to call `h2o_buffer_reserve` and `h2o_buffer_init` on the same call-site. I think it would be better to call one of them only if `inbufs[i].fill_cb` is non-NULL.",
+      "reference": "@i110 Do you have an idea on how we should propagate errors that occur in this block? The error condition happens when i) `h2o_buffer_reserve` fails to allocate space (happens in master as well), or ii) `fill_cb` returns an error (unique to this PR)."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Does this mean the times will show up in emails as UTC?"
+    },
+    {
+      "prediction": "I don't think we need to add `line_color='black'` to `Options('style', color=Cycle(), line_color='black')`",
+      "reference": "What's the motivation for having one of these be a cycle and the other be a fixed color?"
+    },
+    {
+      "prediction": "I think this should be `foreach (var line in table.ToMarkdown().Split(new string[] { Environment.NewLine },StringComparer.InvariantCultureIgnoreCase).Split(Environment.NewLine))`",
+      "reference": "What will be an empty entry now?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this method. I think it should be moved to `BasePlugin#isEnabled` method.",
+      "reference": "Can I ask you to add jsdoc tag that would mark the prop as `@private`?"
+    },
+    {
+      "prediction": "I think this should be `os.IsNotExist(err)` instead of `err != nil`",
+      "reference": "Do you need this? Don't think you use it?"
+    },
+    {
+      "prediction": "I think this should be `android.widget.RemoteViewsService` instead of `com.fsck.k9.widget.list.MessageListWidgetService`",
+      "reference": "`this.` seems unnecessary"
+    },
+    {
+      "prediction": "I don't think we need to change this line.",
+      "reference": "nit: can be removed before merging..."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that the `id` passed to `$(...).appendTo('body')` is the same as the `id` passed to `$(...).appendTo('body')` below.",
+      "reference": "Maybe a single quote would be compatible with airbnb style."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it would be better to put it in `SessionMapImpl` since that is where the implementation of this method is.",
+      "reference": "This is not the right approach. The `Distributor` maintains a model of the current state of the Grid. That model already contains the information about every active session. We don't need to modify `SessionMap` to expose it further."
+    },
+    {
+      "prediction": "I think this test should be moved to `test/testutil.go` so that it can be reused by other tests.",
+      "reference": "@Marwes @wolffcm This is the file where I made a change the the expected output. Its a little bit weird, but I think that the final output makes sense for the most part."
+    },
+    {
+      "prediction": "I think this should be `return nil, fmt.Errorf(\"Neither SiteLocalSettingsPath nor SiteSettingsPath is set\")`",
+      "reference": "I think we probably need a util.Warning() here."
+    },
+    {
+      "prediction": "I think this should be `from url` instead of `from .url`",
+      "reference": "Let's make have a signature that's consistent with `PhysicalKey.from_path`. Users should also have access to PhysicalKey since that class is part of the API (e.g., `Package.resolve_hash`)."
+    },
+    {
+      "prediction": "I think this should be `exports` instead of `namespace`",
+      "reference": "What is this directive for?"
+    },
+    {
+      "prediction": "I think this should be `chained_comparisons`.",
+      "reference": "I'd rephrase it as `Simplify chained comparison between the operands`."
+    },
+    {
+      "prediction": "Why not 1?",
+      "reference": "so in the performance data fixtures, there will be references for the second domain only?"
+    },
+    {
+      "prediction": "I think this should be `LOG.info`",
+      "reference": "This is wildly unhelpful to users --- they need to know when a proxy has been registered."
+    },
+    {
+      "prediction": "I think this should be `hf.request.headers.pop(\":authority\", \"\")` instead of `hf.request.headers.pop(\":authority\", hf.request.pretty_host)`.",
+      "reference": "If there is no authority header (i.e. someone intentionally deleted it), I would argue we probably don't want a Host header in the replay either. How about we only add it if it exists, and do nothing otherwise?"
+    },
+    {
+      "prediction": "I think we can remove the `seelog.Warnf` call here as well.",
+      "reference": "is there any unit test that can be updated to verify this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this check. I think it should be in the `if (ctx->protected_mode) {` part of the `cb_lua_filter` function.",
+      "reference": "exiting at this point is leaking memory, take a look at the valgrind output. the sbuffer must be destroyed."
+    },
+    {
+      "prediction": "I think this should be `--pdftoppm` instead of `--pdftoppm3`",
+      "reference": "i think you want a different flag and help like `--ffmpeg` or something; and you need to mark any tests you want skipped"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "this needed to change? or it was just some formatting thing?"
+    },
+    {
+      "prediction": "I think this should be done in a separate commit.",
+      "reference": "What's the advantage of doing it this way over, say, changing the `.circleci/config.yaml` to have `TRACE` as the `root.log.level`?"
+    },
+    {
+      "prediction": "I think this should be `fetch_project_iam_policy` and `fetch_bigquery_iam_policy` should be renamed to `fetch_project_iam_policy` and `fetch_bigquery_iam_policy`",
+      "reference": "typo: if -> of"
+    },
+    {
+      "prediction": "I think this should be `if err != nil { return \"\", errors.Wrap(err, \"failed to convert Markdown input\") }`",
+      "reference": "It'll be interesting to see who or what this breaks..."
+    },
+    {
+      "prediction": "I think this should be `log.finest(\"In development mode. Copying required files in case we're using a WebDriver-backed Selenium\");`",
+      "reference": "This change is incorrect: the current log level is correct."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to have a separate logger for each test.",
+      "reference": "There are few cases like this where default logger is used from CLI/db tools. I would copy `NewTestLogger` to `NewCLILogger` and use it everywhere in CLI. In future these two might be different."
+    },
+    {
+      "prediction": "I think it would be better to return `null` instead of throwing an exception.",
+      "reference": "While it provides re-use, this method does not make sense in `AbstractAuthenticationToken` because it knows nothing of a hash key. Instead, we should move this to a private method within each subclass."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to keep `id(x)` as it was before, and just use `x.get(id_field)` in the `for` loop below.",
+      "reference": "index by memory address? I realize I don't understand why we don't just build a list :)"
+    },
+    {
+      "prediction": "I think we can remove this empty line.",
+      "reference": "Not part of this commit, but we thought that this may not be a good place for this url, as it is in the `/user/` namespace (effectively preventing us having a user called `lastfmscraper`, however rare it may be)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this instrumentation. I think it should be in `System.Web.Mvc.Instrumentation`, not `System.Web.Mvc.Instrumentation.Abstractions`.",
+      "reference": "Does not appear to be used."
+    },
+    {
+      "prediction": "I think this is the only place where we're using the `on` method, so it should be the only place where we're using the `execute_powershell_script_on` method.",
+      "reference": "I think this starts a new `powershell.exe` interpreter each time, which is pretty slow to run a single command. Can we just do `on(bolt, \"cmd /c #{bolt_command}\")`? /cc @Iristyle"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the collection name, but it's the only way I could get to work.",
+      "reference": "Can we add a `collection?` method to the `VideoPage` so we don't violate Law of Demeter here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "Deleting unused code."
+    },
+    {
+      "prediction": "Why do you put this before `bbox2delta` and `build_assigner`?",
+      "reference": "`PseudoSampler` can also be imported from `..bbox`"
+    },
+    {
+      "prediction": "I think we should use `github.com/ethereum/go-ethereum/core/types` instead of `protocol`",
+      "reference": "Could you move this import to where the other `github.com/algorand/go-algorand` imports are?"
+    },
+    {
+      "prediction": "I don't think you need to add `l2f` to the `load_layers` list here.",
+      "reference": "I'd rather not activate this protocol by default for now, unless there is a good reason for that."
+    },
+    {
+      "prediction": "I think we can remove the DataAccess class now, since it's not used anymore.",
+      "reference": "Remove these 2 blank lines."
+    },
+    {
+      "prediction": "I think this should be `debug` level so we don't flood the logs with this every time.",
+      "reference": "Want to remove this?"
+    },
+    {
+      "prediction": "I don't think we need this.",
+      "reference": "is this a golint requirement? huh. mixed feelings :)"
+    },
+    {
+      "prediction": "I think this is a BCryptPasswordEncoder test that should use BCryptPasswordEncoder.BCRYPT_VERSION_2Y",
+      "reference": "Generally with these tests please split up these tests so that there is a different test for each method"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that the `lazy_pull_flush` function is called from multiple threads, and the `unique_lock` is used to ensure that only one thread is modifying the `pulls` list at the same time. This means that if two threads call `lazy_pull_flush` at the same time, they will both modify the `pulls` list at the same time, and the `unique_lock` will not be able to keep track of who is modifying it. Instead, you should use a `std::set` or a `std::unordered_map` to store the `pulls` list, and then iterate over it in `lazy_pull_flush`. This should ensure that only one thread is modifying the `pulls` list at the same time, and the `unique_lock` should be able to keep track of who is modifying it.",
+      "reference": "Should something use the lock passed in?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I've been able to get this to work.",
+      "reference": "Is this dependant on some version of Docker? Do we need to include both or `volumes` is deprecated (or just wrong!?)?"
+    },
+    {
+      "prediction": "Is it possible that the highest term is not the same as the term of the leader? If so, we need to make sure that the highest term is the same as the term of the leader.",
+      "reference": "I know this is not new code, but should we change `leader.getState() == State.ACTIVE` to `leader.isActive(liveNodes)`?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to check if the node is an `AbstractAnyTypeDeclaration` and then check if the scope is a `ClassScope` and then check if the class name is not `null`.",
+      "reference": "Why not use ASTAnyTypeDeclaration? AbstractAnyTypeDeclaration is deprecated"
+    },
+    {
+      "prediction": "I think this should be `from .util import save` instead of `from .tools.util import save`",
+      "reference": "A blank line at the very end of each file should eliminate the \"No EOF\" warning we see above"
+    },
+    {
+      "prediction": "I think we can remove the `|| WorkflowConditionFailedError` check here.",
+      "reference": "These would both be more robust with `errors.As` so that we could safely chain errors."
+    },
+    {
+      "prediction": "I think this should be `valobj_typename.startswith(SwiftLanguageRuntime::GetCurrentMangledName(\"_TtCs23_ContiguousArrayStorage\") || valobj_typename.startswith(\"_TtCs23_ContiguousArrayStorage\") || valobj_typename.startswith(\"Swift._ContiguousArrayStorage\"))`",
+      "reference": "This is checking the same condition twice?"
+    },
+    {
+      "prediction": "I think it would be better to use `std::type_index(typeid(::Al::NCCLBackend));` instead of `std::type_index(typeid(::Al::MPICUDABackend));`",
+      "reference": "Aren't we able to handle the case where we have MPI-CUDA without NCCL, even if it's suboptimal?"
+    },
+    {
+      "prediction": "I think this should be `static const uint32 PhantomId = 999999;` instead of `static const uint32 PhantomStatId = 999999;`",
+      "reference": "Wouldn't we want to namespace this `AA` or something similar"
+    },
+    {
+      "prediction": "I think this should be `envCfg.EnableSharedConfig` instead of `envCfg.Creds.HasKeys()`?",
+      "reference": "This probably should look beyond envConfig, and include `aws.Config.Credentials` as well."
+    },
+    {
+      "prediction": "I think this should be called `NotEnoughBalanceFilter` instead of `NotEnoughTransactionsBalanceFilter`",
+      "reference": "How does this filter differ from TooExpensiveTxFilter?"
+    },
+    {
+      "prediction": "This method is not used anywhere else, so I think it should be removed.",
+      "reference": "Use \" List<String>\" instead of ArrayList<String> in declaration."
+    },
+    {
+      "prediction": "I think this should be `runtime_type.name` instead of `runtime_type.name.name`",
+      "reference": "we might consider hard throwing when name is None since that is explicitly not working right now and then linking to issue in the exception error message"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in `org.infinispan.commons.persistence.jpa`",
+      "reference": "Just noticed this - why \"like\" and not straight \"=\""
+    },
+    {
+      "prediction": "I don't think we need to check for `commit` and `commit_range` here, since they're only used in the `describe` block below.",
+      "reference": "is this a new test? if yes, isn't it better to check for each env var is present, and the value, instead of a count?"
+    },
+    {
+      "prediction": "I think it would be better to use `AppVersion` instead of `MicroserviceVersion` here.",
+      "reference": "when will lost exception? by my test, never happened."
+    },
+    {
+      "prediction": "I think this should be `url.toLocalString()`.",
+      "reference": "You should re-stringify it here with `QUrl.FullyEncoded`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should go in `Wrap`.",
+      "reference": "i'm not sure why the quit channel is needed in this context"
+    },
+    {
+      "prediction": "I think this should be `resource.MustParse(\"1000Mi\")` instead of `resource.MustParse(args.MemoryRequest)`.",
+      "reference": "When `args.MemoryLimit` is empty, will `MustParse` panic?"
+    },
+    {
+      "prediction": "I think this should be `r.inSync = false` instead of `r.inSync = true`",
+      "reference": "How soon after this will Apply() be called? It would be a shame if there was still a significant delay before a missing or superfluous route was corrected."
+    },
+    {
+      "prediction": "I think this should be `uint32` instead of `int`.",
+      "reference": "I add a `idx` property, and it is only use for benchmark `heap.Remove`, is it acceptable?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. This is not a part of NV Access, it is a part of NVDA as a whole.",
+      "reference": "daemonic t got inserted."
+    },
+    {
+      "prediction": "I think this should be `Specify a different data directory with --data-path or specify the original genesis file with --genesis-file.`",
+      "reference": "This edit is non-essential."
+    },
+    {
+      "prediction": "I think we can remove this function altogether and just use `Object.setPrototypeOf`.",
+      "reference": "Might be safer to ponyfill this?"
+    },
+    {
+      "prediction": "I think this should be `os.MkdirAll(filepath.Join(dir, backend.Paths.Config), backend.Modes.Dir)`",
+      "reference": "why was this join not necessary before?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems like it should be enough to get the point across.",
+      "reference": "Here we compare to the expected call count 3. Because 3 TermsEnum are created: it is a PrefixQuery and there is one TermsEnum created for AutomatonQuery.intersect() (the next call timeout check is skipped once), then 2 TermsEnum created for the 2 matching terms \"one\" and \"ones\"). Would it be clearer to have a separate test method? We could index more docs, for example 50 with a prefix-generated term (e.g. \"term\"+increment). There would be a PrefixQuery for \"term\", the same test code, and we would test the TIMEOUT_CHECK_SAMPLING: - 1 TermsEnum for Automaton.intersect(), and the next calls would be sampled (50/TIMEOUT_CHECK_SAMPLING=3) => call count +4 - 1 TermsEnum for each enumerated term => call count +50 We could verify call count = 54 (instead of 101 without sampling)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `config/initializers/bolt.rb`",
+      "reference": "Add `--format` as an option. Does it make sense to have all display options available? Also need to add `--inventoryfile` and possibly the global_config_options."
+    },
+    {
+      "prediction": "Nit: I think it would be better to call this `RecordExceptionWhen` instead of `RecordException`.",
+      "reference": "Any thoughts on making this false by default? My suggestion is make this opt-in. Storing exception is somewhat expensive, so lets do this only if users opt-in"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `parseResponse` can throw `IOException` if the server responds with a non-200 status code, so we need to check for that here as well.",
+      "reference": "usually I do either ` IOException|RuntimeException` or if doing Throwable, use propagateIfFatal"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. The problem is that `other` is a `CompositeOverlay`, which means it has many `Overlayable`s in it. So we can't just call `dynamic_mul` on the `Overlayable`s in `other`. Instead, we need to make `other` an `Overlayable`, and then call `dynamic_mul` on the `Overlayable`s in `other`.",
+      "reference": "Wondering whether it should be `self.clone`, or `other.clone` or maybe a new `DynamicMap` declaration entirely. I see this is in the condition where `other` is a `DynamicMap`but is this definitely right in terms of `kdims`? I need to think about it more..."
+    },
+    {
+      "prediction": "I don't think this is the right place to add security. I think it should be in the solr-info bean.",
+      "reference": "I chose a new category because over time I'd like to add metrics also for Authorization plugins and Auditlog plugins (all components registered in security.json). An alternative could have been `CONTAINER` I guess?"
+    },
+    {
+      "prediction": "I think this should be `IoTeXAddressLength` instead of `IotxAddressLength`",
+      "reference": "IoAddrLen Golang prefers short variable names"
+    },
+    {
+      "prediction": "I think this should be `$query = $this->getModelManager()->createQuery($this->getClass(), 'a', $this->root);`",
+      "reference": "PHPCR-ODM does not allow blank aliases.."
+    },
+    {
+      "prediction": "I think it would be better to use `context.getRequestType()` instead of `request.getRequestType()`.",
+      "reference": "This seems trivial to forge. Can we tie this into the authentication code paths somehow so that we can really trust that the request is coming from a server instead of just some application claiming to be a server?"
+    },
+    {
+      "prediction": "Is this needed? I don't see it being used anywhere in this file.",
+      "reference": "Why are you including `<Kokkos_Core.hpp>`?"
+    },
+    {
+      "prediction": "I think this should be `hs->qid_health = mk_mq_create(hs->ctx, \"/health\", cb_health, NULL);` instead of `hs->qid_health = mk_mq_create(hs->ctx, \"/health\", cb_health, NULL);`",
+      "reference": "So previously health check was consuming from the normal metrics queue, now we duplicate the data to a new queue? I feel like there should be more changes to ensure the health check handler reads from the new queue? @DrewZhang13"
+    },
+    {
+      "prediction": "Can we remove this line?",
+      "reference": "Yep, looks right; your new selector _should_ disqualify the rule from being XDP accelerated."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should be in `Microsoft.Extensions.Primitives` or `Microsoft.Extensions.Hosting`.",
+      "reference": "cc @anurse This exists now. Kestrel currently passes the current time to all callbacks as an optimization. Just an FYI."
+    },
+    {
+      "prediction": "I think this should be `raise ValueError(f\"mode={mode} is not an allowed option for normalization\")`",
+      "reference": "remove 3.8+ f-string format to allow use in google colab"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `sigframe_rt.c` since it's only used there.",
+      "reference": "Should we assert that xstate is aligned here?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to have a separate test for each of these.",
+      "reference": "any particular reason for suppressing this change?"
+    },
+    {
+      "prediction": "I think this should be called `DIFFICULTY_BOMB_REMOVED` instead of `DIFFICULTY_BOMB_ARE_REMOVED`.",
+      "reference": "Hot take: could this instead become `NO_DIFFICULTY_BOMB` and have` HOMESTEAD` wrap it? So this comes first then homestad with the bomb adjustment, making no difficulty the base case?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "I suggest to use a scheme like \"blobvar\". \"blob\" would potentially collide if we have another API that use the blob package as a driver."
+    },
+    {
+      "prediction": "I think this should be `latest_listen_ts = max(latest_listen) if len(latest_listen) > 0 else 0` instead of `latest_listen_ts = latest_listen[0].ts_since_epoch if len(latest_listen) > 0 else 0`",
+      "reference": "This assumes that the user has at least one listen, it should have a guard around it, because this won't always be the case"
+    },
+    {
+      "prediction": "I don't think this is a good summary of what is going on here.",
+      "reference": "Seems like a lot of unintended changes. Are your line break settings causing these?"
+    },
+    {
+      "prediction": "I think this should be `ca-certificates=2019.11.27=1` and not `ca-certificates=2019.11.27=0`",
+      "reference": "@StrikerRUS what if we just removed build numbers and did `r-pkgdown=1.3.0`? Won't we have to fix this again if the RTD build machines change?"
+    },
+    {
+      "prediction": "I don't think we need to change this line.",
+      "reference": "I'm not arguing with it, but how did you decide to make this change?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this.",
+      "reference": "1: D400 First line should end with '.', not 'd' 4: I102 copyright year is outdated, expected 2014 but got 2012 25:29: E126 continuation line over-indented for hanging indent 31: D103 Docstring missing 40: D103 Docstring missing 57: D103 Docstring missing 68: D103 Docstring missing 101: D103 Docstring missing 115: D401 First line should be imperative: 'Call', not 'Calls' 115: D400 First line should end with '.', not 't' 134: D400 First line should end with '.', not 'e' 134: D205 Blank line missing between one-line summary and description"
+    },
+    {
+      "prediction": "I think we should add a note here about the fact that this value is not guaranteed to be the same as the `preferred_chain` setting in acme/config.yaml.",
+      "reference": "Explain that this must match the `CommonName` of the first (or is it last?) certificate in the chain, where the first certificate is the chain root CA and the last certificate is the leaf certificate."
+    },
+    {
+      "prediction": "I think this should be `bool` and not `string`.",
+      "reference": "do we really need this behavour to be configurable?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "Hmm interesting. Do the tests check out on this?"
+    },
+    {
+      "prediction": "I think this should be 128 instead of 8.",
+      "reference": "For another PR, another `--X` candidate."
+    },
+    {
+      "prediction": "I think this should be renamed to something like \"Updates\" or \"Snapshots\" or \"Releases\" or something like that.",
+      "reference": "Do we need to add this constant for `group_windows.go`? Otherwise, I believe the build will fail on windows. Also what do you think of adding the emoji   at the end? The windows one could be plain."
+    },
+    {
+      "prediction": "I think this class should be renamed to `XpathExpression`.",
+      "reference": "this is a junit 3 import, not a junit 4"
+    },
+    {
+      "prediction": "I'm not sure this is the right test. I think we want to make sure we can start a hibernating cluster without any syncsets. If we can do that, then we should be able to start a cluster with syncsets.",
+      "reference": "shouldn't there be a reque expected here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this property. I think it should be in the `NLog.Layouts` assembly.",
+      "reference": "I prefer to make it `protected set`. What do you think?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `json:\"-\"` here.",
+      "reference": "Maybe rename this to JetStreamStoreDir?"
+    },
+    {
+      "prediction": "I think we can remove the TODO.",
+      "reference": "You can start by logging out the error"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. The fallback gas price should be calculated based on the base fee per gas, not the gas price of the head of the chain.",
+      "reference": "In general, what is the reason for the fallback price to be public? If it is public, why does it take baseFeePerGas instead of a block header?"
+    },
+    {
+      "prediction": "I think it would be better to use `Serializable` as the return type instead of `PartialFunction`.",
+      "reference": "follow-up of #2002"
+    },
+    {
+      "prediction": "I think this should be `conf.getTableScanConcurrency() / (conf.getPartitionPerSplit + 1)`",
+      "reference": "could you give some explanations of what problem do you want to solve by this change, and how/why it works?"
+    },
+    {
+      "prediction": "I don't think we need to make this `LongLivedMarshalByRefObject` since it's just a wrapper for `IXunitTestCase`.",
+      "reference": "LongLivedMarshalByRefObject exists in both the netstandard and netfx xunit?"
+    },
+    {
+      "prediction": "I think this should be `bridge.print_page(*options)`, since it's a method on the bridge object.",
+      "reference": "the bridge here isn't defined as an accessor / reader to try mask it better. So you need to directly call the iVar `@bridge` here."
+    },
+    {
+      "prediction": "I think this should be `cmd.Env = append(cmd.Env, env...)` and not `addDeathSig(cmd)`",
+      "reference": "Testing on linux with this change, containerd no longer stops when k3s stops."
+    },
+    {
+      "prediction": "I think this should be `AllowConnectionWindowIncreaseFunc` or `AllowConnectionWindowIncreaseFuncFunc` to be consistent with the other config keys.",
+      "reference": "Maybe [u]int64 to be consistent with other byte-valued vars in this struct?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `app/views/menus/shared/_menu.html.erb`",
+      "reference": "This brings up the question... do we need to start assigning users to the different use cases in some way? We don't want this link to show up for Ric, for example."
+    },
+    {
+      "prediction": "I don't think you need to create a new `Batch<Dtype>` object here. You can just use `prefetch_[i]` to access the underlying `Batch<Dtype>` object.",
+      "reference": "Unless I'm misunderstanding something, this should just be `prefetch_(param.data_param().prefetch())`, no? That should just use the default value if there's no explicit `data_param` or `data_param.prefetch` set, removing the need to also duplicate and hardcode `PREFETCH_COUNT`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle unknown transaction types. I think it would be better to have a function like `transactionType(tx *types.Transaction)` that returns the type of the transaction based on the type of `tx`",
+      "reference": "nit : move this into a default clause in the previous switch statement."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `@` and `\\\\` are not escaped by default. You need to add `@` and `\\\\` to the list of special characters that are escaped by default.",
+      "reference": "It should be return \"''\"; I will fix it"
+    },
+    {
+      "prediction": "Why do we need a separate type for this? Can't we just use the existing `NewMemberEvent` one?",
+      "reference": "Go standard linter hates `Id`s and likes `ID`s, please do not resist."
+    },
+    {
+      "prediction": "I'm not sure why we're retrying here, but it seems like it would be better to just always retry.",
+      "reference": "I don't think retrying here is as useful, since `git-checkout` doesn't access the network AFAIK, and if it fails once it's probably going to fail all three times."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the renewal time. I think it should be in the `Renewable` field of the `Certificate` object.",
+      "reference": "Fixing this typo caused a build failure in `cmd/ctl/pkg/status/certificate/certificate_test.go:238` Could you also update it there?"
+    },
+    {
+      "prediction": "I don't think you need the `extension_condition` function here. You can just use `pkt.has_field(2)`.",
+      "reference": "Please note that adding this will import `extension_condition` in the entire scapy (as you are located in a layer). You might want to move this inside the RTP class ? Would be better i think"
+    },
+    {
+      "prediction": "I think this should be `options[:seed] = seed.to_s` instead of `options[:seed] = seed`",
+      "reference": "This will need updating, as I'm not sure we'll want to turn on random ordering if they've specified a seed for programmatic use."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `privmod_static[i].init` and `privmod_static[i].cleanup`.",
+      "reference": "Are you sure these are all client libs? This seems fragile: if we later load sthg else early this will break. Seems best to have some kind of sanity check: at least add an assert in the current location where this is set."
+    },
+    {
+      "prediction": "I think this should be a `Dictionary<string, string>` instead of a `Map<string, string>`.",
+      "reference": "Picking nits here, but as a Dictionary already is a collection of KeyValuePairs there's undoubtably a way to code this that doesn't involve constructing any new ones."
+    },
+    {
+      "prediction": "I don't think this is a good change. It's not a good idea to import the same class again and again in the same file. I think it's better to create a new file with the same name, but with a different name.",
+      "reference": "As Miklas stated, please avoid wildcard imports."
+    },
+    {
+      "prediction": "I think this should be `const { app, companionOptions } = server()```",
+      "reference": "Maybe change it to `module.exports = server` ? Becase there's nothing else to export here"
+    },
+    {
+      "prediction": "Is this change necessary?",
+      "reference": "This may be an automated change, but the import seems to fit better where it was previously."
+    },
+    {
+      "prediction": "Do we need this? I don't see it being used anywhere.",
+      "reference": "Might we want to separate including this snippet and sending the data for signed in users into different partials?"
+    },
+    {
+      "prediction": "I think it would be better to have a `ClosestPeerer` interface that has a `ClosestPeer` method. The `ClosestPeer` method can call the `ClosestPeer` method on the given `addr`.",
+      "reference": "closest to the address"
+    },
+    {
+      "prediction": "I think it would be better to say that by default it's disabled.",
+      "reference": "It should be much more clear for end-user"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to just use a `namedtuple` for this.",
+      "reference": "Isn't this an example that this will be prone to false positives ? We had nothing to fix in the codebase (And I think there is dict that could be named tuple in the MessageStore/MessageIdStore), but we have a false positives on our checkers."
+    },
+    {
+      "prediction": "I'm not sure if this test is still relevant. I think we can remove it now.",
+      "reference": "We control `octest`, so it shouldn't be necessary to provide struct literal keys. This is a case of vet being too picky."
+    },
+    {
+      "prediction": "I think this should be `50` instead of `50.00`",
+      "reference": "I would put that constant under config/initializers/constant.rb and I would leave application.rb for rails specific."
+    },
+    {
+      "prediction": "I think this is the only place where `doNotUnref` is used, so I think it should be moved into the `addUnrefToFinalResOp` function.",
+      "reference": "Is there a reason not to do something like `if _, ok := doNotUnref[ptr.ID]; ok { return ops }` here?"
+    },
+    {
+      "prediction": "I think this should be `self._execute(Command.GET_ELEMENT_RECT, return_base64=True)`",
+      "reference": "This would return a tuple of two dictionaries. You need to combine them and return a dictionary"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in `nni/algorithms/optimizers.py`",
+      "reference": "Why not always forward it to `_data`?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure I like the fact that we're sending POST requests to `/refresh-spotify-token` when we could be sending a `PATCH` request to `/spotify/token`.",
+      "reference": "there are tests that use `client.get` which will fail now"
+    },
+    {
+      "prediction": "I think this should be `libkbfs.WriteToReadonlyNodeError` instead of `libkbfs.WriteAccessError`",
+      "reference": "Whoa, TIL about this syntax."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `applicationErrorHandler` function in the `applicationErrorHandler.go` file.",
+      "reference": "This is a smell. When is this valid? Perhaps this should be `nil, nil, ErrNoResponse` or some such. While it is usually invalid to have neither a return value or error, we rarely check for the case in other idiomatic Go. Nothing wrong with extra defenses here, but think about it."
+    },
+    {
+      "prediction": "I don't think this line is needed.",
+      "reference": "This URL is outdated, add a `TODO` to fix it."
+    },
+    {
+      "prediction": "I think we can remove the `ChainID` and `SetChainID`",
+      "reference": "`ChainID()` is a getter method, move up to `Nonce()`"
+    },
+    {
+      "prediction": "I think it would be better to move this into the `catch` block above.",
+      "reference": "If this condition is not met, shall we just throw the exception rather than logging it? @li-ygerchikov Your thoughts?"
+    },
+    {
+      "prediction": "I'm not sure if this is a requirement or not, but it seems like it would be better to add this as a dependency in the `tests` group, since that's where the rest of the tests depend on yaml.",
+      "reference": "nit: alpha-sort `ruamel.yaml` in this list please"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for the default value of `lookback` and `openFileCost`.",
+      "reference": "Remove this because we already added it in another pr"
+    },
+    {
+      "prediction": "I don't think this is the right solution. The problem is that the shortcut icon is not always located in the same folder as the shortcut file.",
+      "reference": "Can we maybe move the `File` creation to a new method? I mean both changes look equal to me with the difference of two parts, which could be specified using two input parameters."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I think we want to require the association with the user...?"
+    },
+    {
+      "prediction": "I'm not sure this is the correct way to determine a signed messages size on chain.",
+      "reference": "I am unsure the below implementation is correct, but need this method to not panic else it borks the chain-validation tests."
+    },
+    {
+      "prediction": "I think we should use `collections.OrderedDict` instead of `OrderedDict` here.",
+      "reference": "nit: no need this change anymore."
+    },
+    {
+      "prediction": "I think this should be `$expr_type->parent_nodes[$source->id] = $source;`",
+      "reference": "This isn't really that closely related to everything else in this PR, but I'm pretty sure it's a correct change. I'm not familiar enough with the taint feature to know how to test it, but I worry that without this change some of the taint stuff might mess up the `parent_nodes` for a reference."
+    },
+    {
+      "prediction": "`Vector` is not used in this file.",
+      "reference": "[Checkstyle] ERROR: Unused import - javaslang.collection.Stream."
+    },
+    {
+      "prediction": "I think this should be `count(*) != 4`",
+      "reference": "The numbers were changed since now _root_ is also added to the doc, increasing the field count by 1."
+    },
+    {
+      "prediction": "Why is this removed? It seems to me that this is the only place where it is used, so it should not be removed.",
+      "reference": "TTemplateParamClass is handled above with the same content in the conditional"
+    },
+    {
+      "prediction": "I think this should be `unset($newConfig['Captcha'][$key], $newConfig['Captcha']['recaptcha_' . $key]);`",
+      "reference": "Do we need to add `theme` here?"
+    },
+    {
+      "prediction": "I think we can remove the `InternetGatewayID` field from the `Network` struct.",
+      "reference": "i think you have to regen the api stuff, `make generate`"
+    },
+    {
+      "prediction": "I think this should be `final` as it's only used once in this method.",
+      "reference": "DataHash shouldn't be needed here"
+    },
+    {
+      "prediction": "I think this should be `LOG.info(String.format(\"Command request: %s%s on session %s\", cmd, printableArgs, sessionId));`",
+      "reference": "`info` was chosen deliberately to mirror the old behaviour that users expected."
+    },
+    {
+      "prediction": "I'm not sure this is the right change, but I don't have a better one right now.",
+      "reference": "For some reason, Safari handles ember arrays differently than Chrome/Firefox. To get around that, we cast the ember array to a regular array to verify the deep equal. (Note: I tried wrapping the expected value in an emberA - which would look like `expect(someEmberArray).to.deep.equal(emberA(['a', 'b']));` but that didn't work either)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to store icebergType. I think icebergType should be stored in icebergTypeHolder. I think icebergTypeHolder should be initialized with icebergType when icebergTypeHolder is null, and icebergType should be set to icebergTypeHolder.getValue() when icebergTypeHolder is not null.",
+      "reference": "`icebergType` indicates the corresponding iceberg type of the `FieldVector` `originalIcebergType` indicates what the real type we will read from the `FieldVector`. For example, the parquet int backed decimal, the `icebergType` should be integer, `originalIcebergType` is decimal. I intend to only keep the `originalIcebergType`, but listening to your ideas."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems to be the only place where we're using the `**` syntax, so I figured I'd add it here as well.",
+      "reference": "The double (and thus triple as seen below) pointer gives me pause. I could imagine providing some larger API surface to smooth it out, but it's likely not worth it."
+    },
+    {
+      "prediction": "I think we can remove this test now that we have `MongooseError` defined in `lib/error.js`",
+      "reference": "The test name as written was grammatically correct, but now it isn't. Both 'MongooseError' and 'instance' should be both singular or both plural."
+    },
+    {
+      "prediction": "Shouldn't this be `bundle == null ? false : bundle.getBoolean(IS_GLOBAL_STORE, false);`?",
+      "reference": "SmartStoreInspectorTest was throwing a NPE here"
+    },
+    {
+      "prediction": "I think this should be `CloseableIterable.transform(records, record -> new InternalRecordWrapper(deleteSchema.asStruct()).wrap(record));`",
+      "reference": "We prefer line wrapping that keeps arguments to the same method aligned rather than aligning arguments to different levels. Here, the lambda to create an `InternalRecordWrapper` is an argument to `transform`, but it is aligned with `deleteSchema.asStruct()` that is an argument to the outer `toEqualitySet` call. Instead, can you add a newline for each argument to `toEqualitySet`? If the line with the lambda is too long, then you can also add a newline for it that it indented from the start of the line with `transform`, so it is clear that it is an argument to `transform` and not `toEqualitySet`."
+    },
+    {
+      "prediction": "I think this should be `translateReason` instead of `translateReasons`",
+      "reference": "I think it would be a good idea to wrap this in an if like the stats above, just in case."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `selectAll` method.",
+      "reference": "@angmas1 just make one more change here too. Move the else line to the line above it, where the if block ends and remove the braces like in line 280."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `BlkioGroup` struct itself, and not in `BlkioGroup.Apply`",
+      "reference": "Looks like the actual rename is missing in this commit edit: Arf, nevermind, I was at the wrong commit"
+    },
+    {
+      "prediction": "I don't think this is the best way to do this. We need to make sure that we don't pass in any options that are different than the default options. I think we should add a check for `this.opts.getAssemblyOptions === defaultOptions.getAssemblyOptions` and throw an error if it's not passed in.",
+      "reference": "I'm a little stuck understanding, how would this result in an error when `this.opts.params` is not set? It will just not do `AssemblyOptions.validateParams(this.opts.params)`"
+    },
+    {
+      "prediction": "I think this should be `compute_service_account` instead of `compute_service_account_pattern`.",
+      "reference": "networkVarPattern (etc) then?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should go in `spec/fake_spec.rb`, not `spec/spec_helper.rb`.",
+      "reference": "I think it makes sense to do `metadata[:block].clone`, as thats what `new_metadata` is, but somehow it's missing the block from its clone."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it should go in `FunctionLikeAnalyzer`.",
+      "reference": "Just realized this should be fixed instead of changing here. Concatenating int should still result in lowercase string."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do this, but it's the only way I could get this to work.",
+      "reference": "For Se 3.x we shouldn't need to specify `marionette: true`? I'll see if I can figure out where this is getting used that it might be necessary here."
+    },
+    {
+      "prediction": "I think this should be `*chain.DefaultStore` and not `*chain.Store`",
+      "reference": "This should be renamed to `chain.Store`. It was named `DefaultStore` as an alternative to `StoreImpl`, which is the (bad) Java convention."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. If the device name is not given explicitly, its device name will be the same as disk name.",
+      "reference": "ad.Source is disk name? Shouldn't the order of initialization be reversed here then? First set ad.DeviceName, if it's empty, set ad.Source? Or this is a different use case? This is the third place where this device/disk default logic is implemented, so another reason to move it to regAttach, if it's possible to be generalized."
+    },
+    {
+      "prediction": "Should this be `int` instead of `int.MaxValue`?",
+      "reference": "typo: \"be hold\" -> \"be held\""
+    },
+    {
+      "prediction": "I think this should be `long.MinValue` instead of `1`",
+      "reference": "I do not think we should have any logic inside the null storage. 1 has a very specific magical meaning. We should either return null here or long.maxValue."
+    },
+    {
+      "prediction": "I think we should bump this to `>=1.15.1`",
+      "reference": "Why did this change?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in `PlansController#edit_plan`, but I'm not sure if that's the right place to put it.",
+      "reference": "Removed this method. Doesn't seem to be used in the code, and the variable set here doesn't seem to be used in the views."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Remove the initializers (\"`= null`\") from both parameters. Remember that that's why we added the overloads: to ensure that all languages could have the convenience of the abbreviated argument lists (not all languages understand default parameters)."
+    },
+    {
+      "prediction": "I think this should be `[]float64` instead of `[]float64` and `[]int` instead of `[]int64`",
+      "reference": "I think you can just remove most of these attributes outside of `ElementType()`. Instead, this should expose the arrow array. We're not going to use these methods anyway."
+    },
+    {
+      "prediction": "I think this should be `import { remove } from './remove';`",
+      "reference": "Thinking this should be `removeNode()` exported from `dom/index.js`. Thoughts?"
+    },
+    {
+      "prediction": "This should be `memoryCpuLimitsEnabled`",
+      "reference": "Does this need to be saved in the state file? Or are we recomputing when the task is read from the state file?"
+    },
+    {
+      "prediction": "I think this should be `st.CacheHitRate = float64(st.NumMatches) / float64(st.NumInserts)`",
+      "reference": "Looks like `s.inserts` and `s.removes` are updated under sublist's lock, so I don't think you need atomic for those 2."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems to be the only place I could think of where this would be useful.",
+      "reference": "@TD-er use bitRead/bitWrite macros, do you use here inversed logic?"
+    },
+    {
+      "prediction": "Why is this being imported here?",
+      "reference": "The previous `fetch` mock was exposed globally before - can we do the same with `fetchMock` so we don't need to import it in every file?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the behavior as it was before, which is to return `MessageBody.ZeroContentLengthClose` when `context.EndStreamReceived` is true. This way, we don't need to make any changes in the `MessageBody` class.",
+      "reference": "Is ExpectData still used anywhere?"
+    },
+    {
+      "prediction": "I think this should be `[^.]` instead of `.[^.]`",
+      "reference": "Do you want to allow the character `.` or any character? Because inside a regex `.` means any character, if you want to authorize `.` then you need to add `\\.` ."
+    },
+    {
+      "prediction": "I think this test is a bit too specific to this PR. Can we move it to a separate PR?",
+      "reference": "`eachAsync()` should pass a doc, not an array of docs, to the callback. This would be a massive breaking change."
+    },
+    {
+      "prediction": "I think this should be `klog.Infof(\"message: %s process successfully\", msg.GetOperation())`",
+      "reference": "please start log with caps letters"
+    },
+    {
+      "prediction": "Do we need this? I don't see it being used anywhere in the PR.",
+      "reference": "Is this where we want it in relation to the other scripts?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this feature. I think it should be in `core/feature.js`",
+      "reference": "Please wrap this variable within a self invoking anonymous function. (function(){ FEATURE_NAME = \"reports\"; //and rest of the js goes here. })();"
+    },
+    {
+      "prediction": "I think this should be `c.cgroupManager.Freeze(configs.Thawed)` instead of `_ = c.cgroupManager.Freeze(configs.Thawed)`",
+      "reference": "Haven't tried, but what happens currently if `-f` is used and it's not running? (thinking: `-f` should ignore the case and just proceed?) I see we have a special case for `all`"
+    },
+    {
+      "prediction": "I think this should be `throw new MongoURIError('skipping SSL tests, MONGODB_URI environment variable is not defined')`",
+      "reference": "no need for custom errors in tests unless the tests are intended to mock a specific sort of error"
+    },
+    {
+      "prediction": "Can we use `kubernetes.io/role/cluster-admin` and `kubernetes.io/role/cluster-admins` instead of `kubernetes.io/role.kubernetes.io/cluster-admin` and `kubernetes.io/role.kubernetes.io/cluster-admins`?",
+      "reference": "This const has already existed in the code, no need to define a new one"
+    },
+    {
+      "prediction": "I think this should be `logging.Base().Errorf(format, args...)`",
+      "reference": "Would Errorf be better than Warnf?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `getFilesContent` will return an empty string if the path doesn't exist. I think you need to check if the path exists and throw an exception if it doesn't.",
+      "reference": "A good candidate for StringBuilder? Also make sure to add white space around the operators (`path + \"/\" + file`)"
+    },
+    {
+      "prediction": "I think this method should be renamed to `uploadFlowFile` and `getUploadedFlowFile` to be consistent with the rest of the API.",
+      "reference": "Can we have the same argument order as `getUploadedFlowFile`?"
+    },
+    {
+      "prediction": "I think we should rename this to `IdleTaskQueueCheckInterval`",
+      "reference": "this dynamic config should still be valid, i.e. operator should have the ability to control each individual task queue just in case"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. This is for the case where you are in the middle of a document and you want to scroll into view. I think this should be in the `DocumentTextInfo` class.",
+      "reference": "Should these cases be logged? `if not obj` used to be"
+    },
+    {
+      "prediction": "I'm not sure if `sudo` is required here, but it seems like it should be. I'm also not sure if `awk` is required, but it seems like it should be as well.",
+      "reference": "Is it OK to assume that keys.gnupg.net has the new key?"
+    },
+    {
+      "prediction": "I think we should add this to the changelog.",
+      "reference": "A note for however is curious. I think these two notes add negative value only. In general we want people to implement this method (in particular after this patch). If we leave these old (and currently mis-rendered) notes, file system implementors might not implement this method referring to this."
+    },
+    {
+      "prediction": "I think this test should be renamed to `shouldKeepInstanceOfPriorityQueue`",
+      "reference": "Comparators (or functions in general) cannot be compared for equality. Therefore `PriorityQueue.of(comparator, ...)` always has to return a new instance."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `Concurrent::CachedThreadPool`",
+      "reference": "It's a little surprising that CachedThreadPool overrides the max_threads argument. This makes sense as a solution though."
+    },
+    {
+      "prediction": "`avg_factor` is not used in `weight_reduce_loss`, right?",
+      "reference": "```python # if avg_factor is not specified, just reduce the loss if avg_factor is None: loss = reduce_loss(loss, reduction) else: # if reduction is mean, then average the loss by avg_factor if reduction == 'mean': loss = loss.sum() / avg_factor # if reduction is 'none', then do nothing, otherwise raise an error elif reduction != 'none': raise ValueError('avg_factor can not be used with reduction=\"sum\"')"
+    },
+    {
+      "prediction": "Why do we need to set stride=1 here?",
+      "reference": "`padding` should be 0 instead. Actually you can just remove padding and dilation to use default values."
+    },
+    {
+      "prediction": "I think this should be done in a separate commit.",
+      "reference": "$balance is initialized to 0 above in line 1428 and then assigned in line 1437. If we're no longer manipulating that value, is there a reason to use a variable at all? It might make the overall code less confusing to make some changes there."
+    },
+    {
+      "prediction": "`kind` should be `kind='bar'` and `colormap` should be `colormap='Paired'`",
+      "reference": "Why do we need to explicitly use keyword arguments?"
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "Might still want to log this as `TargetBuilding` but with a slightly different message?"
+    },
+    {
+      "prediction": "This should be `const std::vector<std::string> &` instead of `std::vector<std::string> &`.",
+      "reference": "ref. above, this also can be omitted in case `txt = std::move(other.txt);` could work (or copy), consider `std::array<String, VARS_PER_TASK>;`? or a custom object implementing `Object& operator=(Object&&) noexcept;'"
+    },
+    {
+      "prediction": "I don't think you need to import astroid here, since you're not using astroid in this file.",
+      "reference": "Well done, we forget that one apparentely."
+    },
+    {
+      "prediction": "I think this should be `dask` instead of `dask.array`",
+      "reference": "Guess it isn't used. The dask thing was just a prototype so removing it is probably the right thing to do."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "This code is correct, but it's a bit confusing to see how (e.g., without the if startswith(\".') the or below would break.) I think it will be clearer for the long run if you refactor this just a bit. CONTENT_INDEX_EXTS (all caps) looks like a constant, but is now being set by the environment. Instead, replace the reference to CONTENT_INDEX_EXTS in index.py with a simple call to a method \"get_content_index_exts\" (or similar). Then you can write all the logic into one clean function."
+    },
+    {
+      "prediction": "I think it would be better to have a constant for the default value of `principal_state_updater_frequency` instead of hard-coding it here.",
+      "reference": "please update the zms conf file to have a small description of these options since by default feature is off and must be enabled"
+    },
+    {
+      "prediction": "I think we should throw an error if both `lazy` and `waitForThumbnailsBeforeUpload` are set.",
+      "reference": "Should be `this.opts` instead of `defaultOptions`"
+    },
+    {
+      "prediction": "I think this should be `users` and not `users`",
+      "reference": "You can reuse `USERS` above as well."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to make this change. I think it should go in the same commit as the other changes.",
+      "reference": "Whoa, was this opto breaking something? I guess if this final path is filtered out, we still need the queue? Have you run the suggest benchmarks to see if removing this opto hurt performance?"
+    },
+    {
+      "prediction": "`List.ofAll` and `Stream.ofAll` are not the same. `List.ofAll` returns a new list, while `Stream.ofAll` modifies the original iterable.",
+      "reference": "I'm not sure about this, please check the usages. It's only used currently to reverse it, maybe we should eliminate this method completely instead."
+    },
+    {
+      "prediction": "I think this should be `protected function listTagsEnabled()`: `protected` means it can only be called from within this class, not from the `AbstractBase` class.",
+      "reference": "Since this is only used in the MyResearchController, does it need to be placed at the AbstractBase level?"
+    },
+    {
+      "prediction": "@deads2k I think this should be renamed to `PhaseSelector`",
+      "reference": "why `PhaseSelector` ?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I'm not sure if we want people to be able to specify their own list of modules that they don't want to include in the analysis.",
+      "reference": "Does` AllowlistedModules` have the same meaning of `WhitelistedModules`?"
+    },
+    {
+      "prediction": "I think this should be `list<array>`, not `list<array>`.",
+      "reference": "I missed the $preserve_keys=true case when adding this to Phan. For psalm, two separate signatures may make sense"
+    },
+    {
+      "prediction": "I don't think we need to import `DynamicPropertyFactory` here, since it is not used in this PR.",
+      "reference": "TcpServer will be created for each HighwayServerVerticle instance so the counter number in server is not correct."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "Why change this from a foreach? I can't see it gaining anything here and code styles shouldn't change just for the sake of it."
+    },
+    {
+      "prediction": "I think this should be `if len(subsystem) == 0 { return hybridMountpoint, nil }`",
+      "reference": "I'd say \"cgroup2 hybrid path\" instead."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Bolt::Configuration` class, since that's where the configuration is stored.",
+      "reference": "I dont see a reason to print URI for `Target`. we do not expose that in inventory v1"
+    },
+    {
+      "prediction": "I'm not sure if we want to make this `bool` or `int`, but I think it makes sense to make it `bool`.",
+      "reference": "Do we still need this?"
+    },
+    {
+      "prediction": "I think this should be `campaignSourceString` since it's a string.",
+      "reference": "`campaignSource` is unused (from `structcheck`)"
+    },
+    {
+      "prediction": "I think this should be `0.2.1-SNAPSHOT` instead of `0.2.1`",
+      "reference": "we should discuss how to not hard-code this version"
+    },
+    {
+      "prediction": "I think we can remove `fullDocument` from here.",
+      "reference": "do we have a doclet for this event?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I've been able to get it to work.",
+      "reference": "Why don't we want to catch segfaults by default?"
+    },
+    {
+      "prediction": "I'm not sure about this suppression, but I think we should be able to get away without it. The `Dispose` method is marked as `[System.Diagnostics.CodeAnalysis.DynamicallyAccessedMembersAttribute(System.Diagnostics.CodeAnalysis.DynamicallyAccessedMemberTypes.PublicConstructors | System.Diagnostics.CodeAnalysis.DynamicallyAccessedMemberTypes.PublicFields | System.Diagnostics.CodeAnalysis.DynamicallyAccessedMemberTypes.PublicProperties)]` and the `Dispose` method is marked as `[System.Diagnostics.CodeAnalysis.DynamicallyAccessedMembersAttribute(System.Diagnostics.CodeAnalysis.DynamicallyAccessedMemberTypes.PublicFields | System.Diagnostics.CodeAnalysis.DynamicallyAccessedMemberTypes.PublicProperties)]`",
+      "reference": "You could probably dispose request & content if you `await` the SendAsync."
+    },
+    {
+      "prediction": "I think this test is a bit too specific to the `to_html` function. I think it would be better to have a test for `to_clipboard` as well, or at least to have a test for `to_html` that uses `to_clipboard` as a helper.",
+      "reference": "why is this test case here? this file is for testing functionalities in utils.py"
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "reference": "why do we have an env var for this? not sure why we'd want to suppress welcome emails but not any others"
+    },
+    {
+      "prediction": "I think this should be `p.errs = append(p.errs, ast.Error{...})` instead of `p.errs = append(p.errs, errs...)`",
+      "reference": "drop the todo?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Maybe use { } here like we are most other places now."
+    },
+    {
+      "prediction": "I think this should be `partitionType.fields().get(pos).fieldType()`.",
+      "reference": "I want to get PartitionData field type, I don't know how to get it in other way."
+    },
+    {
+      "prediction": "nit: I think it would be better to call this `PutEx` since it's a little bit more specific.",
+      "reference": "I think that a single `Get` method would be preferable, that would have the following parameters: Get(addr basics.Address, withPendingRewards bool, cidx basics.CreatableIndex, ctype basics.CreatableType) where we ignore cidx of -1, and adding support for ctype of \"AssetParams\" or something like that. (i.e. so that this Get call would be good for asset holding/asset params and applications )"
+    },
+    {
+      "prediction": "I think this should be `if (((len == 0) && (tgtAttr && (NOT DFS2REC::isSQLVarChar(tgtAttr->getDatatype())))) || (len > 0) && (memcmp(sourceData, HIVE_DEFAULT_NULL_STRING, len) == 0))`",
+      "reference": "This can handle the '\\\\' problem, but if (not possible) the user define '\\\\' as null, this logic will break. So here, it should be to compare the whole HIVE_DEFAULT_NULL_STRING, and make sure length is equal, as I understand. \\N is NULL, but \\NA is not NULL, what will happen if there is \\NAA ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this policy. I think it should be in the `Observation` model, since it's specific to that model.",
+      "reference": "Hmm, I would think that anyone who can edit the request should be able to delete the observation, in case they accidentally add the wrong person or something."
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "I know we are doing this other places already. It would be good to refactor this and the paginable publicly_visible so that we are DRY. Can wait until after MVP though when we do general cleanup"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `ibl_table_update()` function, since that is the place where `ibl_table_update()` is actually called.",
+      "reference": "I think this should read \"is always observed before\" or \"is never observed after\"."
+    },
+    {
+      "prediction": "I'm not sure if this is the correct way to do it.",
+      "reference": "You might consider adding yourself as an author of the class."
+    },
+    {
+      "prediction": "I think this should be `freopen` instead of ` freopen`",
+      "reference": "If this fails, then 'stderr' is not a valid file handler anymore, and then any further 'write' operations will fail. Maybe some handling of such situation should be added here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to get the timestamp from the `Span` object itself.",
+      "reference": "Break out tests for events into their own tests rather than placing them in other ones. That makes it easier for us to figure out where problems lie and to do a TDD-driven implementation over new APIs."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You need to check if the key exists in the dictionary, and if it does, you need to convert the value to a string. If it doesn't, you need to set the value to `None` and then convert the value to a string.",
+      "reference": "I think your indentation is wrong here."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to put it in the `L3RouteResolver` struct itself.",
+      "reference": "UT spotted that we weren't marking routes dirty when they targeted at Node and that node's IP changed. e.g., the case where a WEP appears in the syncer before the corresponding node does, so we don't know the node's IP. I added a new map to track the CIDRs for each node so that when the node IP changes we can mark those CIDRs dirty. Right now it's only used for the \"WorkloadIPs\" scenario. Need to decide if the same issue applies in the \"CalicoIPAM' case. I suspect it might. It looks like we have some logic present to detect when our own node IP changes, but we don't seem to handle when a remote node's IP changes."
+    },
+    {
+      "prediction": "I'm not sure why this is necessary, but it seems like it should be removed.",
+      "reference": "Given this will 'pause' bolt for a few seconds on older rubies, perhaps emit a debug message saying \"Warming up OpenSSL\" or something to that effect"
+    },
+    {
+      "prediction": "I think this should be `self._format = pkgformat or 'auto'` instead of `self._format = pkgformat`",
+      "reference": "Initializing _format to None, but asserting that it's not None later, seems unnecessarily fragile. We shouldn't architect the package class to rely on classes or methods that use it (e.g., build). Let's at least set the format to the default in case we don't create all packages through build.py."
+    },
+    {
+      "prediction": "I think this method should be renamed to `findElementsByXPath` or `findElementsByXPathInWebDriver` or something like that.",
+      "reference": "This change should also probably go into the corresponding method of the abstract By class?"
+    },
+    {
+      "prediction": "I don't think we should have a constant for this. We should have a constant for the error code.",
+      "reference": "Should this be set to something else?"
+    },
+    {
+      "prediction": "This file should not be touched.",
+      "reference": "Whats up with this? It looks like this is rendered on purchases/new for subscribers, so it would result in a dead end?"
+    },
+    {
+      "prediction": "I think it would be better to pass the `backend` to `propagate_ids` instead of passing `cls` to `propagate_ids`.",
+      "reference": "Glad to see this generalized to support the backend argument."
+    },
+    {
+      "prediction": "I think we should be able to get the deployment chains for a deployment without needing to know the deployment chain ID.",
+      "reference": "`ctx` is unused in ListDeploymentChains"
+    },
+    {
+      "prediction": "Should this be `fmt.Errorf` instead of `fmt.Errorf`?",
+      "reference": "I'd say the format like `\"xxx: %w\"` is more convention when wrapping an error basically. You refered to anything like this?"
+    },
+    {
+      "prediction": "I think this should be 10 seconds.",
+      "reference": "This seems relatively high, is it the recommended value from the etcd folks? Or is this debugging cruft?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `peertracker` package.",
+      "reference": "This requires updating the name in the `endpoints` tests, please run `make test`."
+    },
+    {
+      "prediction": "I think this should be `LOG(FATAL) << \"Meta Data Path should not be empty\";`",
+      "reference": "LOG(FATAL) means coredump, LOG(ERROR) is better here."
+    },
+    {
+      "prediction": "I think this should be `handleName` instead of `handle: string`.",
+      "reference": "/javascript/firefox-driver is the Selenium implementation of a WebDriver for Firefox. Since it generally isn't W3C compatible, it shouldn't change. We can just drop this change."
+    },
+    {
+      "prediction": "I think this should be `return nil, fmt.Errorf(\"unmarshal to %s: %w\", typeVal, err)`",
+      "reference": "It's a bit weird to have `svc.go` to include a `ScheduledJobType`...should we rename this file?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "For some reason `protoc` doesn't copy license header from `proto` files to generated code. But this code will never be checked in, so it is ok."
+    },
+    {
+      "prediction": "I think this should be `if (trList2.size() > 0 && old.getPumpId() == 0) {`",
+      "reference": "I don't know what the implications of this change are for pumps other than the insight but i would add `|| trList2.get(0).pumpId == temporaryBasal.pumpId` in case we see the same pump event again, in order to not duplicate it in the database."
+    },
+    {
+      "prediction": "I think we should rename this to `AddressWithChecksum`",
+      "reference": "is there anything still referencing `checksumAddress` or can we just delete it?"
+    },
+    {
+      "prediction": "I think we should remove this condition from here.",
+      "reference": "The reason I didn't suggest it before is because I didn't want alerts for every tried - but not updated provision, but I can see a value in it from OSD perspective"
+    },
+    {
+      "prediction": "I think we can remove the `APP_DOMAIN` environment variable from the `config/initializers/application.rb` file as well.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to check for nilkeys. I think it's better to check for `a.Key == nil` instead.",
+      "reference": "I believe Key is not necessarily nil, it could be empty string. Probably better check a.Kind == Create"
+    },
+    {
+      "prediction": "I think we can remove the `T state` parameter from this method as well.",
+      "reference": "This is only used for headers and therefore isn't used."
+    },
+    {
+      "prediction": "Why is this here? I don't see it being used anywhere in the PR.",
+      "reference": "Wouldn't it make sense to add a files_import_command and a db_import_command, which could be empty? I guess that leads to potential backward-compatibility problems, but it's worth thinking about. Perhaps add an import-api version to solve that? Overall, I think the actual db import logic and files import logic should be moved into the yaml file. What you you think of that?"
+    },
+    {
+      "prediction": "I think we should rename this package to `csv_test` to be consistent with the rest of the test packages.",
+      "reference": "Why do we have this here? I'm not concerned about it really, just curious."
+    },
+    {
+      "prediction": "I think it would be better to have a `DeleteBytesAction` and `DeleteUintAction` instead of these two.",
+      "reference": "@justicz Why are these switched here?"
+    },
+    {
+      "prediction": "Line is too long. [85/80]",
+      "reference": "Maybe verify that an anchor was created -- that the markdown was processed?"
+    },
+    {
+      "prediction": "I think this should be `this.approximation = Iterables.getOnlyElement(twoPhaseIterator.approximation());`",
+      "reference": "Is this just a minor improvement or is it necessary? If just some minor improvement, I recommend you don't touch Lucene in a Solr PR."
+    },
+    {
+      "prediction": "I think we can remove this method as it's not used anymore.",
+      "reference": "why delete this method?"
+    },
+    {
+      "prediction": "I think it would be better to have a separate class for this. It would make it clear that this class is only used by Java.",
+      "reference": "Does this transform only the `MethodContext`, or also the `OutputContext`? (Looking at the other files, I gather it's the latter.) Might be helpful to mention that here."
+    },
+    {
+      "prediction": "I think it would be better to move this import to the top of the file.",
+      "reference": "Could you please clean-up this line while at it?"
+    },
+    {
+      "prediction": "I think this should be `ServiceAccountName` or `ServiceAccountName` to be consistent with the other fields.",
+      "reference": "no need to use a pointer. Just string and check for != \"\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set this. I think it should be in the `if (endpointObject.isHttp2Enabled()) {` block, since that's where we set the `useAlpn` option.",
+      "reference": "this means h2 mode? but how h2c can work?"
+    },
+    {
+      "prediction": "I don't think you need to add this file to the `appModules/` directory.",
+      "reference": "This line should not be there"
+    },
+    {
+      "prediction": "I think we can remove the lock here.",
+      "reference": "why do we need lock here? if needed, then we also need to lock in Start()?"
+    },
+    {
+      "prediction": "This should be `addGameTask(&Game::playerExecuteParsePacketEvent, player->getID(), recvbyte, new NetworkMessage(msg));`",
+      "reference": "this should be on default-switch, uhm? else you will handle twice (c++ and lua). I'm just saying..."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we should be able to add a new metric stream configuration for each metric stream we want to add. If we want to drop aggregations, we can just add a new configuration that has Aggregation = Aggregation.Drop. I don't think this is the right way to do it. I think we should be able to add a new metric stream configuration for each metric stream we want to add. If we want to drop aggregations, we can just add a new configuration that has Aggregation = Aggregation.Drop. I don't think this is the right way to do it. I think we should be able to add a new metric stream configuration for each metric stream we want to add. If we want to drop aggregations, we can just add a new configuration that has Aggregation = Aggregation.Drop. I don't think this is the right way to do it. I think we should be able to add a new metric stream configuration for each metric stream we want to add. If we want to drop aggregations, we can just add a new configuration that has Aggregation = Aggregation.Drop. I don't think this is the right way to do it. I think we should be able to",
+      "reference": "unrelated minor fix"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do this in the `setup` method.",
+      "reference": "can you add `, echo: false` to the end of this as well. I don't think we need to echo this to the log. You might need to update the spec as well."
+    },
+    {
+      "prediction": "I think this should be `assert headers[\"Accept\"] == \"foo/bar\" * 2`",
+      "reference": "This test does not really verify that we're doing the correct thing now, is it? It seems to pass with and without the patch."
+    },
+    {
+      "prediction": "I think it would be better to use `log.Error` instead of `log.FinishWithErr` here.",
+      "reference": "should just be able to defer the call directly here too"
+    },
+    {
+      "prediction": "I think this should be `y=1.05` instead of `y=1.0`",
+      "reference": "Is this a magic number? I assume this makes it look better."
+    },
+    {
+      "prediction": "I think this should be `uint64(gasLimit)` instead of `uint64(gasPrice.AsBigInt().Uint64())`",
+      "reference": "Would welcome a stringification method on big.Int in specs-actors"
+    },
+    {
+      "prediction": "I think we should rename this method to `PersistBlockLevelInfo` to be consistent with the other methods.",
+      "reference": "PersistBlockLevelInfo -> UpdateBlockLevelInfo"
+    },
+    {
+      "prediction": "`_base_` should be `BASE_` and `model` should be `model_`.",
+      "reference": "Type is redundant since it is inherited."
+    },
+    {
+      "prediction": "I think we can remove this method and just check `viewInfo != null` in `isNotView()`",
+      "reference": "better use `!isView()` so that we don't need to modify multiple lines in future."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `spec/common.rb` or `spec/initialization.rb` or `spec/runner.rb` or `spec/runner_test.rb` or `spec/world.rb` or `spec/world_test.rb` or `spec/definitions.rb` or `spec/definitions_test.rb` or `spec/config.rb` or `spec/config_test.rb` or `spec/runner.rb` or `spec/runner_test.rb` or `spec/world.rb` or `spec/world_test.rb` or `spec/definitions.rb` or `spec/definitions_test.rb` or `spec/config.rb` or `spec/config_test.rb` or `spec/runner.rb` or `spec/runner_test.rb` or `spec/world.rb` or `spec/world_test.rb` or `spec/definitions.rb` or `spec/definitions_test.rb` or `spec/config.rb` or `spec/config_test.rb` or `spec/runner.rb` or `spec/runner_test.rb` or `spec/world.rb` or `spec/world_test.rb` or `spec/definitions.rb` or `spec/definitions_test.rb` or `spec/config.rb` or `spec/config_test.rb` or",
+      "reference": "I know you don't usually open up the singleton class, but I was getting a `private method called` error when trying to use `self.attr_writer` - any idea?"
+    },
+    {
+      "prediction": "I think this should be `!==` instead of `===`",
+      "reference": "There's no need to create a new variable here, the other checks just use the item type directly."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [122/120]",
+      "reference": "Any other possible names? Feels like going with \"workshop\" for this product_type could increase the confusion between course/workshop in the codebase."
+    },
+    {
+      "prediction": "I think you can remove the `errorStoreActions` from here, since you are not using it anymore.",
+      "reference": "@eugene-manuilov Isn't this already taken care of by `fetchGetURLChannels` via `createFetchStore`? Why is the extra `receiveError` call needed here? (This was already in the IB, but just struck me while reviewing here.)"
+    },
+    {
+      "prediction": "I think this should be `import googlecloudprofiler as gcp`",
+      "reference": "nit: should go before `grpc`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `integration_test.go`",
+      "reference": "Why not using the Cancel? I thought cancelling a \"parent\" context would cascade to \"children\" contexts, cancelling them as well. In short, how does cancel work?"
+    },
+    {
+      "prediction": "I think this should be `strings.HasPrefix(ty, \"text/\") || strings.Contains(ty, \"html\")`",
+      "reference": "Could you explain why we default to binary now? It seems it's safer to default to text to avoid stuff getting executed or parsed by browsers accidentally. If we need to add support for particular types, we can add them explicitly right?"
+    },
+    {
+      "prediction": "I think this should be `string` and not `string;;local`",
+      "reference": "Bit worried about the prefix since this seem to be being used for non-Openstack data too (host endpoints). Should we just leave host endpoint status at the old path (or remove it since AIFAIK, it's not used anywhere)?"
+    },
+    {
+      "prediction": "I think this should be `(?:(?:[01]?\\\\d|2[0-3]):?([0-5]?\\\\d):)?([0-5]?\\\\d))` instead of `([01]?\\\\d|2[0-3]):?([0-5]?\\\\d):)?([0-5]?\\\\d)`",
+      "reference": "This might be a silly question, but what happens to files with durations > 24 hours?"
+    },
+    {
+      "prediction": "I think this should be `Func<ActivitySourceAdapter, TInstrumentation?>`",
+      "reference": "`AddDiagnosticSourceInstrumentation` method can now be eliminated and simply use `AddInstrumentation`"
+    },
+    {
+      "prediction": "I'm not sure if we want to raise an error here or if we want to return an error. I think raising an error would make more sense here, as it's an error in the `Bolt::Project` class, not in the `Bolt::Project#validate` method. The `Bolt::Project#validate` method is called by the `Puppet::Manager` and `Puppet::Manager#validate` methods, which are called by the `puppet init` command, so it makes sense that we would raise an error in the `Bolt::Project#validate` method.",
+      "reference": "I think we also need to add `name:` to the bolt-project.yaml files in `spec/fixtures/projects`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should go in `getMDUpdates`.",
+      "reference": "worth mentioning the same warning here as in `md_util.go`, I think."
+    },
+    {
+      "prediction": "I think we can remove the check for `nodeList.Items == nil || len(nodeList.Items) == 0` here.",
+      "reference": "Do we need both the checks, for `Items` not nil and `len(Items)`"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Do you know what this is coming from? It keeps getting deleted/created @nickbristow"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for this.",
+      "reference": "Nit: leaving these in place would have reduce the number of files that this needed to touch, and avoided a possible problem removing public fields. I don't think it's worth blocking for this change, but we like to keep patches as small as possible by not breaking references like these."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to remove the deprecated method. I think it should be moved to the top of the file.",
+      "reference": "`The %s is deprecated...` -> `The %s method is deprecated...`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `blockchain` package.",
+      "reference": "No, it wouldn't go there because that code won't reference or enforce it. Here is ok for now."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "How will we show that test run aborted (because of a crash) if we don't set aborted to true?"
+    },
+    {
+      "prediction": "I think this should be `KestrelMemoryPool.MinSegmentSize / 2`",
+      "reference": "Nit: Might as well make this same change to AdaptedPipeline."
+    },
+    {
+      "prediction": "I think we can remove the `ZoneOffset.UTC` from `OffsetDateTime.ofInstant` and `OffsetDateTime.of`",
+      "reference": "Just to clarify: I see that only the hive2 withZone object inspector is changed. Does that mean that the predicate pushdown problem only occurred on hive2?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Based on username requirements for forge.puppet.com - only letters and digits permitted."
+    },
+    {
+      "prediction": "I think this should be `if (struct.size() != that.struct.size()) { return false; }`",
+      "reference": "Was this removed to ignore the extra columns coming from the file projection?"
+    },
+    {
+      "prediction": "I think this should be `// +build windows` instead of `// +build !windows`",
+      "reference": "Please can you pull out the shared function into a shared file? I think that'd be pretty easy to do for this module. I think you could: - pull out a function `configureSyslog` that is implemented on Linux, but stubbed on Windows - pull out a function `openLogFile` that is implemented differently on each - share everything else"
+    },
+    {
+      "prediction": "Shouldn't this be `new EmailNotificationService(new AWSEmailProvider())`?",
+      "reference": "Instead of hardcoding the AWSEmailProvider here, it should come from properties, so that it can be replaced with another EmailProvider for ZMS vs ZTS"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Align the elements of a hash literal if they span more than one line.",
+      "reference": "+1 for creating this script to help people get boot strapped"
+    },
+    {
+      "prediction": "I think this should be `ELT_KEY_PREFIX` instead of `ELT_`",
+      "reference": "Do you need the prefix?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to call this function. I think it should be called in the same place as deleteFromTextTable.",
+      "reference": "if retcode is < 0, it should return -1 indicating an error."
+    },
+    {
+      "prediction": "I don't think we should disable pylint here. I think we should keep it disabled for the rest of the PR.",
+      "reference": "Are those necessary? I feel like `useless-suppression` could be avoided here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that the `codecProfile.Conditions` and `codecProfile.Conditions[0]` will be `null` if `codecProfile.Condition` is `LessThanEqual` and `AudioBitrate` is a property of `codecProfile.Conditions[0]`.",
+      "reference": "Is this equivalent in JavaScript?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to log the error. I think it's better to log the error in the `onFinished` method of the `GetEdgeIndexSpaceID` processor.",
+      "reference": "get/list operation is not schema change, it is supposed to use VLOG"
+    },
+    {
+      "prediction": "I think this is the only place where we need to use `Span<T>` instead of `ReadOnlySpan<T>`.",
+      "reference": "we'll need to delete this whole file, right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but I think it makes sense to move it to the catalog layer.",
+      "reference": "\"this table hasn't mutated within this transaction\" may sound like implying that if this transaction contains table mutation changes, `refresh` may have impact, which I think is not true? I guess what you were saying was if other transactions committed to this table successfully when this transaction is half way through, refresh in this transaction will still return the same state as when this transaction begins."
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "Can we move these into some kind of experimental API group? I don't think we want to call them `v1` just yet :)"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "I think this makes a lot of sense but we may want to highlight the change for people doing UAT in case the ordering of sections is off anywhere when customizing or using the drag-drop feature"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "These files should in `/javascript/brython`"
+    },
+    {
+      "prediction": "I don't think this is the right place to break out of the loop.",
+      "reference": "I'm not sure this is the right fix - I'd be more happy with a `continue`. That way, we don't enter the code at the bottom of the run loop (e.g. sending packets). Keep in mind that this error here may be triggered by a peer doing something security-relevant, so I don't think we should do much more work other than sending the close frame."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to load all the weights in one go, and then filter out the ones in `weight_list` that are not present in `ckpt_dir`.",
+      "reference": "Can we fix this, or is it going to stay broken."
+    },
+    {
+      "prediction": "I think this should be `@attr('postgres', 'testdb')`",
+      "reference": "Is there a reason to remove this attr?"
+    },
+    {
+      "prediction": "I think this should be moved to `config.go`",
+      "reference": "log message should be started with upper-case word."
+    },
+    {
+      "prediction": "I think this should be `512MiB` instead of `512 * KiB`",
+      "reference": "I guess this is not needed anymore now?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "I think I would prefer retaining the argument. It's true that we are not using it now, but it's harmless, it's good to have a constructor function that accepts all the field values as arguments. Performance-wise, it does not matter."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with that doesn't break anything.",
+      "reference": "Is this a different fix?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Why not add a `EnterNS` filed instead of `WithoutNS`?"
+    },
+    {
+      "prediction": "nit: I think it would be better to have a constant for the port number instead of hardcoding it.",
+      "reference": "Same points as in other test file."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "I have a PR pending, can you change this bit index to `16`?"
+    },
+    {
+      "prediction": "I think this should be `if (typeName.indexOf('.') > -1 || typeName.contains('.') && typeName.length() > typeName.indexOf('.') + 1)`",
+      "reference": "we should probably have some utility class for this sort of things"
+    },
+    {
+      "prediction": "I think this should be `pylint: disable=expression-not-assigned` instead of `pylint: disable=missing-docstring, no-else-return, invalid-name, unused-variable, superfluous-parens, try-except-raise`",
+      "reference": "Curious that I'm not seeing any occurrence of the new error check in this file. Why was it disabled?"
+    },
+    {
+      "prediction": "I think this should be `@return {boolean}` instead of `@throws {Error}`",
+      "reference": "does it return something?"
+    },
+    {
+      "prediction": "I think this should be `return ( 'string' !== typeof error ) ? error : getDataErrorComponent( moduleName, error, layoutOptions.inGrid, layoutOptions.fullWidth, layoutOptions.createGrid, data );`",
+      "reference": "This function is also called in various Site Kit module components, where the error object (here `data`) also needs to be passed, otherwise the link to fix the issue won't appear."
+    },
+    {
+      "prediction": "I think this should be renamed to `DiffSpec` or `DiffSpecs` to be consistent with `Revision` and `RevisionSpec`.",
+      "reference": "super nit: `revision string` (the passed value might not be a SHA hash)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `NLog.Layouts` assembly.",
+      "reference": "@snakefoot why is internal needed?"
+    },
+    {
+      "prediction": "I think this should be `DisableHealthCheck: false` instead of `DisableHealthCheck: true`",
+      "reference": "Why were we disabling health checks before?"
+    },
+    {
+      "prediction": "`conftest.py` is not a test file.",
+      "reference": "did `conftest.py` actually show up for you? `test_files` should never contain it..."
+    },
+    {
+      "prediction": "I think this should be `&& (ComIsTrafodionExternalSchemaName(schemaQualName_.getSchemaName()))`",
+      "reference": "Should this be a ! on line 565? I would have expected the condition to be similar to line 564. If I am wrong please excuse my mistake."
+    },
+    {
+      "prediction": "I think this should be `ALTER TABLE ` + PodDBAdapter.TABLE_NAME_FEEDS + ` ADD COLUMN ` + PodDBAdapter.KEY_FEED_VOLUME_REDUCTION + ` INTEGER DEFAULT 0;`",
+      "reference": "Please change to `1070400`. I promise to look into this PR in more detail before the 1.7.4 release ;)"
+    },
+    {
+      "prediction": "I think this should be `LogicalType` instead of `RowType`",
+      "reference": "Can we change to use `LogicalType.getChildren`?"
+    },
+    {
+      "prediction": "Layout/AlignParameters: Align the parameters of a method call if they span more than one line.",
+      "reference": "I'm confused as to how these are both passing."
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "Prefer single-quoted strings when you don't need string interpolation or special symbols."
+    },
+    {
+      "prediction": "I think this should be `string[]|bool[]`",
+      "reference": "there is `\"` sign in commit message and colon. can you rename also `descrition` to plural `descriptions` and add there `$` ?"
+    },
+    {
+      "prediction": "I think this should be `metricsScope.RecordDuration(metrics.ServiceLatencyUserLatency, userLatencyDuration)`",
+      "reference": "timerNoUserLatency is not being used?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in `utils.py` under the `isodate` module.",
+      "reference": "@lmazuel - I've made a couple of minor changes to serialization.py, if you could do a quick review :) They should not be breaking, and I doubt are used by the existing clients so may not need a new release yet. Effectively it's a change to support default/constant date and datetime values by allowing strings to passed in instead of date or datetime objects. The strings would need to already be in the correct ISO formatting that the request will serialize to. Let me know if you think this is okay."
+    },
+    {
+      "prediction": ">\", true, true, structuralModificationType);\n         }\n \n         [Test]\n```\nReview comment?",
+      "reference": "Why change the name in the XML?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in `User` model, not in `User#update` method.",
+      "reference": "This is much cleaner. makes it easier to tell what happens on a save."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `mysql` package.",
+      "reference": "Could you add return error unsupported operator here"
+    },
+    {
+      "prediction": "I don't think you need to import `Empty`, it's already defined in `StreamModule`.",
+      "reference": "(Mhh, we seem to use different formatters - we should unify them. I like the wildcards)"
+    },
+    {
+      "prediction": "I think this should be `enumspb.TASK_QUEUE_KIND_TASK`.",
+      "reference": "Should `kind` be a command line parameter also?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `TryGetAccessListItems` method.",
+      "reference": "can we have it inside FromAccessList?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `TelemetryServer` package.",
+      "reference": "Why not embed filterASCII in this function?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in a different place, e.g. in `FlowRunner` or `FlowRunnerImpl`.",
+      "reference": "I am curious, shouldn't \"currentContinuousFlowProcessed++;\" be added right after line 1873? otherwise we will count 1 extra when a exflow wakes up from the sleep section,.even though it hasn't been assigned"
+    },
+    {
+      "prediction": "I think this should be `# pylint: disable=protected-access`",
+      "reference": "You can probably move this up after the `._Numeric` line."
+    },
+    {
+      "prediction": "I think this should be `super(self.__class__, self).__init__(*args, **kwargs)`",
+      "reference": "@Tarrasch this is another change you need to pay attention, it seems `SGEJobTask` is not calling `super.__init__` which break the test case some how."
+    },
+    {
+      "prediction": "Line is too long. [101/90]<br>Use the new Ruby 1.9 hash syntax.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Do we need indices for this table?"
+    },
+    {
+      "prediction": "(style) please, no space before tabs",
+      "reference": "what about making this function static inline to avoid the extra declaration ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/identity` since it's related to identity stuff.",
+      "reference": "I forget the difference between type definition with \"=\" and without \"=\" again, but can we unify the style? I believe the difference doesn't apply to TableIDType and GroupIDType whatever it is?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `contractor.cpp` file.",
+      "reference": "would be `DEPRECATED Percentage of the graph (in vertices) to contract [0..1]` better?"
+    },
+    {
+      "prediction": "I think we should rename this to `START_CAST_TAB_MIRRORING` as it is not related to media-routers.",
+      "reference": "These command names are specific to Chromium-based browsers. Please move to `ChromiumDriverCommand`"
+    },
+    {
+      "prediction": "I'm not sure we need to create a new `Dictionary<string, object>` for each `Resource` instance. I think we can just use a `Dictionary<string, object?>` here.",
+      "reference": "if we won't use, should we remove this?"
+    },
+    {
+      "prediction": "I think it would be better to use `nameof(ViewNamespaceAbbreviation)` instead of `nameof(ViewNamespaceAbbreviation)` and `nameof(ViewBindingNamespaceAbbreviation)` instead of `nameof(ViewBindingNamespaceAbbreviation)`.",
+      "reference": "Having 2 items with the same key (Mvx) won't work in a dictionary"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "@cpytel how does the flow work right now? I expect to go to the sign up as customer page after putting in my code, but it goes to the longer landing page?"
+    },
+    {
+      "prediction": "This should be `snprintf(event_msg, sizeof(event_msg), \"%s: %.512s\", _(\"Message from server\"), message);`",
+      "reference": "`event_msg` has the same size (1024) as a `message`. Maybe should be increased to 2048 instead? Then this `\"[error] %.512s\", message` could be changed to this: `\"[error] %.*s\", sizeof(message), message`"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Nit: unnecessary whitespace change."
+    },
+    {
+      "prediction": "I think this should be `CombinedOutput()` instead of `CombinedOutput()`",
+      "reference": "Hit a flake here where I think there may have been some output to stderr that got mixed in with the output from Stdout. Hence switching to `Output()`, which does also capture stderr as `err.Stderr`"
+    },
+    {
+      "prediction": "I think this should be `LBANN_ERROR_IF` instead of `LBANN_ERROR`.",
+      "reference": "I would use static_cast<label_t>(0) and static_cast<label_T>(m_num_labels) just in case."
+    },
+    {
+      "prediction": "This method is not used anywhere, please remove it.",
+      "reference": "Even if we want to keep in original case - domain will be lower-cased."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "Based on discussion with Chad it feels like trail_map_json could be a good column name."
+    },
+    {
+      "prediction": "I think this should be `1 => 'Allow', 0 => 'Inherit', -1 => 'Deny'`",
+      "reference": "This array should be logic-less, just the language string (without `Lang::get()`) should appear. Then `trans()` is [or should be] used when the balloon selector renders the values."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "This looks like it runs both pre 3.0 rake tasks and 3 rake tasks, and since pre 3.0 we didn't use bundler, my guess is it will fail then."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/manager` since it's a manager and not just a device manager.",
+      "reference": "I'm not sure the replacement is a good idea, and as I said in #336 , this \"parent-name.scope\" is not a good idea in the first place, specially when we support assigning a slice as the parent, so we should change this, WDYT?"
+    },
+    {
+      "prediction": "I don't think we need to commit here. If we want to commit, we need to make sure that all the transactions are committed before we commit the batch.",
+      "reference": "sloppyReassign: re-assignment to `err` can be replaced with `err := kv.WriteBatch(batchForBlock)` (from `gocritic`)"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the `sendPart` method of `VertxHttpServletResponse`",
+      "reference": "using factory method. return CompletableFuture.completedFuture"
+    },
+    {
+      "prediction": "I think this should be `_x( 'Analytics', 'Service name', 'google-site-kit' )` instead of `_x( 'Analytics', 'Service name', 'google-site-kit' )`",
+      "reference": "The `_x` function needs to be imported at the top of the file (in addition to `__`)"
+    },
+    {
+      "prediction": "I think we can remove the `ofctrl.` prefix here.",
+      "reference": "this looks like the only controversial change potentially. @MatthewHinton56 Could let us know why you made this change (which `go vet` warning was triggered)? @wenyingd could you take a look and let us know if you're okay with this change?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure it's necessary.",
+      "reference": "This seems like left-over debugging code. I think I fixed this to no longer be necessary, switching to UserControl for the designer."
+    },
+    {
+      "prediction": "I think we can remove this file and just use the one in `moto.elbv2`",
+      "reference": "As we're currently not applying any logic in the proxy listener, I guess we can simply remove this file (and remove the `listener` argument from the `register_plugin(..)` call above)."
+    },
+    {
+      "prediction": "I think this should be `types.Uint64(time.Now().UnixNano())` instead of `time.Now().Unix()`",
+      "reference": "Isn't the point of the clock module to encapsulate access to `time.Now()`? Coming later?"
+    },
+    {
+      "prediction": "I think we need to add this to the `consensus_test_fast_upgrade.go` as well.",
+      "reference": "as before - if you've set this, you should be able to get rid of the `os.Setenv(\"ALGOSMALLLAMBDAMSEC\", \"500\")`. make sure that the various tests still takes have the same execution time."
+    },
+    {
+      "prediction": "Line is too long. [87/80]",
+      "reference": "Pretty sure the single quotes here should be double quotes, single quotes are not expanded by the shell."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `org.infinispan.commons.test`",
+      "reference": "these are not really constants, right?"
+    },
+    {
+      "prediction": "I don't think we should be touching this file.",
+      "reference": "import below copyright statement"
+    },
+    {
+      "prediction": "Line is too long. [105/80]",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `string|array|Markup` and not `string|Markup`",
+      "reference": "`if (is_string($value) && $this->getDefinition()->get('allow_html')) {`"
+    },
+    {
+      "prediction": "I think this should be `Debug.Assert(!string.IsNullOrWhiteSpace(key), $\"{nameof(key)} must not be null or empty\")`",
+      "reference": "I think we have the same problem here - if folks called the `public void Add` with a `null` key, we need to use `Guard`."
+    },
+    {
+      "prediction": "This file should be removed.",
+      "reference": "I thought that the main was in catch_qt.cpp?"
+    },
+    {
+      "prediction": "I don't think we should change the list of fields here. We should keep it as it was in the previous version.",
+      "reference": "I would rename the field to IMAGE_TYPES and make it an array. I don't think we need list operations. Also, if possible, I would move the field to the ApiFields class. What do you think?"
+    },
+    {
+      "prediction": "I think this test can be removed now that we have `verifyContainerRunningStateChange` and `verifyExecAgentRunningStateChange`",
+      "reference": "I'd prefer to timeout on this, but I know it's not a pattern that is being followed"
+    },
+    {
+      "prediction": "`*` is a reserved word, please use `org.apache.lucene.util.*` or `import org.apache.lucene.util.*`.",
+      "reference": "hmm let's not use * imports please"
+    },
+    {
+      "prediction": "I think this should be `am.BearerTokenFile == \"\"` instead of `am.BearerTokenFile != \"\"`",
+      "reference": "is this configuration key already exist in prometheus ?"
+    },
+    {
+      "prediction": "Layout/ExtraSpacing: Unnecessary spacing detected.<br>Layout/SpaceBeforeBlockBraces: Space missing to the left of {.<br>Layout/SpaceBeforeBlockBraces: Space missing to the left of }.",
+      "reference": "As @gravitystorm said new tests need to be using factories, not fixtures."
+    },
+    {
+      "prediction": "I think this should be `%s@%s:%d) [%s]>`",
+      "reference": "You can replace IP by hostname."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `getDriversForConnectedUsbDevices`",
+      "reference": "This should say \"why\" NVDA should do things in this order."
+    },
+    {
+      "prediction": "This test is failing because `CreateBucketConfiguration` has a default value of `CreateBucketConfiguration(&s3.CreateBucketConfiguration{}`.",
+      "reference": "nit these will get changed back during next release."
+    },
+    {
+      "prediction": "I think this should be `const Version = \"1.0.0\"`",
+      "reference": "it is actually not needed to make the member public. the build tags can also set a package scoped variable."
+    },
+    {
+      "prediction": "Shouldn't this be `$this->contentExtension->getIcon($content, $type)`?",
+      "reference": "We already have a method for it 'icon' => $content->getIcon()"
+    },
+    {
+      "prediction": "I think we should use `context.WithValue` instead of `context.WithValue(context.Background(), key, value)`",
+      "reference": "Instead of importing `golang.org/x/net/context` The SDK should use `aws.BackgroundContext()` instead of `context.Background()`"
+    },
+    {
+      "prediction": "I think this should be `findOneBy` instead of `findOneBy`",
+      "reference": "I think this is an agnostic change for the ORM right? Why was it doing this? - this does not work on PHPCR-ODM as the `id` is not a field - although I guess it could be mapped as such."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [124/120]",
+      "reference": "I'm not 100% sure, but I think this should be `explorable`. If not, I think `explorable` can be removed entirely."
+    },
+    {
+      "prediction": "I think this should be `if six.PY3 or sys.stderr.isatty()`",
+      "reference": "Note to self: Why was this not originally not using `stderrbytes`?"
+    },
+    {
+      "prediction": "I think this should be `LOG(INFO, ...)`.",
+      "reference": "Is betterLOG_IF(ERROR, status != 0) ?"
+    },
+    {
+      "prediction": "I think this should be `self.bucket_name = 'default'` and `self.collection_name = 'default'` instead of `self.bucket = 'default'` and `self.collection = 'default'`",
+      "reference": "I wouldn't call it default too."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better idea.",
+      "reference": "Why disable the rule for this file?"
+    },
+    {
+      "prediction": "I don't think we need this prop. The `onClick` prop will always be present.",
+      "reference": "Thanks for adding this :+1:. Could you please move it to go after the `ctaLink` prop?"
+    },
+    {
+      "prediction": "Why is this needed? `Current!` is the same as `Current`?",
+      "reference": "The other way around Current can be null."
+    },
+    {
+      "prediction": "I think it would be better to move this to the `mail` sub-menu.",
+      "reference": "why does it have to be here? Due to breadcrumb navigation?"
+    },
+    {
+      "prediction": "I think this should be `resource.NewQuantity(capacity)` instead of `resource.ParseQuantity(capacity)`",
+      "reference": "I fear moving this to some common func. We are swallowing the error here. This might be ok in UT but not in actual source code."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `tests/conftest.py`",
+      "reference": "Please don't use \\ as a line continuation. Do an implicit string join instead with parens: ``` (\"Message ...\" \"and ...\")"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "there needs to be a change here. setting up the batchstore with localstore.Unreserve hook"
+    },
+    {
+      "prediction": "I think we can remove this method and use `setConfig` directly in the `ConfigUtil` class.",
+      "reference": "I think if we can provide a more convenient method to add configs . e.g. public static void addConfig(String k, Object v)"
+    },
+    {
+      "prediction": "I think we need to change the test name to `TestActPool_AddActs_VI`",
+      "reference": "line is 165 characters"
+    },
+    {
+      "prediction": "I think we should remove this file.",
+      "reference": "Please add a copyright notice. Also, could you tell me please why do we need this empty provider here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in `PrivacyController#isGroupAdditionTransaction`.",
+      "reference": "I feel like this would be easier to read if we join the two if's together: `if (maybePrivacyGroup.isEmpty() && !privacyController.isGroupAdditionTransaction(privateTransaction))`"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `loadSoupBlob` returns a `JSONObject`, but `Exception` is not a subclass of `Exception`. So the `catch` block will catch both `JSONException` and `Exception` and log only the `JSONException`. I think you need to change the `catch` block to catch `Exception` and handle it appropriately.",
+      "reference": "Tests are failing because the exception type is NullPointException here if the soupBlob is deleted (so instead of a mal-format json, it's a null)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this check. I think it should be in the `if (this.targetConnection) {` block above.",
+      "reference": "Glad we fixed this as well. Blockly is going to be left as-is for this case, right?"
+    },
+    {
+      "prediction": "I think we can remove this as well.",
+      "reference": "I had accidentally deprecated this class. This is meant to stick around. Only `SalesforceObjectType` goes away."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. It seems like it should be in `utils.py`, but I can't find where it is used.",
+      "reference": "`hex(i)[2:]` is trying to do the same thing above. what is `i` in your malformed case?"
+    },
+    {
+      "prediction": "Shouldn't this be `if (shouldEscape)`?",
+      "reference": "I think this change is redundant; see the previously defined method \"writeUnescapedXML\"."
+    },
+    {
+      "prediction": "I think this should be `CommitmentLength` instead of `Commitment`",
+      "reference": "Ah now I have at least some idea what these are for."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. It seems like it should be in `DataWriter`, not `DataReader`.",
+      "reference": "I believe in previous PRs @mccheah frequently used `fileSchema` -> `newFileSchema` type of renames to avoid hiding fields in builders. Would it make sense to make it consistent?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `connect` function, but I'm not sure if there is a better place to do it.",
+      "reference": "Do we really need separate error in manager if utils.ErrRequestCancelled is the only error which indicates cancelation ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the same place as the other temporal parameters.",
+      "reference": "Leave as `cpp` since that still gives better results."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this predicate. I think it should go in `pkg/apiserver/server.go` under the `processIfUnpaused` function. The reason is that the `processIfUnpaused` function is only called from within the `server` package, and the `PausedPredicates` function is only called from within the `controller-runtime` package. The `PausedPredicates` function should be called from within the `controller-runtime` package, not the `server` package.",
+      "reference": "temp change so the predicate can be used by the `exp` package"
+    },
+    {
+      "prediction": "I think this should be `self.vkCode in self.TOGGLE_KEYS or self.vkCode not in self.TOGGLE_KEYS` instead of `self.vkCode in self.TOGGLE_KEYS or self.vkCode not in self.TOGGLE_KEYS` because `self.vkCode not in self.TOGGLE_KEYS` is the same as `self.vkCode not in self.TOGGLE_KEYS or self.vkCode in self.TOGGLE_KEYS`",
+      "reference": "Just as a tiny clarification, this isn't affected in any way by \"speak typed characters\". That is, \"caps lock on\", etc. is always spoken, even if speak typed characters is off."
+    },
+    {
+      "prediction": "I think we can remove this line now that we've added it to the TODO above.",
+      "reference": "@dgoodwin Will it cause an issue when the issue get fixed in OLM?"
+    },
+    {
+      "prediction": "I think this should be `DumpParityTraces | DumpGetTraces`",
+      "reference": "This option is now unnecessary"
+    },
+    {
+      "prediction": "I think this should be `b := &bucket{dir: subdir}` and then you don't need the `b := &bucket{dir: subdir}` line.",
+      "reference": "Don't jump down to the unexported interface: just set the content type to `\"application/octet-stream\"` explicitly when creating the `Writer`."
+    },
+    {
+      "prediction": "I think this should be `bytesUsed.addAndGet(-BYTES_PER_DEL_TERM + term.bytes.length - (Character.BYTES * term.field().length()));`",
+      "reference": "Instead of counting this here on clear, can we use a second counter for the deleteTerms next to `bytesUsed`? This would be great. It doesn't need to be thread safe IMO"
+    },
+    {
+      "prediction": "I think we can drop the `required_with:password` part here.",
+      "reference": "@daftspunk @bennothommo I wonder if we need to go as deep as detecting what the default varchar length is with a special character to be parsed by the validation trait since we've introduced the config for it."
+    },
+    {
+      "prediction": "I think this should be `setParameter('text', keys.join('')).`",
+      "reference": "Also update line 2205 below"
+    },
+    {
+      "prediction": "I think this should be `STATE_FIRST_TIME_EMPTY_SUBJECT` instead of `STATE_FIRST_TIME_EMPTY_SUBJECT`",
+      "reference": "Typo, should read `firstTimeEmptySubject`"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this. It's not a handler, it's a request. I think it should be something like `Handler for handling a single, transport-level, unary request`.",
+      "reference": "The more I see this the more I think the transport layer _should_ actually refer to this as `UnaryHandler`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the same file as the `logger.warn` above.",
+      "reference": "The `warn` message here seems to indicate that the forced closure of the SSH connection should raise an error; is that getting swallowed up somewhere and not raising?"
+    },
+    {
+      "prediction": "I think this should be `self._expands_user = True` instead of `self._expands_user = False`",
+      "reference": "I wonder if this is a fitting name. Why \"expands user\"? Wouldn't be `user_expanded` or so be more fitting?"
+    },
+    {
+      "prediction": "nit: I think it would be better to move this to the top of the function, so that it's visible to the user of `virtualDOMfromNode`",
+      "reference": "I see what you are doing here, but to stay with the `getter/setter` pattern, should we introduce `set isHidden(value)` & `get isHidden()` which them maintains `_isHidden` with in `vNodeCache`. This will avoid what looks like accessing an internal property like `_isHidden` from `axe.utils.isHidden` & keeps things neat."
+    },
+    {
+      "prediction": "I think we should rename this to `UNITTEST_CSPROJ_TEMPLATE_FILENAME` and `SAMPLE_CSPROJ_TEMPLATE_FILENAME`",
+      "reference": "nit: for consistency, have this line and the previous formatted similarly"
+    },
+    {
+      "prediction": "I think it would be better to put these in a `private static final` field.",
+      "reference": "As before, shouldn't mention these as being orion options I don't think. Perhaps just enclaveKeyStoreFile etc."
+    },
+    {
+      "prediction": "I think this should be `console.log(...args)` instead of `console.log(...args);`",
+      "reference": "What's the thinking here? Only logging when the test has not ended yet?"
+    },
+    {
+      "prediction": "I think this should be `scol = F.when(scol.isNull(), False).otherwise(scol)` instead of `scol = F.when(scol.isNull(), False).otherwise(scol)`",
+      "reference": "@HyukjinKwon @ueshin (cc @itholic @charlesdong1991 ) Not sure if this is the right implementation ..."
+    },
+    {
+      "prediction": "I don't think this is the best way to do this. I think it would be better to have the callback return an error object with the error as the only property. This will allow us to pass the error to the caller and make it easier to debug.",
+      "reference": "Do we want to add any specific error on how the url is malformed?"
+    },
+    {
+      "prediction": "`self->lastLogIdSent_` should be `self->logIdSent_` and `self->lastLogIdSent_++` should be `self->logIdSent_++`",
+      "reference": "why NOT self->lastLogIdSent_ = self->logIdToSend_"
+    },
+    {
+      "prediction": "I think this should be a warning, not an error. I don't think we want to report an error to the user if they are already provisioned.",
+      "reference": "Shouldn't this be in the above `if` block where we are setting the installedtimestamp to ensure this only happens for already installed (and/or adopted) clusters?"
+    },
+    {
+      "prediction": "I think this should be `tabbed_browser.tabopen(background=True)`",
+      "reference": "This seems like another unrelated change I've done in `master`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it would be better placed in `spec/formatters/rails_spec.rb` since that is where the `add_formatter` method is.",
+      "reference": "Would be good to have another context `\"when a duplicate formatter exists for a different output target\"` that shows that it keeps both. As this stands, the specs could pass w/o the `formatter.output == new_formatter.output` check."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to deprecate this function.",
+      "reference": "Why not just remove? The CUDA one has been around for while. This one has not been part of any release (?)"
+    },
+    {
+      "prediction": "I think this should be called `mIsFirstContentfulPaint`.",
+      "reference": "nit: `Contentful` is the word used in the GV API, without the uppercase in F"
+    },
+    {
+      "prediction": "I think this should be `LOGGER.error(swagger.getInfo().getTitle() + \" with path \\\"\" + swagger.getBasePath() + \"\\\" is an empty interface, please delete it or fill with one method!\");`",
+      "reference": "It just does the check, maybe we can name it as checkOperations."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/unit/test-utils.js`",
+      "reference": "We no longer have karma output in the test file so these selectors were now unique and the target didn't need a child selector. Updated to force non-unique nodes"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it seems like it makes sense to me.",
+      "reference": "This moved from a partial to not being in a partial. How about cleaning this view up further by moving it back into a partial?"
+    },
+    {
+      "prediction": "I think this should be `recieveChanges` instead of `recieveUpdates`.",
+      "reference": "add some form of docs in docstring or on :74 for what the effect is"
+    },
+    {
+      "prediction": "I think this should be `from kinto.core.utils import COMPARISON, find_nested_value`",
+      "reference": "Couldn't you import `json` from `kinto.core.utils` here too?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Minor: Hmmm...is it worth setting this explicitly for the tests that it affects?"
+    },
+    {
+      "prediction": "Why is this lock needed?",
+      "reference": "This was added to fix a race condition and you are reverting it here, right?"
+    },
+    {
+      "prediction": "I think this should be `Coinbase` instead of `Coinbase is the address that mining rewards will be sent to`",
+      "reference": "Could you add some extra text here, so that it reads \"eth_coinbase function is not available, please use --private.api.addr option instead of --chaindata option\", so that it is clear that the function can work, but different options"
+    },
+    {
+      "prediction": "I think this is a good example of how we can use a `let` block to ensure that the values we're setting on the page are what we want.",
+      "reference": "`new_ncr_work_order_path` ? (I've been slowly moving specs over to that way of calling paths)"
+    },
+    {
+      "prediction": "`font` is not a valid file extension for web pages.",
+      "reference": "Should flash be included in the patterns too? It currently isn't."
+    },
+    {
+      "prediction": "I think this should be `sealed` instead of `seemed`.",
+      "reference": "I suspect I will have to keep looking whether Left/Right refers to contract/Implementation while working on the rules code."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "(ignore this. testing a codeflow bug)"
+    },
+    {
+      "prediction": "I think this should be `CASSANDRA_DISABLE_TLS`.",
+      "reference": "[Nit] can you change the Usage wording to indicate that the flag is used to opt-out of server certificate verification? (e.g. \"disables validation of the Cassandra cluster's server certificate.\")"
+    },
+    {
+      "prediction": "I don't think we need to import `java.util.List` here.",
+      "reference": "Sorry :) Would be really helpful to integrate something like Eclipse Code Formatter, so it will fail if the code style is broken (the same as eslint fails on the frontend) I use IntelliJ IDEA and their vision of imports is a bit different :)"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "this will still fail on systemd"
+    },
+    {
+      "prediction": "I think this should be `if type(body) is bytes:`",
+      "reference": "The convention in most of the codebase is to use `if isinstance(body, bytes):` instead."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to find a better way to do it.",
+      "reference": "I think this would be better as: `label = comp.rsplit('.', 1)[-1]`"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `RubyGapicContext` implementation.",
+      "reference": "We have been creating a language-level context interface for each language instead of parameterizing the generate function."
+    },
+    {
+      "prediction": "Shouldn't this be `sprintf('%s/%04s/%02s', $media->getContext() ?? '', substr($id, 0, 4), substr($id, 4, 2));`?",
+      "reference": "Same thing about context here."
+    },
+    {
+      "prediction": "I think this should be `handler.namespaceCache.GetNamespaceByID(handler.mutableState.GetExecutionInfo().NamespaceId)`.",
+      "reference": "there is a function to get namespace entry from mutable state directly"
+    },
+    {
+      "prediction": "I think this should be `@hiera_config.transport_data_get` instead of `@inventory.transport_data_get`",
+      "reference": "The `Transport::Config` objects don't serialize properly. We probably want to just turn them into hashes at this point."
+    },
+    {
+      "prediction": "I think this should be `ddev config` instead of `ddev config global`.",
+      "reference": "Sorry, typo s/unusuable/unusable/"
+    },
+    {
+      "prediction": "I think it would be better to add a check for hive metastore type here.",
+      "reference": "If this is needed, should we check that the configured `uri` isn't already equal to the value of the metastore URI configured via `spark.hadoop.hive.metastore.uris` or any of the other ways of setting it. This would be a breaking change for people who have `uri` configured on the SparkSessionCatalog and have it correctly set (which I believe should be everybody as `uri` is currently required if I'm not mistaken)."
+    },
+    {
+      "prediction": "I don't think this is the right way to check for this. I think we should check for the presence of the `Content-Type` header, and if it's not present, we should return `0`.",
+      "reference": "see above - this is only used to select the correct view, we don't need to handle the boundary information here."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "These tests use a self-signed cert so the CA *is* the cert. I considered making a proper chain to use in the tests, but wasn't sure it was necessary to test here, since we test it in the E2E tests anyway."
+    },
+    {
+      "prediction": "I think this should be `dom` instead of `excessDomChildren`",
+      "reference": "We could, maybe, just directly do `excessDomChildren[excessDomChildren.indexOf(dom)] = null;`. Would this improve the size in any way? This will end-up with a property on the `excessDomChildren[\"-1\"]` but maybe we could live with that?"
+    },
+    {
+      "prediction": "I think this should be `unmount` instead of `unbind`",
+      "reference": "this looks like it may be something that needs additional refactoring"
+    },
+    {
+      "prediction": "I don't think we need this blank line.",
+      "reference": "Nit: Remove whitespace (and below)"
+    },
+    {
+      "prediction": "I think it would be better to call this `unzip` or `unzip_file` to be consistent with the `unzip` command.",
+      "reference": "\"Failed opening\" or \"Failed to open\"."
+    },
+    {
+      "prediction": "I'm not sure this is the right error to return here. I think we should be returning an error that says we don't know what to do with the source image.",
+      "reference": "and -> or"
+    },
+    {
+      "prediction": "I think this test should be `it('should do a dry-run feature-10316', function() {`",
+      "reference": "Add a `yield User.init()` here to wait for all indexes to finish building"
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `\"\"`, right?",
+      "reference": "I don't think this belongs here. If the application can take a local_ip as input, the application should take care of it, not in this common library"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be moved to the `Theme` class.",
+      "reference": "@DanHarrin please add a `return true` at the bottom of this method, as the docblock indicates a boolean return value."
+    },
+    {
+      "prediction": "I think this should stay final.",
+      "reference": "This is used in o.a.l.sandbox.search.BM25FQuery."
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. It should be moved to `XYShapeBoundingBoxQuery` since it is specific to the `XYShapeBoundingBoxQuery`.",
+      "reference": "Shouldn't this work as well for Disjoint?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that we don't know what `m` is, so we can't tell what `fn` is. I think we need to pass `m` to `fn` and see what it returns.",
+      "reference": "Same here. This was included to quiet the test for the 3D descriptors. As we removed them, this exception handling is no longer required"
+    },
+    {
+      "prediction": "I think this should be moved to `lbann/data_store/data_store_jag.hpp`",
+      "reference": "Why do you need to load a specific data reader in lbann_library?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "why not ValueTask?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this logic into the `resolveInstallerImage` function.",
+      "reference": "I'm nervous about this line, I don't want to go regenerate a bunch of imageset jobs for clusters that are old, already installed, but don't have a CLIImage set (which they wouldn't because they're old) Adding the Installed guard is meant to address this. Otherwise this *should* recreate the imageset job due to the code in this function that deletes the job if it's finished. (but we don't have our images set due to this clause)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be moved to `TableMetadata#newTableMetadata`",
+      "reference": "Is it necessary to change this file? Doesn't `tableMetadata` call `newTableMetadata`?"
+    },
+    {
+      "prediction": "I think it would be good to have a test for the case where both nodes upgraded at the same time.",
+      "reference": "I think that I have a better proposal for this test - improve it so that it would know how to read the list of SupportedProtocolVersions and dynamically use these. The motivation here is that I expect to have another network version soon, and this test seems to be a good test case for that."
+    },
+    {
+      "prediction": "I think this should be `LOG.debug(\"Listing of namespace [{}] resulted in the following tables: [{}]\", namespace, tableIdentifiers);`",
+      "reference": "What is the purpose of this debug message? I'm not sure how it would help."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it's fine for now, but it's something to think about in the future.",
+      "reference": "Is there a corresponding BUILD dependency to prune?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `testhelpers.go` since it's only testing `SignedURL` and not `PutURL` or `GetURL`.",
+      "reference": "The \"good\" and \"bad\" descriptions confused me a bit. Maybe `allowedContentType` and `disallowedContentType`? Not sure if that's better."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "It would be better to just change RWD to throw IllegalStateException if you attempt to execute a command after quit (unless it's a second call to quit())"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `server_test.go` under the `nats` package.",
+      "reference": "You may want to have a look at `func checkClusterFormed(t *testing.T, servers ...*Server)` in this file that I created some time ago. It will timeout at 5 seconds, but will return as soon as cluster is ready."
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "I see a few places where this blob of code is required due to the removal of `errnum` parameter from `flux_respond` -- trading 2 lines of code for 8. The improvement to the function seems like a good idea, but I wonder if we need a convenience macro or function to do it the old way? You went through and made all the changes, was it overall a net win?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should go in `src/service/analytics-impl.js`",
+      "reference": "I know it's not related to your change, but it's a bit odd that this import uses a relative path rather than the ones setup by webpack. Might be nice to change this one in case we refactor those paths in the future; this one wouldn't get caught in any kind of find+replace."
+    },
+    {
+      "prediction": "I think it's better to use `schedule := schedule` instead of `newSchedule := schedule` to avoid confusion.",
+      "reference": "Good catch, and if we don't declare the newSchedule here, will the last schedule be executed in multi goroutine?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could come up with that worked.",
+      "reference": "`project` should never be null here, unless we have projects that aren't in the solution? Which we shouldn't have"
+    },
+    {
+      "prediction": "This should be `#include \"extractor/restriction_compressor.hpp\"`, not `#include \"extractor/restriction_compressor.hpp\"`.",
+      "reference": "Simple `for` loop instead of `for_each` would help here by making it more readable."
+    },
+    {
+      "prediction": "I think this should be `thread_role::name::confirmation_height`",
+      "reference": "Looks like this falls through, should break."
+    },
+    {
+      "prediction": "I think this should be `manifest_file_overwrite` instead of `manifest_file_overwrite`",
+      "reference": "Let's default manifest lists and manifests to overwrite. These use UUID-based file names and should never conflict."
+    },
+    {
+      "prediction": "I think this should be `showSpeechViewerAtStartup` instead of `showSpeechViewerAtStartup = boolean(default=false)`",
+      "reference": "Please rename [speechView] to [speechViewer]."
+    },
+    {
+      "prediction": "I think it would be better to use `json:\"percpu_usage_in_kernelmode\"` and `json:\"percpu_usage_in_usermode\"` instead of `json:\"percpu_usage_in_kernelmode\"` and `json:\"percpu_usage_in_usermode\"`",
+      "reference": "I would use shorter yet still descriptive names, e.g. `KernelPerCpu` and `UserPerCpu`."
+    },
+    {
+      "prediction": "I think this should be `if not (format.lower() == 't') and check_user_can_edit_record(user_info, recID):`",
+      "reference": "if we are seeking for the content of only one field from metadata, no need to check whether a user could edit record or not"
+    },
+    {
+      "prediction": "I think we should rename this to `executeCdp` or something like that, since it's not just a single command.",
+      "reference": "Should this be camelCase to match the above commands? I am not the expert here so maybe Lucas or David can chime in."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set these tags. I think it should be in the `ValidateOperationName` method.",
+      "reference": "This is also missing in CreateScopeFromExecuteAsync. Can you add that there too?"
+    },
+    {
+      "prediction": "I think this should be `date = date.replaceAll(\"([+-]\\\\d\\\\d):(\\\\d\\\\d)$\", \"$1$2$3\");` and not `date = date.replaceAll(\"([+-]\\\\d\\\\d):(\\\\d\\\\d)$\", \"$1$2$3\");`",
+      "reference": "I hope that this does not break anything... Probably needs detailed beta tests."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this spec. I think it should go in `spec/spec_helper.rb` under `context \"when recording spec start time\"`",
+      "reference": "@JonRowe in what scenario do you see this being manually set?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `process_block` function.",
+      "reference": "I think the work \"trigger\" makes more sense than \"check\" because this is taking an action based on status."
+    },
+    {
+      "prediction": "Why is this change needed? I don't see where this property is used.",
+      "reference": "add a configuration , default put public key in Microservice, if the configuration is set, then put in Instance."
+    },
+    {
+      "prediction": "I think we should keep the distinction between `DEFINITION` and `ASSIGNMENT`.",
+      "reference": "WDYT about the idea above of making these all verbs that correspond to the input spec?"
+    },
+    {
+      "prediction": "I think this should be `STORE_TYPE = ServerSettings.Type.IMAP`",
+      "reference": "With the introduction of the enum(s) the constants in the store classes should be removed."
+    },
+    {
+      "prediction": "(style) space required after that ',' (ctx:VxV)",
+      "reference": "I would say `!= 5`; A packet with <5 would be malformed."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to create a new integrator for each resource.",
+      "reference": "I think you missed removing `Stateful` from `push/config.go`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I'm not sure if we want to make this a public API or not.",
+      "reference": "why would you make it internal? if so then at least protected internal"
+    },
+    {
+      "prediction": "I think this test should be moved to `test_http.py` instead of `test_https2.py`",
+      "reference": "We should check if the response (`d`) contains bar as response, screw the log. :smile:"
+    },
+    {
+      "prediction": "This should be `sparse_feature_group_map_` instead of `sparse_feature_groups`.",
+      "reference": "@Laurae2 good call. This is the only one I found (with `git grep transfered`)"
+    },
+    {
+      "prediction": "I think this should be `node.statement(future=True) == node.frame().body[-1]`",
+      "reference": "I don't understand how this work, could you explain ?"
+    },
+    {
+      "prediction": "I'm not sure this is correct. The number of simd_used slots should be NUM_REGPARM, not MCTX_NUM_SIMD_SLOTS.",
+      "reference": "The abbreviation for \"context\" used extensively inside DR is \"cxt\", not \"ctx\", so s/MCTX/MCXT/."
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "here is `:seconds` missing again"
+    },
+    {
+      "prediction": "Shouldn't this be `Optional<String>`?",
+      "reference": "initializing to null makes me uncomfortable. let's have a default value"
+    },
+    {
+      "prediction": "I think this should be `if (pathStart == -1 || pathStart == start) { // Empty path is illegal } else if (ch == BytePercentage) {`",
+      "reference": "This is the `GET % HTTP/1.1` scenario right?"
+    },
+    {
+      "prediction": "This should be `use Backend\\Facades\\BackendAuth;`",
+      "reference": "No need to use the fully qualified path the BackendAuth facade, just `use BackendAuth` is fine."
+    },
+    {
+      "prediction": "This should be `nano::epoch::epoch_0`, not `nano::epoch::epoch_1`.",
+      "reference": "Because it's similar for send/open/change/receive types, then probably it can be just common `nano::epoch nano::block::epoch () const` with override for state_block (like nano::block::link (), account (), representative ())"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `src/base/webview`.",
+      "reference": "Please give a more descriptive name and add a command to explain what it does."
+    },
+    {
+      "prediction": "I think this should be prefMobileUpdateAllowed, not prefMobileUpdate",
+      "reference": "we can never get rid of this..."
+    },
+    {
+      "prediction": "I think this should be `0.05, 0.1, 0.5, 1, 5, 10, 25, 50, 75, 100, 125, 150, 200, 250, 300, 375, 400, 425, 450, 475, 500, 525, 550, 575, 600, 625, 650, 675, 700, 725, 750, 775, 800, 825, 850, 875, 900, 925, 950, 975, 1000, 1025, 1050, 1075, 1100, 1125, 1150, 1175, 1200, 122",
+      "reference": "what is the effect of changing these buckets in existing data that is available in the monitoring system? if there are any existing dashboards that use previous histogram buckets they are probably going to be wrong ot invalid? also any reason why we chose these specific values?"
+    },
+    {
+      "prediction": "I think this should be `yarpc-with-sanitize-tchannel` instead of `yarpc-sanitize-tchannel` to be consistent with the rest of the code.",
+      "reference": "This won't work. For the flag to be passed to the plugin, it should be passed as part of the --plugin argument. --plugin \"yarpc --sanitize-tchannel\""
+    },
+    {
+      "prediction": "I think this should be `booster->GetBoosting()->MaxFeatureIdx() + 1` instead of `ref_booster->GetBoosting()->MaxFeatureIdx() + 1`",
+      "reference": "@shiyu1994 @StrikerRUS what do you think about this addition to `c_api.cpp`? I think it's a really useful addition to be able to get this type of information from the `Booster`, but I want more opinions since `c_api` is the main public API for the library."
+    },
+    {
+      "prediction": "I think this should be `dbRound` instead of `Round`",
+      "reference": "rename Round -> requestedRound and DbRound -> dbRound. we don't need to export the fields here, only the error struct."
+    },
+    {
+      "prediction": "I think this test should be moved to `testutil/testutil.go` since it is not related to the `ClusterDeploymentReconcile` function.",
+      "reference": "This expected value should probably be what you literally expect, otherwise there's a chance getClusterRegion is doing something wrong and the test wouldn't catch it because it's being run for both expected and actual."
+    },
+    {
+      "prediction": "I don't think we should be using `base64.StdEncoding` here. Instead, we should be using `encoding.Base64`",
+      "reference": "TODO: update unit tests to exercise base64-inline path"
+    },
+    {
+      "prediction": "I think this method should be called `shouldShowOnboarding` instead of `shouldShowOnboarding`",
+      "reference": "this method knows too much - it is kind of a strange side effect I'd prefer if we had separate method for acknowledging that the onboarding was done and should not be shown again"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Do we need to increase the DB version? Also, should probably be a constant, no?"
+    },
+    {
+      "prediction": "I think we can remove this function now, since we don't use it anywhere.",
+      "reference": "We could utilize the new isZero method in this if-statement, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `testutil` since it's only used in `testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which is in `pkg/testutil/testutil.go` which",
+      "reference": "should probably have config have a `t` or a `logger.NewTestLogger(t)`, and return the latter"
+    },
+    {
+      "prediction": "I think this should be `make([]PlanNode, 0, len(inputPlan.Roots))`",
+      "reference": "Does this not do the same thing as `copy`? I thought that `copy` just did an elementwise assignment, but maybe I was wrong."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the gapic config. I think it should be in the proto annotations config.",
+      "reference": "Here we are merging the configs from proto annotations and gapic config. But in other places in this PR we are using the new enum to pick one or the other, right? Or is that not the correct way to think about this?"
+    },
+    {
+      "prediction": "I think this should default to 8080.",
+      "reference": "not related to this change?"
+    },
+    {
+      "prediction": "I think this should be `process.env[\"ANDROID_NDK_PATH\"]` instead of `process.env[\"ANDROID_NDK_HOME\"]`",
+      "reference": "Why is this needed?"
+    },
+    {
+      "prediction": "I think this should be `if (calculatedStepSpeed < 0) {`",
+      "reference": "This one could be replaced with a `std::max`, no?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Couldn't you just write this as an if/else or a guard clause like on line 87? Just seems a bit weird doing this conditional assignment for essentially an if/else."
+    },
+    {
+      "prediction": "I don't think this is the best way to do it. I would suggest to do something like: if ($scopeType=='checkbox' || $scopeType=='switch') { $scope->value = $this->getScopeValue($scope); } else { $scope->value = isset($config['value']) ? $config['value'] : $this->getScopeValue($scope); }",
+      "reference": "Please use strict type comparisons (`===`), not loose type comparisons."
+    },
+    {
+      "prediction": "I think this should be `!getFromBlock().isLatest() && !getToBlock().isLatest() && getBlockhash() != null`",
+      "reference": "The filter parameters are a bit confusing. I understand that one of them has to be set, but what takes precedence if a from/to is set and the hash as well? Should that fail?"
+    },
+    {
+      "prediction": "Please remove the `#include` of `tuple` from this file.",
+      "reference": "Is that necessary?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "yess so much easier to read in order like this!"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `org.apache.lucene.test.TestRuleChain`",
+      "reference": "question: would this convention automatically and always apply to all classes derived from `LuceneTestCase` including any non-`org.apache` name spaces or would it be possible to opt-out (without an exclusion list) somehow for custom code that might perhaps have chosen a different convention?"
+    },
+    {
+      "prediction": "I think this should be 6, since it's the minimum length for a 6 digit passcode.",
+      "reference": "Also found this bug, where we were setting the min passcode length to 6, but the min length for a connected app is 4. This can cause problems when the app is force closed or the shared pref is removed."
+    },
+    {
+      "prediction": "I think this should be `int(cell * self.cellsPerColumn) / self.cellsPerColumn`",
+      "reference": "Why was this change necessary?"
+    },
+    {
+      "prediction": "I think this should be `SeleniumHost` instead of `Selenium Distributor`",
+      "reference": "I kind of did this deliberately, so everything from the distributor appears in the same set of logs."
+    },
+    {
+      "prediction": "I don't think this is correct. The `seq` value should be in the `TCP` payload, not in the `TCPSession` payload.",
+      "reference": "We already have a `seq = pkt[TCP].seq` so you can just re-use it."
+    },
+    {
+      "prediction": "I think this import is not needed.",
+      "reference": "We avoid wildcard imports because it isn't clear where symbols are coming from and there is potential for collision. Could you roll back this change? (As well as the additional newline, we don't use breaks in imports either.)"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `send_resource_list.at(0)->send(message, 5, &input_begin, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end, &input_end,",
+      "reference": "Why are these variables renamed? Is there a conflict with some Windows specific variable?"
+    },
+    {
+      "prediction": "This file should be included in the release.inc.php file.",
+      "reference": "Do you test for file existence simply to reduce the warning message that may occur using the include? Also - I think the practice used in BOINC is to go ahead and use include_once even though it isn't strictly necessary in this case."
+    },
+    {
+      "prediction": "I think this should be `spec.duration` instead of `spec.action`",
+      "reference": "`TimeChaos` does not contains a field .spec.action"
+    },
+    {
+      "prediction": "I think this should be renamed to `RSArchive` or `Archive` instead of `RSAbstractArchive`",
+      "reference": "You need to override where possible in rs-api"
+    },
+    {
+      "prediction": "I don't think we need this. We already have the `AdditionalData` property in the `IReporter` interface. I think we can remove the `AdditionalData` property from the `IReporter` interface and just use the `AdditionalData` property in the `IReporter` implementation.",
+      "reference": "Should this have an add/update/etc pattern? What happens if I need to change some set of data? May be better to just let the callsite manipulate the dictionary."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. It's only used by the `getAlgorithmInstance` method, which is only implemented by the `Region` class. I think it should be moved to the `Region` class, not the `PyRegion` class.",
+      "reference": "I don't think this should be added to PyRegion. Adding it here requires that every region have an \"algorithm\" which may not always make sense and the return value type will be different in every case. But fine to use the same name in the regions where we choose to implement it as a convention if that makes the API more discoverable."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think we should be able to detect a connection timeout on the server side and not worry about the client knowing about it. If we do that, we should be able to remove the `proxyErr` field from the `proxyError` struct.",
+      "reference": "Returning here bypasses all the health check and load balancing features."
+    },
+    {
+      "prediction": "I'm not sure if this file should be part of this PR, but I think it should be part of this PR.",
+      "reference": "Should revert this copyright change, when the file isn't changing in any other way."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the same place as the `if err := os.MkdirAll(dir, DirPermissions); err != nil {`",
+      "reference": "Interesting that filepath.Dir(\"thing\") returns \".\" whereas this returns \"\""
+    },
+    {
+      "prediction": "I don't think this is what we want to do here. We want to get the exact same result as in the docstring.",
+      "reference": "Nice, I did not know about that"
+    },
+    {
+      "prediction": "I think it would be better to use a `std::unordered_map<std::string, std::unordered_map<hsa_isa_t, std::unordered_map<hsa_agent_t, std::vector<hsa_executable_t>>>>` instead of a `std::unordered_map<std::string, std::unordered_map<hsa_isa_t, std::unordered_map<hsa_agent_t, std::vector<hsa_executable_t>>>>`",
+      "reference": "Since the lock is have moved outside of this function, I think it would be simpler and efficient to just return a ref to the std::vector<hsa_executable_t>? That way, we don't need to make a new copy on read and we won't actually need a write operation."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `ReadState` instead of `ReadStateByBlock`.",
+      "reference": "also fetch block hash of this height and return hash"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `SendHttpClientRequestAsync` method on the `HttpClientHandler` class.",
+      "reference": "Why do we need this in this sample app and not the others?"
+    },
+    {
+      "prediction": "I don't think we should add new methods to an existing interface. I think we should just remove the old methods and add the new ones.",
+      "reference": "We don't need this additional method."
+    },
+    {
+      "prediction": "Why did you change this to `return 0;` instead of `return 1;`?",
+      "reference": "Shouldn't this return code conditionally? The above line seems like an error occurred."
+    },
+    {
+      "prediction": "I think this should be `map[string]bool` and not `map[...]bool`",
+      "reference": "What is the need to use string as key? Can we use the apis type as the key?"
+    },
+    {
+      "prediction": "I think this should be `duration: typeof duration === 'undefined' ? 3000 : duration`",
+      "reference": "how about a default parameter instead?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do this in the same way we do for agent install.",
+      "reference": "i don't think we need to add duplicate check for clusterinstallref here, the function already assumes that it was invoked for clusterprovision"
+    },
+    {
+      "prediction": "`if self.with_semantic and (not 'semantic' in self.semantic_fusion or len(self.semantic_fusion['semantic']) == 0):`",
+      "reference": "\\`bbox_feats.shape[0] > 0\\` requires the number of proposal is not 0."
+    },
+    {
+      "prediction": "I think this should be `ImmutableList` instead of `Encoders.STRING()`.",
+      "reference": "Same here. Any way to fit on one line?"
+    },
+    {
+      "prediction": "I think this should be `std::stol(log.subpiece(3).toString())` instead of `log.subpiece(4).toString()`",
+      "reference": "using folly::to is better ?"
+    },
+    {
+      "prediction": "I'm not sure why this was here in the first place, but I'm not sure why it's gone now either.",
+      "reference": "I'm guessing this is unintended?"
+    },
+    {
+      "prediction": "I think we can remove the `&& !this.hasReceived` here.",
+      "reference": "This looks a little suspicious to me, can you explain what's going on here?"
+    },
+    {
+      "prediction": "Style/SymbolArray: Use %i or %I for an array of symbols.",
+      "reference": "Can you talk about this? I'm not sure I follow why this is necessary."
+    },
+    {
+      "prediction": "I think this should be `elseif (isset($params['layout']) && $params['layout']=='simple')`",
+      "reference": "Is there really a need for this 'simple' layout? Is there a reason you can't use 'lightbox'? The only difference seems to be that the lightbox layout includes Piwik/Google Analytics tracking and simple does not. If tracking needs to be disabled for some reason, perhaps there is a way to do that without creating a whole new layout."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it should go in `MergePolicy`.",
+      "reference": "Can you say what exception will be thrown in that case (or add an `@throws`, below)?"
+    },
+    {
+      "prediction": "I think this should be 2.7 or greater is required",
+      "reference": "Is this meant to be \"2.7\" or \"2.6\". The two lines in this commit don't agree with one another. Looks like everything else in this PR is \"2.6\" and that seems sensible."
+    },
+    {
+      "prediction": "I don't think you need to import `ConsecutivePatternMachine` in `nupic/data/generators/__init__.py`.",
+      "reference": "Should we put these tests in a `generators` directory?"
+    },
+    {
+      "prediction": "I think it would be better to use `std::isfinite` instead of `!std::isnan`.",
+      "reference": "I think that it makes more sense to keep these with the standard definition of non-zero is true and zero is false."
+    },
+    {
+      "prediction": "I think this should be `self.prefixed_userid, 'system.Everyone'` and not `self.prefixed_userid, 'system.Authenticated'`",
+      "reference": "Any idea why you want to change the behavior here? Is there a security risk not to have the prefix in the principal here."
+    },
+    {
+      "prediction": "I think this should be called `mockDirtyBlockCache` instead of `mockDBcache`",
+      "reference": "this reads like \"mock database cache\", maybe a clearer name"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Nit: move this with the other CM imports."
+    },
+    {
+      "prediction": "I think this should be `Symfony\\Component\\HttpFoundation\\File\\UploadedFile` instead of `Symfony\\Component\\HttpFoundation\\BinaryFileResponse`",
+      "reference": "This change should be mentioned in the CM"
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. I think it should be something like `EmptySpanContext` or `EmptySpan` or something like that.",
+      "reference": "Do we really want a new type, or can we just use `SpanContext`? The `ISpanContext` interface is read-only, but there's nothing stopping users from casting this to `SpanContext` and modifying it."
+    },
+    {
+      "prediction": "Why do we need both NET461 and NET452?",
+      "reference": "Should we define a const like NETFRAMEWORK or NETFULL which will be set for NET452 NET461 and anything like NET47 etc in future?"
+    },
+    {
+      "prediction": "I think it's better to make it a constant in `types.go` instead of a variable in `ProvingPeriodBlocks.go`.",
+      "reference": "maybe start an issue to capture params that we need final values for and link to it here?"
+    },
+    {
+      "prediction": "I think this should be `if not backend.get('group') or not backend.get('group').startswith('instance:')`",
+      "reference": "Can we instead do the check in find_instance_group_by_url() ? if we return None when group is not found, it will hit the next if statement (if not instance_group) and continue as expected, in this case we don't need to add 3 separate if statements"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for `PM` and `a.m.` and `p.m.`",
+      "reference": "on my android \".a.m\" and \".p.m\" is used add `||` here to support both variants"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to keep the check for `FormatIsMultiplane(image_format)` in `PreCallValidateCreateImageView` and remove it from `PreCallValidateCreateTextureView` and `PreCallValidateCreateCombinedTextureView`",
+      "reference": "`!FormatIsMultiplane(image_format)` will always be true at this point due to the test on line 4639."
+    },
+    {
+      "prediction": "I don't think we need to import `CachedAccessor` from `pyspark.core.accessor` here.",
+      "reference": "Maybe inline the CachedAccessor code? Is it similar to our lazy property? If yes, can we reconcile the two?"
+    },
+    {
+      "prediction": "I think we should rename this function to `AttachSmokeTestsToAPI` since it's not just attaching smoke test cases to the API model.",
+      "reference": "Nit this file has unintended changes."
+    },
+    {
+      "prediction": "I think this should be `selectedItem.isPlayable()`.",
+      "reference": "Potential NPE? `hasMedia && selectedItem...`"
+    },
+    {
+      "prediction": "`snapshot` is not used in this method, it should be removed from the docstring.",
+      "reference": "PEP-8/consistency with the rest of the code."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this instrumentation. I think it should go in the `MongoDb` assembly, since that's where the integration is defined.",
+      "reference": "nit: Would you mind moving the `ParameterTypeNames` assignment into each of the method-specific attributes? That could reduce confusion if we later decided to instrument other methods in MongoDb"
+    },
+    {
+      "prediction": "I think this interface is not used anywhere, so it can be removed.",
+      "reference": "Is exposing SpecProvider needed here?"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to do this, but I don't have a better idea off the top of my head.",
+      "reference": "So this will only be executed when it is true. I'm not sure that makes it always usable. Why not make the afterAction, `Action<bool>` and always call it, with the result as parameter."
+    },
+    {
+      "prediction": "I think this should be `if (result == PFS_RESOLVE_LOCAL) {`",
+      "reference": "Tim, please add { } to this 'if'."
+    },
+    {
+      "prediction": "I think this should be `DefaultHost is the host that will be used when none is specified.`",
+      "reference": "Suggest moving to `enable.go` since thats only place used."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "This changed when I rebuilt my docker container. Not quite sure why, but might have to do with what my host network configuration looks like when it's rebuilt."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Local cache of what? What's stored in it?"
+    },
+    {
+      "prediction": "I don't think this is the right place to set the sort order. I think it should be done in the `copyList` method.",
+      "reference": "If the copied `DataFile` returns null, shouldn't the copy also return null? Why not make the builder use `Integer` instead of a primitive here?"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(path, 'html_files', filename)` instead of `os.path.join(HTML_FILES_PATH, filename)`",
+      "reference": "You could just do this test in the `save_html` function above, and then the users of the save html function don't have to worry about it."
+    },
+    {
+      "prediction": "I think this should be `fmt.Errorf(\"get environment %s config: couldn't find environment dev in the application my-app\", \"dev\")`",
+      "reference": "I think I'd prefer not using the formatted string when it comes to unit test, since we don't pass in any string variable as params."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this change. It's not related to the other changes in this file, so I think it should be in the `mimxrt1062` file.",
+      "reference": "Maybe it would be better to explicitly include devices instead of explicitly excluding devices?"
+    },
+    {
+      "prediction": "`for (auto& atom: atoms()) {` ?",
+      "reference": "This could be `for (auto bond: bonds()){`, right?"
+    },
+    {
+      "prediction": "Shouldn't this be `supportsConnectingLibraryCard`?",
+      "reference": "Is $config no longer being used? Can we remove this line?"
+    },
+    {
+      "prediction": "I think this should be `char* remove_ext(const char* path, const char dot, const char sep, size_t* allocated_size)`",
+      "reference": "can you revert changes to this file."
+    },
+    {
+      "prediction": "Is there any reason to remove the domain name from `templateRole.getName()`?",
+      "reference": "I don't believe the change is sufficient to correctly handle variable substitutions in the name. As part of the process command we pass the original role object that was retreived without taking into account the substitution. So while the first template apply command will work fine because the original role does not exist thus it's a new add operation, if you try to re-apply the same template again, we'll pick up and process a wrong original name thus we'll try to execute an add operation again instead of modify. So we need to take into account the substituted name when we retrieve the original role as well. Same change should be applied to both policy and service blocks as well."
+    },
+    {
+      "prediction": "I think this should be renamed to `IsRunningOnMainThread` instead of `IsOnMainThread`",
+      "reference": "Please add this to IMvxMainThreadDispatcher as well"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "Is it possible for opt not to be a string? I couldn't tell if this was a typo or if there is actually a case where it is not a string."
+    },
+    {
+      "prediction": "I think this should be `googlesitekit-user-input__notification--mobile`",
+      "reference": "Let's move this up next to the `id` as we usually have `className` as one of the first props."
+    },
+    {
+      "prediction": "I think this should be `The `quilt3` PyPi package allows you to build, push, and pull data packages in Quilt using Python. Visit the `documentation quickstart <https://docs.quiltdata.com/quickstart>`_ for more information.`",
+      "reference": "While you're in here, \"build, push and install\"?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Unit never made it to spec...so removing."
+    },
+    {
+      "prediction": "I'm not sure if we want to add a new field here, or if we want to keep the existing one, but I think it would be better to keep the existing field, and just add a new field for the type and id.",
+      "reference": "We have RequestStart which seems redundant a bit to this one, WDYT?"
+    },
+    {
+      "prediction": "I think we can get rid of `createElement` now that we're using `createElementWithAttributes`, right?",
+      "reference": "I think we can remove this corresponding export from `preact` now! Double check no other s using though lol"
+    },
+    {
+      "prediction": "Extra empty line detected at block body end.",
+      "reference": "I'm not sure why this would be in here. Were you working off a branch based off of master? These lines were removed in a previous commit because the table is actually called 'cart_item_traits'."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `app/services/travis.rb` since it's only used there.",
+      "reference": "Maybe `DEFAULT_GITHUB_API_ENDPOINT` would be a better name, since we refer to this as an API endpoint elsewhere?"
+    },
+    {
+      "prediction": "nit: I think we should rename `ecs.capability.` to `ecs.attribute.` to be consistent with the rest of the codebase.",
+      "reference": "If/when you rebase from `dev`, this is going to cause a merge conflict. I'd suggest that soon after merging this PR as the capabilities code has been moved to \"agent/app/agent_capabilities.go\""
+    },
+    {
+      "prediction": "I think this test is a bit too specific to this PR. Can we move it to a separate PR?",
+      "reference": "What is the test case here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to start the cache sync. I think we should start the cache sync in `flowAggregator.Start` and then wait for it to finish in `flowAggregator.WaitForCacheSync`.",
+      "reference": "You could try moving this closer to the call, where we request label info. We might be doing the check very early.. all the resources may not be present with the informer at this point."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put these fields. I think they should be in the `ManifestFile` class itself, and not in the `ManifestFileSchema` class. The reason is that the `ManifestFileSchema` is only used by the `ManifestFile` class, and not by any of the subclasses. So it would be better to move these fields to the `ManifestFile` class itself, and not to the `ManifestFileSchema` class. This would make it easier to maintain the schema for the `ManifestFile` class, and make it clear that the schema is specific to the `ManifestFile` class, and not to any subclass of `ManifestFile`.",
+      "reference": "Can we add these up by the data files counts?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `resolveChildren` function, since that's where the children are resolved.",
+      "reference": "Most of the time assigning won't save bytes unless used 3+ times (var adds 3bytes)"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `metadata` is a `JSON` object, not a string. So when you try to read it with `JSON.parse`, you get a `ReferenceError` because `JSON.parse` expects a string, not a JSON object. You need to convert the `JSON` object to a string before you pass it to `JSON.parse`.",
+      "reference": "I would rather find the code that inverts the logic and remove that."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to just set it to `socket._GLOBAL_DEFAULT_TIMEOUT` and have it be the default.",
+      "reference": "The else doesn't return anything?"
+    },
+    {
+      "prediction": "Why do you need this change?",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "This should be `std::vector<std::string>`, not `std::vector<std::string*`",
+      "reference": "why not simply `paths = search_path_opts.LibrarySearchPaths` ?"
+    },
+    {
+      "prediction": "I think this should be `globalize._translate(item.Name)` since `_translate` is a private method.",
+      "reference": "Are we sure this should be translated by the web client? It was unclear in chat exactly what's getting translated here."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in `PutWrite`.",
+      "reference": "same oneof issue."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "If we resolve here, isn't the whole `module_install` config setting or any subkeys also pluggable? I think that's totally fine, just want to make sure that that's known, and we should also update the data in `options.rb` for those options"
+    },
+    {
+      "prediction": "I think this should be `return hipErrorInvalidConfiguration;`",
+      "reference": "it would be better to change `kds` into a `std::vector` then we don't need to explicitly free it"
+    },
+    {
+      "prediction": "I think this should be moved to the top of the file, so that it's clear what's going on.",
+      "reference": "Looks like you're also changing logic and not only tests."
+    },
+    {
+      "prediction": "I'm not sure if this is a valid assumption, but I think it should be `1.15`",
+      "reference": "Is it possible that test failures here are real and that we should improve the rate limiting logic? Or are we ok with rate limits to be exceeded by 15%?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the `MediaProviderInterface::getBox()` method.",
+      "reference": "Shouldn't there be some kind of validation somewhere? IMO there should be an exception if `$options['lazy']` is not a boolean."
+    },
+    {
+      "prediction": "I don't think this is necessary. The `using` statement is only necessary if there is an explicit `namespace` declaration. In this case, there is no explicit `namespace` declaration, so the `using` statement is not necessary.",
+      "reference": "nit: please move it inside namespace."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could come up with that worked.",
+      "reference": "The inventory schema needs to be regenerated to include these options. Looks like the CI job didn't get triggered since the paths don't include `lib/bolt/transport/**`."
+    },
+    {
+      "prediction": "I think it would be nice to have a link to the NVDA docs for these commands.",
+      "reference": "This might break if a user runs the system tests with a system language other than English, in which the user default language differs."
+    },
+    {
+      "prediction": "I think this is the right fix, but I don't know if it's the right fix for the right reason.",
+      "reference": "This was raising intermittent errors, since Lorem ipsum contains the word \"error\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place to check for participant removal. I think it should be in `RTPSDomain::addRTPSParticipant`. The `removeRTPSParticipant` function removes an existing participant from the domain, and the `addRTPSParticipant` function adds a new participant to the domain. The `removeRTPSParticipant` function doesn't check for the validity of the existing participant.",
+      "reference": "Shouldn't we then add `mp_impl = nullptr` to the destructor of `RTPSParticipant`"
+    },
+    {
+      "prediction": "`@patch('mmdet.datasets.CocoDataset.load_annotations', MagicMock())` and `@patch('mmdet.datasets.CustomDataset.load_annotations', MagicMock())` are redundant.",
+      "reference": "Are the additional brackets necessary?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. It means that you can't get the operation meta if you don't know the method name. I think it would be better to have a method `findOperationMeta(String method)` which would return the operation meta associated with the given method name.",
+      "reference": "consumerMeta only belongs to this invoker instance only relate to this one consumerIntf class seems no need to build a complex key?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it should be moved to `spec/factories/formatters/base_text_formatter_spec.rb`.",
+      "reference": "this one is unrelated right? (Don't mind including it, just making sure I understand)"
+    },
+    {
+      "prediction": "`lr_policy` should be set in `base_runtime_config.py`",
+      "reference": "8, 11 actually achieves similar performance, we should use our default config if [9,11] is unnecessary."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `yarpc.go` since it's the only place that uses it.",
+      "reference": "For a separate PR: Can we make this a JSON/Thrift procedure instead? It could accept the amount of time it needs to sleep as an argument."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it's better to leave it as it is.",
+      "reference": "Please remove this blank line."
+    },
+    {
+      "prediction": "This should be placed in the `Thelia\\Core\\Event` namespace.",
+      "reference": "the cartTrait is not used anymore"
+    },
+    {
+      "prediction": "I'm not sure if we want to add the `Action<TracerBuilder>` to the `configure` parameter here. I think it would be better to have the `configure` take a `Func<TracerBuilder>`.",
+      "reference": "Cn you please make the message more human friendly."
+    },
+    {
+      "prediction": "I don't think you need to import `QWebEnginePage` here.",
+      "reference": "Please do `import enum` instead, then use `enum.Enum` and `enum.unique` - we do this everywhere to see where things are coming from, except for Qt (everything begins with a `Q` anyways) and typing (mostly used in type annotations, so it's clear without the namespacing)."
+    },
+    {
+      "prediction": "I think this should be removed. It's not used anywhere.",
+      "reference": "Isn't ENCFAIL used in codec.c below? Wouldn't this make it no longer compile?"
+    },
+    {
+      "prediction": "I think this should be `public final boolean`",
+      "reference": "this looks very weird"
+    },
+    {
+      "prediction": "I think we can remove this completely.",
+      "reference": "So, the culprit was actually this class \"ConsoleLogHandler\", correct? Because it simply wrote to stdout..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `NLog.Config` assembly.",
+      "reference": "Still curious why you need to modify this method? Why not in the future just have a method called `LogFactory.ReloadConfiguration()` instead of the config assigning itself?"
+    },
+    {
+      "prediction": "I think this should be `cleanup, verify, create, prepare, converge, idempotence, side_effect, verify, cleanup`",
+      "reference": "Should this say \"lint, cleanup, destroy, dependency, syntax, create, prepare, converge, idempotence, side_effect, verify, cleanup, destroy\"?"
+    },
+    {
+      "prediction": "I'm not sure this is the right error code to use here. I think we should be using `pb.Error` instead of `err` to make it clear that this is an error returned from the server.",
+      "reference": "I am not sure why these log messages are changed."
+    },
+    {
+      "prediction": "I think this should be `unsafe.Pointer` instead of `*unsafe.Pointer`.",
+      "reference": "This panic is incorrect - this is a full implementation of a slice equality check. This should be something more like `type x is not comparable` maybe?"
+    },
+    {
+      "prediction": "I'm not sure why this change is needed, but I'm not sure it's correct either. The `MIN` function is supposed to return the minimum of two values, but here it's returning `len + 1` instead of `len`.",
+      "reference": "See below: let's use the existing convention _tcsnlen."
+    },
+    {
+      "prediction": "I don't think we need this anymore?",
+      "reference": "Why don't we need this any more?"
+    },
+    {
+      "prediction": "I think it would be better to put this at the top of the file, so that it's easy to find when you're working on this file.",
+      "reference": "clean unnecessary comma"
+    },
+    {
+      "prediction": "Why do we need `enumerable: false` and `writable: true`?",
+      "reference": "This is the default value for enumerable."
+    },
+    {
+      "prediction": "`auto begin = getStartIterator(rxn, type);` and `auto end = getEndIterator(rxn, type);`",
+      "reference": "Just above, you have `const auto` for a `RDKit::MOL_SPTR_VECT::const_iterator`, here it is only `auto`. Why is this?"
+    },
+    {
+      "prediction": "`release()` is not needed here, it is only used in `registerEvent()`.",
+      "reference": "@djarek just a quick question, wouldn't be `ptr.reset()` here better as it does not memory leak if we do not call delete on the raw pointer aswell? it maybe makes sense to use `ptr.release` if the event is guaranteed to be loaded again but on a reload it would memory leak if for example I removed that script from my files before reloading or not?"
+    },
+    {
+      "prediction": "I think we can drop the `*cli.StringSlice` here.",
+      "reference": "Why to `Get()` value twice, while u have it 2 lines before"
+    },
+    {
+      "prediction": "I think we should rename this to `october.fresh` to be consistent with the other console commands.",
+      "reference": "There is a typo here, this needs to state `october.fresh`"
+    },
+    {
+      "prediction": "I think this test should be moved to `org.apache.gobblin.util.TestHadoopTables`",
+      "reference": "Metadata tables loaded through `HadoopTables` will have their names as location.type (which is weird as we normally use location#type). I am not sure whether it is a big deal or not."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `onScroll` event handler.",
+      "reference": "Can you explain why you went from a positive value (metrics.viewTop) to a negative value?"
+    },
+    {
+      "prediction": "I think this should be `withLockedTasks` and `withUnlockedTasks`",
+      "reference": "Let's use a dedicated DTO instead of `List<String>` to remain consistent with all other existing REST API endpoints: * Introduce a new DTO class (e. g. `ExternalTaskTopicNameDto`) located under `org.camunda.bpm.engine.rest.dto.externaltask` * The class should have the attribute `topicName` of type `String` * Introduce a getter & setter for the attribute * Convert the list of strings to a list of `ExternalTaskTopicNameDto`s * Change the method signature to return `List<ExternalTaskTopicNameDto>`"
+    },
+    {
+      "prediction": "I think this method should be called `getScriptById` instead of `getScript(String id)`.",
+      "reference": "Do you plan to remove the `ScriptDTO getScript(List<String> path);` method long-term?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `enabledTypes` property is an array of objects, not an array of strings. You need to iterate over the array of objects and check the `enabledItem` property for truthiness.",
+      "reference": "@itsiprikshit I used a custom v-model to app count component because the sourceapps component was not reacting to user app count selection, e.g. whenever user changed app count selection, the source apps selection multiplicity remained the same. As a matter of fact, user was able to select one application only. Please let me know if you have any thoughts on this."
+    },
+    {
+      "prediction": "`from mmdet.models.utils import ResLayer, SimplifiedBasicBlock` -> `from mmdet.models import ResLayer, SimplifiedBasicBlock`",
+      "reference": "Similarly, we think we may keep `fused_semantic_head.py` unchanged. Then, we could add a new mask head for the desired function."
+    },
+    {
+      "prediction": "I think this should be `UnsupportedOperationException` instead of `NotSupportedError`",
+      "reference": "Why implement this? It duplicates the value in FeatureConfig."
+    },
+    {
+      "prediction": "I think we can remove the TODO here.",
+      "reference": "@aokolnychyi / @RussellSpitzer / @holdenk Spark3 gives ScanBuilder - abstraction - to define all types of Scans (Batch, MicroBatch & Continuous). But, the current implementation / class modelling - has SparkBatchScan as the Scan implementation. Looking at some of the concerns of BatchScan - all the way from the State maintenance of a single SnapshotId to read from, the asOfTimeStamp & features like VectorizedReads - all of these don't seem relevant to Streaming Scans. So, I feel that we need to divide out Streaming Scans into a different class. Does this thought process - make sense? If we go by this route - do you folks know - how to pass different Scan objects to Spark based on Batch vs Streaming?"
+    },
+    {
+      "prediction": "I think we can remove the `claimName` key from the `result` map as well.",
+      "reference": "Will you please update these files to use a copyright end date of 2021?"
+    },
+    {
+      "prediction": "I think this should be `var originalChecks, originalAudit, originalRules, originalCommons = axe.commons;`",
+      "reference": "Adding fixture to the body as we don't have the `runner.tmpl` anymore"
+    },
+    {
+      "prediction": "I think this should be `apis/core/v1` instead of `k8s.io/apis/core/v1`",
+      "reference": "nit: just for consistency we can move this import block down. Right above openshift/hive imports"
+    },
+    {
+      "prediction": "I think this should be `name: Optional[str]`. The `of the linter` part is redundant.",
+      "reference": "Perhaps clean up the commit history and don't squash?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `content.length()` returns the number of bytes in the array, not the size of the array. So if the array is empty, `content.length()` will return 0, not the number of bytes in the array, which is `content.array().length()`.",
+      "reference": "Just `writeBytes(content.array())` Back to my original, somewhat broken version ;)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test cross-docking.",
+      "reference": "I couldn't tell from reading the test - do we every criss-cross transports in the same test? I want to make sure we aren't breaking the chain..."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "File is not `goimports`-ed"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I could come up with.",
+      "reference": "do we want to still validate that the context had a nonzero TTL? i suppose it's fine to let clients determine whether or not their streams will timeout (though we still don't have the means to cancel them from the server other than EOF), and they can still cancel their own contexts - so maybe not a huge deal, but thought it was worth bringing up."
+    },
+    {
+      "prediction": "I don't think this is a good place to put this method. I think it would be better to put it in `NameFormatter` itself.",
+      "reference": "Go ahead and provide the same implementation for all of the languages - I'm not sure there's a reason for it to be different."
+    },
+    {
+      "prediction": "I'm not sure if this is a good test. I think it would be better to have a test that checks that the response is what you expect. For example, if you want to test that the response is a list of settings, you could do something like `assert self.get_json(\"/options\") == [{\"anticache\": True}, {\"wtf\": True}]`",
+      "reference": "We should also test type confusion here (e.g. `{\"anticache\": \"foo\"}`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `unformatted` message, but I'm not sure if that's the best place to put it.",
+      "reference": "I'd love to trim the content of `unformatted` too. Any ideas?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "can you remove this unrelated change?"
+    },
+    {
+      "prediction": "I think this should be `background-color: {{ conf.colors.hints.bg }}, color: {{ conf.colors.hints.fg }}, font: {{ conf.fonts.hints }}, border: {{ conf.hints.border }}, padding-left: -3px;`",
+      "reference": "Would using `HintLabel` here instead (without `setObjectName`) work?"
+    },
+    {
+      "prediction": "Nit: I think it would be nice to have a `cloudFlare` constructor that takes `serviceAccountFile`, `serviceAccount`, `apikey`, `dns01Nameservers`, and `ambient`.",
+      "reference": "From what I can tell, the `serviceAccountFile` is never set to anything except \"\"? It'd be best to remove this altogether, so we don't mislead future developers"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `app/shared/components/navbar/navbar.html.erb`.",
+      "reference": "Why do we remove the Sign in link?"
+    },
+    {
+      "prediction": "I think we can remove this method altogether, it's not used anymore.",
+      "reference": "holy crap, were we really missing a 'break' statement here?"
+    },
+    {
+      "prediction": "I think we can remove `debugForcedLogMaker` interface and just use `debugLogMaker` directly.",
+      "reference": "This doesn't seem to be used anywhere, probably doesn't need to be a new interface."
+    },
+    {
+      "prediction": "I think this should be `statfs` instead of `statfs`",
+      "reference": "This is a good start, but what do you think about just using `libkbfs.GetCurrentSessionIfPossible()` to avoid calling this altogether if there's no session? That way we can avoid continuous RTTs to the server when logged out."
+    },
+    {
+      "prediction": "Use 2 spaces for indentation in a hash, relative to the start of the line where the left curly brace is.",
+      "reference": "This `aria-labelledby` stuff is all new to me, but my reading of the spec is that the value should be a list of element id values, and `dLabel` doesn't seem to be the id of any element?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in the `Upload` class, since it's only related to uploading a single file.",
+      "reference": "Does this fail validation if you now use the `upload` key instead of `source`?"
+    },
+    {
+      "prediction": "I think this should be `const char*` instead of `const std::string`.",
+      "reference": "Can we use GetCliGlobals(false)->exCollHeap() here instead"
+    },
+    {
+      "prediction": "I think this should be `std::unique_ptr<kvstore::KVStore>(TestUtils::initKV(rootPath.path(), partitions, {0, network::NetworkUtils::getAvailablePort()}))`",
+      "reference": "I suggest you move the waitUntilAllElected into init::KV() method."
+    },
+    {
+      "prediction": "I'm not sure if we want these to be configurable, but I think it's worth it to make them configurable.",
+      "reference": "no need of pub;lic."
+    },
+    {
+      "prediction": "I think it's better to make this a `bool` and set it to `true` by default. This way, we don't have to worry about the future of `UseBuggyProposalLowestOutput` and we can just set it to `true` by default.",
+      "reference": "Do you think we can name this variable in a less ...negative.. language ? maybe `ProposalTieBreakerEqualizer`"
+    },
+    {
+      "prediction": "I'm not sure why we need to pass in the inputSchema here. I think it's enough to pass in the spec. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why we need to pass in the inputSchema here. I'm not sure why",
+      "reference": "Can you add back these newlines? We try to avoid non-functional changes like this because they can cause commit conflicts."
+    },
+    {
+      "prediction": "`filter` is not a function, so you can't call it directly. You need to use `map` or `filter` on an empty array.",
+      "reference": "Why not just `if (!searches.includes(action.search)) searches.push(action.search)`? That way you don't need to remove then add the element."
+    },
+    {
+      "prediction": "I think this should be `substr(str_shuffle($codes), 0, rand(0, 3));` and not `substr(str_shuffle($codes), 1, rand(0, 3));`",
+      "reference": "@demiankatz I'm not sure that this is very useful. I just did it to get started. If we had a small handful of predetermined prefixes that were randomly chosen, it would be easier to distinguish them from the rest of the call number. Would that be preferable?"
+    },
+    {
+      "prediction": "I think we can remove this line too.",
+      "reference": "Why is this deleted too?"
+    },
+    {
+      "prediction": "I don't think we need to dispose of the response here, since it's only used in the constructor.",
+      "reference": "Given that IApiResponse is already made disposable, this is not necessary."
+    },
+    {
+      "prediction": "This should be `if (!menuEntry.getTarget().contains(\"High Level Alchemy\") || plugin.isShowAlchProfit()) return null;`",
+      "reference": "client.getSelectedSpellName() can be used for this"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `tests/testdata/`.",
+      "reference": "Please alpha sort these filenames."
+    },
+    {
+      "prediction": "I think we can remove this check as well, since it's now handled by the validator.",
+      "reference": "is this already called by actPool.validate()?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this check. I think it should be done in `Mol::updateStereoBonds`.",
+      "reference": "It is relatively cheap to tell whether a bond is a ring bond, we don't have to perform an SSSR."
+    },
+    {
+      "prediction": "I think this should be `appTypeFlag = \"app-type\"` and not `svcTypeFlag = \"svc-type\"`.",
+      "reference": "Maybe we should add flag aliases for service as well."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `Block` struct.",
+      "reference": "This LGTM but don't we need a winning PoSts field to fully implement the protocol? No need to add here as I'm working on this in another PR, but curious how we can interop with lotus blocks without post proofs."
+    },
+    {
+      "prediction": "I think we can remove this interface and use the buffered channel directly in the subscriber.",
+      "reference": "can you move the [] into pubSub? so Blockchain just contains a pubSub, not []"
+    },
+    {
+      "prediction": "I think this should be `circuitBreaker/useCircuitBreakers`",
+      "reference": "I don't think 100 is a safe default here, since later we check that the value is between 50-95."
+    },
+    {
+      "prediction": "I don't think this is correct. I think it should be `ThreadSafe = 1 | 2`, where `2` is `ThreadAgnostic` and `1` is `ThreadSafe`.",
+      "reference": "I would be nice if we could describe this without the word \"agnostic\""
+    },
+    {
+      "prediction": "I don't think this is the right place for this check. I think it should be in the function that generates the sparse input pattern.",
+      "reference": "Pick more specific exception types for these cases. Not sure what would be best for this one (`ValueError`?) but the next one could be `IndexError`."
+    },
+    {
+      "prediction": "I think this should be `testMatch: [ '<rootDir>/.storybook/*.test.js' ],`",
+      "reference": "This seemed like the easiest solution We need to include this path in `testMatch` to run the test AFAIK *BUT* we don't want this test ran when running them all normally I'm sure there are other ways to do this!"
+    },
+    {
+      "prediction": "I think this should be `Cannot flush closed transport, message, yo`",
+      "reference": "Most of the following test changes can be avoided if need be."
+    },
+    {
+      "prediction": "I think it would be better to move the `CellStorage` construction into the constructor of `MultiLevelPartition`, so that the `CellStorage` can be constructed in a way that is consistent with the other parts of the constructor.",
+      "reference": "Eventually we might want to move this back as generating this data structure takes some time (for large datasets) and we don't want it in the hot-path of `osrm-customize` runs. Thinking long-term making `osrm-customize` eventually digest `.osrm.cells` will will enable only updating cells selectively and using an earlier customization for the rest."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think we should wait for the NAT service to stop first, and then wait for the VPN server to stop. This is because the NAT server is responsible for stopping the NAT service, and the VPN server is responsible for stopping the VPN service. If we wait for the NAT server to stop first, we can then safely stop the VPN server, which would prevent the NAT server from becoming stuck waiting for the VPN server to stop.",
+      "reference": "You're not really using named result variable - you can just leave `error`."
+    },
+    {
+      "prediction": "I think this should be `SortOrder.unsorted()` instead of `SortOrder.unsorted()`.",
+      "reference": "It wasn't possible to call this method with `PartitionSpec.unpartitioned()` as the schema in the spec was empty."
+    },
+    {
+      "prediction": "Why do you need to use `Hook` here?",
+      "reference": "this namespace is never used"
+    },
+    {
+      "prediction": "I think we should add this to the changelog as well.",
+      "reference": "Well, if you call `Nack`, it _will_ be redelivered. So maybe reword."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `wistia` repo.",
+      "reference": "Why did we make this a static string?"
+    },
+    {
+      "prediction": "I think this should be `//cloudresourcemanager.googleapis.com/projects/{}'.format(parent['projectNumber'])` and not `//cloudresourcemanager.googleapis.com/projects/{}'.format(parent['projectNumber'])`",
+      "reference": "Please re-run PYTHONPATH=. python tests/services/inventory/update_cai_dumps.py following the instructions at the top of this file to ensure the test files have the correct CAI asset type."
+    },
+    {
+      "prediction": "I think this should be `NewStateDBAdapter(sm, 0, !cfg.Genesis.IsAleutian(0), cfg.Genesis.IsGreenland(0), cfg.Genesis.IsKamchatka(0), cfg.Genesis.IsLordHowe(0), hash.ZeroHash256)`",
+      "reference": "`stateDB := NewStateDBAdapter(sm, 0, hash.ZeroHash256, NotFixTopicCopyBugOption())`"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "There are unneeded parens here, but I'll fix it up when merging."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `h2o_ruby.h` instead.",
+      "reference": "Just one more thing... How about doing all of the following? 1. renaming h2o.rb to bootstrap.rb 2. require and include the acl code _from_ bootstrap.rb"
+    },
+    {
+      "prediction": "I think this function is still used in `flux_monitor.c` and `flux_monitor.h`, so it should be removed from there as well.",
+      "reference": "Commit message: \"remove\" is misspelled."
+    },
+    {
+      "prediction": "I think this should be called `trailingZeros` instead of `toTrim`.",
+      "reference": "This is a pretty idiosyncratic method - I'd move it into the JSON-RPC package rather than adding it to `BytesValue`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to `Rspec::Matchers` so that it can be used by all of the `RSpec::Matchers` and not just the derived ones.",
+      "reference": "Is there a way to deduplicate this somehow? Could we roll it into `RSpec.configuration` with a bit of memoization or is there a chicken and egg problem."
+    },
+    {
+      "prediction": "I think this should be called `PROP_UPDATE_IN_PROGRESS` to be consistent with `PROP_UPDATE_NOT_IN_PROGRESS`",
+      "reference": "Generally, I wouldn't say I like the idea of providing properties that return negative/inverted values. It would be more convenient to return whether the update is **in** progress or not."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix, but it's the only thing I can come up with right now.",
+      "reference": "Converting playbackManager to operate in ms (preferred solution, imo) would allow us to completely drop this function."
+    },
+    {
+      "prediction": "I think this should be `10000` instead of `1000`.",
+      "reference": "The default value should be a bigger one, I'd like it is the maximum of int32. That means we will not cut-off anything by default. And users would config it with an appropriate value in config-file, such as 1000 or 5000, etc."
+    },
+    {
+      "prediction": "I think this should be `clio.Info(\"Waiting for settlement to complete\")`",
+      "reference": "Probably missing return here as well"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be under `caddy.config` since it's a global setting.",
+      "reference": "On second thought... this affects how often the function specified by the `load` property is run. So I wonder if, for consistency, we should call this `LoadInterval` instead. Does that make sense?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `common` module since it's used by both `countly-common.js` and `api/lib/countly.common.js`.",
+      "reference": "Now you don't need the localization module here then :)"
+    },
+    {
+      "prediction": "I think this should be `eq(ref, toLiteral(v))` instead of `eq(ref, toLiteral(v)) && eq(ref, toLiteral(v))`",
+      "reference": "This should not convert each value to a literal on every invocation. Instead, this PR can add a `literals` method to `BoundSetPredicate` to access the values as literals. That way, the conversion happens once and can be lazy."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `find_lib_path` function returns a list of paths, but the `lib_path[0]` should be the first path in that list.",
+      "reference": "I think we still need this exception if doesn't have LIGHTGBM_BUILD_DOC"
+    },
+    {
+      "prediction": "I think you can remove the `h.describeCMObject(certificate)` call here.",
+      "reference": "Does this make it easier to read? Previously this would have run `kubectl describe certificate` right? In which case it would have included events related to the cert. Will we miss that? Would the old `kubectl describe` output have gone to stderr?"
+    },
+    {
+      "prediction": "I don't think we need to expose this. It's only used for testing, right?",
+      "reference": "Just an understanding question: do we need a map here, would a list/set of Strings be sufficient? we are not adding entries with `false` here, are we? Being in the list already means it's used, right?"
+    },
+    {
+      "prediction": "I think this should be `removed = self._state.prune(task, self._config) and task.id not in necessary_tasks`",
+      "reference": "Ah, so the change is that you want the `prune()` side effect before doing the `task.id not in necessary_tasks` membership check? Stupid non-Haskell programming languages :p"
+    },
+    {
+      "prediction": "I think this file should be renamed to `testutil.go`",
+      "reference": "This change was not really intended but made by the linter of VS Code. And looking at other packages this looks like a best practise to place interal packages on the top and gh imports afterwards."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `findAndModify` function.",
+      "reference": "Do we normally force ReadPreference primary on other write operations? Or is it just that we normally ignore it?"
+    },
+    {
+      "prediction": "I think we can remove this method altogether, since it's not used.",
+      "reference": "Instead of returning null here and discarding nulls automatically, I think this should return null (or some other signaling value) and fix the part that consumes the output of the visitor."
+    },
+    {
+      "prediction": "I think this should be done in `resolveValidatorProvider` and not in `getValidatorsForBlock`",
+      "reference": "Really, we want to be setting the validatorsForPreviousBlock against the yet to be created block, so it might make sense to update the cache once the block is created/imported instead."
+    },
+    {
+      "prediction": "This should be `supported_options = sizeof(found_opts) / sizeof(found_opts[0]) - 1;`",
+      "reference": "It seems to me that the original implementation was correct. Can you share what the scanner was pointing out?"
+    },
+    {
+      "prediction": "I think this should be `./assets/js/modules/subscribe-with-google/bulk-edit.js` instead of `./assets/js/googlesitekit-modules-subscribe-tagmanager.js`",
+      "reference": "Instead of doing that, please create an `assets/js/googlesitekit-subscribe-with-google-bulk-edit.js` file as entry point - all our entry points should be located in `assets/js`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Widgets` module, since it's specific to the dashboard widget.",
+      "reference": "This is not valid, a `div` shouldn't be in a `span`. You could apply the `className` attribute to the `span` tag."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/config` since it's a config file.",
+      "reference": "Would it be possible to reuse RcJobExpireDuration and RcJobExpireInterval and auto delete them when job is deleted?"
+    },
+    {
+      "prediction": "I don't think we should be importing `print_function` here.",
+      "reference": "Python wants **future** imports to come first."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. If we're detaching from a stack, shouldn't we be detaching from the stack's parent as well?",
+      "reference": "Better to call this before EXITING_DR which is supposed to be last."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's ok, but I don't know why it was there in the first place.",
+      "reference": "Does this work at runtime?"
+    },
+    {
+      "prediction": "I think this should be 3 instead of 2",
+      "reference": "Hmm why do we need to make this change? Like which one requires this change?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "why remove this and other piece?"
+    },
+    {
+      "prediction": "I think this test should be renamed to `hasContentVirtualWithoutAria`.",
+      "reference": "Yikes... That was probably me."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `Tools` namespace.",
+      "reference": "It would be better to move the ... outside of the l8n block. This way it won't need to be translated for each language twice."
+    },
+    {
+      "prediction": "This should be moved to the bottom of the file.",
+      "reference": "That's a huge load of new dependencies. I think it's better to create a reload function on ScriptManager and control with enum instead."
+    },
+    {
+      "prediction": "I think this should be `NewUpgradeTaskJob`",
+      "reference": "`NewUpgradeTaskJob` -> `NewUpgradeResource`"
+    },
+    {
+      "prediction": "This should be `glibtop_get_meminfo` instead of `glibtop_get_mem`",
+      "reference": "Is that the memory usage of the system or the process / awesome?"
+    },
+    {
+      "prediction": "I think this should be `callback(Realm, Schema)` instead of `callback(Realm, string)`",
+      "reference": "\"change\" *and* \"schema\" events (A bunch of these; not marking them all)."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `SnapshotUpdate` since it's related to snapshots, but I'm not sure if there is a better place to put it.",
+      "reference": "I think there is probably a shorter, more descriptive name for this. Something like `commitAtSequenceNumber`?"
+    },
+    {
+      "prediction": "I think this should be `allow_nil: true` instead of `allow_nil: false`",
+      "reference": "The `url_name` should definitely be more than just 1 character long at the shortest. How about 3 or 4 characters?"
+    },
+    {
+      "prediction": "This is not a typo, it's a typo in the function name.",
+      "reference": "Please revert this change"
+    },
+    {
+      "prediction": "Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "Nice. I hadn't even heard of presenters in Rails before (learn something new every day) :)"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It should be in the `Pool` class, not in the `PoolContext` class.",
+      "reference": "`string[]` ? not sure, can you verify this @core23 ? Thank you"
+    },
+    {
+      "prediction": "I think this should be `clearTimeout(playPauseClickTimeout)` instead of `setTimeout(playPauseClickTimeout, 0)`",
+      "reference": "are we aware that this breaks older browsers like WebOS? :) they don't handle `let`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to update the TTL. I think it should be in the `set` method, since it's the only place where we're modifying the object.",
+      "reference": "This isn't really a Kinto record but a Memcached record. I could kind of go either way on this."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `OAuthWebviewHelper#setAccount` method.",
+      "reference": "Kick off the download after login."
+    },
+    {
+      "prediction": "This should be `string[]`, not `array<string, string>`.",
+      "reference": "I don't think you need to change this, `array<int, string>` should be accepted whenever `string[]` is expected. (`string[]` is an alias of `array<string>` itself an alias of `array<array-key, string>` which is a parent type of `array<int, string>`"
+    },
+    {
+      "prediction": "I think this should be `return (serial == wlr_pointer_state.grab_serial || serial == wlr_touch_state.grab_serial);`",
+      "reference": "TODO: fix this, probably in another PR"
+    },
+    {
+      "prediction": "I think this should be `if tab is not None` instead of `if tab is None or tab.tab_id != self._tab_id`",
+      "reference": "Is there any case that tab is None?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `Apply` method of the `Endpoint` struct.",
+      "reference": "pls use `err` instead of `e`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I'm not sure if there is a better way to check for the presence of the gamepads.",
+      "reference": "`attachGamepadScript` above can also be placed inside."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `libp2p/crypto.go`",
+      "reference": "when we use RegisterWriteCommand,this func should be exported,so we can use this to get signer at the outside action package"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think it would be better to log the line before sending it to the middlewares.",
+      "reference": "Dead code - we should remove it or use `log.Trace` for lower logging level. Same applies for change at line 138."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to catch this exception. I would expect it to be in the `catch` block of the `parseParameters` function.",
+      "reference": "How does this make sure invalid coordinates making the parser fail? Relying on the Coordinate type's boost numeric_cast calls in the ctor? Shouldn't we rely on a coordinate's `.IsValid()` member function instead?"
+    },
+    {
+      "prediction": "I think we should remove the `#` from the beginning of this line.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "This should be `if (!defined('CACHE_ENABLED')) { define('CACHE_ENABLED', $useCache); }`",
+      "reference": "Does this line (and the corresponding one in the test bootstrap) actually do anything? I don't see where CACHE_ENABLED is used as a global constant."
+    },
+    {
+      "prediction": "I think we should rename this to `VerifyTransitionDependenciesSatisfied` to be consistent with the other tests.",
+      "reference": "has this changed? are we testing \"all possible situations\" now?"
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `.`.",
+      "reference": "ZK is properly spelled 'ZooKeeper' (capital 'K')."
+    },
+    {
+      "prediction": "I think this should be `~/.gcloud/application_default_credentials.json`",
+      "reference": "Guessing this locates the key you get when you run `gcloud auth application-default login`? What was the previous behavior/the behavior for the manual Veneer creds?"
+    },
+    {
+      "prediction": "`columnSelected && this.countRows() > 1` -> `columnSelected ? this.countRows() > 1 : selected[0] < 0`",
+      "reference": "looks good to me, but it would be better if you assigned `this.countRows()` into a variable, so it doesn't trigger twice in the return statement"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "For now, don't add \"use strict\" to any existing files. That's a separate project that we'll do later. (New files can use \"use strict\".)"
+    },
+    {
+      "prediction": "I think this is the only test that needs to be changed to use `os.Setenv` instead of `os.SetenvNoninteractive`.",
+      "reference": "I was confused by this env name variable, assuming it was the value, not the name. Silly nit, but maybe name it noninteractiveEnv?"
+    },
+    {
+      "prediction": "`posix` is not a valid shell option.",
+      "reference": "`sys.platform` could be equal to `darwin` which is posix. Use `not startswith('win')`?"
+    },
+    {
+      "prediction": "I'm not sure we need to pass in the `Overrides` here. I think we can just use the `config.Overrides` directly.",
+      "reference": "let's move this into `slack_helper.go`"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to make sure we're passing the correct path to the control path.",
+      "reference": "I doubt it will work, but without space between -o and ControlPath it should. Lets see. What I do not understand is why we did not see a failure on CI related to this?"
+    },
+    {
+      "prediction": "I think this should be `if (shifted_bearing > 360. || shifted_bearing < 0.)`",
+      "reference": "this is the reason we might be seeing 360 as discrete bearing"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think it should go in the same package as `astroid` itself.",
+      "reference": "`FULL_VERSION` is available at module level for others to use, but it won't have `%(prog)s` substituted with pylint. Will that be a problem. Also put it here instead of __pkginfo__ with other versions because it isn't pkginfo related."
+    },
+    {
+      "prediction": "I think this should be `long` instead of `L`",
+      "reference": "Is this needed? I don't see any uses."
+    },
+    {
+      "prediction": "I think this should be `builds` instead of `builds`",
+      "reference": "Shouldn't this be V1beta1?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I think there's an `approval_for`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `org.infinispan.container.impl`",
+      "reference": "Here we should use <code>List</code> in generic"
+    },
+    {
+      "prediction": "I think this should be `javaCompile.getToolChain().getVersion()` instead of `javaCompile.getTargetCompatibility()`.",
+      "reference": "The `JavaCompile#getToolChain` method got removed and I am not sure what a good replacement is. I replaced it with `targetCompat` for now but they are not equivalent. Maybe we can use `JavaVersion#current` here?"
+    },
+    {
+      "prediction": "I think this should be `public function fileExists(string $name): ?MultimediaId;`",
+      "reference": "The method name is incorrect. It does not check the existence of the file. I'd suggest deprecating both methods `fileExists` and `findIdByHash` and use findIdByFilename instead of fileExits"
+    },
+    {
+      "prediction": "nit: I think it would be better to have this as a `static const u8 NAT_LEVEL_1_DROP = 0;` instead of a `__u8 NAT_LEVEL_1_DROP = 0;`",
+      "reference": "We have `stdbool` imported, might as well use that for clarity."
+    },
+    {
+      "prediction": "This file should be renamed to `batchstore.go`",
+      "reference": "File is empty, consider removing?"
+    },
+    {
+      "prediction": "`writeConcernMax()` returns a `WriteConcern` object, not a `WriteConcern` instance.",
+      "reference": "`writeConcernMax` was changed to return a `writeConcern` formatted the new way-- `writeConcern: {w:1, ...}`. Bulk execute takes an actual `WriteConcern` object as its first parameter (this was changed in master), so we have to un-wrap the `writeConcernMax` result here."
+    },
+    {
+      "prediction": "I think this should be `@frozen_sort = false` instead of `@frozen_sort = nil`",
+      "reference": "How about passing this through the `config`?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be using `pytest.mark.skipif` instead of `pytest.fixture`.",
+      "reference": "Out of curiosity - did we make this change to allow dynamically assigning a value to `os.environ[\"TEST_TARGET\"]` during test execution? I kind of liked the decorator style `@only_localstack` - makes the condition a bit more explicit. Looks like `skipif` also allows to specify a condition string, e.g. `pytest.mark.skipif('os.environ.get(\"TEST_TARGET\") == \"AWS_CLOUD\"')` - could that be an option? (not sure if that gets lazily evaluated at runtime right before the execution of the annotated test method starts, though..)"
+    },
+    {
+      "prediction": "I think this should be `null` instead of `0`",
+      "reference": "I noticed (SonarCloud noticed actually) that the implementation of `::edit` method is the same as `__construct` is. Does it make sense to call the `edit` method from the constructor?"
+    },
+    {
+      "prediction": "I think this should be `flb_debug(\"syslog-prot: incomplete message! (Enable trace log to see the message.)\");`",
+      "reference": "would you please rename the message prefix to: [in_syslog] ...\""
+    },
+    {
+      "prediction": "I think this should be `const MCS& mcsIdx, RWMol **molExt) const`",
+      "reference": "The pointer to a pointer is kind of gross. How about either taking the `ROMOL_SPTR` directly or, preferably, returning an `std::pair`?"
+    },
+    {
+      "prediction": "I think we can remove this function and just call `pool.pendingCount()`",
+      "reference": "the name confusing, please rename to `pendingCountNoLock` or similar"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to have a function like `DecodeHexStringToPublicKey` that does the following: 1. Decode the hex string into a byte slice 2. Create a new keypair 3. Create a new block using the public key 4. Sign the block with the private key 5. Add the block to the blockchain",
+      "reference": "line is 175 characters"
+    },
+    {
+      "prediction": "I think you can remove the `client != restClient` check here.",
+      "reference": "@ivanbogdanov Does this fix the first time load gray screen issue that @wmathurin noticed?"
+    },
+    {
+      "prediction": "I don't think we should remove this warning. It's very important for users who don't want to spend too much time and effort to figure out what's going on under the hood when they're building their own wheels.",
+      "reference": "Is `2.3.0` version OK?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "I tested it only in Firefox (`./go //rb:firefox-test`)"
+    },
+    {
+      "prediction": "I think this is the wrong place to put this.",
+      "reference": "separate third party and builtin"
+    },
+    {
+      "prediction": "I think this should be `if (!isHybrid() || !BootConfig.getBootConfig(frontActivity).shouldAuthenticate())` instead of `if (!isHybrid() || BootConfig.getBootConfig(frontActivity).shouldAuthenticate())`",
+      "reference": "@smcnulty-sfdc We do want to finish the hybrid activity in our hybrid apps though. Could the caller not pass in `frontActivity` instead? Or start it up again post-logout?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `Outbound` package, under the `Call` method.",
+      "reference": "isn't this going to return a timeout error if the context is canceled?"
+    },
+    {
+      "prediction": "I think this should be `if (argNameVal.length == 2 && !argNameVal[1].isEmpty())`",
+      "reference": "`argNameVal.length` will still be !=2, so line 22 will give true and then continue on line 24"
+    },
+    {
+      "prediction": "I think it would be better to use `preg_match('/\\w/', $chars[$i + 1])` instead of `preg_match('/\\w/', $char)`.",
+      "reference": "It was `$chars[$i]` I believe."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "This has the likely potential to break existing tests that are relying on the old beaker behavior. If we are going to release this in beaker 3.x, then we need to preserve the old behavior as well (so set both the `upcase` and original values). On Windows, they env variables will overwrite each other, with the same value, so not an issue. On *nix, there will be two env variables, but they will have the same value, so again no issue."
+    },
+    {
+      "prediction": "I think this should be `var variableOptions = variableOptions || {}` instead of `var variableOptions = variableOptions || {};`",
+      "reference": "Thanks for fixing this! I probably didn't test the playground when making changes here for cloud variables."
+    },
+    {
+      "prediction": "I think it would be better to check `selectedIds.size() > 0` instead of `selectedIds != null`.",
+      "reference": "I will also check if `nameMapping` needs a precondition null check."
+    },
+    {
+      "prediction": "I think this should be `file.getParentFile().getParentFile().getAbsolutePath()` instead of `file.getParentFile().getAbsolutePath()`",
+      "reference": "Since RuntimeIOException is deprecated and you are touching this code, why not replace it?"
+    },
+    {
+      "prediction": "I think it would be better to use `assertRaisesRegex` instead of `assertRaisesValue`.",
+      "reference": "You should also have a test that calls it with tm_cpp and make sure it returns the correct JSON file."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `actionsheet` file.",
+      "reference": "One more minor change, the string `ForAdditionalLiveTvOptions` is probably also unused."
+    },
+    {
+      "prediction": "I think this should be `LOG(WARNING) << \"Unsupported row reader version \" << (readerVer > 0 ? std::to_string(readerVer) : \"unknown\") << \" (current: \" << currentReader->version() << \")\";`",
+      "reference": "How to distinguish between warning and error?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I've been able to get it to work.",
+      "reference": "Perhaps you want to check if UIAHandler.handler is not None. The current check will break NVDA if you call UIAHandler.terminate() for some reason."
+    },
+    {
+      "prediction": "I think this should be `set(\"replace-partitions\", \"true\")`",
+      "reference": "can we make `replace-partitions` property a static variable in `SnaphotSummary.java`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be in `Selenium::WebDriver::Chrome::Bridge` since Chrome is the only supported browser.",
+      "reference": "Maybe call it `WireBridge`?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `activate` function in `saleor/account/emails/activate_email.py` file.",
+      "reference": "Should really be `one()`, not `one_or_none()`"
+    },
+    {
+      "prediction": "I think this should be `out.NodeRegistration.KubeletExtraArgs[\"node-labels\"] = fmt.Sprintf(\"%s,%s,%s\", labels[0], labels[1], nodeRole)`",
+      "reference": "Should we use strings.Split and strings.Join instead of manual concatenation?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `InputDefinition` class, since that's where it's defined.",
+      "reference": "rough that this needs to exist, but it is what it is"
+    },
+    {
+      "prediction": "I think this test should be moved to `custom-elements.spec.js`",
+      "reference": "FYI, I pulled your branch and hacked a little on your test case. This passes for me (you may want to tweak further): <pre> it('should not set empty string for null props in custom elements', () => { customElements.define('x-bar', class extends HTMLElement { val; }); // @ts-ignore render(&lt;x-bar val={null} /&gt;, scratch); expect(scratch.childNodes).to.have.length(1); expect(scratch.firstChild).to.have.property('nodeName', 'X-BAR'); expect(scratch.firstChild.attributes.length).to.equal(0); expect(scratch.firstChild.val).to.equal(null); }); </pre>"
+    },
+    {
+      "prediction": "I think this should be `512 * 1024 * 1024`",
+      "reference": "It is the maximum value that the configured emulator can support."
+    },
+    {
+      "prediction": "I think this should be `session_manager = objreg.get('session-manager') or '_autosave'` instead of `session_manager = objreg.get('session-manager') or '_autosave'`",
+      "reference": "Here, I've been giving priority to the session saved by the user. This means, that if the user quits with `:wq`, then restarts `qutebrowser`, and then `qutebrowser` crashes for some reason, next time `qutebrowser` is restarted, the session saved lastly with `wq` will be restored, and not the one autosaved. What do you think?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in `BalanceRecord`.",
+      "reference": "this line got me confused for few seconds, as the `AccountData` is both the name of the data structure as well as the function name. I think that `GetAccountData` is a better choice for a name. ( note that the same applies for the previous `BalanceRecord` function name )"
+    },
+    {
+      "prediction": "I think this should be `wait_for_at_least_entries` instead of `wait_for_at_least_entries_nts`",
+      "reference": "I think this mutex is only used here. Remove it."
+    },
+    {
+      "prediction": "I think this is the only place where we need to set the DB version.",
+      "reference": "so this is now init to 1?"
+    },
+    {
+      "prediction": "`version = -1` -> `version = 0`",
+      "reference": "For future note; not prefixing the table in a join can throw an error in the query parser depending on how the query is built I am making an assumption this was tested"
+    },
+    {
+      "prediction": "I think this is the only place where we're using the fake cluster UUID. Is this something we want to be consistent with the rest of the codebase?",
+      "reference": "Had to bring this up, will abandon this style of testing based on Joel's work soon."
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "What does our source validation complain about here? Many of the logok/nowarn places look fine to me at a glance but I'm no match for the logging policeman ;-)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `stats.go` since it's only used there.",
+      "reference": "It feels like this dependency should not be here. Some deep internal openvpn package (middleware) knows about one of possible stats consumers. Maybe as an alternative, statistics and state callbacks can be extracted as separate interfaces in top openvpn package and reused by both sides"
+    },
+    {
+      "prediction": "I think this should be `newTopBlock := topBlock.DeepCopy()`",
+      "reference": "Can you please remove `codec` from the parameters too? I think it's unused now."
+    },
+    {
+      "prediction": "I think this should be `when(typeSolver.getSolvedJavaLangObject()).thenReturn(new ReflectionClassDeclaration(Object.class));`",
+      "reference": "mocks needed this change so that it returns the \"right\" thing"
+    },
+    {
+      "prediction": "I think this should be `shiftControl` and `shiftAlt` instead of `shiftClick` and `shiftAltClick`.",
+      "reference": "I don't think the getters for this are needed, but looks good other than that"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get to work. I'm not sure if there is a better way to do this.",
+      "reference": ":+1: to inlining this."
+    },
+    {
+      "prediction": "I think this should be `mixed|null` instead of `mixed:bool|null`",
+      "reference": "Does psalm support the same logic internally for object-like arrays? Just want to be certain I can indicate this is a contextual return item `crypto?:mixed`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to remove this message, but I don't have a better suggestion.",
+      "reference": "I can't recall if there was a fix specific to this in the past?"
+    },
+    {
+      "prediction": "I think we can remove the TODO here.",
+      "reference": "Not sure what the todo is here? Is it to support some kind of defaulting based off of that?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `NLog.Targets` namespace.",
+      "reference": "Not sure about this name. I think \"explicit\" is also a bit difficult here (I think it should be implicit then) Proposal: FlushOnEvents. Or, It would be cool if we could split into 2 options, FlushOnShutdown and FlushOnReload, but I expect that's far more difficult to implement?"
+    },
+    {
+      "prediction": "I think this should be `@pytest.mark.selenium` instead of `@pytest.mark.qt_log_ignore`",
+      "reference": "I don't think that works - you can't mark a fixture."
+    },
+    {
+      "prediction": "I think this should be `ENV['PULPFILE_PATH']` or `ENV['PULPFILE_PATH'] || ENV['USERPROFILE']`",
+      "reference": "We should not include this path."
+    },
+    {
+      "prediction": "I think this should be `assert self.assertEqual(expected_output.strip(), actual_output.strip())` instead of `self.assertEqual(expected_output.strip(), actual_output.strip())`",
+      "reference": "Add newline at end of file"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to put it in the `onUpgrade` method of the migration class.",
+      "reference": "This should be done when updating to the next release (2.5). You currently only perform the upgrade when users go from 1.4 to 1.5, so it will lead to crashes for existing users."
+    },
+    {
+      "prediction": "I think this should be `ClusterName` instead of `CentralMachineManagementAnnotation`",
+      "reference": "suggest hive.openshift.io/cluster-machine-management or something to make it more obvious what it is."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this review. I think it should be in `RegisterRpcModules` and not in `InitializeBlockchain`",
+      "reference": "We explicitly don't want to do that. This was a complaint from users before."
+    },
+    {
+      "prediction": "I think this should be `int[] getExcludeCode()`.",
+      "reference": "Hello, The PHPDoc is `string[]` not `int[]`"
+    },
+    {
+      "prediction": "nit: I think it would be better to move this function to `pkg/mock/mock.go` and rename `applyTc` to `mockApplyTc`",
+      "reference": "Same issues with parameters order in `ipset_server.go`"
+    },
+    {
+      "prediction": "Shouldn't this be `SystemNamespaceRetentionDays = time.Hour * 24 * 7`?",
+      "reference": "Wow, did it literally mean the retention days is a huge number?"
+    },
+    {
+      "prediction": "I think this should be `if (DirExists(opts.project_root.c_str())) Error(arg + \" is not a directory: \" + opts.project_root);`",
+      "reference": "you probably didn't intend to touch those files in `scripts/`"
+    },
+    {
+      "prediction": "what is the meaning of `tpfp_func`?",
+      "reference": "Similar to `collate_fn`, we may rename it to `tpfp_fn`."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Awesome! We may also use negative rule `never` to fix new lines after `return`: `{ blankLine: \"never\", prev: \"return\", next: \"*\" }`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in libkbfs.",
+      "reference": "Why can't this be a function on the `Config` interface like all the others?"
+    },
+    {
+      "prediction": "I think this should be `if (feed != null && feed.getLink() != null)`",
+      "reference": "If the feed is null, the menu items should still be setup. Just the feed title can be left out. That prevents possible flickering when menu items are displayed/hidden for some feeds."
+    },
+    {
+      "prediction": "I think you can remove the `allowIncompatibleChanges` part here.",
+      "reference": "I'm not sure that this should call `allowIncompatibleChanges()` because adding a required column when there are no existing values will break reading the new column in any table with data in it. The only time it is safe to add a required column is if there is no data in the table. What about throwing an exception here instead? I agree that the column should not be optional if NOT NULL was specified. Another alternative is to check whether the table has data and allow the incompatible change if it doesn't have any rows."
+    },
+    {
+      "prediction": "This should be `Lang::translate('backend::lang.dashboard.status.widget_title_default')`",
+      "reference": "Early translation , should be logic-less"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in `proxy/interceptor.go`",
+      "reference": "I don't understand why these functions return a request/response, respectively. In all implementations we actually _modify_ the request/response given as a parameter. Do you envisage situations where we'd want to construct completely fresh request/response objects? Even if we do, it's not something needed atm, so I'd favour in keeping the API minimal. Also, the fact that these functions _may_ mutate the request/response should be noted."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "Wasn't there a subsequent reason why we had to keep the `/p/project_graph` route? Outside references or is the proposed solution to the original proposal we us `/p/g` as the `compares#project_graph` route?"
+    },
+    {
+      "prediction": "I think you can remove this test now.",
+      "reference": "Any suggestions on how I would even _approach_ writing tests for arbitrary values...."
+    },
+    {
+      "prediction": "I think we should rename these to `ACTIVE_KEY` and `EDIT_KEY` to be consistent with the rest of the codebase.",
+      "reference": "Scope declarations for class constants was not added until 7.1, this will not be accepted."
+    },
+    {
+      "prediction": "`activeClientMap_.empty()` is faster than `activeClientMap_.size() == 0`",
+      "reference": "The assertion here is that find should never return end() because this is the only mechanism that reaps items from the activeClientMap. If it == end something went horribly wrong."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `models`.",
+      "reference": "The `unless the collection ...` part reads a little bit hard, maybe separate into its own sentence and explain what it means by `cannot generate a fresh key`?"
+    },
+    {
+      "prediction": "This should be `ErrorCode = 7` instead of `FailedPrecondition ErrorCode = 7`",
+      "reference": "Doesn't `gcerr_string.go` need to be updated for this?"
+    },
+    {
+      "prediction": "I think this should be `types.RuneSlice` instead of `types.Rune`.",
+      "reference": "Both cases allowed by the Go spec are now supported, so it's a bug in the compiler if we get here. You can replace the `todo:` error with a panic. (Note: getting here would be a bug because when we get to SSA level the code has long been type checked and has already been verified as being valid Go)."
+    },
+    {
+      "prediction": "I think this should be `tab.history_prepared = []`, and the `#end if` should be `tab.history_prepared = []` as well.",
+      "reference": "You can simply do `if tab.history_prepared:` here as empty lists are falsey."
+    },
+    {
+      "prediction": "nit: can we move this to the top of the file?",
+      "reference": "Remove this whitespace"
+    },
+    {
+      "prediction": "`one_img = mmcv.imresize(one_img, input_shape[2:][::-1]).transpose(2, 0, 1)`, `one_img = one_img.transpose(2, 0, 1)`, `one_img = mmcv.imresize(one_img, input_shape[2:][::-1]).transpose(2, 0, 1)`, `one_img = mmcv.imresize(one_img, input_shape[2:][::-1]).transpose(2, 0, 1)`, `one_img = mmcv.imresize(one_img, input_shape[2:][::-1]).transpose(2, 0, 1)`, `one_img = mmcv.imresize(one_img, input_shape[2:][::-1]).transpose(2, 0, 1)`, `one_img = mmcv.imresize(one_img, input_shape[2:][::-1]).transpose(2, 0, 1)`, `one_img = mmcv.imresize(one_img, input_shape[2:][::-1]).transpose(2, 0, 1)`, `one_img = mmcv.imresize(one_img, input_shape[2:][::-1]).transpose(",
+      "reference": "`show_img` is not normalized while `one_img` is normalized. And line 139 pass `show_img` for pytorch2onnx function. Is this expected behavior?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the offset. I think it should be done in the `playbackManager.setSubtitleOffset` function.",
+      "reference": "Why is this bounded between -30 and 30?"
+    },
+    {
+      "prediction": "Is it possible to have both `EXCLUDE_HIP_PLATFORM` and `EXCLUDE_NVCC_PLATFORM`?",
+      "reference": "I think even you can exclude to run it on nvcc"
+    },
+    {
+      "prediction": "Can we use `string(cStorVolume.Spec.Capacity)` instead of `cStorVolume.Spec.Capacity.String()`?",
+      "reference": "G104: Errors unhandled. (from `gosec`)"
+    },
+    {
+      "prediction": "I think this should be `if response['failures'] > 0` instead of `if response['failures']`",
+      "reference": "will `failure` always include `arn` and `reason` in its dictionary? If so, :+1:"
+    },
+    {
+      "prediction": "I'm not sure I like the `action` argument here. It seems like we should be able to just pass in the event type and let the caller decide what to do with it.",
+      "reference": "Should this be optional, or default to 'action' as well?"
+    },
+    {
+      "prediction": "I think we can remove the `ctrl` package now.",
+      "reference": "how about formating this import?"
+    },
+    {
+      "prediction": "I think this should be `braille.handler.handleReviewMove(isAutoTether=not isCaret)`",
+      "reference": "Comma police. :) Also, the docstring needs updating."
+    },
+    {
+      "prediction": "I think this should be `volume.CloudMigrateDriverQuiesce` to be consistent with the rest of the driver.",
+      "reference": "Implement an in-memory implementation of this in the fake driver."
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundBlockBody: Extra empty line detected at block body end.",
+      "reference": "Same question as the other view spec regarding not actually saving records."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `if (extensionDll != null)` block above.",
+      "reference": "Rethrow for `MustBeRethrown()`-exceptions?"
+    },
+    {
+      "prediction": "I think you can remove the `device` arg now, it's not used.",
+      "reference": "May clean the unnecessary modification."
+    },
+    {
+      "prediction": "This should be `flux_sec_incdir=` and not `flux_sec_incdir=`",
+      "reference": "Should this line set the value to `x` since that is checked below?"
+    },
+    {
+      "prediction": "I think we should rename this method to `getSlimCopy()` or `getSlimCopy()` to be consistent with `copy()` and `copyWithoutStats()`.",
+      "reference": "I don't particularly love the terminology here. Why not simply add `copy(boolean stats)` or a copy with an enum to indicate what portions of the datafile to include? At some point we may want just some of the values (e.g. CBO may want counts, but not lower/upper bounds). Just a thought."
+    },
+    {
+      "prediction": "I think you can remove the `if (ann.indexOf('=') == -1) {` check here.",
+      "reference": "you could probably remove keyValue and just compare idx vs ann.length (ex I think the goal here is to ensure it works with a value like `foo=`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could come up with.",
+      "reference": "we do this to differentiate between the multiple cc files for when the recording is stopped / restarted multiple times within a single meeting"
+    },
+    {
+      "prediction": "I think this should be `Item::Lost` instead of `Item::Waiting`",
+      "reference": "You've deleted this code but not added it anywhere else. Should this be used as my proposed example in the .ini file? Do we need a mechanism for configuring a global fallback string independent of the numeric codes?"
+    },
+    {
+      "prediction": "nit: remove unused import `import com.google.common.collect.ImmutableSet;`",
+      "reference": "nit: strip away those imports? seems not used in the new code."
+    },
+    {
+      "prediction": "I think this file should be renamed to `generated.go` or something like that.",
+      "reference": "redundant whitespace :octocat:"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to test this. I think we should just assert that the type is `dict` and that the value is `{'SetStackPolicyResult': 'content'}`.",
+      "reference": "I don't believe that the `assertDictEqual` call can pass without `api_response` being a `dict`, so this second check isn't needed!"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to release the memory. I think it should be done in `h2o_hpack_decode_huffman()`",
+      "reference": "Fuzzer did not like this change. Looks like I misunderstood how the pool works... if allocated from a pool we should never free it manually, right? I think I'll need to drop this patch."
+    },
+    {
+      "prediction": "I think this class should be renamed to `AbstractConfirmDialog` instead of `ConfirmDialog`",
+      "reference": "Wouldn't it make sense to already have title, owner etc in this class?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "New generators should have version `next`"
+    },
+    {
+      "prediction": "Shouldn't this be `metricsConfiguration.getProtocol().getProtocolName()`?",
+      "reference": "Wrap these two lines inside an `if (node.isMetricsEnabled() || metricsConfiguration.isPushEnabled()) { ... }`"
+    },
+    {
+      "prediction": "what is the meaning of `end_level == -1 or end_level == self.num_ins`?",
+      "reference": "if end_level == -1 or end_level == (self.num_ins - 1): for end_level is the Index of the end input backbone level."
+    },
+    {
+      "prediction": "I think this should be `Path.Length + arrayOverhead` instead of `Path.Length / 2 + Path.Length + arrayOverhead`",
+      "reference": "No CLR overhead?"
+    },
+    {
+      "prediction": "I think we can remove the `z` and `p` variables from this test.",
+      "reference": "Consider faking the TestClient's GetProject/GetZone/GetMachine methods."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `transport.go`.",
+      "reference": "Returning a private type as part of the public API? Isn't this impossible for users to implement?"
+    },
+    {
+      "prediction": "I don't think this is correct. `getAtomWithIdx` returns `nullptr` if it couldn't find the atom with the given index, so you can't be sure that `_canonicalRankingNumber` is set if it's not `getPropIfPresent`.",
+      "reference": "prefer common_properties::_canonicalRankingNumber if possible, should be quicker in general."
+    },
+    {
+      "prediction": "what is the meaning of `-1`?",
+      "reference": "The meaning of `loss_normalizer` and `loss_normalizer_momentum` should be reflected in docstring."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it's the only way I could get it to work on Windows.",
+      "reference": "Call to `cloexec_pipe` (or an equivalent function) is missing. I presume that this is the reason why you are seeing timeout errors."
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "There are no changes in this file. Please reset."
+    },
+    {
+      "prediction": "I don't think this is correct. The Q-function is the probability that a normally-distributed random variable is greater than `x`, not the probability that a random variable is greater than `x` and less than or equal to `x`.",
+      "reference": "Should we rename this to tailProbability?"
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to do, but it seems like it should be removed.",
+      "reference": "Add the issue number too i#4400"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `List<E> copy = Lists.newArrayListWithExpectedSize(list.size());` will create a new list with the same capacity as `list`. So if `list` is empty, it will return an empty list, which is not what we want. The fix should be to add a check for `list.size() == 0` before calling `Lists.newArrayListWithExpectedSize`.",
+      "reference": "Why make this modifiable?"
+    },
+    {
+      "prediction": "0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "Why this file is changed?"
+    },
+    {
+      "prediction": "Shouldn't this be `revert` instead of `revert_to`?",
+      "reference": "I'm not sure that we need this"
+    },
+    {
+      "prediction": "I think this should be `dependencyVersion = package.Value?.[\"version\"]?.Value<string>();`",
+      "reference": "Is this going to cause issues in any other places were someone forgets the version? Should we consider also checking for type=project?"
+    },
+    {
+      "prediction": "I think we can remove this function and just call `GetSecretValue` directly in `GetSecretFromAWSSecretsManager`",
+      "reference": "Does this have retries? Is there a possibility that customer would hit throttle errors here?"
+    },
+    {
+      "prediction": "This should probably be `#if defined(PLATFORM_IS_LINUX) || defined(PLATFORM_IS_BSD) || defined(PLATFORM_IS_FREEBSD)`",
+      "reference": "i noted that we are inconsistent with when we indent includes (see atomics.h which is rather different) and for example cpu.c which is also different. we seem to have 3 styles."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `afterBuild` callback of the `gulp serve` task. The `afterBuild` callback is called after the build is complete, but before it is sent to the browser. You can use `fs.existsSync` to check if the file exists, and then call `fs.unlinkSync` to remove it.",
+      "reference": "Deleting the files before starting the audit helps avoid stale results if the new file is not created."
+    },
+    {
+      "prediction": "I think we can remove the `var newSpcLease` and `newSpcLease = &spcLease{spcGot, SpcLeaseKey, c.clientset}` line as well.",
+      "reference": "Remove this blank line"
+    },
+    {
+      "prediction": "I think this should be `import com.google.api.codegen.util.SymbolTable;` instead of `import com.google.api.tools.framework.model.TypeName;`",
+      "reference": "This is... quite unconventional (use a map entry as a key). I would suggest making your own data class for the key."
+    },
+    {
+      "prediction": "I'm not sure this is the right approach. I think we should use the same color scheme as we do for the other outputs. For example, if the output is a balance, then we should use the same color scheme as we do for balances. If the output is a fee, then we should use the same color scheme as we do for fees. If the output is a transfer, then we should use the same color scheme as we do for transfers.",
+      "reference": "why are these colors needed to be specified here? when are they used?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `MetadataTableImpl` since it's only relevant to metadata tables.",
+      "reference": "Nit: no need to change this method."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Why a blank line?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `feedback_email` section.",
+      "reference": "Why remove the `.html_safe` here? Should we use `sanitize` or `raw` instead? This info comes off of the org edit page and is entered by users."
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "I think we need some docs about how to set this parameter correctly. Are we OK with changing the public API like this? We probably at least want to make a change notice to the release doc."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure I like that we have two different names for the same thing. Is there a reason we can't just call it `OverlayControlMsg`?",
+      "reference": "When we undo this for 1.3, do we not need to leave these three entries intact to avoid renumbering of the subsequent constants? If so perhaps the changes to this file should be pulled into a separate initial commit so we can just `git revert` the remainder..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `AddShapeValidations` since it's related to the validation of the shape methods.",
+      "reference": "This is never set anywhere in the `cli/gen-api` folder. Should it be?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a good suggestion either.",
+      "reference": "All the changes in this commit are mechanical. Produced by the script in the previous commit."
+    },
+    {
+      "prediction": "I think this should be `raw` instead of `cmd` since it's a single-line command.",
+      "reference": "What's the difference between these 2 versions?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be moved to `tests/functional/kmsscan/test_kmscan.py`",
+      "reference": "@red2k18 Are we sure its correct to only have 1 now?"
+    },
+    {
+      "prediction": "This should be `newB->setProp(RDKit::common_properties::_unspecifiedOrder, 1);`",
+      "reference": "@greglandrum interesting, is this related to #3307 by any chance?"
+    },
+    {
+      "prediction": "Style/SymbolArray: Use %i or %I for an array of symbols.",
+      "reference": "Minor: how about moving the trailing `if` to an `elsif` above?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `onUpgrade` method of the migration class.",
+      "reference": "That's only executed when users switch from 2.2 to 2.3. Please create a new block with code `2050000` for the next release :) Please also adapt the version number in PodDbAdapter"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `conn` struct.",
+      "reference": "Safe to reference c.srv without capturing it first under a client lock?"
+    },
+    {
+      "prediction": "nit: I think we should move this to the top of the file.",
+      "reference": "@vincepri blame tells me this was you, any objections to the change?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `server`.",
+      "reference": "@abhinav here is where i need to be able to block/wait until the server is started"
+    },
+    {
+      "prediction": "I don't think this is correct. If there is no such parameter, it should return `null`, not `\"\"`.",
+      "reference": "It seems to me as an unrelated change. Could you please revert it?"
+    },
+    {
+      "prediction": "I think this should be `interface{}`",
+      "reference": "~Why are putting all these APIs in their own ~packages~ files? So many more ~directories~ files, to what end? Why not just have them all be a part of the same API file and all live alongside each other so you can easily see them?~ Edit: nevermind, probably works best in separate files."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be moved to the `Node` interface.",
+      "reference": "This likely needs to be double checked -- I recall being convinced at the time of writing this, but now I am less sure"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "For a followup - it is better to use acceptlist/denylist. I realize this is a historical artifact of our codebase."
+    },
+    {
+      "prediction": "I think you can remove the unused imports.",
+      "reference": "Looks like glassfish went a bit nuts doing a refactor"
+    },
+    {
+      "prediction": "I think this test should be `it('should return false when element is not inline', function () {`",
+      "reference": "Couldn't you use position:absolute or float to move inline elements outside their parent? Through clipping an child element can also be outside it's parent. There are probably some other ways to do it too. So I'm not sure the assumption you're making here is right."
+    },
+    {
+      "prediction": "I think it would be better to use `Long.MIN_VALUE` and `Long.MAX_VALUE` instead of `Integer.MAX_VALUE` and `Integer.MIN_VALUE`",
+      "reference": "This could be copied into iceberg code to avoid the spark internal dep? We could use a version check to adjust the method used if needed."
+    },
+    {
+      "prediction": "I think this should be `flux_subprocess_pid (p, \"flux_t\")`",
+      "reference": "This one had me stumped!"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it's the only way I could get it to work.",
+      "reference": "Note that this name differs from the name used as default (and thus used now that you removed this). That should be fine, but wanted to flag it."
+    },
+    {
+      "prediction": "I think it would be better to move this method into the `getQueryBuilder` method of the `EntityRepository`",
+      "reference": "Why public and not protected?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for `current_tab == 'organisation-templates'`",
+      "reference": "We definitely need to address this duplication after MVP"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `KeywordInput` component.",
+      "reference": "This will always generate a label for the first keyword, which means this will be a label linked to no element after the first keyword is entered. This label should be generated inside the `values.map` next to the `<TextField><Input /></TextField>` it is for, because the ID changes based on the index. Also, we have a `VisuallyHidden` component I'd prefer we use over the CSS class. Building as many things out of components is more React-y. Admittedly it's sort of a preference, but it's why we have the `VisuallyHidden` component. So let's use that instead"
+    },
+    {
+      "prediction": "This should be `desktop->xcursor_theme = roots_xcursor_theme_create(NULL);`",
+      "reference": "Does it make sense for this to be fatal? Could there be an embedded application that does not have xcursor themes at all? For instance, compositors without a pointer (like a touch-screen kiosk) will never show a cursor. I think I'm ok with this assumption since it simplifies the code and rootston is assumed to have desktop features, but we should always make sure xcursor is not required to be used in the library."
+    },
+    {
+      "prediction": "I'm not sure if this is the best abstraction for this interface. I think it would be better to have a `TopicsFilter<T>` interface, where `T` is the type of the object being filtered. This would allow us to have a `TopicsFilter<T>.Accepts(T)` method, and a `TopicsFilter<T>.Accepts(ref T)` method, and a `TopicsFilter<T>.Matches(Bloom)` method, and a `TopicsFilter<T>.Matches(ref Bloom)` method, and a `TopicsFilter<T>.Contains(T)` method, and a `TopicsFilter<T>.Contains(ref T)` method, and a `TopicsFilter<T>.Any(T)` method, and a `TopicsFilter<T>.Any(ref T)` method, and a `TopicsFilter<T>.All(T)` method, and a `TopicsFilter<T>.All(ref T)` method, and a `TopicsFilter<T>.Except(T)` method, and a `TopicsFilter<T>.Except(ref T)` method, and a `TopicsFilter<T>.ExceptAll(T)` method, and a `TopicsFilter<T>.ExceptAll(ref T)` method, and a `TopicsFilter<T>.ExceptAny(T)` method",
+      "reference": "would rename to TopicsFilter and original one to SequenceTopicsFilter"
+    },
+    {
+      "prediction": "I think this should be `if len(self) == 1 and (key == 0 or key in self.keys) and not self.drawn`",
+      "reference": "Is the idea that ``self.keys[0]`` here normally matches ``init_key``?"
+    },
+    {
+      "prediction": "I think it would be better to check if `hwnd` is not the same as `winUser.getDesktopWindow()`.",
+      "reference": "Can this be clarified to \"Exit early when hwnd is the windows desktop handle, UiaHasServerSideProvider would return false anyway.\" ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Container` struct, under the `desiredStatus` field.",
+      "reference": "I'm okay with leaving VCPULimit and MemoryLimit on our internal model"
+    },
+    {
+      "prediction": "I think this should be `maxSupportedTier` to be consistent with the rest of the code.",
+      "reference": "5 or 10?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should be able to do this outside of the scope, but I'm not sure how to do it.",
+      "reference": "Should we explain more here?"
+    },
+    {
+      "prediction": "I think we should rename this to `CreateOrGetInstance`",
+      "reference": "I've been struggling with this on my cloud-init integration work as well. In addition to info that is currently stored in the cluster providerstatus, we also need some of the info that is available within the base cluster object as well. I think it would make sense to unify the machine and cluster info needed into a consolidated struct rather than continuing to add additional disparate structs."
+    },
+    {
+      "prediction": "I don't think we should deprecate an internal API.",
+      "reference": "Why not `@InternalApi`?"
+    },
+    {
+      "prediction": "I don't think we need this import. We can just use `typing._GenericAlias` directly.",
+      "reference": "FYI: `GenericMeta` is renamed to `_GenericAlias` in Python3.7."
+    },
+    {
+      "prediction": "I think this should be `template.latest?` instead of `template.sections.all?(&:modifiable?)`",
+      "reference": "So this hides the link to re-order sections on Historic Templates? Good catch, Just checked on DMPonline and hitting that throws a 404"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to reset the reporter, but I don't have a better suggestion.",
+      "reference": "Is `Reporter#reset` no longer used? If so, can we remove it?"
+    },
+    {
+      "prediction": "I think this should be `out.writeLongScalar(header.getBaseFee().get() * 1000L);`",
+      "reference": "same concern here about difference in the write scalar implementation. I am not sure if this could present a consensus problem or not"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should go in `MvvmCross.UI.Views`.",
+      "reference": "Can we remove this file?"
+    },
+    {
+      "prediction": "I think it would be better to log the reason of the error instead of the message.",
+      "reference": "Why listing only first error reason (`ErrStatus.Details.Causes[0].Message`)?"
+    },
+    {
+      "prediction": "I think this should be renamed to `USE_NEW_SNAPSHOT_SEQUENCENUMBER_FOR_COMPACTION` or something like that.",
+      "reference": "Now that this is true, do we have to ignore it with V1 Tables?"
+    },
+    {
+      "prediction": "I think this should be `if (Blockly.Colours.hasOwnProperty(colourProperty))` instead of `if (!Blockly.Colours.hasOwnProperty(colourProperty)) continue;`",
+      "reference": "prefer if (condition) { stuff; } to if !(condition) { continue; } stuff"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Use the new Ruby 1.9 hash syntax.<br>Use the new Ruby 1.9 hash syntax.<br>Use the new Ruby 1.9 hash syntax.<br>Use the new Ruby 1.9 hash syntax.<br>Use the new Ruby 1.9 hash syntax.<br>Use the new Ruby 1.9 hash syntax.",
+      "reference": "Should this section not be converted to a secondary action list?"
+    },
+    {
+      "prediction": "I think this should be `result = message.ask(\"Are you sure you want to close a pinned tab?\", ...)`",
+      "reference": "This askes a blocking question, which means a local Qt eventloop will be running. While this is unavoidable sometimes, the async functions should be used whenever possible (`message.confirm_async` in this case). This means: - Split everything after this question in a separate private method (you can probably just call it `_tab_close`) - If the tab is not pinned, just call `_tab_close` directly - If it is pinned, call `message.confirm_async` with `yes_action=self._tab_close`"
+    },
+    {
+      "prediction": "olicy.EnableConversion), $(\"#chkEnableAudioConversion\", page).checked(user.Policy.EnableAudioConversion), $(\"#chkEnableVideoConversion\", page).checked(user.Policy.EnableVideoConversion), $(\"#chkEnableAudioConversionTranscoding\", page).checked(user.Policy.EnableAudioConversionTranscoding), $(\"#chkEnableVideoConversionTranscoding\", page).checked(user.Policy.EnableVideoConversionTranscoding), $(\"#chkEnableVideoMuxing\", page).checked(user.Policy.EnableVideoMuxing), $(\"#chkEnableAudioMuxing\", page).checked(user.Policy.EnableAudioMuxing), $(\"#chkEnableVideoMuxing\", page).checked(user.Policy.EnableVideoMuxing), $(\"#chkEnableAudioMuxing\", page).checked(user.Policy.EnableAudioMuxing), $(\"#chkEnableVideoMuxing\", page).checked(user.Policy.EnableVideoMuxing), $(\"#chkEnableAudioMuxing\", page).checked(user.Policy.EnableAudioMuxing), $(\"#chkEnableVideoMuxing\", page).checked(user.Policy.EnableVideoMuxing\"), $(\"#chkEnableAudioMuxing\", page).checked(user.Policy.EnableAudioMuxing\"), $(\"#chkEnableVideoMuxing\", page).checked(user.Policy.EnableVideoMuxing\"),",
+      "reference": "Can we deuglify this?"
+    },
+    {
+      "prediction": "`flatMap` is not the right thing to do here. The `partialFunction.isDefinedAt()` check should be done before `apply` is called. If `partialFunction.isDefinedAt()` returns true, then `apply` will be called, and the result of `apply` will be added to the `Option`. If `partialFunction.isDefinedAt()` returns false, then nothing will be added to the `Option`.",
+      "reference": "@skestle Just recognized that the `::apply` is not necessary. Does it compile if you leave it away? Could you please check that, I'm on vacation and have no IDE at hand... Thx!"
+    },
+    {
+      "prediction": "I think this should be `ma` instead of `newSpec`",
+      "reference": "Looks like this is an unnecessary change."
+    },
+    {
+      "prediction": "I think this should be `2` instead of `1`",
+      "reference": "This one won't change as the resource is not included in the project getting tested with the composite root."
+    },
+    {
+      "prediction": "nit: I think it would be better to have this as a `static const u8 NAT_LEVEL_1_DROP = 0;` instead of a `__u8 NAT_LEVEL_1_DROP = 0;`",
+      "reference": "We have `stdbool` imported, might as well use that for clarity."
+    },
+    {
+      "prediction": "I'm not sure if we want to change this, but I think it would be better to use `SqlServer` instead of `sql-server` as the name of the service.",
+      "reference": "Why do we not also have integration tests for postgres?"
+    },
+    {
+      "prediction": "Why is this being skipped?",
+      "reference": "I think its safe to remove this if we're removing `reIndex` outright. The description shows us that the test is \"An example showing how to force a reindex of a collection\""
+    },
+    {
+      "prediction": "I think this should be `Errno = 22` instead of `EINVAL`",
+      "reference": "Can you make this hexadecimal to match the other constants?"
+    },
+    {
+      "prediction": "I think this should be `Bolt::Logger.warn(\"project_yaml\", msg)`",
+      "reference": "Can this get moved to `Bolt::Project#validate`?"
+    },
+    {
+      "prediction": "I think it would be better to just do `Collections.reverse` on the right hand side of the `Tuple.of` instead of creating two new lists.",
+      "reference": "I'm pretty sure it doesn't harm, but don't we have a second pass here with the reverse() operation?"
+    },
+    {
+      "prediction": "I think this should be `TAKES_HEAP_SNAPSHOT = \"takeHeapSnapshot\";`",
+      "reference": "This is not a standard capability and should be hidden behind a vendor prefix: -chromium-takesHeapSnapshot"
+    },
+    {
+      "prediction": "I'm not sure this is the right solution. The problem is that the `sgd_execution_context` needs to know about the `sgd_training_algorithm`, so it needs to be passed as an argument to the constructor of the `sgd_execution_context`. This means that the `sgd_training_algorithm` needs to be passed as an argument to the `sgd_execution_context` constructor, and then the `sgd_training_algorithm` needs to be passed as an argument to the `sgd_execution_context` constructor.",
+      "reference": "This shouldn't work with pointers -- prefer references unless you can meaningfully pass `nullptr`. The address-of operator here is clunky at best."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Sidenote: some day we'll drop the avr-gcc and avr-libc dependencies so that this case becomes supported."
+    },
+    {
+      "prediction": "I think this is the right thing to do, but I don't know if it's the right thing to do in general.",
+      "reference": "do we still need this method?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `if` above.",
+      "reference": "It's better to not leave this old code here... it just clutters things up and makes reading the code more confusing. We can always recover the old code from the repository if needed."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Again, you don't want to depend on glassfish."
+    },
+    {
+      "prediction": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "This shouldn't be required. What was going on here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `ConnectionPolicies` struct, since that's where the connection policies are defined.",
+      "reference": "Why add this here?"
+    },
+    {
+      "prediction": "Is it possible to have `IsXNullableReturnType` set to `true` here?",
+      "reference": "nullability does not apply to headers type (that is always an object)"
+    },
+    {
+      "prediction": "I think this should be `uint64_t` instead of `uint32_t`",
+      "reference": "Do we also need to add the iova, or is that done somewhere else?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should go in `identity` package.",
+      "reference": "why change this?"
+    },
+    {
+      "prediction": "I think this should be `(time_t *)&ctx->_timestamp_cache.tv_at` instead of `(time_t *)&ctx->_timestamp_cache.tv_at.tv_sec`",
+      "reference": "If there is no guarantee that tv_sec is of type `time_t`, IMO you should copy the value instead of enforcing a pointer typecast."
+    },
+    {
+      "prediction": "I think this should be `opts.DockerfilePath != \"\"` instead of `opts.DockerfilePath != \"\" && !strings.HasSuffix(opts.DockerfilePath, \"/Dockerfile\")`",
+      "reference": "Should we validate the suffix before going to the filesystem?"
+    },
+    {
+      "prediction": "I think this should be `ddev_default_site_name` and `ddev_default_service_type` instead of `siteName` and `serviceType`",
+      "reference": "It feels like we should be working to reduce/eliminate our global vars. I'm not sure it adds much here. Any objection to just removing it?"
+    },
+    {
+      "prediction": "I think this should be `t.integer :user_id, null: false, foreign_key: { to_table: 'users' }`",
+      "reference": "did you consider using AR enums for this? would give us all the boolean methods (`weekly?` etc) for free"
+    },
+    {
+      "prediction": "I think this should be `if not filename or not os.path.exists(filename): return None`",
+      "reference": "This may change \"\" into None. Better return filename."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think we should move it to `kubelet` since it's only used by kubelet.",
+      "reference": ":+1: That does resolve my security concerns about this feature."
+    },
+    {
+      "prediction": "I think this should be `get_if_raw_hwaddr(iff)`, not `get_if(iff, SIOCGIFHWADDR)`.",
+      "reference": "Could you add a docstring?"
+    },
+    {
+      "prediction": "nit: I think we should be able to use `aws.StringValue(resp.Region)` instead of `resp.Region`",
+      "reference": "Shouldn't this still be `Region()`?"
+    },
+    {
+      "prediction": "I don't think this should be a member of the `CompletionView` class. It should be a member of the `QCompleterView` class.",
+      "reference": "This shouldn't be here but `self.connected = None` in `__init__` instead, otherwise that'd be an attribute which is set this way in _every_ instance of that class (i.e. a class rather than an instance variable). I'd also say let's make it \"private\" (i.e. `_connected`) and please add a quick note about what it is to the docstring of the class."
+    },
+    {
+      "prediction": "I think this should be `start_error_message=None, create_no_window=False`",
+      "reference": "This would be better served as a method/property that is set when people don't want to a window."
+    },
+    {
+      "prediction": "I think we can remove the `context.getInterface().getMethods()` call as well.",
+      "reference": "I was dubious about putting getNonStreamingMethods in the context, but after some thought, I have decided to embrace the approach. In this particular case, the loop is supposed to be over page streaming methods. So, could you add a getPageStreamingMethods() method in the context, use it here, and remove the isPageStreaming() check below?"
+    },
+    {
+      "prediction": "I think it would be better to use `int(version.split(\".\")[0])` instead of `str(self.settings.get('version', self.JMETER_VER)).split(\".\")[0]`",
+      "reference": "LooseVersion class can help here"
+    },
+    {
+      "prediction": "I'm not sure this is correct. I think it should be something like `zeros.at(i) = 0;` and then `foo.m_AvailableStepsCount = foo.m_AvailableStepsCount + 1;`",
+      "reference": "`const Dims zeros(shape.size(), 0);`"
+    },
+    {
+      "prediction": "I think this should be `--profile` and not `--profile [COUNT]`",
+      "reference": "Would be good for this not to say `10` anymore..."
+    },
+    {
+      "prediction": "I think this should be `adsense` instead of `adSense`",
+      "reference": "See above, we don't need to pass the module name here."
+    },
+    {
+      "prediction": "I think this should be `BadMethodCallException` instead of `Overblog\\GraphQLBundle\\Exception\\BadMethodCallException`",
+      "reference": "This is random line :-) overridden String type by custom trimmed - please change that commit message to something like \"introduced custom String type with automatic trimming\""
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. The `setFeatureToken` should be in the `Sync` module, not in the `realmConstructor.Sync` object.",
+      "reference": "My personal taste: \"depreciated\" -> \"deprecated\""
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "this includes the port as well right? should the name be `destinationServicePortName`?"
+    },
+    {
+      "prediction": "I don't think it's necessary to import `configuration` here, since it's not used in this file.",
+      "reference": "We can import all of them in one line."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `AppendElements` class.",
+      "reference": "this was wrong, the rest were just inconsistent :)"
+    },
+    {
+      "prediction": "I think this should be `proposals-completed` instead of `proposals-...`",
+      "reference": "I gave them the EXACT SAME NAMES in my branch :)"
+    },
+    {
+      "prediction": "I think this should be `0755` rather than `0756`",
+      "reference": "What would you think stat'ing the directory first before doing the log+mkdirall and only proceeding if the directory does not exist? The logging might be less confusing (I'd be wondering why it was logging that it was creating the directory when I knew it already existed)."
+    },
+    {
+      "prediction": "I think we can remove the `InternalContext` from the `TraceContext` array as well.",
+      "reference": "Sorry about the confusion, I'm trying to address that in the PR follow-up. If we ever have `Span` objects, then accessing their properties is going to be safe. The only question remaining is \"What is the runtime type for Scope.Span?\" and we just have to account for it when it is `Datadog.Trace.Span` and when it is not"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "I was using this imports for writing the results to file to more easily check that it was valid JSON. You can probably remove these `java.io` imports."
+    },
+    {
+      "prediction": "I think this should be `logentry` instead of `logEntry`",
+      "reference": "You should probably include a `json:...` annotation here."
+    },
+    {
+      "prediction": "nit: I think this should be `ecsMaxRuntimeIDLength` and not `ecsMaxReasonLength`",
+      "reference": "can container ID > 255 chars? why are we doing this check?"
+    },
+    {
+      "prediction": "I think it would be better to use `prohibits_from` instead of `prohibits`.",
+      "reference": "*\"This **field field** prohibits :other from being present.\"*, **field** word repeated is right?"
+    },
+    {
+      "prediction": "I think we should rename this to `npl` as it's not related to `nfl`.",
+      "reference": "nit: no need to add an alias for this case."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `quic` package.",
+      "reference": "Or `QUIC_LOG_LEVEL`. Which one do you prefer?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to parse the allocation units from the disk section itself, rather than passing it in as a string.",
+      "reference": "Minor: you can set it to byte here and only set it to *virtualDiscDesc.CapacityAllocationUnits if that's not nil/\"\". Saves two lines"
+    },
+    {
+      "prediction": "I think this is a good test, but I think it would be better to have a test that checks that the fingerprint is the same as the one it was given when it was created, rather than checking that it's there at all.",
+      "reference": "While reading this I found that it makes actually little sense for \"beers\" to have a \"fingerprint\". We might want to do another pass on the examples here to use something that actually makes more sense to the reader. This could be done in another issue."
+    },
+    {
+      "prediction": "I think this test is failing because `mkString(\",\")` is not being called. I think it should be `mkString(\", \")`.",
+      "reference": "Thank you! That's better, especially the conversion test should not test the mkString method :)"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The idea here is that we don't know where we are in the window yet, so we can't tell if the location is relative to the top or the left of the window. I think we need to do something like this: 1. Get the scroll position of the window. 2. Get the scroll position of the scroll bar. 3. Add the scroll position of the window to the scroll position of the scroll bar. 4. Add the scroll position of the scroll bar to the scroll position of the window. 5. Add the scroll position of the window to the scroll position of the scroll bar. 6. Add the scroll position of the scroll bar to the scroll position of the window. 7. Add the scroll position of the window to the scroll position of the scroll bar. 8. Add the scroll position of the window to the scroll position of the scroll bar. 9. Add the scroll position of the window to the scroll position of the scroll bar. 10. Add the scroll position of the window to the scroll position of the scroll bar. 11. Add the scroll position of the window to the scroll position of the scroll bar.",
+      "reference": "You're basically converting client to screen coordinates here, doing it manually. Is there a specific reason why you aren't using clientToScreen here? Does it fail?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to check for the max transfer limit. I think it should be in the `if doMultiThreadCopy(f, src)` block, but I'm not sure if that's the right place to check for the max transfer limit.",
+      "reference": "This needs to be done for server-side copies too, earlier in the function."
+    },
+    {
+      "prediction": "I think we can remove offline_access now, since we don't need to support it.",
+      "reference": "Do we want to add in `profile` here as well so we request access to the end user's default profile claims like name?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `Proposal::ApprovalsController#update` action, since that's where the individual approval objects are being updated.",
+      "reference": "I am not sure what the difference between approving and full approving is..."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "Please back out the import changes in files you haven't touched. I think these happen if you run goimports without having the vendor directory populated"
+    },
+    {
+      "prediction": "I don't think this is the right place for this test.",
+      "reference": "Many tests in this class seem to have just been fortunate that `SolrIndexSearcher` ignored `fl` and retrieved all fields when not using lazy loading."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this description. I think it should be in the `scapy.contrib` module.",
+      "reference": "Is this really needed?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's the only way I've been able to get it to work. The problem is that `RSpec.world` is an instance of `RSpec::Core::RSpec`, which has a `register` method that takes two arguments: the name of the example group to register and a list of arguments to pass to the `register` method. The `register` method returns a new instance of `RSpec::Core::ExampleGroup`, which has a `__send__` method that takes two arguments: the name of the example group to register and a list of arguments to pass to the `register` method. The `register` method returns a new instance of `RSpec::Core::ExampleGroup`, which has a `__send__` method that takes two arguments: the name of the example group to register and a list of arguments to pass to the `register` method. The `register` method returns a new instance of `RSpec::Core::ExampleGroup`, which has a `__send__` method that takes two arguments: the name of the example group to register and a list of arguments to pass to the `register` method. The `register` method returns a new",
+      "reference": "Is this necessary with line 38?"
+    },
+    {
+      "prediction": "I think this should be `pytest>=3.2`",
+      "reference": "DO we want to have >= here? Or =="
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `CoreTags`.",
+      "reference": "Can you also add this `\"version\"` tag to the CoreTags? Now that we're targeting the service/env/version trio"
+    },
+    {
+      "prediction": "Why do we need this test?",
+      "reference": "This entry will be incompatible with the currently implemented formatting rules. Maximum 120 characters per line."
+    },
+    {
+      "prediction": "I think this should be `if @options[:preserve_host_output]` instead of `if @options[:preserve_host_output] && @options[:keep_host_output]`",
+      "reference": "Let's call this color_host_output or some such - preserve_host_output makes it sound like you won't get any output without this being set."
+    },
+    {
+      "prediction": "I think this should be `axe.utils.getNodeAttributes(node, 'aria')` instead of `axe.commons.aria.implicitRole(node)`",
+      "reference": "suggestion: if we perhaps make `attributes` a getter in `virtualNode`, it does look seamless to access the property, same as what we have done for `isFocusable` or `tabbableElements`. `node.attributes` can become `virtualNode.attributes`"
+    },
+    {
+      "prediction": "Shouldn't this be `final`?",
+      "reference": "I think a float value fits better."
+    },
+    {
+      "prediction": "I don't think we need this.",
+      "reference": "This should be (POST,DELETE) `/pss/subscribe/{topic}` for subscriptions and there should be `/pss/send/{topic}` for sending."
+    },
+    {
+      "prediction": "I think this should be `this.setState`, not `this.shouldComponentUpdate`.",
+      "reference": "Closing over the closure arguments is a neat trick :+1: Love it :100:"
+    },
+    {
+      "prediction": "I think it would be better to use `assertThrows` instead of `assertThrowsAndStops`.",
+      "reference": "Does this need to be a block or can it be an expression?"
+    },
+    {
+      "prediction": "I think this should be `if (req.getQueryString() != null && allowedPostRequest(req)) {`",
+      "reference": "req.getQueryString() != null is not necessary since same check is already done in allowedPostRequest"
+    },
+    {
+      "prediction": "I think this should be `public abstract boolean isRestOnlyTransport()`",
+      "reference": "As per go/actools-regapic-design, in the final product, generated GAPICs must be able to support multiple transports if supported by the API. For Java, we'll support this in the microgenerator; the monolith generates single-transport GAPICs. For PHP, the situation is likely similar, though on a longer timescale. All this to say that I would appreciate structuring the code in terms of \"supports REST\" rather than \"REST only\", but only where this doesn't require undue work. For everything else, please add a TODO to the effect of \"we're generating single-transport GAPICs for now, though in the fullness of time we'll generate multi-transport GAPICs.\""
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "here is `:seconds` missing again"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "Changed this helper to also get class name."
+    },
+    {
+      "prediction": "I think this should be called `applicationName` and not `applicationName`",
+      "reference": "I think there's another spot for this in DefaultCapabilityMatcher"
+    },
+    {
+      "prediction": "This should be `#define PONY_ACTOR_DEFAULT_BATCH 10`",
+      "reference": "Why the rename from PONY_SCHED_BATCH ? ACTOR_DEFAULT_BATCH is less meaningful to me than SCHED_BATCH."
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "Let's be sure we have an issue that tracks this, filed against testnet."
+    },
+    {
+      "prediction": "I don't think this is the right way to check for buffers. I think it should be `if (typeof payload == 'string') {`",
+      "reference": "There should be no space before !Buffer"
+    },
+    {
+      "prediction": "I think this should be `padding-bottom: 16px;` instead of `padding: 0`.",
+      "reference": "Same here, we need the default padding for this story."
+    },
+    {
+      "prediction": "I think this should be `time.perf_counter()`",
+      "reference": "I think this should go away, its noise in the grand scheme of things."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in `UpdateRepository`.",
+      "reference": "Maybe just call `ioutil.WriteFile` with 0 bytes? And Make an error log `log.Error` if any error occurs."
+    },
+    {
+      "prediction": "nit: I think it would be better to have this method return a new instance of the `ServiceBuilder` and then call `this.ieService_ = service;` directly in the `setIeService` method. This way you don't have to pass in a new instance of the `ServiceBuilder` in the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `setIeService` method, and you don't have to pass in a new instance of the `ServiceBuilder` to the `set",
+      "reference": "`this.ieService_` should be initialized to null in the constructor."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to apply the policy. I think it would be better to apply the policy on the inbound traffic to the workloads.",
+      "reference": "I just noticed that we use Return here, when we have logically allowed a packet, whereas in the forward chain we use AcceptAction. Do you know why that is?"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it should be in the `BeanPostProcessor` interface.",
+      "reference": "This is public method , we need to deprecated this method first and add new updated method for it."
+    },
+    {
+      "prediction": "Why do we need to do this? If the node is not a relay, then we don't need to specify a netaddress. If it is a relay, then we don't need to specify a netaddress.",
+      "reference": "I think that you don't want to have these workarounds; you want to make sure that the relayTemplates is configured correctly. If not, we need to fix it there."
+    },
+    {
+      "prediction": "I don't think we need to pass the block number here, since the block number is already passed in the constructor of the `ProtocolSchedule` class.",
+      "reference": "This field is not needed."
+    },
+    {
+      "prediction": "I'm not sure I like the idea of having a `reason` parameter in the cancellation controller. I'd rather have a `reason` field on the `Subscription` model.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "move to line 18 below"
+    },
+    {
+      "prediction": "I think we should remove the purpose of this snippet.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I think this test is a bit too specific to this PR. Can we move it to a new PR?",
+      "reference": "Thats odd, I'd actually not expect this output at all..."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do this, but it's the only way I could get Travis to pass locally. I'm not sure if it's the right way to do this in general, but it's the only way I could get Travis to pass locally. I'm not sure if it's the right way to do this in general, but it's the only way I could get Travis to pass locally. I'm not sure if it's the right way to do this in general, but it's the only way I could get Travis to pass locally. I'm not sure if it's the right way to do this in general, but it's the only way I could get Travis to pass locally. I'm not sure if it's the right way to do this in general, but it's the only way I could get Travis to pass locally. I'm not sure if it's the right way to do this in general, but it's the only way I could get Travis to pass locally. I'm not sure if it's the right way to do this in general, but it's the only way I could get Travis to pass locally. I'm not sure if it's the right way to do this in general, but it",
+      "reference": "As a side note, I get why this is up here (sort order), but it'd be easier to not forget to update it if it were down next to `1.x` -- would it be acceptable to make that change the next time I make this sort of PR? :smile: :innocent: (don't want to hold this one up since folks are blocked on getting this one in, it's a pretty superficial change, and I probably won't have cycles again for a few days to update the PR with the change :sweat_smile:)"
+    },
+    {
+      "prediction": "I think this should be `SmartContractPermissioningConfiguration.createDefault()`",
+      "reference": "this class is permissioning so let's keep \"Onchain\" here"
+    },
+    {
+      "prediction": "Shouldn't this be [assembly: AssemblyTitle(\"AutoRest\", \"0.0.0.0\")]?",
+      "reference": "we can be more specific to call out the it is \"AutoRest C# code generator\"?"
+    },
+    {
+      "prediction": "I think this should be `axe.VirtualNode = VirtualNode;` instead of `axe.VirtualNode = new VirtualNode;`",
+      "reference": "Ditto for not adding this to the axe namespace."
+    },
+    {
+      "prediction": "This file should be renamed to `registration.go`",
+      "reference": "Should we just remove this file? Having it present but empty feels misleading"
+    },
+    {
+      "prediction": "I think this should be `logPeerTracker.Infof(\"Trusting peer=%s\", pid.Pretty())`",
+      "reference": "Would it make sense to include an `Untrust` as well? If I were playing around with this on the CLI I wouldn't want to make a change I couldn't undo."
+    },
+    {
+      "prediction": "I think this should be `getSolrRecord` instead of `getSolrRecordCallback`",
+      "reference": "It might be better to call this `$createRecordMethod` since it's not a full PHP callback, just a method name for the plugin manager."
+    },
+    {
+      "prediction": "Why not using `Tlog::info` instead of `Tlog::notice`?",
+      "reference": "remove this line"
+    },
+    {
+      "prediction": "I think it would be better to use `array` instead of `MainMenuItem[]`",
+      "reference": "I've never seen this syntax before, it's still just an array, the difference is that it's an array of MainMenuItems instead of an array of arrays."
+    },
+    {
+      "prediction": "I think this should be `API-MS-Win-Core-Kernel32-private-L1-1` and not `API-MS-Win-Core-Appinit-L1-1`.",
+      "reference": "nit: to match the others: s/private/Private/"
+    },
+    {
+      "prediction": "I don't think we need to make this public.",
+      "reference": "Maybe just IEnumerable? Not sure myself."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "do we need this change?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to move this logic to `registerAccountNoLock` so that we don't have to duplicate the logic here.",
+      "reference": "Would want @matthiashanel to have a look since if I recall he had to add the shallowCopy() to fix some bugs during reload. That being said, since I believe the $G account cannot referenced in configurations, this should not be a problem, but Matthias has looked at this in more details in the past."
+    },
+    {
+      "prediction": "I think we can remove this function as it is not used anywhere.",
+      "reference": "The descriptions are incorrect"
+    },
+    {
+      "prediction": "I think this should be `if (debuggerAddress.match(/\\/se\\/cdp\\/ws\\//))` instead of `if (debuggerAddress.match(/\\/se\\/cdp\\/ws\\//))`",
+      "reference": "if we are returning the `ws` here when passing in `se:cdp` we can just return it straight or do we have to make a request to get the `ws` address?"
+    },
+    {
+      "prediction": "I think this should be `replace=True, color=False`",
+      "reference": "Please only indent this by four spaces."
+    },
+    {
+      "prediction": "I think this is a good idea, but I don't think we need to change this line.",
+      "reference": "What's the reason behind these characters?"
+    },
+    {
+      "prediction": "I don't think we need to wait for the next line to be executed for this to be true.",
+      "reference": "Minor: The name of this specs seems a little off now"
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to say, but I think it should be `Apply methodrahse not yet supported`.",
+      "reference": "I'd suggest using `ast_error_continue` here to get the `apply` message in the `Info` part of the error report."
+    },
+    {
+      "prediction": "I think it would be better to have a `SearchTabDetailHelper` instead of a `SearchTabsHelper`.",
+      "reference": "If we decide to change the name of the .ini setting, we might also want to review the naming of this property and related code. Maybe a simple `getOptions` or `getSettings` would be more concise than `getTabDetailConfig`. At very least, I don't think we need Tab in the method name since the method belongs to a helper with \"Tabs\" in the name."
+    },
+    {
+      "prediction": "I think we should add a check here to see if the user is an admin of the group. If so, we should return `true`. Otherwise, we should return `false`.",
+      "reference": "Use const for \"user\", same as below"
+    },
+    {
+      "prediction": "I think this should be set to `1` instead of `0` so that we don't have to keep track of this in the DB.",
+      "reference": "->setUnsubscribed(false) would be better :)"
+    },
+    {
+      "prediction": "Unused method argument - `opts`. If it's necessary, use `_` or `_opts` as an argument name to indicate that it won't be used.<br>Unused method argument - `type`. If it's necessary, use `_` or `_type` as an argument name to indicate that it won't be used.",
+      "reference": "I now think singular `Sport` is better, but I will wait for the first round of feedback before updating."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move it to the `models.py` file, under the `created_at` and `updated_at` fields.",
+      "reference": "Don't you need = something ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "Looks like this could be `const` (I realize it wasn't like that before)?"
+    },
+    {
+      "prediction": "I think this should be `FAILED_TO_LOAD`, not `failedToLoad`",
+      "reference": "Maybe just `FAILED`. We can't be sure loading was the thing that failed."
+    },
+    {
+      "prediction": "I think we can remove the `use strict;` from the top of the file.",
+      "reference": "You have replaced single quotes with double ones, this fails linting and our current coding style. Please fix this, otherwise it's a whopping of 1200 LoC while in reality it should be rather small."
+    },
+    {
+      "prediction": "Should this be prefQueueAddToFront, not prefQueueAddToFront?",
+      "reference": "@danieloeh @TomHennen. Some of these files are using spaces, the others use tabs. Is there a project wide preference? I personally like spaces, because they are consistent across editors"
+    },
+    {
+      "prediction": "I don't think we need to pass in the counter here, since we're only using it in the `initialize` method.",
+      "reference": "I guess this is ok for backwards-compatibility? Maybe it'd be better to check arity in the helpers? Or just call it out in the release notes, because there are at least a couple projects on github that overrode `initialize`.."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `TriggerManager` class.",
+      "reference": "At some point, we would like to get rid of these custom injections as well. Looks Ok for now."
+    },
+    {
+      "prediction": "I think this should be `DD_TRACE_HEADER_TAG_NORMALIZATION_ENABLED` to be consistent with other settings.",
+      "reference": "nit: Not a fan of \"fix\" in the name, but don't have a better suggestion"
+    },
+    {
+      "prediction": "I think this should be `fields['protocol'] = self.tab_url[idx].scheme()`",
+      "reference": "Since we're calling `self.tab_url(idx)` a third time here already, can you please add a `url = self.tab_url(idx)` somewhere above (e.g. after all the `fields[...] = ...` assignments) and change the calls here to `url.host()` etc.?"
+    },
+    {
+      "prediction": "I think this should be `constants` instead of `gcloud`",
+      "reference": "alphasort this import"
+    },
+    {
+      "prediction": "Indent the first statement one step more than the start of previous line.",
+      "reference": "remove this file!"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this logic, but it seems like it should be in the same place as the `if len(name) == 0` check.",
+      "reference": "You don't need to say `rune(' ')` - `' '` is already a `rune`."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "We can't freeze this and then add something to it in the next line. The `freeze` on line 43 is sufficient. If Rubocop flags this we need to exclude it."
+    },
+    {
+      "prediction": "I think this should be `public interface Kind<TYPE extends Kind<TYPE, ?, ?>, E, T>`, not `public interface Kind<TYPE extends Kind<TYPE, ?, ?>, E, T>`.",
+      "reference": "We call it Kind2 and move out of the `javaslang/control` package into the `javaslang`package. Maybe I will later generate Kind1..Kindn but that's another story."
+    },
+    {
+      "prediction": "I think this should be `db.logger.Infof(\"db capacity in chunks (and in Bytes): %v (%v)`, since it's not the capacity in chunks but the capacity in bytes.",
+      "reference": "i think it might be nicer to have a message like: `database capacity: %d chunks (%d bytes, %d megabytes)`. counting in bytes is so 90s :)"
+    },
+    {
+      "prediction": "I think it would be better to have a prefix like `azkaban.hadoop.spark.job.link` and `azkaban.spark.hadoop.job.link`",
+      "reference": "The '.' in the key names are used to separate namespaces, NOT to separate words. How about azkaban.external_resources.resource_manager? ---- Why is it better than using the full name in the variable name e.g. RESOURCE_MANAGER_LINK ?"
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "This seems to be duplicated with line 19?"
+    },
+    {
+      "prediction": "I think this should be `#include <Locale.h>`",
+      "reference": "This is wrong. xlocale.h is needed on OSX and maybe elsewhere too. This needs a test in configure to see if xlocale.h and uselocale() are available and an #if test here. Also, the NO_PER_THREAD_LOCALE stuff in configure.ac needs to be updated to include test for uselocale()."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this snippet. I think it should go in the `aws-iam` package, not the `aws-java-iam` package.",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for the supported MIME types. I think it would be better to check if the MIME type is in the `supportedMimeTypes` array in the `ResourceModel` class.",
+      "reference": "Why did you have multiple function calls here instead of having one long array of supported mimeTypes? In line with your PR here, I think we might eventually see inline videos or inline pdf. It would be great to generalize this a bit."
+    },
+    {
+      "prediction": "nit: you can use `from typing import Union` instead of `from typing import NumericType, Union`",
+      "reference": "nit: shall we keep alphabetical order?"
+    },
+    {
+      "prediction": "I think this should be `objectForPrimaryType`",
+      "reference": "We could just move the check for function into `typeForConstruction` to make this a one liner every time we need to do this."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it would be better to move this test to `org.mule.runtime.core.internal` or `org.mule.runtime.core.internal.dsl`",
+      "reference": "A correct by unrelated change. We'll slide this one in without another word ;)"
+    },
+    {
+      "prediction": "Why do you need this?",
+      "reference": "I feel that a slightly safer way is to do if (param_.has_device_id()) { Caffe::SetDevice(param_.device_id()); } just in case a user has hard-coded a device id outside the solver and does not specify the device id in the solver param. Currently, if nothing is set, the solver will always use the 0th device, which might not be desired."
+    },
+    {
+      "prediction": "I don't think this is the right test to run. I think we want to make sure that `rawCode` is working as expected, not that `multi` and `line` and `block` are working as expected. I think we want to make sure that `rawCode` is working as expected, not that `rawCode` is working as expected. I think we want to make sure that `rawCode` is working as expected, not that `rawCode` is working as expected. I think we want to make sure that `rawCode` is working as expected, not that `rawCode` is working as expected. I think we want to make sure that `rawCode` is working as expected, not that `rawCode` is working as expected. I think we want to make sure that `rawCode` is working as expected, not that `rawCode` is working as expected. I think we want to make sure that `rawCode` is working as expected, not that `rawCode` is working as expected. I think we want to make sure that `rawCode` is working as expected, not that `rawCode` is working as expected. I think we want to make sure that `rawCode` is",
+      "reference": "let's replace this example with something that won't fail in Python. Like \"for+if+print\""
+    },
+    {
+      "prediction": "I think this should be `if precision is not None and not isinstance(value, str)`.",
+      "reference": "Please remove the brackets and use `string_type` from `compat.py` module instead of `str` in `isinstance()`."
+    },
+    {
+      "prediction": "Why do we need to pass the `gasPrice` here?",
+      "reference": "line is 143 characters (from `lll`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with.",
+      "reference": "It should follow a `continue`."
+    },
+    {
+      "prediction": "I think this should be `working_dir: \"{{ .WebWorkingDir}}` and not `working_dir: \"{{ .WebWorkingDir }}```",
+      "reference": "Should this be DBWorkingDir?"
+    },
+    {
+      "prediction": "Shouldn't this be `TriggerEventType.valueOf(eventTypeStr.trim().toUpperCase(Locale.ROOT));`?",
+      "reference": "I think you can remove the whole line."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I don't think we should change the enum name here. It's fine to change the enum name in a follow-up PR.",
+      "reference": "Not a big deal, but what is the value of adding this status condition for a failure to delete a client? It has some value for a dataverse admin, to know that the last attempt to harvest from a certain server resulted in a failure. Is it really useful to know that an attempt to delete a client failed? - should the startup check simply remove the \"delete in progress\" flag quietly instead? After all, seeing how the client is still there makes it somewhat clear that the attempt to get rid of it didn't work out, so they should try again? I may be missing some situation where it could actually be useful - so I'm open to hearing it."
+    },
+    {
+      "prediction": "MetricsConfig.forTable(table) -> MetricsConfig.of(table)",
+      "reference": "Can we also update the equality delete branch below?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `StepExecutionContext`, not in `IStepContext`.",
+      "reference": "This doesn't solve the entire problem we talked abut yesterday, right? Because it's possible that the step itself didn't skip, but rather that it chose not to yield the output in question?"
+    },
+    {
+      "prediction": "I think this should be `from dateutil import rrule`",
+      "reference": "I also do not see any usages of this import. Is this a fix for a crash?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `engine.startTask` method.",
+      "reference": "Feels like this needs to be invoked in `PostUnmarshalTask` since that's where most of the task setup happens."
+    },
+    {
+      "prediction": "`coalesce` is not needed here.",
+      "reference": "I wonder what would happen if the granteeName becomes DB__ROOT but the granteeID is not DB__ROOT ID (33333) but something else."
+    },
+    {
+      "prediction": "I think this should be `if f := os.LookupEnv(\"CFLAGS\"); f != \"\" {`",
+      "reference": "Feel like this should probably come into this tool as a flag rather than it just grabbing an env var internally?"
+    },
+    {
+      "prediction": "I think this should be `setParameter('nameOrHandle', nameOrHandle)`",
+      "reference": "This should only be sent if the driver is speaking to a W3C conformant remote, so we need an if-condition check like we have in the Python bindings."
+    },
+    {
+      "prediction": "Layout/AlignHash: Align the elements of a hash literal if they span more than one line.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think it would be better to use `self.assertEqual` instead of `self.assertListEqual`",
+      "reference": "nit: folders are always a numeric ID, projects as given by CAI use project number for ancestry"
+    },
+    {
+      "prediction": "This should be `ErrNetworkBad` instead of `ErrBadNetwork`",
+      "reference": "is this used anywhere?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "change pkg_errors to errors. Keep it consistent across all the files."
+    },
+    {
+      "prediction": "I think this should be `[\"--review-status\", \"unreviewed\", \"confirmed\", \"false_positive\"]` instead of `[\"--review-status\", \"unreviewed\", \"confirmed\", \"false_positive\"]`",
+      "reference": "I don't see any test cases which would test that fix dates are set properly on review status changes / storage events. For this reason please create some more test cases and also check my scenarios above."
+    },
+    {
+      "prediction": "I don't think we need to add the default response to the list of things that need to be normalized. I think we should just add it to the list of things that are already normalized. This way we don't have to worry about adding the default response to the list of things that need to be normalized, and we don't have to worry about removing the default response from the list of things that are already normalized.",
+      "reference": "This belongs to SwaggerModeler.cs `public override ServiceClient Build()`"
+    },
+    {
+      "prediction": "I don't think this is the right type. It should be `number|number[]|string|string[]|Function`",
+      "reference": "We could go further and directly give the tips of the `columns` property for someone (like me) who would like to specify some column width and let the others be autosized."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `m_additionalSubpassDependencies.size() > 0` block.",
+      "reference": "I think a `std::copy` with `std::back_inserter` will get this done in one call."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this error. I think it should be something like `UnsupportedHOLExperiment` or `UnsupportedFHL2Experiment`.",
+      "reference": "In the future: s/QUIC 36/Version36/ to make grepping easier"
+    },
+    {
+      "prediction": "`value` is a string, not a boolean, so this will always evaluate to `true` or `false` and not `on` or `off`. I think you want to do `value = ev.target.checked ? 'on' : 'off'` instead.",
+      "reference": "This is an example of why I prefer a flexible custom `render()` solution   'on'/'off' may not be the right value for every application. You have to make a lot of decisions even for very simple form fields :("
+    },
+    {
+      "prediction": "I think it would be better to move this to a separate PR.",
+      "reference": "we should also skip the e2e flow aggregator tests if the Flow Exporter feature is disabled."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `console.py` under the `script` module.",
+      "reference": "We should initialize the observation in the constructor of the Script class - otherwise, we don't have this feature for mitmdump."
+    },
+    {
+      "prediction": "I think this should be `tuple` instead of `(tuple, type(None))`",
+      "reference": "I would remove this line. Otherwise looks fine cc @ueshin"
+    },
+    {
+      "prediction": "I think this should be `1 == EXPECT_EQ (ledger_walker.walked_blocks.bucket_count (), 1);`",
+      "reference": "Would it be better to remove that line alltogether if it not directly relevant?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it seems like it should be.",
+      "reference": "Does \"FPU\" here include XMM/YMM/ZMM SIMD and not just x87 FPU/MMX?"
+    },
+    {
+      "prediction": "I think this should be `strutils.bytes_to_escaped_str(self.path)` instead of `strutils.bytes_to_escaped_str(self.path.encode('utf-8'))`",
+      "reference": "I'm not sure if we need to store path as bytes, and then decode/escape it when printing or just store it as a unicode string (as done here.)"
+    },
+    {
+      "prediction": "I think this should be `assert 'instance' == host.check_output('hostname -s', cwd='~/.m2/repo')`",
+      "reference": "Why is this flipped? Looks unrelated and our pattern is `expected == returned`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I'm not sure if it's the best way to differentiate between an in-person and an online workshop. I'm not sure if it's the best way to show the difference between an in-person and an online workshop. I'm not sure if it's the best way to show the difference between an in-person and an online workshop. I'm not sure if it's the best way to show the difference between an in-person and an online workshop. I'm not sure if it's the best way to show the difference between an in-person and an online workshop. I'm not sure if it's the best way to show the difference between an in-person and an online workshop. I'm not sure if it's the best way to show the difference between an in-person and an online workshop. I'm not sure if it's the best way to show the difference between an in-person and an online workshop. I'm not sure if it's the best way to show the difference between an in-person and an online workshop. I'm not sure if it's the best way to show the difference between an in-person and an online workshop. I'm not sure if it's the best way to show the difference",
+      "reference": "Can the `h5` tags be pulled outside of the `if` block?"
+    },
+    {
+      "prediction": "Why not use the `DirectoryStructureCreator` from `Symfony\\Bundle\\FrameworkBundle\\KernelBundle\\KernelBundle`?",
+      "reference": "From @PetrHeinz review: this should be also aliased (`ImageDirectoryStructureCreator`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make a copy of the original map and then deep copy the values.",
+      "reference": "Isn't AdditionalTags already of `Tags` type? If not, we should make it so, if it's not a breaking change"
+    },
+    {
+      "prediction": "I think this should be `MIT_LICENSE` instead of `MIT`",
+      "reference": "Add blank line below license header."
+    },
+    {
+      "prediction": "I think this should be done in the `init` function, since it is called after the machine is initialized.",
+      "reference": "Perhaps this could be moved into `InitPWM`? I suspect not initializing the clocks will reduce power consumption."
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "I believe this `order` isn't tested."
+    },
+    {
+      "prediction": "I think this should be `#if defined(WEBSERVER_METRICS) && defined(ESP32)` instead of `#if defined(ESP32) && defined(WEBSERVER_METRICS)`",
+      "reference": "No idea why it ended up in the code. You can remove the entire line."
+    },
+    {
+      "prediction": "I think this should be `iFromPeers := n.toAntreaPeerForCRD(ingressRule.From, cnp, controlplane.DirectionIn, namedPortExists)` and then `iFromPeers = append(iFromPeers, *n.toAntreaPeerForCRD(ingressRule.From, cnp, controlplane.DirectionIn, namedPortExists))`",
+      "reference": "\"i\" and \"From\" is a bit duplicate, and should it be singular given the method name? I mean \"fromPeer\" or just \"from\"?"
+    },
+    {
+      "prediction": "Shouldn't this be `scopes` instead of `scope`?",
+      "reference": "Did you verify this isn't going to break any existing test assumptions? We probably shouldn't modify this as we don't know what impact this is having on existing tests. It may invalidate a test that has different expectations of the scopes."
+    },
+    {
+      "prediction": "I think this should be `if claims.Namespaces == nil || len(claims.Namespaces) == 0 {`",
+      "reference": "technically not necessary since reading from a nil dictionary will always return found = false, so this should already be handled"
+    },
+    {
+      "prediction": "I think this should be called `RelaxedReclaimPurgeByNode` for consistency with the rest of the API.",
+      "reference": "This can return bool. RelaxedReclaimPurge(nodeID string) (bool, error)"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this, but I don't have a better suggestion.",
+      "reference": "Is there something we can do to make this automatic? At the very least, can we add Shanghai, Cancun, etc. here now so that we don't forget them?"
+    },
+    {
+      "prediction": "I think this should be `False` by default.",
+      "reference": "For boolean values, I'd rather have `argparse.SUPPRESS` as default value. No need to have a `(default: False)` in the help if it's a toggle."
+    },
+    {
+      "prediction": "I think it would be better to move this to `initSyncTest`",
+      "reference": "The genesis block needs to exist in the store the fetcher pulls from, this is because the fetcher will stop fetching when it finds a block it has seen before and this can sometimes be the genesis block."
+    },
+    {
+      "prediction": "I think this should be `0644` instead of `0600`",
+      "reference": "should we do the same for os.Mkdir(criuOpts.WorkDirectory, 0755)?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this change. I think it should be in `rebaseBraveStringFilesOnChromiumL10nFiles`",
+      "reference": "@bbondy Two messages because the label and description differ very slightly in their wording..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this check. It seems like it would be better to put it in `get_ymm_caller_saved` itself.",
+      "reference": "\"currently\" implies it should and will be changed: so TODO i#?"
+    },
+    {
+      "prediction": "This should be `t('imaging.pageTitleRequested')`",
+      "reference": "This should be `computed('i18n.locale'....`"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "Wait, does this not work if we do `btl.reload`? That seems odd that we have to have the object in the same state."
+    },
+    {
+      "prediction": "I think this should be `var eventListenerCount = 0; function on(scope, fn) {`",
+      "reference": "I seem to recall this being the cause of a bug at some point."
+    },
+    {
+      "prediction": "I don't think this is the right place to hide the settings menu item.",
+      "reference": "Can you remove it from the menu instead of setting it hidden? Or are there any other trouble?"
+    },
+    {
+      "prediction": "I think this should be `github.com/temporalio/temporal/.gen/proto/persistence`",
+      "reference": "And for \"execution \"."
+    },
+    {
+      "prediction": "I think this should be `type_files = mod.path + \"/types\"` and then `search_dirs << type_files if File.exist?(type_files)` instead of `search_dirs << type_files if File.exist?(type_files)`",
+      "reference": "Do agents usually download these from puppetserver when applying a catalog?"
+    },
+    {
+      "prediction": "I think this should be `return (ev->flags & ASIO_WRITEABLE) != 0;`",
+      "reference": "This should be `return (ev->flags == ASIO_DISPOSABLE);` because `ASIO_DISPOSABLE` has a value of `0` as defined in the enum in `asio.h` and because when `ev->flags` is assigned `ASIO_DISPOSABLE` it is assigned as `ev->flags = ASIO_DISPOSABLE` unsetting any other flags previously assigned."
+    },
+    {
+      "prediction": "I think this should be `o.PeriodicExportingMetricReaderOptions.ExportIntervalMilliseconds = options.DefaultCollectionPeriodMilliseconds;` and not `o.PeriodicExportingMetricReaderOptions.ExportIntervalMilliseconds = options.IsDelta ? options.DefaultCollectionPeriodMilliseconds : options.DefaultCollectionPeriodMilliseconds;`",
+      "reference": "this is not required, right? its the default already..."
+    },
+    {
+      "prediction": "I'm not sure this is the right error code for this case. I think it's more appropriate to use `StatusCodeConflictLockIsNotHeld` or `StatusCodeConflictRequiredLockIsNotHeld` or something like that.",
+      "reference": "\"contingent to\" -> \"contingent on\""
+    },
+    {
+      "prediction": "I don't think we need to define a new exception here. We can just use the existing `GroupByError` one.",
+      "reference": "Does pandas throw an exception like this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `DescriptorSet::DescriptorSet()`.",
+      "reference": "The point of the exercise has been to eliminate CoreChecks as an object dependency for DescriptorSet et. al."
+    },
+    {
+      "prediction": "I think this class should be renamed to `ScanContextImpl`.",
+      "reference": "Why do we need to make so much more of this ScanContext public?"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "is choice_name not needed anymore?"
+    },
+    {
+      "prediction": "Why do we need to validate the properties here? I don't think it's necessary.",
+      "reference": "Is this the right place to do the validation? If a user adds a bad property or performs some schema update that causes a validation error, that would break all writes to the table. To me, it doesn't seem like we are catching the problem early enough and possibly allowing a typo to break scheduled jobs. What do you think about adding this validation when altering the table? `UpdateProperties` could check whether any properties starting with `write.metadata.metrics` were modified and run this. Similarly, `UpdateSchema` could run this as well, although I think that we should probably modify `UpdateSchema` to simply update the properties for column renames (if that's easily done)."
+    },
+    {
+      "prediction": "I think this should be `if (std::find(g_hip_visible_devices.begin(), g_hip_visible_devices.end(), (i - 1)) == g_hip_visible_devices.end()) continue;`",
+      "reference": "Why is this necessary? Just in order to get an indexed loop?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this metric. I think it should be in the pool sync metrics, but I'm not sure if there is a better place to put it.",
+      "reference": "instead of using `prometheus.Gauge`, using `promethus.Counter` will be better."
+    },
+    {
+      "prediction": "I think this should be `code.match(/P11[7J4T1ACZ]....$/)` and not `code.match(/^P11[7J4T1ACZ]....$/)`",
+      "reference": "maybe we should have a unit test for this and then just test one case in `spec/services/ncr/approval_manager_spec.rb` ?"
+    },
+    {
+      "prediction": "Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "Why is this necessary?"
+    },
+    {
+      "prediction": "I think this method should be moved to `org.apache.iceberg.Table#getProperties()`",
+      "reference": "Minor: If this is going to be public, we should fix the name. `get` doesn't add any value. How about `distributionModeFor` instead?"
+    },
+    {
+      "prediction": "I think this should be `urwid.Frame(focus, window=window)`.",
+      "reference": "This needs a more self-explaining name. Maybe `.is_focused`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `gulp/tasks/azurereport.js` since it's related to Azure Reporting.",
+      "reference": "minor: I'd add a space between `,` & `'paginggroup'`."
+    },
+    {
+      "prediction": "I think this should be `--reupload-all` instead of `--reupload`.",
+      "reference": "\"Re-upload all fragments (even if fragment is already in registry)\""
+    },
+    {
+      "prediction": "I'm not sure if we want to expose this in the API, but I think it's fine to expose it in the CLI.",
+      "reference": "I don't see us ignoring the field. Are we sure Prometheus wouldn't refuse to load the rules?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `ViewportManager` class.",
+      "reference": "I wonder if we can just check if the mSession is active and then only call `setActive(true)` and `callSurfaceChanged()` if it isn't?"
+    },
+    {
+      "prediction": "I think this should be `EXE_MEMORY_FOR_UNPACK_ROWS_IN_MB`, not `EXE_MEMORY_FOR_PROBE_CACHE_IN_MB`.",
+      "reference": "A default value of 100 MB maybe too small. I understand that this is good for mixed workloads, but do consider than plans with Unpack, especially when used for insert/upsert are simple. Unpack is always serial and part of master exe. Often there is only one in a query. The cost of having a low value here seems to be that queue length from unpack could be limited by this when we are loading a table with wide rows (even if actual data in row is not wide). We know from performance runs that queue length is critical to upsert load performance. This is a hunch on my part and could be misguided. It will be good to performance test for upsert of table with wide rows, if you think that is worthwhile."
+    },
+    {
+      "prediction": "I don't think we need to add these two functions to the interface. They are only used by the validator, and the validator does not need to know about the bucket pool at all.",
+      "reference": "think we should pass in ctx, and use ctx.GreenlandHeight inside bucketPool to determine if create or not"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it would be better to move this test to `org.mule.runtime.core.internal` or `org.mule.runtime.core.internal.dsl`",
+      "reference": "A correct by unrelated change. We'll slide this one in without another word ;)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "Why remove this?"
+    },
+    {
+      "prediction": "Why did you remove the `Twig_Environment` import?",
+      "reference": "I still see some usages of not namespaced variants (look for Twig_ in project, about 42 matches). Is it intentional?"
+    },
+    {
+      "prediction": "I think this is the only place where we're using the new `specs` package.",
+      "reference": "Is this a new dependency? Does this require a dep update?"
+    },
+    {
+      "prediction": "I think this should be `else if (opts.generate_hexfloat_in_json && type.base_type == BASE_TYPE_FLOAT || type.base_type == BASE_TYPE_DOUBLE)`",
+      "reference": "> opts.generate_hexfloat_in_json && **IsFloat**(type.base_type)"
+    },
+    {
+      "prediction": "I think this should be renamed to `AppVersion` or `AppVersionString`",
+      "reference": "As an alternative to this, we could also create a new method, like `Common() *AppManifest` This way we don't have to add a new method to the interface everytime we add a new field to the `AppManifest`. This is a nit though so it's up to you."
+    },
+    {
+      "prediction": "I think this should be `pip>=10.0.0` and not `harparser`",
+      "reference": "As long as we have this feature as an inline script, I'm inclined to remove the dependency here. I'd suggest that we just try to import harparser and pytz and throw an error instructing the user to install the dependencies (catch for `ImportError`). In the long term, we probably want to include that in the mitmproxy core, but we first want to introduce a clean extension interface before that's going to happen. As soon as said interface is there, porting the inline script and adding the dependency again should be a breeze."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this, but it works for now.",
+      "reference": "makes it work nice with vscode. all logic is now in `.gulp/*.iced` files"
+    },
+    {
+      "prediction": "I don't think this is the best name for this method. `List<WebElement>` is a bit vague. What about something like `findMatchingElements` or `findMatchingElementsBy`?",
+      "reference": "should a similar change be done for findElement?"
+    },
+    {
+      "prediction": "I'm not sure this is right. The `requests_cached_late_votes` count should only be incremented if the `requests_generated_votes` count is also incremented.",
+      "reference": "`+` -> ` + ` I guess. Same below"
+    },
+    {
+      "prediction": "I don't think this is the right place to set the givenFeeSat",
+      "reference": "`typeof` is not a function - its an operator, and the standard way to use it is as `typeof foo !== ...` (i.e. no parenthesis). Also, I would personally use `opts.fee != null` instead (with a non-strict comparison, which also identifies `undefined` values because `null == undefined`)."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this test. I think it should go in `spec/reader/chapter_a_spec.rb` under `it \"does not warn about unreadable files\" do`",
+      "reference": "Is there a reason why you're not using Rspec `skip`? It allows to specify message and makes it clearly visible in test results which tests were skipped."
+    },
+    {
+      "prediction": "I think this should be `+optional` as well.",
+      "reference": "how about \"all matching SyncSets and SelectorSyncSets\""
+    },
+    {
+      "prediction": "I think this should be `VerifyChecksum` instead of `VerifyFingerprint`",
+      "reference": "What is the logic behind `ReadVectorSize` reading `CountElement64`, then skipping `T` (in this case `unsigned`)? Naming doesn't cover what's actually happening here. Ideally I'd also take this through a `files` layer, any input as to how?"
+    },
+    {
+      "prediction": "I don't think you need to import javax.servlet. You can just use `javax.servlet.http`",
+      "reference": "This looks like an unused import to me?"
+    },
+    {
+      "prediction": "I think this should be called `TransferTaskInfoToProtobuf` instead of `TransferTaskInfoFromProtobuf`",
+      "reference": "`RW` means read/write. Why do we have it here, as part of a func name?"
+    },
+    {
+      "prediction": "I think this should be `LOG.info(String.format(\"Connecting to %s and %s\", publishConnection, subscribeConnection));`",
+      "reference": "I'd keep this at `info` level..."
+    },
+    {
+      "prediction": "Is it possible to use `h2o_buffer_reserve` here instead of `h2o_buffer_try_reserve`?",
+      "reference": "I think you missed this."
+    },
+    {
+      "prediction": "I think we can remove the `NETCOREAPP2_1` ifdef here.",
+      "reference": "List all of the current TFMs so we know if this gets outdated. E.g. this breaks if we add 2.2."
+    },
+    {
+      "prediction": "I think this should be `spec.Xattr = []byte{req.GetXattr()}` instead of `spec.Xattr = req.GetXattr()`",
+      "reference": "here, you want to use req.GetXattrOpt() referring to the in line 514 of api.proto. This will be nil if not provided. See the example above on line 801 on this file"
+    },
+    {
+      "prediction": "I think this should be `nextMedia = callback.getNextInQueue(currentMedia);` not `nextMedia = currentMedia; nextMedia.setPosition(0);`",
+      "reference": "I think this should be done outside LocalPSMP, but in `getNextInQueue`. The reason is that I want to reduce the dependence of the media players on the preferences and database. Also, it will then probably work on Chromecast."
+    },
+    {
+      "prediction": "I think this should be `name = std::string(name);` instead of `name.assign(demangled == nullptr ? \"\" : demangled);`",
+      "reference": "Please remove it or remove all `std::` namespace prefixes."
+    },
+    {
+      "prediction": "I don't think we need this file anymore.",
+      "reference": "wait! I'm confused - is that an empty file ?!"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "These should all call `findUpVirtual`."
+    },
+    {
+      "prediction": "I think this should be `file_dict.get('mime-type', None) or 'application/octet-stream'` instead of `file_dict.get('mime-type', mime) or 'application/octet-stream'`",
+      "reference": "This changes the behavior. Original behavior was \"set if not set\", while new is \"just set\"."
+    },
+    {
+      "prediction": "I think it would be better to keep the `substring(7)` as it was in the previous version and just change the `7` to `8` in the next line.",
+      "reference": "Let's keep using the regular expression to make sure that the token is well-formed. I think the regular expression should be altered instead of doing a substring."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this change. The `Array` constructor should take a `T[]` as an argument, not an `Object[]`. If we want to make this change, then we should make it in the `Array` constructor.",
+      "reference": "All the operations are delegated to this entity, hence the rename. `back` can have too many meanings."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "@trevor-vaughan it looks like the spec failures are caused by the fact that although it was a great idea to put the guard clause here first & get the error case out of the way, the main code path has been erased when I assume it should be just below the guard clause."
+    },
+    {
+      "prediction": "I think we should move this to `protocol.go`",
+      "reference": "Could you delete this constant?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "looks like this snuck in from your other change?"
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "These env vars are not set in any CF environment, on purpose, because we are moving away from using env vars to store role-based information and instead using the database. So in a CF environment, the wrong emails get used (the defaults, rather than what is in the db)."
+    },
+    {
+      "prediction": "I think it would be better to keep the locals as `template, question, answer, readonly, locking`",
+      "reference": "this partial is used also for previewing a template, did you test if still works?"
+    },
+    {
+      "prediction": "nit: I think we should use `fmtAppInitDockerfilePrompt` and `appInitDockerfileHelpPrompt` instead of `fmtAppInitAppNamePrompt` and `appInitDockerfileHelpPrompt`.",
+      "reference": "Would it be better to put like `Which Dockerfile would you like to use for %s?`"
+    },
+    {
+      "prediction": "I think we can remove the `strict` option from `update()` and `updateMany()` and just have the `strict` option in `updateOne()` and `updateMulti()`.",
+      "reference": "You mistakenly removed `/docs` here, please add it"
+    },
+    {
+      "prediction": "I think it would be better to say that the returned list is sorted in ascending order.",
+      "reference": "I missed this earlier, but why does this say \"file blocks\"? This should probably be \"recommended split locations\"."
+    },
+    {
+      "prediction": "I think we can remove FilterDeleter pointer from AlertsServer struct.",
+      "reference": "Please change this to non-exported."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to remove the assert. I think it should be removed in the next PR.",
+      "reference": "This is ignoring compatibility: it needs to check the size to ensure this field exists in the client."
+    },
+    {
+      "prediction": "I think this should be `minimumPrepareMessages + 1` instead of `minimumPrepareMessages`",
+      "reference": "Do all the prepare authors also need to be unique? Or is it enough the we have minimumPrepareMessages. I guess I'm wondering if need a hasDuplicateAuthors check like in the RoundChangeCertificateValidator."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be under `github.com/influxdata/flux/stdlib`",
+      "reference": "we had some situations where we merged some promql bits then took them out. i'm not sure what's going on here, but unless your code actually needs promql (probably not?) let's remove this line."
+    },
+    {
+      "prediction": "Shouldn't this be `@var \\Shopsys\\ShopBundle\\Model\\Article\\Article`?",
+      "reference": "Annotations in this hunk should follow PhpDoc style (above the occurrence, type first)"
+    },
+    {
+      "prediction": "I don't think this is the right place to throw this exception. I think it should be in `BoundType` and `BoundType.EXTENDS`.",
+      "reference": "This is an improvement - thanks! Could we rephrase slightly to not use the phrase \"is not supported\" -- instead stating _WHAT_ isn't valid please? e.g. maybe `String.format(\"Unsupported BoundType provided: %s\" type)` or something like that."
+    },
+    {
+      "prediction": "I think this should be a `uint32` instead of an `int`.",
+      "reference": "nit: `procfd` seems more appropriate now?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `auth.py`",
+      "reference": "Let's not call it django. Maybe `auth_provider_headers`? `auth_headers`?"
+    },
+    {
+      "prediction": "I think it would be better to have a separate function for this, so that we don't have to repeat the code in `isClientAuthorized` if we decide to add a new auth method.",
+      "reference": "Should we mention TLS map here?"
+    },
+    {
+      "prediction": "I don't think you need to import net.runelite.api.ClanMemberRank",
+      "reference": "remove this empty line pl0x"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Why not have this in the `Dispatcher`? Doesn't seem like NCR-specific functionality."
+    },
+    {
+      "prediction": "I think this is the wrong place to do this. I think it should be in `test/testkv.go`",
+      "reference": "If you need only test db, use `memdb.NewTestDB(t)`"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "I'm not sure I like setting the IsEmpty property true for upgraded connections since it feels a middle misleading. Maybe we can leave the ForUpgrade class as is and change this condition to `if (!messageBody.IsEmpty && !messageBody.RequestUpgrade)` to make things more explicit."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "git checkout since this is unrelated to this PR? (running migrations also changes this for me -- not sure why it keeps going back and forth"
+    },
+    {
+      "prediction": "I'm not sure about this change. It seems to me that the TLSv1.0 spec doesn't say anything about the TLSv1.1 protocol, and it seems to me that the TLSv1.1 spec doesn't say anything about the TLSv1.2 protocol either. So I'm not sure what the best thing to do here is. If we're going to change the TLSv1.0/1.1/1.2 fields, I think we should do it in a way that doesn't change the TLSv1.0/1.1/1.2 fields at the same time.",
+      "reference": "Is this change needed?"
+    },
+    {
+      "prediction": "This should be `if (tmpInt >= 0 && tmpInt <= 127)`",
+      "reference": "Isn't the % 128 redundant here?"
+    },
+    {
+      "prediction": "I think we can remove this now, since it's not used in this test.",
+      "reference": "Are you testing that the values are always false? I think you should test for setting values to true or false, right? Who is going to take action with these values?"
+    },
+    {
+      "prediction": "I think this should be `extern const struct batch_queue_module batch_queue_mpi = { /* ... */ };`",
+      "reference": "Please change MPI to CCTOOLS_WITH_MPI"
+    },
+    {
+      "prediction": "I think this should be `if (!(cond)) { ... }`",
+      "reference": "Why you can't use function here?"
+    },
+    {
+      "prediction": "I think we should remove this file. It's not used anywhere.",
+      "reference": "We should scope this down to only the methods we use."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `mkString` doesn't take a space between the parentheses, so this is not a valid call to `mkString`. I think we need to fix `mkString` to take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mkString` to not take a space between the parentheses, or fix `mk",
+      "reference": "Oh, thanks for catching - I thought I've catched all after changing `mkString(infix, prefix, suffix)` to `mkString(prefix, infix, suffix)`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `authz.go` since it's only related to the authz service.",
+      "reference": "Suggestion: `.. the CertificateRequest is ready for signing` - could we perhaps word this differently? I understand that in this case it will be the associated X.509 certificate that can now be signed, so maybe `the certificate is ready for signing` ? (Same with `CertificateRequestReasonDenied`)."
+    },
+    {
+      "prediction": "I think this should be `accountID` instead of `accountID || 0`",
+      "reference": "We should use `! isValidAccountID( accountID )` for the `accountID` checks."
+    },
+    {
+      "prediction": "I think it would be better to make `inventory` a `Proc` and pass it to `#inventory` instead of `#inventory`.",
+      "reference": "That alternative does seem better. Did you want to try to do it in this PR? It makes sense to me that the inventory would always be available before creating the executor."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Missing the `-y` argument which all other `apt-get install` commands have."
+    },
+    {
+      "prediction": "This should be `%u` instead of `%u32`.",
+      "reference": "So I think we are going to settle on the \"concise and elegant\" `PRI` macros as they are the safest option for now. Even though they make my eyes bleed a little..."
+    },
+    {
+      "prediction": "I think this should be `!strings.Contains(input, string(phOpen)) && !strings.Contains(input, string(phClose)) && !strings.Contains(input, string(errOnUnknown)) && !strings.Contains(input, string(errOnEmpty))`",
+      "reference": "Interesting, was this needed for a test case to pass? I figured if there is no opening brace, there is definitely no placeholder -- we don't even have to check for a closing one."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "I think we want to provide getAction API instead"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this check. I think it should be in the `SynthDriver.__init__` method, since that's the place where the driver is instantiated.",
+      "reference": "There is also `winVersion.isWin10`. I think this should be converted to use the helper function. The helper has a note that it doesn't work in source copies, but looking at the implementation it looks equivalent to what you have here."
+    },
+    {
+      "prediction": "I think this should be `config.fail_fast = false` and not `config.fail_fast = true`",
+      "reference": "This doesn't seem like it should be part of this pull request."
+    },
+    {
+      "prediction": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Is the idea that `@product` (and `@workshop` for `workshops_controller`) would eventually go away here?"
+    },
+    {
+      "prediction": "I think this should be `04055304e432294a65ff31069c4d3092ff8b58f009cdb50eba5351e0332ad0f6` instead of `0d21840abff46b96c84b2ac9e10e4f5cdaeb5693cb665db62a2f3b02d2d57b5b`",
+      "reference": "We should add those only once we have successfully synced to the BSC main net, we have a utility to generate those. Please remove for now"
+    },
+    {
+      "prediction": "I think it would be better to have a test that verifies that the original env is not modified when the env file is not found.",
+      "reference": "And how is that related?"
+    },
+    {
+      "prediction": "I think we can remove the `EnableRemoteClusterConnection` flag now.",
+      "reference": "this should be required"
+    },
+    {
+      "prediction": "I think this should be `@NotNull` instead of `@Null`",
+      "reference": "This appears to be unused"
+    },
+    {
+      "prediction": "I think we can remove this method as it is not being used anywhere.",
+      "reference": "I think that \"table's state\" isn't clear enough. How about \"history of the table's current snapshot\" like the one below?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should go in `src/obj/include/obj/instr.h` and `src/obj/include/obj/instr_imem.h`.",
+      "reference": "This looks like a bug: pretty sure there are scatter-gather instructions on AArch64. Ditto below."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that the `t.tracer` is shared by all the transports, and we don't want to share the same `t.tracer` with multiple transports. I think we need to find a way to share the same `t.tracer` with multiple transports, but in a way that is independent of the underlying transport.",
+      "reference": "No need to change this: id love if we changed as a team to unkeyed fields, it ends up catching a lot more at compile time, at minimal cost"
+    },
+    {
+      "prediction": "I think this should be `AppName returns the name of the application.`",
+      "reference": "Should this be in the parent struct? `AppManifest` since it's embedded to `LBFargateManifest` it'll get the method as well."
+    },
+    {
+      "prediction": "I think we should use `logp` instead of `github.com/cihub/seelog`.",
+      "reference": "\"Openvpn check\" we could move this to separate prefix."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The problem is that we don't know what `body` is until we get the response, so we can't tell if it's `None` or something else.",
+      "reference": "body is being used here for the first time without every being populated. This will error. To run tests do `./go clean test_py` and that will run the Firefox tests"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `onFinished` function.",
+      "reference": "We don't return now?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I've been able to get Travis to pass.",
+      "reference": "This might end up not doing exactly what we want (the retry only picks up the `go get`, not the `go build`, due to the `&&`)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `OpenTelemetry.Console.Exporters` namespace, under the `ConsoleExporterOptions` class.",
+      "reference": "Is there a BOM change?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "It's not a good practise to delete the public static constant."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to check for signedness. I think it should be in the verification pool, but I'm not sure if that's the right place to do it.",
+      "reference": "I deleted this because `WellFormed` is immediately called by `verify.TxnPool` below. Can someone please double check this for me since it's... pretty important"
+    },
+    {
+      "prediction": "I'm not sure about this one. I think it should be `ply>=3.5, <3.12`",
+      "reference": "Did you actually test this with ply 3.4? That release is pretty old (2011), so I think we can bump this to at least 3.6 (2015) or even 3.10 (2017)..."
+    },
+    {
+      "prediction": "I think it would be better to have this check in `deployHiveCluster` instead of `deployHive`",
+      "reference": "This can be simplified somewhat to `pointer.Int32Ptr(0)`. But it is not necessary."
+    },
+    {
+      "prediction": "I think this should be `bs1 > b2 > 0 && !(bs2 > b1 > 0)`",
+      "reference": "Clever but perhaps confusing."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be in the same place as the `deleteSyncSetResources` call.",
+      "reference": "@abutcher Is it a safe assumption that secrets are the only objects that need to get attached to syncsets?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Should you capture o.asflr before releasing consumer's lock?"
+    },
+    {
+      "prediction": "This should be `!= null` instead of `== null`.",
+      "reference": "Just asking: Should this happen, that null is passed? If this is the case I think we should think about passing an `Optional` object to the method."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "what about making a `ProfilesController` or `UserProfilesController` and having this be a `show` action instead? That would be more Railsy (although that can be considered a compliment or a dis, depending on who you are :hamburger: )"
+    },
+    {
+      "prediction": "I think this should be `device_data->enabled_features = *enabled_features_found;` instead of `device_data->enabled_features.core = *enabled_features_found;`",
+      "reference": "I'd move the empty assignment and non-null case into the if check directly above (adding an else clause as needed)"
+    },
+    {
+      "prediction": "Shouldn't this be `Lang::get('system::lang.locale.ar', 'flag-sa')`?",
+      "reference": "Please remove this extra line of whitespace"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "maybe it's time to rename `ProductSearchExportWithFilter` to something better, what do you think?"
+    },
+    {
+      "prediction": "I think we can remove the `attributePrefix+capabilityFirelensFluentd` part from the `appendTaskEIACapabilities` and `appendFirelensFluentdCapabilities` functions.",
+      "reference": "Instead of adding new code here - can you move these methods to agent_capability.go, so the same is used for unix and windows as well. this will need removal of these methods from agent_capability_unix.go as well."
+    },
+    {
+      "prediction": "I think this should be `hash=hash` instead of `hash=hash or None`",
+      "reference": "Make it the last parameter, just in case someone uses the API with non-keyword args."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this method. It's a bit confusing because we're not removing a team member, but also not removing a collab. Maybe something like `remove_collaborator` or `remove_collaborator_from_team`?",
+      "reference": "Don't extend an instance initialized by `Struct.new`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `pmix_list_remove_item` and `PMIX_LIST_FOREACH_SAFE`. The reason is that `tdes` is called from `pmix_list_remove_item` and `PMIX_LIST_FOREACH_SAFE`.",
+      "reference": "Just wondering - would it make more sense to simply replace `PMIX_DESTRUCT(&t->nslist)` with `PMIX_LIST_DESTRUCT(&t->nslist)` here, and then add `PMIX_RELEASE(p->jobbkt)` to the `pmix_nspace_caddy_t` destructor on line 154 of src/include/pmix_globals.c? Seems to me like we always want to have these things removed/destructed."
+    },
+    {
+      "prediction": "I don't think you need to add this line. You can just use `RuleSetFactoryTest::VALID_XPATH_CLASS_NAMES`.",
+      "reference": "I think, we should fix/improve AbstractRuleSetFactoryTest. I guess, both apex and visualforce rules are now tested, which is unnecessary."
+    },
+    {
+      "prediction": "I think this should be moved to the `ActionController` class.",
+      "reference": "In php8 named parameters were introduced and now it is required to match called method parameter name when setting parameters by array destructing or `call_user_func_array()` etc."
+    },
+    {
+      "prediction": "I don't think you need to import `Try` here.",
+      "reference": "easily possible to get that information without exposing internal information"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Why do we need tristate logic (pe, foss, nil)?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "this change needs to be reverted to what is on `master`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "Do you still need to call the superclass `createVariable` if you've already determined that the variable exists?"
+    },
+    {
+      "prediction": "I think this should be called `EachByPackFinalized` to be consistent with the other methods.",
+      "reference": "Ignoring the pack entry from an existing entry but using the new entry from a non-finalized index, is subtle enough that it needs explaining."
+    },
+    {
+      "prediction": "This should be `auto it = std::min_element(instants.begin(), instants.end(), [](const uint32_t a, const uint32_t b) { return a < b; });`",
+      "reference": "This is a revert, is it really an issue?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I can't think of a better way to do this.",
+      "reference": "can we delete this file altogether?"
+    },
+    {
+      "prediction": "I think this should be `flag.CommandLine.Parse([]string{\"-test.coverprofile\", tempCoveragePath()})` and not `flag.CommandLine.Parse([]string{\"-test.coverprofile\", tempCoveragePath()});`",
+      "reference": "Suggestion: log some additional info here so we know where we are i.e 'Failed to prepare coverage framework..'"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `System.Net` or `System.IO` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or `System.Net.Sockets` or",
+      "reference": "There didn't seem to be any good place to put constants that are .NET Tracer-specific, so I created this internal static class. If there's a better place, let me know."
+    },
+    {
+      "prediction": "I think this should be `UpdateDeviceStatusBufferSize` to be consistent with the rest of the code.",
+      "reference": "Log message should be started with upper-case word."
+    },
+    {
+      "prediction": "I think this should be `Layout<EventId>` instead of `Layout<int>`. The reason is that `EventId` is a `Layout<EventId>`, not a `Layout<int>`.",
+      "reference": "changes for example usage in this class"
+    },
+    {
+      "prediction": "I think we can remove the `emby-collapse` css file.",
+      "reference": "I know frameworks that support current directory when loading dependencies, is this a limitation of the requirejs loader or can we fix it somehow?"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "`ctx` is unused in EnableApplication"
+    },
+    {
+      "prediction": "I think this should be `return \"\";` instead of `return stringtab(\"\")`.",
+      "reference": "@jemc , this is one change needed in `generate_multi_dot_name`. I believe this * will do no harm to working pony code: The new code was added in a place where it is throwing an assert. So no working code will get affected. * is in sync with the rest of the method: if the parent ast node has null `data`, we are supposed to return empty string. That is possible when we have expressions wrapped in braces."
+    },
+    {
+      "prediction": "I think you can remove this import.",
+      "reference": "Should not be a static import."
+    },
+    {
+      "prediction": "I think we can remove the `change.Attachments` and `change.Attachment` fields from the `SubmitTaskStateChange` call, since they are no longer used.",
+      "reference": "should we be using `aws.String` here? it seems there are a few other spots we could use `aws.String` in the `if change.Attachment != nil { ...` block."
+    },
+    {
+      "prediction": "I think it would be better to use `getServiceReportURL( 'content-publisher-overview', _.defaults( args, { '_u.date00': startDate, '_u.date01': endDate } ) )` instead of `getServiceReportURL( 'content-publisher-overview', { _u.date00: startDate, _u.date01: endDate } )`",
+      "reference": "We can't pass raw dates like this because they need to be formatted as `YYYYMMDD` as noted in the IB. We added the `generateDateRangeArgs` utilities to handle this for us, as well as abstracting the relationship between arg name and specific dates which is not obvious just by looking at it. Let's update it to use `generateDateRangeArgs( { startDate, endDate } )` instead   Use care to make sure we're sourcing the utility from Analytics utils though since AdSense has its own version by the same name, which is different."
+    },
+    {
+      "prediction": "I think it would be better to call this `save_execution_data` or `save_execution_summary` to make it clear that it saves data related to the execution itself, and not just the summary.",
+      "reference": "`default=True` shouldn't be used for BoolParameters iirc."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this.",
+      "reference": "It would be better to use `return refStr[12:]`. or `11`... I have problem with counting.."
+    },
+    {
+      "prediction": "I think this should be `for _, testOutput := range target.TestOutputs`",
+      "reference": "Pretty sure these should only contribute to the runtime hash."
+    },
+    {
+      "prediction": "I think this should be `<div id=\"gsa18f-layout\"></div>`, not `<div id=\"ncr-layout\"></div>`.",
+      "reference": "Just verifying: this change doesn't break the 18f layout, right?"
+    },
+    {
+      "prediction": "I think this should be removed as well.",
+      "reference": "we should keep this counter using the \"isYarpcError\" api"
+    },
+    {
+      "prediction": "nit: I think this should be `new CaseInsensitiveMap<>()` instead of `new CaseInsensitiveMap<>()`",
+      "reference": "Could we just use case insensitive strings as keys instead of creating a new map implementation? e.g. simply using a treemap with a comparator `Map<String, Integer> idToPos = new TreeMap<>(String.CASE_INSENSITIVE_ORDER);` could work I think"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do this in a way that is resilient to changes to `RuntimeContextSlot<T>` in the future.",
+      "reference": "I think this should be a throw. Or the method should be TryGetSlot?"
+    },
+    {
+      "prediction": "I think this should be `catch (LuaError &e)` instead of `catch (const std::exception& e)`",
+      "reference": "We don't usually put spaces between ( and [, and I think `path` should be passed by reference."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get this to work.",
+      "reference": "Make sure to `snake_case` locally scoped variable names in the future"
+    },
+    {
+      "prediction": "I think this should be `HealthCheckPath string `yaml:\"healthcheck_path\"`",
+      "reference": "Maybe we should use underscores like `health_check` - what do you think?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could find that worked.",
+      "reference": "Do a TrimSpace before setting."
+    },
+    {
+      "prediction": "nit: I think we can remove the `SubnetGatewayIPV4Address` field from the `CIDRBlock` and `CIDRBlockIPv4` structs as well.",
+      "reference": "Are the fields in this struct grouped by IP address family or by function? There are separate fields for IPv4Addresses and IPv6Addresses, but a single field for DomainNameServers. Should this new field be named SubnetGatewayAddress with type array consisting of both IPv4 and IPv6 gateways?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": ".find(id) is being deprecated to Rails 5."
+    },
+    {
+      "prediction": "I don't think we need to do this. `equalise_attributes` is only used in `equalise_coords.py`, and that module is only used in `equalise_coords.py` because it has to do with equalising the attributes of a single `Cube`, not the attributes of a whole `CubeList`. So I think we can just do this in `equalise_coords.py` and leave `equalise_coords.py` unchanged.",
+      "reference": "Will be good to have the iris interface moved to geoviews. Could this be done for 1.10.6?"
+    },
+    {
+      "prediction": "I think this should be `return err` instead of `return nil`",
+      "reference": "does this get wrapped in the calling method?"
+    },
+    {
+      "prediction": "I think this test should be moved to `tests/authorization/test_group_authorization.py`",
+      "reference": "Do we still need this test case? If your new test cases test this, then we can remove it."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this method, but it's the name I've seen used in the past, so I'm going to go with it.",
+      "reference": "If you want to express \"any role that is not creator\" you could use the following statement: Role.where(plan: self).where(Role.not_creator_condition).any? which would be less verbose and a bit more efficient if it is used for Yes/No shared?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "Do we need any update on the test side?"
+    },
+    {
+      "prediction": "I think this should be done in `notifyOneOpLocked` instead of `notifyOneOp`.",
+      "reference": "There's a few `return nil` cases above this -- we should probably move this above the big switch."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this error. I think it should be in the `TransactionInLedgerError` struct, since that's where the transaction is already in the ledger.",
+      "reference": "`lile *LeaseInLedgerError` to reduce copying?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `NewSessionQueuer` class itself, since it's only used by the NewSessionQueuer.",
+      "reference": "should it be `newsessionqueue` or `newsessionqueuer`? In case we'd like to be consistent"
+    },
+    {
+      "prediction": "I don't think we should be changing the signature of this method. We should keep the old signature and add a new parameter for access.",
+      "reference": "We don't have a requirement anywhere in Test Platform for GetStream() with write access. It is ok to directly change `return new FileStream(filePath, mode, FileAccess.Read)`."
+    },
+    {
+      "prediction": "I don't think this should be in this PR.",
+      "reference": "FWIW `from qutebrowser.browser import hints` is okay - it's just things like `from qutebrowser.browser.hints import HintManager` I try to avoid."
+    },
+    {
+      "prediction": "I think this should be `esc_attr( get_theme_mod( 'neve_transparent_header', false ) )`.",
+      "reference": "@preda-bogdan let's use a filter for the header classes, something like, hfg_header_classes where we hook this logic, in neve pro"
+    },
+    {
+      "prediction": "Why is this change required?",
+      "reference": "Out of curiosity.. is there a rule to determine the package name as prevDirectory+currentDirectory."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `get_xmm_caller_saved` and `get_ymm_caller_saved` since they are the only places where we need to do it. The rest of the logic should be done in `get_xmm_vals` and `get_ymm_vals` since they are the only places where we need to do it.",
+      "reference": "It seems this needs to check the lazy cxt switching flag. It is used on the initial thread for early injection and in other places where the lazy switch should apply."
+    },
+    {
+      "prediction": "I think we can remove the `@since 1.8.0` here.",
+      "reference": "Why not use the JSdoc annotation as it did before?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think this should be done in the `ProductPropertiesUtil` class.",
+      "reference": "Why don't we drop the packageImplementationVersion in general? That way the version will be fetch all that time only from the properties file and it will be consistent."
+    },
+    {
+      "prediction": "I think this should be `std::vector<std::unique_ptr<FieldDef>>`.",
+      "reference": "Please use the style of the rest of the code, a space between the type and `*`"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "nit: remove extra blank lines."
+    },
+    {
+      "prediction": "I think this should be `taskGroupNameDefault = \"copilot\"`",
+      "reference": "nit: What do you think of `copilot-task-group`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `input` event, since that's where the value is being changed.",
+      "reference": "We should check whether or not the input node has a value since we can just have an onChange on an uncontrolled component as well."
+    },
+    {
+      "prediction": "I think this should be `(2021, 1, 1)`",
+      "reference": "I think there should be some reasoning below this line about the version updating."
+    },
+    {
+      "prediction": "I think this should be `H2O_COMPRESS_HINT_AUTO`",
+      "reference": "Please use `h2o_lcstris` instead of `strncasecmp` so that the comparison would be a complete match (instead of a prefix match; current code would yield true if [val,len] is [\"O\", 1]) and that it would be locale-independent."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we should be able to get the node cache from `fb` directly.",
+      "reference": "I prefer if possible for `if`/`else` statements to have the positive case first."
+    },
+    {
+      "prediction": "I think this should be `return isinstance(x, type) and isinstance(x, t)`",
+      "reference": "Docstring would be nice"
+    },
+    {
+      "prediction": "I think this should be moved to `BaselineErrorPronePlugin` since it's not specific to a single plugin.",
+      "reference": "Low risk to fix this by default because existing baseline consumers pass this check. We don't attempt to fix checks that have been opted out of."
+    },
+    {
+      "prediction": "This should be `#if defined(KOKKOS_ARCH_VOLTA70) || defined(KOKKOS_ARCH_PASCAL60) || defined(KOKKOS_COMPILER_CLANG) && (KOKKOS_COMPILER_CLANG >= 1300)`",
+      "reference": "What about `VOLTA72`, `TURING75`, `AMPERE80` and `AMPERE86`? We only want to set the number of teams for these two architectures or for all the architectures newer than Maxwell?"
+    },
+    {
+      "prediction": "I think it's better to add `from torchvision import ...` at the top of the file.",
+      "reference": "Since we have specified `load_from`, `pretrained` can be left None."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the rule controller, not the log controller.",
+      "reference": "I prefer to translate these `else if` into `switch` for more readability."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to update the trails state for a user. I think it would be better to move this logic to the `StatusesController#update_trails_state_for` method, since that's where the logic for updating the state of a single status is located.",
+      "reference": "This can also be a local variable in `create`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to remove the file. I think it should be done in the same place where we check if the file exists, but I'm not sure if there is a better place to do it.",
+      "reference": "Why not use `path.Join`?"
+    },
+    {
+      "prediction": "I think this should be `global.googlesitekit.api = API;`",
+      "reference": "Why not `export default API` instead? This ensures we export the same that we export on the global."
+    },
+    {
+      "prediction": "I think this should be `itk.ImageIOFactory.CreateImageIO(io_filename, itk.ImageIOFactory.FileModeType_ReadMode)`",
+      "reference": "This is a good change I think. It greatly helps reduce name conflicts."
+    },
+    {
+      "prediction": "I think we can remove this now that we have the `Placement` struct.",
+      "reference": "cluster not having omitempty, does this imply that cluster is required? seems empty is valid so just checking"
+    },
+    {
+      "prediction": "This is a bit of a hack, but I think it's the best we can do.",
+      "reference": "@aws/aws-ecs-agent, @yumex93: how concerned should we be about user provided envvar names clobbering existing envvars? i'm not convinced we should be doing additional validation here."
+    },
+    {
+      "prediction": "I think it would be better to have this in a separate file, since it is not part of the public API.",
+      "reference": "nit: aspects are defined in `OperationBase`, should `defineAspects` live there as well?"
+    },
+    {
+      "prediction": "I think this should be `os.path.join` instead of `os.path.abspath`",
+      "reference": "Why the `os.path.abspath`? `os.path.dirname` should always give a valid directory path."
+    },
+    {
+      "prediction": "I think this should be `getWineEnginesPath` and not `getWineEnginePath`",
+      "reference": "I think we should not consider special engine types on this level. I would prefer to build the specific engine path based on `application.user.engines` later on."
+    },
+    {
+      "prediction": "I don't think we need to check the endBound_ here, as the endBound_ is only used to filter the range scan.",
+      "reference": "Is `rbegin` correct? We can't make sure that the last index column is the `rbegin` of `scanItems`."
+    },
+    {
+      "prediction": "I think this should be done in `config.New()` instead of here.",
+      "reference": "check it is != 0, just like `EVMNetworkID` above"
+    },
+    {
+      "prediction": "I think this should be `Log.DroppedExportProcessorItems<T>(typeof(BatchExportProcessor<T>), typeof(T).Name, this.exporter.GetType().Name, this.droppedCount);`",
+      "reference": "if you just do \"this.GetType().Name\" and \"this.exporter.GetType().Name\", we get what we need."
+    },
+    {
+      "prediction": "I think this should be `public static final String VERSION_NUMBER = SalesforceSDKManager.getInstance().getAppContext().getResources().getString(R.string.api_version);` instead of `public static final String VERSION_NUMBER = \"v29.0\";`",
+      "reference": "That's the code that would prevent any SalesforceSDKTest from running. At class loading time, SalesforceSDKManager.getInstance() would throw a RuntimeException because init() had never been called."
+    },
+    {
+      "prediction": "I don't think we need this constant. We can just use `0` as the default revision.",
+      "reference": "there is no \"DEFAULT_REVISION\" logic, no need to define this."
+    },
+    {
+      "prediction": "I think it would be better to move this to the `else` branch of the `if` statement above.",
+      "reference": "so this is to fix a scenario where the app is first launched normally, then background and foreground through the link, we are not updating loginOption"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `componentDidMount` or `componentWillMount` since it's only called when the component is initialized.",
+      "reference": "It doesn't seem like the right way to implement this, because you add a new element that doesn't really exist (noteTextViewer). Instead you should modify the command handler `if (command.name === 'focusElement' && command.target === 'noteBody') {` in NoteText.jsx. Then focus either the editor or the viewer depending on what's currently visible."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "We don't add blank lines in imports."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this file.",
+      "reference": "Please add your name to the copyright list."
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `\"\"`, right?",
+      "reference": "You can elide the `if` check for empty string, because if it's not a value that is set, the string will be empty anyway."
+    },
+    {
+      "prediction": "`messagesSentLabel` and `actionsPerformedLabel` are not used in this file. Please remove them.",
+      "reference": "Different i18n keys. By mistake or on purpose?"
+    },
+    {
+      "prediction": "I don't think we should have this constant here.",
+      "reference": "Given the description for this is `// Describes how the ACME challenge self check behaves when it fails.`, perhaps `RetryForever` is better terminology here as when the check fails, we don't just wait, we retry periodically."
+    },
+    {
+      "prediction": "I think this should be ot_gpgme_error_to_gio_error (gpg_error, error, \"Loading data from fd %i\", fd);",
+      "reference": "Nit: let's add a more descriptive prefix in line with the other ones? E.g. `Unable to import key` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `spec/lib/bolt/ssh/test/test_helper.rb`",
+      "reference": "This looks like it will pass an array as `:key` in the default case. That doesn't reflect how this will actually work in practice."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in the `IntegrationTests` folder under the `IntegrationTests.Program` class.",
+      "reference": "Do we need this?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think we should move this to the `ofFlowBuilder` struct itself.",
+      "reference": "I am a bit confused about this function. If it is used to set priority, we actually use function \"Table.BuildFlow(priority uint16)\" to set the value. If it is used to parse the priority value, I think adding a property(if you don't want to consume Match.Priority directly) to cache it and consume that property should be simpler than using string regex."
+    },
+    {
+      "prediction": "I think this should be `node` instead of `packaged`.",
+      "reference": "In this case we probably meant \"package\"."
+    },
+    {
+      "prediction": "`dirname` is not available in IE11, so we need to use `getDirName` instead.",
+      "reference": "Just export `createFile` directly."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I would expect it to be in the `buildid-replacer` package, but I don't see it there.",
+      "reference": "This probably needs to get shipped with plz; you shouldn't have to redefine it within the user's repo. It should use `defaultPath` like the other things up the top of the file."
+    },
+    {
+      "prediction": "I think this should be `['btc'].testnet` instead of `['btc'].network.testnet`",
+      "reference": "interface for bitcoin should not change if possible. i.e: networks.testnet should return networks['btc'].testnet"
+    },
+    {
+      "prediction": "I think we should use `io.ReadCloser` instead of ` ioutil.ReadCloser`",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I think this should be `wlr_list_for_each_child(c, &surface->current->frame_callback_list, link)`.",
+      "reference": "We have this functions in a couple of places. It should probably live in util or something."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `after_save` callback for the model, since that's where the object is actually updated.",
+      "reference": "maybe this should be inside the destroy?"
+    },
+    {
+      "prediction": "`if torch.onnx.is_in_onnx_export()` -> `if not torch.onnx.is_in_onnx_export()`",
+      "reference": "Do not import torch in the test function because it will import torch every test iteration and will slow down the testing speed."
+    },
+    {
+      "prediction": "I think this should be `aws.StringPtr(id)`",
+      "reference": "I think we want to return an error here, not nil."
+    },
+    {
+      "prediction": "I think this is the only change that should be in this file.",
+      "reference": "Nitpicking, but I think it would be more accurate to use an actually supported value, i.e. `completed`."
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "How about renaming the `ApiJob` class to `JobApi`. As per rails convention, get method is to get a single object, not a collection, can we change that to `fetch`/`where`. It would be great if we can move the `get` method to a class method."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `factory_test.go`",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "This should be `nil, cache.ErrNotImplemented`.",
+      "reference": "`k` is unused in PutHash"
+    },
+    {
+      "prediction": "I think this should be `public function getSourceCode() : StatementsSource;`",
+      "reference": "None of the child return null here, plus it was creating an incoherence between interfaces."
+    },
+    {
+      "prediction": "I think this should be `min_threads = MIN(threads, 2000000000);`",
+      "reference": "Should we remove this entirely now then?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface. I think we should move it to `monitoring` package.",
+      "reference": "That is way too generic a name."
+    },
+    {
+      "prediction": "I think this test should be moved to `test/unit/dom/test.js`",
+      "reference": "This might be overkill for test code...but as stewards of accessibility it would be appropriate to have a label in the fixture."
+    },
+    {
+      "prediction": "I think this should be `secondary_full_clear` and `secondary_small_clear`?",
+      "reference": "Are these allocations necessary, or can these be instantiated normally (i.e., `VkCommandBufferObj secondary_full_clear(...)`). If the allocations _are_ necessar, I'd vote for using something like `std::unique_ptr` and then remove the associated `delete`s."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `ElasticsearchHttpStorage` itself.",
+      "reference": "es.clear should call this"
+    },
+    {
+      "prediction": "nit: can you move this to the `test_...` class below?",
+      "reference": "please remove. this is true by the definition of the Python language"
+    },
+    {
+      "prediction": "`<Ctrl-Return>` is the shortcut for opening the settings dialog.",
+      "reference": "Hmm, I wonder if binding `<Return>` is a good idea... this means it wouldn't be passed to webpages anymore, which makes me wonder how many people rely on that... I guess trying it is the only way to find out :wink:"
+    },
+    {
+      "prediction": "I think this should be `packetWithPorts(1, src: \":0\", dst: \":0\")` instead of `packet{protocol: 1, srcAddr: src, srcPort: 0, dstAddr: dst, dstPort: (icmpType << 8) | (icmpCode)}`",
+      "reference": "Golang naming convention is to use camel case `icmpPktWithTypeCode` Often the linter will complain"
+    },
+    {
+      "prediction": "I think we can remove the `RequestBodyMinimumDataRate` and `RequestBodyMinimumDataRateGracePeriod` properties as well.",
+      "reference": "nit: Probably should renamed to `InitializeBody`"
+    },
+    {
+      "prediction": "I think this should be `icon` instead of `settings.icon`",
+      "reference": "is `settings.icon` always null/undefined when there's no icon or does is it get set to a blank string? It might be worth wrapping it in an `isBlank()` anyway"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to register the bootstrap container. I think it should be in the `BootstrapContainer` provider.",
+      "reference": "put invoke to the end of the list, and other module at beginning of the list (line 86)"
+    },
+    {
+      "prediction": "I think this should be `float64(db.capacity*swarm.ChunkSize*9.5367431640625e-7)` instead of `9.5367431640625e-7`",
+      "reference": "I think that bytes is too precise. Can we just calculate the approximate value in appropriate units? Something like MB, GB depending on the value, `db capacity: 5000000 chunks (approximately 20GB)` ."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `src/test/javascript/jest-dom-test-utils.js`",
+      "reference": "This check shouldn't look at the content, you're already doing this with `none-empty-text`. I also think this check should be renamed to `none-empty-alt` or something like it. There is nothing specific to applets in this check to warrant putting `applet` in the check ID."
+    },
+    {
+      "prediction": "I think this should be done in a separate PR, as it's not related to the current PR.",
+      "reference": "Why did you change the signature of this method ? You can break BC doing that"
+    },
+    {
+      "prediction": "Why do we need to read the file again?",
+      "reference": "can you catch the error here please and return err if it can't read the template"
+    },
+    {
+      "prediction": "Shouldn't this be done in `onStart`?",
+      "reference": "Is this method called at another location too? I'm just asking because you didn't remove another call to `updateLibrary`"
+    },
+    {
+      "prediction": "I think this should be `@object.present? && ( @object.is_default || @object.org.funder? || @object.published )` and not `@object.is_default || @object.org.funder?`",
+      "reference": "Not necessary but you should be able to add a `?` on `is_default?` and `published?`"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to have a dedicated method for this.",
+      "reference": "here is us decoding directly the bytebuffer in benchmarks"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't have a better suggestion. The problem is that if we don't set the default values here, we will get a SerializationException when the Tracer is initialized in the Asp.Net HttpModule startup. The default values are added to the underlying logging framework, and, upon handling a new request, Asp.Net tries to copy the type log4net.Util.PropertiesDictionary from one AppDomain to another. Even though the type does not implement MarshalObjectByRef, Asp.Net is still trying to copy the type from one AppDomain to another.",
+      "reference": "When the Tracer is first created AND `DD_LOGS_INJECTION=true`, this is invoked and causes the issues described."
+    },
+    {
+      "prediction": "I think this should be `OFFCHAIN_EXTERNAL_PRIVACY_GROUP_DOES_NOT_EXIST`",
+      "reference": "NIT: If we are renaming the JsonRpcError, should we also rename this enum value to be consistent?"
+    },
+    {
+      "prediction": "I think this should be `VectorizedExpression`.",
+      "reference": "This could be made into a function expression, yes?"
+    },
+    {
+      "prediction": "I think this should be moved to the top of the function.",
+      "reference": "Oops, I missed this initially."
+    },
+    {
+      "prediction": "I think this should be `lambda_client.create_function(FunctionName=lambda_name)`",
+      "reference": "Can we add an assertion here the policy actually contains both statements created above?"
+    },
+    {
+      "prediction": "I think this should be `xpress|disable` and not `xpress`",
+      "reference": "a space after comma?"
+    },
+    {
+      "prediction": "I think this should be `if (skewFromParent != null && isLocalSpan(node.value())) {`",
+      "reference": "nit formatting here and below"
+    },
+    {
+      "prediction": "I think we can remove `mock_registration` and `mock_entry` now that they are unused.",
+      "reference": "I think that there is a general consensus of trying to avoid this kind of mocks in the new tests that we write. I would suggest to have tests using fake service implementations. In this case, we can have a fake entry service. Examples of how tests have been written this way are the tests for the `spire-server agent` and `spire-server bundle` commands. It would be great if we can have that kind of testing here also. What do you think?"
+    },
+    {
+      "prediction": "I think this should be `RunQuery(ctx context.Context, query string) (bool, string, error)`",
+      "reference": "`ctx` is unused in RunQuery"
+    },
+    {
+      "prediction": "I think this should be `os.path.relpath(os.path.basename(report.files[event['location']['file']]))`",
+      "reference": "Couldn't we use type hints to express this?"
+    },
+    {
+      "prediction": "I think we can remove the `icon` variable and just use `icon.classList.add('ratingbutton-icon-withrating')` below.",
+      "reference": "why you do `.classList.add()` here but `.replace()` in other places?"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "What is this sleep here for?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/unit/components/handsontable/test-auto-row-size.js`",
+      "reference": "I think that this test doesn't cover this bug correctly. When I attached the older version of the handsontable to this test it passes, should fail. Can you check that?"
+    },
+    {
+      "prediction": "Can we make this a `sync.RWMutex`?",
+      "reference": "Maybe this should just be passed in on construction time"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to create the monitor. I think it should be done in the `wlr_drm_backend_monitor_create` function.",
+      "reference": "We should create the monitor after the `if (!primary_drm)` check."
+    },
+    {
+      "prediction": "I think this should be `yield 'qutebrowser.browser.tab.TabData._initializing',`",
+      "reference": "Can be removed now that slots are used"
+    },
+    {
+      "prediction": "I'm not sure if we want to change this or not. I think it's fine to leave this as it is for now. We can change it in the future if we decide we want to change this.",
+      "reference": "The description should match the test"
+    },
+    {
+      "prediction": "I'm not sure if this should be done in `xwayland_start_server()` or `xwayland_start_display()`. I think it should be done in `xwayland_start_server()`, but I'm not sure if it should be done in `xwayland_start_display()`.",
+      "reference": "rootston needs to unset this now"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `blockBuffer` struct itself.",
+      "reference": "why do you rename this function?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this helper. I think it should go in `app/helpers/mailer_helper.rb` since it's only used in that file.",
+      "reference": "since we are no longer using these numbered icons can we remove them from source control?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the `RestProducer` implementation.",
+      "reference": "both edge and normal need to set this, so set it to be default action?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `WriteToXml` method of the `JObject` class.",
+      "reference": "Is doing a straight string compare of the files the right way to determine this? I guess if you expect to be the only one writing this file it could work but it does seem like it might be a large string compare and if we are doing this hundreds of times that might cause some memory issues."
+    },
+    {
+      "prediction": "I think this should be `#if defined(__linux__) && __linux__` instead of `if (!system(buf)) {}`",
+      "reference": "What's the purpose of this change? It basically changes nothing until we want to put smth between curly braces"
+    },
+    {
+      "prediction": "I think this should be `dev->type == WLR_INPUT_DEVICE_TABLET_TOOL || dev->type == WLR_INPUT_DEVICE_TABLET_UI`",
+      "reference": "Hmm, why is this needed? `wlr_input_device_destroy` should destroy the tablet tool."
+    },
+    {
+      "prediction": "I don't think this method is used anywhere.",
+      "reference": "I think it's better to throw `UnsupportedOperationException`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `createElementWithAttributes` since it's only used there.",
+      "reference": "`props.children` is not always guaranteed to be an array. When only one child is passed this will break. We can use `toChildArray()` to turn it into an array :tada:"
+    },
+    {
+      "prediction": "I think this should be `--mode=%s' % buildStyle if buildStyle == 'Debug' else ``configureCmd += ' --mode=release' configureCmd += ' --mode=Debug' configureCmd += ' --builddir=%s' % buildDir",
+      "reference": "Please just remove this line. We have the git history if we want the old version."
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in this test.",
+      "reference": "Can we put some of the code on a new line to make it more readable. The max line length was 120 now, right?"
+    },
+    {
+      "prediction": "I think this should be `v := runtimevar.OpenVariable(ctx, \"etcd://myvarname?client=my.etcd.server:9999\")` instead of `v, err := runtimevar.OpenVariable(ctx, \"etcd://myvarname?client=my.etcd.server:9999\")`",
+      "reference": "Remove the handling here as well."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "Do we want to make this configurable?"
+    },
+    {
+      "prediction": "I think this should be `throw new IllegalArgumentException(message, argName, number);` instead of `throw new IllegalArgumentException(message, argName, number);`",
+      "reference": "Prefer adding a second `positive(String, double)` that delegates down to this three-param version. Using `null` in code is generally Not A Great Idea, and it looks ugly."
+    },
+    {
+      "prediction": "I think this should be `client_data` instead of `client_data_attributes`",
+      "reference": "Is this going to break things for 18F proposals, or will these fields just be ignored?"
+    },
+    {
+      "prediction": "I think it would be better to say something like \"If the provided key is none of those, the modifier key cannot be released.\"",
+      "reference": "Keys.COMMAND seems to be an alias to Keys.META. That isn't mentioned?"
+    },
+    {
+      "prediction": "I think this should be `json:\"caCertificate,omitemptuy,omitempty\"`",
+      "reference": "What's the reason for moving this to config from status?"
+    },
+    {
+      "prediction": "I think this should be `return suppress_file_handler.write_to_suppress_file(self.suppress_file, bug_id, file_name, comment)`",
+      "reference": "Shouldn't we re validate/update the in memory suppress data here?"
+    },
+    {
+      "prediction": "Why do we need to change this?",
+      "reference": "curious why the renumbering is required here and elsewhere?"
+    },
+    {
+      "prediction": "nit: I think we can simplify this to `if aws.CredentialsAssumeRole != nil && aws.CredentialsSecretRef.Name != \"\" {`",
+      "reference": "s/must specify secrets/must specify secrets or Role info/"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put Blazer, but I don't see a better place.",
+      "reference": "let's leave this as a dev-only feature, and copy prod db to local env when needed. that keeps blazer security issues to a minimum."
+    },
+    {
+      "prediction": "I think this should be `AlwaysOnSampler` since that's what we call it in the rest of our codebase.",
+      "reference": "We'll be renaming ActivitySampler to Sampler anyway, so this change will be non-required. Prefer to avoid changes here to avoid merge conflict with my PR doing the rename."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. If the rule value is empty, we don't want to replace it, right?",
+      "reference": "This applies only in the \"set\" headers case, but what about \"adding\" a header (field prefixed with `+`)?"
+    },
+    {
+      "prediction": "I don't think we should deprecate this class.",
+      "reference": "I'm not sure we should deprecate this. I kinda like the idea, that we would provide all exceptions (if we throw any) with a common super type. Of course, the exception should be not a checked exception like this one, but rather a runtime exception. We maybe need additionally an internal exception that we would convert into such a public API exception. But that's out of scope of this PR - that's the question of exception handling."
+    },
+    {
+      "prediction": "I think this should be `TabbedBrowserTab` instead of `TabbedBrowser`.",
+      "reference": "You'll need to adjust the docstring :wink:"
+    },
+    {
+      "prediction": "I'm not sure that this is the right way to do it. The problem is that we don't know the process id of the writer at this point, so we can't guarantee that the process id of the writer is the same as the process id of the reader. I think we need to do something like this: 1. Get the process id of the writer 2. Get the process id of the reader 3. Check that the process id of the writer is the same as the process id of the reader. If they are the same, then we can start the reader. Otherwise, we can't start the reader.",
+      "reference": "Add a TODO explaining why this is being done"
+    },
+    {
+      "prediction": "`VarDelete` and `VarRename` are not used in this file, so remove them.",
+      "reference": "This file still uses Blockly.VariableModel..."
+    },
+    {
+      "prediction": "I think this should be `bool` and not `string`",
+      "reference": "Small nit, in the PR description its `publicAccess` but here its `public`. Guessing the preferred naming is public?"
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. It's a bit confusing because we have a `Decoder` type, but also a `Decode` function. What do you think about `DecoderFunc` or `DecodeFunc`?",
+      "reference": "Maybe `driver.Watcher` -> provider (2x)? This is the concrete type, this user doesn't really know anything about the driver."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in `DocValuesWriter` and `DocValuesReader` instead of `DocValuesWriter` and `Sorter.DocMap`.",
+      "reference": "can we remove this since `getDocValues` already returns an iterator? (we might need to do `T extends DocIdSetIterator` above)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in the `Encoding` module, under the `utf_8` and `binascii` classes.",
+      "reference": "Don't check this in. This is already implemented in core_extensions/ruby/string.rb"
+    },
+    {
+      "prediction": "I'm not sure this is the right change. I think we need to change the `hash` property to `hashBuffer` or `hashValue` or something like that.",
+      "reference": "This is used for both `id`, and `hash` we may not want these both to be enumerable."
+    },
+    {
+      "prediction": "`dup` is expensive, so I don't think it's worth it to do it here.",
+      "reference": "Is this necessary? I would expect YARD to treat it as public anyway... (Don't hold off merging on this...I'm mostly just curious)."
+    },
+    {
+      "prediction": "I think this should be `latestVersion` instead of `ai.Version`",
+      "reference": "@wxing1292 to double check this is the right version to use?"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I would put it in `IntSetTrait`.",
+      "reference": "Shouldn't there be another case for when both min and max are specified?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this file.",
+      "reference": "Don't miss this one `2: I102 copyright year is outdated, expected 2014 but got 2013`. Thanks"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that `AppendTail` doesn't take into account the case where `tail_` or `blockPt_` is `nullptr`.",
+      "reference": "the tail_ maybe nullptr?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should go in `syncer.go` since it's only used there.",
+      "reference": "The PR description mentions that `The controller code is extended to react to \"add\" events`, but I don't see any additional cases added here beyond the existing Create and Delete... is there something I'm missing?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "Does it matter that #include changed to #include? ? I think the change makes sense since it aligns with the method name on Hash but unsure if anything called #include that would break with the change."
+    },
+    {
+      "prediction": "I think this should be `time.Hour` instead of `time.Second`.",
+      "reference": "LGTM, curious though why exactly this is in place, and are there any possible issues from not waiting long enough?"
+    },
+    {
+      "prediction": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "?: is unnecessary because we already force preserve_hosts to be a string with to_s."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `NewSessionQueuer` class itself, since it's only used by the NewSessionQueuer.",
+      "reference": "should it be `newsessionqueue` or `newsessionqueuer`? In case we'd like to be consistent"
+    },
+    {
+      "prediction": "I'm not sure this is correct. I think it should be `warnings[2].includes('MyLazyLoaded')`",
+      "reference": "Since lazy is re-rendered when mounting the fallback, these checks get triggered an additional time."
+    },
+    {
+      "prediction": "I think this should be `def is_content_editable(self) -> bool`.",
+      "reference": "There is a `is_content_editable()` method just above this, what's the difference?"
+    },
+    {
+      "prediction": "I'm not sure if we want to deprecate the `ServiceMonitorSelector` or not. If we want to deprecate it, I think we should also deprecate the `PodMetadata` and `ServiceMonitorSelector` fields in the `PodSpec` and `ServiceMonitorSelector` fields in the `ServiceMonitor` structs.",
+      "reference": "Hmm, I'm not sure this is exactly what we want to get across. The behavior of leaving both values unset allowing the entire config to be managed is what is indeed deprecated. It's just that not the entire field is deprecated, just that one behavior."
+    },
+    {
+      "prediction": "(style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (style) 'bas' may be misspelled - perhaps 'bas'? (",
+      "reference": "Given the slightly vague discussion in basename(3) about POSIX basename (modifies arg) vs GNU (doesn't), I always assumed it was advisable to pass a string copy However, I guess you'd get a \"discarding const\" warning promoted to error by our build system here if you were getting the POSIX implementation. Hah! Cool, I probably can go remove some memory copies in other parts of the code where those conditions exist."
+    },
+    {
+      "prediction": "I think we should use `Date.now() + MINUTE.toMillis(10)` instead of `DateTime.now().minusMinutes(10).toDate()`",
+      "reference": "Do you know somehow that this was the culprit? Even 1 minute is a lot, so I would expect the bug to lie somewhere else."
+    },
+    {
+      "prediction": "I think this should be `LOG(WARNING) << \"put failed, ret \" << code;`",
+      "reference": "what to do if some failed? just logging?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "since these thoughts are all Jack Handey's the generator should probably be called `jack_handey` to reflect that."
+    },
+    {
+      "prediction": "I think this should be `locationName:\"environmentFiles\" type:\"list\"` and not `locationName:\"environment\" type:\"map\"`",
+      "reference": "this file is autogenerated. you should edit `model/api/api-2.json` and then go generate this file. otherwise the next people generating api.go will remove the changes you added here"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this controller. I think it should be in `users/progress_bars_controller.rb`",
+      "reference": "Extra empty line detected at class body beginning."
+    },
+    {
+      "prediction": "I don't think we need to synchronize on exFlowKey here, since it's not being used anywhere outside of this method.",
+      "reference": "This is smart but hacky! I would probably prefer an alternate solution that would be more obvious to read/understand."
+    },
+    {
+      "prediction": "I think it would be better to call this `UpdateTableCapacityStatus` to make it clear that it returns the status of the update.",
+      "reference": "i would not return a pointer for status here which is an int"
+    },
+    {
+      "prediction": "I think it would be better to put them in alphabetical order. Also, I think it would be better to put them in the same order as they are defined in the code.",
+      "reference": "Those functions are not currently wrapped, so I don't think it is necessary to exclude them (at least for now)."
+    },
+    {
+      "prediction": "I'm not sure if this is the best error message to display, but it's the only one I could come up with that gave me a passing grade on my last test.",
+      "reference": "Should we specify the name of the folder?"
+    },
+    {
+      "prediction": "why do you need to import rdqueries and rdmolops here?",
+      "reference": "it's not wrong, but you don't technical need `rdmolops` here since it's imported as part of `Chem`"
+    },
+    {
+      "prediction": "I think this should be `pxz.wait(timeout=10)`",
+      "reference": "I think the absence of this might have been the cause of the dump file corruption. We didn't wait for the pxz command to exit, leading to a race condition between the cp and this, leading to corrupted files in some places. I came across this because the hashes created and printed were different from the hashes of the actual files in dev also."
+    },
+    {
+      "prediction": "I think this should be `*logging.Level` since it can be `nil`",
+      "reference": "Out of the scope."
+    },
+    {
+      "prediction": "I think it would be better to use `this.$(columnIndex)` instead of `this.$container.find(`th span.columnSorting:eq(${columnIndex})``",
+      "reference": "Maybe `sortByColumnHeader` would be more precise? The plugin has a method with the same name and it could be confusing."
+    },
+    {
+      "prediction": "I think we can remove the `|| EMPTY_ARR` from the `oldParentVNode!=null &&` and `oldParentVNode != EMPTY_OBJ` branches.",
+      "reference": "@developit I vaguely remember you mentioning something about some JS engines having an escape analysis feature that makes objects (and arrays?) locally declared that never leave the function essentially free from a GC perspective or something. Do you think we could benefit from that here? Or should we leave this as it is?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the same place as the `if op.IsBlockUpdate() {` check above.",
+      "reference": "Remind me why this is needed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `QueryVisitor`.",
+      "reference": "same as above, can this be null now?"
+    },
+    {
+      "prediction": "I think we should use `Long` instead of `int` here.",
+      "reference": "This doesn't make sense to me. Should `cv.lengths` be an integer array instead of longs? It looks like they should always be integers since the values are assigned from `ArrayData#numElements()`. That's very likely an int."
+    },
+    {
+      "prediction": "I think this should be `precursor = double.Parse(id.Substring(MsDataFileImpl.PREFIX_PRECURSOR.Length), CultureInfo.InvariantCulture);`",
+      "reference": "str.Substring(3).Split(' ')[0] would also work and is more concise."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but I don't have a better idea. The problem is that the `allow_local_peers` config option is set to `false` by default, and the `preconfigured_peer_a` flag is set to `false` by default. This means that the `send_keepalive` function will never be called, so the `keepalive` function will never be called, and the `send_keepalive` function will never be called, and so on and so forth.",
+      "reference": "Did you mean to negate the not_a_peer check?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It's for the DBeaver project, not DBeaver itself.",
+      "reference": "Copyright (C) 2010-2021 DBeaver Corp and others"
+    },
+    {
+      "prediction": "I think this line should be removed.",
+      "reference": "This seems like an unrelated change"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think this should be done in the `init` method of the `linuxStandardInit` struct.",
+      "reference": "Just to double-check -- are you sure this needs to be done *before* we create a new session? (Is `SetKeyLabel` setting what the label will be for all future keys or the label for the current key?)"
+    },
+    {
+      "prediction": "Is this needed? I don't see it being used.",
+      "reference": "I thought we were removing the FAIL policy altogether...."
+    },
+    {
+      "prediction": "I think this should be `--outfile`, not `--outfile`.",
+      "reference": "Could you elaborate on why you made this change?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "before the fix, does this sentence make service crash? My point is can you re-produce the problem. I'm not sure the bug is due to non-asiic code"
+    },
+    {
+      "prediction": "I think this is a good idea, but I think it should be done in a separate PR, as it's not related to the other changes.",
+      "reference": "This comparison looks a bit strange, we should first check existence of key \"IsApiToken\" and then check if it is equal to true. ~~Besdies, s/IsApiToken/IsAuthedByToken/~~"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this.",
+      "reference": "This might be working as is, but I'm a bit thrown by the || and then && and how things get evaluated. Testing real quick on the go playground, true || false && false seems to either evaluate the && first, or start on the right side. Anyhow could you group with braces, it looks to me like it should be ( a || b) && c in this case?"
+    },
+    {
+      "prediction": "I think this is the wrong place for this. It should be in `pkg/crypto/sm2/main.go`",
+      "reference": "add an entry in makefile to generate binary with diff name?"
+    },
+    {
+      "prediction": "I think this should be `validation.IsNil` instead of `validation.NewError(\"one of 'routingKey' or 'serviceKey' is required\")`",
+      "reference": "hmm so we didn't validate the HTTP config for all receivers?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should go in `SpanContext` since it's only used by `SpanContext` and `SpanContext` is the base class for all propagating contexts.",
+      "reference": "ITextFormatActivity had this, can I add to have the same effect?"
+    },
+    {
+      "prediction": "I think this file should be renamed to `ecs.go`",
+      "reference": "not blocking: can we remove extra lines here?"
+    },
+    {
+      "prediction": "This should be `if (typeof(data) === 'string' || data === 'random')`",
+      "reference": "This just feels weird... can we use some kind of constant here? something like: `if (_.isUndefined(data) || data === PrivateKey.Random)`"
+    },
+    {
+      "prediction": "I think it would be better to use `isObject` instead of checking if the class is defined.",
+      "reference": "if does not have a class I guess the exception should be thrown because we cannot recognize the type based on it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this TODO. I think it should be in the same place as the other TODOs.",
+      "reference": "LOL: use your GitHub username?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. I think it should be in the key class itself.",
+      "reference": "any tests for nextPrefix() ?"
+    },
+    {
+      "prediction": "This should be `Optional.empty()`.",
+      "reference": "In general I'm saying \"if there is no common type I'm just going to skip some stuff and fail silently.\""
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "did we move the eventing logging here? Probably the eventing logging was removing and they are now using the pkg logging in eventing. If that is the case, we should do the same here"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `WithRouteTag` since it's used by all handlers, not just route tags.",
+      "reference": "I think that `Handler` should preserve the `operation` member and the span formatter should receive the operation name string as a parameter too, otherwise the `operation` parameter in the `NewHandler` function becomes useless if we pass a custom span formatter. Also, shouldn't it be called `spanNameFormatter`?"
+    },
+    {
+      "prediction": "I think this should be `IPField(\"sourceip\", \"127.0.0.1\")` instead of `IPField(\"sourceip\", None)`.",
+      "reference": "Can you merge #466? That would make the current `None` default value working and more relevant that `\"127.0.0.1\"`."
+    },
+    {
+      "prediction": "I think this should be `customCategory|predefinedCategory` instead of `predefinedCategory|customCategory`",
+      "reference": "is reserved filed name the only possible reason this could fail?"
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `true`.",
+      "reference": "Would a KBFS_DISABLE_JOURNAL make more sense?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add this warning. I think it should be in `DropSpaceExecutor::dropSpace()`.",
+      "reference": "BTW. What's the message meaning?..."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this test. I think it should go in `spec/mongoid/extensions/object/numeric_spec.rb`",
+      "reference": "this should say false"
+    },
+    {
+      "prediction": "I think this file should be removed.",
+      "reference": "for my understanding - are this line and the next line both necessary? they seem to contain duplicate information. same for the other test files"
+    },
+    {
+      "prediction": "I think this should be `qurl_from_user_input(config.val.url.searchengines[term], None, None, None, None, None, None)`",
+      "reference": "No need for the `.keys()`, iterating over a dictionary gives you its keys (and thus you can also do `key in some_dict`). With that, it also fits on one line :wink:"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "If we want to avoid the user calling constructor and destructor, we should add `= delete` to their declarations. If we just want to avoid writing the default behavior, we should add `= default` to the declarations. I'm more in favor of the second option to avoid an API break."
+    },
+    {
+      "prediction": "I think this should be renamed to `Observation` instead of `Materialization`.",
+      "reference": "My preference here is to keep a stricter hierarchy. We should have a mixin or something that is an AssetEvent that Observation and Materialization can both inherit from. That way we can check the type in the frontend if we need to."
+    },
+    {
+      "prediction": "I think this should be `sorted(spOutput, key=lambda x: x == 0)`",
+      "reference": "Can also cast them as `set`s and compare them."
+    },
+    {
+      "prediction": "I think this file should be in the `trigger` package.",
+      "reference": "Still TODO I guess?"
+    },
+    {
+      "prediction": "Is this needed? I don't see it being used anywhere.",
+      "reference": "Another cleanup. There are two additional `using namespace std::chrono;` under this one. Should also remove std::chrono:: from the full file. Please do this on a single commit."
+    },
+    {
+      "prediction": "I think this should be `false, false, false, false` instead of `false, false, false, false, false`",
+      "reference": "Looking at this list of boolean flags I wonder if we should create enums for all of these options, so that the code becomes much more readable"
+    },
+    {
+      "prediction": "Is this needed?",
+      "reference": "Is this necessary?"
+    },
+    {
+      "prediction": "Why do we need this? I don't see it being used anywhere.",
+      "reference": "How about this NS inside query NS?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to load the account. I think it should be in `LoadAccount`",
+      "reference": "in line 79, sm also can be stateReader"
+    },
+    {
+      "prediction": "I think we can remove the `std::vector<PartitionID>` from the `partsBySpace` map and just use `partsOnHost_` directly.",
+      "reference": "maybe can not return a map directly, it will be modified by inner thread."
+    },
+    {
+      "prediction": "`diff_res` is a dict, so `test_res` should be a dict as well.",
+      "reference": "What is the actual change here, why is this test change needed? Now the diff will send back more data?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. The RGroupDecomposition class is only used by the RGroupFactory, which is only called from the RGroupFactory::add function. The RGroupFactory::add function is only called from the RGroupFactory::addHs function, which is only called from the RGroupFactory::add function. I think this should be moved to the RGroupFactory::addHs function.",
+      "reference": "Might be better to use a tag here. I for one have used 1000 isotopes as a tag in the past..."
+    },
+    {
+      "prediction": "I'm not sure if this test is still relevant?",
+      "reference": "This tests seems it was broken from the start :joy:. It \"worked\" because the old sample chance was small enough that this have always be off. Thanks for fixing this. Could you also change the strings on this test? `\"Recording off\"` and `\"new span is recording events\"`"
+    },
+    {
+      "prediction": "I think it would be better to have this method return `true` in this case, so that it's clear that there is no configuration to be produced.",
+      "reference": "This is not actually the case, you need to modify `toConfiguration` below to prevent a configuration from being produced"
+    },
+    {
+      "prediction": "I think this should be `a.c.Log.Debug(\"Private key recovered, but no SVID found\")`",
+      "reference": "I'm worried about we hide some important log here"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep it as it was before. The reason is that we don't have a way to get the current time, so it's hard to know what is the \"latest\" time when we want to get the results.",
+      "reference": "Python 3 will throw an exception for inf, but not providing a timeout will result in no timeout exception."
+    },
+    {
+      "prediction": "I don't think we need to expose these endpoints.",
+      "reference": "As this method changes the state, it should be `POST`. Also, maybe to rename it to `createTag`?"
+    },
+    {
+      "prediction": "Why not just return `model.StageStatus_STAGE_SUCCESS`?",
+      "reference": "`ctx` is unused in rollbackTraffic"
+    },
+    {
+      "prediction": "I think this should be `public const string guidVSPackageString = \"b97edb99-282e-444c-8f53-7de237f2ec5e\";`",
+      "reference": "Roslyn conventions have const in PascalCase."
+    },
+    {
+      "prediction": "I don't think we need to check for null here. If the client port is not specified, then it should be 2181.",
+      "reference": "This is not bullet proof if e.g. user has `clientPort=1234` in `zoo.cfg` and in zkHost connection string. Then we'll add a warning that dynamic config differs from zkHost, which is not entirely true since we just lack the port part. We have no way from client to read the `clientPort` from server except from connecting to the server with 4LW *ont the clientPort* which is a chicken and egg. This hack will make the comparison work for default port, which is a compromise."
+    },
+    {
+      "prediction": "I'm not sure if we want to use `MessagePack` or `MessagePack.Formatters` here. I think we should use `MessagePack` since we're using `MessagePackExtensionMethods` and `MessagePack.Formatters` is used in `MessagePackExtensionMethods`.",
+      "reference": "We can probably delete this entire file. It's not used now and we'll (probably) write a custom serializer before we ever switch to MessagePack 2.1."
+    },
+    {
+      "prediction": "I think this should be `--no-identity` instead of `--o-identity` to be consistent with the rest of the CLI.",
+      "reference": "Aren't you duplicating this option?"
+    },
+    {
+      "prediction": "I think this should be `strings.HasSuffix(filepath, \".build_defs\")`",
+      "reference": "I'm not sure we should be doing this based on the extension? Calling them `.build_defs` is just a convention"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pandas/_utils.py`, since it's only used in `pandas/_utils.py`.",
+      "reference": "Should remove in `_MissingPandasLikeMultiIndex:` too"
+    },
+    {
+      "prediction": "I think we can remove this field since it's not used anywhere.",
+      "reference": "We should scope this down to the bits that we're actually using, otherwise the API is going to be problematic, as it includes references to NAT gateways and public and/or private subnets. Copying the types to be more local to the task in hand is fine."
+    },
+    {
+      "prediction": "I think this should be `+optional` as well.",
+      "reference": "Is it is a hard requirement for this field to be 'cn-northwest-1' when wanting to interact with AWS China? It appears that putting in 'cn-north-1' would also result in using the alternative API endpoint (with the region overridden to use 'cn-northwest-1' for the created AWS client)."
+    },
+    {
+      "prediction": "I think this test is flaky because of the `await Task.Delay(1000);` at the end of the test.",
+      "reference": "Stephen already fixed this one. Only the OverSize test is flaky now right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this, but it's the only one I could think of.",
+      "reference": "if the `activity_mailer` has a layout, should we include the footer in that? Realize there may also be conflicts with work @rememberlenny is working on..."
+    },
+    {
+      "prediction": "I think this should be `Puppet::Util` instead of `Puppet[:disable_i18n]`",
+      "reference": "This makes me nervous, mostly because I don't know very much about it. Will users run into the gettext error in the wild? I can't reproduce it locally, but it's hard to be 100% sure. Is disabling gettext in Pal an option?"
+    },
+    {
+      "prediction": "I think this should be `List<ExecutableFlow> fetchRecentlyFinishedFlows(long lifeTimeMs, Predicate<ExecutableFlow> predicate) throws ExecutorManagerException`",
+      "reference": "Is Java smart enough to know the generic type here? Never knew this."
+    },
+    {
+      "prediction": "I think this should be `BackupRepository.newInstance(BackupRepository.class, repo)`",
+      "reference": "no thought given to reload"
+    },
+    {
+      "prediction": "I think this method should be called `isMetricAttributeEnabled` instead of `isMetricAttributeDisabled`",
+      "reference": "Please, don't remove the `getDisabledMetricAttributes` method. All changes must be backward-compatible."
+    },
+    {
+      "prediction": "I think this should be `err.message.match(/Failed to kill on some hosts/)`, not `err.code === 11601`.",
+      "reference": "Is this left over from debugging?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to close the client. The client should be closed in the `done` callback of the `done()` function.",
+      "reference": "I've fallen for this before myself :) We _do_ call `close` inside of `done` on L65"
+    },
+    {
+      "prediction": "I think this should be `uplink.Config{UserAgent: \"rclone\"}`",
+      "reference": "That looks fine! You could use `\"rclone/\"+fs.Version` too if you wanted. BTW does tardigrade use http under the hood? If you were using rclone's http Client then you'd get a User-Agent and you'd also get support for `-vv --dump bodies` and other nice things."
+    },
+    {
+      "prediction": "I think you can remove the `min=1e-12` from `neg_prob = 1 - prob`",
+      "reference": "Adding one line `prob = prob.clamp(min=EPS, max=1-EPS)` already works."
+    },
+    {
+      "prediction": "I think this should be `ConcurrentQueue` instead of `ConcurrentQueue`",
+      "reference": "another option would be to create a `CreateRequestQueue(bool lockingQeueue)`, and set it only in `InitializeTarget`, isn't? The would prefer having the creation and \"business\" rules for the creation in one region, I think `InitializeTarget` would be the best place."
+    },
+    {
+      "prediction": "I think this should be `parent_pending_link` instead of `sibling->parent_pending_link`.",
+      "reference": "Wait, I think the `subsurface_handle_place_above` code was correct before. `place_above` means \"place it above in rendering order\" right?"
+    },
+    {
+      "prediction": "I think this should be `1. hey\\n2. there`",
+      "reference": "why did this change?"
+    },
+    {
+      "prediction": "I think this is the only test that needs to be changed to use `MakeHandlePrivate`",
+      "reference": "Do you think we should have some tests here for `tlf.SingleTeam` too?"
+    },
+    {
+      "prediction": "I think this should be `import \"github.com/influxdata/flux/table\"`",
+      "reference": "No problem with this but might as well revert this file since nothing else changed."
+    },
+    {
+      "prediction": "I think we can remove the `IsIntegrationEnabled` method altogether, since we don't need to distinguish between http and https.",
+      "reference": "In my changes, I did not create a new `AspNetScopeManager`. I'm not familiar with it enough to know if this is a valid change or not"
+    },
+    {
+      "prediction": "I think this should be `throw new Error` instead of `throw new Error(message);`",
+      "reference": "Not cool with this. I think we talked about it in past. This is a side effect. If we want to introduce the side effect in our functions like `inRange` (which I am for) the side effect (error) should originate in that function and not in some internal `throwError` function. Every stacktract will start at line `2` of `trowError.js`. Been there, done that and falled back to throwing the error from the place where it should be thrown."
+    },
+    {
+      "prediction": "I don't think we should add new methods to an existing interface. I think we should just remove the old methods and add the new ones.",
+      "reference": "We don't need this additional method."
+    },
+    {
+      "prediction": "I don't think this is the right test to cover the new `loadUser` method. I think it would be more appropriate to test that the `loadUser` method is called with the correct `OAuth2UserRequest`.",
+      "reference": "The purpose of this test is to ensure that the `OAuth2UserRequest` passed into `loadUser` is **not** null else throw `IllegalArgumentException`. Changing the `List` of `OAuth2UserService` mocks to `DefaultOAuth2UserService` doesn't really apply to what is being tested here. Please revert this. Thank you."
+    },
+    {
+      "prediction": "Why do we need to make this change?",
+      "reference": "nit: rename to nativeStk? proto might lead to think protobuf"
+    },
+    {
+      "prediction": "I don't think this is a good name for this. It's not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really a query id, it's a shard id. It's also not really",
+      "reference": "Why not `queryUUID` (and reference the same constant as in other places)?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this error message. I think it should go in the `agent.go` file, since that's where the registration entry is created.",
+      "reference": "This is also used to log, of which our convention is leading uppercase..."
+    },
+    {
+      "prediction": "I think this should be `less_precise=True`",
+      "reference": "What about we just name it `check_exact`?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good test. I think it would be better to have a test that creates a span with mutually exclusive endpoints in binary annotations and annotations and verifies that the service names are in the correct order.",
+      "reference": "before, a binary annotation had the same service name as a regular annotation, masking the bug where we weren't querying the latter"
+    },
+    {
+      "prediction": "I think this should be a constant in the `ex` module.",
+      "reference": "Where does the `7` come from?"
+    },
+    {
+      "prediction": "I think this should be `value={ value }` and not `value={ setValues.country }`",
+      "reference": "I don't know how we haven't got bugs from some of these! As here this would have had a stale `value`"
+    },
+    {
+      "prediction": "I think we should add a warning here saying that we don't support this property.",
+      "reference": "One nit here I forgot about this before but we usually structure error messages as \"Cannot X because Y. Then the recommendation goes here\" I would also recommend not using \"it\" in the message since it the pronoun is a bit ambiguous. \"to specify write sort-order\" may be clearer"
+    },
+    {
+      "prediction": "nit: can we move this to the top of the file?",
+      "reference": "I'm surprised this worked at all! It was importing the default export but `eslint-plugin-import` warned me `warning Using exported name 'initialState' as identifier for default export import/no-named-as-default`"
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "This was used in `app/views/video_tutorials/_video_tutorial_details.html.erb` how are we handling that now?"
+    },
+    {
+      "prediction": "nit: I think we can drop the `:assumed-role/` part of this regex.",
+      "reference": "Is `\\S+` really the right thing? Anybody know what kind of encoding this is? Perhaps if we know it doesn't have commas, we can just use `[^,]+`. Also, parens around the comma are unnecessary, since we're not using the capture group."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `tests/unit/test_numpy.py`",
+      "reference": "This block of changes looks like you removed tests without replacing them anywhere. Did I miss something?"
+    },
+    {
+      "prediction": "I think it would be better to add a `GetPropsAsDict` method that takes a `const char *` argument for the format string, and then call `GetPropsAsDict` with a `const std::string &` argument. This way you don't have to pass `\"\"` as an argument to `GetPropsAsDict` if you don't want to use the default format string.",
+      "reference": "n.b. private and computed values are now exposed to the API. They were hidden/not exposed before."
+    },
+    {
+      "prediction": "`remove_target` is not a good name. `remove` is a verb, `target` is a noun. It should be `remove_target` or `remove_link_target` or something like that.",
+      "reference": "Nitpick: Please add a period after `link` :wink:"
+    },
+    {
+      "prediction": "I think we should log the error here as well.",
+      "reference": "we got no chance that `err` is not nil here, need to revisit how `NewUpstreamController` is defined."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. It's not a command, it's a property. The heap snapshot is a property of the session, so it should be under the `/session/:sessionId/heap_snapshot` path.",
+      "reference": "Instead of building in routing for a browser-specific command, could you refactor the HttpCommandExecutor to allow arbitrary commands to be registered?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this test. I think it would be better placed in `spec/spec_helper.rb` since that is the file that has all of the spec files for a particular module.",
+      "reference": "It'd be nice to assert the right deprecation is being raised, just add a third argument of `/\"its\" method/`"
+    },
+    {
+      "prediction": "I think we should rename this to `WRITE_LOCATION_PROVIDER_IMPL` to be consistent with `OBJECT_STORE_LOCATION_PROVIDER_IMPL` and `OBJECT_STORE_ENABLED`.",
+      "reference": "Do we want to add this to the docs?"
+    },
+    {
+      "prediction": "I think this should be `code.String()` instead of `code.`",
+      "reference": "This isn't really related but makes debugging test failures much easier since you see the string representation of the code instead of the integer value."
+    },
+    {
+      "prediction": "I think this should be `if (var_type.IsValid())` instead of `if (var_type.IsValid() && var_type.GetByteSize(exe_ctx.GetBestExecutionContextScope()) == 0)`",
+      "reference": "Because of this new early exit."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be in `Selenium::WebDriver::Chrome::Bridge` since Chrome is the only supported browser.",
+      "reference": "Maybe call it `WireBridge`?"
+    },
+    {
+      "prediction": "I think it would be better to move this check to `PrivacyApiGroupJsonRpcMethods#createPrivacyMethod` instead of `PrivacyApiGroupJsonRpcMethods#createPrivacyMethodForMultiTenancy`.",
+      "reference": "can this be removed now?"
+    },
+    {
+      "prediction": "(style) space required after that ',' (ctx:VxV)",
+      "reference": "This is going to create a problem - we are already encountering issues with the length of the usock pathname on recent Linux kernels as the temp directory base has moved down to the var directory under a name created by a hash. So adding another field to it is going to add to the problem."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `apps/main.go` file.",
+      "reference": "What do you think about moving this to the `cloudformation` pkg?"
+    },
+    {
+      "prediction": "I think this should be `array` instead of `array|null`",
+      "reference": "Please make this private by default"
+    },
+    {
+      "prediction": "I don't think this is the right place for this logger. It should be in the engine impl.",
+      "reference": "We can add `@Override` annotations to all of the implementations of this method. I think this is a best practice for us, but it would be nicer in any case."
+    },
+    {
+      "prediction": "I think it would be better to have a separate interface for `Detector` and `Cache` and implement the other in the `Resolver` interface.",
+      "reference": "Most of uses of `RefreshAndGet` seems to be made only for `Refresh` part, result is ignored. We can simplify this method to assigning single responsibility to it - just `Refresh()`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `add_option` method of the `OptiKContainer`.",
+      "reference": "@Pierre-Sassoulas Are we sure this doesn't create problems? Without it for some reason we get an error on `accept-no-param-docs` being a duplicate error. I couldn't figure out why it did this. This solves the issue and passes the tests, but I wonder if this creates other issues.."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Does it make sense to check for \"Created\" here as well? Or should it return an error in such case?"
+    },
+    {
+      "prediction": "I think we can remove the `identifier` parameter from the `createTableTransaction` method, since it is not used anymore.",
+      "reference": "Did you mean to use `identifier.toString` here as well?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "interesting! didn't realize this was needed in RSpec 3"
+    },
+    {
+      "prediction": "nit: `ks.Series` is a bit of a mouthful, maybe `databricks.koalas.series` or `databricks.koalas.spark_series`?",
+      "reference": "This is to avoid circular reference"
+    },
+    {
+      "prediction": "I think this should be `self._notebook = False` instead of `_notebook = False`",
+      "reference": "As the bokeh renderer also has a ``load_nb`` classmethod (and notebooks are explicitly mentioned in that method name), I would consider making ``_notebook`` into a constant ``notebook_context`` class parameter. The docstring can say it is set to True if the ``load_nb`` classmethod has been called."
+    },
+    {
+      "prediction": "This should be renamed to `DropCollectionError` since it's a specific type of error that occurs when trying to drop a collection that doesn't exist.",
+      "reference": "Where is the actual command generated here? Shouldn't this be taking in `constructor(db, collectionName, options)` and then constructing the command off of that?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "Mirror the `OPTIONAL OUT` of fullpath"
+    },
+    {
+      "prediction": "I think it's better to use ConsensusV1 instead of ConsensusVNext.",
+      "reference": "Please merge with master and replace with with \"future\" version."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Nit: Change is not needed for this PR :slightly_smiling_face:"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it would be better placed in `CloudFoundry::Private::CapnpProtocol` since that's the only module that uses this method.",
+      "reference": "if we are returning `nil` from an `else` I think we can just remove the `else` (and this method will still return `nil`)"
+    },
+    {
+      "prediction": "Nit: I think it would be better to have a dedicated class for this test.",
+      "reference": "side note: this change is ok as of today because we have not exported the results for this new type to BenchView yet. After we to that the namespace, type name and method name should not be changed (they create a benchmark ID which is used in BenchView)."
+    },
+    {
+      "prediction": "I think this should be `return false;` instead of `return null;`",
+      "reference": "How exactly are you expecting the user to see issues with their Midas library? It seems like all error information is being swallowed and not clearly reported to the user. Even if the eventual result is to report that loading the file failed, it seems like the exception, in this case, might have more information about why."
+    },
+    {
+      "prediction": "I think this should be `wsc.connection.SetWriteDeadline(time.Now().Add(float64(wsc.config.WriteDeadline.Seconds())))`",
+      "reference": "Looks good, but seems `ReadMessage` hadn't used this Deadline in Underlying `WSConnection`?"
+    },
+    {
+      "prediction": "I think this should be `azkaban_project_dir_size_in_bytes.txt` instead of `_azkaban_project_dir_size_in_bytes_`",
+      "reference": "Make it \"___*\" three underscores to make it even less likely to have a collision? I assume it will be hard to change the file name once this change is deployed..."
+    },
+    {
+      "prediction": "I think this should be final.",
+      "reference": "We don't need this anymore after #2151"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this class. I think it should be in `NLog.Layouts`",
+      "reference": "this is this removed? I don't understand in the context of this PR"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You should be able to do `ip6 = inet_pton(socket AF_INET6, ip6)`",
+      "reference": "`isinstance(ip6, Net6)` instead"
+    },
+    {
+      "prediction": "I think this should be `ErrorOr<ResultCode, std::vector<Status>, bool>`",
+      "reference": "Do we really need the param here? If only partial results returned, we could return ResultCode::PARTIAL_RESULTS;"
+    },
+    {
+      "prediction": "I think this should be `tmpOutput := path.Join(tmpDir, fmt.Sprintf(core.TmpOutputFormat, output))`",
+      "reference": "this logic seems duplicated from the other new function. I think it should probably be a member function on `BuildTarget`"
+    },
+    {
+      "prediction": "I think this should be `assert.Equal(430, len(stream)-1)`",
+      "reference": "size reduce to 1/4 of using Gob"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `io` since it's an interface.",
+      "reference": "The Hash interface states that a call to `Write()` never returns an error. Does this also apply to the number of written bytes?"
+    },
+    {
+      "prediction": "I think this should be `%w(2 3 4 5 6 7 8 9 10 11).map(&:to_s).join(' ')` instead of `%(\"$(sw_vers -productVersion 2>/dev/null | cut -d . -f 2) -lt 12\").map(&:to_s).join(' ')`",
+      "reference": "In my tests, I found that `[[ \"\" -lt 12 ]]` evaluates to true, but `[[ -lt 12 ]]` is an error, which is why the subshell is wrapped in `\"`."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I don't see a validation on `published_on`, we generally shouldn't specify it in the base factory unless the model would be invalid without the model. Is there a different way we can handle this?"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for `ScoreMode.BOTH` instead of having a bunch of ifs.",
+      "reference": "maybe add a `isExhaustive()` method on the enum to avoid these large conditions?"
+    },
+    {
+      "prediction": "nit: I think we should rename this to `substitutionRegex` or `substitutionPattern` to be consistent with the rest of the codebase.",
+      "reference": "so those regexps were wrong previously? e.g. they would match `/v\\/foo`? If so, raise a bug and fix on the 1.0 branch."
+    },
+    {
+      "prediction": "I think it would be better to use `empty($params) && null === $source`",
+      "reference": "I think with MultiBackend we could have a slightly different logic: return true if there are no configured login targets ( = getLoginDrivers returns an empty array). MultiBackend could also check all configured login targets for loginIsHidden support and verify that at least one of the configured login targets allows login, but I wouldn't go that far without a use case. In a related note, I'm not quite sure if loginIsHidden works with ChoiceAuth at all, but that's a different issue."
+    },
+    {
+      "prediction": "I think this should be renamed to `binary_dump` or `binary_dump_file` to be consistent with the other functions.",
+      "reference": "Should we add error info for not supporting the compressed binary?"
+    },
+    {
+      "prediction": "I think this should be `clusterRegionGetter = utils.NewClusterRegionGetter()` and not `clusterRegionGetter = new utils.ClusterRegionGetter()`",
+      "reference": "I suggest injecting this into the reconciler. For this and the others. Have Wire generate it and push it into the controller creation."
+    },
+    {
+      "prediction": "I think this should be `EthereumJsonSerializer()` instead of `_serializer = JsonSerializer.Create(_settings);`",
+      "reference": "Can't we just make _serializer static? We use same settings every time. I would also put those fields on top of the class for readability."
+    },
+    {
+      "prediction": "I think this should be `f.websocket = True` instead of `f.metadata['websocket'] = True`",
+      "reference": "Can't we just use `metadata['websocket_flow']` to identify handshake flows and not add another attribute?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to format the output. It might be better to use `#` instead of `#-'` in order to make it clear that it's a single character.",
+      "reference": "Funny that we were wrapping this with string interpolation before..."
+    },
+    {
+      "prediction": "This should be `alias('model.id')`",
+      "reference": "I had an eslint error saying I must \"alias\" my model - so I copied this from controllers/site.js"
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. I think `MetricPointsAccessor` would be better.",
+      "reference": "unsure if the methodname can still be `GetMetricPoints()` as before..."
+    },
+    {
+      "prediction": "I think this should be `violation.get('rule_name', violation.get('rule_name', ''))`",
+      "reference": "This file contains the functional changes, the rest is for testing."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "I know you didn't do this, but there's no need for the `nethttp` alias and it was confusing to me - just remove it and s/nethttp/http/ everywhere (it's fine that the package here is http itself, I do the same thing in transport/grpc)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in `SmartStoreTest#setUp` and then call `store.getDatabase().query` after the test.",
+      "reference": "I felt that this was the most non-invasive way to fix the test, but this does expose some raw DB queries and knowledge about how SQLite works which may be a code smell. The alternative to getting this to pass is to perform enough writes to trigger a checkpoint, but that threshold is determined in the config stage and this test may not be able to know what that page threshold is set to. This could lead to inconsistent test runs which is why I elected to force the checkpoint regardless of the page threshold."
+    },
+    {
+      "prediction": "I don't think this is the right model version for GBDT. I think it should be `v1` or `v2`.",
+      "reference": "Is new model format backward compatible with current v2?"
+    },
+    {
+      "prediction": "I think this test should be moved to `spec/models/user_decorator_spec.rb`",
+      "reference": "should we perhaps write a spec that ensures we don't send a welcome email to a user on login when the user is not new?"
+    },
+    {
+      "prediction": "I think we should rename this to `ContainerStopTimeoutSeconds` to be consistent with the other config keys.",
+      "reference": "I think my preference would be to have the type be a `time.Duration` and use `time.ParseDuration` for parsing."
+    },
+    {
+      "prediction": "I don't think we need to change anything in this file.",
+      "reference": "JMX offers normal APIs for this. I don't think you want the glassfish one."
+    },
+    {
+      "prediction": "This should be `name: goss`",
+      "reference": "We should really leave a TODO or something or else we will forget them ..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be in `dm.define_dagstermill_solid`",
+      "reference": "This exposes what a bad name `dagster_type` was, but is it crazy to want this to just be `type` -- do we gain usability by being super-explicit that this is a `runtime_type`? If so, would it make sense to rename the `config_field` to be `config_type`?"
+    },
+    {
+      "prediction": "I think this should be `return self._isSelectionAnchoredAtStart` and not `return self._get_isSelectionAnchoredAtStart`",
+      "reference": "It might help here if you give a brief explanation of what you mean with a selection being anchored at the start."
+    },
+    {
+      "prediction": "I think this is the wrong place for this function. It should be in `pkg/transfer.go`",
+      "reference": "move cachedStates to handle function"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this setting. I think it should be in `settings.xml` under the `android` folder.",
+      "reference": "I don't think this can just be renamed; it's saying that privacyMode is a BooleanSetting, which it's not."
+    },
+    {
+      "prediction": "I think this should be `Thread.currentThread().getName() + \".\" + stage.name` instead of `Thread.currentThread().getId() + \".\" + stage.name`",
+      "reference": "what was the reason for this change?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be under `System.Diagnostics.Tracing`, but I'm not sure if that's the right place to put it.",
+      "reference": "Minor, but these two nullable and the logic around them could be replaced by a Lazy<Bool>."
+    },
+    {
+      "prediction": "I think this should be `k8s.io/kubernetes/pkg/apis/meta/v1` instead of `k8s.io/client-go/kubernetes`.",
+      "reference": "I think I would just pass in `len(peers)`, on the principle of minimum information."
+    },
+    {
+      "prediction": "I think this should be `read.object-storage.in-predicate-limit`",
+      "reference": "Spark's default value is 10 but it also rewrites IN as OR/EQUALS. We can do better that this because of the way we evaluate IN predicates. On datasets we tested this, we saw performance improvements on IN predicates with up to 200 elements (on sorted column). We may increase the default value a bit but I am very reluctant to make it more than 300 since our jobs were 20% slower with IN predicates that contained 300 elements. That being said, the optimal value still depends on a particular query and table."
+    },
+    {
+      "prediction": "I don't think this is a good idea. I'd rather have a `@author` tag that says who wrote it.",
+      "reference": "Please update copyright year."
+    },
+    {
+      "prediction": "I think this should be `static_cast<std::uint64_t>(number_of_node_based_nodes / TIMER_SEC(expansion)) / (max_edge_id + 1)`",
+      "reference": "Same here, could just be `std::setprecision`."
+    },
+    {
+      "prediction": "I think we can remove this now, too.",
+      "reference": "Do we still want the \"admins\" gate?"
+    },
+    {
+      "prediction": "I think we should remove the `# snippet-start` and `# snippet-end` tags here.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I think this should be `DisableAutoObservabilityMiddlewareEnabled bool` since it's a boolean.",
+      "reference": "nit: `it is assumed`"
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "I would appreciate it if you could refrain from using `stdbool.h`. We allow the header files of H2O to be included from C++ (which means that `bool` might be a C++ type), and therefore my preference is to not use `bool` in our code (but instead use `int` or `char` for the purpose) to avoid confusion."
+    },
+    {
+      "prediction": "I think this should be `manage[searchScope]` and not `manage[searchScope][0]=searchScope`",
+      "reference": "@danielbidala What happens if `searchScope` is not defined in `config_relation.yaml` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `MDServer` struct, since that's where the `RefreshAuthToken` method is.",
+      "reference": "Do we want maybe a 1min timeout?"
+    },
+    {
+      "prediction": "I think we can drop the `else` here, since it's the same as in the `else` above.",
+      "reference": "I guess code never reaches this line, you can remove `else` block and just `return a[0], a[1]`. Also, probably rename `a` to `auth` as well."
+    },
+    {
+      "prediction": "I think it would be better to have a constructor that takes a `LinkedHashMap` instead of a `LinkedHashMap` that takes a `SerializableByteBufferMap`",
+      "reference": "Doesn't this no-arg constructor need to be public for serialization to work?"
+    },
+    {
+      "prediction": "I think this should be `Reeds aangeheg` instead of `reeds aangeheg`.",
+      "reference": "Just use the word \"veld\" instead of \"gebied\" here"
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "This change (import org.shredzone.flatter4j.model.User;) does not seem relevant to this fix / commit."
+    },
+    {
+      "prediction": "I think this should be `hydratedBuild` and not `hydrated-by-hydrating-rpm`",
+      "reference": "This should be a Bool() rather than a String(). (See the other PR for an example)"
+    },
+    {
+      "prediction": "Shouldn't this be `1 <= len()`?",
+      "reference": "This should be a failure if we don't get the exact number of expected rules. You can use a constant if you don't want to update several lines any time you update the test rule strings."
+    },
+    {
+      "prediction": "I think you can remove the whitespace changes in this file as well.",
+      "reference": "We need to switch from BeamSync to FullSync when we download all the needed headers, blocks, receipts and state"
+    },
+    {
+      "prediction": "This should be `g_value_dup_object` instead of `g_value_get_object`.",
+      "reference": "Why you duplicate it? The underlaying transport should live as long as the multiplexed one. And must be destroyed after protocol is destroyed. Duplicating the transport may lead to object references hold and maybe memory freeing problems. I think this property must hold a reference to it and not a copy. The copy can lead to memory freeing problems."
+    },
+    {
+      "prediction": "Is there a reason why we don't want to use the `Status` field from the `{{.Status}}` struct?",
+      "reference": "`in.Status.Experiment.Phase`. we can omit `ChaosStatus`"
+    },
+    {
+      "prediction": "I think this file should be renamed to `testutil.go`",
+      "reference": "Just curious... why the alias here?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Seems like this should conflict with changes I made during merging of the 20.2 RESX file translation"
+    },
+    {
+      "prediction": "I think this should be `enable_optimize_collapse_project_rule`",
+      "reference": "Why disable this rule?"
+    },
+    {
+      "prediction": "I think this should be `editor[1:]` instead of `editor[1:]`",
+      "reference": "I think you don't need the `... if '{}' in arg else arg` part - if the arg doesn't contain `{}`, `arg.replace('{}', ...)` will return the unchanged string anyways."
+    },
+    {
+      "prediction": "I think this should be a constant in `google.cloud.security.inventory.pipelines`",
+      "reference": "nit: Should we make this an attribute of the LoadGroupMembersPIpeline class?"
+    },
+    {
+      "prediction": "I think this should be `SELECT name, system_suspended FROM principal WHERE system_suspended = ?;`",
+      "reference": "need to update this command to use the name field instead of principal_id"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be moved to `PrivacyApi`",
+      "reference": "nit: I can't help but think the privateTransactionHandler should be a base-class member ... every Priv Json RPC seems to need it..."
+    },
+    {
+      "prediction": "I'm not sure about this change. The problem is that we have a lot of code that uses the flag `FLAG_ACTIVITY_NEW_TASK`. I think it would be better to leave it as it was and just add a new flag `FLAG_ACTIVITY_REORDER_TASKS`.",
+      "reference": "Why does this start the main activity and not the preferences? With `overridePendingTransition(0, 0)`, this could instantly switch the theme without the user being disrupted"
+    },
+    {
+      "prediction": "I think this should be `public boolean isDocker() { return up; }` and not `public boolean isDocker() { return up; }`",
+      "reference": "Prefer a human-readable string rather than querying specific technologies. How would I indicate a session is running on BrowserStack? Or some custom thing?"
+    },
+    {
+      "prediction": "I think this should be `msrestazure>=0.2.1,<0.2.2` and not `msrestazure>=0.2.1`",
+      "reference": "It appears as though whoever checked in python changes didn't re-run regenerate:expected. I am modifying these files as a result of running that after a sync and build."
+    },
+    {
+      "prediction": "I think this should be `hasAnyResourceNameParameter( flatteningGroup.getFlattenedFieldConfigs().values().stream().anyMatch(FieldConfig::useResourceNameType)`",
+      "reference": "This was copied from JavaMethodViewGenerator; only the `public static` method modifiers were added."
+    },
+    {
+      "prediction": "I don't think we need this method, we can just use `getMaxPriorityFeePerGas` and `getMaxFeePerGas`",
+      "reference": "this is going to throw for frontier transactions post-london"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just remove the `useDebug(true);` statement from the `propel.php` file. The `useDebug(true);` statement should only be used for testing purposes.",
+      "reference": "The catch clause is not needed. The finally clause alone is enough."
+    },
+    {
+      "prediction": "I think this should be `public boolean isDocker() { return up; }` and not `public boolean isDocker() { return up; }`",
+      "reference": "Prefer a human-readable string rather than querying specific technologies. How would I indicate a session is running on BrowserStack? Or some custom thing?"
+    },
+    {
+      "prediction": "I'm not sure why this is the case, but I think it should be `isClosed = true;`",
+      "reference": "Changing the default value for isClosed from TRUE to FALSE can have other repercussions. We might miss throwing error. Can you please confirm this change."
+    },
+    {
+      "prediction": "I think this should be a `StopTask` event, not a `TaskStopped` event. The reason is that this is a task that was stopped by the user, not a task that was stopped by the service.",
+      "reference": "Please use more meaningful names than `te`here and in other places."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/plugin/plugin.go`",
+      "reference": "Perhaps this would be more idiomatic as `ConfigFor(interface{}) (*PluginConfig, bool)`?"
+    },
+    {
+      "prediction": "I think we can remove the `AWS_JSON_CONTENT_TYPE` constant now.",
+      "reference": "nit: We could import `APPLICATION_AMZ_JSON_1_1` from `constants.py` here."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface. I think it would be better to move it to `MvvmCross.Interfaces`.",
+      "reference": "Should this not inherit from IMvxMainThreadDispatcher?"
+    },
+    {
+      "prediction": "I think it would be better to move this to `roundCtx` itself, so that it can be accessed by all callers of `UpdateRound` without having to pass `roundCtx` as an argument.",
+      "reference": "line is 167 characters (from `lll`)"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "nit: consider adding a blank line between L15 and L16."
+    },
+    {
+      "prediction": "I think this should be `extractor::StorageManeuverOverride` instead of `NodeID`.",
+      "reference": "This is another bug fix: Without skipping the bytes of the vector this would read garbage data."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `NewGenesisData` function, but I'm not sure if there is a better place to do it.",
+      "reference": "why need this? i don't see Gen.CreatorPubKey being used?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "As previously mentioned, best to drop this locale override. It's not something we only provide in exceptional circumstances. Moreover, it doesn't work for this PR anyway, while massively increasing the code complexity!"
+    },
+    {
+      "prediction": "I think we can remove the `{{.}}` in the `VersionDetailsPatch` as well.",
+      "reference": "is this supposed to be `Desired`?"
+    },
+    {
+      "prediction": "I think this test is still valid, but I think it would be better to remove the `Scorer` class and just use the `Weight` class directly. The `Weight` class has a `getScorer()` method that returns the `Scorer` object. The `Weight` class also has a `getWeight()` method that returns the `Weight` object. The `Weight` class has a `getWeight()` method that returns the `Weight` object. The `Weight` class has a `getWeight()` method that returns the `Weight` object. The `Weight` class has a `getWeight()` method that returns the `Weight` object. The `Weight` class has a `getWeight()` method that returns the `Weight` object. The `Weight` class has a `getWeight()` method that returns the `Weight` object. The `Weight` class has a `getWeight()` method that returns the `Weight` object. The `Weight` class has a `getWeight()` method that returns the `Weight` object. The `Weight` class has a `getWeight()` method that returns the `Weight` object. The `Weight` class has a `getWeight()` method that returns the `Weight` object.",
+      "reference": "The iterator is not wrapped when the score mode is set to `COMPLETE_NO_SCORES` so you don't need to change this assertion anymore ?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You need to check if the folder exists before you try to create it. If it doesn't exist, you should create it. Otherwise, you should raise an exception.",
+      "reference": "Is there too much spaces there? (should be 4 I think)"
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `0`.",
+      "reference": "IS_FIRST_PULL revision is better to use instance property, not static. When KieClient has only one instance, instance property is better. When KieClient has many instances(not possible), static is not good eitheir."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "This needs to be removed since we removed the mixin."
+    },
+    {
+      "prediction": "It seems that we don't need to set weight_ here anymore.",
+      "reference": "Can `weight_` be deleted?"
+    },
+    {
+      "prediction": "I think this should be done in a separate PR, as it's not related to the other changes.",
+      "reference": "This is fine, but by Symfony convention this part is responsability of the compiler pass class, i.e `DependencyInjection\\Compiler\\?`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `Puppet::Util` since it's only used by `Puppet::Manager` and `Puppet::Manager` is only used by `Puppet::Pillar` and `Puppet::Pillar` is only used by `Puppet::Pillar`.",
+      "reference": "The host object already has a nice way of querying configprint. Try `puppet('master')['user']`"
+    },
+    {
+      "prediction": "I think this should be `from .addons import` instead of `from mitmproxy.addons import`",
+      "reference": "Let's call this `share` and not `upload` - the user wants to share their flows, uploading is just the implementation of that. :)"
+    },
+    {
+      "prediction": "I don't think this is necessary. The `config` module can be imported from `gui` in this file.",
+      "reference": "It would be better to keep focus on the main text control. But to get around the fact that Dialogs focus their first child on show, even when not active, something like Dialog.isActive should be chcked when appending text, rather than whether the text control has focus."
+    },
+    {
+      "prediction": "I think this should be done in the `emit_exit` function, not here. The `emit_exit` function is responsible for emitting the exit instruction.",
+      "reference": "So is this invoked every time we extend the trace?"
+    },
+    {
+      "prediction": "I think this should be `else if (kv.first == \"microsecond\" || kv.first == \"millisecond\")`",
+      "reference": "Why not use switch here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `auth` package.",
+      "reference": "I think `NoAuth` is a better name"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "reference": "This does create a minor bifurcation in how closing the InputWriter is done in success vs failure cases. Is there anyway to merge this with the success exit path? This is something that seems like it would be better as a function closure instead of method on the `$esapi.Name` type. Can the `es.Close` not be used instead of this method?"
+    },
+    {
+      "prediction": "I don't think we need this change.",
+      "reference": "Please keep this blank line. PEP8 says: > Method definitions inside a class are surrounded by a single blank line."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "You are not passing in a `:params` key here."
+    },
+    {
+      "prediction": "I think this should be done in the `prepareHostNetwork` function.",
+      "reference": "Probably change \"ovs\" to \"OVS\"."
+    },
+    {
+      "prediction": "Why do you need to check here if the category is set?",
+      "reference": "Please linebreak this"
+    },
+    {
+      "prediction": "I think this should be `T?` since it can be `null`",
+      "reference": "`= default` is implied here, so having it is redundant. Is it a stylistic choice to include it, or just an oversight?"
+    },
+    {
+      "prediction": "I think we should keep the old name for the old APIs until we have a good plan for the future of the Project and App managers. This code is still valid and will continue to work just fine with the new `AppManager` and `ProjectManager` interfaces. We just need to make sure to remove the old `wsProjectManager` and `wsAppManager` APIs in this PR.",
+      "reference": "Maybe it's time to fix the param name for this interface?"
+    },
+    {
+      "prediction": "I think we can remove the `ClientLibraryVersion` and `ClientFeatureVersion` fields as well.",
+      "reference": "Why are we removing it? Looks like useful info. Obviously, field names should change."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the `NLog.LayoutRenderer` class.",
+      "reference": "Missing fallback to MessageTemplates.ValueFormatter.Instance"
+    },
+    {
+      "prediction": "`axis` should be `int` or `None`",
+      "reference": "The parameter in the docstring should be fixed too. Actually, why don't you try to implement the other axis? It wouldn't be impossible to do if we use pandas UDF from a cursory look. We have enough time before the next release currently."
+    },
+    {
+      "prediction": "I think we can remove this file.",
+      "reference": "This also runs a patch if its in the middle of one."
+    },
+    {
+      "prediction": "I think this should be `seelog.Debugf(\"Updating task: %s\", task.String())`",
+      "reference": "Could we rename this method to `updateKnownStatus` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the origin tag. I think it should be done in the `StartSpan` method. The `StartSpan` method is called for every span that is added to the context, so it should be the right place to set the origin tag.",
+      "reference": "Maybe it would make sense to move that to DecorateRootSpan? Currently it only has Azure stuff, but given the name of the method I feel like it would be semantically appropriate"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like we should be able to move it to the `extensions` package.",
+      "reference": "nit: these imports were in a separate group to ones from this repo"
+    },
+    {
+      "prediction": "I think it would be better to have a method `isPushContainerizedMethodEnabled(String dispatchMethod)` that returns `true` if the dispatch method is `PUSH_CONTAINERIZED` and `false` otherwise.",
+      "reference": "Is it possible to rename the methods to `isXXXMethod` to keep the usage by the caller generic? The caller could use the result of these methods to verify whether a feature is enabled or to validate an input value for example."
+    },
+    {
+      "prediction": "I think we should use `setTimeout(attemptRm, attemptTimeout);` instead of `attemptRm();`",
+      "reference": "Don't penalize everyone with 250ms delay b/c some machines have problems."
+    },
+    {
+      "prediction": "Shouldn't this be `log.error(\"Failed to parse config file after update: \" + filename, e);`?",
+      "reference": "This try-catch should be here so that watcher thread doesn't just exit in case of an exception. This change alone would probably be enough to fix the error as well, assuming that there's another `ENTRY_MODIFY` event when the file write is finalized. But of course not a perfect fix because it doesn't protect against possibly reading a non-empty but only partially written file?"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "Weird that we have to check for `s == nil` here.. I would instead have fixed the call stack to find out when this gets invoked with a nil server."
+    },
+    {
+      "prediction": "I think this should be `message += Ember.Handlebars.Utils.escapeExpression(erroredEmails.join(', '));` instead of `message += erroredEmails.join(', ');\nReview comment?",
+      "reference": "The `import` statement for `Ember` is missing in this file."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be moved to `getDerivedStateFromProps`",
+      "reference": "Awesome :tada: I'd love to have a test case for this so that we don't regress on this feature in any future refactorings :+1:"
+    },
+    {
+      "prediction": "I think this test should be moved to `test_series_sql.py`",
+      "reference": "For testing some of the rounding functions"
+    },
+    {
+      "prediction": "Shouldn't we use the default rounding type here?",
+      "reference": "This should be present in UPGRADE notes. Also on currently running projects you should suggest users to update these settings."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I'm not sure if it's the right thing to do in general, but I'm not sure if it's the right thing to do in this case.",
+      "reference": "Are you sure you want to do this? This will affect the working set block cache too, such that it'll now be tracking global free space. It throws off the fraction calculation."
+    },
+    {
+      "prediction": "Why is this lock needed?",
+      "reference": "Why would the read access need to be mutexed? The caller might still get a wrong answer since the value might change immediately after the mutex is unlocked before the caller receives the value."
+    },
+    {
+      "prediction": "I think this should be `tab = self.currentWidget() or None` instead of `if tab and tab is not None:`",
+      "reference": "This fits on one line without the parentheses :wink:"
+    },
+    {
+      "prediction": "I think this should be `4` instead of `2`.",
+      "reference": "Looks like `TestCloudRecovery` changes are unrelated?"
+    },
+    {
+      "prediction": "Why not use the `DomainEventInterface` here?",
+      "reference": "Try to separate it to different methods. Invoke is huge :)"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea, but I can't think of a better way to do this.",
+      "reference": "Looks like another spot that would benefit from `truncate`"
+    },
+    {
+      "prediction": "I think this should be `(i - H2O_MAX_TOKENS + 1)` instead of `(i - H2O_MAX_TOKENS)`",
+      "reference": "This change is not related to what this PR wanted to fix, but undefined behavior sanitizer warns without this parenthesis because `h2o__tokens + i` overflows."
+    },
+    {
+      "prediction": "I think this should be `if no_docstring_rgx.pattern and re.match(no_docstring_rgx, node.name).group(0)`",
+      "reference": "This is actually incorrect and should be the other way around. Going to fix this tomorrow. If `no_docstring_rgx == re.compile(\"\")` no function should be checked, since every function matches the pattern. If `no_docstring_rgx == re.compile(\"^$\")` all functions should be checked."
+    },
+    {
+      "prediction": "I think this should be `'user-read-recently-played'` instead of `'user-read-recently-played user-read-currently-playing'`",
+      "reference": "Should this be in a config file?"
+    },
+    {
+      "prediction": "I think this test should be moved to `utils.test.js`",
+      "reference": "purely preference, but have you considered using parentheses to indicate that a method is being tested? e.g., `describe('eachAsync()')`"
+    },
+    {
+      "prediction": "`fs::remove` will throw an exception if the file doesn't exist.",
+      "reference": "Why not return success?"
+    },
+    {
+      "prediction": "I think you can use `ini_get` instead of `get_cfg_var` here.",
+      "reference": "Please use snake_case for properties to keep it consistent with the rest of the codebase."
+    },
+    {
+      "prediction": "I think this should be `s.logger.Infof(\"successfully connected to peer (inbound) %s\", i.BzzAddress.ShortString())`",
+      "reference": "i don't understand why we need this change. what information is this providing when running in less than Debug loglevel? I prefer to revert this"
+    },
+    {
+      "prediction": "I think this test should be moved to a separate commit.",
+      "reference": "Oh, so this test leaks DateTimeUtils offset? It would be safer to place the reset in an `@After` method to not cause side effects even if the test case fails."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Footer` component.",
+      "reference": "Kind of a nit-pick, but is there no way to solve this while still not rendering an empty div if `footerText` is empty? We could still use the `Row` to maintain the same layout but then only render the `Cell` for the pagination - I think it's possible to use specific classes to horizontally offset?"
+    },
+    {
+      "prediction": "I think this enum should be renamed to something more meaningful, like `DomainType` or `DomainTypeEnum`",
+      "reference": "Template is also not an object type so I don't expect to notify on templates. Instead when applying templates, we'll be updating roles/policies/services/groups."
+    },
+    {
+      "prediction": "I think this should be `rows` and `cols` instead of `rows, cols`.",
+      "reference": "Had to fix this to get my tests passing (should have been a new PR sorry)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `IssuerConfig` struct.",
+      "reference": "Small nit, and sorry for not spotting sooner.. this should have `omitempty` on it, else when marshalling nil values into json, it will be `caBundle: null` which trips up some JSON parsers"
+    },
+    {
+      "prediction": "I think this should be `transposed = param.Boolean(default=False, doc=\"\"\"\"Whether to transpose the layout when plotting. \"\"\")`",
+      "reference": "Would be good to say the normal behavior is scanline order (left to right, top to bottom) and that transposing makes it work top to bottom and left to right."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this flag. I think it should be in the `--shared-date` section, since that's where the shared date is coming from.",
+      "reference": "Perhaps note here that `--drive-use-created-date` takes precedence if both set?"
+    },
+    {
+      "prediction": "`eps` should be a float, not a `1e-15`.",
+      "reference": "Use EPS=1e-15 as that in atss_head or FCOS head."
+    },
+    {
+      "prediction": "I think this should be `storage.NewMasterIndex()`",
+      "reference": "Is there a need to replace the Index with a MasterIndex?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in `testExecCommandAgent` since that is where the `mockExecCommandAgent` is defined.",
+      "reference": "are we adding stop state change in a different PR?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [122/120]",
+      "reference": "How much difference is there between sort and sort_by ?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "count doesn't seem to be used. Let's remove it."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `Mongoid::QueryCache` class.",
+      "reference": "Mongo is a hard dependency for Mongoid. Why do we need an if-statement to if it's defined? We should always use Mongo::QueryCache"
+    },
+    {
+      "prediction": "I'm not sure if we want to add a foreign key here. I think it would be better to just add a `foreign_key: true` to the `org_id` column in the `Plans` table.",
+      "reference": "Does this point out at a ROR funder id? If so, could we get rid of the `funder_name` field on the plan and just use the `name` of the funder with id `funder_id`?"
+    },
+    {
+      "prediction": "I think this should be `global.isTsTest = process.env.FB_TS_TEST ? true : false;` instead of `var isTsTest = process.env.FB_TS_TEST ? true : false;`",
+      "reference": "Use !! to cast to bool, rather than the ternary bool antipattern."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I found that worked.",
+      "reference": "why is it in the list twice to begin with?"
+    },
+    {
+      "prediction": "I think this should be `if (!g_creatureEvents->playerLogout(this) && !g_creatureEvents->playerPong(this))`",
+      "reference": "This way you will **never** be kicked while monsters are around you."
+    },
+    {
+      "prediction": "I don't think this is the right place for this test. It should be moved to `org.sonar.plugins.java.api.tree.AssignStmtTree`.",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "I think we can remove `moduleSlugs` from the props since it's no longer being used.",
+      "reference": "This just fixes an unrelated prop type error, correct?"
+    },
+    {
+      "prediction": "I think this should be `for k := range dirCacheEntry.dels {`",
+      "reference": "This was a bug in KBFS-2071 -- oops."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. It's not a channel, it's a payment. It's not a voucher, it's a condition. It's not a redem, it's not a close. What do you think?",
+      "reference": "// Condition is a condition for ... Condition"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [131/120]",
+      "reference": "Might be good to test that no emails were sent out, either. You should be able to use `deliveries` for this"
+    },
+    {
+      "prediction": "`LOG.finest` should be used in the `finally` block, otherwise it will not be logged if an exception is thrown.",
+      "reference": "Right now this is experimental and deeply flaky. We left this at `info` to make debugging user reports a lot easier."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a list of files that should not be downloaded.",
+      "reference": "Is this new list being consumed yet?"
+    },
+    {
+      "prediction": "I think we can remove this line now.",
+      "reference": "moved to `errors.go` which is the only places `math/rand` is used in this package"
+    },
+    {
+      "prediction": "I think this should be `Symfony\\Component\\Environment\\EnvironmentType` instead of `Symfony\\Component\\Filesystem\\Filesystem`",
+      "reference": "This use is wrong, corrent is: `Shopsys\\FrameworkBundle\\Component\\Environment` It is the same in classes below"
+    },
+    {
+      "prediction": "Shouldn't this be `if (pendingConfiguration.remoteHost != null && pendingConfiguration.remoteHost.length() > 0)`?",
+      "reference": "This is not really needed, the line added in `GridNodeConfiguration.java` is what really fixes the `remoteHost` regression."
+    },
+    {
+      "prediction": "I think we can remove this line as well.",
+      "reference": "For my understanding, why was this removed?"
+    },
+    {
+      "prediction": "I think this should be `plugin.Contains(\"/\")` instead of `Directory.Exists(pluginsDirectory)`",
+      "reference": "the whole idea was to display information on the plugin directory in case it was missing - need to add if...else... and display the plugin dir if it is configured (non empty) but cannot be found"
+    },
+    {
+      "prediction": "I don't think this is the right place for this error message. I think it should be in the `LoginActivity` class.",
+      "reference": "Can you please set this as an error on the password view, rather than a toast, just to make sure that it is kept consistent. Check out a couple of lines below."
+    },
+    {
+      "prediction": "I'm not sure why we need to specify `openvpn` here. I think we can just use the default.",
+      "reference": "Shouldn't we pass here a proposal too?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move it to the `solid` module.",
+      "reference": "im not a huge fan of exact text match in these tests, not sure what a good solution is that solves the same problem"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this route. I think it would be better to move it to the `/stake` route, since it's only used for top ups and not for regular staking.",
+      "reference": "this should _not_ be in the api. all postage APIs have moved to the debugAPI"
+    },
+    {
+      "prediction": "I think this should be `not node or node.get(RESERVED['file'])`",
+      "reference": "Including both functions seems like overkill for this PR since only _is_internal_node is ever used. Keep them if you think they'll both be used in the future, but if not, the code will be easier to read if you collapse the logic into a single function."
+    },
+    {
+      "prediction": "I think this should be `tester`, not `tester`.",
+      "reference": "It seems that the two words mean the same thing."
+    },
+    {
+      "prediction": "Why did you change this to proposedTerm_ instead of proposedTerm_ = lastLogId_?",
+      "reference": "Do we need to set lastLogTerm_ to committedLogTerm when `lastLogId_ < committedLogId_`, on line 286"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the word hashes. I think they should be added in the same way as the words.",
+      "reference": "About ~7% memory usage increase on average, at most 512KB"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this TODO. I think it should be in the `Handle` interface, but I'm not sure if that's the right place to put it.",
+      "reference": "Was this too big of a pain to do now?"
+    },
+    {
+      "prediction": "I think you can remove the `arr_eigvec`, `arr_eigval`, `arr_means` and `arr_vars` variables now.",
+      "reference": "Why need to spend time for initializing of array? when will we rewrite the contents anyway? This can take a lot of time in some algorithms. Especially if the filling is in sequential mode."
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.",
+      "reference": "Can this be `null:false, default: ''` to avoid the nil vs blank issue?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in the `windows` test group.",
+      "reference": "Had to add this to get macOS to work."
+    },
+    {
+      "prediction": "nit: can we rename this to something more descriptive? `flowContainerTestBase` sounds a bit like it's the base class for all test classes, which is not the case here.",
+      "reference": "What is the usage of this class?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `MetadataTable` class, since that's where the schema is read from.",
+      "reference": "Could we keep the original signatures as well, so we do not have to rewrite the method calls everywhere and add `null, null`? I think this could greatly reduce the changes needed in this patch"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in `ClearTombstonesForNewStorage`.",
+      "reference": "Ideally this error should not be swallowed (can do in the next PR)"
+    },
+    {
+      "prediction": "I think it would be better to pass the `timeout` as a parameter to the `requests.post()` call. This way, we don't have to pass the `timeout` as an argument to the `requests.post()` call every time we want to update the auth.",
+      "reference": "Wait a minute... You're passing it as a POST parameter. There's no way that can work."
+    },
+    {
+      "prediction": "I think this should be `LogWorkflowInfo(\"Workflow %q finished clean up.\", w.Name)`",
+      "reference": "minor: to be consistent with other logs, either use \"cleaning up\" (preferred) or \"cleanup\"."
+    },
+    {
+      "prediction": "I think this should be `INFLUX_NAMESPACE = \"listenbrainz\"` and not `REDIS_NAMESPACE = \"listenbrainz\"`",
+      "reference": "Adding the config changes to the consul config template `consul_config.py.ctmpl` would be helpful too."
+    },
+    {
+      "prediction": "I think it would be better to keep the old ones and add a new one for \"Laboratory\"",
+      "reference": "While we're at it should we add Department? See Dataverse Category: Add Department #2829"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Files should have an extra line at the end of them."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `detachedClone` function, since it's only used there.",
+      "reference": "TODO: check whether this is needed or not"
+    },
+    {
+      "prediction": "I don't think we need this variable at all. We can just use `IsInsecure` directly.",
+      "reference": "`ReadConfig` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "This is duplicated with `+build !cgo`?"
+    },
+    {
+      "prediction": "I think this should be `expect { driver.find_element(id: 'box0', visible: false) }.not_to raise_error(WebDriver::Error::NoSuchElementError)`",
+      "reference": "This should just be `.not_to raise_error` otherwise it potentially hides errors"
+    },
+    {
+      "prediction": "I think this should be `sfun: either an 1-arg function that takes a Column and returns a Column, or a 2-arg function that takes a Column and its DataType and returns a Column, or `sfun: either an 1-arg or 2-arg function that takes a Column and returns a Column, or `sfun: either an 1-arg or 2-arg function that takes a Column and returns a Column, or `sfun: either an 1-arg or 2-arg function that takes a Column and returns a Column, or `sfun: either an 1-arg or 2-arg function that takes a Column and returns a Column, or `sfun: either an 1-arg or 2-arg function that takes a Column and returns a Column, or `sfun: either an 1-arg or 2-arg function that takes a Column and returns a Column, or `sfun: either an 1-arg or 2-arg function that takes a Column and returns a Column, or `sfun: either an 1-arg or 2-arg function that takes a Column and returns a Column, or `sfun",
+      "reference": "Line 175 needs to be indented so that it's under `sfun`."
+    },
+    {
+      "prediction": "This should be `public override string ToString() => string.Empty;`",
+      "reference": "It looks strange a bit. What is the reason?"
+    },
+    {
+      "prediction": "I think this should be `protected final Logger LOG = Logger.getLogger(CollectionStrategy.class.getName());`",
+      "reference": "Is this one of those `LexicalPreservingPrinter` issues?"
+    },
+    {
+      "prediction": "I think this should be `buf.WriteString(fmt.Sprintf(\"<binary>\", strtype))`",
+      "reference": "Can we also add the length of the slice here? Some like `<binary> len %d`"
+    },
+    {
+      "prediction": "I think it's better to use `zap` from `github.com/olivere/zipkin` instead of `go.uber.org/zap`.",
+      "reference": "move this line together to after line 16, group internal packages together"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Line is too long. [81/80]",
+      "reference": "How about using a reference(project_id) instead of project name"
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "File is not `gofmt`-ed with `-s`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the DataFileService.",
+      "reference": "@qqmyers this looks fine but have you seen any performance issue here? If so, we could right some helper method like doesExternalToolNeedDataset(externalTool). But if there's trivial performance impact, not worth it."
+    },
+    {
+      "prediction": "I think this should be `staticbox_use_local_prefs` or something like that.",
+      "reference": "Make this a member variable instead (move to header inside `class` and rename to `m_bUsingLocalPrefs`)."
+    },
+    {
+      "prediction": "I think this should be `testSuite.name === 'pin-mongos' ? it.skip : it;` instead of `testSuite.tests.forEach(testData => {`",
+      "reference": "Generally I'd say we should factor this out into something more extensible (check an array of potentially skipped tests, for examples), but since we're likely to remove this soon for scheduled work I think this is fine. What do you think @daprahamian?"
+    },
+    {
+      "prediction": "I think this should be `NVCC_OPTIONS -std=c++11` instead of `NVCC_OPTIONS -std=c++11 EXCLUDE_HIP_PLATFORM clang`",
+      "reference": "Why are we skipping these tests? they should be passing in HIP-Clang."
+    },
+    {
+      "prediction": "I don't think this is correct. The `range(256)` is used to create a string of 256 chars, not a random string of 256 chars. The `random.choice(self.chars)` is used to pick a random character from the string.",
+      "reference": "I think a \"string\" in Scapy's spirit (if such a thing exists) is actually a `bytes` object in Python 3 (see `Str*Field`s). So maybe `RandString._fix()` should return a `bytes` object. What do you think?"
+    },
+    {
+      "prediction": "I think this should be `const jSDOM = require('jsdom');`",
+      "reference": "This can be combined to 1 line, just tested, seems to work."
+    },
+    {
+      "prediction": "I think we can remove the `if (chainId.equals(EthNetworkConfig.MAINNET_NETWORK_ID))` check here.",
+      "reference": "what needs to happen if it is not present?"
+    },
+    {
+      "prediction": "I think this should be `ds.DeploymentConfig.PipedSpec.IsInsecureChartRepository(chartRepoName)` instead of `ds.DeploymentConfig.PipedSpec.IsInsecureChartRepository(chartRepoName)`",
+      "reference": "`DeploymentConfig` is only for deployment configuration not Piped configuration so `ds.DeploymentConfig.PipedSpec` is always nil. Instead of that, you can have Piped config with `e.PipedConfig` because it is placing inside `executor.Input`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this code example. I think it should go in the `s3.rb` file, under the `s3_object_acl` method.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I think this should be `Func<HttpRequestBase, bool>`.",
+      "reference": "This line doesn't seem to be right?"
+    },
+    {
+      "prediction": "I think this is the only change that should be in this file. The rest of the changes in this file are unrelated to this PR.",
+      "reference": "this need to be moved up after getting policy.."
+    },
+    {
+      "prediction": "I think this should be `platform.gcp.instance.* invalid project ID`",
+      "reference": "I'm curious - is there a way to test these?"
+    },
+    {
+      "prediction": "I think we can remove the `catalogName` argument from this method.",
+      "reference": "This was broken before as the name of the metadata table started with its type, not catalog."
+    },
+    {
+      "prediction": "I think this should be `BuildConfig` instead of `BuildInfo`.",
+      "reference": "Maybe remake `VersionAsString()` function, so that that we would have build info in all places"
+    },
+    {
+      "prediction": "I think this should be guarded by `if (handshakeCompleted_)`",
+      "reference": "This should probably say something other than SSL_peek?"
+    },
+    {
+      "prediction": "I don't think we need to update copyright year here.",
+      "reference": "Nitpicking, but this should be capitalized since it's a product name :)"
+    },
+    {
+      "prediction": "I don't think we need this anymore, right?",
+      "reference": "Why do we need RawTransactions?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I don't think we need to check for `isDelta` here, since we are only interested in the last sgv, not the delta between the last sgv and the current sgv.",
+      "reference": "should not be this lessThan15MinAgo ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `catch` block of the `DecodeRequestLine` method, since that's where the exception is being thrown.",
+      "reference": "Would it work if we changed the return type of 'RejectRequest*' methods to Exception and instead did `throw RejectRequest(...` ?"
+    },
+    {
+      "prediction": "I think this should be `videoTestElement.canPlayType('video/mp4; codecs=\"dts\" || codecs=\"dts+\"').replace(/no/, '')`",
+      "reference": "It is not a \"perfect\" test, but: webOS 1.2 emulator says `probably`. webOS 3 emulator says \\``. webOS 4 emulator says \\``. Why did you remove `options.supportsDts`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. It seems to me that the check should be done in `hipErrorInvalidValue` instead.",
+      "reference": "This check should be after the HIP_INIT_API to ensure HIP tracing & lazy init works correctly. Also return should be wrapped in ihipLogStatus() so that logging works correctly."
+    },
+    {
+      "prediction": "I think this should be `protected $indexingConfigurationsToReIndex = array();`",
+      "reference": "Can't use short array syntax for the 3.1 release branch."
+    },
+    {
+      "prediction": "(style) space required after that ',' (ctx:VxV)",
+      "reference": "@rhc54 Is this intentional or for debug purposes?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have the `on_resource_changed` function take a `Resource` object as its first argument, and then use `config.registry.statsd.timer(key)(on_resource_changed(resource))` as its second argument. This way, the `statsd` middleware will be able to track the execution time of the listener, but it will only know about the listener for resources that are passed to it as its first argument. The `on_resource_changed` function will be able to use the `Resource` object to get more information about the resource, such as its ID, and then use that ID to get the `statsd` middleware to track the execution time of the listener for that resource.",
+      "reference": "Maybe it should be `plugins.history` here instead of `listeners.X`, since we use those for listeners configured via `.ini` files?"
+    },
+    {
+      "prediction": "Metrics/CyclomaticComplexity: Cyclomatic complexity for team_plan_quantity_select_attributes is too high. [7/6]<br>Metrics/PerceivedComplexity: Perceived complexity for team_plan_quantity_select_attributes is too high. [8/6]",
+      "reference": "Hmm, I also don't have a better idea here. As a small thing, could you make this `@team_page.present?` or similar to make the intent slightly more clear?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. I think it should be done in `Encryptor.getDecryptingCipher`",
+      "reference": "Without that check it would fail later (probably in the getDecryptingCipher method) but the error could be hard to make sense of."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I don't think that this belongs in the host code. A host is pretty much unaware of what hypervisor is running it and I don't want them to be so coupled to their hypervisor."
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "I'm here to tell you that this doesn't make sense because of the way it works. `setuptools.setup()` installs this only for setup-time and then discards those. But the problem is that it will not pick up a newer version because the older version is already in runtime and module cannot substitute itself. The proper place for adding build deps nowadays is `pyproject.toml`->`[build-system]`->`requires'."
+    },
+    {
+      "prediction": "I think this should be `fakeConnectionManager` instead of `fakeManager`",
+      "reference": "fakeConnectionManager -> mockConnectionManager. We probably should not use the `fake` anymore."
+    },
+    {
+      "prediction": "I think this should be `self.base_class_path = [os.path.abspath(x) for x in self.base_class_path]` and not `self.base_class_path = [os.path.abspath(x) for x in self.base_class_path]`",
+      "reference": "Code style. Btw, it's weird Codacy didn't catch that."
+    },
+    {
+      "prediction": "I think this is the only place where `NewEthAddress` is used, so it should be `NewEthAddress` and not `NewEthAddress(bytes)`.",
+      "reference": "no need in BigInt just m.Unwrap().Big().Bitlen() == 0"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `flowAdd` operation.",
+      "reference": "I assume you want to change NetworkPolicy to use bundles in a separate PR later?"
+    },
+    {
+      "prediction": "I think this should be `a4adc8f4da9c7ec0d93b50ca5b73dd14df791c0e`",
+      "reference": "nit: We could parameterize the commit hash, as it's used in multiple places."
+    },
+    {
+      "prediction": "`_ = require('underscore');` should be removed.",
+      "reference": "move this above the \"Constants\" as this is 3rd party ones"
+    },
+    {
+      "prediction": "I think it would be better to use `self.app.get(self.collection_url, headers=self.headers, status=200)` instead of `app.get(self.collection_url, headers=self.headers, status=200)`",
+      "reference": "`assert not in`"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If we're going to emit a state change event when we restart, we should do it when we restart, not when we fail to restart.",
+      "reference": "just curious, why was this modified?"
+    },
+    {
+      "prediction": "Please add a space after `hiveTableName`.",
+      "reference": "I'm curious why the table name is bound at compile time? Is it just to save the table create/drop overhead in a prepare-once-execute-many situation?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "quick question, isn't ProcessJob used by all job types? Then why is JOB_ID not found?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `General`",
+      "reference": "looks like you forgot to actually increase the version"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it would make sense to put it in the `HttpServerInstrumentation` class.",
+      "reference": "I think it might be helpful if we add `Default value: False.` on the end of the summary."
+    },
+    {
+      "prediction": "I think this should be `!== undefined` instead of `!== config.sync.fullSynchronization`",
+      "reference": "I think the check should be `config.sync.fullSynchronization === false` - otherwise this will get triggered even when full sync is `true`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `piped.SealedSecretEncryption` or `piped.SealedSecret`.",
+      "reference": "this blown my mind"
+    },
+    {
+      "prediction": "I think this should be `vsm.openebs.io/replica-container-status` instead of `vsm.openebs.io/replica-cont-status`",
+      "reference": "this should be controller-container-status"
+    },
+    {
+      "prediction": "nit: remove unused imports",
+      "reference": "Sort import. You can use IDE's organize import feature."
+    },
+    {
+      "prediction": "This import is unused.",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "I think this is the only test that needs to be skipped. The rest of the tests should be able to run locally.",
+      "reference": "This PR is off to a rough start. :)"
+    },
+    {
+      "prediction": "I think this should be `config.pacUrl = config.proxyAutoconfigUrl ? config.proxyAutoconfigUrl : config.proxyAutoconfigUrl;`",
+      "reference": "Actually, it looks like the entire `normalizeProxyConfiguration` function isn't needed anymore"
+    },
+    {
+      "prediction": "I think this should be `excluded_portal_link` instead of `excluded_portal_links`",
+      "reference": "How about `unless current_path == carts_path`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems to be the only place where this is used, so I added it here.",
+      "reference": "You need to modify the copyrite aliases so it's done automatically."
+    },
+    {
+      "prediction": "I think we can remove the `key.toLowerCase()` part here, since it's the same as `key`.",
+      "reference": "Quick question: Is this true for all `onAnimation*` and all `onTransition*` events?"
+    },
+    {
+      "prediction": "This should be `else if (typeof(data) === 'object' && data !== null && data.constructor && (data.constructor.name && data.constructor.name === 'Address'))`",
+      "reference": "Should sending an address into address error? Since the call isn't needed, and could be fixed easily."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "This is the wrong MDC; see MDCLoggingContext which imports `org.slf4j.MDC`"
+    },
+    {
+      "prediction": "Shouldn't this be `initial(0)`?",
+      "reference": "Maybe we can clarify a bit what that means."
+    },
+    {
+      "prediction": "I think this is the only place where `TcsTask` is used, so I think it should be marked as `Obsolete`",
+      "reference": "This entire method can be deleted; it isn't used and it isn't an override of Stream."
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "We avoid static method imports. Can you call `MetadataTableUtils.createMetadataTableInstance` instead?"
+    },
+    {
+      "prediction": "I think this should be `return self._pac_url == other._pac_url` instead of `return self._pac_url == other._pac_url or self._pac_url == other._pac_url`",
+      "reference": "No need for this, as pylint already only turns things off for this function and it's needed for the entire function."
+    },
+    {
+      "prediction": "I think this should be `to: :purchaseable` instead of `to: :purchaseable, prefix: :purchaseable, allow_nil: true`",
+      "reference": "I'm actually surprised this is valid Ruby. I thought you'd need either `delegate(...)` or `delegate \\ ...`."
+    },
+    {
+      "prediction": "I think this should be `Assert.assertEquals`",
+      "reference": "Why change this?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. We should be able to do it in the same way as we do for local uploads, which is to use the `isSomeFileRemote` variable.",
+      "reference": "Maybe change it to `'Can't UPLOAD remote files when bundle: true is set'`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle the case where the circular buffer is empty, but it seems like it's the only way I could come up with.",
+      "reference": "Do we need to drop the remaining items from the batch? Otherwise we might end up with a dead loop. Add @CodeBlanch for awareness."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `applyMessageDirect` function, right?",
+      "reference": "Because power is now added during `submitPoSt` this is needed for setting power in the genesis block. Again let me know if this bootstrapping solution is flawed."
+    },
+    {
+      "prediction": "I think this should be `ReflectionMethod::getDeclaringClass`",
+      "reference": "To be consistent, these should go at the top between `DateTimeZone::listIdentifiers` and `SQLite3::createFunction`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `FileSystem`.",
+      "reference": "I think we should avoid changing any lucene classes for the moment - fileModified() can probably stay where it is?"
+    },
+    {
+      "prediction": "I don't think we should be changing the visibility of an `ImageView` from `Invisible` to `Gone`. I think we should be changing it from `Gone` to `Invisible`.",
+      "reference": "GONE breaks formating of \":\". or is it fixed below?"
+    },
+    {
+      "prediction": "I think we should deprecate all methods from `ASTFormalParameter`.",
+      "reference": "What do you mean with this? Won't the deprecation be inherited?"
+    },
+    {
+      "prediction": "I think this should be `--cpp-suffix` instead of `--cpp-suffix-string`.",
+      "reference": "line between \" and \" should be <80 chars."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "What will happen if neither `rvm` nor `ruby` is present in the .travis.yml?"
+    },
+    {
+      "prediction": "I think this should be `p[2:5] + struct.pack(\"<H\", len(pay)) + p[4:]`",
+      "reference": "@p-l- Hey it seems that this was len-4 ?!"
+    },
+    {
+      "prediction": "nit: can we move this to the top of the file, so it's more obvious what's going on?",
+      "reference": "This is something I added which resolved a few large errors in the output, although not related to the refactor here I don't think."
+    },
+    {
+      "prediction": "I think this was a leftover from the previous version of the code.",
+      "reference": "i have removed this line because whenever a video is played in AntennaPod this line was making play button invisible"
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think it's better to use `org.apache.logging.log4j.core.util.SystemUtils#java.io.tmpdir`",
+      "reference": "remove temporary code"
+    },
+    {
+      "prediction": "This should be `Long` instead of `Hash`",
+      "reference": "* Should be tagged `@Unstable` * Should be a default method, returning `null` * Also, perhaps `Optional<Long>` instead of just `Long`? Always empty when the BASEFEE isn't relevant? If so the default is `Optional.empty()`"
+    },
+    {
+      "prediction": "I think this should be `test-store-name` instead of `test-store-name-'test-' + MODULE_SLUG`",
+      "reference": "Minor detail, but let's use the interpolated template string syntax instead."
+    },
+    {
+      "prediction": "I think we can remove the ClockIds field and use the default value instead.",
+      "reference": "this line is so long, split it to multi lines."
+    },
+    {
+      "prediction": "I think this should be `device_extensions.vk_ext_scalar_block_layout || enabled_features.core12.scalarBlockLayout == VK_TRUE`",
+      "reference": "Looks like this might be one of those \"promoted features\" where you can enable it by _either_ enabling the extension _or_ enabling the feature bit. If that is the case, I think this needs to be: `(device_extensions.vk_ext_scalar_block_layout == kEnabledByCreateinfo) || (enabled_features.core12.scalarBlockLayout == VK_TRUE)`."
+    },
+    {
+      "prediction": "Is this needed for all platforms or only for macOS?",
+      "reference": "Why should we define it for v1? This is preserved only for backward compatibility, all further modifications must be done in the latest vX"
+    },
+    {
+      "prediction": "I think this should be `return dataset.shape[0] > 0`",
+      "reference": "This is a definite improvement! Hardcoding nonzero is vastly better than hardcoding length. Even so, is there no way to determine the actual value of nonzero in a way that doesn't load the entire dataset?"
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this method. I think it should be something like `GetSpanUriTagValueFromRequestUri` or `GetSpanUriValueFromRequestUri` since it's a method that returns a span uri value.",
+      "reference": "do we have a way to avoid the string concats, if there is no username/password in the Uri? if (uri has UsernameInfo) { do what is done in this PR. } else { existing behavior. }"
+    },
+    {
+      "prediction": "I think this should be `flb_plg_debug(ctx->ins, \"HTTP data=%s\", c->resp.data);`",
+      "reference": "do not print the payload since it might be corrupted, you can get the data with _debug.http.response_payload configuration property"
+    },
+    {
+      "prediction": "I think this should be `Expect(\"./copilot/.workspace\").To(Not(BeAnExistingFile()))` and not `Expect(\"./copilot/.workspace\").Should(BeAnExistingFile())`",
+      "reference": "maybe \"app init creates an copilot directory and a workspace file\"? since you validate for the file as well."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "Why do we need this namespace? I don't see it being used anywhere.",
+      "reference": "Nit: Could rename this file from SabreEvent.php now that it's not for Sabre"
+    },
+    {
+      "prediction": "I'm not sure if we want to do this now or later, but I think we should do it now.",
+      "reference": "I believe this `Stop` function _is_ called with deferred."
+    },
+    {
+      "prediction": "I think we can remove the `parseAccountID` function from this file as well.",
+      "reference": "This entire clause should now be removed. It was only relevant if `accountID` couldn't be parsed from `clientID`, which is now no longer needed."
+    },
+    {
+      "prediction": "I think this should be `https://testinfra.readthedocs.io/en/latest/`.",
+      "reference": "FTR: Unlike commonly known practice, `latest` in Ansible docs corresponds to the stable version, not to the latest state of the main Git branch. Is this your intention? Just checking..."
+    },
+    {
+      "prediction": "I think this should be `charsetTarget = CharInfo::GB18030;` instead of `charsetTarget = CharInfo::UTF8;`",
+      "reference": "Why do we have || CharInfo::UnknownCharSet here? I do not see it it in neighbouring statements. This is just for my understanding."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `NLog.Targets` assembly.",
+      "reference": "isn't `HttpWebRequest` an expensive key value?"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "couldn't we run this query whether there are observations or not?"
+    },
+    {
+      "prediction": "I think it would be better to use `luigi.run(['--local-scheduler', '--no-lock', 'AmbiguousClass', '--n', '1000'])`",
+      "reference": "This shouldn't be needed to change right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I'm not sure if it's possible to have multiple bundles for the same file.",
+      "reference": "This seemed broken to me. Why would browserify only check the (previously `cdn`, now) `bundle` example? And even so, the path is incorrect here. Fixing this is unrelated so should not go into this PR I feel. But when we fix this in master, perhaps that solves the reload issues that you experienced @hedgerh?"
+    },
+    {
+      "prediction": "I think we can remove the `Timeout` field from the `DDWAF_RET_CODE` struct as well.",
+      "reference": "Same question, do we have potential version-conflict crashing scenarios here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. The RGroupDecomposition class is only used by RGroupDecomposition.cpp, which is only called from RGroup.cpp, which is only called from RGroup.cpp, which is only called from RGroup.cpp, and so on. I think it would be better to move this code to RGroupDecomposition.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp, which is called from RGroup.cpp,",
+      "reference": "Definitely not required, but it would be better if you had a constexpr for `\"INPUT_DUMMY\"`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `TryAutoArchive` method, since that is the place where the `TryAutoArchive` method is called.",
+      "reference": "It seems ok. Because already created a file name at line 1815 and already checked if it is null or empty at line 1816. If file name is null, then already this line will not executed."
+    },
+    {
+      "prediction": "I think we can remove this method and just use `getArrowGlacierBlockNumber()` and `getBaseFeePerGas()`.",
+      "reference": "Can all the BaseFeePerGas methods that are added be removed? It's not needed for the bomb and not referenced anywhere else in this PR."
+    },
+    {
+      "prediction": "I think this should be `Warnf` instead of `UserOut.Warnf`.",
+      "reference": "util.Warning()? Easier to say."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the same place as `numEntry` and `numBranch`.",
+      "reference": "these 2 no longer needed after refactor"
+    },
+    {
+      "prediction": "I think this should be `SpecCSIRawBlockMountOptions` or `SpecCSIMountOptionsRawBlock` to be consistent with the other mount options.",
+      "reference": "most spec options appear to be snake case - `fa_direct_raw_block`, let's stick to that convention"
+    },
+    {
+      "prediction": "I think this should be `require('selenium-webdriver/chromedriver')` instead of `require('selenium-webdriver/chrome')`",
+      "reference": "Looks like we can then drop this dependency."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think this should be done in the `FlowRunner.run()` method.",
+      "reference": "3) Subscribe to the script change signal in `FlowMaster.__init__`. The event handler should call `self.masterq.put((\"script_change\", script))`. 4) Add a `handle_script_change` function, that once called, takes the script object and calls `script.reload()`."
+    },
+    {
+      "prediction": "I think this should be `public String getSubject() { return _subject; }` and `public String getBody() { return _body.toString(); }`",
+      "reference": "This method is for unit testing only, right? How about making it package private? This way the readers would know that this is not a public API outside this package and would reduce the search space."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "accidental? i guess it's the same as any other boilerplate"
+    },
+    {
+      "prediction": "I think this should be `logger.warning` instead of `logger.debug`. The former is for informational purposes, while the latter is for error reporting.",
+      "reference": "shadowing the builtin, let's use function or func or callable_"
+    },
+    {
+      "prediction": "I'm not sure this is a good example of what you're trying to do here. I'd rather see a concrete example of what you're trying to do, and then we can figure out how to make it work.",
+      "reference": "Can you add some good examples, for instance integers, strings and whatnot?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Can we just have 2017 here @kmova if possible, as i seen in other projects as well( kubernetes etc..), they mentioned only the year when the file has been created."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `DebugPrintf::ParseFormatString`.",
+      "reference": "@Tony-LunarG I just realized that this differs from the previous behavior in that spirv-opt will run on the byte code if there are any \"group decorations.\" If this is a problem, I can add an additional constructor to keep the pre-existing behavior."
+    },
+    {
+      "prediction": "I think we can remove the `column.getPrimitiveType()` call here as well.",
+      "reference": "Nit: continuation lines should be indented 4 spaces from the start of the statement."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. It seems like it would be better to do this in the `sysfs_init()` and `sysfs_cleanup()` functions, since those are the places where we initialize and clean up the sysfs data structures.",
+      "reference": "Are these protected by any kind of lock?"
+    },
+    {
+      "prediction": "I think it would be better to call this `firewall_rule_hierarchical_name_prefix` or `firewall_rule_hierarchical_name_suffix`.",
+      "reference": "This should probably default to a string, as get_resource_ancestors is causing the tests to fail due to the rsplit on a None hierarchical_name."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should go in `extensions/v1beta1/types.go` under the `Ingress` type.",
+      "reference": "I prefer not to have markdown here, while it is nice for the site this is also shown for `kubectl explain` where this will look weird"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to change `mock.DEFAULT_ISMASTER` to `mock.DEFAULT_ISMASTER_36`",
+      "reference": "@durran Was this change intended to be included in this PR?"
+    },
+    {
+      "prediction": "I don't think you need to pass `null` here.",
+      "reference": "Feels like this could have a default null value to remove the need for this explicit \"null\" use."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `if err != nil` block above, right?",
+      "reference": "\"failed to create ...\""
+    },
+    {
+      "prediction": "I think this should be `PodError` instead of `PodObjectError`",
+      "reference": "No formatting directives, `glog.Error` will do just fine."
+    },
+    {
+      "prediction": "I think this should be `expect { driver.find_element(id: 'box0', visible: false) }.not_to raise_error(WebDriver::Error::NoSuchElementError)`",
+      "reference": "This should just be `.not_to raise_error` otherwise it potentially hides errors"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "Need an `if err == nil {break}` above this line so that we don't log/sleep if the retry succeeds."
+    },
+    {
+      "prediction": "Why did you remove the `AbstractId` here?",
+      "reference": "Separate it to different methods :D Invoke method look's like old fashion portal class :D"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it's better to use `getBoundingClientRect` to get the actual size of the element and then use `offsetHeight` to get the height from the DOM.",
+      "reference": "Collapse/expand animation of filter is broken - not smooth."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in the `if` block of the `install` method above.",
+      "reference": "so during the `script` stage we run `mvn install`?"
+    },
+    {
+      "prediction": "I think it's better to use `github.com/iotexproject/iotex-core/state/factory` as the package name.",
+      "reference": "we may need to move CandidateNamespace to this protocol as it is only used here."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to set the node status to KILLED when the flow is in EXECUTION_STOPPED state. I think it's better to set the node status to KILLED when the flow is in either EXECUTION_RUNNING or EXECUTION_KILLED state. The reason is that if the flow is in EXECUTION_RUNNING or EXECUTION_KILLED state, it means that the flow is still running, but it's not killed yet. If the flow is in EXECUTION_STOPPED state, it means that the flow is still running, but it's already killed. In either case, the node status should not be changed when the flow is in EXECUTION_STOPPED state. Instead, the node status should be set to KILLED when the flow is in either EXECUTION_RUNNING or EXECUTION_KILLED state.",
+      "reference": "Just to confirm, this will take care of all the nodes in graph. right?"
+    },
+    {
+      "prediction": "I think this test should be moved to a new file, since it is not related to the other tests in this file.",
+      "reference": "nit: remove this line. All \"arrange\" may be in one code block."
+    },
+    {
+      "prediction": "I don't think we need this type. We can just use `xattrName` and `xattrValue`.",
+      "reference": "This name is not a good fit, what about `ExtendedAttribute`? Or even just `Attribute`?"
+    },
+    {
+      "prediction": "This should be `assert` instead of `TEST_ASSERT`.",
+      "reference": "lets keep our \"testing framework\" header only if possible. since there is an `assert(0)` in there, it is intended to halt on the first test failure."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `action/protocol` since it's a part of `protocol` package.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "Can you revert this change?",
+      "reference": "I was saving this white-space for my retirement :nit:"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "What if you don't want to continue with the behavior, but you just want the deprecation to go away? Should we add `skip_after_action :discard_flash_if_xhr`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `AccountStateByHash160` and not in `AccountState`.",
+      "reference": "change `accountNonceMap` to map[address.Address][]uint64"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in the test.",
+      "reference": "The `CLIENT_CONFIG` is \"client_config\" with underscore, so we can't reuse it in the second arg."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `if` block above, right?",
+      "reference": "Does this mean all admins get the default batch size (which is 20) no matter the value of `-sync-batch-size`? It seems the command line flag doesn't do anything then does it?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. DownloaderException should only be thrown when the CurlDownloader is not supported.",
+      "reference": "Set default value as empty array instead of null"
+    },
+    {
+      "prediction": "I think this should be `if routeUpd.Dst != nil`",
+      "reference": "Would be good to UT this case"
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix, but it's the only thing I can come up with.",
+      "reference": "this was updated after linking the latest ethereum/tests after a submodules recursive update"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It's not a data property of the component, it's a data property of the component instance.",
+      "reference": "Todo: find a way to only do this in one spot, ideally with defaults for the stuff we use"
+    },
+    {
+      "prediction": "I think this should be `org.apache.avro.generic.GenericDatumReader` instead of `org.apache.parquet.hadoop.ParquetFileReader`",
+      "reference": "Imports should not have blank lines. Sorry there isn't a checkstyle rule running for this yet, we still need to update the build for this one."
+    },
+    {
+      "prediction": "feval should be a list of function, not a function object.",
+      "reference": "Please move this to L3115 to group the code logically - all code for `feval` in one place."
+    },
+    {
+      "prediction": "I think this should be `public final`",
+      "reference": "This was done to wrap the operation so it could be spied upon during testing. Likely needs to be undone once an alternative test is discovered."
+    },
+    {
+      "prediction": "I think we can remove the `time.Now().Add(time.Duration(cr.Expires) * time.Millisecond)` part.",
+      "reference": "I feel we need to impose a lower limit here maybe? What about a max and default for when none is set?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `cds.cdsObjectToUpnpObject`",
+      "reference": "ineffectual assignment to `err` (from `ineffassign`)"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think it's better to have a single `Replacer` object that can be used by both `ReplacerWithDefault` and `ReplacerWithValue`.",
+      "reference": "Ahhh... this makes sense. Does using functions to defer evaluation of these values have anything to do with #916 (where the `Host: {host}` was 'cached' between different hostnames)? If not, could you help me understand what this change does?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Perhaps the only test needed is to see if the socket is not an invalid socket? Is the >= 0 test necessary at this point?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the subscription form, but it's the only one I could come up with right now.",
+      "reference": "Could this line use `url: current_user.subscription`?"
+    },
+    {
+      "prediction": "This should be `preg_quote($quoted_remove_var_id, '/`",
+      "reference": "I was seeing `Warning: preg_match(): Unknown modifier 'a' in /path/to/project/vendor/vimeo/psalm/src/Psalm/Context.php on line 480` here"
+    },
+    {
+      "prediction": "I think you can remove the `static` keyword here.",
+      "reference": "The changes in this file are not necessary."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "run `make lint` to fix"
+    },
+    {
+      "prediction": "I'm not sure this is a good test. I think it would be better to have a test that actually executes the pipeline and verifies that the result is as expected.",
+      "reference": "we will need `assert result.success` instead. result will always be not null because it returns an execution result including several metadata - when the execution fails, it'd return an execution result whose `success` attribute is false."
+    },
+    {
+      "prediction": "I think we can remove `ZvolWorkers` field from `CStorVolumeReplicaSpec` struct.",
+      "reference": "Can we mention how can one determine the best value given an infrastructure? Does it default to any value?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Leave this formatting the way it was please, that's the standard across all of October's files."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `libcontainer/selinux`",
+      "reference": "@runcom says there is no need to use `selinux` here. `selinux` \"github.com/opencontainers/selinux/go-selinux\" The code should work fine without it."
+    },
+    {
+      "prediction": "I think this should be `enum('ACTIVE','DELETE_IN_PROGRESS','DELETED')`",
+      "reference": "nit: It might be worth defining these elsewhere as a python-type?"
+    },
+    {
+      "prediction": "I think this should be `quartz.scheduler.enabled`",
+      "reference": "prefix it with azkaban. Maybe azkaban.server.schedule.enable_quartz?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Puppet::Data` module, since that's where the `apply` method is.",
+      "reference": "It'd be nice if we also caught `Unknown variable $trusted / $server_facts / $settings::` here - while I highly doubt anyone is using them (and if they are they're surely misusing them) it'd be crazy-making to have a working plan suddenly start failing with no clue as to why."
+    },
+    {
+      "prediction": "I think we can remove this panic.",
+      "reference": "Probably need to understand/implement this one now."
+    },
+    {
+      "prediction": "I think this should be `if (d != null && depValidationStatuses.get(d) == null)`",
+      "reference": "It will be better to add contains check for key in hashAndFileNameToDep instead of null check over here. All of these should be inside that contains if case."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "perhaps these magic numbers could go into a Defaults module somewhere? 8140 implies the default is for puppet server. i guess that's somewhat reasonable. but if we're trying to decouple beaker from puppet, maybe this should be 80?"
+    },
+    {
+      "prediction": "This should be `uint64_t` instead of `uint32_t`.",
+      "reference": "style violation: { on own line"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add compression hint. I think it should be in `AddHeaderDuped` function.",
+      "reference": "Could we accept caseless strings (e.g. `off`, `on`, `auto`) instead of numbers to indicate the mode? Of course, we should use an enum internally (as we already do)."
+    },
+    {
+      "prediction": "I think we can remove `GetTipSetAndState(tsKey types.SortedCidSet)` and `GetTipSetStateRoot(tsKey types.SortedCidSet) (cid.Cid, error)` in favor of `GetTipSet(tsKey types.SortedCidSet) (*types.TipSet, error)` and `GetTipSetStateRoot(tsKey types.SortedCidSet) (cid.Cid, error)`",
+      "reference": "The sender doesn't use `GetTipSet`, so please remove it from this interface."
+    },
+    {
+      "prediction": "I think we should revert this change.",
+      "reference": "We can get rid of this import then."
+    },
+    {
+      "prediction": "I think this should be `KubernetesContainerizedImpl` instead of `KubernetesContainerizedImpl`",
+      "reference": "These configurations are currently overwritten in the props we provide in our config files. They need to be overridden there as well after you make this change."
+    },
+    {
+      "prediction": "I think this should be `if *scratchBucketGcsPath == \"\" && mgce.OnGCE() {`",
+      "reference": "Why is error ignored here?"
+    },
+    {
+      "prediction": "This should be `PartitionFlagBootLegacy`",
+      "reference": ">PartitionFlagBiosGrubLegacy PartitionFlag = \"bios-grub\" [](start = 1, length = 55) I presume this is to fix a breaking change...and it's too late to fix the breaking change (since it already escaped into the world?)"
+    },
+    {
+      "prediction": "I think this should be `Iterable<ManifestFile> manifests = table.currentSnapshot().dataManifests().get(0).getManifests();`",
+      "reference": "Is this explicitly ignoring the effect of deleted rows on partition metrics or is it just that you are short circuiting any delete files (as we can't use them anyways)"
+    },
+    {
+      "prediction": "I think this should be `if (!string.IsNullOrEmpty(Tools[toolName].osSpecific[Os][\"defaultParameters\"]))` instead of `if (!string.IsNullOrEmpty(Tools[toolName].osSpecific[Os][\"defaultParameters\"]))`",
+      "reference": "I would also check whether or not this \"defaultParameters\" is null or empty to allow for people to omit it from the config file."
+    },
+    {
+      "prediction": "I think this should be `fmt.Sprintf(\"%v\", string(event.Content))`",
+      "reference": "I think event.Content is of type interface{}, so we may need %v here?"
+    },
+    {
+      "prediction": "I think this should be `RemovedInInvenio22Warning` instead of `RemovedInInvenio23Warning`.",
+      "reference": "Note: `editor` depends on legacy `BibEdit`"
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "we should probably deprecate (and display a warning) the misspelled keyword arg here rather than removing it... and then add the new one. This changes a public API and will break any code that is currently using the misspelled version."
+    },
+    {
+      "prediction": "I think this test should be moved to `ks` instead of `GroupBy` since it's not related to `GroupBy`.",
+      "reference": "I'm wondering if a positive test case is needed here? Then we might compare the result DataFrame with a pre-created Koalas DataFrame."
+    },
+    {
+      "prediction": "Shouldn't this be `@return true, if no underlying value is present, false otherwise.`?",
+      "reference": "I think it was reversed"
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "reference": "The setter \"setName\" can be package-private as well"
+    },
+    {
+      "prediction": "I don't think this is correct. You are changing the value of `beta1` and `beta2` at the same time. I think you need to change `beta1` and `beta2` to be `beta1_prev` and `beta2_prev` respectively.",
+      "reference": "Not sure if this is intentional or snuck in through find/replace; while this is a correct style fix (and is welcome in another PR), it should not be done here so that the commit maintains a logical changeset."
+    },
+    {
+      "prediction": "I think this should be `std::vector<uint8_t>(1073741824, 0)` instead of `std::vector<uint8_t(1073741824)`",
+      "reference": "This makes me wonder if the TBufferTransport should have a size limit that is configurable, with a default of INT32_MAX, and then the test can make a smaller one like 4KB, and write 4KB and then one byte more, instead of using up 2GB of memory."
+    },
+    {
+      "prediction": "I don't think we should be adding these properties to the method view. I think we should be adding them to the `Method` class.",
+      "reference": "If you use an enum for the streaming type, you wouldn't need to add each boolean as a separate attribute."
+    },
+    {
+      "prediction": "I don't think we need to import `MySQLdb` here.",
+      "reference": "can you check if you can use `sqlalchemy.exc.IntegrityError` instead?"
+    },
+    {
+      "prediction": "Why not use `std::unique_lock<std::mutex>`?",
+      "reference": "Why we need to use recursive_mutex here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `certmanager_test.go` under the `ACME` package.",
+      "reference": "am I correct these changes are related to: > I have removed a bunch of legacy functions for issuer generation ... if so, I can see why that's a valuable change but this is already a pretty huge PR and these changes to use `gen` here feel quite distant from the goal of this specific PR; they should maybe be in their own PR, to make everything easier to review? or have I missed how these changes are linked to this PR?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "I think that this is an unused import"
+    },
+    {
+      "prediction": "I think it would be better to use `datasetVersionDTO.getVersionState().toString()` instead of `datasetVersionDTO.getReleaseTime().substring(0, 10)`",
+      "reference": "The string \"DVN_3_0\" appears in `if (!SOURCE_DVN_3_0.equals(xmlr.getAttributeValue(null, \"source\"))) {` in ImportDDIServiceBean. It's not an exact match so I don't *think* this will break dataset import from DDI but I thought I'd at least mention it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `CompareCommits` method.",
+      "reference": "Can you add docstring to this? Also, this might be a nit but I think `CompareCommits` or `CommitRange` might be a better name. What do you think?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `CheckImmutableFields` method.",
+      "reference": "were all these ones missed in the previous PR so you are adding them now? the sources v1 and some of this? Is that right? If so, can you update the PR title"
+    },
+    {
+      "prediction": "I think this should be `VERSION_75`",
+      "reference": "hmm should this be `VERSION_81` instead?"
+    },
+    {
+      "prediction": "I'm not sure this is the right number. I think it should be something like 1024 or 2048 or 4096 depending on how big the trie is going to be.",
+      "reference": "What is the reason for increasing this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could get it to work.",
+      "reference": "is there any reason we are not adding this to companion as well?   There are no CORS concerns there, so adding probably shouldn't be a problem, no? **One question about the requestId;** Are the request Ids unique to each tus-js-client request, or are they unique to an upload instance instead?"
+    },
+    {
+      "prediction": "I think this should be `private final String cookieName;` and `private final String cookiePath;` instead of `private String cookieName; private String cookiePath;`",
+      "reference": "The formatting is a little off here."
+    },
+    {
+      "prediction": "I don't think this change is correct. It should be `pw.toString().contains(\"JavascriptCompiler$CompiledExpression.evaluate(\" + source + \")\"));`",
+      "reference": "Perhaps we should import assertj for tests. These assertions are so much cleaner with assertj. Don't know whether hamcrest equivalent exist (maybe it does)."
+    },
+    {
+      "prediction": "This should be `for (std::unordered_map<Slot, InventoryItem>::iterator i = slots.begin(); i != slots.end(); ++i) {`",
+      "reference": "You should use C++-type casts instead."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `CreateWorkflowExecution` method of the `InternalCreateWorkflowExecution` struct.",
+      "reference": "Looks like each AppendHistoryNodes is its own transaction - what happens if `len(request.NewWorkflowNewEvents) == 5` but we only succeed in writing 3 of them? Will the system be able to pick up from there later? I'm _guessing_ the answer is related to the `ON CONFLICT` and `INSERT IGNORE` changes below? What does Cassandra do?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `SVIDStore` service, not the `SVIDStore` service. The `SVIDStore` service is responsible for managing the SVID store, and it should be the one that deletes the SVIDs.",
+      "reference": "why are you replacing log here? it is defined above and contains EntryID and SPIFFEID"
+    },
+    {
+      "prediction": "I think this should be `cmd.vpnServerFactory` instead of `cmd.openvpnBinaryCheck`",
+      "reference": "Why not `checkOpenvpn` as in client command?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The `Base` constructor is called by `Base._reporter` which is called by `reporter` which is called by `mocha` which is called by `it` and `describe` which is called by `run`. I think it would be better to move the `Base._reporter` call out of `reporter` and into `Base._reporter` which would then call `Base._reporter` and `Base._reporter` which would call `Base._reporter` and `Base._reporter` which would call `Base._reporter` and `Base._reporter` which would call `Base._reporter` and `Base._reporter` which would call `Base._reporter` and `Base._reporter` which would call `Base._reporter` and `Base._reporter` which would call `Base._reporter` and `Base._reporter` which would call `Base._reporter` and `Base._reporter` which would call `Base._reporter` and `Base._reporter` which would call `Base._reporter` and `Base._reporter` which would call `",
+      "reference": "> Note: Non-headless tests (that run on selenium-webdriver) need to wire up to mocha reporter to collect all the results and report them. (See `test-webdriver.js`). But for other tests this can be skipped."
+    },
+    {
+      "prediction": "I think this should be `b'CONNECT %s:%s HTTP/1.1\\r\\n' % (connect_to[0].encode(), connect_to[1])` instead of `b'CONNECT %s:%d HTTP/1.1\\r\\n' % (connect_to[0].encode(), connect_to[1])`",
+      "reference": "I think we should apply IDNA encoding for the host here."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better idea.",
+      "reference": "No need to use `this->` here"
+    },
+    {
+      "prediction": "I think this should be done in the `ReconcileClusterDeployment` method, not in the `ReconcileClusterInstall` method.",
+      "reference": "I can't quite tell, is there a reason why we're finding this condition again when it was done a few lines above? Or was this a mistaken paste?"
+    },
+    {
+      "prediction": "Can you remove this copyright file from this PR?",
+      "reference": "doh. My fault, sorry."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `types.go` since it's used in both `types.go` and `models.go`.",
+      "reference": "Why do we returning `interface{}` here? Don't we have a predefined type for `ID`?"
+    },
+    {
+      "prediction": "I think we can remove `__` and `sprintf` now that we're using `wp.i18n` and `lodash`",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/config/provider/viper` or `pkg/config/provider/gocfg`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think we should use `===` instead of `==` here. The reason is that `===` is a faster comparison than `==`.",
+      "reference": "Not sure why you want to make the distinction between null and undefined? And also `null == undefined` anyway so I think you could simply have `!noteIds` here."
+    },
+    {
+      "prediction": "I think it would be better to have a constant for this, so we don't have to change it in the future.",
+      "reference": "Almost makes me wonder if this should default to Rpc-Context so everything defaults under Rpc-"
+    },
+    {
+      "prediction": "nit: I think it's better to move `accountutil` to the same folder as `account.go`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I don't think we need to check for `len(mset.consumers) >= mset.config.MaxConsumers` here, since we already check for `mset.config.MaxConsumers > 0` above.",
+      "reference": "would these require any account lock here?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to just set it to `socket._GLOBAL_DEFAULT_TIMEOUT` and have it be the default.",
+      "reference": "The else doesn't return anything?"
+    },
+    {
+      "prediction": "I think this should be `if (opts.fee == 0 || opts.feeSat == 0) this.givenFeeSat = opts.fee * opts.feeSat;`",
+      "reference": "if (typeof (opts.fee) !== 'undefined' || typeof (opts.feeSat) !=='undefined') {"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in `device.go` under the `if d.Permissions == \"\" {` block.",
+      "reference": "I not sure we need to change this method's logic. Isn't no perms and invalid config?"
+    },
+    {
+      "prediction": "Why not 10 seconds?",
+      "reference": "this is an abysmal amount of time. can you explain under which circumstances you should wait for a minute for a message?"
+    },
+    {
+      "prediction": "I think we should be able to get this from the `targetEnvironment` in the `Apply` method of the `deployapi.DeploymentConfig` resource.",
+      "reference": "Can we add this code to `job_package.go` and `svc_package.go`?"
+    },
+    {
+      "prediction": "I think this is the wrong place to put this. It's not a plan node, it's a physical node.",
+      "reference": "While you're here, can you change the casing on this so it is unexported? `nextPlanNodeIDKey` instead."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think we should move this to `pkg/dispatch` or `pkg/dispatch/remote`.",
+      "reference": "thanks for tidying all these up, as you can tell they've been used pretty arbitrarily in the past to just exclude things from bootstrap..."
+    },
+    {
+      "prediction": "I think this should be `plot.dimensions is not None` instead of `not unbound_dimensions(plot.streams, plot.dimensions, False)`",
+      "reference": "Again, `no_duplicates=False` would be clearer here..."
+    },
+    {
+      "prediction": "I think this should be `closed <-chan struct{}` and not `closed <-chan struct{}`",
+      "reference": "shouldnt we stop these routines BEFORE the db closed?"
+    },
+    {
+      "prediction": "I think this import should be removed.",
+      "reference": "It fills me with endless sadness that we can't use Selenium's own `SessionId` here."
+    },
+    {
+      "prediction": "nit: `LOG.info(String.format(\"Pulling %s:%s\", name, tag));`, `LOG.finest(String.format(\"Pulling %s:%s\", name, tag));`, `LOG.trace(String.format(\"Pulling %s:%s\", name, tag));`, `LOG.warn(String.format(\"Pulling %s:%s\", name, tag));`",
+      "reference": "This will always need to be displayed to users."
+    },
+    {
+      "prediction": "Style/PercentLiteralDelimiters: %w-literals should be delimited by ( and ).",
+      "reference": "Maybe we make an `ORIGINS` constant on the Cart model instead?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add this field. I think it should be in `apicontainer.Container`, but I'm not sure if that's the right place to add it.",
+      "reference": "looks like this is not needed anymore with boltdb migration. lets confirm with @fenxiong"
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "shadow: declaration of \"err\" shadows declaration at line 408 (from `govet`)"
+    },
+    {
+      "prediction": "This should be `LOG.info(\"When using the W3C Action commands, offsets are from the center of element\");`",
+      "reference": "Again, the existing level is correct --- we're letting users know about something that might cause there tests to fail."
+    },
+    {
+      "prediction": "I think this should be `conversion.Converter` and not `conversion.NewConverter`",
+      "reference": "Required by the libcalico-go changes"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "You want \"current_min_count\""
+    },
+    {
+      "prediction": "I think we can remove the `=` from the end of the `selectionArgs` array.",
+      "reference": "Why changes in this file?"
+    },
+    {
+      "prediction": "I think this should be `\\\\[([^\\\\]]+)\\\\]\\\\[([A-Za-z_][A-Za-z_.0-9]*)*\\\\]`",
+      "reference": "This doesn't look right. I think what you want here is `\\\\[([^\\\\]]+)\\\\]\\\\[[A-Za-z_][A-Za-z_.0-9]*]*\\\\]`."
+    },
+    {
+      "prediction": "I think it would be better to use `raise_with_backtrace` here.",
+      "reference": "@AndrewRayCode thanks for contributing   Could you write/modify the specs to make sure that this description is working properly?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle redirects in this case, but it's the only way I could come up with.",
+      "reference": "I would have tested 304 only there."
+    },
+    {
+      "prediction": "Why is this timeout set to 10 seconds? Is there a reason we don't want to wait longer?",
+      "reference": "`TimeoutAfter` to be safe."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. If the hook fails, we're going to end up with an uninitialized variable here. I think we need to do this in the `else` branch of the `if` statement above, before the `start_pc` is set to `vsyscall_sysenter_displaced_pc`.",
+      "reference": "So the 32-bit tests on the new kernel hit this case? Won't they fail in debug build?"
+    },
+    {
+      "prediction": "I think you can remove the `type='text/css'` here.",
+      "reference": "Is there a recommended quote style for attributes? I see single and double here, double further down."
+    },
+    {
+      "prediction": "I think this should be `load_time`, not `load_time`",
+      "reference": "The description of `load_time` here is different from the description below..is that intentional?"
+    },
+    {
+      "prediction": "`collections` and `queue_music` are not in the `case` statements above.",
+      "reference": "Actually it could also be a video playlist. But music is used more often."
+    },
+    {
+      "prediction": "I think this should be `!== 0` instead of `=== 0`",
+      "reference": "From the ticket: > I understand why a session ID would be silently omitted for implicit sessions, but what is the reasoning behind omitting it for explicit sessions instead of raising a logic error to the user? So what this change is doing is \"silently omitting\" the session if its an unacknowledged write. I think we want to actually return an error in this case."
+    },
+    {
+      "prediction": "I think this should be `TimeSpan` instead of `double?`.",
+      "reference": "You can simplify this by doing: return (EndTime - StartTime) * 60 / PointsAcrossPeak;"
+    },
+    {
+      "prediction": "This should be `mLineEdit.setFixedWidth ( 10*fm.width ( \"a\" ) );`",
+      "reference": "I think it's the time for bumping `REQUIRED_QT_VERSION` to the last LTS, 5.12. `QFontMetrics::horizontalAdvance()` doesn't exist in 5.10."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this change. I think it should go in `lib/bolt/states` since it's only used by `lib/bolt/states/bolt.rb`, which is in `lib/bolt`.",
+      "reference": "I already set up cd10 for \"inventory_version in google analytics. I've added cd11 for Boltdir Type now"
+    },
+    {
+      "prediction": "I think this should be `ENV.fetch(\"TRAVIS_INTERNAL_RUBY_REGEX\", \"ruby-2.7\")`",
+      "reference": "@henrikhodne We are lazily skipping `jruby-*` here (assuming that JRuby all of a sudden start using these version numbers)."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. It seems to me that if the precision is 2, then the output should be 2 digits after the decimal point. If the precision is 3, then the output should be 3 digits after the decimal point. I'm not sure what precision is being used here, but if it's 2, then the output should be 2 digits after the decimal point. If it's 3, then the output should be 3 digits after the decimal point. I'm not sure what precision is being used here, but if it's 2, then the output should be 2 digits after the decimal point. If it's 3, then the output should be 3 digits after the decimal point. I'm not sure what precision is being used here, but if it's 2, then the output should be 2 digits after the decimal point. If it's 3, then the output should be 3 digits after the decimal point. I'm not sure what precision is being used here, but if it's 2, then the output should be 2 digits after the decimal point. If it's 3, then the output should be 3 digits after the decimal",
+      "reference": "How does this cope with values like `5.001`? Shouldn't that set precision to 0? Instead it is set to 2."
+    },
+    {
+      "prediction": "I think this should be 10, not 15.",
+      "reference": "Rather than having a sleep here, we should check to see if the service we're waiting for is up yet, using something like dockerize. Not critical this second, but would be nice for later."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to run this. I think it would be better to move this to the `swap-package-links` command.",
+      "reference": "I'm not sure if there are any links to swap for Handosntable package. Should this be a top lvl script?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this alignment check. It seems like it should be done in `sigpending_t`'s `init()` method.",
+      "reference": "It seems like it's too big now: can we remove signal_frame_extra_size from line 537? That should only be needed when placing xstate separately. It seems like it isn't needed at all for pending? Also, if we have special heap align forward for us, we don't need this align here either."
+    },
+    {
+      "prediction": "`Get-Command` doesn't have a `Parameters.Keys` property, but it does have a `Parameters.Keys | % { $_.ParameterGroupNames }` property, which should be what you want here.",
+      "reference": "Just double checked the `-in` operator. It's PS3 only, so we might want to change `$_ -in $allowedArgs` to `$allowedArgs -contains $_`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. It seems like it should be in the UIA, not the UIA itself.",
+      "reference": "Might as well use translate here, as @jcsteh suggested"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `plot` module.",
+      "reference": "Definitely an improvement as long as the old tests pass (i.e backwards compatible)."
+    },
+    {
+      "prediction": "I think this is not needed anymore.",
+      "reference": "nit: non functional change"
+    },
+    {
+      "prediction": "I think this should be `os.Getenv(\"HOME\")` instead of `os.Getenv(\"AWS_SHARED_CREDENTIALS_FILE\")`",
+      "reference": "The file is not required - if you're using an IAM instance profile, for example. I'm not sure we can error 100% of the time if it's missing. @randomvariable any suggestions?"
+    },
+    {
+      "prediction": "I think this should be `expr.High.Type().Underlying().Underlying().(*types.Basic)`",
+      "reference": "There is a very similar line `lowType = expr.Low.Type().(*types.Basic)` a few lines above this line. Can you change that in the same way?"
+    },
+    {
+      "prediction": "why do you need this interface?",
+      "reference": "New files should be copyright Hyperledger Besu Contributors."
+    },
+    {
+      "prediction": "I think this should be `execute(\"chmod u+x #{remote_path}\")` and not `execute(\"chmod u+x \\\"#{remote_path}\\\"\")`",
+      "reference": "if I give a non-existent script, then I don't get an errors. I would have expected the `copy` method to raise, but maybe `net-sftp` silently exits?"
+    },
+    {
+      "prediction": "I think this should be `list(to_replace) + list(value)`.",
+      "reference": "@itholic can you also update the parameters in the docs?"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it should be in `SpliteratorProxy`.",
+      "reference": "The super impl Spliterator.getComparator() throws an IllegalStateException by default. Is it really necessary to return null? If null is used somewhere it will throw a NPE, which is roughly the same as throwing an IllegalStateException. I'm just curious - I'm sure there is a reason!"
+    },
+    {
+      "prediction": "I think it would be better to move this to `AvroTable.Builder` instead of `Avro` class.",
+      "reference": "Do we need to do the same in Avro `WriteBuilder` too? I don't think we use that method right now but should make sense for consistency. We already handle that for Parquet."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add this flag. I think it would be better to add it to the Java compiler.",
+      "reference": "can you make all the `_` into `-` to be consistent with the other options?"
+    },
+    {
+      "prediction": "I think this is the wrong place for this interface. It should be in the `cloudformation` package, not the `aws-sdk-go` package.",
+      "reference": "What do you think of renaming this file to `stack.go` or `cfn_stack.go`? `common.go`/`util.go` don't provide us anything descriptive about the contents of the file."
+    },
+    {
+      "prediction": "d7b434c9e5ccb7259bd606e66f9660676afc6692b8cdc6793d190904551d2103b7b22fa636dcbb8208839785ba402ea08fc00c8f1500ccef28bbf599aa64ffb1e1d5dc1bf3420a3777badfe697856e9d52070a50c3ea5821c80bef17ca3acffa28f89dd413f096f898\")]\n #else\n+[assembly: InternalsVisibleTo(\"OpenTelemetry.Exporter.OpenTelemetryProtocol.Tests, PublicKey=00240000048000009400000006020000002400005253413100040",
+      "reference": "It isn't a very effective example if it requires access to the internals   I can't tell just looking at the diff why this is needed, can you provide a little context?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this spec. I think it should go in `spec/spec_helper.rb` under `context \"Bolt::Transport::Local\" do`",
+      "reference": "Why does this include `bolt/inventory`?"
+    },
+    {
+      "prediction": "This should be `Collections.emptyMap()`.",
+      "reference": "We called `t._1.equals(...)` where `t._1` potentially could be `null`."
+    },
+    {
+      "prediction": "nit: `aria-*` should be `aria-`",
+      "reference": "I don't think we should include the `hasAttribute` test here. Even without aria-valuenow, it's still an aria range element. This check is going to make reuse of this function problematic. Better to move the attribute check part outside this function IMO."
+    },
+    {
+      "prediction": "I think this should be `ergonode:fixture:load --file`",
+      "reference": "Shouldn't it be optional?"
+    },
+    {
+      "prediction": "I think this should be `APIGateway` instead of `API Gateway`",
+      "reference": "would be helpful to make these a package level constant so they are accessible to the user. similar to Service Name. Not directly related note, v2 SDK ServiceName should be renamed to ServiceEndpointPrefix."
+    },
+    {
+      "prediction": "I think this should be `file_exists($this->realpath($path)) ? include $this->realpath($path) : null;` instead of `return include $this->realpath($path) : null;`",
+      "reference": "Duplicate call to the method. Better to put in a variable."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `prefetcher.go` under the `Prefetcher` interface.",
+      "reference": "\"and returns that\" -- it looks like this method has no return value."
+    },
+    {
+      "prediction": "I think this should be `assert flip_direction in ('horizontal', 'vertical', 'diagonal', 'none')`",
+      "reference": "Modifications are also needed for PolygonMask."
+    },
+    {
+      "prediction": "I don't think we need to use `internal` here, since it's only used in one file.",
+      "reference": "Does this really make a difference?"
+    },
+    {
+      "prediction": "I think this should be renamed to something like `_root` instead of `_id`.",
+      "reference": "Added a validator for _root_, which is now added automatically since the schema used here declares _root_."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It's not a property of the table, it's a property of the rewrite operation.",
+      "reference": "The other properties are `file-open-cost`, not `open-file-cost`."
+    },
+    {
+      "prediction": "I think it would be better to use `test x\"$have_libunwind\" != xno` instead of `test x\"$enable_easylogging\" != xno`",
+      "reference": "confused by this: what we need here is not libunwind but libunwind-dev right? Why skipping detection?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "just in case helps, it this related with this PR?"
+    },
+    {
+      "prediction": "I don't think this is correct. The `address` key should be a `dict` and not a `list`.",
+      "reference": "This is incomplete I think (at least source_address and ip_address are missing)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to have a `try` block around the `fastSyncDownloader.get().start().whenComplete(this::handleFastSyncResult);` line. If the `fastSyncDownloader.get().start()` fails, then the `handleFastSyncResult` method will not be called.",
+      "reference": "We have a `stop` method here, we should probably be calling that instead of just exiting. There could be important resources to close or cleanup nicely now or in the future."
+    },
+    {
+      "prediction": "I think this should be `string[] = null`",
+      "reference": "fixes ArgNullEx when using both `x-ms-parameter-grouping` and media type `application/xml` in the same operation (issue #2236)"
+    },
+    {
+      "prediction": "I think we should use `ErrCodeBucketNotFound` instead of `ErrCodeNoSuchBucket` here.",
+      "reference": "`NotFound` is a generic error code derived from the HTTP response message's status code, and can be returned for any S3 operation that responds with a 404 status code and no other error code present. Due to this the constant `ErrCodeBucketNotFound`. In addition, these constants are generated based on the API model defined by Amazon S3, modifications to the file directly will be lost during regeneration of the API. With that said, we are looking at ways to work with the Amazon S3 team to better represent this error in a way that is more easily consumed."
+    },
+    {
+      "prediction": "I'm not sure if this is something that should be done at the package level, or if it's something that should be done at the host level. I think it's more likely to be done at the host level, but I'm not sure if there's a better place to do it.",
+      "reference": "@johnduarte I was curious about the case when the operating system was neither `osx` or `win`; in this case, the test will indeed pass, but nothing will have actually really been tested. Should this have a `skip_test` condition at the top?"
+    },
+    {
+      "prediction": "Why not use `log.Fatal()`?",
+      "reference": "I think this example was included by accident? It doesn't seem to belong in this PR (but a separate PR with this would be nice!)."
+    },
+    {
+      "prediction": "I don't think we should change the name of the parameter.",
+      "reference": "Just fixing a bunch of outdated java docs in this file"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to move it to the `TxPool` class.",
+      "reference": "I think it may be safer not to touch Concurrent to normal."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `pool` event handler.",
+      "reference": "is this for everything, or just legacy?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `changeStream` module, since that's where it's used.",
+      "reference": "I think this should be a `MongoError` right?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm also not sure about the other changes in this file.",
+      "reference": "This is currently a db.tar.gz with just one .sql file in it. It might be worth another test (or maybe I'll find one) that has more than one sql file in it."
+    },
+    {
+      "prediction": "I think you can remove the `policy.Spec.Selector = \"has(host-endpoint)\"` line here.",
+      "reference": "Why these changes to existing tests?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `ServiceNameMapping` property on the `SourceInfo` class, and then have the `ConfigurationKeys.ServiceNameMappings` contain a list of `ServiceNameMapping` instances. This would allow us to easily add/remove mappings without having to change the source info class.",
+      "reference": "It should be `IsNullOrWhitespace` I believe, since we're going to trim the value afterwards (and I just realized the other configuration keys have the same issue)"
+    },
+    {
+      "prediction": "I think this should be `target.provided_features + additional_features` instead of `target.features + provided_features`",
+      "reference": "This is just for consistency + searchability with the transports"
+    },
+    {
+      "prediction": "I think this should be `/tmp/kubeedge/` instead of `/tmp/kubeedge/testData`",
+      "reference": "please run gofmt."
+    },
+    {
+      "prediction": "Why did you remove this package?",
+      "reference": "thats kind of a bummer, why do we have to give it full access? if it is just for the private interfaces i'd personally rather have those interfaces pollute the public exported symbols than open the tests up like this."
+    },
+    {
+      "prediction": "Why do we need to check `appNodes` here?",
+      "reference": "Because this is an unexpected situation so can you add a log here to help us figure out what resource is causing this problem?"
+    },
+    {
+      "prediction": "I think this should be `Collections.emptyMap()` instead of `Collections.emptySet()`.",
+      "reference": "just observing that this innocent looking change seems important to this PR. Previously this data had disappeared from the state."
+    },
+    {
+      "prediction": "I think this should be `cond->getOptionalInteger(0) >= errorRowCount`",
+      "reference": "Actually, I do have one question: You mention that each warning is for a different range. Should we add the rowcounts instead of using the max?"
+    },
+    {
+      "prediction": "I think it should be `Name() string` instead of `Name() string`",
+      "reference": "it is confusing to return ID as Name"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The problem is that we don't know what `body` is until we get the response, so we can't tell if it's `None` or something else.",
+      "reference": "body is being used here for the first time without every being populated. This will error. To run tests do `./go clean test_py` and that will run the Firefox tests"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "commit mesasge: I would append \"...Kernel::loadClassCache() method call\""
+    },
+    {
+      "prediction": "I think it would be better to say something like \"If the provided key is none of those, the modifier key cannot be released.\"",
+      "reference": "Keys.COMMAND seems to be an alias to Keys.META. That isn't mentioned?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "You can move this import down to before line 8 (import org.junit.Test;)."
+    },
+    {
+      "prediction": "I think this should be `scanner_type_id` to be consistent with the rest of the code.",
+      "reference": "It's okay to add the `scanner_index_id` here. But we still should keep the `inventory_index_id` reference because it will help the user to know right away, which inventory the violation is coming from, without having to do another lookup."
+    },
+    {
+      "prediction": "Shouldn't this be `afterClickOn(element, driver)`?",
+      "reference": "this shouldn't be beforeClickOn, but rather beforeSubmit? and added to WebDriverEventListener. Since submit does not synthesize the 'click' events, this isn't accurate."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "Should this be I18n'd?"
+    },
+    {
+      "prediction": "I think this should be `ImmutableList.copyOf(fieldErrorLists, 0)`",
+      "reference": "We don't typically use `ImmutableList` to avoid leaking it in methods that are accidentally public or purposely part of the API. I'm +1 for returning `ImmutableList.copyOf(errors)` below, but I don't think we should guarantee the list will be an `ImmutableList`."
+    },
+    {
+      "prediction": "I don't think this is the right test. I think you want to test that `m` and `m2` are the same.",
+      "reference": "@greglandrum Is this test sufficient?"
+    },
+    {
+      "prediction": "I think this should be `GITHUB_CLIENT_ID` and `GITHUB_CLIENT_SECRET` instead of `GITHUB_ACCESS_TOKEN` and `GITHUB_CLIENT_ID`",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Use 2 (not 4) spaces for indentation.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I think Rails complains about the `null: false` part if we ever decide to roll this migration back. Might need to split this up into separate `up/down` methods to handle that."
+    },
+    {
+      "prediction": "I think this should be `seelog.Debugf(\"Managed task [%s]: done waiting for container transitions\", mtask.Arn)`",
+      "reference": "I don't think we need this line. You're logging in `waitForContainerTransition()` anyway"
+    },
+    {
+      "prediction": "I'm not sure this is the right translation. I think it would be better to say `The :other must be :value when :attribute is :value.`",
+      "reference": "Shouldn't the `:attribute` placeholder appear?"
+    },
+    {
+      "prediction": "I think you can remove the `struct` field from the `ResidualVisitor` and just pass it to the `ExpressionVisitors` `visit` method.",
+      "reference": "Could you update this to `structLike` instead of `arg`?"
+    },
+    {
+      "prediction": "I think this should be `var heapStart, heapEnd uintptr = alloc(heapSize)`",
+      "reference": "I only now spot this `freebsd,!baremetal` in the code, which doesn't make a lot of sense. The fact that there is `linux,!baremetal` is because baremetal targets pretend to be linux/arm to get the standard library to compile. Such a `!baremetal` exclusion is not necessary for FreeBSD. But this is not something that needs fixing in this PR."
+    },
+    {
+      "prediction": "I think this should be `/var/tmp/ostree-deltaobj-XXXXXX` instead of `/var/tmp/tmpostree-deltaobj-XXXXXX`",
+      "reference": "Ugh, we were using cwd? =("
+    },
+    {
+      "prediction": "I think this should be removed as it's not used anymore.",
+      "reference": "This can probably be removed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `onResult_` function.",
+      "reference": "put this sentence in `if (onResult_) { }`"
+    },
+    {
+      "prediction": "I think this should be done in a separate PR, as it's not related to the rest of the PR.",
+      "reference": "Will this always be started up as default? Is there any impact to the VM in terms of load and memory usage?"
+    },
+    {
+      "prediction": "I think it would be better to say \"qualifiers that are defined AFTER a parameter set will apply only the the parameter set that precedes them.\"",
+      "reference": "Thanks for fixing the typo's but given that this is a fork of the other commandline.cs file they will still exist there :( Part of the reason forks suck."
+    },
+    {
+      "prediction": "I don't think this is the best way to handle this. I think it would be better to have a `sqlalchemy_connection.py` file in the root of the project that imports `sqlalchemy` and `sqlalchemy.engine`. Then you can just import `sqlalchemy` here and `sqlalchemy.connection` in `sqlalchemy_target.py` instead of `sqlalchemy_connection.py` in `sqlalchemy_target.py`.",
+      "reference": "This doesn't seem related to docs? :)"
+    },
+    {
+      "prediction": "I think this should be `private static boolean notificationDuringQuietTimeDisabled = true;` and not `private static boolean notificationDuringQuietTimeEnabled = true;`",
+      "reference": "Having a negative in the variable/method name makes the code harder to read. Also, you inverted the logic but didn't invert the default value. I suggest sticking to the original name."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `Trace` assembly.",
+      "reference": "Think you need to keep `RawWrite` around as obsolete until NLog6 (Property that just assigns `ForceTraceWriteLine`)"
+    },
+    {
+      "prediction": "I think this should be `/* cache_line_size= */ &cache_line_size, /* icache_line_size= */ NULL)`",
+      "reference": "I would expect this to be named *get* not *set*: it's a query; it's not setting some persistent state."
+    },
+    {
+      "prediction": "I think this should be `from dagster.core.definitions.pipeline_failure_sensor import PipelineFailureSensorContext`",
+      "reference": "maybe should rename this to be `run_status_sensor_definition.py`"
+    },
+    {
+      "prediction": "I think it would be better to have a `SparkWriteOptions` class with a `static SparkWriteOptions` factory method that returns the default options. This way, we don't need to repeat the option names in the SparkWriteOptions class, and we can just use `SparkWriteOptions.fanoutEnabled` instead of `SparkWriteOptions.FANOUT_ENABLED`.",
+      "reference": "Do we want to add the other write options in this PR or keep the refactor separate?"
+    },
+    {
+      "prediction": "I think this should be `if new_token is not None` instead of `new_token is not None`",
+      "reference": "either make this a config or a constant we can define at the top. Burying this in the code is no good."
+    },
+    {
+      "prediction": "I don't think we need to create a new class for this. We can just use the existing `recall_score` and `precision_score` functions.",
+      "reference": "Same about new metrics. Can we avoid adding them?"
+    },
+    {
+      "prediction": "nit: I think you can use `--vpc-name` and `--subnet-name` instead of `--project-id` and `--vpc-name` and `--subnet-name`.",
+      "reference": "I know this is a port from the previous PR, but I am wondering if we can take the chance to improve the naming? `--vpc-host-project-id` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test.",
+      "reference": "The test description says it's \"by reference\". We should change the description"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the `ValidationRules` class.",
+      "reference": "A C# newbie question here: would it make sense to declare Id as an abstract property so any subclass Must have its own Id?"
+    },
+    {
+      "prediction": "I think it would be better to return a `map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]map[string]",
+      "reference": "just looks like you should return an **empty map** with nil error"
+    },
+    {
+      "prediction": "I think this should be `if version.ShouldPrintVersion() || version.PrintOnStderr() {`",
+      "reference": "could you also update the starting log at L146?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the `startUnit` function, but I'm not sure if that's the right place to do it.",
+      "reference": "Also this probably should be `error.As()` or something like it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this command. I think it should go in `selenium-webdriver/src/main/java/org/mozilla/selenium/webdriver/commands/`",
+      "reference": "This is Safari specific and should be defined somewhere in the `safaridriver` namespace"
+    },
+    {
+      "prediction": "I think this should be `statat` instead of `fstatat`.",
+      "reference": "Let's maybe be nice and `unlinkat()` here in the interest of having `/run` be less littered. Or in addition/alternatively, make the file `/run/ostree/initramfs-mount-var` since we already made `/run/ostree/` for the deployment staging bits."
+    },
+    {
+      "prediction": "This should be `#include <boost/numeric/ublas/format.hpp>`",
+      "reference": "Not sure if this doesn't require explicitly adding this library to cmake."
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "we should probably deprecate (and display a warning) the misspelled keyword arg here rather than removing it... and then add the new one. This changes a public API and will break any code that is currently using the misspelled version."
+    },
+    {
+      "prediction": "I think this should be a `FileAlreadyExistsException` instead of an `IOException`.",
+      "reference": "seems reasonable to also want to include the 'to' location?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "This `if` statement can be merged with the above `if opts[:collect_perf_data]`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this case-insensitive match. I think it should be in the `Document` struct, not in the `Document` struct itself.",
+      "reference": "But won't we do case-insensitive matching for all the drivers? For example, we will get the revision field case-insensitively. Let's make this case-sensitive. I think the way the mongo driver works, by lowercasing field names, will be OK with that."
+    },
+    {
+      "prediction": "I think this should be `<h2 class=\"product-title\">...</h2>`",
+      "reference": "I believe this should be on `workshops/show` now, not `products/show`"
+    },
+    {
+      "prediction": "I think this test is failing because it doesn't handle the case where the IP is not a valid IPv4 or IPv6 address.",
+      "reference": "If we return error when we are unable to found country in database, using country detector would be much easier - if error was not returned, that means country was returned :) This doesn't have to be solved in this PR, but since you're adding such case, we can add a `TODO` just to track this :)"
+    },
+    {
+      "prediction": "I think we can remove the `if (string.IsNullOrWhiteSpace(...))` as well.",
+      "reference": ".NET Framework: Fallback to `DD_INTEGRATIONS` if `DD_DOTNET_TRACER_HOME` was not set."
+    },
+    {
+      "prediction": "I think this should be `path_ = paths[0] if paths else b\"\"`",
+      "reference": "Can you use `authority, _, path = rest.partition(b\"/\")` here? That should make stuff a bit cleaner."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `spec/common.rb` or `spec/clients.rb` or something.",
+      "reference": "I don't think this needs to be a float if its a number of seconds... :)"
+    },
+    {
+      "prediction": "This should be `array $values of values`",
+      "reference": "Please add \"Array\" back, right now it's not a proper sentence."
+    },
+    {
+      "prediction": "I think this should be `queryOccurrence = params.getBool(QueryElevationParams.ELEVATE_ONLY_DOCS_MATCHING_QUERY, false) ? BooleanClause.Occur.SHOULD : BooleanClause.Occur.MUST;`",
+      "reference": "@tkaessmann when I ran tests I saw this line has a bug. It inverts the SHOULD and MUST. Hopefully I fixed it and merged it without the bug."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to do this. I think it would be better to do it in the `StartTelemetryURIUpdateService` function.",
+      "reference": "There may be another bug, it looks like we're already in a `if telemetryConfig.Enable` block at this point, maybe that block should be unnested?"
+    },
+    {
+      "prediction": "I don't think we need to check for the existence of `content-type` here, since we are already checking for the existence of `multipart` in line 942.",
+      "reference": "This... looks like it plainly didn't work before? Good catch. I would suggest we change the logic here to 1. Check if `self.headers[\"content_type\"].startswith(\"multipart/form-data\")`, and if that's not the case, add a content-type header with a random (?) boundary. The point here is that if someone assigns to `.multipart_form = ` in their addon, the header should be updated accordingly if the request was not multipart before. 2. *After* that, call `self.content = multipart.encode(self.headers, value)` (as we do right now)"
+    },
+    {
+      "prediction": "I think we should use `require.NoError(registry.Register(account.ProtocolID, account.NewProtocol()))` instead of `require.NoError(registry.Register(account.ProtocolID, account.NewProtocol(), registry))`",
+      "reference": "importShadow: shadow of imported from 'github.com/stretchr/testify/require' package 'require' (from `gocritic`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it's better to use a loop invariant here.",
+      "reference": "Wouldn't it be enough to change the condition in the for loop to `link_state != IBL_UNLINKED`?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "I'm not familiar with signed setups, but this does not seem to offer a way to choose between GPG and signapi verification. Are they usually either both enabled or both disabled? Would this be better suited as a verb with flags for different methods?"
+    },
+    {
+      "prediction": "Why do we need this change?",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this route. I think it would be better to move it to `users/registrations` or `users/sessions`.",
+      "reference": "would it make sense to mock oauth in a way that we direct a user to this endpoint on login (in a test)?"
+    },
+    {
+      "prediction": ".Id + '\" data-toggle=\"modal\" data-target=\"#paperInfo\">' + entry.Overview + \"</button>\", html += '<div class=\"listItemBodyText secondary listItemBodyText-info\">', html += entry.Overview, html += \"</div>\", html += \"</div>\");\n```\nReview comment?",
+      "reference": "Since these two lines use the same logic, it may be cleaner to use an `if` statement rather than duplicating it."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "unnecessary blank line"
+    },
+    {
+      "prediction": "I think this should be `Integer.MAX_VALUE - (101 - 200) = Integer.MAX_VALUE - 200`",
+      "reference": "Nit: no need for \"to\" at the end of the doc because it already uses \"to which\"."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "I would just say \"Signed before publishing.\""
+    },
+    {
+      "prediction": "I think this should be `using UDPv4Transport = eprosima::fastdds::rtps::UDPv4Transport;`",
+      "reference": "We can just include the public UDPv4TransportDescriptor header here ..."
+    },
+    {
+      "prediction": "I think this should be `self.settings.get(\"data-address\", None)`, not `self.settings.get(\"data-address\", None)`.",
+      "reference": "Why do we need this change?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in the `Tag` struct.",
+      "reference": "These three atomic functions done separately are not atomic. There is a possibility of data race, as nothing is guarding Total in between lines 182 and 183. Total got on line 182 may be changed before new value is stored on line 183 by some other goroutine resulting an incorrect value. Mutex should be used."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix, but I don't have a better idea.",
+      "reference": "Is this because you don't have % on your version of Slot? I don't really care either way."
+    },
+    {
+      "prediction": "I think we can remove the `angular.extend` here as it's already done in line 218.",
+      "reference": "Please use var declaration for variables you are declaring. I could not find one for userInjectedExpandableRowHeaderColDef and finalExpandableRowHeaderColDef. Also code styling in the if block needs to be corrected."
+    },
+    {
+      "prediction": "I think this should be `sprintf(errString , \"Error %d creating directory : %s\", errno, cachedLibPath.data());`",
+      "reference": "If cachedLibPath > 200 bytes, it will overflow errString."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `grid-group` component, since it's specific to grid items within a group.",
+      "reference": "Talked this through with a few more folks. I think it would be better to flag prohibited attributes for review, instead of outright failing them. ARIA labels are used fairly liberally. We don't really know if they are actually needed whenever they are used."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. It seems like it should be in the validator, but I'm not sure if that's the right place to do it.",
+      "reference": "Let's keep it to avoid the diff"
+    },
+    {
+      "prediction": "I think this should be `RESERVED_PROPERTIES_SET` or `RESERVED_PROPERTIES_SETTER`",
+      "reference": "nit: not directly related to this PR, but shall we use `ImmutableSet` for consistency?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it's the only way I could get it to work.",
+      "reference": "Layout/DotPosition: Place the . on the previous line, together with the method call receiver."
+    },
+    {
+      "prediction": "I think this should be `final String encryptedInstanceUrl = bundle.getString(AuthenticatorService.KEY_INSTANCE_URL); if (encryptedInstanceUrl != null) { newInstanceUrl = SalesforceSDKManager.decryptWithPasscode(encryptedInstanceUrl, SalesforceSDKManager.getInstance().getPasscodeHash()); }`",
+      "reference": "We send the `instanceUrl` encrypted, but never bothered to decrypt it. I guess it was working because we never did any org split testing where the `instanceUrl` actually changes."
+    },
+    {
+      "prediction": "I think this should be `forCmsSignature` instead of `forBftCmsSignature`",
+      "reference": "Can this be moved to QbftBlockHeaderFunctions class as it is only used for qbft"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "`Tags` now comes from the `PropertyBagHolder` base class."
+    },
+    {
+      "prediction": "I don't think this should be in the base class. It should be in the derived classes.",
+      "reference": "I'm not sure where this should go, but one of the `get_extents` methods should mention that `range_type` can be `'data'` or `'combined'` (are there others?). I found out those are the two expected values by searching the code..."
+    },
+    {
+      "prediction": "I don't think we should be using `org.slf4j` for logging. We should use `org.slf4j.Logger` and `org.slf4j.LoggerFactory` instead.",
+      "reference": "please revert all of this stuff in core.. I'll take another pass after. cheers and thanks for the help!"
+    },
+    {
+      "prediction": "I think this should be `drop table if exists` instead of `drop table if EXISTS`",
+      "reference": "Why was this needed?"
+    },
+    {
+      "prediction": "I think we should be able to get rid of the `hasPrefetched` variable here.",
+      "reference": "I see that CI found a couple test hangs, maybe your `TogglePrefetcher` change below wasn't enough and this still needs to be a `go` invocation for some reason?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could get to see all the properties of a domain config.",
+      "reference": "I do not understand this change"
+    },
+    {
+      "prediction": "nit: I think we can remove the `type:\"string\"` on the `Header1` field here.",
+      "reference": "It would be helpful to clarify this is only for unmarshaling a response. not marshaling a request."
+    },
+    {
+      "prediction": "I think this is the only change that should be in this file. The rest of the changes are unrelated to this PR.",
+      "reference": "I have no explanation for why this test wasn't erroring before... There are two applicable nodes in this tree, one passes, the other fails."
+    },
+    {
+      "prediction": "This should be indented 4 spaces to match the rest of the file.",
+      "reference": "Remove the \"rm\" from the end of \"HAVE_CONFIG_H\""
+    },
+    {
+      "prediction": "I don't think this is a good name for a test. I think it should be something like `FakeStorageFaultSlasher` or something like that.",
+      "reference": "Non-Blocking: This might be too trivial. It doesn't allow us to test that it's being callled."
+    },
+    {
+      "prediction": "I think this should be `return extFormat ? (char *)\"BOOLEAN\" : (char *)\"BOOLEAN\";`",
+      "reference": "Not sure why the ternary expressions are used here, since the true and false results are the same."
+    },
+    {
+      "prediction": "I think it would be better to have a separate function for each database driver. It would make it easier to switch out one of them without having to change all the other functions.",
+      "reference": "Is `vertigo` another name for Vertica databases?"
+    },
+    {
+      "prediction": "I think this should be `urlstr = urlstr.lstrip('mailto:')` and not `urlstr = urlstr.lstrip('mailto:')`",
+      "reference": "That doesn't do the right thing - it strips any of the characters m, a, i, l, t, o and :. It'd probably be cleaner to do this before converting the URL to a string."
+    },
+    {
+      "prediction": "I think it would be better to move this to the top of the function.",
+      "reference": "Why the white space changes?"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `ZoneDB.StartUpdatingName` instead of `ZoneDb.startUpdatingName`.",
+      "reference": "This is now outside the mutex, so could race."
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "hmmm did this pass spotless check? I don't think we typically use wildcard imports"
+    },
+    {
+      "prediction": "I think we should be able to get rid of this `lockBeforeLock *keybase1.LockID`.",
+      "reference": "`lockBeforeLock` -> `lockBeforeGet` (here and everywhere below)."
+    },
+    {
+      "prediction": "I think this test should be moved to `org.pmd.lang.java.test.JavaParserTests`",
+      "reference": "Do we need to internalize net.sourceforge.pmd.lang.java.ast.JavaParserConstants on master, so that we can rename it?"
+    },
+    {
+      "prediction": "I don't think you need to change this line.",
+      "reference": "The GUID here was incorrect, because of that, Vault plugin loading was failing. Should I change to any correct GUID, or it has some special meaning?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "just confirming: if we remove explicit Collect(), then we are relying on the provider dispose... which has a hard-coded 5000 msec to finish flushing. On the other hand, if we keep explicit Collect(), we can pass an explicit timeout to it. (we were not doing it now and was relying on Infinite timeout). Net effect of this change in test code: Previously we would wait indefinitely for Collect() to be over. With this PR, we hope Collect() will be over in 5 secs. If we ever need more than 5 sec, we need to bring back Collect(maxTimeInMilliSecs..)."
+    },
+    {
+      "prediction": "I think this should be `azkaban.logging.kafka.global.disable` and `azkaban.logging.kafka.brokerList` and `azkaban.logging.kafka.topic`",
+      "reference": "What do you think of a name like azkaban.server.logging.kafka.brokerList? This way the name signals that this is a server config."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable across all platforms. I think we should be able to do this in a way that is portable",
+      "reference": "Maybe change into `return _concurrentWrites ?? PlatformDetector.SupportsSharableMutex` ?"
+    },
+    {
+      "prediction": "I don't think we need to escape the `*=` wildcard here.",
+      "reference": "indentation of resulting file is different between 11 and 8 where 8 doesn't indent lines and 11 does"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the error details. I think it should be in the `DecodeError` handler, since that's where the error details will be added.",
+      "reference": "Do we still need to return the `Response` here?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Here `global.VERDACCIO_API_URL` I'd use something different as `global.TEST_VERDACCIO_API_URL` then when you search by `VERDACCIO_API_URL` we don't confuse with `window.VERDACCIO_API_URL`"
+    },
+    {
+      "prediction": "I think it would be better to have this as a separate option, like `--auth-url` instead of `--auth-url`",
+      "reference": "This is totally fine, but consider whether in the future a YAML input format would be better. If so, we can plan for it."
+    },
+    {
+      "prediction": "I think this should be `Conventions` instead of `Conventions`.",
+      "reference": "This should be `DD_TRACE_CONVENTION` to follow our ... conventions."
+    },
+    {
+      "prediction": "This should be `alias_` instead of `*alias_`.",
+      "reference": "Be careful memory leaks. memory leaks occur when getters.getAliasProp == nullptr . right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the per-account cache.",
+      "reference": "Why not just a time.Time?"
+    },
+    {
+      "prediction": "Why do we need this change?",
+      "reference": "Please do not submit a commit that is not part of this PR."
+    },
+    {
+      "prediction": "Is it possible to have an empty description and an empty owner email?",
+      "reference": "This is small braking change: there is no way to clear description any more. If value is empty string, description will not be updated at all. It will affect existing Thrift endpoints also."
+    },
+    {
+      "prediction": "I think this should be `return \"\", \"\", 0, err`",
+      "reference": "Can you open a separate PR with this patch? It seems valuable outside the context of this PR!"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to move it to the `chunking` router.",
+      "reference": "I would suggest to have plurals in the api `\"/chunks/{address}\"`"
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to do, but it seems like it should be changed to `Tuple.Create(LabelAtoms.C15, BioMassCalc.C, BioMassCalc.C15)`.",
+      "reference": "Let's just not do this. We don't have LabelAtoms.O17 here. This is just a short-cut to avoid needing to write out a more verbose function like: 5O\" - 5O i.e. Add 5 x 17O atoms to replace 5 x 16O atoms. This was the original implementation in Skyline before I added the checkboxes to denote simply labeling all atoms in the amino acid."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I'm not sure we should be using the `objectid` as a parameter at all. I think we should just pass the `objectid` into the `approval` function and let it do its thing.",
+      "reference": "1:D100: Docstring missing 23:D400: First line should end with '.', not 'p' 23:D200: One-line docstring should not occupy 3 lines 36:D400: First line should end with '.', not 'n' 36:D200: One-line docstring should not occupy 3 lines 48:D400: First line should end with '.', not 'p' 48:D200: One-line docstring should not occupy 3 lines 63:D400: First line should end with '.', not 'p' 63:D200: One-line docstring should not occupy 3 lines 76:D400: First line should end with '.', not 'p' 76:D200: One-line docstring should not occupy 3 lines"
+    },
+    {
+      "prediction": "I think this should be `file_exists($carousel->getUploadDir() . DS . $carousel->getFile(), true);`",
+      "reference": "Adding a warning or an error in the Thelia log would be a nice idea."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `./service/ampdoc-impl.js` since it's only related to ampdoc.",
+      "reference": "nit: `Set` (and swap `forEach`s for `for (const i of CONNECTION_EVENTS)`)"
+    },
+    {
+      "prediction": "I don't think we need to change anything in this file.",
+      "reference": "nit: Is this `using` statement still needed?"
+    },
+    {
+      "prediction": "I think this should be `string(id.Id)`",
+      "reference": "Dont need to cast `string` -> `string`"
+    },
+    {
+      "prediction": "I think it would be better to move this to the `if err != nil` block above.",
+      "reference": "why doesn't the API return an error in this case?"
+    },
+    {
+      "prediction": "I think this should be `NoRegionError` instead of `BaseEndpointResolverError`.",
+      "reference": "Not sure how particularly useful these exceptions are. Seem to be exceptions required in copying and pasting the EndpointResolver over. I wonder if it makes sense to move these out of ``exceptions.py``"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Can we re-add this? I think that grouping is useful (though would be open to putting them in a mixin or something)."
+    },
+    {
+      "prediction": "I think we can remove the `timeFields` and `partitionFields` and just use `fields` and `schema` here.",
+      "reference": "Overall LGTM, one nit is that I think `partitionFields` here would be good to be renamed so that it's easy to tell it's just for collision detection. Also I wonder if we want to do this check for other transformations too (e.g. bucket, and record numBuckets in the string), so that we might be able to combine `fields` and `partitionFields` into potentially something like a LinkedHashMap?"
+    },
+    {
+      "prediction": "I think this should be `None` by default. If you want to use a service account key, you should set this to the path of the file that contains the service account key.",
+      "reference": "Don't need the `or None` here as the flag will already default to None. Also, `some_dict.get('foo')` will also default to `None`."
+    },
+    {
+      "prediction": "I think this should be `padding: 16px` instead of `padding: 0`.",
+      "reference": "Please, use the default padding here."
+    },
+    {
+      "prediction": "I think we can remove `serverLacksFeature` now, since it's not used anymore.",
+      "reference": "let's actually use the direct include: `require('../core/error').MongoError;`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this setting. I think it should be in the same place as `new-instance-open-target` and `new-instance-open-target.window`.",
+      "reference": "You're missing a space before the `\"` here and below, but I'll fix it up when merging."
+    },
+    {
+      "prediction": "This should be `std::endl` not `std::cout`",
+      "reference": "remove this debugging code?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the same place as `textStarts` and `intStarts`, but I'm not sure if that's a good idea.",
+      "reference": "s/`terms text`/`terms's text`"
+    },
+    {
+      "prediction": "I think it would be better to have this as an option in the `uppy-form` component itself, instead of in the `Form` class. This way, we don't have to add a new option to the `Form` class, and we also don't have to add a new method to the `uppy-form` component.",
+      "reference": ":bike: :derelict_house: , but maybe the default option should be `multipleResults: false` or `combineMultipleResults: false`? i feel like `replaceResultInFormWithNew` is very verbose but also doesn't immediately clarify what it's for."
+    },
+    {
+      "prediction": "I think we can remove this scenario now that we have the email and username form in the previous scenario.",
+      "reference": "Line is too long. [87/80]"
+    },
+    {
+      "prediction": "I don't think you need to change this line.",
+      "reference": "Are you using an automated tool for sorting the imports?"
+    },
+    {
+      "prediction": "I think you can remove the `project_id` argument from this method as well.",
+      "reference": "If this is `project_number` now, can we also update the `id` the description?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be calling `Stop()` on the `deadlineTimer` instead of resetting it to `nil` and then calling `Stop()` again.",
+      "reference": "* This will defer until the function returns. Are you sure this won't happen multiple times. * Alternatively, why is the deadline timer not defined outside the outer loop? * More generally, why are we looping in the first place instead of reading one frame and returning?"
+    },
+    {
+      "prediction": "I think this should be `contentType = \"application/msgpack\";`",
+      "reference": "Should we rename the traces parameter now that this isn't just traces?"
+    },
+    {
+      "prediction": "I think this should be `drmModeAtomicCommit(drm_fd, atom->req, flags, NULL)`",
+      "reference": "I don't think it makes sense to have both `TEST_ONLY` and `NONBLOCK`. We should probably leave `NONBLOCK` out."
+    },
+    {
+      "prediction": "I think this should be `point->surface = NULL;` instead of `point->surface_destroy.link = NULL;`",
+      "reference": "Need to remove the surface destroy listener (and `wl_list_init` it so that `touch_point_destroy` still works)"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "create ticket to investigate this"
+    },
+    {
+      "prediction": "I think this should be `socklen_t *address_length(ipaddress_t* ipaddr)` or `socklen_t* address_length(ipaddress_t* ipaddr)`.",
+      "reference": "If we want to expose this for FFI use in the standard library, it needs to get either a `pony_` or `ponyint_` prefix to its name, for cleanliness of the function namespace. `pony_` means it is a public API meant for use by third party code, whereas `ponyint_` means it is internal. Unless there is a good reason to make it public, we should prefer to keep it internal because public functions require us to treat any changes to them as breaking changes in Pony. So, this function should end up named `ponyint_address_length`, or something else with that prefix."
+    },
+    {
+      "prediction": "I think this should be `OptionsDeviceFuseMountName` to be consistent with the other options.",
+      "reference": "Fuse for shared volumes is a px specific implementation. libopenstorage doesn't know about it, right? So should this be called DeviceVirtualMount (or something similar) instead?"
+    },
+    {
+      "prediction": "I think this should be `[MarshalAsAttribute(UnmanagedType.U1)]` instead of `[MarshalAs(UnmanagedType.U1)]` and remove the `[MarshalAs(UnmanagedType.U1)]` attribute on TargetMethodArgumentsToLoadLength and TargetMethodArgumentsToLoad.",
+      "reference": "`bool` vs `BOOL` in native side :)"
+    },
+    {
+      "prediction": "nit: `const lint = require('../lib/lint')`",
+      "reference": "I think maybe you moved it to util but this is still around and it should get it from util instead."
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "Do we need to test the redirection?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in the `onItemLongClick()` method of the `EpisodesApplyActionFragment`.",
+      "reference": "Tsts, that's why... activity leak about to happen"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "You should do a select here IMO just to be safe so we do not block forever."
+    },
+    {
+      "prediction": "I think this should be `Outbox *core.MessageQueue` and not `MsgQueue *core.MessageQueue`",
+      "reference": "I'll replace this with the actual outbox object soon."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be moved to `AnnotationTest`.",
+      "reference": "Providers are called *before* `setUp()`, so I don't think you can access properties here."
+    },
+    {
+      "prediction": "I think this should be `AttributeTargets.Class | AttributeTargets.Struct | AttributeTargets.Event | AttributeTargets.Property | AttributeTargets.Field | AttributeTargets.Constructor | AttributeTargets.Event | AttributeTargets.Method | AttributeTargets.Enum | AttributeTargets.Property | AttributeTargets.Field | AttributeTargets.Constructor`",
+      "reference": "While this allows it to be applied does it correctly cause the entire class to be skipped?"
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "clean unnecessary modification."
+    },
+    {
+      "prediction": "I think this test should be `expect(<div a=\"a\" />).to.have.property('key').that.is.undefined;`",
+      "reference": "I think these assertions want to be `.to.not.exist` which would pass for `null` or `undefined`"
+    },
+    {
+      "prediction": "I think it would be better to have a separate PR for this. It's not related to the other changes in this PR.",
+      "reference": "Don't think you need this since this is not a URL example."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Maybe a test to check if the generated value is within the boundaries would be nice, what do you think?"
+    },
+    {
+      "prediction": "I think we should use `Encoding.UTF8` instead of `UTF8Encoding`",
+      "reference": "Suppresses UTF-8 BOM in outputs"
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this. I think `Inbound` and `Outbound` would be better names.",
+      "reference": "order wrong for \"outgoing and incoming\""
+    },
+    {
+      "prediction": "I think this should be `private final boolean isMemoryCircuitBreakerEnabled;` instead of `private static final boolean IS_MEMORY_CIRCUITBREAKER_ENABLED = false;`",
+      "reference": "We know it's a boolean and it's in the MemoryCircuitBreaker, why not simply call it `enabled` (like many other Solr plugins do)?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Should team be a \"feature\" that can be \"included\" or a type? In code it seems to me that a `team` flag makes sense."
+    },
+    {
+      "prediction": "I think we can remove the mutex here.",
+      "reference": "Why is this called update params when it's updating project info? Should p called project?"
+    },
+    {
+      "prediction": "I think this should be `ON DUPLICATE KEY UPDATE data=VALUES(data), data_encoding=VALUES(data_encoding)`",
+      "reference": "`upsertHistoryTreeQuery` is a better name for this query now. Is it ok to change history?"
+    },
+    {
+      "prediction": "I think we can remove the `git-hash` file altogether.",
+      "reference": "b did not mean branch - it meant the next version - so probably better to be able to release from a tag on the branch so we can create a hotfix branch of the 1.4.1 tag and tag it 1.4.1b and then version is picked as 1.4.1b"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Keep the mpi.h include, just move it to after adios2.h and guard with the ifdef. Otherwise everything else looks good."
+    },
+    {
+      "prediction": "I think this should be `ms_shared` or `ms_private` or `ms_slave` or `ms_shared` or `ms_private` or `ms_slave` or `slave` or `shared` or `private` or `shared` or `private` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `slave` or `shared` or `",
+      "reference": "Shouldn't this be something like `rootmountmode` to fit the pattern of the other fields' serialized representations?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Why were we initializing here in the first place?"
+    },
+    {
+      "prediction": "I think this should be `internal sealed class NullLogProvider : MvxBaseLogProvider`",
+      "reference": "This should be removed, and instead set the logger to None."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the undo/redo logic, not here.",
+      "reference": "Storing a reference to source data isn't the best choice. Maybe you can find a different way (without storing the reference) to save removed data?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this.",
+      "reference": "Is the tutorial going to move to the new Trafodion website or will it stay on the Confluence wiki? (Check with Gunnar.)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this conversion. It seems like it should be in the `errordetails` package, not the `status` package.",
+      "reference": "maybe call it `fromRetryTaskV2Failure`"
+    },
+    {
+      "prediction": "I think this should be `com.google.api.codegen.transformer.php.PhpGapicCodePathMapper`",
+      "reference": "The diff for this class is difficult to walk through because I reorganized it to be much more clear. There are improvements throughout, but the most important are in the new `createSmokeTest.*` methods."
+    },
+    {
+      "prediction": "`None` is not a valid value for `img_norm_cfg['mean']` and `img_norm_cfg['std']`, so you should use `img_norm_cfg.get('mean', None)` and `img_norm_cfg.get('std', None)` instead.",
+      "reference": "We suggest keeping this logic in lines between 173-175 to restrict its influence."
+    },
+    {
+      "prediction": "This should be `@return string|void`",
+      "reference": "why not ? maybe another phpstan plugin"
+    },
+    {
+      "prediction": "I don't think we need to pass the timeout here, since it's already in the config.",
+      "reference": "That was a really bad CR from me before if I did not spot it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `setup/gcp` directory, under the `forseti-env.sh` file.",
+      "reference": "Does this resolve the scenario when the user-triggered forseti process is running, and it would be killed by the cron job restarting the server?"
+    },
+    {
+      "prediction": "I'm not sure why this was removed.",
+      "reference": "Does this happen somewhere else now?"
+    },
+    {
+      "prediction": "I think we can remove the `videoTestElement.canPlayType` check here.",
+      "reference": "Remove trailing space at this line, amend the commit, and then force-push."
+    },
+    {
+      "prediction": "Can you remove this empty line?",
+      "reference": "This should be grouped together with the import of \"go.opentelemetry.io/{api/core,sdk/export}\" below."
+    },
+    {
+      "prediction": "I don't think you need to change this line. The `rules_filename` should be the same as the `module_name`.",
+      "reference": "What's the impact to people upgrading to the new version? If they don't change their file names, they'll suddenly be broken. While I agree that your naming makes more intuitive sense, I think we need to maintain backwards compatibility. Please choose a new rules file name for the new scanner and keep ke_rules.yaml for the existing version scanner."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `data_spark_columns` list.",
+      "reference": "Shall we add some docstrings clearly to describe when to use which clearly? For example, `spark_frame` now should always be used via `df._internal.applied.spark_frame` for Spark DataFrame APIs that internally creates new query execution plan with the different output length. For expressions and/or functions, `df._internal.spark_frame` should be used together with Spark column instances, in order to avoid the operations on different DataFrames."
+    },
+    {
+      "prediction": "I don't think this is the right place to add this note. I think it should be in the docstring for the `DataFrame` class.",
+      "reference": "nit: `.` at the end of line?"
+    },
+    {
+      "prediction": "I don't think this is the best way to do this. You should be able to get the number of messages in the `MessageManager` by calling `MessageManager.getMessagesInFolder(mCurrentFolder.id)`.",
+      "reference": "I'm not sure I agree that this should take priority over \"Loading\"."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with that doesn't break anything.",
+      "reference": "Was this an intentional deletion? Why not just allow overwrite of `deploy_mode`?"
+    },
+    {
+      "prediction": "I think we can remove the `syncCvr` function and just call `c.syncCvr(cvrGot)`.",
+      "reference": "better to call it as syncCVRStatus.. just syncCVR is very confusing"
+    },
+    {
+      "prediction": "This is a bit of a nitpick, but I think it would be better to have a `deleteApp` function that takes in the `deleteAppOpts` as an argument. This way, we don't have to pass in both the deleteAppVars and the deleteAppOpts in the same function.",
+      "reference": "should we make this not a pointer? so that when we embed the struct a copy is made instead of modifying the original object."
+    },
+    {
+      "prediction": "I think this should be `kinto.redis.storage` and `kinto.redis.cache` and `kinto.redis.permission`",
+      "reference": "Does this mean that `kinto_redis` is required to run tests?"
+    },
+    {
+      "prediction": "I think we can remove `originalZeroIndex` and `originalIndexes` from `mergeBatchResults`",
+      "reference": "will this ensure the indexes for ordered writes as well?"
+    },
+    {
+      "prediction": "I think this should be `('ByteArray', 'Int16Array', 'UInt16Array', 'Int32Array', 'UInt32Array', 'Int64Array', 'UInt64Array', 'Real32Array', 'Real64Array', 'BoolArray')` instead of `('ByteArray', 'Int16Array', 'UInt16Array', 'Int32Array', 'UInt32Array', 'Int64Array', 'UInt64Array', 'Real32Array', 'Real64Array', 'BoolArray')`",
+      "reference": "Put `BoolArray` first?"
+    },
+    {
+      "prediction": "I think this should be `import('./standalone');` and not `import('./site');`",
+      "reference": "I believe this whole file can be deleted now."
+    },
+    {
+      "prediction": "I think it would be better to use `uint64` instead of `int64` here.",
+      "reference": "We should specify format for a user. I.e: Integer, 1000 == 1 MYST"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way I could get this to work.",
+      "reference": "Does the errors returned from these controllers indicate which controller threw the error? If not, there might be some value in making the \"CustomControllers\" type a `map[string]func(ctx context.Context, sc *server.Context) error` with the name of the controller as the key and include the key in this error string. This would apply for below as well."
+    },
+    {
+      "prediction": "I think this should be `builder_.FinishSizePrefixed(Offset<Table>(toff), file_identifier_.length() ? file_identifier_.c_str() : nullptr);` and not `builder_.FinishSizePrefixed(Offset<Table>(toff), file_identifier_.length() ? file_identifier_.c_str() : nullptr);`",
+      "reference": "maybe at least pull the file identifier arg out of the if?"
+    },
+    {
+      "prediction": "I think this should be `self._tabbed_browser.moveTab(cur_idx, new_idx, icon, label)`",
+      "reference": "You can most likely remove that (and the try/finally) too, as we shouldn't have any flickering from removing/inserting tabs anymore now."
+    },
+    {
+      "prediction": "I think it would be better to return `this._localHooks[key]` instead of `this._localHooks[key].push(callback)`",
+      "reference": "Could you add `s` to `@return`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Selenium::WebDriver::Chrome::Selenium::Remote::Chromium` module.",
+      "reference": "Moving this code inside the last case statement doesn't seem right. Why are we defining `name` and `is_relative` variables there if we aren't using them anywhere?"
+    },
+    {
+      "prediction": "I think we should use `err` instead of `ok` here. `ok` means the message is a byte slice, but `err` means the message is a string.",
+      "reference": "Will cause any bugs without this change? I've seen this code block(L293-302) many times... ;-)"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I'm not sure what the best way to do this is, but I don't think we should be adding properties to the context that aren't added to the context at all.",
+      "reference": "You are opening a door to race-condition-hell by returning an unprotected dictionary. I recommend that you return `IReadOnlyDictionary` that only works on the platforms where it is known."
+    },
+    {
+      "prediction": "Why is this test skipped?",
+      "reference": "also seems we should not skip this test"
+    },
+    {
+      "prediction": "I think this should be `DeleteFile[] deletes = deletes != null ? deletes : new DeleteFile[0];`",
+      "reference": "This is mostly for my understanding: is `DeleteFile[] deletes` a mandatory builder param now for file scan tasks? If not, from a v1 / v2 compatibility standpoint would it make sense to add an overloaded constructor?"
+    },
+    {
+      "prediction": "I think it would be better to have a method `fetchAgedQueuedFlows(long minAge, String projectId, String flowId)` instead of `FetchAgedQueuedExecutableFlows.FETCH_FLOWS_QUEUED_FOR_LONG_TIME`",
+      "reference": "Can the error message reflect the purpose of the query more closely? Something like \"Error fetching executions queued for a long time\""
+    },
+    {
+      "prediction": "I think this should be `ci.StartupScript != \"\" && sourceExists(ci.StartupScript)`",
+      "reference": "Are you making it so startup script HAS to be in sources?"
+    },
+    {
+      "prediction": "I think this should be `|| vals.LocalWidth() == 1 || vals.LDim() == vals.LocalWidth() || vals.LocalHeight() == 1`",
+      "reference": "Might be more clear to just have one if statement?"
+    },
+    {
+      "prediction": "I think it would be better to use `traceback.format_exc()` instead of `exc_info=True`",
+      "reference": "remember that logger methods will do string interpolation automatically anyway, so you should be able to do `logger.error('message %s', var, exc_info=True)`"
+    },
+    {
+      "prediction": "I think this should be `render_args_struct` instead of `render_args_structs`.",
+      "reference": "Perhaps rename to `render_service_args_struct`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems to be the only place where crypto2 is used in this file.",
+      "reference": "clean the name"
+    },
+    {
+      "prediction": "I think we can remove this now, right?",
+      "reference": "I think we would expect StatusNotFound instead of StatusForbidden?"
+    },
+    {
+      "prediction": "This should be `snprintf (rankstr, sizeof (rankstr), \"%d\", s->rank);`",
+      "reference": "Since this is a recurring theme, would it make sense to have an ioencode interface for it like `ioencode_rank()` that takes an integer rank like before?"
+    },
+    {
+      "prediction": "I think this should be `agent.dockerClient` and not `agent.dockerClient`",
+      "reference": "Is it a concern that we initialize these fields irrespective of whether resources like cgroup/volumes are enabled or not?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in `OutboundMiddleware` itself.",
+      "reference": "Wouldn't this be the first panic in yarpc? What do we do for transport validation? Return errors?"
+    },
+    {
+      "prediction": "I think this should be called `cancel` instead of `cancelReason`.",
+      "reference": "have you envisioned other cancellation reasons?"
+    },
+    {
+      "prediction": "I think this should be `array<string, string|int>`",
+      "reference": "For the record, class-string is now suported by PHPStorm. Don't know if it's worth moving it to `@phpstan-param` then."
+    },
+    {
+      "prediction": "I think this should be `cmd.exc.CommandError` instead of `cmdexc.CommandError`.",
+      "reference": "This prevents negative IDs, but it doesn't check for too large IDs. You should probably just do the same check you do for `QWebengineScript` here as well."
+    },
+    {
+      "prediction": "Nit: I think it would be more consistent to call this `tags` rather than `awsTags` as `aws` isn't the name of the AWS service.",
+      "reference": "This might be very confusing. Maybe `resource-tags`? I"
+    },
+    {
+      "prediction": "I think this should be `self.dilation = _pair(self.dilation)` and not `self.dilation = _pair(self.dilation)`",
+      "reference": "We may use `_pair` to wrap `dilation`."
+    },
+    {
+      "prediction": "I think this should be `maskedPaths = []string{\"/proc/kcore\", \"/proc/sys\", \"/proc/sysrq-trigger\", \"/proc/irq\", \"/proc/bus\"}` instead of `readonlyPaths = []string{\"/proc/kcore\", \"/proc/sys\", \"/proc/sysrq-trigger\", \"/proc/irq\", \"/proc/bus\"}`",
+      "reference": "what's with `setReadonly`? Why we deleted it totally?"
+    },
+    {
+      "prediction": "I think this should be `const char * szTypeRef0 = \"typename\";`",
+      "reference": "Currently I think this is not acceptable. I think the value should be chosen by the author of the parser. (Personally \"type\" is better than \"typename\".)"
+    },
+    {
+      "prediction": "I think we can remove the ec2 sdk import here.",
+      "reference": "Is this excessive?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that the initramfs is mounted as read-write, so the initramfs won't be able to write to the /etc directory. I think you need to do something like this: `if (mount (\"etc\", \"/etc\", NULL, MS_BIND, NULL) < 0) {` and then `if (mount (\"/etc\", \"/etc\", NULL, MS_BIND, NULL) < 0) {` and then `if (mount (\"/etc\", \"/etc\", NULL, MS_BIND, NULL) < 0) {` and then `if (mount (\"/etc\", \"/etc\", NULL, MS_BIND, NULL) < 0) {` and then `if (mount (\"/etc\", \"/etc\", NULL, MS_BIND, NULL) < 0) {` and then `if (mount (\"/etc\", \"/etc\", NULL, MS_BIND, NULL) < 0) {` and then `if (mount (\"/etc\", \"/etc\", NULL, MS_BIND, NULL) < 0) {` and then `if (mount (\"/etc\", \"/etc\", NULL, MS_BIND, NULL) < 0) {` and then `if (mount (\"/",
+      "reference": "Ohhhh I see, this change was previously having *no effect* - I had thought you meant we were doing something like bind mounting the initramfs' `/etc` as the real `/etc` but we'd clearly notice if that happened, we'd be missing all sorts of config files etc."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it would be better to put it in the `SnapshotWriter` class.",
+      "reference": "I think this needs a more specific name, like `sequenceNumberOverride`"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Ought we just raise an exception if the adapter isn't defined?"
+    },
+    {
+      "prediction": "I think we can remove this now that we have `ap-southeast-1` and `ap-south-1` in `stsLegacyGlobalRegions`",
+      "reference": "Should this have been removed? We still set the region to \"aws-global\" in v3model.go#L115"
+    },
+    {
+      "prediction": "I think this should be `qs` instead of `querystring`.",
+      "reference": "You could shorten this to `qs or None`. But why not just build a dict of parameters we want to include and only add `querystring` if there's something here, similar to the way you do in the error view?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to just acquire the lock and not worry about releasing it.",
+      "reference": "sorry if I'm missing it somewhere else in the code, but do you need to create a `ticker := time.NewTicker`?"
+    },
+    {
+      "prediction": "nit: I think you can remove the `dataresidencySs` variable and just pass it in as an argument to the constructor.",
+      "reference": "Inline this, as we don't use it again."
+    },
+    {
+      "prediction": "I think this should be `ScanlineFilter` instead of `ScanlineFilterCommon`",
+      "reference": "I think the instantiation of the new objects i.New() and T.New() need to be done in a function to have local variable and reduce the amount of memory used."
+    },
+    {
+      "prediction": "This should be `0.0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "This is probably more digits than needed - perhaps rename OneOverK0 to IonMobility instead"
+    },
+    {
+      "prediction": "I think this should be `FlattenedMethodAsyncCallSettings` and `FlattenedMethodAsyncCancellationToken` instead of `PagedOptionalArrayMethod` and `FlattenedMethodAsyncCancellationToken` respectively.",
+      "reference": "End with 'Method', so: - FlattenedAsyncCallSettingsMethod - FlattenedAsyncCancellationTokenMethod"
+    },
+    {
+      "prediction": "I think this should be `gocloud.dev/internal/secrets`",
+      "reference": "`localsecrets` should be updated in the same way."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "We'll want to PR a spec change into github.com/algorandfoundation/specs. (Side note: It might or might not make sense to combine this protocol upgrade with Tsachi's protocol upgrade for fixing the reward rate calculation.)"
+    },
+    {
+      "prediction": "I think this is the wrong place to do this. This function is called from the worker, so it should be called from the worker and not the worker and the worker should be the one that generates the blocks.",
+      "reference": "Use `FullBlock`, it comes from the same package."
+    },
+    {
+      "prediction": "I don't think we need to import `rwlock` here.",
+      "reference": "do we plan to use anything else than flask for making HTTP requests inside localstack? if so, it maybe makes sense not to strongly couple to flask for now, and just leave the type of the `request` function parameters open for now."
+    },
+    {
+      "prediction": "I think this should be `errors.New(\"Writer.As failed to get Writer\")`",
+      "reference": "I think we should make this a separate test case: it seems like making the precondition fail is the test case that would ensure that this escape hatch worked. Otherwise, if it's always true, then it would be the same as if the escape hatch didn't modify the outgoing request."
+    },
+    {
+      "prediction": "I think it's better to use `s.index` here, so that we don't need to create a new `s` object every time.",
+      "reference": "Error return value of `s.idx.BuildIndex` is not checked"
+    },
+    {
+      "prediction": "I think this should be `yield functools.partial(member.__call__, fields=['a', 'b'])` instead of `yield functools.partial(member, fields=['a', 'b'])`",
+      "reference": "What's the reason for excluding those from the fuzzing here? You probably could just add some inner type via `functools.partial` like below with `FormatString`."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `kill` instead of `SIGKILL`",
+      "reference": "Its this golint or govet stuff?"
+    },
+    {
+      "prediction": "I think this should be `dr_get_current_drcontext()` instead of `dr_mcontext_t mcontext;`",
+      "reference": "These are good cleanups: I didn't realize there were so many like this in the samples and tests. But given that there are quite of few of these mcontext changes I would separate these into their own PR (no need for an issue: iX branch) since they are logically distinct. Cleaner history, simpler revert paths, etc."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this benchmark. I would expect it to be in `BenchmarkTests` or `BenchmarkTests<Color>`, but not `BenchmarkTests<Color>(ColorEquality)`. What do you think?",
+      "reference": "This is just a random struct equality, no value here."
+    },
+    {
+      "prediction": "I think we should rename this to `NatPortRange` to be consistent with the other configs.",
+      "reference": "Just spotted the `local` on here; that shouldn't be needed - no reason to limit this config to env vars only"
+    },
+    {
+      "prediction": "I don't think we need to import `process_dimensions` here, since it's imported in `..util` below.",
+      "reference": "I had some weird issues when importing ``from .. import util`` getting the wrong utilities, hence I did this."
+    },
+    {
+      "prediction": "This file is not covered by the GNU General Public License. See the `LICENSE` file in the project root for a full list of individual source files.",
+      "reference": "Please remove this line."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "should this state that we're falling back to `:acceptable_exit_codes`?"
+    },
+    {
+      "prediction": "`value.dtypes` is a `pandas.core.series.Series` object, not a `pandas.core.dataframe.DataFrame` object.",
+      "reference": "I don't think this works. The dataframe object is encapsulated in the value parameter. This might be the root of the failing checks."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check if there's only one server. I think it would be better to check if the number of servers is equal to 1 or not.",
+      "reference": "Why was it necessary to add this check against `this.servers.size`?"
+    },
+    {
+      "prediction": "I think this should be `<= 0` instead of `< 0`.",
+      "reference": "This tests for < 0 but the error message says it must be > 0. The `if` should probably be `<=` to match the message."
+    },
+    {
+      "prediction": "I think we can remove the `HttpWebRequest` check here, since it's already checked in the `IsTracingEnabled` method above.",
+      "reference": "Short circuits when it's something like a WebPackRequest, which we should instrument and test specifically."
+    },
+    {
+      "prediction": "Why did you change the name of this flag? I think it should be `flag.DiskCertCache`.",
+      "reference": "Should this be true now by default?"
+    },
+    {
+      "prediction": "I think it would be better to put this under `ExecutorServlet` since it's not specific to Azkaban.",
+      "reference": "/DispatcherStatistics ? /statistics is a bit confusing with /stats"
+    },
+    {
+      "prediction": "I think this should be `or \"tail: unrecognized file system type\" in raw_message`",
+      "reference": "nit: Wondering if we should simply filter on `\"tail: unrecognized file system type\"`, or is the type identifier `0x794c7630` always the same?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/installer` since it's only used by `pkg/installer/installer.go` and `pkg/installer/installer_test.go`, which are in `pkg/installer`.",
+      "reference": "I think the goal of using `fake-cluster-id` was to make it sure clear that this is a fake cluster, replacing this with UUID only now makes these clusters look very similar to real ones which can cause problems.. any reason why we didn't go for the original recommendation from slack thread of `fake-cluster-UUID` @twiest"
+    },
+    {
+      "prediction": "I think it would be better to use `get_node_first_ancestor_of_type_and_its_child_of_type` instead of `get_node_first_ancestor_of_type_and_its_child`",
+      "reference": "I need to spruce up my `mypy` knowledge but I think this is incorrect. Using `T_Node` twice makes it so that both should be the same type. I think this would be better: `Tuple[Optional[T_Node], Optional[nodes.NodeNG ]]`. It might even better to do: `Union[Tuple[None, None]. Tuple[T_Node, nodes.NodeNG ]]` as that shows that `None` should always pair with `None`"
+    },
+    {
+      "prediction": "I think this should be `controller.Sync(context.Background(), test.certificateRequest, nil)` and the next line should be `err = controller.Sync(context.Background(), test.certificateRequest, errors.Wrap(err, \"unexpected error\"))`",
+      "reference": "Nit: below on line 449 we use `%v` to print an error - perhaps would be nice to do it in the same way in both places (but probably not that important). Similar in other places in this PR."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `addons.py`.",
+      "reference": "I'm not against adding a kill event, but we need to do it properly, please. That means listing it in events.py, adding docs for it, and so forth."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `cpp_generator.cpp` file.",
+      "reference": "this line can be lifted outside of the if-then? In fact only CreateString / CreateSharedString needs to be inside of it."
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "I think the `EpisodesApplyActionFragment` is unused now. So please delete it :)"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "It seems like we should leave `notice` here since it can be configured in Bolt config files, and that's what sets the console config for messages from Puppet. For example, this will break plans that call `notice()` in their apply blocks."
+    },
+    {
+      "prediction": "I think this meter should be added to `CommonMetrics` instead of creating a new one.",
+      "reference": "This metrics is only exposed in executors. Should it be defined in azkaban.execapp.ExecMetrics instead?"
+    },
+    {
+      "prediction": "I think we can remove this `safeDateFormat` function and use `parseDate` directly in the `getEffectiveMaxDate` function.",
+      "reference": "I originally factored out these functions from date_input. Currently, they are only used in datepicker, but they seemed more general purpose. If you decide we should keep the current date_input API (in case anyone is using it directly), then I think it makes sense to factor these out in order to be able to maintain consistency, but otherwise they could be inlined into datepicker.jsx if you prefer."
+    },
+    {
+      "prediction": "I think this should be `enumspb.CONTINUE_AS_NEW_INITIATOR_UNSPECIFIED` instead of `enumspb.CONTINUE_AS_NEW_INITIATOR_CRON_SCHEDULE`",
+      "reference": "I think this should never happen and should return an error also. Having `WorkflowBackoffTimerTask` with `UNSPECIFIED` type looks weird."
+    },
+    {
+      "prediction": "I think it would be better to check if `e.keyCode === 32` instead of `e.keyCode === 13` and `e.keyCode === 32 && browser.tizen`",
+      "reference": "Add `browser` import"
+    },
+    {
+      "prediction": "I think this should be `GenerateSHA256PasswordHash` instead of `GenerateSHA256PasswordHashes`",
+      "reference": "Would it be good practice to add a nonce in here (that's also stored in the config), so it's harder to tell when passwords are being reused?"
+    },
+    {
+      "prediction": "I think this should be `@Produces(MediaType.APPLICATION_FORM_URLENCODED)` instead of `@Produces(MediaType.APPLICATION_JSON)`",
+      "reference": "For consistency, the return type could be `TelemetryDataDto` here. This would be more consistent with the other REST API interfaces. From a functional perspective, the current code is perfectly fine though."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `SolvingDriver<float>` is a `shared_ptr`, so when you pass it to a function that takes a `shared_ptr`, you're allocating a new one every time. I think the right fix is to make `SolvingDriver<float>` a member of `SGDSolver<float>`, and then pass a pointer to it to the `SolvingDriver<float>` constructor.",
+      "reference": "I've never seen this `boost::make_shared`...why use it instead of `reset`?"
+    },
+    {
+      "prediction": "I think this method should be renamed to `getDispatchMethodForFlow()`.",
+      "reference": "Wasn't this added as a part of different PR already?"
+    },
+    {
+      "prediction": "Why do you need this array?",
+      "reference": "hmmmmm, should not be in constructor in this case `$this->toDelete = [[]];` ??? maybe not"
+    },
+    {
+      "prediction": "I think this should be `if response.get('ResponseMetadata', {}).get('HTTPStatusCode', 'Unknown') == 200:`",
+      "reference": "We can just update it to be as follows: `if response: print(\"Success\") else: print(\"Error\")`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to store the state. I think it would be better to store the state in the server block.",
+      "reference": "Maybe this should be exported so (external/third-party) plugins can also use it."
+    },
+    {
+      "prediction": "Why do we need to call `service.ipForward.Enable()` again?",
+      "reference": "Warning logging whould be exactly here in high logic. So that everybody understands why we swallow errors"
+    },
+    {
+      "prediction": "I don't think we need to pass in the url here, since it is passed in in the constructor of `JsonRpcWebSocketProcessor`",
+      "reference": "if its optional, move it to last item an use JsonRpcUrl? url = null"
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this method, but it's the only one I could come up with when I was trying to think of a better name for this method.",
+      "reference": "how is this different from the `client_slug` method already available on a `user` ?"
+    },
+    {
+      "prediction": "I think this should be `LOG.info(String.format(\"Connecting to %s and %s\", publishConnection, subscribeConnection));`",
+      "reference": "I'd keep this at `info` level..."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "I believe this env var won't be set anywhere. Do we need this condition?"
+    },
+    {
+      "prediction": "nit: `window.getParameterByName` -> `getWindowLocation.getParameterByName`",
+      "reference": "why add function explicitly?"
+    },
+    {
+      "prediction": "I think this should be `return &Sublist{root: newLevel(), cache: make(map[string]*SublistResult)}`",
+      "reference": "Is that intentional? If you don't create the cache here, it will never be created therefore used."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it seems to work.",
+      "reference": "I think we can now remove this test case since we can ensure that multiple calls to `Fetch` will double the interval on each call"
+    },
+    {
+      "prediction": "I think this should be `/__temporal_sys/` instead of `/__cadence_sys/`",
+      "reference": "Not part of this PR but this const seems to be defined in multiple places. Needs to be extracted somewhere."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Remove this file from the PR"
+    },
+    {
+      "prediction": "I think this needs to be changed to `ConstantinopleInstance` instead of `ConstantinopleFix`",
+      "reference": "need to add it for ropsten, rinkeby, goerli as well"
+    },
+    {
+      "prediction": "I think we can remove the `v.Command` field from the `VolumeProperty` struct now, since it's not used anymore.",
+      "reference": "G204: Subprocess launching should be audited (from `gosec`)"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "curious why only making these 3 public. what about other package scope methods like `times`, `timestamps`, `array`?"
+    },
+    {
+      "prediction": "Please add a license header here.",
+      "reference": "The Email field is not translated into Russian, and into any other. It's kind of a \"standard\" word. Therefore, we may face the fact that you have to create files with exceptions for all languages. @caouecs, what do you say?"
+    },
+    {
+      "prediction": "I think this should be `boost::program_options::value<std::size_t> (0, boost::program_options::value_options::read_only | boost::program_options::value_options::min(1, 256000), \"\")`",
+      "reference": "Should it say `default 0` here? looks a bit confusing. Also following BODMAS we should put: `2048 * (signature checker threads + 1)`"
+    },
+    {
+      "prediction": "I think this should be `from ruamel.yaml import yaml`",
+      "reference": "Add a newline. We group by 1: standard library, 2: 3rd party library, 3: application library"
+    },
+    {
+      "prediction": "I'm not sure if this is the right endpoint to use for the range query. I think it would be better to have a separate endpoint for the range query. The reason is that the range query endpoint is only available if the Prometheus agent is running in a Kubernetes cluster. If you don't have access to the Kubernetes cluster, then you can't run the range query endpoint, which means you can't use it to validate the range query.",
+      "reference": "So `queryRange.Step` will be ignored?"
+    },
+    {
+      "prediction": "I think this should be `#if defined(_X86_) || defined(_AMD64_) || defined(_ARM_)`",
+      "reference": "It would be nice to replace these _HOST_* defines with HOST_*."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to replace the `rootDir` with the `rootDirWithSeparator` in the first place. The `rootDirWithSeparator` should be the same as the `rootDir` but with the `rootDir` replaced with the `rootDirWithSeparator`.",
+      "reference": "The problem here was that the record file started with something like `/tmp/whatever/myproj` and was scrubbed to `[ROOTDIR]/myproj`, but the `/` is a `\\` on Windows."
+    },
+    {
+      "prediction": "This should be `using ParameterList = eprosima::fastdds::dds::ParameterList;`",
+      "reference": "Don't use relative paths. src directory is already on the include search path"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be under `builds` since it's about the build itself, not the build target itself.",
+      "reference": "if we're introducing this now can we make this private?"
+    },
+    {
+      "prediction": "I think this should be `ScanProvider.FileExists(_documentFilePath, f.FilePath)` instead of `ScanProvider.FileExists(_documentFilePath, f.FilePath)`",
+      "reference": "Something is wrong with the indentation here."
+    },
+    {
+      "prediction": "I think you can remove this line.",
+      "reference": "This should go to the external imports group below"
+    },
+    {
+      "prediction": "I think this should be `log.Println(\"processing yaml folder\", p)`",
+      "reference": "Intended to be checked in?"
+    },
+    {
+      "prediction": "Shouldn't this be `LOGGER.warn(\"There are schemas only existing in service center: {}, which means there are interfaces changed. It's recommended to increment microservice version before deploying.\", scSchemaMap.keySet());` ?",
+      "reference": "I think this warning message is the same as above one. Anyway, it's fine to keep it."
+    },
+    {
+      "prediction": "I think this should be `Path.Combine(logsDir, Path.GetFileName(fileName))`",
+      "reference": "This is done few times, replace with some well named method"
+    },
+    {
+      "prediction": "I think it would be better to use `List<String>` instead of `String[]` here.",
+      "reference": "`configs` isn't very descriptive, especially when there are other config-ish things like `generatorConfigs`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move this test to `gatling/test/integration/src/test/java/org/apache/gobblin/gatling/test/integration/TestGatlingExecutor.java`",
+      "reference": "In my point we have to check conversion with specific (non-ASCII) characters in unicode string."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `tests/unit/` since it's a unit test.",
+      "reference": "Given that file still contains doctests (and should still contain doctests), they should be run. Please turn this back on."
+    },
+    {
+      "prediction": "I think this should be `public static final String ATTR_SELF_SERVE = \"selfServe\";`",
+      "reference": "we should keep the camel case format - selfServe"
+    },
+    {
+      "prediction": "I think we can remove the `metadata` variable and just use `metadataItems` here.",
+      "reference": "consider the alias `compute` in case this ever references other apis."
+    },
+    {
+      "prediction": "I think this should be `cgroups.PathExists(filepath.Join(m.paths[\"devices\"], dirName))`",
+      "reference": "It probably doesn't matter in practice, but we're not supposed to access a map without holding a lock. This is why I have suggested using `m.Path(\"devices\")` earlier -- it takes a lock before accessing m.paths. Alternatively, you can leave this code as is but add taking a lock (same as `Path()`)."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `baseModuleStore.js` since it's only used by `baseModuleStore.js`.",
+      "reference": "Since `ga` is an acronym here, we should rename the generated actions and selectors to adhere to our naming conventions as it will no longer be capitalized properly. See below for how this is already done for the amp container ID settings. We should add `GA` to our list of checked acronyms as well in `packages/eslint-plugin/rules/acronym-case.js`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in the `emit` block above.",
+      "reference": "uploading_finished and upload_meta_data_extracted can fire very quickly after another, and there is not much difference in the Assembly status that's useful to us. I kept only the Assembly fetch after metadata is extracted, which ensures that we'll have all the correct `uploads.*.meta` properties on the client side."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check if we're in rootless mode.",
+      "reference": "Does it make sense to keep trying to read the file every time the function is called, or should we maybe use `sync.Once()` here?"
+    },
+    {
+      "prediction": "This should be `uint64` instead of `int64`",
+      "reference": "`DbCapacity` -> `DBCapacity`"
+    },
+    {
+      "prediction": "I think this should be `0x757a1c47` instead of `0x757a1c470000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "For Berlin I confirmed all the fork hashes with the Geth team. Would you do the same with Martin? He responded quickly the last time."
+    },
+    {
+      "prediction": "I think this should be `cmdParts := strings.Split(cmd, \" \")` instead of `cmdParts := strings.Split(cmd, \" \")`",
+      "reference": "Does `sudo` required for `darwin` only? Don't we need it for `linux` too?"
+    },
+    {
+      "prediction": "I think this should be `AWS::CloudFormation::Stack`",
+      "reference": "Can you add description to this exported struct?"
+    },
+    {
+      "prediction": "I think this should be `component.base || component.nextBase`",
+      "reference": "Although this is needed for `getSnapshotBeforeUpdate` this also fixes a bug with `componentDidUpdate`. During rendering, the `state` variable is mutated. This has the consequence, that `previousState` would never hold the previous state, but the most current one."
+    },
+    {
+      "prediction": "elb is not used in this PR, can we remove this line?",
+      "reference": "Should this maybe directly start with ELBv2 aka NLBs or ALBs?"
+    },
+    {
+      "prediction": "I think this should be `else if` instead of `else`",
+      "reference": "This was the request that was often causing the test to fail as an unexpected API failure (or at least one of them   )."
+    },
+    {
+      "prediction": "I think it would be better to have this as an instance property of the `NetworkError` class, so that we don't need to pass the `xhr` around.",
+      "reference": "I understand why `xhr = null`, but I think adding `error = null` implies that this method should work even if we don't pass the `error` argument. Should we remove it?"
+    },
+    {
+      "prediction": "I think it would be better to pass the mapping as a parameter to the constructor. This way you don't have to pass in the mapping when you don't want to, and you don't have to pass in the mapping when you do want to. I think it would be better to pass the mapping as a parameter to the constructor. This way you don't have to pass in the mapping when you don't want to, and you don't have to pass in the mapping when you do want to. I think it would be better to pass the mapping as a parameter to the constructor. This way you don't have to pass in the mapping when you don't want to, and you don't have to pass in the mapping when you do want to. I think it would be better to pass the mapping as a parameter to the constructor. This way you don't have to pass in the mapping when you don't want to, and you don't have to pass in the mapping when you do want to. I think it would be better to pass the mapping as a parameter to the constructor. This way you don't have to pass in the mapping when you don't want to, and you don't have to pass in the mapping when you do want to. I",
+      "reference": "It is not your code, and it is used like this all over the place in gapic generator, but the general rule is we should prefer using the least specific class/interface in the hierarchy, which still satisfies our requirements. Here, unless we really need anything specific from `ImmutableSet`, please use `Set` (less specific type) instead."
+    },
+    {
+      "prediction": "This should be `ctx->aws_sts_endpoint = (char *) tmp;` instead of `ctx->aws_sts_endpoint = (char *) tmp;`",
+      "reference": "Sorry... I just realized... since we use sts_endpoint in the EKS Provider, this error message is no longer true. Role_arn is not required."
+    },
+    {
+      "prediction": "I think this should be `static <T> Tree<T> fill(int n, T element) { return io.vavr.collection.Collections.fillObject(n, element, empty(), Tree::of); }`",
+      "reference": "(...), where each element ~are~ **is the** given {\\@code element}."
+    },
+    {
+      "prediction": "I don't think we need this. We already have a `SortOrder` field in the `File` object.",
+      "reference": "I think this can be an int because we have a default that is always valid, `0`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this test. I think it should go in `spec/models/section.rb`, but I'm not sure.",
+      "reference": "I think you can use `build_stubbed` here for the same results but with more speed."
+    },
+    {
+      "prediction": "I think this should be `duration = max(hold, rampup)`",
+      "reference": "what if one of them is property and one is not?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. This means that if there is an error, we don't log it, but we don't ignore it either. I think we need to do something like: `catch (error) { if (error) { logger.error(error) } else { logger.error('An error occurred') } finally { process.exit(1) } }`",
+      "reference": "> UI plugins swallow rejection errors so they don't end up in the console with no way to fix But it will still catch upstream in `addFile` and restrictions to show the informer?.."
+    },
+    {
+      "prediction": "I'm not sure why we need to pass in the `owner` here. I think we can just set `owner = { session: session }` in the `if` block above.",
+      "reference": "can't this just be left undefined/null?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "changed bc we are using the `error` var below"
+    },
+    {
+      "prediction": "I think we can remove the `status` field from the `assert` as it is already in the `expected_result` field.",
+      "reference": "Is there a reason why this is change is in the same commit?"
+    },
+    {
+      "prediction": "I think we can remove the `dir` argument from the `Command` type and use `dest` directly in the call to `exec.Command`.",
+      "reference": "@nghialv btw, perhaps was this method originally created for copying the repo root to the given `dest` as a subdirectory?"
+    },
+    {
+      "prediction": "I think this should be `ILoadMonitor`",
+      "reference": "Just curious - why not var?"
+    },
+    {
+      "prediction": "I think this should be `import '../images/logo-g_white_small.png';`",
+      "reference": "Why would we import this here and not use it? Is there a side-effect of this import?"
+    },
+    {
+      "prediction": "I think this should be `@attr message` instead of `@attr errors`.",
+      "reference": "The name `errors` is a bit mis-leading -- usually I'd expect a field called `errors` to be an array of errors but here it's just a count. And it's not a count of _all_ errors -- it's a count only of some errors."
+    },
+    {
+      "prediction": "Why is this bundle removed?",
+      "reference": "there was no usage of the bundle in the framework? everything works exactly the same after the removal?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Why this difference?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this error handling. I think it should be in the `SQLSrvr_ExtractLob` function.",
+      "reference": "Surround code issue. RequestError may not be null terminated when the RequestError size is less than the length of the string in p_buffer->errorText. Also, this can cause core dump due to segment violation if length of errorText is less than the size of RequestBuffer."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pfs/transaction.go`",
+      "reference": "`AccountMaxVersionPrefix` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think we can remove this file too.",
+      "reference": "I see it is not part of this PR but I somehow overlooked it before. Can you use same import aliases as we use everywhere in the project: `apicommon` -> `commonpb` `apireplication` -> `replicationpb` `\"go.temporal.io/api/enums/v1\"` -> `enumspb \"go.temporal.io/api/enums/v1\"`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this exception. I think it should be in `azure-storage-blob/__init__.py` or `azure.storage.blob/__init__.py`",
+      "reference": "Nit: Maybe merge these into one string while you are here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `runtime/async.go`",
+      "reference": "Why make a new file for this? I think keeping everything related to the scheduler in a single file increases readability. Especially when there is no API-boundary in between. Functions like `getCoroutine` and `yield` also live in the normal scheduler files so it doesn't seem consistent."
+    },
+    {
+      "prediction": "I think this should be `ifaceConfigurationEnabled` instead of `ifaceConfigurationEnabled`.",
+      "reference": "tiniest of nits: why would you abbreviate \"interface\" but not \"configuration\" ?"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "Removing since all of the domain-neutral testing will be done in the new `Samples.MultiDomainHost.Runner` app"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `process` module, since it's only used there.",
+      "reference": "is there a better way to detect this rather than just always assuming it's cygwin if everything else fails? i think you can check `if sys.platform == 'cygwin'`"
+    },
+    {
+      "prediction": "I think this should be `except GitCommandError as exc`",
+      "reference": "duplicate code - can you move into build_from_path() ?"
+    },
+    {
+      "prediction": "I think this should be `@pytest.mark.parametrize('name', ['foo', 'bar', 'baz'])` instead of `return 1`.",
+      "reference": "very important test as we want to make sure the old name still work."
+    },
+    {
+      "prediction": "I think this should be 0, not 1.",
+      "reference": "I think returning 1 should some how be moved in to specific support for LibreOffice.."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. It's not used in any of the generated IRs, and it seems like it should be moved to a place where it's actually used.",
+      "reference": "Should be inside `ifdef X86` I would think (or moved to ... I guess arch/x86/emit_utils.c)"
+    },
+    {
+      "prediction": "I think we can remove `EdgeNodes` and `Services` fields from `LocationCache` struct.",
+      "reference": "Using sync.Map(key: nodename; value: state) instead of this \"EdgeNodes\" string slice here would be much better. Reasons: 1. Using sync.Map makes the time complexity of function UpdateEdgeNode and IsEdgeNode O(1), while using string slice with for loop makes it O(n). 2. Later we could be checking node state whether it's \"ready\" or still \"not ready\" by directly comparing \"value\" of sync.Map and doing the node state updating."
+    },
+    {
+      "prediction": "I think this should be `UnwrapAnchorTagsPattern = new Regex(\"^([^<>]*)<a\\\\s*.*\\\\shref\\\\s*=\\\\s*[\\'\\\"]([^\\'\\\"]*)[\\'\\\"][^>]*>(.*)</a>$\", RegexOptions.IgnoreCase);`",
+      "reference": "nit, does it make sense to get this variables outside the func where they are used?"
+    },
+    {
+      "prediction": "I think this should be `Amazon-SNS` instead of `Amazon-Simple-notification-service`",
+      "reference": "to **an** Amazon... Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "I think this should be `if err != nil { ... }` and not `if err != test.WantErr { ... }`",
+      "reference": "Nit: is \"top\" a short name we use frequently? Seems a bit mysterious, and saves only 2 chars."
+    },
+    {
+      "prediction": "I think this should be `cassandraProtoVersion` instead of `ProtoVersion`",
+      "reference": "This name is really confusing. May be just remove this const completely and hardcode `4` where it is used?"
+    },
+    {
+      "prediction": "This test is no longer necessary, is it?",
+      "reference": "Why did you remove this spec?"
+    },
+    {
+      "prediction": "I think this should be `goog:driver`",
+      "reference": "The vendor prefix is still being used on Chromium based browsers like Edge Chromium and Chrome. Did you mean to remove this?"
+    },
+    {
+      "prediction": "I think this should be `self.composite_root_resource = kwargs.get('composite_root_resource') or []`",
+      "reference": ".get will return None if there is no project_id key, so the or None part is redundant"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `OpenTelemetry.Metrics.Histogram`",
+      "reference": "Do they need to be `internal`?"
+    },
+    {
+      "prediction": "I think this should be `CombinedOutput()` instead of `CombinedOutput()`",
+      "reference": "Hit a flake here where I think there may have been some output to stderr that got mixed in with the output from Stdout. Hence switching to `Output()`, which does also capture stderr as `err.Stderr`"
+    },
+    {
+      "prediction": "I think this should be `{\"cuMemcpy2D_v2\",                                        {\"hipMemcpyParam2D\",                                        \"\", CONV_MEMORY, API_DRIVER, HIP_UNSUPPORTED}}`",
+      "reference": "Is there any official CUDA API called cuMemcpy2D_v2?"
+    },
+    {
+      "prediction": "I think we can remove this interface and just use the constants directly in the code.",
+      "reference": "Can you please move this interface as class in common Constants file in az-core module?"
+    },
+    {
+      "prediction": "I think this should be `fs::path` instead of `folly::stringPrintf`",
+      "reference": "Why not TempDir?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should be in `MvvmCross.Core.Views`.",
+      "reference": "IMvxFormsView -> IMvxElement is a breaking change, isn't it?"
+    },
+    {
+      "prediction": "I don't think we need this import. You can just use `ansible_galaxy_install` from `molecule.provisioners.ansible_galaxy_install`",
+      "reference": "This should be imported alphabetically."
+    },
+    {
+      "prediction": "This is a breaking change. I think it should be done in a separate PR.",
+      "reference": "Please apply same changes as for Payment"
+    },
+    {
+      "prediction": "I think we can remove the `useHasScrolledEffect` here.",
+      "reference": "This should only be used in the `ScrollEffect` component, as you have in the `DashboardMainApp` below"
+    },
+    {
+      "prediction": "I think this should be `g.token = token` instead of `g.token_str = token`",
+      "reference": "I'd call it `g.auth_token`, just to make it more clear"
+    },
+    {
+      "prediction": "I think this should be `1 << 20` instead of `(1 << 20) * 2`",
+      "reference": "Not sure if this is what we really need, we might need to adjust this constant after more extensive benchmarking."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put it. I think it would be better to put it in `api/teams` or `api/teams/v2`",
+      "reference": "Shouldn't we use `:branchname` here? Or just `:branch`?"
+    },
+    {
+      "prediction": "I think this should be `RandUUID` instead of `RandNum`",
+      "reference": "`AutoTime` is a number but calling `int()` would make it crash. Tested through the docs"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `EvictingQueue` class, since it's only used by that class.",
+      "reference": "@eddynaka Looks like the same block more or less 4 times, should we have a helper method? private void AddOrReplaceAttribute or something?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `JettySolrRunner` since it's related to the Jetty solr runner.",
+      "reference": "Please do the same for the DebugFilter a few lines above."
+    },
+    {
+      "prediction": "I think this file should be in `src/preact` instead of `src/preact`. I don't think it should be in `src/preact` because it's not related to the `preact` component.",
+      "reference": "Minor nit: Just inline both functions here. No need to make a file per function (that's a popular approach for some projects though, just not for us :) )."
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. We should be using the default configuration source, not creating a new one. I think we need to add a new method to the `TracerSettings` class that takes in a `IConfigurationSource` and returns a new instance of `TracerSettings` that uses the default configuration source. Then we can remove the `CreateDefaultConfigurationSource` method from the `TracerSettings` class.",
+      "reference": "There are several uses of the word \"default\" in here, perhaps it can be reworded slightly?"
+    },
+    {
+      "prediction": "I think this should be `SUBMIT_FORM` instead of `SUBMIT`",
+      "reference": "`function (` is correct. A space must be there. Please run `make cs` to have correct coding style."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in `querySelectorAll` instead of `querySelector`.",
+      "reference": "I think this is now going to fail the scenario where there is a `main` with aria-hidden=\"true\". I don't think that should fail."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just copy the subject into the slice.",
+      "reference": "Is this specifically that if we capture the cast in another variable, even if it passes escape analysis, that the runtime will copy?"
+    },
+    {
+      "prediction": "I think this should be `KMSAlias` instead of `KMSAliasPrefix`.",
+      "reference": "Should this be optional for existing configs to work?"
+    },
+    {
+      "prediction": "I think it would be better to do this in `EmitCheque` instead of here.",
+      "reference": "note that `balance` does not necessarily fit into `int64`, especially with the new bzz token. so you might want to avoid the `int64` conversion step using `big.NewFloat(0).SetInt(balance).Float64()`."
+    },
+    {
+      "prediction": "I think it would be better to have a `flatbuffers::IDLOptions::kJsonSchema` constant in the `flatbuffers` namespace.",
+      "reference": "lets leave out the `-S` ?"
+    },
+    {
+      "prediction": "I think this method should be called `toLower` instead of `toLowerCase`.",
+      "reference": "Is this method needed?"
+    },
+    {
+      "prediction": "Please add a blank line between the C++ code and the .cpp file.",
+      "reference": "How about, instead of reading the param every time, just add a class member variable to hold the value of the flag, and just set it once during `LayerSetUp`?"
+    },
+    {
+      "prediction": "I think it's better to call it `DesiredTasks` instead of `ServiceTasks`.",
+      "reference": "Does this mean it's only returning tasks associated with the primary deployment? or something else"
+    },
+    {
+      "prediction": "I think this should be `awserr.NewErrorString(err.Error())` instead of `awserr.OrigErr()`?",
+      "reference": "nit, would be good to include prefix for this log statement."
+    },
+    {
+      "prediction": "`aria-required-attr` doesn't seem to be used in this test.",
+      "reference": "Same here, you should use a made up role. May I suggest `role=\"McCheddarton\"`? :)"
+    },
+    {
+      "prediction": "I think this should be `validateNoConflictingData` instead of `validateNoConflictingDeletes`.",
+      "reference": "@aokolnychyi, shouldn't this check whether the operation is a delete? If this is invoked by `DELETE FROM` then we don't need to validate conflicting deletes."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Any reason to split up this import block? I think most files have them together in one block, right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think we should move it to the `AppSec` namespace, since it's only used there.",
+      "reference": "`Tags.AppSecEnabled` should probably move to `Metrics.AppSecEnabled` I guess?"
+    },
+    {
+      "prediction": "This should be `super(AtomicLocalFile, self).__init__(self, path, mode)` and not `super(AtomicLocalFile, self).__init__(self, path, mode)`",
+      "reference": "What happens if you remove this? Can't you still initialize this atomic_file class, since it's just AtomicLocalFile with an additional method (move_to_final_destination)?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `layouts/partners.html.erb`",
+      "reference": "As previously discussed in #1944 it's not safe to remove IC like this as it will break every language that still has a `%{ic}` marker in the translation."
+    },
+    {
+      "prediction": "I think we can drop this rule now, since we don't need to support VXLAN traffic anymore.",
+      "reference": "I believe these rules will be enforced both (1) on egress from a local workload, and (2) on ingress **to** a local workload. Right? I understand that we definitely want (1), but do we really want to enforce (2) as well?"
+    },
+    {
+      "prediction": "I don't think we need to add this to the `COMMANDS` list, since it's not really a command. It's more of a `POST` request, right?",
+      "reference": "Update after command rename"
+    },
+    {
+      "prediction": "Line is too long. [102/80]",
+      "reference": "What about adding an optional parameter that allows users to choose whether they want to use the format or not?"
+    },
+    {
+      "prediction": "I think this test should be moved to `integration_test.go`",
+      "reference": "Lint didn't like this context being before the skip"
+    },
+    {
+      "prediction": "I think it would be better to have this as a separate file, so that it can be easily changed in the future without affecting the rest of the package.",
+      "reference": "so now basically anybody could drink my tequila? currently electron dev-env runs on `http://localhost:9080`"
+    },
+    {
+      "prediction": "This should be `qWarning() << \"Can't load plugin \\\"%1\\\". Plugin can't build ILXQtPanelPlugin.\"`",
+      "reference": "`mPluginLoader` is nullptr in static plugins, which causes segmentation faults"
+    },
+    {
+      "prediction": "I think we can remove `ThemeManager` from this class as well.",
+      "reference": "Better use `Optional<Consumer<String>> onSearch` here."
+    },
+    {
+      "prediction": "I think this test is no longer necessary.",
+      "reference": "this is asserted as invalid input"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `parse` package.",
+      "reference": "Why do you need to create a new parser? Can't you just use `state.Parser` to answer this question?"
+    },
+    {
+      "prediction": "I think this should be renamed to `CreateBlockForProcessingForTesting` or something like that.",
+      "reference": "remove processing references from Nethermind.Core where possible, keep clean"
+    },
+    {
+      "prediction": "I don't think this is the right exception to be thrown in this case. The user should not be able to send a message with an unsupported content transfer encoding. This exception should only be thrown if there is a problem with the content transfer encoding itself.",
+      "reference": "Why not make this a subclass of `MessagingException` so we don't need to change all `throws` clause?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "it should instead be `from .hmac_drbg import DRBG` also, please put it at the bottom of the imports (PEP 8 order), so near line 36"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the `findColumnName` method.",
+      "reference": "Nit: indentation is 2 indents (4 spaces) from the previous line."
+    },
+    {
+      "prediction": "I think this should be `ddev`, not `ddev`",
+      "reference": "I don't understand why this is changing to a hard-coded string."
+    },
+    {
+      "prediction": "I think this should be `DevnetConfig` instead of `Devnet`",
+      "reference": "nit: perhaps calling this `config` would be more clear?"
+    },
+    {
+      "prediction": "This should be `Lade hoch`, not `Lade hoch`.",
+      "reference": "I would rather recommend using 'Wird hochgeladen' since the imperative form of 'Lade hoch' confuses me whether this is an action I need to do or whether it's an action that Uppy is currently doing. Just my two cents."
+    },
+    {
+      "prediction": "I think this should be 00755.",
+      "reference": "Hmm I would have expected this to go away? What role does it play now that we have a minimum?"
+    },
+    {
+      "prediction": "I think this should be 11601",
+      "reference": "if we are removing this, is it useful to have the `isGetMoreError` message below? If that's not required, can we also remove the `mongoErrorContextSymbol` outright?"
+    },
+    {
+      "prediction": "I'm not sure if `wait_for_at_least_entries` is the right name for this method. I think it should be something like `wait_for_enough_entries` or something like that.",
+      "reference": "I think this mutex is only used here. Remove it."
+    },
+    {
+      "prediction": "I think this should be `if (executor == null) {` instead of `if (executor == null || executor.isShutdownNow()) {`",
+      "reference": "it's unclear to me which method you are recommending overriding?"
+    },
+    {
+      "prediction": "I think this should be `not field` instead of `not field == \"\"`.",
+      "reference": "This make me think that if the typing of a variable is inconsistent it could cause problem. For example here, if `field == 0`, the behavior is not exactly the same. A `0` and is implicitelty false. Well it works because summing 0 is neutral, but what if it was a multiplication ?"
+    },
+    {
+      "prediction": "please remove empty line",
+      "reference": "Remove this file from the PR"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "I cannot find the changes / results of the eip2565 benchmark"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to move this logic to the `useEffect` hook.",
+      "reference": "@tofumatt now that I see this in context, we're now resetting 2/3 of the keys in the `NewComponent` state - the other being `error`. It seems that `error` should probably also be reset - essentially resetting all of the component state when the date range changes. What do you think? cc: @felixarntz @adamsilverstein"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with that doesn't break anything.",
+      "reference": "This one and the ones below are for shutdowns but if it appears in some method called frequently during the lifetime of the process it should be cached in a static."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move it to `unit/test/models/user.js`.",
+      "reference": "Better to just do `const err = yield User.bulkWrite(ops).then(() => null, err => err);` so you don't need to write out the whole `try/catch` block."
+    },
+    {
+      "prediction": "I think this should be `if (math.MaxInt64-action.ExecutionBaseIntrinsicGas)/action.ExecutionDataGas < dataSize || action.ExecutionDataGas == 0 { return 0, action.ErrInsufficientFunds }`",
+      "reference": "`action.ExecutionDataGas = 100`"
+    },
+    {
+      "prediction": "I think this test should be moved to the `GalleryControllerTest` class.",
+      "reference": "@supersmile2009 please review. Is it right to mark this as legacy? Are we testing the legacy path here? Are there tests for the non-legacy path?"
+    },
+    {
+      "prediction": "I think we can remove this line as well.",
+      "reference": "getSpcResource should not be bothered about deletion. Can we have `separation of concerns` for this logic?"
+    },
+    {
+      "prediction": "I think this should be `ks.DataFrame({'A': [1, 2], 'B': ['x', 'y']})` instead of `ks.DataFrame({'A': [1, 2], 'B': ['x', 'y']})`",
+      "reference": "This is actually matched with pandas's result since the output is sorted."
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `0700`",
+      "reference": "struct field tag `yaml:\"endpoint\"\"` not compatible with reflect.StructTag.Get: key:\"value\" pairs not separated by spaces (from `govet`)"
+    },
+    {
+      "prediction": "I think this should be `unique_ptr` instead of `std::unique_ptr`.",
+      "reference": "This is not necessary. Topic inherits from DomainEntity which already has instance_handle_"
+    },
+    {
+      "prediction": "I think we should be able to remove this `runBpfUnitTest` function altogether and just call `runBpfTest` directly.",
+      "reference": "Do we need the forXDP parameter in runBpfUnitTest? If not, I think better to revert in order to save a few changes."
+    },
+    {
+      "prediction": "I think we can remove the `getSnapshotObj` function and use `casSnapshot` directly.",
+      "reference": "All the code that reference to this function must also be changed."
+    },
+    {
+      "prediction": "I think this method should be called `validateUser` instead of `validateUserGroupMembership` to be consistent with other methods in this class.",
+      "reference": "How is this new interface different from existing `validateProxyUser` method? If I understand correctly the `proxyUser` parameter is essentially a group and the method should verify if `realUser` belongs to it."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Do we have namespace conflicts?"
+    },
+    {
+      "prediction": "I'm not sure why we need to pass `ctx` here. I think we can just use the one passed in to `newWebAPI`",
+      "reference": "We have several groups of caches so I think we should not use this shared configuration value for all of them. Instead of that, each cache group should have a separate configuration field, and the added `validation cache` should have its own field too. But this point, they can be fixed by `24h`."
+    },
+    {
+      "prediction": "I think this should be `ReplicateInterface` and `MetadataSupporter` should be `MetadataSupporterInterface`",
+      "reference": "This is a BC break for extending classes, it can't be done on 3.x"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. I think it should be done in the `get_etag_from_s3_response()` method.",
+      "reference": "Line length (PEP8, should be 80 chars max)"
+    },
+    {
+      "prediction": "I don't think we need this constant. We can just use `X-Priority` and `priority_high` everywhere.",
+      "reference": "this is duplicate now. I suggest moving it into `MimeHeader`, we already have a bunch of other headers there"
+    },
+    {
+      "prediction": "I think we can remove the `fmt.Fprintln(os.Stderr, ...)`, since we're printing to stderr anyway.",
+      "reference": "Can you expand the warning slightly \"Your password contains leading/trailing whitespace - in previous versions of rclone this was stripped\" Just so users know they might have to type their password without a leading space."
+    },
+    {
+      "prediction": "This should be `MessageDlg.Show(this, Resources.SkylineWindow_FIND_IRT_DATABASE_The_database_file_specified_could_not_be_opened);`",
+      "reference": "I think \"this\" should be \"parent\"."
+    },
+    {
+      "prediction": "(style) 'self' may be misspelled - perhaps 'self'? (style) 'self' may be misspelled - perhaps 'self'? (style) 'self' may be misspelled - perhaps 'self'?",
+      "reference": "Why not init this at variable decl time instead?"
+    },
+    {
+      "prediction": "I think this should be `ProjectIDEnvConfig` instead of `ProjectIDEnvKey`.",
+      "reference": "Do we still need this if we're doing everything through `envconfig`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think we should move it to the `testnet.go` file.",
+      "reference": "I would like these options to have `experiment` prefix or suffix. Because we will be deleting them later"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "Since we have the same logic in both forms, can we make a shared partial/helper for it?"
+    },
+    {
+      "prediction": "I think this should be `marketAddr + \"/ask_plans\"` instead of `blockchain.MarketAddr().Hex() + \"/ask_plans\"`",
+      "reference": "Why to introduce this intermediate variable?"
+    },
+    {
+      "prediction": "I think it would be better to use `client_type` and `token` instead of `kerberos` and `insecure`",
+      "reference": "Maybe use ChoiceParameter?"
+    },
+    {
+      "prediction": "Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "Not positive is this is needed, given the `TokenAuth` include"
+    },
+    {
+      "prediction": "I'm not sure if this should be a `NamedTuple` or a `List[Tuple[str, ...]]`, but I think it should be a `Tuple[str, ...]`.",
+      "reference": "Why not use a docstring for it?"
+    },
+    {
+      "prediction": "I think this should be `--p2p-port=9 --rpc-http-port=10 --rpc-ws-port=10` instead of `--p2p-port=9 --rpc-http-port=10 --rpc-ws-port=10`",
+      "reference": "Two more tests - no ports specified and every possible port specified - p2pPort, graphQLHttpPort, rpcHttpPort, rpcWsPort, metricsPort, metricsPushPort, stratumPort"
+    },
+    {
+      "prediction": "I think this should be `Copyright (c) 2020 Eli Fine <ejfine@gmail.com>`, not `Copyright (c) 2020 Eli Fine <ejfine@gmail.com>`",
+      "reference": "Don't worry about that next time, it's automated :)"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "No need for this, as pylint already only turns things off for this function and it's needed for the entire function."
+    },
+    {
+      "prediction": "nit: I think we can remove `AWSVPCHostValue` and `opentelemetryNetworkHost` as they are no longer used.",
+      "reference": "Are these values expected to change over time? Why not inject these values in ECS backend instead of agent doing it?"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "as best i can tell, you are setting this here in order to make it available in `(e *ETCD) Snapshot(ctx context.Context, config *config.Control) error`. It feels wrong to pass state like this through an env var. Could you instead add a `nodeName` property to the ETCD struct?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be in the same place as `drop_non_existent` and `keep_non_existent`.",
+      "reference": "These seem to support the new test framework below, but the `.in.csv` files are still committed."
+    },
+    {
+      "prediction": "I think this should be `if (txReceipts[i].Removed == false)` and not `if (txReceipts[i].Removed == false)`",
+      "reference": "!txReceipts[i].Removed and We can check it for first item outside of for loop, we don't expect mixed batches"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to have a single flag for `url` instead of `github-url` and `github-access-token-url`.",
+      "reference": "We can't unfortunately change any of our old flags :( it'd be a breaking change. Instead, we can keep both `githubURLFlag` and `repoURLFlag` and make the description of `githubURLFlagDescription` as \"(Deprecated) Use --url instead. GitHub respository URL for your service.\" It would be even better if we can make `githubURLFlag` as hidden but keep in the command. This way old commands will still work with `--github-url` but it won't appear anymore in the help menu."
+    },
+    {
+      "prediction": "I think this should be `Puppet::Pops::Issues.issue(:plan_not_found) { Bolt::Error.plan_not_found(plan_name) }` instead of `Puppet::Pops::Issues.issue(:UNKNOWN_PLAN) { Bolt::Error.unknown_plan(plan_name) }`",
+      "reference": "This provides file and line number if applicable."
+    },
+    {
+      "prediction": "please use `import matplotlib.pyplot as plt` instead of `import numpy as np`",
+      "reference": "This is already imported. Please check the file."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `SelectionActionDelegate`.",
+      "reference": "One thing we can do in a follow up is use the shorthand calls, so You can just call `aSelection.unselect()` instead of calling the more verbose `execute()`. The fact that you have to check if the action is available sort of sucks though."
+    },
+    {
+      "prediction": "I don't think we need the `break` here.",
+      "reference": "I don't think adding this value to _every_ connection policy is needed or useful... why is this necessary?"
+    },
+    {
+      "prediction": "I think we should use the same name for this interface as we do for the concrete class.",
+      "reference": "adding this constrain for null AggreagateId is redundant"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it works for now.",
+      "reference": "Why when current user is owner but still check access? Should we use `else if` here?"
+    },
+    {
+      "prediction": "I think this should be 1 minute. We don't want to miss any events.",
+      "reference": "60 seems high. 10? 20?"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(args.config_directory, 'Default.sqlite')` instead of `os.path.join(cfg_dir, 'Default.sqlite')`",
+      "reference": "I think this is not the best way to handle this problem. This way every relative path will be converted to absolute path. What if the user would like to move his config directory to a different directory?"
+    },
+    {
+      "prediction": "I think this should be called `ControlPlaneScheme` instead of `ControlPlaneLoadBalancerScheme`.",
+      "reference": "Let's make this a pointer given that's optional"
+    },
+    {
+      "prediction": "Can you add a license to this file?",
+      "reference": "Why is CircularJSON imported here? It is never used in this file."
+    },
+    {
+      "prediction": "I don't think this is the right fix. I think we need to change the `OpenApiDocumentType` to `OpenApiDocumentType.MergeState`",
+      "reference": "Should be ARM"
+    },
+    {
+      "prediction": "Why did you remove the `is not a directory` part?",
+      "reference": "I don't see how removing \"open file bucket\" helps your purpose."
+    },
+    {
+      "prediction": "I'm not sure this is the right version. I think we should be using the latest version of solr, which is 6.6.2, and then check against the newest version of solr, which is 7.0.0.",
+      "reference": "Solr dependency was downgraded?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "or should we just autoload everything in `lib` ?"
+    },
+    {
+      "prediction": "I think this should be `slots.size() > 0` instead of `slots.stream().anyMatch(slot -> slot.isSupporting(caps))`",
+      "reference": "Prefer `Stream.anyMatch` instead of iterating over all slots."
+    },
+    {
+      "prediction": "I think this should be `blacklist_violation` instead of `BUCKET_VIOLATION`",
+      "reference": "create_tables isn't used in 2.0 (should probably be deleted?)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `OpenTelemetry.Requests` or `OpenTelemetry.Headers` or `OpenTelemetry.RequestHeaders` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers` or `OpenTelemetry.Headers",
+      "reference": "Now I start to wonder, do we want to have `80` as the default or `9090` (or depending on whether we are exposing it via `PrometheusExporterMiddleware` vs. `PrometheusExporterHttpServer`)?"
+    },
+    {
+      "prediction": "I'm not sure this is a valid query. I think it should be `CREATE USER IF NOT EXISTS user1 WITH PASSWORD \\\"aaa\\\" , FIRSTNAME \\\"a\\\", LASTNAME \\\"a\\\", EMAIL \\\"a\\\", PHONE \\\"111\\\" , ACCOUNT LOCK, MAX_QUERIES_PER_HOUR 1, MAX_UPDATES_PER_HOUR 2, MAX_CONNECTIONS_PER_HOUR 3, MAX_USER_CONNECTIONS 4`",
+      "reference": "try an illegal case. and the result is syntax error."
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "Initialize `needs_reshape_` with true and call `Reshape` method"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `before` block of the `RandomizeOrdering` method, since that's where the `Kernel.rand(list.size)` is called.",
+      "reference": "This is a _very_ complicated line. I'm not a fan of ternaries to begin with (although, I allow them in very simple situations), but this ternary has compound conditionals and would really need to be broken up. That said: if I'm reading this right, it sounds like this logic can make it where this lambda (which is called `RANDOM_ORDERING`) can make it not random and basically do the opposite of what it says. In general, it's good to avoid making methods or lambdas do the opposite of what they say."
+    },
+    {
+      "prediction": "I think this should be `backend\\web\\info\\apacheSolrDocument` and not `backend\\web\\info\\ApacheSolrDocument`",
+      "reference": "We should remove this"
+    },
+    {
+      "prediction": "I think this should be `IptablesNATOutgoingInterfaceFilter: string`",
+      "reference": "We should validate this, I think. To add a validator: - change `string` to, say `iface-pattern` - add a new `case \"iface-pattern\"` to the `switch` in the `loadParams()` function: - in that `case`, you should be able to use a RegexpPattern like the one for `iface-list`. However, you'll need a different regexp. I think this regexp should work: `^[a-zA-Z0-9:._+-]{1,15}$` i.e. 1-15 characters consisting of alphanumerics, colon, dot, underscore, hyphen and plus."
+    },
+    {
+      "prediction": "I think this is not correct. The balance should be increased by 40, not decreased by 40.",
+      "reference": "File is not `goimports`-ed with -local github.com/iotexproject/iotex-core (from `goimports`)"
+    },
+    {
+      "prediction": "nit: I think we can remove this method since we don't use it anywhere.",
+      "reference": "Nit: Here we don't need the extra `this` in iceberg because we usually use the `this` to distinguish whether it is a member variable or local variable when assigning value."
+    },
+    {
+      "prediction": "I think this should be `ENGINE-16002 Exception while closing command context: ENGINE-03110 Required admin authenticated group or user or any of the following permissions.`",
+      "reference": "Why do we have that extra \"ENGINE-16002 Exception while closing command context:\" now? Is that done intentionally by us or where does that now come from?"
+    },
+    {
+      "prediction": "I think this class should be called `MetricReporter` instead of `LoggerReporter`.",
+      "reference": "Please don't use star imports, in the codebase we tend to use direct imports."
+    },
+    {
+      "prediction": "I don't think you need this line.",
+      "reference": "Filename can be `config.functional`"
+    },
+    {
+      "prediction": "I think this should be `std::this_thread::sleep_for (100ms);` instead of a `joinable` check.",
+      "reference": "Should the thread join be done in the stop function maybe? I am simply wondering and sharing my thoughts. This is my thinking: * The join used to be done in the stop function of node class and now it has moved in the destructor. * The class state_block_signature_verification joins its thread at the stop function too."
+    },
+    {
+      "prediction": "I think this should be `WalletsCmd.AddCommand(walletCreateCmd)` and `WalletsCmd.AddCommand(walletListCmd)` instead of `WalletCmd.AddCommand(walletCreateCmd)` and `WalletCmd.AddCommand(walletListCmd)`",
+      "reference": "don't use `init` function (from `gochecknoinits`)"
+    },
+    {
+      "prediction": "I think this should be `Amazon::Sdk::Resource::SimpleNotificationService` instead of `Amazon::Sdk::Resource::SimpleNotificationService::V3`.",
+      "reference": "Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "I think this should be `DEFINE_int32_t_crash_test, 1024, \"The flag could not be read\"`.",
+      "reference": "You are fixing the crash problem, and you have fixed it, so it won't crash anymore. So `crash_test` is not a proper name."
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. I think it should be moved to `NameFormatterDelegator`.",
+      "reference": "If this default doesn't depend on the argument, shouldn't it be delegated as the default for the no-arg version instead?"
+    },
+    {
+      "prediction": "I think this should be `get_or_create_file(path, permissions=0o600)`",
+      "reference": "I did not use the new save_config_file method in here, because I am not sure whether this whole logic is still necessary? There is an TODO about it there as well."
+    },
+    {
+      "prediction": "This should be `itemId` instead of `items[0]`",
+      "reference": "Why was this changed?"
+    },
+    {
+      "prediction": "I don't think this is correct. The `floor_thr == 0` check should not be necessary here.",
+      "reference": "I think you meant, self.floor_thr = 0"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea.",
+      "reference": "Can you revert these changes?"
+    },
+    {
+      "prediction": "I don't think it's necessary to import `StatusBarIOS` here.",
+      "reference": "Do we need to change this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to check if swap is enabled in the cgroup manager and set the swap limit accordingly.",
+      "reference": "With this commit, `MemoryUnlimited` is defined as a `unit64`, so I think you can drop the redundant cast from this line (and the later lines which also have `uint64(configs.MemoryUnlimited)`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "`props.totalFileCount < props.maxNumberOfFiles` makes the intent a bit more clear I think. And a check to see if maxNumberOfFiles even exists?"
+    },
+    {
+      "prediction": "what is the meaning of `nextEpoch`?",
+      "reference": "`epochNum` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. We should be looking for the entity in the oldRows array.",
+      "reference": "This needs to be fixed before we can accept the PR"
+    },
+    {
+      "prediction": "This should be `constexpr size_t` instead of `size_t`.",
+      "reference": "Are these forward declarations needed if we `#include <node/common.hpp>` which includes the definition ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `Message` doesn't have a `ConnectID` field, but the `MessageType` field. I think we need to change the `MessageType` field of `Message` to something like `MessageType: string` or `MessageType: string`, and then we can remove the `String()` method.",
+      "reference": "can you use either `connectID` or `ConnectID` i would preffer `connectID`"
+    },
+    {
+      "prediction": "Shouldn't this be `array|null|bool`?",
+      "reference": "It seems like it cannot be null?"
+    },
+    {
+      "prediction": "I think we can remove this todo.",
+      "reference": "should we consider add deprecated annotation here?"
+    },
+    {
+      "prediction": "I think this is the wrong place for this. This is for testing, not for production use.",
+      "reference": "Why can't the implementor get the username from the config?"
+    },
+    {
+      "prediction": "I think this should be `uint64(1000000000000000000)` instead of `50`",
+      "reference": "@CoderZhi Please confirm if this change makes sense"
+    },
+    {
+      "prediction": "I don't think we need this error code.",
+      "reference": "Do you think you can come up with a better message? :-) It looks like this is not used anywhere. So maybe just delete the line."
+    },
+    {
+      "prediction": "I think this should be `iterator().reduceLeft(op)`.",
+      "reference": "`tail()` is an expensive operation for certain collections"
+    },
+    {
+      "prediction": "I think this should be `../../../TestServer/swagger/azure-parameter-grouping.json` instead of `../../../TestServer/swagger/model-flattening.json`",
+      "reference": "do we need to add this one in this PR?"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "where is a test for it?"
+    },
+    {
+      "prediction": "Why do we need to change the regex?",
+      "reference": "Why has this spacing been adjusted?"
+    },
+    {
+      "prediction": "I think it would be better to use `LOGGER.error` instead of `LOGGER.warning` here.",
+      "reference": "Sorry to do this, since you're touching this file can you resolve the pylint doc messages above?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to add the recording_msid to the insertId. The recording_msid should not be part of the insertId.",
+      "reference": "@alastair, because the `insertId` for two listens with different metadata and same ts was the same, only one of them would get written into BQ. A question is what would be the ideal way to write tests for stuff like this, so that this doesn't break again?"
+    },
+    {
+      "prediction": "This should be `Gesamtweg-Strecken\u00e4hler Offset`",
+      "reference": "Wondering where this breaks btw (although pretty understandable)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "is this needed? 10 is already the default"
+    },
+    {
+      "prediction": "I think this should be `if 'text_font_size' not in style: style['text_font_size'] = '%dPt' % element.fontsize` and not `if 'text_font_size' not in style: style['text_font_size'] = '%dPt' % element.fontsize`",
+      "reference": "One day we can try to allow the user to specify font sizes in something other than points. For now this is fine though..."
+    },
+    {
+      "prediction": "I think this should be `security_attributes_` and `plugin_security_attributes_` should be `security_attributes` and `plugin_security_attributes` respectively.",
+      "reference": "duplicated in both block of preprocessor condition. Can you join in one?"
+    },
+    {
+      "prediction": "I think this should be `node.ownerDocument.querySelectorAll('*#' + axe.utils.escapeSelector(node.id))`",
+      "reference": "Can we put the first part of the expression inline with the `if(` for consistency's sake?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/api/server` since it's only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `pkg/api/server` and `pkg/api/server` is only used by `",
+      "reference": "Do servers have no need of accessing this information?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the `formatTimeAgo` method as it was before, since it's not used anywhere else in the codebase.",
+      "reference": "It feels like this will break some stuff, but let it break. We need \"text\" version anyway."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `intersectColumns` function.",
+      "reference": "Here are two things I wonder: First, would it make sense to replace the TRUE here with \"leftExpr->getValueId().getType().supportsSQLnull() && rightExpr->getValueId().getType().supportsSQLnull()\". I'm not sure we have optimizations elsewhere that set the \"special nulls\" semantics back to FALSE if one of the operands it not nullable. Second, I wonder whether we have bugs elsewhere in the code that may not check for the \"special nulls\" semantics. For example, hash and merge joins probably can't use such predicates as equi-join predicates - one of the reasons for the optimization above. When I look at method ItemExpr::isAnEquiJoinPredicate() in core/sql/optimizer/OptItemExpr.cpp, I don't see a check for that. You have an example with a NULL value that is working, so hopefully this is not an issue, but I'm not sure how and why it works."
+    },
+    {
+      "prediction": "I think this should be `TD(NumericConstant, 260, \"nan, infinity or function name (signed)\")`",
+      "reference": "This will cause someone writing a schema field like `inf:string` to get a pretty confusing error? If they intended to use `inf` as short for `information` or whatever :) Might it be better to keep it as `Identifier` and explicitly recognize the few identifiers we care about only when parsing values (not while parsing field names)?"
+    },
+    {
+      "prediction": "I think we can remove the `1000` as well.",
+      "reference": "Mmm .... this affect #168 I'll need to research the collateral damage"
+    },
+    {
+      "prediction": "I think we can remove the `beta` entry from here as well.",
+      "reference": "nit: refactor into a const map lookup."
+    },
+    {
+      "prediction": "I think we can remove the `this.GetType().Assembly;` call in `AddEventListeners` as well.",
+      "reference": "I don't think this is going to work. We need to get the actual assembly of the app project. @johnnywebb thoughts?"
+    },
+    {
+      "prediction": "I don't think we need memoize here. If the value is not in the cache, we fetch it and add it to the cache. If the value is already in the cache, we return the value from the cache.",
+      "reference": "@MSusik be careful about changing existing API. I think in this case you shouldn't rename the argument."
+    },
+    {
+      "prediction": "(style) 'kafkatopic' may be misspelled - perhaps 'kafka'? (style) 'kafkatopic' may be misspelled - perhaps 'kafka'?",
+      "reference": "This is bad. It is not a NULL-terminated string here."
+    },
+    {
+      "prediction": "I think this should be `@var bool|string|bool`",
+      "reference": "This probably should be @param string|false."
+    },
+    {
+      "prediction": "I think it would be better to have a test that checks that the result is not empty.",
+      "reference": "Unrelated test fix. Already in cordova34 branch."
+    },
+    {
+      "prediction": "I think it would be better to move this to the `stakingV2` command.",
+      "reference": "validate() will be called by either V1 or V2"
+    },
+    {
+      "prediction": "I think this should be `#if defined(_MSC_VER) || defined(__GNUC__)`",
+      "reference": "rather than generate code for this every time, stick it in `flatbuffers.h` (and call it `FLATBUFFERS_NOEXCEPT` to avoid clashes)."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea.",
+      "reference": "@AskAlexSharov @tjayrush , I'm still feeling uncomfortable with this change: - because it works by accident. For example in next lines `tx` object used as: `ReadBlockByNumber(tx)`. If you go inside `ReadBlockByNumber` you can find `!errors.Is(err, ethdb.ErrKeyNotFound)` - but ethdb.Tx doesn't return this error - and it satisfy `DatabaseReader` by accident - (I faced this problem last week - when passed Tx to place which expected Db - and I promised to change tx API - rename Get to GetOne or even remove the method). - we can merge this PR if it doesn't break things and solve problem of localDb use of RPCDaemon. But I will make PR now which will allow open Read tx by TxDb/ObjectDb - and will ask you to use it. Then will take a look how hard to change LmdbTx to not satisfy rawdb.DatabaseReader interface."
+    },
+    {
+      "prediction": "I think this should be `anomaly_mode` instead of `anomalyMode`",
+      "reference": "I think this means the file doesn't have a newline character at the end."
+    },
+    {
+      "prediction": "I think this should be `if (uid == 0) {` instead of `if (uid != 0) {`",
+      "reference": "Hi Ramya, rather than checking here in the perf counter library, we should add the privilege check in the host_exerciser app."
+    },
+    {
+      "prediction": "I think this should be `if notifyObjectDeletion is not None:`",
+      "reference": "Could you elaborate on this some more? Has this behavior changed with Python 3.8?"
+    },
+    {
+      "prediction": "I think this should be `r.Config.Logger.Log(fmt.Sprintf(\"Error dumping request: %s\", err))` and not `r.Config.Logger.Log(fmt.Sprintf(\"Error dumping request: %s\", err.Error()))`",
+      "reference": "Do we want to continue on logging the `dumpedBody`, if an error was thrown? It is probably an empty string. This would make the log after this one pretty much useless."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if there is a better way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if there is a better way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if this is the best way to do this",
+      "reference": "the text here is derived from the object type and not getting properly handled by get text"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "This will make other pods that run antctl will also connect its localhost? Maybe only do it when it's in antrea-agent and antrea-controller"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `menuItems` array above, under the `other` key.",
+      "reference": "The corresponding action for this key needs to be removed as well, along with any unused translations."
+    },
+    {
+      "prediction": "Nit: I think we should be able to use `SpanContext` directly here instead of `SpanInternal`",
+      "reference": "Can't Span cannot be null anymore? I assume it was a useless check as there are discrepencies within integrations, but as you explicitly removed this one, I was wondering"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. This is a transaction trace, not a block trace.",
+      "reference": "hmmm, ok, this is interesting -> Lukasz definitely uses rewards traces"
+    },
+    {
+      "prediction": "I think this should be `read.split.by-partition.enabled` to be consistent with the other properties.",
+      "reference": "Why do we need this flag? can this be detected if the table is bucketed/partitioned and enabled/disabled automatically? Is this for backwards compatibility?"
+    },
+    {
+      "prediction": "I think this should be indented by 4 spaces, not 2.",
+      "reference": "can you revert changes to this file."
+    },
+    {
+      "prediction": "I think this should be `// Package soc provides the single-owner chunk implemenation and validator.`",
+      "reference": "suggestion to add: An soc is a chunks whose address is derived of (...)"
+    },
+    {
+      "prediction": "I think we should rename this to something like `USERNAME_ALREADY_REGISTRED_FOR_EMAIL`",
+      "reference": "It doesn't make sense. if we are going to have the profile page where the user can change the password and he / she is already logged in ... I do not need to register a new username. only: new password..confirm new password..something similar...and \"USERNAME_ALREADY_REGISTERED\" only if I have a register page..are we going to have? We need to talk about it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `StateRootUpdater` class, since that's where it's used.",
+      "reference": "not a great fan of this change - maybe worth splitting into CommitTree() and CloseBlock(long blockNumber)?"
+    },
+    {
+      "prediction": "I think this should be `ConnectionAbortedException` instead of `ConnectionClosedException`",
+      "reference": "Ugh, if we're going to make a breaking change, I'd like this to be moved to ConnectionContext."
+    },
+    {
+      "prediction": "I think we can remove this todo and just call `fsInfo.NewFs(...)` directly.",
+      "reference": "I was planning to add it as a common option, but there is a cyclic import for the rc & fs libs. Thoughts on avoiding it or should I remove this for now?"
+    },
+    {
+      "prediction": "I think this should be `/deploy/Nuget/Datadog.Trace.nuspec` instead of `/deploy/Nuget/Datadog.Trace.dnn.csproj`",
+      "reference": "I think this entry can be removed now, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. It seems like it should be in the `Key` class, not the `Bucket` class. The `Key` class is the base class for all Key objects, and the `Bucket` class is the base class for all buckets. The `Key` class should have a `handle_version_headers` method, and the `Bucket` class should also have a `handle_version_headers` method. The `Key` class should also have a `handle_storage_class_header` method, and the `Bucket` class should also have a `handle_storage_class_header` method. The `Key` class should have a `handle_version_headers` method, and the `Bucket` class should also have a `handle_version_headers` method. The `Key` class should have a `handle_storage_class_header` method, and the `Bucket` class should also have a `handle_storage_class_header` method. The `Key` class should have a `handle_storage_class_header` method, and the `Bucket` class should also have a `handle_storage_class_header` method. The `Key` class should have a `handle_storage_class_header` method, and the `Bucket` class should also have a `handle",
+      "reference": "I don't understand why you want to populate the storage class in this case, if the S3 docs say they won't populate the header in this case?"
+    },
+    {
+      "prediction": "I don't think we need to add any new capabilities here. We already have them in the existing capabilities.",
+      "reference": "These are really meant to be the keys in the capabilities, not the keys of values within the capabilities"
+    },
+    {
+      "prediction": "I think it would be better to say \"If the regionHint parameter is an empty string, GetBucketRegion will fallback to the ConfigProvider's region config. If no region is specified, an error will be returned.\"",
+      "reference": "Should the last sentence be `If no region was found` rather than `specified`?"
+    },
+    {
+      "prediction": "I think this should be `if (Victim.Type.TELEPORT.equals(tpMatcher.group(1)))`",
+      "reference": "Text.sanitize just removes images from names, for instance, the hardcore ironman symbol when someone talks. A better option would be text.standardize, or text.toJagexName"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could think of.",
+      "reference": "How about `ApplicationKindFromConfigKind`? And I think this function should be better in the config package. The reason is `config` package can import and refer things from the model package but not vice versa."
+    },
+    {
+      "prediction": "nit: I think we should rename this to `SignedTokenExpiration` or `SignedTokenExpirationTime` to be consistent with `CredentialExpiration`.",
+      "reference": "Nit: missing closing \".\"."
+    },
+    {
+      "prediction": "I don't think this test is necessary, it's already tested in `testClickingALinkMadeUpOfNumbersIsHandledCorrectly`",
+      "reference": "I believe it's misleading name for the condition. I prefer \"element_to_be_disable\" We can have a condition, when element is enabled but we can't click it, because another element overlays above it. So, If we use \"unclickable\" we might mislead people, who use that condition to verify if element can be clicked"
+    },
+    {
+      "prediction": "I think it would be good to add a note in the README that this is a binary representation of the pipeline manifest.",
+      "reference": "Same here. Should this be pipeline YAML file? Like `MarshalPipelineManifest`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in the `Stop` function, since it's the main responsibility of the instance.",
+      "reference": "I will add this to my PR"
+    },
+    {
+      "prediction": "I think this should be `@role.user_id == @user.id` instead of `@role.owned_by?(@user.id)`",
+      "reference": "Again not 100% sold on the name"
+    },
+    {
+      "prediction": "I don't think we should have a `/settings/xxx` endpoint for anything other than the avatar. I think we should just have a `/settings` endpoint for everything else.",
+      "reference": "Use `m.Combo` for `Get` and `Post` methods."
+    },
+    {
+      "prediction": "I think this should be `self.cancel(None)`",
+      "reference": "Is there a reason you're not using `self._die(\"The file {} ...\")` here?"
+    },
+    {
+      "prediction": "I think this should be `exec [app_name] [environment_name] [cmd]`.",
+      "reference": "I think we'll want @rickmanelius (or somebody) to go through all the help and make it more accessible. Probably later in the cycle. But \"Run a command in an app container\" doesn't do it for me :)"
+    },
+    {
+      "prediction": "I think this should be `[]string{\"/home/weavewait/weavewait\", \"/home/weaveworks/weaveworks\"}`",
+      "reference": "shouldn't this be a const?"
+    },
+    {
+      "prediction": "I think we can remove the `includes` here.",
+      "reference": "This is a bit of a throwaway field in the xunit output, it doesn't impact anything on EVG, should we just name it `integration` now?"
+    },
+    {
+      "prediction": "I think this should be renamed to `SolrSingleThreadedThreadUnsafe`",
+      "reference": "Please change the name of the file too."
+    },
+    {
+      "prediction": "This should be `./tools/actioninjector/gentsfaddrs.yaml` instead of `./tools/actioninjector.v2/gentsfaddrs.yaml`",
+      "reference": "line is 161 characters (from `lll`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `@pal` class, since that's where the `generate_types` method is.",
+      "reference": "When is this false?"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "note: also here ;)"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just remove the `Runtime.getRuntime().removeShutdownHook(shutdownHook);` line from here. The `Runtime.getRuntime().shutdownNow();` call is supposed to be the last thing that is done before the JVM is shut down, so it should be the last thing that is done even if the `Runtime.getRuntime().removeShutdownHook(shutdownHook);` line is removed.",
+      "reference": "I don't think we need to necessarily check if that returned true or not, we should just remove the shutdown hook. Since nothing would check or do anything with this flag anyways."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `kythe` package, since it's specific to the `kythe` package.",
+      "reference": "nit: blank between, for consistency."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `rdocifyProtoMarkdownLinks` and `rdocifyCloudMarkdownLinks` are both called with the same pattern, so they'll always match. I think we need to find a better way to escape the `$` character, but I don't know what that would be.",
+      "reference": "ditto for RDoc or YARD link"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think we should be able to get the timestamp from the `Date` object. If we can get the timestamp from the `Date` object, then I think we should be able to get the timestamp from the `Date` object. If we can get the timestamp from the `Date` object, then I think we should be able to get the timestamp from the `Date` object. If we can get the timestamp from the `Date` object, then I think we should be able to get the timestamp from the `Date` object. If we can get the timestamp from the `Date` object, then I think we should be able to get the timestamp from the `Date` object. If we can get the timestamp from the `Date` object, then I think we should be able to get the timestamp from the `Date` object. If we can get the timestamp from the `Date` object, then I think we should be able to get the timestamp from the `Date` object. If we can get the timestamp from the `Date` object, then I think we should be able to get the timestamp from the `Date` object. If we can get the timestamp from the `",
+      "reference": "This method is designed to allow the condition runner to tell users something interesting. The current `INFO` log level is correct."
+    },
+    {
+      "prediction": "I think this should be `uint64` instead of `int64`",
+      "reference": "Again, would not change that."
+    },
+    {
+      "prediction": "I think it would be better to have a `Status` enum with a constant for each of the statuses and then use that constant in this method instead of the `switch` statement. It would make the code easier to read and understand.",
+      "reference": "Is canceled considered failed?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "These ineffectual assignment changes intrigue me. Our CI tests have been checking for these for months; how come the tests haven't been failing? How did you catch these?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it should be in `TestMintX509CA` since this is the place where we actually mint the X509 certificate.",
+      "reference": "We should add an additional call to Recv() somewhere after here that we assert returns io.EOF."
+    },
+    {
+      "prediction": "I think this should be `ctx, cancel := context.WithCancel(ctx)`",
+      "reference": "Why not return a `ctx` from `setupJournalBlockServerTest` and use it everywhere, like in the other test files? And maybe put a test timeout on it while you're at it? (Same for the other journal test files below.)"
+    },
+    {
+      "prediction": "`InstanceReportUndestroyedObjects` should be `ObjectLifetimes::InstanceReportUndestroyedObjects`",
+      "reference": "Do we need HeadlessSurface and DisplayPlaneSurface here too?"
+    },
+    {
+      "prediction": "I think this should be `MongoDriverError: Cannot call abort() on a stream twice`",
+      "reference": "Would it be appropriate for this to be a `MongoStreamClosedError`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/apis/core/v1`",
+      "reference": "Can you raise new issue that talks about refactoring webhook code. It should follow `idiomatic maya` patterns."
+    },
+    {
+      "prediction": "I don't think we need this interface. We can just use `ISpanDecorator` in the `ISpanDecorator` implementation.",
+      "reference": "(Probably repeating myself) Instead of `Datadog.Trace.Interfaces`, should we move this and several other files into a `Datadog.Trace.Decorators` namespace instead?"
+    },
+    {
+      "prediction": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Should we generate `gem \"rsolr\"` into the application Gemfile?"
+    },
+    {
+      "prediction": "I think this should be `--verbose` and not `--verbose-v`",
+      "reference": "commit message, \"currently ignored\"? (-ed)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in the `Selenium::WebDriver::Chrome` module.",
+      "reference": "Same here. and all others."
+    },
+    {
+      "prediction": "I think you can remove the `OneTimeChainProcessor` constructor and just use the `OneTimeChainProcessor` instance.",
+      "reference": "can we pass some simple interface instead of a full additional processor? (an imterface with these two events only)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. The linker is supposed to do this, but I don't know how it does it.",
+      "reference": "Use a named constant"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "The `?` is unnecessary since `Assert.Single` will fail if it can't find the stop event."
+    },
+    {
+      "prediction": "I think this should be renamed to `dataLocationAsString()`.",
+      "reference": "What about providers that don't have a reliable location? Do they return null?"
+    },
+    {
+      "prediction": "I think we can remove this `using System;` at the top of the file, since it's only used in one place.",
+      "reference": "I might not be thinking about things right, but would it make sense for a MetricProcessor to have a handle on the ParentProvider kinda like how we do for traces? That way instead of calling `SetGetMetricFunction(this.Collect)` in the MeterProviderSdk you'd have a handle on the provider to call `Collect` directly."
+    },
+    {
+      "prediction": "I think this should be `aws_stack.connect_to_service(\"dynamodb\", endpoint_url=\"http://127.0.0.1:{PORT_DYNAMODB_BACKEND}\".format(PORT_DYNAMODB_BACKEND))`",
+      "reference": "nit: If we want to squeeze out a few more milliseconds from the startup, we may actually be able to remove this line (`wait_for_port_open(PORT_DYNAMODB_BACKEND, http_path=\"/\", ...` should already be sufficient to ensure that the service is up and responding to HTTP requests). Thoughts?"
+    },
+    {
+      "prediction": "I think this should be `if (num_raw & SYSCALL_NUM_MARKER_BSD)`.",
+      "reference": "Convention is to use TEST"
+    },
+    {
+      "prediction": "Can you remove the `skipif`?",
+      "reference": "Just in case: remember to remove this."
+    },
+    {
+      "prediction": "I think this should be `baseFee = Math.max(parentBaseFee - feeDelta, 0);`",
+      "reference": "I am not certain if this should be 0 or 7 here. I think a basefee under 7 is pathological"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `dmstore.c`.",
+      "reference": "it would be better to move the declaration of the variable with its initialization at line 68"
+    },
+    {
+      "prediction": "This should be `from rdkit import Chem` instead of `from rdkit import *`",
+      "reference": "I don't think this is required."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `integration/ethereum/test/unit/eth/test.go`",
+      "reference": "This rename should also ideally be in a separate PR."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `parseResponse` method.",
+      "reference": "Think this needs to go right below line 91. `parseResponse` can return `null` (line 133), which will cause this future to never complete."
+    },
+    {
+      "prediction": "I think this should be `private final String browserStartCommand = null;` and `private final String browserURL = null;` instead of `private String browserStartCommand` and `private String browserURL`",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "I think it would be better to move this check to the top of the method, so that it is always evaluated before the startWebMetrics() call.",
+      "reference": "Yeah, why not use this method instead of checking with `containsKey`?"
+    },
+    {
+      "prediction": "I don't think it's necessary to update `time_forward` and `time_backward` here.",
+      "reference": "Compute `(time_forward + 1e-3) * 1e3 / loop_num` ahead."
+    },
+    {
+      "prediction": "I don't think we should be throwing exceptions in the `InternalLogger.Error` method. I think we should be throwing exceptions in the `InternalLogger.ErrorInternal` method.",
+      "reference": "Mh... `MustBeRethrown()` above already logs `ex`, so `ex` will be logged here for the second time. I think this line must be moved to before `MustBeRethrown()`."
+    },
+    {
+      "prediction": "I think it would be better to use `Generator<Method, String>` instead of `Generator<Method>`.",
+      "reference": "Since this never changes, why not pass it into the constructor instead?"
+    },
+    {
+      "prediction": "I think this should be renamed to `targetDbForSaves`.",
+      "reference": "this is critical to avoid state root saving from beam"
+    },
+    {
+      "prediction": "I think this should be `self.num_classes + 1`",
+      "reference": "To make it more clear, `fill=self.num_classes`."
+    },
+    {
+      "prediction": "I think this should be `interface{}` instead of `func`",
+      "reference": "Nit: \"ByteSize estimates the size in bytes of the message for the purpose of restricting batch sizes.\""
+    },
+    {
+      "prediction": "I think we can remove the `options.ThreadCount = 4;` line.",
+      "reference": "It's weird to have this on the same level as kestrel. This needs to move to options."
+    },
+    {
+      "prediction": "`Blob` is not supported in iOS Safari. You can use `HTMLMediaElement` instead.",
+      "reference": "is it safe for us to use the `blob.type?.length` optional chaining with IE11 support? @goto-bus-stop"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "remove this debug statement"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to check this. I think it should be in the `if (api_version >= VK_API_VERSION_1_1) {` part of the if statement above.",
+      "reference": "This should not be needed. 1) all 1.1 promoted extensions are marked as enabled when a 1.1 device is created (which happens iff both physical device support and apiVersion are set to 1.1) 2) The KHR and non-KHR versions of the bits are aliases of each other, so no different set is required."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `loop_dev_full` environment variable that is set to `1` if we are running a loop with dev mode enabled and `0` if we are not. Then we can check for this in the `makeflow_generate_output_files` function.",
+      "reference": "I can see why you did it this way, but let's not get into the habit of passing information back via an environment variable. Add an element to `batch_job_info` instead, which is where we send back detailed info about jobs."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that the `i` value is not guaranteed to be the same as the `dom.value`, but it is guaranteed to be the same as the `oldProps.value`. If we want to update the progress bar when it is 0, we need to use the `oldProps.value` instead of the `i` value.",
+      "reference": "Turns out my previous fix of doing `!dom.hasAttribute('value')` is not a good one as it leads to all input values always being updated. The new fix only updates it, despite `dom.value === props.value` when the value is `0` for progress elements."
+    },
+    {
+      "prediction": "please don't include `<vector>`, it's not your code, but some other's",
+      "reference": "Why is this being added here? Is this relevant to these changes?"
+    },
+    {
+      "prediction": "This should be `LOG.info(\"Bound legacy RC support\", e);`",
+      "reference": "This informational message is important to users. Please leave."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this, but it's the only one I could think of that would show up in the dashboard.",
+      "reference": "why isn't rubocop failing for single quotes?"
+    },
+    {
+      "prediction": "Why did you remove the `JOIN value_translation vt ON vt.value_id = pv.value_id`?",
+      "reference": "Is `WHERE` should `Language` param ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems to be the only place where this is used.",
+      "reference": "ps - first letter from preview, when we move it into public it will not be relevant. my suggestion is spmd or ds(distributed)"
+    },
+    {
+      "prediction": "I don't think this is the right place to add this limitation. I think it should be in `apply_series_op`",
+      "reference": "Hm, should we show pandas' dtype instead of Spark data type?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be in the `metrics` namespace.",
+      "reference": "@reyang when exponentialhistogram arrives, we won't have any bits left for it.. unless we take Summary or make this 16 bits instead of current 8 bytes"
+    },
+    {
+      "prediction": "I think this should be `@phpstan-taint-sink text $class`.",
+      "reference": "This change is not tested.. I guess this is how these taint-annotations work..?"
+    },
+    {
+      "prediction": "I think this should be `obj.setdefault(k, {})` instead of `obj.setdefault(k, {}) = v`",
+      "reference": "Fails if obj is list. Please fix!"
+    },
+    {
+      "prediction": "I think you can remove the `include_all_instances` parameter from this test.",
+      "reference": "Could you add an assertion to make sure the `IncludeAllInstances` parameter is actually set and passed to the request?"
+    },
+    {
+      "prediction": "I think this should be `output = subprocess.check_output(shlex.split(' '.join(version)), env=check_env, universal_newlines=True)`",
+      "reference": "The name of this argument is not too intuitive. The point is, these kind of subprocess calls will return `str`s in Python2 and 'byte`s in Python3. It would be a lot of work to make CodeChecker handle both `str` and `byte` everywhere. Using the `universal_newlines` argument, we can force Python3 to return text rather then bytes."
+    },
+    {
+      "prediction": "I think we can remove the `Clusters` field from the `DomainReplicationConfig` struct and just use `in.GetClusters()` in the `toProtoClusterReplicationConfigurations` function.",
+      "reference": "These `Get`s are just cosmetic changes for consistency."
+    },
+    {
+      "prediction": "I think this should be `@pytest.mark.skipif(sys.platform == 'win32', reason=\"Broken due to yaml safe load\")`",
+      "reference": "We should never skip unit tests, but instead fix them. Look at `pytest.raises` for cases where we expect an exception."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `Push` method, since that is the place where the actual work is being done.",
+      "reference": "Do we need to call ToList here?"
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "I think this can be moved into the model, at the very least, but I also think you might be able to do a has many through."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a method that returns the string representation of the tree, and then call that method on the result of `toLispString` on each of the children.",
+      "reference": "could you please separate computations from string concatenation?"
+    },
+    {
+      "prediction": "I think we can remove the `errorCode` and `errorMessage` properties from the `ClientModel.ModelTypes` list as well.",
+      "reference": "@NiklasGustafsson - I had to make these changes for the x-ms-client-name extension tests for java and python. The tests were expecting wire format instead of the client name. Hence modified them appropriately. Let me know what you think ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in `h2o_headers_command_t`'s constructor.",
+      "reference": "Could we just set the list of header modification commands to `req->overrides` and apply them only when the request is sent upstream in lib/core/proxy.c? The reason I ask is because an upstream server (accessed using the modified headers) might return 399. In such case, the request would be delegated to the next handler. I think that the headers of the original request should be passed to the next handler, since per my understanding the intended behavior of `proxy.header.*` is to modify the headers passed to the upstream server only."
+    },
+    {
+      "prediction": "nit: I think we can remove `lastUpdateTime` and `lastRefreshTime` since they are no longer used.",
+      "reference": "Can we rename 'refresh' to 'read' and 'update' to 'write'? Until I had read more of the code below, I couldn't tell the difference between 'update' and 'refresh'."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. This is not a fatal error, so we should not log it. Instead, we should check if the error is a fatal error, and if so, log it.",
+      "reference": "can we remove it ?"
+    },
+    {
+      "prediction": "I think this should be: `return (memcmp(colName.data(), TRAF_DIVISION_COLNAME_PREFIX, strlen(TRAF_DIVISION_COLNAME_PREFIX)) == 0) && (colName.data()[colName.length()-1] == '_');`",
+      "reference": "I take it that by the time we get here, colName has already been case-shifted to upper case?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `client.go` since it's only used in that file.",
+      "reference": "Super clear desc - thx."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that we're calling collectLinkLibraries for all visible modules, which is not what we want. I think we need to call collectLinkLibraries for only the modules that are visible to the current module.",
+      "reference": "Ahh the correct fix is to call `collectLinkLibraries` on the SourceFile being compiled, not the module. Sorry for our mistake!"
+    },
+    {
+      "prediction": "I think this test should be renamed to `TestDefaultConfigFromFile` to match the other tests.",
+      "reference": "Can you also modify the `TestEnvironmentConfig` in `config_test.go` to cover this?"
+    },
+    {
+      "prediction": "I think we can remove the `onResponse` method as well.",
+      "reference": "There's no point \"deprecating\" this interface method (the customer still has to implement the new callback that we have introduced which makes it a breaking change). I'd make an exception and simply rename it to `onResponse`."
+    },
+    {
+      "prediction": "I think this should be `logger.info(\"Using delegate supplier: \" + clazz.getName());`",
+      "reference": "We chose `info` in the test code for obvious reasons. Changing to `finest` makes debugging harder and noisier."
+    },
+    {
+      "prediction": "I think this should be `static std::unordered_map<std::string, uintptr_t> coGlobals = {}`",
+      "reference": "I know this is not new here but we need to remove this global or protect access from multiple threads."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could get it to work.",
+      "reference": "Doesn't seem any less confusing or error-prone to me. IMO the solution is to not use \"object\" in the enrich callback signatures. Type safety/compile-time checking is our friend!"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `travis.yml` under the `before_script` section.",
+      "reference": "I think this should be `GOMAXPROCS=$(nproc 2>/dev/null || echo 2)`"
+    },
+    {
+      "prediction": "I think we can skip validation for self-proposed blocks as well.",
+      "reference": "Can we log if this actually happens."
+    },
+    {
+      "prediction": "Shouldn't this be `if (pendingConfiguration.remoteHost != null && pendingConfiguration.remoteHost.length() > 0)`?",
+      "reference": "This is not really needed, the line added in `GridNodeConfiguration.java` is what really fixes the `remoteHost` regression."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "It looks like the import format was only changed from one-per-line to all on one line here. Can we keep it as it was so it's easier to see what changed? It's also more consistent with how we format multiple imports in other components."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be in `Selenium::WebDriver::Chrome::Bridge` since Chrome is the only supported browser.",
+      "reference": "Maybe call it `WireBridge`?"
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "Looks like `org.json.*` can be removed?"
+    },
+    {
+      "prediction": "I think this widget should be renamed to `GAUNTLET_MAP_CONTAINER` or `GAUNTLET_MAP` depending on the context.",
+      "reference": "remove the extra comma and newline"
+    },
+    {
+      "prediction": "I think this should be `application_name` instead of `application_name`",
+      "reference": "Can we try to use Ruby 1.9-style hashes?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `RegisterHandlerDirective` since it's related to the handler directive.",
+      "reference": "Terminology question - these are called \"global options\" in the code, but the parallel non-global versions of these are called \"directives\"... Should this be `RegisterGlobalDirective`? Or does that have different semantics?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `controlTypes.ROLE_PASSWORDEDIT` is used to check if the role is `PASSWORDEDIT` and `STATE_PROTECTED` is used to check if the role is `PASSWORDEDIT` and `ROLE_READ_ONLY`. I think you need to check `focusObject.isReadOnly()` instead of `focusObject.isProtected` to check if the role is `PASSWORDEDIT` and `ROLE_READ_ONLY` instead of `ROLE_READ_ONLY` and `ROLE_READ_ONLY`.",
+      "reference": "I do not care much, but you could as well just return bool(focusObject and focusObject.isProtected) here and avoid the if check? I know, really trivial."
+    },
+    {
+      "prediction": "I don't think this test is necessary, it's already tested in `testClickingALinkMadeUpOfNumbersIsHandledCorrectly`",
+      "reference": "I believe it's misleading name for the condition. I prefer \"element_to_be_disable\" We can have a condition, when element is enabled but we can't click it, because another element overlays above it. So, If we use \"unclickable\" we might mislead people, who use that condition to verify if element can be clicked"
+    },
+    {
+      "prediction": "I don't think this is the right place for this change.",
+      "reference": "Do DiscoveryContext and GoGapicContext have a more specific common ancestor than Object? If so, let's use that."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "is this the intended behavior now?"
+    },
+    {
+      "prediction": "I think this test should be renamed to `testStemmingWithDutchStemmer` to match the name of the underlying test.",
+      "reference": "Hunspell/C++ stems that a bit differently from Lucene, but in a controversial way, so I removed the check instead of changing the expectation"
+    },
+    {
+      "prediction": "I think this should be `1` instead of `0`.",
+      "reference": "I'm curious why this is needed?"
+    },
+    {
+      "prediction": "why do you need to check if target is also caster?",
+      "reference": "I do not see how that changes anything, as the target is operated on many times before that line?"
+    },
+    {
+      "prediction": "I don't think we should change the priority of the `analytics-loyalty` menu item.",
+      "reference": "Same user/visitor thing."
+    },
+    {
+      "prediction": "This is a bit of a hack, but I think it's the best way to ensure that the environment is set up correctly.",
+      "reference": "This is not the correct path on all hosts. How can I correctly determine whether I will need to look in `Program Files` or `Program Files (x86)` for the script?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `JaegerExporterEventSource` class.",
+      "reference": "I think it would be better to move this try/catch block inside of the `FlushAsyncInternal` function as there are a few of places that can call it, eg event timer (here), AppendAsync, FlushAsync and CloseAsync."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to put it in the `IndexingChain` class itself.",
+      "reference": "should we set a smaller initial size to make sure that we exercise the growing logic in our tests?"
+    },
+    {
+      "prediction": "I think this should be `assert str(dt.max(f.A)) == str(f.A.max())`",
+      "reference": "Let's add a new line at the end of this file, so that the last line becomes a valid line of code."
+    },
+    {
+      "prediction": "I think this should be `parser.UnqualifiedName(field->name()->str());`",
+      "reference": "Why was this removed?"
+    },
+    {
+      "prediction": "I think this should be `Path of the SARIF log file to log the results of the validation`",
+      "reference": "I would prefer that this tool follow the driver framework conventions, which use OutputFilePath as an argument, take a look at AnalyzeCommandBase. You can propose changes to that if you'd like, but binskim, sqlskim, this tool should all conform (in the interests of building an eco-system with consistent command-line interface)"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it should be in `io/vavr/collection/CollectionUtils.java` or `io/vavr/collection/CollectionUtilsImpl.java`",
+      "reference": "(...), where each element ~are~ **is the** given {@code element}."
+    },
+    {
+      "prediction": "I don't think we need to add the AspNetCoreInstrumentation and HttpClientInstrumentation as they are already in the AspNetCoreInstrumentation and HttpClientInstrumentation packages.",
+      "reference": "What if `AddOpenTelemetryTracing` is called multiple times?"
+    },
+    {
+      "prediction": "This should be `std::make_unique<folly::RWSpinLock::ReadHolder>(std::move(wal->rollbackLock_));`",
+      "reference": "Why is the condition of \"lastId <= wal _-> lastLogId() \" added here?"
+    },
+    {
+      "prediction": "I think this should be `path.Join(baseURL, \"settings\", \"environment\", \"id\")` instead of `path.Join(baseURL, \"settings\", \"environment\")`",
+      "reference": "`environmentID` is unused in MakeEnvironmentURL"
+    },
+    {
+      "prediction": "I don't think we should be changing the visibility of the text view here. I think it should stay the same as it was before.",
+      "reference": "Just hide the textview"
+    },
+    {
+      "prediction": "I think this should be `PaymentVersionV1` instead of `PaymentVersionV2`",
+      "reference": "'Supports'? Very uninformative naming inside protocol. Why not just paymentVersion, or just recognise supported payment version from generic protocol version supported by the client."
+    },
+    {
+      "prediction": "I think this is the only place where we use `s.p` as a context. I think it would be better to use `s.Ctx` instead.",
+      "reference": "We should add an additional call to Recv() somewhere after here that we assert returns io.EOF."
+    },
+    {
+      "prediction": "I think we should move this import to the top of the file.",
+      "reference": "What is this used for?"
+    },
+    {
+      "prediction": "I think this class should be renamed to `ASTAnnotationTypeDeclaration` as well.",
+      "reference": "this is a breaking API change. Do we really need to do it in 6.2.0? Can't we just deprecate the methods?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to move this logic to the `dropAllergenTable` method in the `AllergenDao` interface.",
+      "reference": "Is dropping the **Allergen table** necessary? Can't we just add the two new columns to the existing table using a raw query?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this test. I think it should go in `spec/features/users/subscriber_views_completed_trails.feature`, but I'm not sure if there's a better place to put it.",
+      "reference": "I'm confused why the expectations here would have inverted? Seems unrelated to topic stuff. Can you clarify?"
+    },
+    {
+      "prediction": "I think we should use `std::numeric_limits<uint64_t>::max()` instead of `std::numeric_limits<int64_t>::max()`",
+      "reference": "I am very sorry that I neglected to resolve the conflict and did not fully unit test edge"
+    },
+    {
+      "prediction": "I think we can remove the `extends CloseableGroup` from this class.",
+      "reference": "I think closeable should be handled by the concrete impl of a catalog and not by the base class, I don't think all catalogs must implement closeable by design. This should be done via composition rather than inheritance. Additionally, I don't know what the effect of this change is on other catalogs so this feels a bit on the risky side."
+    },
+    {
+      "prediction": "Why do you divide by `stride_data[i]` here?",
+      "reference": "Should this be `const`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this benchmark. I think it should go in the `BenchmarkRunner` class.",
+      "reference": "Can we also add the Exporters here by using `DefaultConfig.AddExporter(...)`? At least the Datadog exporter that can't be configured from the command line..."
+    },
+    {
+      "prediction": "I think this should be `if (!opts->has_string(\"ckpt_dir\") || !opts->has_string(\"no_model1_train\"))`",
+      "reference": "@JaeseungYeom I believe you had set this in a previous commit to, by default, not train model1 when loading from ckpt. Would it work for you to have this as a command line option `--no_model1_train` instead of the default behavior?"
+    },
+    {
+      "prediction": "I think this is a real Ruby time span type. We should add it to the list of supported types.",
+      "reference": "@devigned, please clarify"
+    },
+    {
+      "prediction": "I think we can remove this line now.",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I think this should be `pip install -r requirements.txt` and not `pip install -e requirements.txt`",
+      "reference": "Should we break this up into a separate line?"
+    },
+    {
+      "prediction": "I think this should be moved to the top of the test suite.",
+      "reference": "Why the override on the env variable, wouldn't we want to use the value set from Prow?"
+    },
+    {
+      "prediction": "I think this is the wrong place to put this code. It should be in the `team_plans_controller.rb` file, not the `projects_controller.rb` file.",
+      "reference": "I don't see anything here that excludes people on a Team plan from from the revenue report. I see that TeamPlans are counted below, but the individual subscribers on the plan would still be included."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it would be better to move this check into the `slice` method itself.",
+      "reference": "I know that we do not pull the last bit of performance out of the JVM but it is more robust and removes redundant code."
+    },
+    {
+      "prediction": "Shouldn't this be `registry == null || registry != null`?",
+      "reference": "I'd make this a `throw new NullPointerException(\"registry == null\");` instead"
+    },
+    {
+      "prediction": "I'm not sure if we should have a default value for `ClusterType` or not. If we don't have a default value for `ClusterType`, I think we should have a default value for `UsageReportingIntervalSecs`.",
+      "reference": "Does adding things here require us also to extend the FelixConfiguration resource in libcalico-go?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/api` since it's only used by `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api/server` which is in `pkg/api",
+      "reference": "does this need to be public?"
+    },
+    {
+      "prediction": "I think this should be 8, not 4.",
+      "reference": "Please import this from a pre-existing definition."
+    },
+    {
+      "prediction": "`mitmweb` -> `mitmweb_server`",
+      "reference": "File name does not represent class name... if we ever cared for such things?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "I don't mind the `make_spammer` route name. It seems to more clearly convey that the state of the account will change. The \"label as\" route name seems less definitive, as if one is simply adding something to the account -- a new label -- instead of changing the status of the account."
+    },
+    {
+      "prediction": "I think this should be `EnsureTelemetryConfig` instead of `EnsureTelemetryConfig`",
+      "reference": "This exits if `-d` not specified which isn't exactly what we want, right?"
+    },
+    {
+      "prediction": "I don't think this is a good name for a helper function. It's not like it's doing anything fancy, and it doesn't seem to be used anywhere else in the codebase.",
+      "reference": "So a `*` becomes `**`? How does this affect the query? If I have `csordas*marton` as filter and I have `csordas*marton` and `csordasFOOBARmarton` as run names in the database I get only the first as result?"
+    },
+    {
+      "prediction": "I think it would be better to use `schema->getFieldType(i).type` instead of `schema->getFieldType(i).type != SupportedType::VID`",
+      "reference": "I want to know in which situation you need to traverse from \"Timestamp\" data....."
+    },
+    {
+      "prediction": "I think this should be `if err != nil && !aws.IsReaderSeekable(ctx.Body) {`",
+      "reference": "nit: hasBytes should be hashBytes"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `if __name__ == '__main__'` block.",
+      "reference": "If I enable the clangsa analyzer with the `--stats` flag the checkers will be disabled even if they were enabled previously. I think we should check here if stats was enabled. `if 'stats_enabled' in args and args.stats_enabled` we should leave them enabled."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "Note that we probably should set the default gasPrice to `0.1 gwei` considering our minimum gas limit has to be 21k. At 21k the minimum fee would end up being `0.000021 ETH` which is just a little too high"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be.",
+      "reference": "So during a config reload, this function is invoked, but prior to that the o.TrustedKeys slice is set to nil, so this code would work. However, I am wondering what happens to s.trustedKeys, I am not sure this is handled at all. (same with the new map you are adding). I am not sure if this is supposed to be supported with a config reload, but just wondering if this is not putting things into a bad state."
+    },
+    {
+      "prediction": "I think we can remove this line now.",
+      "reference": "I think the logic should be to check for token len of 0 then check for token len of less than 2. The way the logic is now, it is hard to determine if it is empty or malformed."
+    },
+    {
+      "prediction": "I think it would be better to use `strcmp()` instead of `==`.",
+      "reference": "Nothing would change here even if this `if` statement were removed."
+    },
+    {
+      "prediction": "I don't think we need to import `sql` here.",
+      "reference": "`sql` module is imported later."
+    },
+    {
+      "prediction": "nit: `nms` -> `nms_threshold`",
+      "reference": "I think it's strange, whether `assert cfg.nms.type == 'nms'` would be better."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Doing a quick search, it looks like there are other places that would be handling this promise rejection. How did you confirm this is not needed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/testutil.go`",
+      "reference": "let's use greenland in the unit tests as well"
+    },
+    {
+      "prediction": "I think this should be \"The mode that presents text in a flat representation that can be navigated with the cursor keys like in a plain text document\"",
+      "reference": "Please split this into two lines"
+    },
+    {
+      "prediction": "I think this should be `if (SelectedIndex > 6)`.",
+      "reference": "Is there any variable available instead of hardcoding 5?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to handle the result from the PasscodeManager. I think it would be better to handle it in the `LoginActivity`",
+      "reference": "The onNewPasscode method was taking care of creating the user account. Now the user account will already have been created."
+    },
+    {
+      "prediction": "I think this should be `MaxChainNameLength = 255`",
+      "reference": "This is really a minimum, isn't it, so `minPostWriteInterval` perhaps? IIUC it never gets used as a default, because the default IptablesPostWriteCheckInterval is 1s, which is larger than this."
+    },
+    {
+      "prediction": "I think you can remove the `:data => {:ajax_modal => \"trigger\"}}` from the `link_to` as it is already present in the `t('blacklight.tools.#{action}Link')` variable.",
+      "reference": "I wonder if you can use the controller's context instead of using it explicitly?"
+    },
+    {
+      "prediction": "I think this should be `appendErrorf` instead of `appendError`.",
+      "reference": "Can you add this log for `baseline` too?"
+    },
+    {
+      "prediction": "I think this should be `Create a new colour block.`",
+      "reference": "Let's call this \"NEW_BROADCAST_MESSAGE\". Msg.NEW_MESSAGE will be confusing down the line."
+    },
+    {
+      "prediction": "I think we can remove the `Deprecated: Use TransitionDependenciesMap instead.`",
+      "reference": "A better way to do this (in your future PR) would be to: 1. get rid of `TransitionDependencySet` altogether 2. tag `TransitionDependenciesMap` as `\"TransitionDependencySet\"` in JSON 3. update the JSON marshaler to deal with `\"TransitionDependencySet\"` tag. If it's a map, no big deal. Else, construct the `TransitionDependenciesMap` map. Doing that would mean that we don't need to have duplicate code, names for this."
+    },
+    {
+      "prediction": "I think this should be `AttributeError` instead of `KeyError`",
+      "reference": "It's very error-prone to have so large `try` clause with such non-specific exceptions handled. Consider rewriting it with either of these: 1. Check that expected columns are present in `DataFrame` (`{'physical_key', 'logical_key', 'size'}.issubset(df.columns)`). 2. Normalizing output of query by using `COALESCE()` (or normalizing `DataFrame` itself?). Also it looks like a row with package metadata isn't used here, so might make sense to filter out it with `WHERE logical_key IS NOT MISSING`."
+    },
+    {
+      "prediction": "I think this should be `EnableHttpGit`",
+      "reference": "Change to `DisableHTTPGit`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this logic, but it seemed like a good place to put it.",
+      "reference": "org was not used in the mail content"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to set the `disabled` property. I think it should be done in the `loadStyle` function, since it's the responsibility of that function to set the `disabled` property based on the value of the `nightShift` variable.",
+      "reference": "Is this gonna cause a lint issue, because of unused var? Maybe we should log it out?"
+    },
+    {
+      "prediction": "I think this should be `host['vmhostname'] || host.name` instead of `host[:vmhostname] || host.name`",
+      "reference": "My understanding is that this will always be set to the actual containers IP, and never to the `DOCKER_HOST` IP if that env_var is set."
+    },
+    {
+      "prediction": "I think this should be `debouncedSetInProgress = useDebounce( setInProgress, 3000, { trailing: false } );`",
+      "reference": "this dependency to a callback also has to be a callback (eslint now warns us)"
+    },
+    {
+      "prediction": "I think we can remove the `*apis.NewTestCStorPool` here.",
+      "reference": "directory name as well needs change.."
+    },
+    {
+      "prediction": "I think this should be `normalizeIdentifier` instead of `normalizeIdentifierForPlugin`.",
+      "reference": "Why is the getIdentifier() call required here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it would be better to have a function that returns the path of the file being parsed.",
+      "reference": "Looks more appropriate to stick this in `AbsolutePath`.. actually that already has `FLATBUFFERS_NO_ABSOLUTE_PATH_RESOLUTION`"
+    },
+    {
+      "prediction": "I think this should be `h('input')` instead of `h`",
+      "reference": "this might interfere when we have multiple dashboards. maybe ActionBrowseTagline needs to be a Component so we can store the reference to the element on it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `GetBlock` RPC.",
+      "reference": "Should this be inside the `else` below and right after `CDebugf` so that we still get the logs?"
+    },
+    {
+      "prediction": "This should be `dyn.ErrCodeValidationException`.",
+      "reference": "Weird they don't expose the type for dynamo and this seems a pretty common error."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this. I think it should be something like `ConfigureOpenTelemetry` or `OpenTelemetryInitializer` or something like that.",
+      "reference": "@cijothomas I couldn't make the unit tests work without a way to destroy the ActivityListener we create internally so I return it as an IDisposable here. But we'll also need to also stop any ActivityProcessors/ActivityExporters on application shutdown right?"
+    },
+    {
+      "prediction": "`wallet` -> `wallet`",
+      "reference": "The default should be \"False\", as \"True\" would be a significant change in behaviour."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this hack. I think it should be in `pkg/__init__.py` instead of `pkg/__init__.py.c`",
+      "reference": "since goal is to restore old_protocol, shouldn't we get it from pandas if possible? e.g. this will break if their source code changes."
+    },
+    {
+      "prediction": "This should be `\\n    ` instead of `str_replace('&#10;&#9;', \"\\n    \", $matches[2])`",
+      "reference": "This should consume less resources because str_replace don't need to assign arrays with all the values"
+    },
+    {
+      "prediction": "I think this should be `ast_print_type(a_type->type)` and `ast_print_type(p_type->type)` instead of `ast_print_type(a_type)` and `ast_print_type(p_type)`.",
+      "reference": "It would be better to pass `param` as the second argument here, so that the parameter declaration is printed alongside the type. Same thing for the changes in `ffi.c` and `lambda.c`."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Do we need to set the members to final like submitTime, submitUser ...? They are only used by getter method."
+    },
+    {
+      "prediction": "This should be `System\\MediaLibraryItem::class`",
+      "reference": "That makes this error message incorrect; if someone has extended Cms\\Classes\\MediaLibraryItem, then this error message will read \"Class My\\Classes\\CustomMediaLibraryItem has been deprecated\"."
+    },
+    {
+      "prediction": "Why do you need to pass the config here? I don't see where it is used.",
+      "reference": "Is it worth injecting the whole configuration object for just one setting? Would it be more generic to pass an $options array with the override_from in it? I'm not completely opposed to passing $config, but since we've avoided it so far, it seems like it would be cleaner to maintain that separation. Simply adding a getOptions method to the factory would make this a lot more readable."
+    },
+    {
+      "prediction": "This should be `const SPI0_SCK_PIN = 25;` and `const SPI0_MOSI_PIN = 23;` and `const SPI0_MISO_PIN = 24;` and `const SPI1_SCK_PIN = 2;` and `const SPI1_MOSI_PIN = 3;` and `const SPI1_MISO_PIN = 4;`",
+      "reference": "Just curious: where do these pin numbers come from? I see you took the SPI0 pin numbers from the Arduino board layout which is good, but I can't find such a relation for `SPI1`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `MethodTypeResolver`.",
+      "reference": "Isn't that already too late? I'd expect, that the LinkageError occurs the first time, we have the class instance (the classloader tried to load the class) - or do we get a partially linked class back, and we get the exception when retrieving the methods?"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for the header. It doesn't seem to have anything to do with traceparent.",
+      "reference": "nit:s/Traceparent/traceparent/ - As per the spec header name is all lowercase."
+    },
+    {
+      "prediction": "I don't think we need this function at all. We can just use `path.Dir(fullPath)` and `path.Join(fullPath, \"..\")`.",
+      "reference": "We need to have unit tests for it since it is now a public function. Also I'm not a fan for making this movement since `relPath` doesn't consume any info in this pkg."
+    },
+    {
+      "prediction": "I don't think this import should be removed.",
+      "reference": "We generally avoid wildcard imports. If you feel like changing it back, great. Otherwise, no big deal."
+    },
+    {
+      "prediction": "I don't think we need to change this line.",
+      "reference": "We also have a typo on the line above! OUt or Out?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this, but it seems like a good place to start.",
+      "reference": "OH! These are text.... (Ignore above)"
+    },
+    {
+      "prediction": "I think we can remove this as it's not used anymore.",
+      "reference": "How was this set before? Was it just ignored?"
+    },
+    {
+      "prediction": "Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "Should we `save!` here?"
+    },
+    {
+      "prediction": "Please use `snprintf` or `snprintf_s` instead of `strlcpy` and `sizeof` instead of `sizeof(\"65535\")`.",
+      "reference": "sizeof(port) seems ever-so-slightly clearer."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I would expect it to be in `NetworkPolicySpec` or `ClusterNetworkPolicySpec` but not in `Rule`.",
+      "reference": "Not sure if this has been discussed. I wonder why SourceGroups and DestinationGroups are not in From and To NetworkPolicyPeer as NetworkPolicyPeer already describes the source/destination of the traffic."
+    },
+    {
+      "prediction": "I think this should be `parts[cursor_part]` instead of `parts[cursor_part - 1]`",
+      "reference": "This doesn't account for `:bind --mode caret ...` (i.e. `--mode caret` instead of `--mode=caret`; both work). Do you see an easy way to make that work?"
+    },
+    {
+      "prediction": "I think this should be `reflect.TypeOf(&p.WorkflowConditionFailedError{}): 3`",
+      "reference": "Is this map used for durable state anywhere? If so we probably want to keep ConditionFailedError as 2"
+    },
+    {
+      "prediction": "Should this be `HEAD` or `GET`?",
+      "reference": "You can use `http.MethodHead` here."
+    },
+    {
+      "prediction": "I think this file should be renamed to `RemoteSessionTest.java`",
+      "reference": "And another empty test. Please remove until there's content."
+    },
+    {
+      "prediction": "This should be `mediasync: Begin processDownloadStatus` not `mediasync: begin processDownloadStatus`",
+      "reference": "You really didn't need to do any of these, we need to remove the API client from source eventually anyway."
+    },
+    {
+      "prediction": "why is this private?",
+      "reference": "seems pretty strange here because TiSession contains PDClient."
+    },
+    {
+      "prediction": "Align the parameters of a method call if they span more than one line.",
+      "reference": "I created a test method for this scenario. Would you normally split it out like that or bundle this into one of the existing test methods? I'm more accustomed to creating lots of separate `it` blocks in nested `context` blocks in rspec"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `NewSessionQueuer` class itself, since it's only used by the NewSessionQueuer.",
+      "reference": "should it be `newsessionqueue` or `newsessionqueuer`? In case we'd like to be consistent"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this method. I think it would be better to put it in the `IndexWriter` class.",
+      "reference": "Maybe add `@lucene.experimental`? We are exposing (slightly) internal details about `IndexWriter` so maybe we need to reserve the right to change this API in the future ..."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "@johnduarte I know this isn't part of your code change, but it seems to me that this `curl_with_retries` is not necessary. If the `retry_on` host command is successful, isn't puppetdb ostensibly listening to the ssl port as well? What do you think about removing that `curl_with_retries` after the `retry_on`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should be in `libcalico.go`",
+      "reference": "Does an empty struct really take less storage than a bool? I guess it must be, or else you wouldn't have changed this, but I'm curious about the detail."
+    },
+    {
+      "prediction": "I think this should be moved to the `braille` module, since it is only used by that module.",
+      "reference": "nit: This should perhaps be called `waitForAndReportSelectionChange`, as it affects braille as well, not just speech."
+    },
+    {
+      "prediction": "I think this should be `return` instead of `return make_error`",
+      "reference": "nit: I suggest that we rename `tag_error_response` to `tag_resource_success`, as later in line 132 we're returning an error if this value is *not* truthy."
+    },
+    {
+      "prediction": "Why do you need this? Styleguide is not a Frontend controller.",
+      "reference": "I am missing return typehint"
+    },
+    {
+      "prediction": "`_base_` should be removed.",
+      "reference": "_base_ = 'retinanet_pvt-t_fpn_1x_coco.py'"
+    },
+    {
+      "prediction": "Shouldn't this be `onAuthFlowError(String message, Exception e)`?",
+      "reference": "We should remove `throws Exception` here, since we're catching it anyway and triggering `onAuthFlowError`."
+    },
+    {
+      "prediction": "I think this should be `go-libp2p-pubsub.PubSub`",
+      "reference": "a super thin wrapper like this makes me think we should consider not having it, instead just defining in the plumbing.api the publisher and subscriber interfaces we expect a ps dependency passed in to support and then to pass calls directly to that thing rather than through this wrapper"
+    },
+    {
+      "prediction": "nit: `fine` is a verb, `log.fine` is a noun.",
+      "reference": "Waiting for the pull takes a long time. This message informs the user that at least one of the images being pulled is available. Please leave."
+    },
+    {
+      "prediction": "I think we can remove this line now, since we don't need it for 3.6.1.",
+      "reference": "We should set a 1.7 upper bound for this as well. :smiley:"
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this component. It's not a component in the traditional sense since it doesn't have a container or parent. It's also not a component in the sense that it doesn't do anything outside of the context of the sidebar. Maybe something like `SidebarNavigation` or `SidebarNavigationComponent`?",
+      "reference": "Lodash mixed in with WordPress dependencies"
+    },
+    {
+      "prediction": "I think we can remove the `stripe_subscription` association from the `Subscription` model as well, since we're not using it anymore.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this spec. I think it should go in `spec/controllers/users/edit_profile_controller_spec.rb`",
+      "reference": "perhaps we should test that users cannot edit the profiles of others?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to check this. I think it should be in the server tests.",
+      "reference": "Does this mean the content changed with the new pyopenssl version? Is Wireshark already compatible with this new format?"
+    },
+    {
+      "prediction": "I think we can remove the `config.Default` here.",
+      "reference": "line is 147 characters (from `lll`)"
+    },
+    {
+      "prediction": "I think this should be `show_speech_viewer_at_startup = boolean(default=False)`",
+      "reference": "Perhaps move this option into a separate speechViewer section, as we are planning to also save position information as well."
+    },
+    {
+      "prediction": "I think this should be `?:` instead of `?`",
+      "reference": "This is not a segment responsibility this query should be in exporter-File module"
+    },
+    {
+      "prediction": "I think this should be `--export-fixes-file`, not `--export-fixes`.",
+      "reference": "Did you measure the performance with and without this option. Do we want to export fixits even if the `clang-apply-replacements` tool is not available in the user's PATH?"
+    },
+    {
+      "prediction": "I think we can remove the `BlockDevices` field from the `SpcObjectClient` struct.",
+      "reference": "What if we do not change this file ? Are we using this `pkg/disk/v1alpha1` elsewhere for block device requirement?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "regarding dropping the port/host/uri methods on Nod: this would `options[:port] = target.port if target.port`"
+    },
+    {
+      "prediction": "I'm not sure why we need this utility method. I think we can just call `FileUtil.determineFileType(file, fileName)` directly.",
+      "reference": "good question - I vote we just remove, up to you if you want to handle in this PR though. (as it would require changing the places you call it; maybe moving the test? Though my guess is if we don't now, we never will."
+    },
+    {
+      "prediction": "I think this should be `use_directory_cache?` instead of `use_directory_cache? or data.cache?(:bundler)`",
+      "reference": "This is so that if we turn on bundler caching globally it still won't affect python etc. Same inheritance logic as for the cache slug."
+    },
+    {
+      "prediction": "I think this should be `this.SupportsAllocation(contexts) && !IsAllocated(contexts)`",
+      "reference": "[You can ignore it] Maybe a better method name would be IsSupported or HasSupportForAllocation?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix, but I'm not sure what else to do either.",
+      "reference": "Why is this changed to `\\*`? The behavior of the code on line 42, shows that `*` is what's handled?"
+    },
+    {
+      "prediction": "I think this should be `schemaVersion: 1, path: 'another.realm'`",
+      "reference": "I think this should either return `-1` or throw. I think my vote is on the former."
+    },
+    {
+      "prediction": "I think this should default to 30 seconds, since that's how long it takes to get a heartbeat back from the kinto servers.",
+      "reference": "I am confused, I've read 2, 3 and 5 seconds in various places"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to get the content length directly from the header.",
+      "reference": "Any particular reason you chose int64 instead of int? This requires adding type conversions throughout the code below."
+    },
+    {
+      "prediction": "I think this should be `match='The team you specified is not a valid team.'`",
+      "reference": "this string should really be a local constant. it's repeated below."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `JaegerAttributes` since that's where we store all the attributes.",
+      "reference": "Number types are hard  . Should we be considering byte/sbyte? What about unsigned variants that would otherwise fit in a signed 64 bit integer like UInt16, UInt32?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "`Assert.True(true)`...   xunit used to have Assert.DoesNotThrow"
+    },
+    {
+      "prediction": "I think we can remove the `us-east-1` portion of the region, since it's not used.",
+      "reference": "should this be an env var like the other `S3_*` values?"
+    },
+    {
+      "prediction": "`parentNode` can be `null` if the node is not attached.",
+      "reference": "If you do go this route, I think it would be worth adding a note here that IE 11 is the only reason why we're not using the more obvious `node.remove()`."
+    },
+    {
+      "prediction": "I think this should be `if key in dictionary` instead of `if key in dictionary`",
+      "reference": "The added text doesn't really match why it's possible. Maybe it would be better to name that as a separate case for this checker? (e.g. dict lookup is quicker than list comparison)"
+    },
+    {
+      "prediction": "Why did you change this?",
+      "reference": "just whether we are not missing some doctrine extension because above this attribute is nullable position and it shows int instead of int|null but maybe for lvl2 it is OK"
+    },
+    {
+      "prediction": "I think this should be `slots.size() > 0` instead of `slots.stream().anyMatch(slot -> slot.isSupporting(caps))`",
+      "reference": "Prefer `Stream.anyMatch` instead of iterating over all slots."
+    },
+    {
+      "prediction": "I think this should be `else if (drm_format_mod_explict != nullptr) {`",
+      "reference": ":grimacing: yikes, thanks for catching this!"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface. I think it should be in `MvvmCross.Droid.Support.V7.RecyclerView.Model`",
+      "reference": "`Binded` is weird. I think it needs to be something with `Bound` in it (and drop the `Mvx` bit) like `ViewHolderBound` or something."
+    },
+    {
+      "prediction": "Why do you need to pass diskList here?",
+      "reference": "can we have String() and GoString() implementation of `apis.CStorPool` struct to pretty print it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in `ParseTableDelimiters`.",
+      "reference": "Like I said, please replace by `EXPECT(kTokenString)`"
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "Will you please update the copyright message to `2021` for classes that you modify?"
+    },
+    {
+      "prediction": "I think it would be better to use `isArray` here instead of `entity.hasOwnProperty('$$hashKey')`",
+      "reference": "Is there a better way to filter these other than relying on the hashKey. I few like this might now be trustworthy in the long run. If angular ever removes this property, this feature would be broken."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't have a better suggestion. The problem is that if we set `dynamo_exited_synched` to `true` in `os_tls_thread_exit()`, we can't guarantee that `last_thread_tls_exited` will always be `false` when we call `os_tls_thread_exit()`. This is because `os_tls_thread_exit()` can be called concurrently with `os_thread_exit()`, and `os_thread_exit()` can return before `os_tls_thread_exit()` has a chance to set `last_thread_tls_exited` to `false`. In other words, `os_tls_thread_exit()` doesn't guarantee that `last_thread_tls_exited` will always be `false` when we call `os_tls_thread_exit()`, so we can't guarantee that `last_thread_tls_initialized` will always be `true` when we call `os_tls_thread_exit()`.",
+      "reference": "No need to check dynamo_exited"
+    },
+    {
+      "prediction": "I think we should rename this to `SetTLSCertificate` or `SetTLSCertificateFile` to be consistent with the other functions.",
+      "reference": "from linter: >method SetTlsCACertificate should be SetTLSCACertificate"
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "this data had incorrectly aligned timestamps (which wasn't the point of the test)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `Workflow` struct, since it's only used by that class.",
+      "reference": "Since this is workflow agnostic, should we put this in common?"
+    },
+    {
+      "prediction": "I think this should be called `version_compatible_analyzer_version` instead of `version_compatible`",
+      "reference": "using the `@abstractmethod` decorator wouldn't be better?"
+    },
+    {
+      "prediction": "Why do we need to check this? Isn't `isBoolean` enough?",
+      "reference": "This was crashing the app during EDIT on boolean Doctrine Embedded properties"
+    },
+    {
+      "prediction": "I think this should be DFL_SYSFS_SEC_GLOB \"*dfl*/*flash_count*/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash_count/*flash",
+      "reference": "Will this be backwards compatible with the previous path?"
+    },
+    {
+      "prediction": "I think this should be called `CurrentScope` instead of `CurrentSpan`.",
+      "reference": "what is potential use for current scope? It seems you'd only want it to stop it. But if you get current scope you never know if it's yours to stop - i.e. this is not safe or correct to stop current scope. So I wonder should we even try to expose it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move this test to `describe('Search term input', () => {`)",
+      "reference": "I think this is likely the only additional change needed on the original PR."
+    },
+    {
+      "prediction": "Remove empty line.",
+      "reference": "Please revert changes to unrelated file"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `org.mortbay.jetty` since it's only used in that package.",
+      "reference": "Can test use a Mockito mock instance instead?"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this check. I think it should be in `org.graylog2.plugin.invoker.HandlerInvocationTreeMatcher`",
+      "reference": "Bit of a shame about these multi-line strings in annotation parameters - they're not a deal breaker but just make the diff noisier"
+    },
+    {
+      "prediction": "I don't think we need this method. We can just use `IEnumerable<T>.GetType()`",
+      "reference": "we can remove this one as this is not used any more"
+    },
+    {
+      "prediction": "I think you can remove the `error` variable now.",
+      "reference": "we don't really need an extra error flag"
+    },
+    {
+      "prediction": "I don't think you need to import `net.ltgt.gradle.errorprone.javacplugin.CheckSeverity`.",
+      "reference": "any chance you could just fix the processors plugin?"
+    },
+    {
+      "prediction": "`use_depthwise` -> `depthwise`",
+      "reference": "We may keep this config unchanged since `use_depthwise=True` by default."
+    },
+    {
+      "prediction": "I think this should be `#if defined(__FreeBSD__) || defined(__OpenBSD__)` instead of `#if defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__DragonFly__)`",
+      "reference": "Maybe add i#956 in this log as well."
+    },
+    {
+      "prediction": "This should be `inventoryWeight += storeInbox->getWeight() + item->getWeight();`",
+      "reference": "if (StoreInbox* storeInbox = getStoreInbox()) {"
+    },
+    {
+      "prediction": "I think this should be called `importBip39` instead of `importAllFromMnemonic`",
+      "reference": "what is this, is it going to be used later?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "@anodelman test failure is probably due to this, needs to be fixed to `(\"Host #{host.name}\")`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `@include_comments_files` block.",
+      "reference": "I'm not sure that we need this anymore, but that can be a separate discussion."
+    },
+    {
+      "prediction": "I don't think this is the right place for this method. It's not about closing the connection, it's about closing the connection to the server.",
+      "reference": "The formatting seems different from the rest of the code"
+    },
+    {
+      "prediction": "I think this should be `m.Spec.ProberSpec.ProxyURL` and not `m.Spec.ProberSpec.ProxyURL`",
+      "reference": "Could we have a unit test for this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to wait for the service to stop before we stop the app.",
+      "reference": "nit: put params in new line"
+    },
+    {
+      "prediction": "I think this should be `text` instead of `text` and `icon` should be `button-icon--mobile`.",
+      "reference": "The button for the date range selector has incorrect width since it has `padding-right: 8px` which sets the width to: `44px`. Can you review to make the button `36px` on small screens as per the AC?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `if (!d_r_is_avx512_code_in_use())` block above.",
+      "reference": "This should be inside the set routine IMHO: matches the others; all callers need it; simplifies code here."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the commit phase.",
+      "reference": "MAX_RETRY_TIMES makes much more sense."
+    },
+    {
+      "prediction": "I think this should be `*fakeStorage` instead of `*fakeStorage`",
+      "reference": "`saveSession` allows easier mocking - just save function which You need, instead full interface"
+    },
+    {
+      "prediction": "`encoding_type` is not used in this method, it should be removed.",
+      "reference": "No docs for the new param here?"
+    },
+    {
+      "prediction": "Can you add a license header to the top of the file?",
+      "reference": "This file is unrelated to the aims of this PR. Please remove. If it is needed to demonstrate Sonar Deltas then it has been proven and can be removed."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `TraceLogging`, but I'm not sure if that's the right place for it. I think it should be in `TraceLogging`, but I'm not sure if that's the right place for it. I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `TraceLogging`, but I think it should be in `Trace",
+      "reference": "minor: `Export all ended spans to the configured Exporter that have not yet been exported.` - This is the spec description. Lets use something on that line. \"queue\" is not necessarily present for all processor."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this debug log. I think it should go in the `ExecuteAndWaitProgress()` function, since it's related to the `ExecuteAndWaitProgress()` output.",
+      "reference": "is this deliberate? or testing?"
+    },
+    {
+      "prediction": "I think this should be a `List<IBaseProcessor<T>>`.",
+      "reference": "I want to get more perspectives regarding this one. Making it `readonly` makes me feel that we're assuming the head should never change, and my worry is that other code might assume it (e.g. they might cache the value and assume it will never change). While this is true for now, I guess in the future we might want to support something like `PrependProcessor(processor)` or `InsertProcessor(index, processor)`."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that the range query does not include the end of the range.",
+      "reference": "This isn't critical for test coverage, but it seemed off that we had disabled wrapping the reader."
+    },
+    {
+      "prediction": "I don't think we need to import all these modules here. We can import just the ones we need.",
+      "reference": "Please remove Descriptors3D from this file. We just did this in master."
+    },
+    {
+      "prediction": "I think this should be `folly::SemiFuture<StorageRpcResponse<cpp2::ExecResponse>>`",
+      "reference": "You could pass the MetaClient instance in ctor, and create a new instance if nullptr."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "What does this do? Seems like a poor mans mutex/semaphore?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `Write` instead.",
+      "reference": "This is a breaking change, and cannot be made. We have committed to this API for v1."
+    },
+    {
+      "prediction": "I think this should be `Context|null`",
+      "reference": "Can also drop corresponding types from docblock here"
+    },
+    {
+      "prediction": "`log.info(\"Got a request to create a new session: \" + new DesiredCapabilities(request.getDesiredCapabilities()));`",
+      "reference": "This change is unhelpful to users."
+    },
+    {
+      "prediction": "I think this should be `var _test3_a = _o.Test3.A; var _test3_b = _o.Test3.B; return new Vec3(_test3_a, _test3_b);`",
+      "reference": "why is this variable introduced? Please stick `_o.Test3.A` directly in the call below."
+    },
+    {
+      "prediction": "I don't think we need the `gentleVerifyToken` here, since it's the same as the `hasSessionAndProvider` check.",
+      "reference": "do you mind sharing what is the reason for this change?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with right now.",
+      "reference": "thoughts on putting `alter_query { |rel| rel.limit(@closed_proposal_limit) }` in a method that we can call here? That way we can have a test for this logic without needing a controller spec."
+    },
+    {
+      "prediction": "Indent the first parameter one step more than the start of the previous line.",
+      "reference": "Perhaps `@node` throughout"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Shouldn't this default be 'normal' ?"
+    },
+    {
+      "prediction": "I think this should be `num_pipelines_in_repo num_sensors_in_repo num_sensors_in_repo repo_hash python_version metadata version dagster_version os_desc os_platform`",
+      "reference": "nit: type this"
+    },
+    {
+      "prediction": "I think we should rename this to `START_CAST_TAB_MIRRORING` as it is not related to media-routers.",
+      "reference": "These command names are specific to Chromium-based browsers. Please move to `ChromiumDriverCommand`"
+    },
+    {
+      "prediction": "I think we can simplify this to `block device %s doesn't belong to node %s`, but I'm not sure.",
+      "reference": "lets print nodeName that got passed also"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for Python 3 support.",
+      "reference": "Should be `if sys.version_info[0] > 2:`"
+    },
+    {
+      "prediction": "I think it would be better to move this to the top of the file, so that it is loaded before the rest of the code.",
+      "reference": "Is there any need to pass this through `string_metric_parse`? I believe you can either create an off_t or just pass 0, with no need to added a string conversion into the mix."
+    },
+    {
+      "prediction": "I don't think we need this method. We can just use `self.driver.execute(Command.PRINT_PAGE, print_options)`, right?",
+      "reference": "Let's not use a form of hungarian notation in naming our variables"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in `H2O_PROBE_CONN`",
+      "reference": "Seems like we are calling the probe twice?"
+    },
+    {
+      "prediction": "I think this should be `exec [app_name] [environment_name] [cmd]`.",
+      "reference": "I think we'll want @rickmanelius (or somebody) to go through all the help and make it more accessible. Probably later in the cycle. But \"Run a command in an app container\" doesn't do it for me :)"
+    },
+    {
+      "prediction": "I'm not sure if this change is correct. I think it should be `spec.IPFamilies` instead of `spec.IPFamilies`.",
+      "reference": "Have not tested this yet, lets see if tests complain, but I suspect it should be as easy as this"
+    },
+    {
+      "prediction": "I think it would be better to have this as a constant in `pyspark.sql`.",
+      "reference": "no big deal but we don't we just use a list to keep the order? I don't think it's likely to have a duplicated columns if that was the concern."
+    },
+    {
+      "prediction": "I think we should keep the `sitekit` prefix here.",
+      "reference": "See above, same for all similar cases below."
+    },
+    {
+      "prediction": "I think this should be `disable-snapshot-annotations` instead of `disable_snapshot_annotations`",
+      "reference": "where is this coming from?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "We can also drop `plans.includes_screencasts`, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If we're going to do this, we should be doing it in a way that doesn't require us to have a lot of entropy available. I'm not sure how much, but if we're going to do this, we should be able to do it in a way that doesn't require us to have a lot of entropy available. I'm not sure how much, but if we're going to do this, we should be able to do it in a way that doesn't require us to have a lot of entropy available. I'm not sure how much, but if we're going to do this, we should be able to do it in a way that doesn't require us to have a lot of entropy available. I'm not sure how much, but if we're going to do this, we should be able to do it in a way that doesn't require us to have a lot of entropy available. I'm not sure how much, but if we're going to do this, we should be able to do it in a way that doesn't require us to have a lot of entropy available. I'm not sure how much, but if we're going to do this, we should be able",
+      "reference": "wouldn't `int.to_bytes(counter, 32, 'little')` be equivalent, clearer and faster?"
+    },
+    {
+      "prediction": "I think it would be better to move this function to the `storage/deal.go` file, since it is only used by that file.",
+      "reference": "I don't see why this is in porcelain since it's just a free function, with no dependency on plumbing or the `a` receiver. I think it should be moved to `protocol/storage`. It's exposed unnecessarily widely here."
+    },
+    {
+      "prediction": "I think it would be better to have a `clip` function in the `UserEntityStatMessage` class, and call it here and in the `create_messages` function.",
+      "reference": "Could we only do this for all time? Because that's what is causing problems rn?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to use the `resolve` method on the `CommandManager` itself, which would allow us to use the `CommandManager`'s own `resolve` method instead of `view.flows.resolve`. This would allow us to remove the `execute` method from the `CommandManager` class and make it more reusable.",
+      "reference": "This looks better than before, but we'll now likely run into issues with `'` characters in the spec. Maybe we can just use `manager.call_strings` instead?"
+    },
+    {
+      "prediction": "I think we can remove this method and just use `confParser.longConf().option(SparkReadOptions.SPLIT_SIZE).parseOptional();`",
+      "reference": "Should this replace `splitSize` instead of adding a parallel call? The `SparkReadConf` is not yet released, so we can change it still."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I'm not sure we should be relying on `FileNotExistsError` to handle the EEXIST case. I think we should be using `os.path.exists` to see if the path already exists, and if it does, we should raise a `FileExistsError`.",
+      "reference": "To reduce complexity, please use OSError on Python3 as well."
+    },
+    {
+      "prediction": "`cur_tab` is not a tab, it's the current tab. I think this should be `tab.caret.reverse_selection()`",
+      "reference": "That seems wrong."
+    },
+    {
+      "prediction": "I'm not sure if this is needed, but I think it's a good idea to call `session.configureSessionTimeout()` here, since this is the first time we're using the session, and it would be nice to make sure we're using it correctly.",
+      "reference": "Yes, this is needed here. This is a builtin user who just reset their password."
+    },
+    {
+      "prediction": "I think this should be `self._object_path(self._pkg_dir + '.' + buildfile)`",
+      "reference": "This seems to move the storage of temporary files to the CWD. Is that right? I don't think we should do that. If the process gets interrupted, we should try our best to clean up, but if even that fails, it'd be nice if the mess was left in a different directory. Maybe we should have a directory explicitly for builds?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `server/server.go` or `client/client.go`.",
+      "reference": "Please explain why a user would enable this (space savings), and the requirements for this option to be safe."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure if it's the best way to do this. I'm not sure",
+      "reference": "I think the `undefined` checks still need to be removed from here."
+    },
+    {
+      "prediction": "I think this should be a constant in the `kubernetes` module.",
+      "reference": "Is there any particular reason to have 6?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `space.go` since it's only used in that file.",
+      "reference": "The biggest chunk does not guarantee that it contains the biggest CIDR-aligned range. If we don't care too much about a few CPU cycles being wasted, then I'd suggest to merge the if-statements."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems reasonable to me.",
+      "reference": "instead of doing the if err := __; err != nil { return err } thing, you can do errs = addErrs(errs, ___). If you want."
+    },
+    {
+      "prediction": "I think we should rename this file to ecs-cli.go to be consistent with the other files in this package.",
+      "reference": "nit: Is this error message accurate?"
+    },
+    {
+      "prediction": "(style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO_MODE_BARE' may be misspelled - perhaps 'OSTREE_REPO_BARE'? (style) 'OSTREE_REPO",
+      "reference": "Ouch! :man_facepalming: And of course, this worked for me because that evaluated to false when I was testing the bare-user path."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it. I think we should be able to get the status from the response itself.",
+      "reference": "Parentheses aren't needed here. `if len(statusParts) > 0 {` will suffice."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should go in `testing` package, since it's not specific to a particular test.",
+      "reference": "// the test is skipped on given servers because it will fail."
+    },
+    {
+      "prediction": "Layout/AlignParameters: Align the parameters of a method call if they span more than one line.",
+      "reference": "The name of the test should be updated to \"it 'should return bad_request if api key is invalid\""
+    },
+    {
+      "prediction": "I think this should be `if jobInfo.workerCompletionMsg is not None:`",
+      "reference": "Use `==` in this case instead of `is`."
+    },
+    {
+      "prediction": "Can we rename this to `CStorPoolPhaseType`?",
+      "reference": "Please put it like this: `CStorPoolPhase is a typed string for phase field of CStorPool`"
+    },
+    {
+      "prediction": "Why not using `ConfigQuery::getAvailableStock()` ?",
+      "reference": "variable must be in camelCase => $newStock"
+    },
+    {
+      "prediction": "I think this should be `int` instead of `int32`",
+      "reference": "unnecessary conversion (from `unconvert`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `BitMappableTrie` since it's only used by that class.",
+      "reference": "... is probably automatically vectorized, since it has the same speed as `System.arraycopy`"
+    },
+    {
+      "prediction": "I think this should be `errors.New(\"cgroup not enabled\")` instead of `var control cgroup.Control = cgroup.New()`",
+      "reference": "Is there anyway to get rid of the global variable, if it is specific for each task, can you add it to the `mtask` struct?"
+    },
+    {
+      "prediction": "I think this should be `forall [t0, t1, t2] (a: t0, <-b: t1, b: t2, ?c: int) -> int` and not `forall [t0, t1, t2] (a: t0, <-b: t1, b: t2, ?c: int) -> int`",
+      "reference": "Surprised to see this old syntax here, but not _that_ surprised."
+    },
+    {
+      "prediction": "This method is not used anywhere, so I think it should be removed.",
+      "reference": "It is good to mark it! Maybe we can remove the method for Vavr 1.0.0 - but for now it can stay as-is. Thx!"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `pkg/exec` since it's a separate package from `pkg/maya` and `pkg/maya/exporter`.",
+      "reference": "exported func New returns unexported type *pool.pool, which can be annoying to use"
+    },
+    {
+      "prediction": "I think this should be `:parallel` instead of `:transports` to be consistent with the rest of the module.",
+      "reference": "Sorry, I feel like I'm missing something - where does this get called?"
+    },
+    {
+      "prediction": "I think we can remove the `else` block here.",
+      "reference": "@guihecheng Thanks for the fixing! Actually this is not the dup codes, when we first registerNode, then update the nodestatus immediately, or the node will keep not ready for 10s. And do you have other ways to refactor it?"
+    },
+    {
+      "prediction": "This should be `#include <sys/stat.h>`",
+      "reference": "How portable is this? (e.g., Windows). I'm relaunching tests of windows because they failed for other reasons."
+    },
+    {
+      "prediction": "I think this should be `const auto &set_binding_pair = pPipe->active_slots[setIndex];`",
+      "reference": "What this does is narrow down what we are saying the descriptor set is requiring. This is a good thing, but doesn't go far enough, we are still potentially setting unused bindings to either a required state or an invalid one. set_binding_pair.second is a map of the binding/requirements from the bound shaders... and instead of marking all bindings, looking at that will mark only the used ones. Will open a new PR"
+    },
+    {
+      "prediction": "I don't think this is the right place for this error. I think it should be in the `BlockMeta` struct itself.",
+      "reference": "should use `codes.InvalidArgument`, same as above line 1033 and 1036"
+    },
+    {
+      "prediction": "I think this should be `GetDefaultDockerAPIVersion`",
+      "reference": "Please revert this file."
+    },
+    {
+      "prediction": "I think it would be better to say \"See https://godoc.org/gocloud.dev#hdr-URLs for background information.\"",
+      "reference": "Probably in another PR, I think the package should be renamed to `servicebus` to be consistent."
+    },
+    {
+      "prediction": "I don't think this is true. Every swagger/configuration must have a security definitions section and it must adhere to the structure described in: https://github.com/Azure/autorest/tree/master/docs/developer/validation-rules/security-definitions-structure-validation.md.",
+      "reference": "curious: are we planning to add link to each of the rules in the message?"
+    },
+    {
+      "prediction": "I think this should be `SettleDepositOnSubChain`",
+      "reference": "line is 123 characters"
+    },
+    {
+      "prediction": "Why did you change the parameter type from `array` to `array $user`?",
+      "reference": "same here; `@param array $user Array of user data`"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "A quick read of this test indicates that we're changing the behavior of this api, and so the test needs updating. Is that right?"
+    },
+    {
+      "prediction": "I don't think we need to change the name of this class.",
+      "reference": "Register needs lowercasing"
+    },
+    {
+      "prediction": "This should be `public static final boolean DEFAULT_VALIDATE_PROXY_USER = true;`",
+      "reference": "This should belong to FlowContainer class."
+    },
+    {
+      "prediction": "I'm not sure this is a good idea, but I don't have a better suggestion.",
+      "reference": "I think there is a chance that on disable/enable cycling, referring to `self.window` is problematic. In the `__init__` method `self.window` is set to None. If `__init__` gets called before `_run` exits the `while` loop then we wont call `self.window.destroy()`. Can you use non-instance variables after this point? I don't think `self.timer` is used outside this function, if so, then I'd prefer it also be a local variable."
+    },
+    {
+      "prediction": "I think this is the only test that needs to be changed to use `os.Setenv` instead of `os.SetenvNoninteractive`.",
+      "reference": "I was confused by this env name variable, assuming it was the value, not the name. Silly nit, but maybe name it noninteractiveEnv?"
+    },
+    {
+      "prediction": "I think it would be better to have a constant for the default value of this property.",
+      "reference": "nice, having the \"overwrite\" default makes this change backwards compatible, right?"
+    },
+    {
+      "prediction": "Shouldn't this be `onStatus((s) -> !s.is2xxSuccessful(), (response) -> parse(response).map(userInfoErrorResponse -> userInfoErrorResponse.getErrorObject()))`?",
+      "reference": "It doesn't make sense to accept all 2xx status code. For example, `203 Non-Authoritative Information` or `205 Reset Content`, etc."
+    },
+    {
+      "prediction": "`BytesRef` can be replaced with `byte[]`",
+      "reference": "Found a better choice here that doesn't involve object creation: StringHelper.startsWith()"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `nodeattestor` since it's the node attestor that is doing the attestation.",
+      "reference": "Shouldn't we consider the situation where it doesn't find a plugin?"
+    },
+    {
+      "prediction": "I think this should be `throw invalid_argument(dal::detail::error_messages::unsupported_read_mode(mode));`",
+      "reference": "Do not understand why this needed for table. All new classes defined in the `preview` namespace."
+    },
+    {
+      "prediction": "I think this should be called `NOTIFICATION_TYPE_PRINCIPAL_EXPIRY_NOTIFICATION`.",
+      "reference": "Kept the same property value even though it is now in Common to avoid breaking existing clients."
+    },
+    {
+      "prediction": "I think we can remove the `Schema` parameter from the constructor.",
+      "reference": "TODO: this is temporary until we figure out case sensitivity for metrics evaluators."
+    },
+    {
+      "prediction": "I think we can remove the `Syncer.HandleNewTipset` and `Syncer.HandleNewBlocks` calls here.",
+      "reference": "can we just kill this TODO?"
+    },
+    {
+      "prediction": "I think this should be `invalid path \"awful\": selected encoding not supported`",
+      "reference": "This appears to be a change in error handling due to a dag upgrade."
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.",
+      "reference": "These changes look like a hosts file _must_ be provided?"
+    },
+    {
+      "prediction": "Why not use `PyUnicode_FromString(\"O\")`?",
+      "reference": "@akohlmey Just make it a cast like the others. I dislike it either way, but that's the price we pay for Python 2 support."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `build-system/runner` package, since it's related to the runner itself.",
+      "reference": "I wonder if the default here should be nightly? I won't block on that though just a question for a follow up."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with that doesn't break anything.",
+      "reference": "Also not for merge to master."
+    },
+    {
+      "prediction": "I think this should be `len(self._cron_schedule.split(\" \")) == 7`",
+      "reference": "nit `non-standard` might be a bit subjective - maybe \"Cron schedule {blah} is in an unrecognized format. Dagster cron strings must consist of five characters separated by a single space chacater\"."
+    },
+    {
+      "prediction": "I think this should be `clickView(com.salesforce.androidsdk.R.id.sf__picker_custom_url);`",
+      "reference": "ARM emulator gets tripped up on the copy/paste menu when tapping the text. Tapping on the text field isn't necessary for setting text."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. It seems like it should be in the spellchecker, but I'm not sure how to make it work.",
+      "reference": "extra word lookup removed"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it seems like it should be in the same namespace as the `wait_for_condition` function.",
+      "reference": "Will you link to {!webdriver.promise.Promise} here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this route. I think it should be in the `pages_controller`",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think we can remove the `Intents.ACTION_NEW_PROFILE` and `Intents.ACTION_NEW_TREATMENT` from the `else if` above as they are now handled by `else if (isNSProfile)`",
+      "reference": "@AdrianLxM finally found some your bug too :)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the same place as `SetImageViewLayout`.",
+      "reference": "This crashed in an app using descriptor_indexing with some unbound descriptors."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I don't have a better suggestion.",
+      "reference": "I think we might want to keep these marks separate? Then we can create a new `workflow` file with 3 jobs: 1) Prime stdlib 2) Prime batch 1 of external 2) Prime batch 2 of external"
+    },
+    {
+      "prediction": "I think this should be `if (childViewModelCache != null && childViewModelCache.Exists(fragInfo.CachedFragment.ViewModel.GetType()))`",
+      "reference": "Throw `fragInfo.CachedFragment.ViewModel.GetType()` into a variable so it's not evaluated each time."
+    },
+    {
+      "prediction": "Style/HashSyntax: Use the new Ruby 1.9 hash syntax.<br>Style/SymbolArray: Use %i or %I for an array of symbols.",
+      "reference": "Rather than just getting rid of the memoization maybe we should use an `after_save` filter to set it to `nil` so that it will be recomputed on next read?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `src/dom.js`.",
+      "reference": "JS abstraction leaks so this should be `if (el.parentNode !== null) { el.parentNode.removeChild(el) }` @developit"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `ledger` package.",
+      "reference": "nit: change it to `Exchange: unexpected status 0x%x` so that we'll know where this issue came from as well as making it easier to decode."
+    },
+    {
+      "prediction": "I think this is not a good name for this function. It is not a run task. It is a function that returns a yaml string.",
+      "reference": "Name of constant is not appropriate. Name of constant should make use of version no. as suffix. The constant can be a typed string versus a string."
+    },
+    {
+      "prediction": "I think this should be `if options[:tty]` instead of `if options[:tty] && options[:tty]`",
+      "reference": "@glennsarti you want to validate this on Windows? IIRC this won't work"
+    },
+    {
+      "prediction": "I don't think this is the right place to set the lookback. I think it should be in the `ZipkinElasticsearchHttpStorageProperties`",
+      "reference": "ps I'm ok to add this as a elasticsearch property instead. ex zipkin.storage.elasticsearch.names-lookback (we have similar things in cassandra)"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think we need to force a reindex when the database has changed, not when the package list has changed.",
+      "reference": "Curly brackets here `{ (breakLine) xx (breakLine)}`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check if the username is already taken. I think it would be better to check if the username already exists in the database.",
+      "reference": "Can you check this case Assuming you have a dashboard with 2 users, test1 and test2 Currently logged in user is test1. You should not be able to change it to test2. You open dev console and paste in `countlyGlobal[\"member\"].username = \"test2\";` then you enter value test2 in the input and it is allowed and form would be submitted. Would server side validation deal with such case?"
+    },
+    {
+      "prediction": "I think this should be `System.getProperty(DRIVER_CLASS_NAME, \"com.mysql.cj.jdbc.Driver\")`",
+      "reference": "since we're changing this class, let's fix the default class name as well. it should be com.mysql.cj.jdbc.Driver"
+    },
+    {
+      "prediction": "Line is too long. [87/80]",
+      "reference": "At load time, we ensure the config'd objects exist."
+    },
+    {
+      "prediction": "I think this should be `else if (!parsedAddress.Scheme.Equals(\"https\", StringComparison.OrdinalIgnoreCase))`",
+      "reference": "Is there a test for this case?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to display the number of videos, but it's the only way I could get to work.",
+      "reference": "This copy doesn't match what I had in my mockup. I think it's better to call them episodes."
+    },
+    {
+      "prediction": "Why do we need to check for `FrontendServiceName` and `MatchingServiceName`?",
+      "reference": "For some reason it worked for frontend even before this change."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be able to do something like `std::unique_lock<std::mutex> lk (nano::random_pool_mutex);` and then do `nano::random_pool.GenerateBlock (account.bytes.data (), account.bytes.size (), std::ref (lk));`",
+      "reference": "Can we encapsulate this in a function so if we want to change it against we don't have to change all call sites?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it's the way I've always done it.",
+      "reference": "Maybe a good opportunity to do a related improvement: The parameter `total_blocks` in `bootstrap_attempt::process_block` shadows a field with the same name, so maybe rename the parameter to `total_blocks_a` The parameter and field are incremented at the same time, but the field is used by `populate_connections` as well, so I assume they mean different things. Other than that, LGTM"
+    },
+    {
+      "prediction": "I think it would be better to use `DEFINE_int32_t` instead of `DEFINE_int32`.",
+      "reference": "Please revise the naming of these two flags. _**default**_ is the value we take when user doesn't provide. You could infer from the fact how `default_parts_num` are used."
+    },
+    {
+      "prediction": "I think this should be `@Override`",
+      "reference": "I don't think we should have separate \"transformer\" and \"generator\" terms. I would suggest renaming this existing class `ImportTypeTransformer` to `StandardImportTypeTransformer` and renaming the interface `ImportGenerator` to `ImportTypeTransformer`."
+    },
+    {
+      "prediction": "I think we can remove this file as it's no longer used.",
+      "reference": "I believe `tag-matchers` is the proper form for this filename. Only components should use pascal case, I believe everything else should continue to use kabob-case."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/apis/duck/v1alpha1`",
+      "reference": "same here, move this one down"
+    },
+    {
+      "prediction": "Shouldn't this be `createdBy` instead of `createdAsAdministrator`?",
+      "reference": "Some could be created as superadmin instead, to add more special cases. Also, the reason behind this change should be described in the commit msg and the PR description."
+    },
+    {
+      "prediction": "I don't think this should be in this PR.",
+      "reference": "Is this required?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to set this flag. I think it should be set in the `SecurityConfig` constructor.",
+      "reference": "`--public-volume-create-allowed=true` and set the default to `true`, then use `!c.Bool('...')`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `h2o_http2_conn_close()` instead.",
+      "reference": "Could we use `0x7ffffffe` instead, since `INT32_MAX` is an ID of a _pull_ stream? Other than that the PR seems ready for merge. Thank you for working on the fix."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Might as well add a docstring for the new `exit_code` parameter too."
+    },
+    {
+      "prediction": "I don't think we need to change this method.",
+      "reference": "`poll` and `remove` have different meanings."
+    },
+    {
+      "prediction": "I think this should be `write.parquet.bloom-filter-max-bytes-per-partition`",
+      "reference": "What's the perf impact of writing bloom filer? Does it make sense to enable it by default if the perf impact is minor? Would be nice to include benchmarks?"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this code sample. It should be in the `aws-iam` package, not the `aws` package.",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it seems like it should be here.",
+      "reference": "Remove `HIP_UNSUPPORTED` - it is an alias"
+    },
+    {
+      "prediction": "I think this should be `fetchRecentlyFinishedFlows(long lifeTimeMs, Predicate<ExecutableFlow> predicate)`",
+      "reference": "i think we should return an empty list instead of null here."
+    },
+    {
+      "prediction": "I think it would be better to use `prefSocksProxyHost`, `prefSocksProxyPort` and `prefSocksProxyUser` instead of `prefHttpProxyHost`, `prefHttpProxyPort` and `prefHttpProxyUser`.",
+      "reference": "Is it really necessary to store them in a new setting? I think it should be enough to store the credentials in `ProxyConfig.direct()`. Then AntennaPod does not need to store multiple different proxy settings. Also, it will not change existing users' settings (currently the proxy settings would be lost on upgrade)."
+    },
+    {
+      "prediction": "I think this should be `nodeGroup.Children.Any() && nodeGroup.Children.All( node => ((TransitionDocNode)node).ExplicitValues.CollisionEnergy == ((TransitionDocNode)nodeGroup.Children.First()).ExplicitValues.CollisionEnergy)` instead of `nodeGroup.Children.Select(node => ((TransitionDocNode)node).ExplicitValues.CollisionEnergy).Distinct().Count() == 1`",
+      "reference": "This is a mess. Can it be broken out as a separate function using ifs and return statements, which would also make it more debuggable."
+    },
+    {
+      "prediction": "I'm not sure this is the right approach. I think we need to make sure that `item` is a `Plot` before calling `map` on it. If it's a `DynamicMap` or `HoloMap` then we need to call `map` on it. If it's a `Plot` then we don't need to call `map` on it. I think we need to make sure that `item` is a `Plot` before calling `map` on it. If it's a `DynamicMap` or `HoloMap` then we need to call `map` on it. If it's a `Plot` then we don't need to call `map` on it. I think we need to make sure that `item` is a `Plot` before calling `map` on it. If it's a `DynamicMap` or `HoloMap` then we need to call `map` on it. If it's a `Plot` then we don't need to call `map` on it. I think we need to make sure that `item` is a `Plot` before calling `map` on it. If it's a `DynamicMap` or `HoloMap` then",
+      "reference": "This stuff is a bit ugly and should probably become a utility. The main problem is that ``.map`` is being abused a bit."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this. I think it would be better to check if `forced_empty_list` is `True` and only return an empty list if that's the case. Otherwise, I think it would be better to return an empty list if `forced_empty_list` is `True` and raise an error if that's not the case.",
+      "reference": "Maybe there is a better way here so that we don't even bother triggering a storage query since we want an empty list here."
+    },
+    {
+      "prediction": "Nit: I think it would be better to call this `simulateCustomPolicyV2` since it is only used for v2.",
+      "reference": "just curious, `region` is not used in the AWS IAM client, afaik. But I see this is a general field for clutch clients, is it just ignored internally?"
+    },
+    {
+      "prediction": "I think this should be `http://www.apache.org/licenses/LICENSE-2.0`",
+      "reference": "Why are we doing that change?"
+    },
+    {
+      "prediction": "I think this should be `!rawErrorData.containsKey(CLASS) && !rawErrorData.containsKey(STACK_TRACE)`",
+      "reference": "Why && changed to || ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. It seems like it should be in `dask.utils`, but I don't know what the right place is.",
+      "reference": "can you please change this to `_collection_to_single_partition()`, like we've done for alll other helper methods defined in this module?"
+    },
+    {
+      "prediction": "This is not needed anymore.",
+      "reference": "This class is never used, don't import it."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I'm not sure we want to allow exceptions to be caught in the test suite. I'm not sure we want to make it impossible for someone to write a test that passes. I'm not sure we want to make it impossible for someone to write a test that fails. I'm not sure we want to make it impossible for someone to write a test that fails because of an unhandled exception. I'm not sure we want to make it impossible for someone to write a test that fails because of an unhandled exception. I'm not sure we want to make it impossible for someone to write a test that fails because of an unhandled exception. I'm not sure we want to make it impossible for someone to write a test that fails because of an unhandled exception. I'm not sure we want to make it impossible for someone to write a test that fails because of an unhandled exception. I'm not sure we want to make it impossible for someone to write a test that fails because of an unhandled exception. I'm not sure we want to make it impossible for someone to write a test that fails because of an unhandled exception. I'm not sure we want to make it impossible for someone to",
+      "reference": "Missing a `return`. Either way, I think it's better to add a `#if DEBUG throw;` in the catch block instead (so we keep the log in debug mode)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Files should have an extra line at the end of them."
+    },
+    {
+      "prediction": "I don't think this is thread safe.",
+      "reference": "I think we could replace this (well the backing field) with `BlockingCollection` when we drop NET35?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it would be better to use `pd.concat` instead of `pser_or_pdf.drop(input_groupnames, axis=1)`.",
+      "reference": "Can you show before/after output in the PR description? From a cursory look, seems like it will just drop the duplicated column which isn't consistent with pandas' results."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should go in `validation.rb` under the `validation` module, since it's only used there.",
+      "reference": "Added validation that `config: transports` isn't used @adreyer @katelopresti Curious what you think about having Bolt link directly to the docs site."
+    },
+    {
+      "prediction": "I think this should be a `typing.Mapping[str, typing.Any]` instead of a `typing.Dict[str, typing.Any]`.",
+      "reference": "This probably isn't needed anymore now?"
+    },
+    {
+      "prediction": "I think this should be `import { cloneElement } from './vnode'`",
+      "reference": "Would it be easier to kick this import to the top of the file? It'll get stripped out anyway."
+    },
+    {
+      "prediction": "I think this should be `verifyPrivacyGroupContainsEnclavePublicKey` instead of `verifyPrivacyGroupContainsEnclavePublicKeyInEnrollment`",
+      "reference": "are you sure that the privacyGroup cannot be null?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think you need to do `referenceMatch = list(reference.GetNumAtoms(onlyExplicit=True))` and then `referenceMatch.append(i)`.",
+      "reference": "Wrapped with list"
+    },
+    {
+      "prediction": "I think this function should be called `Div` instead of `DivFloat64` to be consistent with the rest of the package.",
+      "reference": "Please move this to `pkg/stats/v1alpha1`"
+    },
+    {
+      "prediction": "I think it would be better to use a guard clause here instead of using a regex.",
+      "reference": "a small nit pick ...this can go into `lib/patterns.rb` ?"
+    },
+    {
+      "prediction": "I think this should be `mdc-tab--active` instead of `googlesitekit-cta-link--danger`",
+      "reference": "The above TODO is no longer necessary I believe"
+    },
+    {
+      "prediction": "I think this should be `\"/bin/sh\" -c \"zfs set io.openebs:livenesstimestamp=\\\"$(date +%s)\\\" cstor-$OPENEBS_IO_POOL_NAME\"`",
+      "reference": "Why is this change required ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the markdown.go file.",
+      "reference": "This really is an excerpt. ;-)"
+    },
+    {
+      "prediction": "I'm not sure why this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "@acud The reason why the default is now `true` is because the chunk must be in the neighborhood of the storer peer with this change. The majority of the tests expect a valid receipt so the closest peers in the tests must be in the neighborhood. As of writing this, there is only one test that expects a peer to be out of depth."
+    },
+    {
+      "prediction": "I think this should be `go.temporal.io/server/common/persistence/sql/sqlplugin/sqlite`",
+      "reference": "This is what breaks the build. Apparently w/o this line, `sqlite` package is not even compiled. With this line it gets compiled but compilation fails because we don't use CGO for our production Linux builds and > ***Important: because this is a `CGO` enabled package you are required to set the environment variable `CGO_ENABLED=1` and have a `gcc` compile present within your path.***"
+    },
+    {
+      "prediction": "I think this should be `uint i = 0;`",
+      "reference": "This should be all slots"
+    },
+    {
+      "prediction": "This method is not used anywhere, so I think it can be removed.",
+      "reference": "Need to add deprecated annotation and call the new fixed method to avoid the API broken issue."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Glean` class.",
+      "reference": "This was just fixed in mozilla-mobile/android-components#4892. @pocmo is cutting a new Android Components 19 dot release today, so you'll be able to jump to the new version and restore the build type."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `inject` controller.",
+      "reference": "Leaving this log line because it helped me to diagnose that the reconciler client was using a different cache than the event sources, so not always seeing the injectable that triggered the reconcile."
+    },
+    {
+      "prediction": "I think this is the only place where we need to lock.",
+      "reference": "Do we need also to protect access to `mset.js` and `mset.jsa` or even `mset.store` down below? (not sure if those are immutable or not)."
+    },
+    {
+      "prediction": "I think this should be `newUserInfo = keybase1.UserInfo{}`",
+      "reference": "I'd rather you do this under `if info.filledInMerkle {` since that's the only place it's relevant."
+    },
+    {
+      "prediction": "I think we can remove the `leaderBaseUrl` and `leaderCoreName` variables.",
+      "reference": "*NULL_DEREFERENCE:* object `leaderUrl` last assigned on line 348 could be null and is dereferenced at line 351."
+    },
+    {
+      "prediction": "Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "reference": "Since the IdentifierScheme's don't change without us making additional code changes, perhaps this should be added as a constant as we previously did with Perms?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "nit: We can revert this now."
+    },
+    {
+      "prediction": "I think this import is not needed.",
+      "reference": "Oh, that's the reason why you have two different data structures here. Does the order of the `selectedItems` list matter? If not, I think it would be more clear if both would be a Set."
+    },
+    {
+      "prediction": "This should be `luaUserdataGet`, not `luaContainerGetContentDescription`.",
+      "reference": "Should be named getDescription, since the other description functions are named that."
+    },
+    {
+      "prediction": "I think this should be 2048 instead of 1920",
+      "reference": "Don't we want to do this for g2 4k also?"
+    },
+    {
+      "prediction": "I think we can remove the `isPublic` field from the `OneDriveApi` class, since we don't use it anymore.",
+      "reference": "I would prefer if we don't add this package as it's unsupported, and I expect not necessary. Node buffer supports many formats - is it not possible to use one of its helper functions to load the content?"
+    },
+    {
+      "prediction": "I think we should rename this to `chainToFullRules` to be consistent with `chainToContents` and `chainToDataplaneHashes`.",
+      "reference": "I don't think this is used any more, please remove."
+    },
+    {
+      "prediction": "I don't think this is correct. The `inner_type.name` is the name of the inner type, not the inner type itself.",
+      "reference": "actually use type_name=print_config_type_to_string(self, with_lines=False) to populate this"
+    },
+    {
+      "prediction": "I think this should be `datastore.settings.delete_local_resource` instead of `datastore.settings.get('datastore.settings.delete_local_resource')`",
+      "reference": "We should be using dependency injection here, instead of fetching the config factory at the last minute from the container. That would allow us to more easily overwrite the \"delete_local_resource\" setting in tests."
+    },
+    {
+      "prediction": "I think you can remove the `LoggerFactory` import and use `org.slf4j.Slf4j` directly.",
+      "reference": "everywhere else in the project, `LOGGER` is used"
+    },
+    {
+      "prediction": "I think this should be `API_VERSION: Settings.api_versions.min_by(&:to_f).to_json`",
+      "reference": "Why is this value set to \"min_by\", and what are the implications of it? Does `&:to_f` play nice with semver (e.g. 1.2.0)?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `src/test/javascript/jest/__init__.js`",
+      "reference": "typo --> gloabl"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do this in the same way we do for the other `Replacer` contexts.",
+      "reference": "Wanted to double-check: does the `log` middleware still set its own \"empty\" value (should default to `-` at least for the default log format)?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with right now.",
+      "reference": "It would be better to just change RWD to throw IllegalStateException if you attempt to execute a command after quit (unless it's a second call to quit())"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could come up with that worked.",
+      "reference": "Is this going to be configurable? If not, it should just be removed for now."
+    },
+    {
+      "prediction": "Please use `import json`, `from json import JSONEncoder`, `from collections import OrderedDict`, `import collections`, `import operator`, `import functools`, `from ConfigParser import NoOptionError`, `from json import JSONEncoder`, `from collections import OrderedDict`, `import collections`, `import operator`, `import json`, `from ConfigParser import NoOptionError`, `from json import JSONEncoder`, `from collections import OrderedDict`, `import json`, `from ConfigParser import NoOptionError`, `from json import JSONEncoder`, `from collections import OrderedDict`, `import json`, `from ConfigParser import NoOptionError`, `from json import JSONEncoder`, `from collections import OrderedDict`, `import json`, `from ConfigParser import NoOptionError`, `from json import JSONEncoder`, `from collections import OrderedDict`, `import json`, `from ConfigParser import NoOptionError`, `from json import JSONEncoder`, `from collections import OrderedDict`, `import json`, `from ConfigParser import NoOptionError`, `from json import JSONEncoder`, `from collections import OrderedDict`, `import json`, `from ConfigParser import NoOptionError`, `from json import JSONEncoder`, `from collections import OrderedDict`, `import json`, `from ConfigParser import NoOptionError`,",
+      "reference": "Maybe just one import line - `from collections import OrderedDict, Mapping` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it would be better to move it to `AbstractRule` since this is a rule that can be used in multiple places.",
+      "reference": "Btw this class probably doesn't belong in PMD. It says it's there to test UIs, but arguably UIs can use their own test fixtures, and this in in src/test so probably not even shipped anywhere."
+    },
+    {
+      "prediction": "I think it would be better to move the `finishResolution` call into the `receiveGetContainers` call, so that we don't have to repeat the `finishResolution` call every time we call `receiveGetContainers`.",
+      "reference": "Maybe we can group each `finishResolution` call with the corresponding `receiveGet...` call? That would make the connection more clear when reading the code."
+    },
+    {
+      "prediction": "I think it would be better to use `self._source_address` instead of `source_address`.",
+      "reference": "Can we take this out of the constructor here and just use the config value? (This would also make the other proxy mode cases obsolete)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle required fields, but it's the only way I could come up with at the time.",
+      "reference": "Please flip the conditional to be `if (field.required) { ... }`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test. I think it should go in `spec/models/user_decorator_spec.rb` but I'm not sure.",
+      "reference": "should we create a fixture without first name and last name and have a spec like this one that uses it to make sure nothing errors out when they are not present?"
+    },
+    {
+      "prediction": "I think it would be better to call this method `generateMethodSample` to be consistent with other methods in this interface.",
+      "reference": "It seems odd to require all Python snippets to have this method when it's not relevant for messages.snip or the discovery snippets. (I see that we're already doing something similar with generateModule/generateBody where some of the implementations are empty. This also seems strange to me.)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to store the hash. I think it should be in the private state, but I'm not sure if it's the right place to store it.",
+      "reference": "What about this: \"// Corresponding PMT does not exist.\""
+    },
+    {
+      "prediction": "I think this should be `if not os.path.exists(config_file):`",
+      "reference": "You don't need af ormat specified here."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Nitpick: import ordering"
+    },
+    {
+      "prediction": "This should be a `Func<IBlockDataRecoveryStep, Func<ITransactionProcessor, IRewardCalculator>>`",
+      "reference": "code smell here, a function that create a reward calculator from transaction processor?"
+    },
+    {
+      "prediction": "I think we should rename this to `RuntimeConfigurator` since it's not just a runtime variable manager.",
+      "reference": "I don't think the `A` is idomatic is it?"
+    },
+    {
+      "prediction": "I think this should be `RemoveReachableFiles` and not `ExpireSnapshots`.",
+      "reference": "nit: the others use the method name in the api and not the class name of the api"
+    },
+    {
+      "prediction": "I think this should be done in the `runSILMandatoryOptPreDiffPasses` function.",
+      "reference": "Does it work to run all the sil diagnostic passes before we serialize? That would be more consistent with what the normal compiler does in `FrontendTool.cpp : performCompileStepsPostSILGen()`"
+    },
+    {
+      "prediction": "`section` is a `belongs_to` relationship with `section_1` and `section_2`, so you should be able to access it with `section.section_1` or `section.section_2` instead of `section.section_1` and `section.section_2`",
+      "reference": "You left in a puts."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it should be in `FilterCollector.java` since it's only used by the `FilterCollector` and not the `CachingCollector`.",
+      "reference": "Hmm, should we also cache the `smoothingScore` for this hit? Or, if we will keep it at returning `0`, couldn't we remove this impl and inherit the default from `Scorable`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right change. I think we need to keep `copy:less_dist` and `cut-release` as part of `baseTasks`",
+      "reference": "I removed that from here because I am hoping that will stop the random unwanted updates to the website with unstable versions."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it's better to put it in the `Process` struct.",
+      "reference": "Nit: missing period."
+    },
+    {
+      "prediction": "nit: remove the period at the end of the sentence.",
+      "reference": "how **to** deny Simple Storage **Service**"
+    },
+    {
+      "prediction": "Why did you change this timeout?",
+      "reference": "Were you having trouble with timeouts in general acceptance tests or only the editor test? It's possible to set timeouts on a per-test basis by using `this.timeout(25000)` within the `it()` function. I'd like to drop the global timeout in the future if possible rather than increase it - in some circumstances a failing test can have a knock-on effect which makes all later tests timeout so Travis can get hung up for a long time if no one notices and cancels the build."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You should be able to use `os.path.relpath` on `fromPath` and `toPath` to get the relative path.",
+      "reference": "I don't think this is necessary."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "Instead of blindly ignoring all errors, this should just ignore the volume not found error."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's fine, but I'm not sure if it's the right thing to do.",
+      "reference": "The `rowPosition` will be ignored if the position column is not projected."
+    },
+    {
+      "prediction": "I think this should be `exec`: `zfs set io.openebs:livenesstimestap=$(date) cstor-$OPENEBS_IO_CSTOR_ID`",
+      "reference": "Is there a possibility of a clash between periodSeconds and timeoutSeconds? For instance, the current probe is not yet timed-out and the next one has started."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure the constructor is not `final` in the first place. If it is `final`, then we can't use `cls.getDeclaredConstructor()` because it will throw an `IllegalStateException`. If it is not `final`, then we can't use `CommandLine.defaultFactory().create(cls)` because it will throw a `NoSuchMethodError`.",
+      "reference": "This logic is already been performed in `CommandLine.defaultFactory().create(cls)` ... whats the point of repeating it here?"
+    },
+    {
+      "prediction": "I think this should be `this.hot.countSourceCols() || this.hot.countCols()`",
+      "reference": "Please change from `let` to `const` here and above. Setting value to `this.hot.countSourceCols()` should be enought probably."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it. If the connection is closed by the client, we should not wait for I/O to finish. I think we need to check if the connection is closed by the client, and if it is, we should not wait for I/O to finish. If the connection is closed by the server, we should not wait for I/O to finish. I think we need to check if the connection is closed by the server, and if it is, we should not wait for I/O to finish. I'm not sure if this is the right way to do it. If the connection is closed by the client, we should not wait for I/O to finish. I think we need to check if the connection is closed by the client, and if it is, we should not wait for I/O to finish. I think we need to check if the connection is closed by the server, and if it is, we should not wait for I/O to finish. I'm not sure if this is the right way to do it. If the connection is closed by the server, we should not wait for I/O to finish. I think we need to check if the connection is closed by the server, and if it",
+      "reference": "Somewhat nitpick: I see the `!=` pattern mentioned in the fluent-bit style guide, but imo, it would be safer to check that a fd is non-negative with ` > -1` or `>= 0`"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Use the new Ruby 1.9 hash syntax."
+    },
+    {
+      "prediction": "Can you remove this empty line?",
+      "reference": "Please host to a GitHub repo."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the `doneSplit` method of the `server` struct.",
+      "reference": "The Error log message should not expose internals. The message should be something like this `\"done split failed for address %v\", tagr.Address`. Also, the Debugf would be more informative with the address in the message."
+    },
+    {
+      "prediction": "I think this should be `loglevel` instead of `loglevel-on-instance`",
+      "reference": "can we rename `fileLogLevelUsage` to be more generic like `instanceLogLevelUsage`? Same goes for other var below like fileLogLevelFlagName, FileLogLevel."
+    },
+    {
+      "prediction": "I think we can remove the `fixed` class from `itemsContainer` as well.",
+      "reference": "Does this one not need the `tabContent` object used in the other files?"
+    },
+    {
+      "prediction": "I think we should rename this to `IsManagedCluster`.",
+      "reference": "Our use of the terms `managed` and `unmanaged` in CAPA is interesting. I think we should probably update the docs (as part of a separate PR) to explain that we are referring to whether its CAPA managed infra. As opposed to meaning AWS managed service."
+    },
+    {
+      "prediction": "I think this is a good change, but I don't think it's related to this PR.",
+      "reference": "Looks like all of these can be put into a list."
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this. Can you change it to `canonicalizedKeyMapping`?",
+      "reference": "nit: \"*mapping\" for a map is unnecessary. Consider calling this `originalNames` or similar."
+    },
+    {
+      "prediction": "I'm not sure this is the right translation. I think it should be something like \"Are you sure you want to delete this profile? This cannot be undone.\"",
+      "reference": "Please split this into two sentences; i.e. \"This profile will be permanently deleted. This action cannot be undone.\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `IobCobCalculatorPlugin` class.",
+      "reference": "NPE here and in other APS plugins"
+    },
+    {
+      "prediction": "I think this should be `return strings.Join([]string{msg.Router.Source, resourceType, resourceNamespace, resourceName}, \"/\")`",
+      "reference": "Thanks for the fixing, and could you please provide more details for this bug? Because we have the deduplication mechanism in cloudhub, so it will has no problem."
+    },
+    {
+      "prediction": "I think it would be better to have these in a `Properties` class. It would make it easier to add new properties in the future. Also, I think it would be better to have these in a `Settings` class. It would make it easier to add new properties in the future. Also, I think it would be better to have these in a `Settings` class. It would make it easier to add new properties in the future.",
+      "reference": "These could be final"
+    },
+    {
+      "prediction": "I don't think we need to add the context to the `Invocation` object. We can just pass it to the `invoke` method.",
+      "reference": "highway have the same problem we can add a new method in invocation: mergeContext 1.if new context have more items, then addAll to new context, and replace old context 2.if new context have less items, then allAll to old context directly."
+    },
+    {
+      "prediction": "I think this should be `Trace.TraceError(sb.ToString(), docId.Substring(\"T:\".Length), type.GetAssembly().Name.Value);` and not `Trace.TraceError(sb, docId.Substring(\"T:\".Length), type.GetAssembly().Name.Value);`",
+      "reference": "Created a single error so it's not interleaved in log"
+    },
+    {
+      "prediction": "Why did you remove this check?",
+      "reference": "`isIE9` was reverted. Shouldn't this condition be reverted as well?"
+    },
+    {
+      "prediction": "`defaultReturn` is not used in this function, so it should be removed.",
+      "reference": "Sorry if I'm being too pedantic here, but wouldn't it be nice to have standard parameter name? Like using `defaultValue` everywhere."
+    },
+    {
+      "prediction": "I think this should be `bserver-0.dev.keybase.io:443,bserver-0.kbfs.keybase.io:443,bserver-1.dev.keybase.io:443,bserver-1.kbfs.keybase.io:443,bserver-1.kbfs.keybase.io:443,bserver-2.dev.keybase.io:443,bserver-2.kbfs.keybase.io:443,bserver-2.kbfs.keybase.io:443,bserver-3.dev.keybase.io:443,bserver-3.kbfs.keybase.io:443,bserver-3.kbfs.keybase.io:443,bserver-4.dev.keybase.io:443,bserver-4.kbfs.keybase.io:443,bserver-4.kbfs.keybase.io:443,bserver-5.dev.keybase.io:443,bserver-5.kbfs.keybase.io:443,bserver-5.kbfs.keybase.io:443,bserver-6.dev.keybase.io:443,bserver-6.kbfs.keybase.io:4",
+      "reference": "Remind me again: what's the point of having new clients connect to both -0 and -1? If we ever have to blackhole -0, we'd have to blackhole -1 at the same time, right? What is supposed to be the difference between the two? Is it just that someday we might want to have two ELBs, and this will help load balance between them?"
+    },
+    {
+      "prediction": "Why do we need to reset the ping timer here?",
+      "reference": "Why not go back to `c.setPingTimer()` here instead so you don't need the new boolean in setFirstPingTimer()."
+    },
+    {
+      "prediction": "I don't think we need the trailing `/*` here.",
+      "reference": "We should figure out why so many of these files have an invisible change on the first line: Are we writing out some sort of byte order mark?"
+    },
+    {
+      "prediction": "This file should be removed.",
+      "reference": "Was this our code @csrwng"
+    },
+    {
+      "prediction": "I think this should be `uname -n` instead of `uname -r`.",
+      "reference": "I think you need `.untaint` here."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to check for the existence of the `tasks` or `files` directories. I'm also not sure if it's the best way to check for the existence of the `bolt-project.yaml` file.",
+      "reference": "Why are we including the `files/` directory in this check? I know `tasks/` and `plans/` make sense, since you can reference content in those directories from the CLI, but I don't think you can for `files/` (unless I'm missing a command)."
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "And again. The reason it's bad is that if someone uses a JDK not produced by Oracle they won't have this class."
+    },
+    {
+      "prediction": "I think we can remove the `X509Name` field from the `CertificateSpec` struct and just use the `Name` field in the `X509Name` field.",
+      "reference": "Can we rename this field to `Subject`? Looking around, it seems like 'subject' is the standard terminology for this stanza"
+    },
+    {
+      "prediction": "I'm not sure what this means, but I think it's a bit confusing. Can you clarify?",
+      "reference": "Remove additional whitespace."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `authn` middleware, not the `customResponseForwarder` middleware. The `customResponseForwarder` middleware is supposed to be able to forward requests to the `authn` middleware, and then the `authn` middleware should forward those requests to the `customResponseForwarder` middleware.",
+      "reference": "We should add `Secure` as well"
+    },
+    {
+      "prediction": "I think this should be `TestWebdavNexcloud` instead of `TestWebdavNextcloud`",
+      "reference": "I presume this isn't a big deal"
+    },
+    {
+      "prediction": "I don't think we should be changing the behaviour of `app_memory_pre_alloc` in this way. It seems to me that we should always call `app_memory_pre_alloc` with `true` as the `update` parameter, and always call `os_raw_mem_alloc` with `false` as the `update` parameter.",
+      "reference": "Maybe {}, even though no multi-line body?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "We can try something like this: response.should render_template(:partial => 'partial_name')"
+    },
+    {
+      "prediction": "I think this should be `if text and caption and hasattr(caption, 'decode')`",
+      "reference": "Please use `isinstance(text, bytes)` instead. Otherwise, this will lead to unnecessary decoding on python 2 unicode strings."
+    },
+    {
+      "prediction": "I think this should be `exe_ctx` instead of `frame`.",
+      "reference": "This one is unrelated to the GetSymbolVendor removal ... the API to GetValueAsData was changed."
+    },
+    {
+      "prediction": "I think this should be `networkMode: bridgeNetworkMode`",
+      "reference": "can you assert the value of this field below similar to other fields?"
+    },
+    {
+      "prediction": "I think it would be better to have a separate class for `ClassLiteral` and `ASTClassLiteral`",
+      "reference": "I'll change that to \"linkplain\" for consistency :)"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `object_id` is a `str` and `object_uri` is a `str`. The `replace` function doesn't accept `str` and `object_id` will be converted to a `str` before it gets passed to the `replace` function, so the `replace` function will try to match a `str` against a `str` and not a `str` against a `str`. I think we need to convert `object_id` to a `str` before passing it to the `replace` function.",
+      "reference": "This is fix. But since the history resource was relying on the bug to work, I had to do some changes regarding the history entries (eg. explicit declare that the permissions inherit from bucket)"
+    },
+    {
+      "prediction": "I think this should be done in the `transformer.finish()` method.",
+      "reference": "can't we leverage Closeable here and get some sugar&warns? Also, line 89 still calls setContext() .. is it right? or I'm missing something?"
+    },
+    {
+      "prediction": "I think this should be `[]byte(peerID)`, not `[]byte([]byte(peerID))`",
+      "reference": "Use explicit swarm.Address when defining peerID, by using NewAddress, and remove conversion here."
+    },
+    {
+      "prediction": "I think this should be `/var/lib/containers/rpm` or `/var/lib/rpm` depending on whether you're talking about the RPM database or the RPM package.",
+      "reference": "`RpmDependenciesDirectory` should start with a lowercase character so it's not exported outside of this package, it looks like its only referenced in this file."
+    },
+    {
+      "prediction": "why do we need to add these three parameters?",
+      "reference": "If we make a new backbone class, we don't need to support `BasicBlock`"
+    },
+    {
+      "prediction": "I think this should be `if configParams.BPFEnabled && (bpf.SyscallSupport() || bpf.IsBPFDisabled()) {`",
+      "reference": "This was an out-of-date dupe of the check on line 335"
+    },
+    {
+      "prediction": "I think we can remove `bPaginate` from `dataTable.defaults` as we don't need it anymore.",
+      "reference": "Let's put this back"
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "Things like these should have been addressed before submitting a pull request."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this check. I think it would be better to put it in `org.graylog2.plugin.metrics.zipkin` since this is the place where the zipkin instrumentation is added.",
+      "reference": "This one as well. Any better suggestion please?"
+    },
+    {
+      "prediction": "I think we can remove this traceInOp now, since it's not used anymore.",
+      "reference": "wonder if we should collect c.trace, c.pa.subject, etc.. while under the lock to prevent data races.. or simply move the tracing under the lock."
+    },
+    {
+      "prediction": "I think this should be `!= strcmp(P->short_name, \"s2\")` instead of `0 !=`",
+      "reference": "Why is this hack needed ? Ideally, we shouldn't need that."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be in `plugin.go` since it's only used there.",
+      "reference": "Minor/Nitpick: replace `1` with a constant, esp. as used in `prog/weaver/http.go` in `{{if eq .Plugin.Version 1}}`"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "nit: Looks like the whitespace got thrown off, can you fix this?"
+    },
+    {
+      "prediction": "This should be a localized string, not a non-localized string.",
+      "reference": "Redo this by adding a string literal and then pressing F6 to have Resharper move it to Properties\\Resources.resx which will also create this property."
+    },
+    {
+      "prediction": "I think this should be `// +build windows` instead of `// +build !windows`",
+      "reference": "Unless a log file has been specified, IMO you should set the server option to enable syslog (windows event log) here, or someplace along the service start code path. We shouldn't really rely on users to specify that when creating the service."
+    },
+    {
+      "prediction": "Why do we need this change?",
+      "reference": "It's strange here, because I saw the `TableColumn` is marked as `PublicEvolving`, but after released flink 1.12.0 it did not have any Interface compatibility guarantee. At least, it should marked as `deprecated`, and keep it a major release."
+    },
+    {
+      "prediction": "I think this should be `CoreMessages.dialog_connection_edit_wizard_conn_conf_network`",
+      "reference": "I'm not sure regarding this approach, for me the whole text including anchors should go to resources like ` netConfigLink.setText(CoreMessages.dialog_connection_edit_wizard_conn_conf_network_link); ` You shouldn't concatenate translated values inside the code. If you need some params, please use NLS.bind()"
+    },
+    {
+      "prediction": "I think this should be `c.logger.Debugf(\"increasing receive flow control window for the connection to %dkB, in response to stream flow control window increase\", c.receiveWindowSize/(1<<10))`",
+      "reference": "How often do we expect this to trigger? Should we maybe put it behind an if logger.Debug()?"
+    },
+    {
+      "prediction": "I think we should be able to remove the `controller.address` part of the URL now that we have the `controller.Address` variable.",
+      "reference": "How was the name `controller` arrived? Can you get a consensus for the name controller ? This may be OK for jiva. However, c-stor does not have a concept called `controller`."
+    },
+    {
+      "prediction": "I think this should be `continue` instead of `switch`",
+      "reference": "i dont think we need to explicitly single out `kernel.domainname` here since its not a TODO for OCI either right ?"
+    },
+    {
+      "prediction": "I think it would be better to use `<=` instead of `<= 100`",
+      "reference": "Should we better change the operator to `<` since the variable defines the minimum size that gets compressed?"
+    },
+    {
+      "prediction": "I think this test should be renamed to `shouldAppendOneMillionTimes` or something like that.",
+      "reference": "I'm really impressed by the AppendElements addition! Very cool! :-)"
+    },
+    {
+      "prediction": "`log.info(version());` should be `log.finest(version());`",
+      "reference": "This change means that users can't easily see which version of the selenium server they're using. This is `info` level information."
+    },
+    {
+      "prediction": "I think this should be `MAX_WORLD_ID = 255`",
+      "reference": "This should be in qtutils and imported to here and the other place instead of declaring it twice."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `FetchCertAlternatives` since that is the place where we fetch the alternatives.",
+      "reference": "Seems a bit weird to add a timeout in \"logging\" middleware, but I see that that is done elsewhere, so fine."
+    },
+    {
+      "prediction": "This should be `admin.customForms.messages.deleteForm` instead of `messages.deleteForm`",
+      "reference": "This code is passing a non localized string when it should be passing in a localized string or it should use the name of the item being deleted."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "Could we combine this with the `setState` call above, so we don't have to call it twice?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `String#valid_encoding?` and `String#force_encoding?`",
+      "reference": "That part that still leaves me mystified is how the String class, when populated with a value from the SQL_ASCII encoded database is set to \"UTF-8\" encoding. Since the database is SQL_ASCII, each byte in the string stored in the database is considered one character. Ruby, however, is using UTF-8, which is writing a series of one-byte code points that are only meaningful when considered in the correct encoding. When Ruby passes a string to Postgresql, the code points are essentially \"flattened\" into bytes as they are written to the DB. When Rails (through Ruby) asks for a string from the DB, where does the information about the encoding come from?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it works for now.",
+      "reference": "Is it a doable way and does it make sense."
+    },
+    {
+      "prediction": "I think it would be better to use `LOG.debug` instead of `LOG.warning` here.",
+      "reference": "max was wrong here. As its buitlt-in, not a number, and a TypeError was thrown."
+    },
+    {
+      "prediction": "I don't think you need to escape the `\\n` here.",
+      "reference": "This change is not needed, please revert it."
+    },
+    {
+      "prediction": "why do we need to set it to \"None\" ?",
+      "reference": "@ClimbsRocks Please add that it could be a string. `feval : callable, string or None, optional (default=None)`"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "now that I have started using, I realized, I could have named my new class singular `Sport` and not having to make this change. let me know, I will update"
+    },
+    {
+      "prediction": "I think this should be `ddev` and not `com.ddev.platform`",
+      "reference": "Not sure why this is changing to a hard-coded string."
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this. I think it would be better to put it in `FlowContentView` since it's only used by that class.",
+      "reference": "Is there a reason why we `str` the result? It would be nice to eventually support more datatypes here, so we want to generally aim for arbitrary JSON."
+    },
+    {
+      "prediction": "I don't think we need to change this import.",
+      "reference": "Why did we update the to this api?"
+    },
+    {
+      "prediction": "I think this should be `throw new Exception(\"Unable to convert preference key [\" + key + \"] value [\" + stringPref + \"] to enum of type \" + defaultEnum.getDeclaringClass() + \", cause: \" + ex.getCause());`",
+      "reference": "should probably just catch `IllegalArgumentException` here"
+    },
+    {
+      "prediction": "I think this should be `const docURL = \"https://v2.docs.influxdata.com/v2.0/reference/flux/stdlib/built-in/transformations/window#calendar-months-and-years\"`",
+      "reference": "What does this error message mean?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `value.encode('utf-8')` will fail if `value` is already UTF-8 encoded. I think we need to do something like this: ```python if isinstance(value, six.text_type): value = value.encode('utf-8') else: value = six.text_type(value) if isinstance(value, six.text_type): value = value.encode('utf-8') ```",
+      "reference": "I don't think this change is necessary. If we cast `value` to type `six.text_type`, then the next if statement should evaluate to True and do the encoding for us."
+    },
+    {
+      "prediction": "Do we need to check that `trait` is not `nil` before calling `handle_trait_values_for` on it?",
+      "reference": "Can we move this into a method(s)?"
+    },
+    {
+      "prediction": "I think this should be `clen([]byte{0}) == 0`",
+      "reference": "You could perhaps call this `strlen` (although I don't particularly care about the name as it is an implementation detail)."
+    },
+    {
+      "prediction": "Why is this method not inlined?",
+      "reference": "should we have a common `Escape()` method that can be used everywhere?"
+    },
+    {
+      "prediction": "I don't think this change is needed.",
+      "reference": "The whitespace removal above here is probably unintentional :slightly_smiling_face:"
+    },
+    {
+      "prediction": "I think it would be better to use `defer wg.Wait()` instead of `defer wg.Done()` here.",
+      "reference": "should we start p2p even after dispatcher? b/c dispatcher handles msg from p2p"
+    },
+    {
+      "prediction": "I think this test can be deleted now.",
+      "reference": "Did we decide once and for all to use req/resp throughout spring?"
+    },
+    {
+      "prediction": "I think we should use `render_to_string` here instead of `render`",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I think this should be `daisy.Errf(\"Failed to delete resources in %q. Check with the owner of %q for more information\", *scratchBucketGcsPath, scratchBucketName)`",
+      "reference": "This will delete the whole bucket, which could have unforseen consequences in normal use cases. We should be deleting args.SourceFile instead."
+    },
+    {
+      "prediction": "I think it would be better to call it `CONTINUE_ON_STOPPED` instead of `RESTART_ON_EXECULATION_STOPPED`",
+      "reference": "Lets move the whole string to next line for better readability"
+    },
+    {
+      "prediction": "I think this should be `Could not save WebCal-calendar` instead of `Error saving WebCal-calendar`",
+      "reference": "Please change this back to `Error saving WebCal-calendar` (and `Error saving WebCal-calendar` only)"
+    },
+    {
+      "prediction": "I think this should be `goog:driver`",
+      "reference": "The vendor prefix is still being used on Chromium based browsers like Edge Chromium and Chrome. Did you mean to remove this?"
+    },
+    {
+      "prediction": "I think this test should be renamed to `NLogLayoutTests` to match the other tests in this file.",
+      "reference": "Not ideal, but I used a test case orderer so I could avoid a bug that occurs when running two tracer's sequentially with different DD_LOGS_INJECTION settings."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `EntrySize` and `len(p.blobs)-1`.",
+      "reference": "I think the condition should be ` uint(headerLengthSize) + crypto.Extension + (len(p.blobs)+1)*uint(EntrySize) > uint(maxHeaderSize)` `headerLengthSize` is only the length field. But the pack header also includes the blob type and its ID. And the result of the check, currently seems to be inverted."
+    },
+    {
+      "prediction": "I think this should be renamed to V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2. Also, I don't think it makes sense to have a type that is used for both V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2. I don't think it makes sense to have a type that is used for both V1 and V2",
+      "reference": "Remove selector if only 1 version is supported"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it works for now.",
+      "reference": "Why did you remove more than the asked command?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get to work.",
+      "reference": "I just realized that we are logging time now - we shouldn't do that here, that's the logger's job. I've opened a new issue for that #444."
+    },
+    {
+      "prediction": "I think it would be better to add `sanitize` as a parameter to `AddMol` and `GetMol`.",
+      "reference": "why is the `sanitize=True` here in the docs?"
+    },
+    {
+      "prediction": "I think this belongs_to should be `user_id` instead of `user`",
+      "reference": "Didn't end up using this, but I think it's useful anyway."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to use the `GetMetadata` method to get the value of the metadata and then use that value in the `call :copyandcheck` command.",
+      "reference": "I'm pretty sure there's already a metadata item that contains what you're calculating here (PackageRelativePath)?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "We want to show this button only when there is a page for us to connect to. There is no reason to put the Browse Security Info button on the page at all unless we've identified the UUID from the KB."
+    },
+    {
+      "prediction": "This file should be renamed to `androidicon.xpm`.",
+      "reference": "Why `2` in this and others?"
+    },
+    {
+      "prediction": "I think it would be better to use `string.Format` here instead of concatenation.",
+      "reference": "This may cause a new allocation, please consider merging the concat in above line. Same applies to change in dotnethostmanager."
+    },
+    {
+      "prediction": "I think this should be `importedname = '.'.join(node.names[0][0].split('.')[1:])`",
+      "reference": "Alternatively (and better) would be to look for the .level attribute of the node. If it's bigger or equal to 1, than that is a relative import. So \"from . import x\" should have level 1, while \"from .. import z\" should have level 2 and so on. The same should happen for \"from .c import z\"."
+    },
+    {
+      "prediction": "I don't think we need to override this in this PR.",
+      "reference": "Btw. this API must definitely change with PMD 7 - we are exposing here a implementation detail (that we use Saxon). And it happens, that the way, how custom functions are registered, changed with Saxon 9.5... which makes the need for a implementation agnostic API relevant..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I would expect it to be in the `Deployment` struct, not in the `GenericDeploymentSpec` struct.",
+      "reference": "nit: `Additional configuration used while sending notification to external services.`"
+    },
+    {
+      "prediction": "I think this should be `setParameter('nameOrHandle', nameOrHandle)`",
+      "reference": "This should only be sent if the driver is speaking to a W3C conformant remote, so we need an if-condition check like we have in the Python bindings."
+    },
+    {
+      "prediction": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "This should have failed before."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `channel` package.",
+      "reference": "the paymentchannel dir is where the paymentchannel manager will live. it will store the types below."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this command. I think it should go in `selenium-webdriver/src/main/java/org/mozilla/selenium/webdriver/commands/`",
+      "reference": "This is Safari specific and should be defined somewhere in the `safaridriver` namespace"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be.",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "Should this be `fontWeight` instead of `fontWeight`?",
+      "reference": "prefer changing the signature by adding params to the end, not the middle, I think."
+    },
+    {
+      "prediction": "I think this should be 10902",
+      "reference": "I guess we aim for hardcoded port for Prometheus operated ports right?"
+    },
+    {
+      "prediction": "I think this should be `node.parent.statement(future=True).fromlineno + 1`",
+      "reference": "This is the same effect as doing `node.parent.fromlineno` but avoids the `StatementMissing` exception from calling `statement`."
+    },
+    {
+      "prediction": "I think this should be `mb_id` and not `mb_id`",
+      "reference": "I think user-ids might be better; also note that options should use - and not _ to separate words. Also, how are more than one id specified? comma seperated? I think the usage statement should indicate this."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I'm not sure we can rely on `timerSleep` returning true or false. I think we should just use `timerSleep` to see if we need to sleep. If we don't need to sleep, then we can just return immediately.",
+      "reference": "This code is necessary. It makes sure that you can sleep for longer durations than the timer supports."
+    },
+    {
+      "prediction": "I think this should be `ExpireSnapshots#executeDeleteWith(ExecutorService executorService)`",
+      "reference": "Can we add the new name in parallel and deprecate this name? We should be careful about changes to the public API and give people at least a release to switch over before removing public methods."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it's better to check if `backend->impl` is not `NULL` before calling `backend->impl->destroy`.",
+      "reference": "If you're going to start testing that impl itself is not NULL you should probably be doing it everywhere."
+    },
+    {
+      "prediction": "I'm not sure if this is the right approach. I think we should be able to use the `config` package to load the yaml config file and then use that to determine which files are manifest files.",
+      "reference": "`ctx` is unused in LoadPlainYAMLManifests"
+    },
+    {
+      "prediction": "nit: can be simplified to `toBuilder = create().fromBuilder(fromBuilder);`",
+      "reference": "why is this called toBuilder when the method is called fromBuilder? (I realise you did not change this...)"
+    },
+    {
+      "prediction": "I think it's better to make this configurable via `tableConfig` instead of `flinkConfigOptions`.",
+      "reference": "nit: seems rather large."
+    },
+    {
+      "prediction": "I think this should be `to_ts = int(time.time())` instead of `to_ts=int(time.time())`",
+      "reference": "I think it'd make sense to only calculate time.time() once (it's also used if max_ts and min_ts aren't set)"
+    },
+    {
+      "prediction": "I think we should rename this to `grpcStubClassName` to be consistent with the naming used in `GrpcStreamingType` and `MockGrpcStubType`.",
+      "reference": "Use existing `grpcStubCallString` instead"
+    },
+    {
+      "prediction": "Is this used anywhere? I don't see it referenced anywhere in the PR.",
+      "reference": "In a recent PR, Bob added a `TracerConstants` class. Do you think this makes sense to put in that class instead so that we can consolidate?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `catchpoint` package, since that's where it's used.",
+      "reference": "I want to propose couple of different names for catchpointWriting: syncInterruptedWaitingCatchpointWriting syncWaitingForCatchpointWriting catchpointWriting is lacking context, and I found it difficult to understand the logic without this context."
+    },
+    {
+      "prediction": "I think this should be `Model` and not `ServiceConfig`",
+      "reference": "Do you still need the `name` in the arg description, if this is not `str` type anymore.?"
+    },
+    {
+      "prediction": "Why do you remove this?",
+      "reference": "We may raise a warning."
+    },
+    {
+      "prediction": "This should be `icd10CMCode` and `icd10Code` should be `icd10CMCode`. Also, `icd10CMCode` should be `ICD10CMCode` and `icd10Code` should be `ICD10CMCode`.",
+      "reference": "Trailing comma caused the eslint test to fail"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way I could get it to work.",
+      "reference": "I think you should use `app.config`. In the usual case, the config file is not overwritten, there is additional config file outside of the source of `Invenio`."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in the resolver.",
+      "reference": "Will you please update the copyright message to now include `2021`?"
+    },
+    {
+      "prediction": "I think this should be `data.begin() + offsets[index], data.begin() + offsets[index + 1];`",
+      "reference": "Potential subscript out of range."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "please undo this change. you didn't really meant to make it, right ?"
+    },
+    {
+      "prediction": "I think it would be better to move this to `src/test/javascript/storybook/test-utils.js`",
+      "reference": "To check! are these millseconds or seconds :thinking: The docs aren't clear"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `if num > int64(u.ctx.MaxUploadParts) || num > int64(MaxUploadParts)` block.",
+      "reference": "Why was this moved?"
+    },
+    {
+      "prediction": "I think we can remove this interface and just use `CStorPoolExpansion` and `CStorVolumeReplicaExpansion`",
+      "reference": "What are these object with suffix Expansion?"
+    },
+    {
+      "prediction": "I think this should be `--rpc-allow-unprotected-txs` instead of `--rpc-allow-unprotected-txs`",
+      "reference": "This breaks backwards compatibility, so it will have to wait for the next quarterly release of the default is to deny. I would recommend adding the flag with the default to allow and then at the next quarterly release rc cycle flip the flag to deny."
+    },
+    {
+      "prediction": "This is a reduction operator which now takes as first argument the TeamPolicy member_type. Every member of the team contributes to the sum.",
+      "reference": "I'm kind of surprised this doesn't require `typename`?"
+    },
+    {
+      "prediction": "I think this should be `test \"$YACC\" != \"bison -y\"` and not `test \"$YACC\" = bison -y`.",
+      "reference": "I think this will break the case (you can try even with /usr/bin/bison as value): `YACC=\"/path/to/bison -y\" ./configure` I wonder if we should use AC_PROG_YACC at all or simply use AC_CHECK_PROGS since we want to use bison and not another yacc"
+    },
+    {
+      "prediction": "I think this should be `Path` instead of `Path to`",
+      "reference": "can we explain to users in the docs what happens if they leave the field blank?"
+    },
+    {
+      "prediction": "I think this should be `kernel_id` and `ramdisk_id` instead of `kernel_id` and `ramdisk_id`",
+      "reference": "We try not to change the function footprints unless absolutely necessary. Why is it necessary to have both secuirty_group_ids and security_groups?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we should be replacing `File.separator` with `File.pathSeparator` instead.",
+      "reference": "Thanks! We're already replacing File.separatorChar with '/' in the call to FileUtil#toCanonicalPath above, so the correct fix here should be: `relativePath.replace('/', '.')` I'll make the change upstream."
+    },
+    {
+      "prediction": "Why do you need to change this import?",
+      "reference": "It could be nice to be explicit about this change in upgrade notes, what do you think?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in the `companion` controller, since it's the controller for this specific provider.",
+      "reference": "spotted a bug."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it would be better to move it to `org.mule.runtime.core.api.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest` or `org.mule.runtime.core.internal.query.impl.QueryTest`",
+      "reference": "We were always using 0000000 prefix"
+    },
+    {
+      "prediction": "`gJobMgr->shutDown();` should be guarded by `if (gJobMgr)`.",
+      "reference": "Here we also need to determine if gJobMgr has called the init function."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that works.",
+      "reference": "Out of curiosity, why is `condition.any?` also necessary here?"
+    },
+    {
+      "prediction": "I think this should be a `new` instead of a `new[]`.",
+      "reference": "Always use `std::make_shared` whenever possible, it will save you one memory allocation."
+    },
+    {
+      "prediction": "I think this should be `router` instead of `router-expose`",
+      "reference": "Update to real tag before pull."
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix. I think we need to make sure that we're not tracking the health of the zipkin collector itself.",
+      "reference": "Is there any better string that only exposed by prometheus? We have changed the meter tags to contain method and service name. For example `scrape()` contains `method=getHealth` and `service=server.internal.health.ITzipkinHealth` which made this test failed."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I'd expect it to be in the core logger, but I don't see where it is used.",
+      "reference": "Adding ICoreLogger here lets us pull this into Core as a strategy"
+    },
+    {
+      "prediction": "I think we can remove the `DockerID` field from the ContainerMetadata struct.",
+      "reference": "why did you change this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should go in `spec/failure_formatter.rb` since that is where the failure colorization is done.",
+      "reference": "This is invalid syntax on 1.8."
+    },
+    {
+      "prediction": "I think you can remove the `[O:2]` from the first line of this test.",
+      "reference": "This isn't part of the PR, but I can't find in the smarts definition that \"[C][C]\" == \"[C]-,:[C]\" There is a line saying essentially not to specify undefined items (but that's a bracket versus non bracket thing)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `UnsignedMessage` struct.",
+      "reference": "Is this required for tuple encoding? This is confusing."
+    },
+    {
+      "prediction": "I think this should be called `flowRunnerManager` instead of `flowPreparer`",
+      "reference": "Does it need to be a member variable? It is currently only used in one method."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `join` module, since it's only used by the `join` module.",
+      "reference": "@MrTimscampi doesn't this need an update to apiclient?.."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `X86_64` and `X64` are not the same. `X86_64` is an alias for `X86_64`, but `X86_64` is not the same as `X64`. So I think we need to use `X86_64` instead of `X64` here.",
+      "reference": "This is a nop: prob best for history to not change the line"
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "Can you delete this whole file now?"
+    },
+    {
+      "prediction": "I think this should be `ArrPassing` not `ArrString`",
+      "reference": "can you combine this passing test, code, solidity binary into existing ArrDelete? the function is much similar, pushing int value into an array, and delete one item in the array"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the keep sorted flag. I think it would be better to set it in the DBWriter.reorderQueue method.",
+      "reference": "I think that could lead to problems when setting it to not sorted again. In that case, it will not be reset. What about using one single method for the adapter (`setDragDropEnabled`)?"
+    },
+    {
+      "prediction": "I think this should be `app.SiteStatus() != platform.AppNotRunning`",
+      "reference": "I'd say the error should be an error, not instructions to the user. So error would be something like \"site should be running and is not\""
+    },
+    {
+      "prediction": "I think we can remove ECSDataVersion 13 and 14 now.",
+      "reference": "Are you sure `v3EndpointIDToContainerName` and `v3EndpointIDToTask` are saved in the state file, can you verify that?"
+    },
+    {
+      "prediction": "Is this mutex needed?",
+      "reference": "This mutex is never used. Should it be, if users can now make calls into the crypto setup?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I've been able to get it to work.",
+      "reference": "Check here for sles-, and i'd prefer a regex since I don't know what an 'include' is going to do exactly. :)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it. I think we should check if `amount` is `None` or `Decimal('0.0')` and then use `Decimal(amount) / self.exchanger.exchange(1, currency)` or `amount / self.exchanger.exchange(1, currency)`. If `amount` is `None` or `Decimal('0.0')` we should use `amount / self.exchanger.exchange(1, currency)` or `amount / currency` respectively.",
+      "reference": "Should this be `None` instead?"
+    },
+    {
+      "prediction": "I think this should be `warn` instead of `info`",
+      "reference": "Not sure if the warning message is enough for this time, if the configuration is wrong, java-chassis may need to stop working instead of sending out the warning message to let the admin know about it. BTW, we may need a default override policy here."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this, but I think it makes sense to move it to the `version` package, since it's not specific to etcd.",
+      "reference": "I see this getting unwieldy as we add more commands with subcommands. Can we perhaps enhance it to handle subcommands properly? Perhaps something like `\"etcd-snapshot:1\"` which would indicate that the etcd-snapshot command may have 1 subcommand after it, and if the 1 next argument after it doesn't start with `--` then the args should be appended after that subcommand."
+    },
+    {
+      "prediction": "This method is called from `getActiveFlows` and `getInactiveFlows`, so I think it should be renamed to `getRunningFlows` as well.",
+      "reference": "why returning a string instead of a list?"
+    },
+    {
+      "prediction": "I think this should be `nameof(IntegrationIds.AdoNetCommandIntegration)`",
+      "reference": "Was there a reason for changing all these integration names?   Are they actually used for anything besides information in logs?"
+    },
+    {
+      "prediction": "I think this should be `Set#copyOf()` instead of `Set#add()`.",
+      "reference": "we should not resurrect this method that we just removed"
+    },
+    {
+      "prediction": "I think this should be `self.get_screenshot_as_png(self, filename)`",
+      "reference": "This needs to be just return self.get_screenshot_as_file(filename)"
+    },
+    {
+      "prediction": "I think this function can be defined in `xhr.js` and imported here.",
+      "reference": "Shouldn't this return false here?"
+    },
+    {
+      "prediction": "I think this should be `omitempty` since we don't want to write to this field if it's empty.",
+      "reference": "Should have json tag"
+    },
+    {
+      "prediction": "(style) Macros with complex values should be enclosed in parentheses (style) Macros with complex values should be enclosed in parentheses (style)",
+      "reference": "Is this needed after all? Looks like the struct was re-defined inside the test file."
+    },
+    {
+      "prediction": "I think this should be `RemoveAllReachableFiles`",
+      "reference": "I think the name of the method should match the name of the action: `removeReachableFiles`."
+    },
+    {
+      "prediction": "I don't think we should have a separate page for managing consent types. I think it should be part of the \"Manage\" menu.",
+      "reference": "can you make the C in Consent lower case to match the other ones in this list?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "fix the group"
+    },
+    {
+      "prediction": "I think this should be `slots.size() > 0` instead of `slots.stream().anyMatch(slot -> slot.isSupporting(caps))`",
+      "reference": "Prefer `Stream.anyMatch` instead of iterating over all slots."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "we have `logging.TestingLog()` that should be used for that purpose."
+    },
+    {
+      "prediction": "(style) code indent should use tabs where possible (style) please, no space before tabs",
+      "reference": "This is also a question of style. src[i] will always be valid because of check if 'for' statement"
+    },
+    {
+      "prediction": "I think this should be `pidof -w 1 %s 2>&1`, where `%s` is the process name.",
+      "reference": "Why change it?"
+    },
+    {
+      "prediction": "nit: remove the trailing semicolon",
+      "reference": "nb: this change is unnecessary"
+    },
+    {
+      "prediction": "I don't think we should have a public link for an object that is not publicly accessible.",
+      "reference": "Is the publicLink mostly empty? I'm just wondering how much this will bloat memory usage as sometimes rclone has millions of Object~s in memory."
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "I don't think the Internal API types need the json annotations. They are never serialised, right?"
+    },
+    {
+      "prediction": "I think this should be `ReportApplicationMostRecentDeploymentCompleted`",
+      "reference": "Maybe this was added incidentally. We don't need this function in the interface."
+    },
+    {
+      "prediction": "I don't think we need to do this. We can just do `t = path.Clean(strings.Split(t, \"?\")[0])`",
+      "reference": "Is this a safe/reliable way to split the URL?"
+    },
+    {
+      "prediction": "Shouldn't this be `System.currentTimeMillis() - 1000` ?",
+      "reference": "do we also need to profile the time to getConnection()?"
+    },
+    {
+      "prediction": "I think this should be `continue` instead of `continue if f[conn]...`",
+      "reference": "Should we move the decode part directly to the actual first-use of this? Or how does this affect if the value gets decoded and we need to get the bytes back later?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Legacy` section.",
+      "reference": "How would you feel about continuing to point this at the MyResearchController's holdsAction, but instead changing that action to force a redirect to the new holds-list route? That way, people will get sent to the new URL instead of having two different URLs that do the same thing."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to have a flag in the router that indicates whether the connection was created or closed.",
+      "reference": "I think if yo use %s then you do not need reason.String() and can just do reason."
+    },
+    {
+      "prediction": "I think we should rename this to `WorkflowTaskTimeoutSeconds` as it is not related to workflow tasks at all.",
+      "reference": "Having \"wf\" prefix doesn't make sense anymore. Please remove."
+    },
+    {
+      "prediction": "nit: `Object.keys` can be replaced with `Object.prototype.toString.call(node).match(/[A-Z][^A-Z0-9]/)`",
+      "reference": "Removed this copy of the `spyAll` function and replaced it with the same function declared in `helpers.js`. Same for `sortAttributes` below"
+    },
+    {
+      "prediction": "I think this should be `DEFAULT_MAX_CPU = \"0\";` and `DEFAULT_MAX_MEMORY = \"0M\";`",
+      "reference": "Default max cpu should be 8 and memory 64GB"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it would be better to put it in `pandas/_utils.py` or `pandas/core/utils.py` or `pandas/core/metadata.py` or `pandas/core/metadata/core.py` or `pandas/core/metadata/__init__.py` or `pandas/core/metadata.py` or `pandas/core/metadata/__init__.py` or `pandas/core/metadata/core.py` or `pandas/core/metadata.py` or `pandas/core/metadata/__init__.py` or `pandas/core/metadata/core.py` or `pandas/core/metadata.py` or `pandas/core/metadata/__init__.py` or `pandas/core/metadata/core.py` or `pandas/core/metadata.py` or `pandas/core/metadata/__init__.py` or `pandas/core/metadata/core.py` or `pandas/core/metadata.py` or `pandas/core/metadata/__init__.py` or `pandas/core/metadata/core.py` or `pandas/core/metadata.py` or `pandas/core/metadata/__init__.py` or `pandas/core/metadata/core.py` or",
+      "reference": "@itholic, can you remove `nbytes = unsupported_property('nbytes')` at `_MissingPandasLikeSeries`?"
+    },
+    {
+      "prediction": "I think this is the only test that needs to be changed. The rest of the tests pass.",
+      "reference": "What happened to the assertion on `conf.TaskCPUMemLimit`?"
+    },
+    {
+      "prediction": "nit: `Object.create(null)` would be more idiomatic.",
+      "reference": "Just curious: Is using `Object` faster than a literal `{}`?"
+    },
+    {
+      "prediction": "Shouldn't this be \"ACME does not support certificate durations\" instead of \"ACME does not support certificate renewals\"?",
+      "reference": "How come we don't allow this to be configured with the ACME issuer? Happy to leave this as-is for now if there's a lot more consideration that needs to be made, but it seems like we could/should be able to allow this?"
+    },
+    {
+      "prediction": "I think this should be `NetworkListener` instead of `NetworkLoadBalancerListener`",
+      "reference": "Should we remove the `Aliases` field as well?"
+    },
+    {
+      "prediction": "Do we need to store transcriptions and samples in the same instance? It seems like it would be better to have a separate instance for each.",
+      "reference": "what's the effect of turning these two off? is there an issue on file for context?"
+    },
+    {
+      "prediction": "I think this should be `GcloudBetaLatestWrapperRelease`",
+      "reference": "The format like gcloud-beta-prod is a bit hard to understand. Maybe find a a more clear way to describe it,"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface. I think it should be in `OpenTelemetry.Context`.",
+      "reference": "Are we renaming this too in a follow up PR?"
+    },
+    {
+      "prediction": "This should be `lang.locale.be`, not `system::lang.locale.be`.",
+      "reference": "The indentation on this is off by one space to the left"
+    },
+    {
+      "prediction": "I think this should be `&& isExternalFormatConversion` instead of `|| isExternalFormatConversion`",
+      "reference": "I guess nothing is wrong with this approach, but more curious if you didn't just go `|| (format != VK_FORMAT_UNDEFINED)) {` As if there ever was another external format system added in Vulkan it would need to be manually added here"
+    },
+    {
+      "prediction": "I think this should be `collect(joining(\" \"))` instead of `collect(Collectors.joining(\" \"))`",
+      "reference": "Is order really irrelevant here?"
+    },
+    {
+      "prediction": "I don't think you need to import `Toast` here. You can just use `snackbar` from the `androidx` package.",
+      "reference": "I think this import will be unused now. If it is unused remove it."
+    },
+    {
+      "prediction": "I think this should be `btree.New(128)`.",
+      "reference": "Let's change to tx"
+    },
+    {
+      "prediction": "I think this should be `defaultDiskBlockCacheFraction = 0.15`",
+      "reference": "Looks like I was wrong about the sync cache size being 15% before. There seems to be some reporting bug related to my use of the sync cache, that makes the limit different from the working set cache limit. For now, I say we keep this at `0.10` (change the param default too), and we can change it later if needed."
+    },
+    {
+      "prediction": "I think this should be `os.Getenv(\"DDEV_NO_SENTRY\") == \"true\"`.",
+      "reference": "I think it'd be a good idea to define the environment variable name as a constant in values.go because it's hard-coded in several places, but that's a small point."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Style/SymbolArray: Use %i or %I for an array of symbols."
+    },
+    {
+      "prediction": "I think this should be `CheckComposeDockerCompose` to match the name of the function.",
+      "reference": "Our habit is to go ahead and put a description line (or more) in front of every function, not just non-test or exported functions."
+    },
+    {
+      "prediction": "I don't think this is necessary. The header is already set in the `OAuthRequest`.",
+      "reference": "I'm slightly concerned about this because doesn't ORCID use XML instead of JSON?"
+    },
+    {
+      "prediction": "I think this should be `datetime.min` instead of `datetime.min.replace(tzinfo=timezone.utc)`",
+      "reference": "brackets around if conditions isn't really pythonic."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "It turn out that when we close server it closed underlying listener itself, so this line always generated and error."
+    },
+    {
+      "prediction": "I think we should keep `breed` and `name` as `string`",
+      "reference": "Looks like this is for debugging? Maybe just remove."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it's the only way I could come up with that worked.",
+      "reference": "Same here - this code is duplicated. I think we can just remove the `exists?` method completely from `Criteria` and it should delegate to the context."
+    },
+    {
+      "prediction": "I think this should be `usm_mem.resize(0x1440);` instead of `usm_mem.reserve(0x1440);`",
+      "reference": "I'm not quite sure that we preallocate memory. Do you have a good reason for that?"
+    },
+    {
+      "prediction": "I think we can remove this function now. It's not used anywhere.",
+      "reference": "If we have added to s.clients, we could just do the following since if its a route will be ok I think. s.mu.Lock() defer s.mu.Unlock() return len(s.clients) <+ s.opts.MaxConn"
+    },
+    {
+      "prediction": "I think it would be better to have a separate function for this, so that we don't have to repeat the logic in the `invoke` function.",
+      "reference": "if the targetLeader is src_ itself, it is really need call transLeader function?"
+    },
+    {
+      "prediction": "I think this should be `getAndSavePageResponseTypeName` instead of `getAndSavePagedResponseTypeName`",
+      "reference": "Can we remove this featureConfig since you removed in L98 (assuming it is not used else where)"
+    },
+    {
+      "prediction": "I think this should be `using System.Linq;` instead of `using System.Linq; using System.Collections.Generic;`",
+      "reference": "namespaces in this file need a sorting"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `ConfigureDefaultMTUs` since we're setting `Wireguard.EncryptHostTraffic` and `c.Wireguard.MTU`",
+      "reference": "This line should actually be: `if c.KubernetesProvider == config.ProviderAKS && c.Wireguard.EncryptHostTraffic {` because we only need to tweak the MTU like this on AKS."
+    },
+    {
+      "prediction": "I think this should be `maxSynapsesPerCell` instead of `maxSegmentsPerCell`",
+      "reference": "`get` has an optional second parameter that is the default if the key isn't found."
+    },
+    {
+      "prediction": "I don't think this is necessary. The `account_out` tag already contains the `$user->id` value.",
+      "reference": "Please also indent the new response line as was done to the existing line above."
+    },
+    {
+      "prediction": "This should be `auto_fire = !this->GetTarget()`.",
+      "reference": "Shouldn't need this-> here."
+    },
+    {
+      "prediction": "This should be `MESSAGE_EVENT_ADVANCE` instead of `MESSAGE_TRADE`",
+      "reference": "I believe this is incorrect, if I'm not wrong, MESSAGE_TRADE should be used when buying/selling items from NPC's"
+    },
+    {
+      "prediction": "I think this should be `DetectionsGraphController` instead of `DetectionsGraphController`",
+      "reference": "Better if you give split[0], split[1], and split[2] descriptive names than using them this much through so many lines of code."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be moved to `amp-base-grid` component.",
+      "reference": "I know we're deleting this soon but I thought I'd fix it anyway :smile:"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to check if the type of the argument is the same as the one of the marker. If they are the same, we should use the marker as the message. Otherwise, we should use the argument as the message. I'm not sure if this is the right way to do it. I think we need to check if the type of the argument is the same as the one of the marker. If they are the same, we should use the marker as the message. Otherwise, we should use the argument as the message. I'm not sure if this is the right way to do it. I think we need to check if the type of the argument is the same as the one of the marker. If they are the same, we should use the marker as the message. Otherwise, we should use the argument as the message. I'm not sure if this is the right way to do it. I think we need to check if the type of the argument is the same as the one of the marker. If they are the same, we should use the marker as the message. Otherwise, we should use the argument as the message. I'm not sure if this is the right",
+      "reference": "I like this, makes it clearer what scope the continuation is in"
+    },
+    {
+      "prediction": "I think this should be `A1` instead of `top-left-corner!`",
+      "reference": "These test checks if the value of the merged cells is correct. Please revert the changes and set `autoWrapCol` and `autoWrapRow` to `false` to the Handsontable instance. This change applies to the entire mergeCells.e2e.js file."
+    },
+    {
+      "prediction": "I'm not sure if this migration is needed. I think we can just remove the `clearHashedChangesets` migration from the changelogs.",
+      "reference": "Don't need to clear them - because nobody have data there, also can don't delete buckets - just leave them, new nodes will not have them if remove bucket from buckets.go"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this interface. I think it should go in `com.google.api.codegen.`",
+      "reference": "Similarly, this has only one implementing class"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to do something like: DEFAULT_HEADERS = {'Accept' => 'application/x-www-form-urlencoded', 'Content-Type' => 'application/x-www-form-urlencoded', 'User-Agent' => 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/14.0 Safari/605.1.15', 'Accept-Language' => 'en-US,en-US;q=0.8,en;q=0.5', 'Accept-Encoding' => 'gzip, deflate', 'Connection' => 'keep-alive', 'Accept-Charset' => 'ISO-8859-1,utf-8', 'Keep-Alive' => 100, 'Strict-Transport-Security' => 'max-age=3111111111111111111111111111111111111111111111111111111111111",
+      "reference": "Does it send requests with urlencoded bodies anywhere? I thought it sends only json. Maybe content-type should be `application/json` by default?"
+    },
+    {
+      "prediction": "I think this should be `?PhpParser\\Node\\NullableType|null`",
+      "reference": "I must have forgotten that one in previous PR"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "This test name doesn't really reflect what the test does any more... Then again I'm not really sure what it is testing now - is it actually just testing that FactoryGirl can create records? or does that itself funnel through the rails code so that we're testing rails can create records?"
+    },
+    {
+      "prediction": "I think it would be better to use `m.Values` instead of `m.Runtask.Spec.Task` here.",
+      "reference": "Can this be `BuilderForYAMLObject`"
+    },
+    {
+      "prediction": "I'm not sure this is the right solution. I think we need to close the pool after we're done with it.",
+      "reference": "better use `with mp.Pool(...) as pool:` here"
+    },
+    {
+      "prediction": "I don't think this is the best way to handle this exception. I think it would be better to return a 500 with a stack trace.",
+      "reference": "Since this is a config problem, BAD_REQUEST is probably not the right response."
+    },
+    {
+      "prediction": "I think this should be `errors = append(errors, err)` instead of `errors = append(errors, err)`",
+      "reference": "We could probably do 1 sec if we have time, and hence exit early if we have small number of clients. So LB = 60 sec. Grace = 10secs. So 50Sec window, if NumClients < 50, Close 1 every second. Then exit when NumClients == 0."
+    },
+    {
+      "prediction": "I think this should be `this_column == that_column and this_name == that_name` instead of `(this_column == that_col) and (this_name == that_name)`",
+      "reference": "can you rename `that_col` to `that_column`?"
+    },
+    {
+      "prediction": "I think we can remove this file and just use `Widgets` in `componenets.js`",
+      "reference": "This is not correct, it will override everything else that is currently in `Widgets`. This file shouldn't be modified at all, we should be adding `components.Widget` in `assets/js/googlesitekit/widgets/index.js`, where the rest of that object is defined. Oh, and there's a typo here :)"
+    },
+    {
+      "prediction": "I think it would be better to move this check to the end of the `if` statement.",
+      "reference": "Each task finished will result in the meta info updated?"
+    },
+    {
+      "prediction": "I think this should be `replace( /\\s+/g, '' )` instead of `/\\s/g`",
+      "reference": "Quick follow-up @aaemnnosttv, why this and not `.trim()`?"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "I don't think we should have file-wide exceptions for this rule, let's annotate the individual instances."
+    },
+    {
+      "prediction": "I think this should be `if (false)`",
+      "reference": "@raprasad you plan to take this \"if true\" out, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to do something like this: `flows = [file(path, \"rb\") for path in paths]` Then you can do `flows[0]` to get the first one, `flows[-1]` to get the last one, etc.",
+      "reference": "Could you adjust this to `with open(path, \"rb\"):` here and below? We should make sure that we close all files."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Does this need its own import block or can it be combined with the imports below as in most other files?"
+    },
+    {
+      "prediction": "I think this should be `False` by default.",
+      "reference": "It defaults to `false`."
+    },
+    {
+      "prediction": "I don't think we need to check for `isPreflightCorsRequest` here, since it's only used in `applyPreflightCorsResponse`.",
+      "reference": "Why conditional check is done? As i understand, later both `applyPreflightCorsResponse()` and `allowAllCorsActions()` does the same"
+    },
+    {
+      "prediction": "I think this should be a warning, not an error.",
+      "reference": "has not yet exit?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `radio.py`",
+      "reference": "I wonder if the version should be part of the deprecation API. It might ease our future selves while debugging issues =)"
+    },
+    {
+      "prediction": "I think this should be `logger.info(\"Using delegate supplier: \" + clazz.getName());`",
+      "reference": "We chose `info` in the test code for obvious reasons. Changing to `finest` makes debugging harder and noisier."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be in the `CreateAnnotatedCodeLocationsFromCodeFlows` method, since that is where the code flow annotations are created.",
+      "reference": "This is just a rename..."
+    },
+    {
+      "prediction": "I think this should be `return \"\"` instead of returning an error.",
+      "reference": "Should probably panic here. I think hashers are contracted not to return errors (and returning \"\" doesn't handle the error)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should be in `pkg/chainlink/chainlink/core/types.go`",
+      "reference": "Aside: this is a confusing name for an interface that contains multiple addresses."
+    },
+    {
+      "prediction": "I think this should be `exit_code = 1` instead of `exit_code = 0`",
+      "reference": "We expect to use Bolt::CommandResult for scripts as well?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `Metadata::ExampleGroupHash` since that's where the `superclass_metadata` and `user_metadata` are created.",
+      "reference": "same thing as above inre commas and args?"
+    },
+    {
+      "prediction": "I think this should be `memoryCircuitBreakerThreshold` instead of `memoryCircuitBreakerThresholds`",
+      "reference": "Should we validate that this is between 0 and 100?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "reference": "`HardKiller` doing `os.Exit()` without proper shutting down other dependencies. I think there should be a better approach for this."
+    },
+    {
+      "prediction": "I think this should be `wlr_drm_format_set_add(&renderer->data_ptr_texture_formats, formats[i].drm_format, DRM_FORMAT_MOD_LINEAR);` instead of `wlr_drm_format_set_add(&renderer->data_ptr_texture_formats, formats[i].drm_format, DRM_FORMAT_MOD_LINEAR);`",
+      "reference": "Why isnt the return value checked?"
+    },
+    {
+      "prediction": "I think this should be `fmt.Println(\"obj\", obj)` and `fmt.Println(\"fields\", fields)`",
+      "reference": "remove this debug"
+    },
+    {
+      "prediction": "Please use LBANN_ERROR(...) instead of LBANN_ERROR(\"This layer has no children\")",
+      "reference": "This should change to a call to `Layer::find_layer_index`, which should be renamed to `find_child_layer_index` and it should return a `size_t` (technically the `difference_type` for whatever iterators are going to be used)."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think you need to use `target.script` instead of `getattr(target, \"script\", None)`.",
+      "reference": "what is this change about?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `TestInfra` class, but I'm not sure if that's the best place to put it.",
+      "reference": "Why did we leave the `stderr` param in this case while we removed it from other bake methods?"
+    },
+    {
+      "prediction": "I think this should be `VerifySameTrustDomainMemberID` instead of `VerifyTrustDomainMemberID`",
+      "reference": "This seems to only be used by the server/ca package.... maybe this can live there instead of being lumped into these API implementation helpers?"
+    },
+    {
+      "prediction": "I think this test should be moved to `test/provisionContainerResources.go`",
+      "reference": "hmm.. this is Hacky, the test will succeed in Windows even if you remove the calls to these, right?"
+    },
+    {
+      "prediction": "I think this should be `2.3.0` instead of `2.3.1`",
+      "reference": "should this block be in the file `05_install_ruby.rb`?"
+    },
+    {
+      "prediction": "I think this should be `yarpc.Filters{UnaryFilter: recorder}` instead of `yarpc.Filters{UnaryFilter: recorder, UnaryFilter: recorder}`",
+      "reference": "To match outbounds, let's just call this `Unary: recorder`, `Oneway: ...`."
+    },
+    {
+      "prediction": "I think this should be `hash.update(string).hexdigest()`",
+      "reference": "`string.encode('utf-8')` is more common but I guess this is mostly cosmetic :)"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the `method` and `from` fields in the log.",
+      "reference": "Consider logging `r.Url.Path`? Either way, ship it!"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the `prepare` function in `consensus/consensus.go`",
+      "reference": "this probably should be done lazely only if logic/app call txn in the group. Or even done in LogicSigSanityCheck?"
+    },
+    {
+      "prediction": "Shouldn't this be `enumspb.CONTINUE_AS_NEW_INITIATOR_WORKFLOW` instead of `enumspb.CONTINUE_AS_NEW_INITIATOR_UNSPECIFIED`?",
+      "reference": "Please move this to `common/enums/defaults.go`."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "In general, if you find unalphabetized imports, you should alphabetize them. Fine for now though."
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "we should probably deprecate (and display a warning) the misspelled keyword arg here rather than removing it... and then add the new one. This changes a public API and will break any code that is currently using the misspelled version."
+    },
+    {
+      "prediction": "I think this should be `if (explicitCompensationVoltage.HasValue && (explicitIonMobility != explicitCompensationVoltage || !Equals(explicitIonMobilityUnits, explicitCompensationVoltage)))`",
+      "reference": "Is this if statement here necessary? It looks like this is the same logic that is taken care of in the setter for the property \"CompensationVoltage\"."
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this class. It's not a queue, it's not a part of a queue, it's a part of the full text index. I think it should be something like `DocumentKey` or `FullTextKey` or something like that.",
+      "reference": "Eliminate this function; handled by mappings and analyzer"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be in `AbstractBase\\JSTree.php`",
+      "reference": "I wonder if this new logic would actually fit better as a support method, both for readability and overriding... e.g. <pre> protected function getRouteNameFromDataSource($route) { if ($route === 'collection') { return $this->getDataSource()->getCollectionRoute(); } elseif ($route === 'record') { return $this->getDataSource()->getRecordRoute(); } return $route; } </pre> That's extremely nitpicky, and you can feel free to leave this as-is if you prefer... but that feels marginally better to me."
+    },
+    {
+      "prediction": "I think we should use `log.info` instead of `log.finest` here.",
+      "reference": "This is in test code: understanding what we're doing is important in this context."
+    },
+    {
+      "prediction": "I think this should be `llvm` instead of `tinygo.org/x/go-llvm`",
+      "reference": "This change (and a few similar ones below) are not related to rpi3 support, and should be removed."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think we need to pass kwargs to the `BoxPlot` constructor, not the `koalas.BoxPlot` constructor. The `koalas.BoxPlot` constructor should be passed the `BoxPlot` instance, not the `BoxPlot` instance itself.",
+      "reference": "Hmm, why did we come to need the `,` at the end?"
+    },
+    {
+      "prediction": "I think this should be `userProvidedFields[paths[i]]` instead of `isPathInFields(userProvidedFields, paths[i])`",
+      "reference": "I'm suspicious of this. For one thing, `query._fields[paths[i]]` may be `0`, `false`, etc. so checking for falsy will catch both cases where both the field isn't in the projection and if the field is explicitly excluded from the projection. For another, I'm not so sure that #6546 is a bug. Let's discuss that more."
+    },
+    {
+      "prediction": "I think this is a good idea, but I don't think we need to add it to the interface. We can add it to the interface if we want, but I think it's fine to leave it as-is for now.",
+      "reference": "What is this? I'm a bit unclear why the parser would have a list of statements in it."
+    },
+    {
+      "prediction": "I think it would be better to use `prometheus.Labels` instead of `[]string`",
+      "reference": "I believe this is Prometheus name, wonder if this is descriptive enough of a label name? @nrchakradhar @simonpasquier wdyt?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's the only way I could get it to work.",
+      "reference": "Are most of our other fields default-zero-value? Would `AllowVXLANPacketsFromWorkloads` be a better formulation of this? I think it would make it easier for golang users of the API (default value matches the type zero-value)"
+    },
+    {
+      "prediction": "I think this should be `RemoveTransaction(Keccak hash, long blockNumber, bool removeSmallerNonces);`",
+      "reference": "void RemoveTransactions(Address sander, long removeBelowThisNonce) and separate these two calls"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should be able to do this in the same way we do it in the other tests, i.e. by using `clientConnectToServer` to connect to the server and then calling `clientConnectToClient` to connect to the client.",
+      "reference": "Not sure which interest you are looking for to propagate here, but keep in mind that we have helpers (maybe not in /test package?) to check/wait for interest on a literal on a server for a given account."
+    },
+    {
+      "prediction": "I think this should be `true` by default. If we don't do this, we'll end up with two processes running at the same time, one as root and one as normal user.",
+      "reference": "nit: should not need this."
+    },
+    {
+      "prediction": "I think this should be `___azkaban_project_dir_size_in_bytes___` instead of `___azkaban_project_dir_size_in_bytes______`",
+      "reference": "Since this constant is an implementation detail rather than a user-facing API, is it better to define it in a place where it is used?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `list(json.dumps(x))` will always result in a list, even if `x` is not a list. This is because `json.dumps(x)` will always result in a string, even if `x` is not a string. The fix should be to make sure that `list(json.loads(x, object_pairs_hook=FrozenOrderedDict))` always results in a list, even if `x` is not a list. This is because `list(json.loads(x, object_pairs_hook=FrozenOrderedDict))` will always result in a list, even if `x` is not a list. The fix should be to make sure that `list(json.loads(x, object_pairs_hook=FrozenOrderedDict))` always results in a list, even if `x` is not a list. This is because `list(json.loads(x, object_pairs_hook=FrozenOrderedDict))` will always result in a list, even if `x` is not a list. The fix should be to make sure that `list(json.loads(x, object_pairs_hook=FrozenOrderedDict))` always results in a list, even if `x` is not a list. This is because `list(json",
+      "reference": "It would be nice if we have some docs explaining this. Core luigi should avoid having hard to understand code."
+    },
+    {
+      "prediction": "I think this should be `loadDictionary(ignoreCase, affix, dictionaries, null)`",
+      "reference": "extract a method to call from a test"
+    },
+    {
+      "prediction": "I think this should be `private final ExecutorService collectorExecutor = ExecutorUtil.newMDCAwareCachedThreadPool(new DefaultSolrThreadFactory(\"coreContainerWorkExecutor\"));`",
+      "reference": "nitpick: regular order is private than final."
+    },
+    {
+      "prediction": "Shouldn't this be `missing-param-doc` instead of `missing-any-param-doc`?",
+      "reference": "We could add an old names here, the ideal would be to not force to disable missing-any-param when the old one was already disabled. But they are not really equivalent so maybe you were right to not add it."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like it should be in the same namespace as the other corefx ones.",
+      "reference": ">public const string LINQ = \"LINQ\"; [](start = 8, length = 34) Are there duplicated benchmarks here? #Closed"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Probably worth taking the opportunity to add yardocs to this method now. Esp. curious about what the options hash accepts."
+    },
+    {
+      "prediction": "Shouldn't this be `if (view && !surface)`?",
+      "reference": "This is Very Meh . We want to get rid of `role_data`, and it's an internal field."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I would expect it to be in `extensions` folder, right?",
+      "reference": "`await this.driver().loadExtension(path);` should work"
+    },
+    {
+      "prediction": "I think this is the right fix, but I don't know why it worked before.",
+      "reference": "Is any of this needed anymore?"
+    },
+    {
+      "prediction": "I think this should be `sizeof(gpointer) * 7) +  /* hole */` and not `sizeof(gpointer) +  /* hole */ +  /* hole */`",
+      "reference": "Can you add a mention here and in `ostree-repo.c` that the holes were found with `pahole`?"
+    },
+    {
+      "prediction": "I'm not sure if we should deprecate the `get_search_results` method or not. I think we should deprecate the `get_search_results` method as it is not used anymore, but I don't know if we should deprecate the `search_results` method as it is not used anymore either.",
+      "reference": "While we're changing this, I wonder if we can do away with `extra_controller_params`.. Maybe a new type of `solr_search_params_logic` that appends the attributes?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I believe that you only need to rescue Beaker::DSL::Assertions, as they include Minitest::Assertions."
+    },
+    {
+      "prediction": "I think this is the only place where we need to set the ActivityTraceFlags. The rest of the code is just setting the ActivityTraceFlags to the correct flags.",
+      "reference": "Do we need `&=` or `=` is sufficient?"
+    },
+    {
+      "prediction": "I think this should be `joinPath($config->get('config_sync_directory'), $config->get('SitePath'), $config->get('SyncDir'))` instead of `joinPath($config->get('config_sync_directory'), $config->get('SitePath'), $config->get('SyncDir'))`",
+      "reference": "Let's keep both of these here. It should work on most any version of Drupal 8 then true?"
+    },
+    {
+      "prediction": "I'm not sure why we're doubling the tolerance here, but 10 seems like a reasonable increase from what I've seen in other tests.",
+      "reference": "@lmolkova This ok? I'm getting random failures from the build checks that don't happen locally from a few non-deterministic time related comparisons."
+    },
+    {
+      "prediction": "Can you add a license header to the top of the file?",
+      "reference": "Q: are we supposed to change this for files that already exist?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `dynamic_map.py` since that is where the `DynamicMap` class lives.",
+      "reference": "I count only four times where `selection_specs` had to be specified as a keyword instead of by position! If that is how often it was used that positional argument in our own codebase, I am pretty certain users barely used it (if at all)."
+    },
+    {
+      "prediction": "I'm not sure why this test is here, but I think it should be moved to `spec/decorators/HashDiffDecorator_spec.rb` so that it can be easily tested independently.",
+      "reference": "is this test for the case above? seems to cover a numeric rather than empty val?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. What happens if the user wants to update `Content-Type` and `Content-Length`?",
+      "reference": "I like the general idea, but this will break: - JSON is not binary-safe, so anything binary will break this. - JSON is super slow for multiple-MB things - We want to have drag-and-drop upload - the easiest way to implement this is FormData upload, so we should have a multipart/formdata endpoint. Can we put to /flow/content?"
+    },
+    {
+      "prediction": "I think this should be `bool dynamo_exited_synched_log_and_stats = false;`",
+      "reference": "_and_synched seems to be more consistent w/ exited_and_cleaned"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "nit: in case of whole file `ifdef` it may be helpful to have `_netcore31` suffix it in the name of the file as well."
+    },
+    {
+      "prediction": "I think you can remove the `typeMap.get(\"airplaneMode\").toString()` line as well.",
+      "reference": "should use Number instead of Long"
+    },
+    {
+      "prediction": "Align the operands of an expression in an assignment spanning multiple lines.",
+      "reference": "Had no idea that method existed!"
+    },
+    {
+      "prediction": "I think this should be `monkeypatch.setattr(tabwidget.objects, 'backend', usertypes.Backend.QtWebKit)`",
+      "reference": "I wonder if we shouldn't just do `w.show()` here, which causes Qt to correctly calculate the sizes."
+    },
+    {
+      "prediction": "I think we should be able to use `object_store.bucket` instead of `object_store.bucket + '/'` here.",
+      "reference": "may be nice to have helper method to generate s3 paths rather than the minor code dup"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it would be better to fetch the private settings first, and then fetch the blog title.",
+      "reference": "This isn't needed here, we still run `this._afterAuthentication` which loads settings & config - the reason the settings fetch is here is to make sure we have all the settings before saving the blog title rather than fetching everything once auth has completed. Probably moot anyway as the oauth code will be removed shortly"
+    },
+    {
+      "prediction": "I think this should be `cfg1.ClientCAs != cfg2.ClientCAs`",
+      "reference": "This line is getting a little long; let's move the conditions to at least two lines (maybe three)."
+    },
+    {
+      "prediction": "Is this needed? I don't see where it is used.",
+      "reference": "Don't use relative paths. src directory is already on the include search path"
+    },
+    {
+      "prediction": "I think this is the only place where `current_resource_owner.completions` is used, so I think it's fine to remove the `only_trail_object_ids` from here.",
+      "reference": "Line is too long. [86/80]"
+    },
+    {
+      "prediction": "Layout/MultilineMethodCallBraceLayout: Closing method call brace must be on the line after the last argument when opening brace is on a separate line from the first argument.",
+      "reference": "We have to navigate up four directories to come down three? I see it, but it's kinda icky. Is something like `load Rails.root + 'app/lib/twitter_digits.rb'` out of fashion nowadays?"
+    },
+    {
+      "prediction": "I think this should be `goto out_dma_close` instead of `res = FPGA_EXCEPTION;`",
+      "reference": "Empty space at end."
+    },
+    {
+      "prediction": "I think this should be `bolt_project: @project,` and not `bolt_project: @project, bolt_executor: @bolt_executor,`",
+      "reference": "I think similar to line 176 here we'll want to call `detect_project_conflict` after overriding this."
+    },
+    {
+      "prediction": "I think we can remove the `const { exec } = require('child_process')` line.",
+      "reference": "How about `const { promises: fs } = require('fs')` and then replacing `fs.promises.` with `fs.`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just close the res.Body after we're done reading from it.",
+      "reference": "This ignores errors closing the body. Not likely to happen, but no reason to take the risk either."
+    },
+    {
+      "prediction": "I think this should be `String.format(\"%s\", query, args)`.",
+      "reference": "Why was this change needed?"
+    },
+    {
+      "prediction": "This should be `ARG_IMPORT_PEAK_BOUNDARIES_PATH`",
+      "reference": "Needs a description added to CommandArgsUsage.resx"
+    },
+    {
+      "prediction": "I think this should be `getQuantityString(R.plurals.download_report_content, successfulDownloads, failedDownloads, successfulDownloads, failedDownloads);`",
+      "reference": "The line is a bit too long. That's why the test currently fails. Please break it into two lines."
+    },
+    {
+      "prediction": "I think this should be `slots.size() > 0` instead of `slots.stream().anyMatch(slot -> slot.isSupporting(caps))`",
+      "reference": "Prefer `Stream.anyMatch` instead of iterating over all slots."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Can this just be a feature flag to check if it's on? This will give us flexibility on launch date (which could be turned on July 1st) and allow us to easily revert back in case the policy is reverted."
+    },
+    {
+      "prediction": "I think this should be `uint8_t` instead of `uint8`.",
+      "reference": "Due to this change, src/examples/adc needs to be modified"
+    },
+    {
+      "prediction": "I think we can remove this now that we have `sortedLabels` as a struct.",
+      "reference": "`s/singlePointer/singlePtr` or please rename the type to `singlePointer`."
+    },
+    {
+      "prediction": "I think this should be 100000.",
+      "reference": "Do we need it? If right now it's toml"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `componentDidMount` or `componentWillMount`",
+      "reference": "You are never calling oldUnmount, this could lead to a plugin chain failing."
+    },
+    {
+      "prediction": "I think this should be `0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0",
+      "reference": "What does setting partitionInfo to null mean exactly?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `plot.js` since it's only used by `bokeh.plotting.js` and `bokeh.plotting.js` should be the place to do this.",
+      "reference": "Not sure you need the initial ``not_updated`` as ``any([])`` evaluates to false."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "Since this migration hasn't been merged to master yet, what do you think about just squashing these two into the migration you really want?"
+    },
+    {
+      "prediction": "I think this should be `BetterErrors.subl = :subl if defined? BetterErrors`",
+      "reference": "This doesn't apply to all developers"
+    },
+    {
+      "prediction": "I think this should be moved to the top of the file.",
+      "reference": "I think `serializer.xml` can be moved into condition below too"
+    },
+    {
+      "prediction": "I think this should be `dotnet run --framework netcoreapp3.1 --configuration Release --filter OpenTelemetrySdkBenchmarksActivity`",
+      "reference": "Probably put this in a simple README.md file?"
+    },
+    {
+      "prediction": "I think this should be `RPCViolation | OneAPIViolation | SDKViolation` instead of `1 << 0 | 1 << 1 | 2 << 2`",
+      "reference": "Super Cool bitwise   Minor: Most likely you don't need `System.Collections.Generic`, `System.Linq` & `System.Threading.Tasks`"
+    },
+    {
+      "prediction": "I think it would be better to pass the microserviceName instead of the appId.",
+      "reference": "If we just return the TCC, maybe we need to update the method name for it."
+    },
+    {
+      "prediction": "Shouldn't this be `string.Format(CultureInfo.InvariantCulture, \"Additional parameters for the {0} operation.\", createOperationDisplayString(method.MethodGroup.Name.ToPascalCase(), method.Name));`?",
+      "reference": "For consistency, shouldn't this use `SwaggerModeler.GetMethodNameFromOperationId(method.Name)` as above? I'd just reuse `\"Additional parameters for \" + SwaggerModeler.GetMethodNameFromOperationId(method.Name) + \" operation.\"` here, `string.Format` with `CultureInfo` is complete nonsense here anyways."
+    },
+    {
+      "prediction": "I think we should use `0.005f` instead of `0.01f` here.",
+      "reference": "As this is just a backport, I guess it is fine to leave these timings, although they are not enough to make tests stable."
+    },
+    {
+      "prediction": "I think this should be `ecs` instead of `ecs`",
+      "reference": "Why this way?"
+    },
+    {
+      "prediction": "I think this should be `os.TempDir()` instead of `t.TempDir()`.",
+      "reference": "Changing this because otherwise the second run of testExtractCompilationsEndToEndWithDatabase will try to overwrite a generated file and fail. Maybe there's a better way?"
+    },
+    {
+      "prediction": "Shouldn't this be `if (pathParam.startsWith(\"?\") && pathParam.endsWith(\"?\")`?",
+      "reference": "This modification is not correct. Suggections: 1. upgread guava to 30.0-jre will fix this issue"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `libkb/keystore/sha256.go` and `libkb/keystore/sha384.go` and `libkb/keystore/sha512.go`",
+      "reference": "All changes to signature code stem from here. The syscalls interfaces expects VerifySignature to return an error. I performed the change here and bubbled it up through the rest of the code - mostly mechanical."
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "This strategy seems good. Just wondering do you think this is a good time to move stuff out of core? For example, I guess storage, since it's for use by server and not client, doesn't need to be Java 6?"
+    },
+    {
+      "prediction": "I think it would be better to have a method `isPartitioned` that returns `true` if the table has partitioned columns and `false` if it has non-partitioned columns.",
+      "reference": "We had a wrong predicate before: we compared a list to an expression."
+    },
+    {
+      "prediction": "I think this is the wrong way around. `mask` should be `mask::rwx`, not `mask::rwx\\ngroup::rwx`.",
+      "reference": "Name copy-pasted from above. \"empty\"?"
+    },
+    {
+      "prediction": "Why do we need to do this?",
+      "reference": "May I ask why change this?"
+    },
+    {
+      "prediction": "I'm not sure we need to add this. I think we can just use the size of `value.Tags` and `value.Metrics` here.",
+      "reference": "Does this mean we now always include the dictionaries even if they're empty? If so, we should make sure that this doesn't break the Agent (even older versions). It's possible that it doesn't handle empty dictionaries well."
+    },
+    {
+      "prediction": "I think this should be `long`",
+      "reference": "The description and name is unclear. Typo in 'which'."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Files should have an extra line at the end of them."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should go in `proc.h` under the `num_simd_saved` definition.",
+      "reference": "> i#1312 AVX-512 support: Add code cache to/from context switching. s|to/from|| (redundant and mildly confusing (called \"enter\" and \"return\" in code))."
+    },
+    {
+      "prediction": "(style) space required after that ',' (ctx:VxW)",
+      "reference": "This stat may fail, right? I think dag_file_exists does not actually check the file is there."
+    },
+    {
+      "prediction": "I think it would be better to use `{}` instead of `{ validationLevel: testValidationLevel }`",
+      "reference": "I think we can make a safe assumption that if only two values are passed in then you have `(optionsToValidate, optionsForValidation)`, if its three then you have `(optionsToValidate, overrideOptions, optionsForValidation)`"
+    },
+    {
+      "prediction": "I think this should be `k.Edges()`, not `edgeSets` and `edgePages`.",
+      "reference": "If there are more items to add to what's being written out here, please add a TODO."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this change. I think it should be in `super_socket.py` since it's only used by `recv()` and `send()`.",
+      "reference": "Can you add an `else:` case to handle an invalid length? That will prevent weird errors."
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "reference": "* We need to deprecate/internalize first on master. * We should directly make the AST node final now * The setter `setName()` can be package-private."
+    },
+    {
+      "prediction": "I think this should be `List<InitCodeNode> paramInits = param.listInInitializationOrder();` instead of `List<InitCodeNode> paramInits = param.listInInitializationOrder();`",
+      "reference": "Where did this functionality move to?"
+    },
+    {
+      "prediction": "Why do we need a separate `splitTableRegion` method? Can't we call `splitTableRegion` from `dropTable`?",
+      "reference": "maybe should firstly check whether current tidb support `split table region`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this error notification. I think it should be in `sitekit.js` since it's a core sitekit file.",
+      "reference": "Please use the same (plural) name for the new component since it renders multiple notifications."
+    },
+    {
+      "prediction": "I think we should log the error here as well.",
+      "reference": "Perhaps show attempt x of y in message ie: Failed to download block %d (attempt %d of %d). %v"
+    },
+    {
+      "prediction": "I think we should keep the `getPrivacyGroup` method signature as it was before, and just change the return type to `Optional<String>`. This way, we don't need to make any changes to the `PrivateTransactionHandler` class.",
+      "reference": "why did you took `Exception` out? Is this method only throwing runtime exceptions?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this information. I think it should be in the `binary` section, since it's specific to that objective, but I'm not sure if that's the right place to put it.",
+      "reference": "please revert all of these unrelated whitespace changes"
+    },
+    {
+      "prediction": "This enum is not used anywhere, so I think it should be removed.",
+      "reference": "PGP_INLINE is clearer IMHO"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Please group imports"
+    },
+    {
+      "prediction": "Why is this increased from 5 to 10?",
+      "reference": "Would it make sense to expose this as a config option in `TestHiveMetastore`? A small number of worker threads helped us to catch a few connection leaks early on."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. It seems like it should be in the same place as `dilation_h` and `dilation_w`, but I'm not sure.",
+      "reference": "Let's fix the extra spaces above."
+    },
+    {
+      "prediction": "I don't think this is the best name for this method. `List<WebElement>` is a bit vague. What about something like `findMatchingElements` or `findMatchingElementsBy`?",
+      "reference": "should a similar change be done for findElement?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "// Keys are stored in an in-memory map, where the key is the name of the // private key, and the value is the structure where the actual private key and the password are // stored."
+    },
+    {
+      "prediction": "nit: I think we should rename this to `Source` or `SourceSpec` to be consistent with the rest of the API.",
+      "reference": "I think this needs a json inline tag here"
+    },
+    {
+      "prediction": "I think this should be `dict:dict:dict()`",
+      "reference": "Hi @cjimison good point but doesn't this break the compatibility with Erlang/OTP prior to 17.0? I think this could be done over an conditional macro defined in the rebar.config: `{erl_opts, [ {platform_define, \"^R\", pre17} ]}` and some additions with `-ifdef(pre17). ... -else. ... -endif.`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in the `choosePeer` function in the `List` struct.",
+      "reference": "should we change/wrap the error?"
+    },
+    {
+      "prediction": "Shouldn't this be `trace`?",
+      "reference": "Are we sure that passing this log in warning does not risk spamming the logs of a node running on the mainnet or on ropsten etc ? Do we often receive invalid messages of this type ?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I'd like to avoid a method call here atm since this is an extremely hot line in the code."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You need to read the whole file and then decode it as hex.",
+      "reference": "Is this a bugfix?"
+    },
+    {
+      "prediction": "I'm not sure why we have both `Sources` and `FallbackSources`.",
+      "reference": "Is this a relevant change?"
+    },
+    {
+      "prediction": "I think we can remove the `hasVariable` matcher now, since it's no longer used.",
+      "reference": "@dansanduleac what do you think of this one? dots were aligned and no longer are"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `app.js` file.",
+      "reference": "Why we dont delete app specific notes when app is deleted?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in the `ConfigBuilder` class, since it's the only place where we have access to the `pluginList` object.",
+      "reference": "The _OverviewPlugin_ is forcefully enabled without the option to disable it, using _PluginDescription.alwaysEnabled_. That should also work for this plugin?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the same place as the other status fields.",
+      "reference": "Can you facet on this new externalStatus Solr field?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to just do `filter(blacklight_config.facet_fields)` and it will do the right thing.",
+      "reference": "I wonder if it would it make sense to push the pivot facet handling down into `FilterField` (or maybe a subclass, `PivotFilterField`?)?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be possible to do this in a way that doesn't leak the result of the plan.",
+      "reference": "Ah I see what you were saying about making it harder to validate types. We'll have to catch the return for that."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it. I think it would be better to use a `BottomNavigationView` instead of a `BottomNavigationItem` and `BottomNavigationView` instead of a `BottomNavigationView`.",
+      "reference": "Try using using `.setDismissOnTouch(true)` instead"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be possible to do this in the `UpdateSpan` method.",
+      "reference": "Zach, is `UpdateSpan` of concern too? I see that it is called from `ExecuteAsyncInternal`."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Nit: its scope can be reduced to package default"
+    },
+    {
+      "prediction": "Should this be `ReadOnly`?",
+      "reference": "Why is this get;set;?"
+    },
+    {
+      "prediction": "I think this should be `ShrinkLocatorLists(wdata.endpoint.unicastLocatorList)` instead of `ShrinkLocatorLists({wdata.endpoint.unicastLocatorList})`",
+      "reference": "Should we also create sender resources for unicast locators? Should we make a similar change on StatelessReader?"
+    },
+    {
+      "prediction": "Can we add a link to the Datadog docs here?",
+      "reference": "Can we use `dotnet` instead of `net`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the `StartSpan` method, since that's where the span is created.",
+      "reference": "Should this be `ForceFlushAsync`? Given we have the `async` suffix elsewhere."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Is this change intentional?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be moved to `SarifParser.cs`.",
+      "reference": "This could be an integer property if we supported explicitly setting non-string-valued properties."
+    },
+    {
+      "prediction": "Why is this prefixed with `route_` instead of just `route`?",
+      "reference": "this is not consistent with the previous annotation - is it because of the usage in `Admin\\Content\\Category\\list.html.twig` twig template? Should not it be at least mentioned in the commit message? :slightly_smiling_face:"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I'm not sure if we want to block until a termination signal is received. I'm not sure if we want to block until the user enters a terminal. I'm not sure if we want to block until the process is done. I'm not sure if we want to block until the process exits. I'm not sure if we want to block until the user enters a terminal. I'm not sure if we want to block until the process is done. I'm not sure if we want to block until the process exits. I'm not sure if we want to block until the user enters a terminal. I'm not sure if we want to block until the process is done. I'm not sure if we want to block until the process exits. I'm not sure if we want to block until the user enters a terminal. I'm not sure if we want to block until the process is done. I'm not sure if we want to block until the process exits. I'm not sure if we want to block until the user enters a terminal. I'm not sure if we want to block until the process is done. I'm not sure if we want to block until the process exits. I'm",
+      "reference": "why add `return` here?"
+    },
+    {
+      "prediction": "I think this should be `deserializedMessage.Mdc == Bytes.FromHexString(\"0xf8c61953f3b94a91aefe611e61dd74fe26aa5c969d9f29b7e063e6169171a772\"));`",
+      "reference": "put this hash in variable"
+    },
+    {
+      "prediction": "I think this file should be renamed to libkbfs_test.go.",
+      "reference": "Why not use `kbfsblock.ZeroRefNonce`?"
+    },
+    {
+      "prediction": "I don't think we need to add `CallbackEnv` to `models.py` and `lightgbm.py` because it is not used in these two files.",
+      "reference": "what mypy warning does this address? I'm confused by this change."
+    },
+    {
+      "prediction": "This function is not used anywhere, please remove it.",
+      "reference": "Note that `hasAttacked` that is called before this function already performs this check. You probably don't need to check `hasAttacked` then."
+    },
+    {
+      "prediction": "I think this should be `require('raw-loader!@ampproject/design-system-react/components/carousel/__examples__/default.jsx')`",
+      "reference": "Unfortunately, the site example only has access to imports from `components/index.js` that means that you can't import the items file `carousel-items.js`. You can import that file in all the other storybook files though."
+    },
+    {
+      "prediction": "I think this should be `electrum`, not `privkey2electrum`.",
+      "reference": "Leave this out and move the script to the scripts folder."
+    },
+    {
+      "prediction": "I think this should be `setParameter('nameOrHandle', nameOrHandle)`",
+      "reference": "This should only be sent if the driver is speaking to a W3C conformant remote, so we need an if-condition check like we have in the Python bindings."
+    },
+    {
+      "prediction": "I think this interface can be simplified to just `IDuplexPipe` and `IFeatureCollection`",
+      "reference": "TransferFormat.Text in general purpose connection interface? :/"
+    },
+    {
+      "prediction": "I think this should be `@executable` instead of `executable_path`",
+      "reference": "Is this `@missing_text` ?"
+    },
+    {
+      "prediction": "I think we need to do this in the `if !s.getOpts().Cluster.NoAdvertise` check above.",
+      "reference": "This work for both sides?"
+    },
+    {
+      "prediction": "`map` is not a function, so it can't be used as a spread operator. You'll need to use `mapElements` instead.",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "I think this test should be moved to the `gce_api_client` module.",
+      "reference": "Include the standard parameterized error tests as well."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `/api/v1/`",
+      "reference": "All lowercase in route rule."
+    },
+    {
+      "prediction": "I think we should check if `accountID` exists before calling `receiveProperties` and `receiveProfiles`",
+      "reference": "I'd suggest we use `response.profiles[ 0 ].webPropertyId` here instead, to make sure it's based on the profiles retrieved. The value _should_ be the same, but I think it's a bit more accurate like that."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep it as it was before. The problem is that if we increase the number of days by 3, we have to increase the `bgGraph`'s size by 3 as well. I think it would be better to keep it as it was before and just increase the `bgGraph`'s size by 6 when the number of days exceeds 24.",
+      "reference": "This doesn't change the default of 6h, but will increase the displayed timeframe by 3h for each long press, before wrapping around to 3h. This requires 7 long-presses - quiet a journey ..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `node/chainlink/chainlink.go`",
+      "reference": "We still need to figure out a way to get rid of this write method. Is this still necessary?"
+    },
+    {
+      "prediction": "I think it would be better to use `assertEqual` instead of `assertItemsEqual`",
+      "reference": "`assertItemsEqual` compares the sequences after sorting them, is that intended?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "I don't think this one needs to be changed since the only previous path is already an absolute one."
+    },
+    {
+      "prediction": "I think we can remove the `XmsPageable` constant and use `AzureHostSignature` directly in the `IsARMServiceDefinition` method.",
+      "reference": "what's the goal of this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/core/types.go` since it's only used by `pkg/core/types.go` and `pkg/core/types.go` should be in the same package as `pkg/core/types.go`.",
+      "reference": "why is this here? We already have a message pool in the node."
+    },
+    {
+      "prediction": "I think this is the only change that should be in this file. The rest of the changes are unrelated to this particular test.",
+      "reference": "Another mess after all renames and enum conversions. Doesn't break anything though."
+    },
+    {
+      "prediction": "I think this should be `storage=json` and not `storage_strict_json`.",
+      "reference": "I don't think it's necessary to change this for the history plugin tests?"
+    },
+    {
+      "prediction": "Shouldn't this be `127.0.0.1` and not `127.0.0.0`?",
+      "reference": "In general, I suggest using the `kind()` getter inhereted from `TransportInterface`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in the `NLog.Layouts` assembly.",
+      "reference": "Is the removal of `[AppDomainFixedOutput]` on purpose? (looks like a copy-paste...euh..remove error)"
+    },
+    {
+      "prediction": "I'm not sure if we want to remove the `ListUsers.java` sample.",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think this should be called `PublicKey` and not `PublicKey`",
+      "reference": "gofmt everything pls"
+    },
+    {
+      "prediction": "I think we can remove the `+kubebuilder:printcolumn` annotations as they are no longer needed.",
+      "reference": "ditto, We do not need add this feature for PodXXXChaos"
+    },
+    {
+      "prediction": "I think this should be `lowerCamelApiTypeName` instead of `lowerCamelApiTypeName.toLowerCamel()`",
+      "reference": "There is a simpler way to do this: 1. Don't bother with the whole lowerCamelApiTypeName in SampleConfig 2. Use this code: `return localVarName(Name.upperCamel(apiTypeName, \"Service\"));` Basically, each of the static creation methods of Name just need to have their arguments all be in the same casing, so you would just pass the upper camel string \"Service\" here."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this change.",
+      "reference": "Can you explain this change to me?"
+    },
+    {
+      "prediction": "I think this should be `artifacts/storageclass.yaml` instead of `artifacts/storageclass-1r.yaml`",
+      "reference": "let us rename the artifact to `storageclass-single-replica.yaml`"
+    },
+    {
+      "prediction": "I think this should be `KestrelTransportLibuvFunctions` or `TransportLibuvFunctions` to be consistent with the other namespaces.",
+      "reference": "Nit: I should have done this before, but rename the file to LibuvFunctions.cs. Make sure no other files are misnamed."
+    },
+    {
+      "prediction": "I think this should be `UUID = (byte) (20 << 5)`",
+      "reference": "We should not add new types unless absolutely required. I'm not sure `UUID` is requires a new type"
+    },
+    {
+      "prediction": "I don't think you need to define this in the base class. You can just override the method in each subclass.",
+      "reference": "If only Gatling uses it, can we skip introducing new entity for now? Looks like overkill..."
+    },
+    {
+      "prediction": "I don't think we need this, it's already in the APIV1WalletsResponse",
+      "reference": "rename -> info ?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It should be in the `scheduler` section, under the `upstream` key.",
+      "reference": "Remove the `config_path=dict(...)`, that is automatic!"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could come up with.",
+      "reference": "I had to create a new flattened tree to support this functionality. The only other place in which a flattened tree is created is during the audit itself. Since `axe.cleanup` can be called at any arbitrary time, we can't reuse the same tree."
+    },
+    {
+      "prediction": "I think we can remove this method and just use `model.setConfig(config);` in the `addConfigs` method.",
+      "reference": "As this method is just set the internal cofigModel, it's better to use setConfigs instead of addConfigs."
+    },
+    {
+      "prediction": "I'm not sure if we want to set this to `10 * time.Minute` or `10 * time.Second`?",
+      "reference": "Is this line needed?"
+    },
+    {
+      "prediction": "I'm not sure if we should be using the v1alpha1 or v1beta1 version of this package here.",
+      "reference": "nit: import formatter"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. I think it should be done in the `BuildResponses` method, which is called by the `method.IsXNullableReturnType = true;` line above.",
+      "reference": "can you use LINQ's `All` instead? Should be a side-effects free one liner with that."
+    },
+    {
+      "prediction": "This should be `#ifdef __ANDROID__`",
+      "reference": "Could you please remove `#include <spawn.h>` as a whole and see what happens? I believe that this is no longer needed; it was necessary when we called `posix_spawnp` directly, but now we use `h2o_spawnp`."
+    },
+    {
+      "prediction": "Why is this `false` and not `true`?",
+      "reference": "`whether this resource is paused` to `whether this resource has been paused` ? To keep same doc style."
+    },
+    {
+      "prediction": "I think this method should be called `hasPermissionToManageHighPriorityFlow`",
+      "reference": "This method is validating permission for hp flow. Can you please rename this method to convey the same?"
+    },
+    {
+      "prediction": "I think we can remove the `beforeHandshake` kwarg from the `options` argument here.",
+      "reference": "Just to confirm, do we actually want to explicitly set the `kBeforeHandshake` symbol on all network errors regardless of whether the options/property is passed in?"
+    },
+    {
+      "prediction": "I think this should be `r.inSync = false` instead of `r.inSync = true`",
+      "reference": "How soon after this will Apply() be called? It would be a shame if there was still a significant delay before a missing or superfluous route was corrected."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to just do `fmt.Fprintf(&buf, \"\\t%s %q\\n\", g.imports[path], strings.Replace(path, g.imports[path], \"\", -1))`",
+      "reference": "While this is commonly correct, the more reliable check is to get the package's identifier from the type checker and check those for equality. The information should be there, just another matter of plumbing it."
+    },
+    {
+      "prediction": "I think this should be `return '{}{}{}'.format(full_parent_name, resource_type_name, full_resource_name)`",
+      "reference": "Just to clarify, should this be 'composite_root/root/' or 'composite_root/'?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just set the subscriptionID to the empty string, and then we can remove the check for subscriptionID in DeletePullSubscription.",
+      "reference": "Any reason not to move it to after line 281? These are all pubsub related fields IIUC."
+    },
+    {
+      "prediction": "nit: `Negative Number` or `Negative number`",
+      "reference": "Checks if value is a negative `Number` primitive or object. Zero is considered neither positive nor negative."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to get the scheme directly from the request.",
+      "reference": "Just a nit, but remove the empty line here. :)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the default corpus. I think it should go in `pkg/compile/rule.go` under the `Apply()` method.",
+      "reference": "A `GoPackageInfo` details message should be added to the `CompilationUnit_FileInput` to ensure the import path remains the same."
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "Please don't create new fields with the `m` prefix. I'd rather mix than introduce more of these."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `query` sub-config, under the `circuitBreaker` section.",
+      "reference": "Can we please simplify these names? they are awfully verbose and repeating the parts that are already unique and obvious."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to make this change. I think it should be done in the `bccRecipient` constructor.",
+      "reference": "Flipped the logic here, was this actually wrong before?"
+    },
+    {
+      "prediction": "nit: should be `res = 0;` instead of `bool nat_lvl1_drop = 0;`",
+      "reference": "I think (a) should be initialized to `NAT_FE_LOOKUP_ALLOW` and (b) it should be probably the first thing in `calico_v4_nat_lookup2`"
+    },
+    {
+      "prediction": "I think this should be `./gradlew testClasses`, not `./gradlew :testClasses`.",
+      "reference": "I'd look as to what this class actually does... seems weird. Classpath below (under \"run\") is wrong for gradle."
+    },
+    {
+      "prediction": "I think this file should be moved to `src/test/composable/algod/algod_test.go`",
+      "reference": "Group w/ algorand imports"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "I don't think this import is being used."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/query/parse.go` since it's only used by that file.",
+      "reference": "This is an interesting way to do this. I like it. An alternative way to implement a multiple kind pattern was to have this return `AnyKind` and then have match perform an actual check to see if it matches."
+    },
+    {
+      "prediction": "I think it would be better to use `null` instead of `master` and `primary`.",
+      "reference": "You did not check this code"
+    },
+    {
+      "prediction": "I think this should be `if len(data) == 0 || len(types) == 0 { return nil, nil }`",
+      "reference": "This was necessary due to a bug uncovered in testing. If an actor method which requires params is called without any params at all, not returning an error here causes a panic."
+    },
+    {
+      "prediction": "I think this should be `amp-story-bookend` instead of `amp-story`",
+      "reference": "Thanks for the pr. Would you be able to edit the file to use spaces rather than tabs? Normally we have prettier run on commit, but I guess file patches don't do that."
+    },
+    {
+      "prediction": "I think this test should be moved to `org.apache.hive.hcatalog.tests.testHiveClientPool#testConnectionFailureRestoreForMetaException`",
+      "reference": "Is it intentional that we removed this annotation?"
+    },
+    {
+      "prediction": "I think this should be `ServiceDefinitionDocumentState.All`",
+      "reference": "should be general?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that the `LowestInsertedHeader` can be `null` if there is no head.",
+      "reference": "if lowest inserted header is non null non zero then pivot should be higher and a better best known on the left. maybe Head?.Number ?? Math.Max(Pivot, LowestInsertedHeader))?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `azkaban.execapp.jmx` since that's where the mbean is registered.",
+      "reference": "Is a FakeApp class needed? Can test use a Mockito mock instance instead?"
+    },
+    {
+      "prediction": "I think it would be better to make this a `bool` instead of a `const bool`.",
+      "reference": "We should make this an enum."
+    },
+    {
+      "prediction": "I think this should be `DisplayedTarget = TestAnalysisTarget.Replace('\\\\', '/');`",
+      "reference": "@lgolding, note I fixed the forward slashes that appeared in the VS output (which was certainly not by design)."
+    },
+    {
+      "prediction": "I think this should be `self.handles['y_axis'] = plot.yaxis[0]`",
+      "reference": "Minor unrelated fix that snuck in."
+    },
+    {
+      "prediction": "I think we should keep the import order as it was in the previous commit.",
+      "reference": "Took the opportunity to fix the import ordering."
+    },
+    {
+      "prediction": "I think this should be `self.radiation = float(wavelength)`",
+      "reference": "I think int and float can be handled in one if statement. The subsequent calculations don't really care whether it si a float or an int."
+    },
+    {
+      "prediction": "I'm not sure if this is the right answer, but it seems like it should be.",
+      "reference": "Can we keep this at 10 to reduce the probability of a flake? Or are you afraid this would be too likely to give a false positive if delayed cancellation is every actually broken?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "nit: add a blank line under this."
+    },
+    {
+      "prediction": "This should be `chainID` not `chainIDFlag`",
+      "reference": "no need to add this flag query the endpoint set-up to determine the chainID"
+    },
+    {
+      "prediction": "I think this should be `return !node.querySelectorAll(':not(textarea)').length`",
+      "reference": "Here is a few scenario - What if there is a submit button with in a form, but is always disabled?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. It would be better to have a separate file for this.",
+      "reference": "wrong code: should be `Wh_e`"
+    },
+    {
+      "prediction": "`static` is not needed here.",
+      "reference": "This leads to a small \"leak\" (mostly theoretical only though), if we don't want it this could be created in `initialize()` and destroyed in `finalize()`."
+    },
+    {
+      "prediction": "I think this should be called `BuildIDEnv` instead of `BuildID`.",
+      "reference": "BuildIDOSEnvVarName or similar, otherwise it sounds it's actually holding actual value of env var"
+    },
+    {
+      "prediction": "I don't think we need to log here. It is already logged in line 103.",
+      "reference": "I like this standard \"container has started\" message. Should we have a \"container exited because FOO\" message as well?"
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "And again. The reason it's bad is that if someone uses a JDK not produced by Oracle they won't have this class."
+    },
+    {
+      "prediction": "`applyToWindow` is not used in this file, can we remove it?",
+      "reference": "We have Vue application at our disposal, and have examples container implemented as component. Shouldn't helpers be imported instead being global? Not the best practice in Vue app I guess"
+    },
+    {
+      "prediction": "I think this should be `DiskCacheClosedError`, not `DiskCacheClosedError struct`.",
+      "reference": "No trailing punctuation in the error message."
+    },
+    {
+      "prediction": "I think this should be `clip.close().cancel();`",
+      "reference": "tested? seems like it could break"
+    },
+    {
+      "prediction": "I think this should be `public static void main(String[] args) throws IOException {`",
+      "reference": "This have to be public because the renamed o.a.l.backward_index.TestBackwardsCompatibility refers this."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be in `tests/consensus/unupgraded_test.go`",
+      "reference": "unrelated to your change, but I don't think that this is correct anymore. We also seen to remove application support from `primaryNodeUnupgradedProtocol` for the test to be correct."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this logic to `aws/config.go` since it's only used there.",
+      "reference": "Should this also validate that the RoleArn env var is provided, or just let the creds fail?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `transport/inbound` and `transport/outbound`",
+      "reference": "fwiw, in tchannel-go I've implemented a similar method on TChannel that either returns a Tracer instance it was initialized with, or returns `opentracing.GlobalTracer()`, which by default happens to return a singleton instance of `NoopTracer`. In Go the use of global variables is not frowned upon as say in Java, so this pattern allows minimal work to enable tracing in a service, one just needs to store a real tracer in the global variable, e.g. via `opentracing.InitGlobalTracer(someTracerImpl)`, while the instances of TChannel do not need to be explicitly given any tracer since they will default to that global."
+    },
+    {
+      "prediction": "I think this test should be moved to `test_series.py` since it's not related to the other tests in that file.",
+      "reference": "Shall we also add `ks.Series([1, 2, 3], name=[\"0\", \"1\"])`?"
+    },
+    {
+      "prediction": "I think we can remove the `enodeList` parameter from the `addNodesToWhitelist` method and just pass it to the `web3jService`",
+      "reference": "Can this be updated?"
+    },
+    {
+      "prediction": "Why did you remove the `quality` and `format` parameters from the `addFormat` method call?",
+      "reference": "this change is to avoid duplication on the defaults. They are already on the Configuration class. (Also I removed the false default, because the admin format does not have it). This is also produces the rest of the diff, changes from false to null."
+    },
+    {
+      "prediction": "I think we should be able to remove the `containerAccess.lockContainer(containerID)` and `containerAccess.unlockContainer(containerID)` here.",
+      "reference": "What does it protect from? If it's subsequent CNI calls, won't containerID be different from the first one? or it's different in containerd? we use `getInfraContainer` to get the lock identity in CNIAdd. And this reminds me what if the first CNI call and the subquent ones run into connectInterfaceToOVSAsync, will duplicate ports being created or it has been handled?"
+    },
+    {
+      "prediction": "Shouldn't this be a 404?",
+      "reference": "In Exceptions' constructors, we always add a optional last parameter `Exception $previous = null`. In the past, this was (still is?) a part of the coding standards."
+    },
+    {
+      "prediction": "I think this should be `time.Second * 10` instead of `time.Second * 25`",
+      "reference": "I think more idiomatic (and consistent with elsewhere in our code base, and more natural to read as \"25 seconds\") is `25 * time.Second`"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "Enabling `YAML` by default would be a breaking change in behavior for some applications if the struct's used by that application already include YAML tags, but their application has been (un)marshaling DynamoDB Attributes based on the struct name."
+    },
+    {
+      "prediction": "I think this should be done in the `finally` block of the `try` block above.",
+      "reference": "For the command executor, which in turn, uses the HTTP client to talk to the WebDriver, the client might have high-timeout values set, so the shutdown can take a long time if we wait for it to complete, especially if multiple-long running threads are there. I think it might be a good idea in general to couple the shutdown() with `awaitTermination()` with a timeout, if the ongoing tasks do not complete within that timeout then call `shutdownNow()`. What do you think?"
+    },
+    {
+      "prediction": "I think this should be done in the constructor of HiveMetastoreTest.",
+      "reference": "Is this change needed? `start()` is still defined and uses `HiveMetastoreTest.class`. The only difference is that this doesn't pass a `Configuration` and the parameterless `start` passes `new Configuration()`."
+    },
+    {
+      "prediction": "I don't think we need to expose the state reader to the staking command.",
+      "reference": "I think it is weird to store sr in protocol struct"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this check. I think it would be better to do this check in `getFromCache` instead.",
+      "reference": "Will this cause a reload _all_ NATable information? It's too bad we don't have some way to limit the reload to just the tables that have a metadata mismatch."
+    },
+    {
+      "prediction": "I think this should be `immediately fail unless value X == 0`",
+      "reference": "'fail unless X is a non-zero number' it will also fail if X is a byte string"
+    },
+    {
+      "prediction": "I think this should be `if self.overlaid and True:`",
+      "reference": "SyntaxError I think"
+    },
+    {
+      "prediction": "This is the wrong VUID. The correct VUID is VUID_VK_ACCELERATION_STRUCTURE_VS_00000001.",
+      "reference": "I've filed an internal spec issue to add these missing VUs."
+    },
+    {
+      "prediction": "This import is unused.",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "I think we can remove the `org.apache.flink.table.data.*` imports.",
+      "reference": "In iceberg, we usually don't use `*` to import package, it's more clear to import the specify package one by one."
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "And again. The reason it's bad is that if someone uses a JDK not produced by Oracle they won't have this class."
+    },
+    {
+      "prediction": "I think we can remove the `maxBufferedBytes` variable since we don't use it anymore.",
+      "reference": "Since we now keep track of all bytes written, and extend the write timeout as needed, it's tempting to no longer add 2 times the max buffer size to the bytes written accounting for the connection drain timeout. As we've discussed before, this add several minutes to the timeout with the default 240 bytes/sec rate limit. I'm thinking instead this line changes to `timeoutControl.BytesWritten(minDataRate, 1);` to add a grace period to the drain if necessary. What do you think @Tratcher?"
+    },
+    {
+      "prediction": "I think it would be better to use `Object.keys` here.",
+      "reference": "Good call. Makes we wonder if we should do this with the ARIA props."
+    },
+    {
+      "prediction": "Please remove this import.",
+      "reference": "Please do not use wildcard imports."
+    },
+    {
+      "prediction": "This should be `wlr_log(L_INFO, \"New wlr_surface %p (res %p)\", surface, res);`",
+      "reference": "Not sure about these added logs, as said in the commit message it's probably not something we want all the time, but it helped me debug a bit."
+    },
+    {
+      "prediction": "I think this should be `compare_points` instead of `compare_dataset`. The name `compare_dataset` is a bit confusing because it implies that the dataset being compared is the same as the dataset being tested, which is not necessarily the case.",
+      "reference": "Should there be an equivalent check or is it ok to remove this comparison?"
+    },
+    {
+      "prediction": "I don't think we need to mention `zzzcomputing.com` in the link, it's implied by the `alembic` domain name.",
+      "reference": "sorry, why this change?"
+    },
+    {
+      "prediction": "I don't think this is the right type of NamedList to return from doAnalysis.",
+      "reference": "I think best practice is to use `NamedList<Object>` as the return type, and `NamedList<?>` as the argument type in methods, but I can't find a reference for it right now."
+    },
+    {
+      "prediction": "Why not using the `couponModule` here ?",
+      "reference": "Be careful, the base model is imported here !"
+    },
+    {
+      "prediction": "I'm not sure if we want to add support for `tg` in the future, but I think it's fine for now.",
+      "reference": "Do we want to make this part of the default? Probably not. In fact, the default should probably be eth, web3 and net (which are the standard namespaces on other nodes)."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "What is the plan when we deploy the Crawler VM project and eliminate the Clump model?"
+    },
+    {
+      "prediction": "why remove these two lines?",
+      "reference": "these two are customized config fields, if we don't need them any more, we probably can use default ControllerManagerConfiguration"
+    },
+    {
+      "prediction": "I think this should be `runtime.GOARCH`, not `runtime.GOOS`.",
+      "reference": "I don't think we should be pulling the runtime name from the Compiler. There are two compiler frontends, but both of them compile the same source to generate the \"go\" runtime. I think until there is a request from an alternative implementation (the embedded world **might** have a different runtime), it's fine to hardcode this as \"go\""
+    },
+    {
+      "prediction": "I think this should be `fields = fields + ' ' + this.schema.options.discriminatorKey`",
+      "reference": "fields may be an object."
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "reference": "i'm not confident these are the right defaults... anybody have input?"
+    },
+    {
+      "prediction": "I think we should rename this to `DiskGroups` or `DiskGroupsList` since it's a list of disk groups, not disks.",
+      "reference": "DiskGroupList can be better name"
+    },
+    {
+      "prediction": "I think we can remove the `metricsHandler` argument from this function.",
+      "reference": "nit: lets leave gateway options at the end of the func signature."
+    },
+    {
+      "prediction": "I think this should be reverted.",
+      "reference": "It will be better to use `commit()` on a separate thread, apart from the UI thread. The reason is `commit()` is synchronous while `apply()` is asynchronous. So in case it might not perform actions immediately as expected."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this validation. I think it should be done in the `ValidateTerminationState` method on the `TerminationState` struct.",
+      "reference": "Took me almost 4 hours to find this."
+    },
+    {
+      "prediction": "Please remove empty line.",
+      "reference": "There should be only a single blank line between imports and `__all__`."
+    },
+    {
+      "prediction": "I think this should be 4, not 5. The number of fields returned is determined by the number of distinct keys in the index, not the number of fields in the query.",
+      "reference": "Doc size increased by 1 since _root_ is also returned by queries."
+    },
+    {
+      "prediction": "I think this should be `if lidx == uidx: uidx = lidx+1`",
+      "reference": "Now the return value has changed (or at least now includes `clip`) it might be worth updating the docstring..."
+    },
+    {
+      "prediction": "Why do we need to pass nil here?",
+      "reference": "shadow: declaration of \"err\" shadows declaration at line 410 (from `govet`)"
+    },
+    {
+      "prediction": "I think this should be `foreach ($msg['dataset'] as $attr => $value) { $html .= ' data-' . $attr . '=\"' . $value . '\"' . PHP_EOL; }`",
+      "reference": "Should we be escaping $attr and/or $value here? Seems like htmlspecialchars() might be in order to be on the safe side."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "why not remove the need for the `quiet_o` method and just add it similar to how the `--branch` is added? also, looks like the indenting is a little off"
+    },
+    {
+      "prediction": "I think we can remove these two constants and just use `VERSION_SET_ID` as the flow execution id.",
+      "reference": "These variables are already created in constants. Please reuse those."
+    },
+    {
+      "prediction": "I think this should be `self.version_id` not `self.version_id_id`",
+      "reference": "query_args is a string, not a list. Therefore you cannot call append to it. Also, line 986-989 of key.py already have code that is appending the versionId query parameter."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The error is coming from `decodeMetaData`, not `readMetaData`. I think we need to fix `decodeMetaData` so that it only returns an error if `info` is not nil.",
+      "reference": "I think this change and the one below will break the integration tests which expect a `nil` object if `NewObject` fails."
+    },
+    {
+      "prediction": "I think we should move this to the `config` module, since it's not specific to a particular `Transport` implementation.",
+      "reference": "Looks like this isn't used in this file?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pandas/_frame.py`.",
+      "reference": "why hidden name? How about `index_name`?"
+    },
+    {
+      "prediction": "I don't think this is needed. The default value of `hiveScanMode` is `HIVE_SCAN_NONE`, which is the same as `HIVE_SCAN_SPECIAL_MODE` in this case.",
+      "reference": "Usually, CQDs are not accessed directly in executor operators directly. It should be passed as a flag in the TDB. It is possible that the query is compiled in a different process, then this CQD setting won't be available in the executor layer."
+    },
+    {
+      "prediction": "Why do we need a new tag for ShardItem and ShardContext? I think we can just use `NewAnyTag(\"shard-item\", shardItem)` and `NewAnyTag(\"shard-context\", shardContext)`.",
+      "reference": "This tag was broken, the value write to log was the memory address. Please verify that this actually write out meaningful content in log."
+    },
+    {
+      "prediction": "I don't think you need this. You should be able to run `flake8 -m` and it will tell you if there are any unused imports.",
+      "reference": "Why do you have to disable all of these checks?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `on_element_edit` event handler of the element.",
+      "reference": "Just a note to myself: After merging this, I should edit the line after this one to use `message.error`, as raising a `CommandError` from here seems wrong!"
+    },
+    {
+      "prediction": "I'm not sure why this is removed. It seems to me that it should be restored in `test28_done` even if `TEST_REG2_ASM` is not set.",
+      "reference": "In this test, we want xax to be dead, so that aflags are not spilled to a slot."
+    },
+    {
+      "prediction": "I think we can remove the `EnablePrometheusMetrics` field from `newOptions` and just set it to `true` by default.",
+      "reference": "I remember there were some misleading code about this but forgot to correct them. Although we initialize `config` here, it was discarded in `complete`, so I guess setting the default value here doesn't take effect. We should change L139-L140 as well to use the initialized `config`. So do antrea-controller."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Stubbin a value object (like `RSpec.configuration`) is a bit of a code smell, I think. You can just set `color_enabled` and `success_color` through the configuration APIs provided by `RSpec::Configuration` -- no need to stub. The `sandboxed` thing in `spec_helper.rb` takes care of preventing changes to the configuration object in one example from leaking into other examples."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not the right way to do it. It's not",
+      "reference": "you can use `toString(true)` that will work for BTC and remove the prefix for BCH."
+    },
+    {
+      "prediction": "I think this should be `config.val.tabs.persist_mode_on_change` instead of `config.val.tabs.persist_mode_on_change:`",
+      "reference": "This line is too long now - please break it after the comma."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `pkg/streamer` since that's where the streamer lives.",
+      "reference": "I find this addition to the interface a bit contentious. I think it would be cleaner to just return the response headers together with the new stream on `NewStream`. I.e. change the method signature for `NewStream` to be: `NewStream(ctx context.Context, overlay swarm.Address, headers p2p.Headers, protocolName, protocolVersion, streamName string) (p2p.Stream, p2p.Headers, error)`. I'd like to hear what other think. cc @zelig @janos"
+    },
+    {
+      "prediction": "Shouldn't this be done in the `beforeDelete` hook of the `Cart` entity?",
+      "reference": "i hope there is some magic where isEmpty or some cron can strip all non listable products from cart, but since we have the situation tested it should be OK and also we'll see during tests"
+    },
+    {
+      "prediction": "I think it would be better to have a method `daysFromDate` or `daysFromTimestamp` that takes a `Date` and returns the number of days that have passed since 1/1/1970 up to the given date.",
+      "reference": "I think we should check the type returned by `getLiteral` and handle that here. Then we won't need separate code for different versions."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be in `suite/` since it's a unit test for a single function.",
+      "reference": "Tests for JWT handler functionality?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "It's weird to set a 101 response and complete without writing anything, but this is still a valid test case right?"
+    },
+    {
+      "prediction": "I think this test should be moved to `tests/unit/http.server/test_http_server.py`",
+      "reference": "Should we move that information in the capability itself?"
+    },
+    {
+      "prediction": "I think this should be `_auto_indexable_1d = True`",
+      "reference": "Is this needed?"
+    },
+    {
+      "prediction": "I think this should be a warning, not an error.",
+      "reference": "This log should not be an error as it communicates that we need to retry till we get the cert information for remote ingress controller."
+    },
+    {
+      "prediction": "I think this should be called `FlushTransactions` instead of `Flush`",
+      "reference": "BlockTree alredy has a polluted API, this one seems very internal"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should be in `describe 'apply_manifest' do`",
+      "reference": "Sorry, I realize my line numbers were probably off after you deleted your variables! I meant for this to go in the `before(:all) do` and `after(:all) do` blocks."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `ToParentBlockJoinQuery` class, since it's specific to that query type.",
+      "reference": "This bit I wasn't really sure about."
+    },
+    {
+      "prediction": "I think it would be better to do this in the `parseChecks` function.",
+      "reference": "From the code, a check's metatdata was only added to `axe._load` if a rule used it. Since `role-none` and `role-presentation` were no longer used in any rule, their metadata was never added. This caused any translation file that passed translations for those checks to fail `axe.configure` with > \"Locale provided for unknown check: \"role-none\"\" To fix I added this line to parse all checks regardless of if a rule uses it or not."
+    },
+    {
+      "prediction": "I think this should be `update_policy=\"incoming\"`, not `update_policy`.",
+      "reference": "We should raise `ValueError` if `update_policy not in ['existing', 'incoming']`. (or `in Enum.__members__` or however we want to express legal values)"
+    },
+    {
+      "prediction": "I think it would be better to pass the `window` to the constructor of `MvxIosSetup` instead of passing the entire `window` to the `MvxIosViewPresenter`.",
+      "reference": "`_applicationDelegate` and `_window` are already set in the call for `: this (applicationDelegate, window)`, no need to assigning them again here"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to pass the `selector` to the `Provider` instead of passing the `props` to the `Provider`. The reason is that the `Provider` should be able to access the `context` and the `context` should be able to access the `selector`. If we pass the `selector` to the `Provider`, then the `context` won't be able to access the `selector`.",
+      "reference": "We might not have a selector prop in the consumer"
+    },
+    {
+      "prediction": "I think this should be `!toBeProcessed.Number > 1 && _blockTree.Genesis != _blockTree.Head`",
+      "reference": "IMO == true is more readable than ?? false"
+    },
+    {
+      "prediction": "nit: `parser` -> `parserUtil`",
+      "reference": "feel free when editing sections like this to introduce modern features like object destructuring."
+    },
+    {
+      "prediction": "I don't think this file should be included in this PR.",
+      "reference": "I think the correct pragma is `# -*- coding: utf-8 -*-"
+    },
+    {
+      "prediction": "I don't think you need to import iec104, iec104_fields, iec104_information_elements, iec104_information_objects here.",
+      "reference": "This fixes some import errors, due to the fact that this file has the same name than its module."
+    },
+    {
+      "prediction": "I think this should be renamed to `preferredName` or something.",
+      "reference": "I think we need to clear this out on logout, and update it on login, right?"
+    },
+    {
+      "prediction": "Shouldn't this be `return glnx_error (error, \"Invalid filename\");`?",
+      "reference": "I think normally we'd make this a precondition (using e.g. `g_return_val_if_fail`), but meh, this works too!"
+    },
+    {
+      "prediction": "I think this should be `x-cse-dest-target-microservice`",
+      "reference": "Is this header used by other outside service?"
+    },
+    {
+      "prediction": "This test is no longer valid, right?",
+      "reference": "that's fine by me to have simple test for provisioning check"
+    },
+    {
+      "prediction": "I think it would be better to have a separate method for this, e.g. `new BasicTimeChecker(\"BasicTimeChecker_1\", s.getFirstSchedTime(), s.getTimezone(), s.isRecurring(), s.skipPastOccurrences(), s.getPeriod(), s.getCronExpression());`",
+      "reference": "These two methods seem to be identical except the names. Any idea why two methods are needed?"
+    },
+    {
+      "prediction": "I think this should be in a try/finally block, to make sure the database is created even if there is an error.",
+      "reference": "I don't see much difference between this and `create_db` module. How about having a function with a `database` argument?"
+    },
+    {
+      "prediction": "I think this should be `analytics` and not `googleAnalytics`",
+      "reference": "See above, no need to pass the module name."
+    },
+    {
+      "prediction": "I think it would be better to have eval_set and eval_names in the base class, and then override them in LGBMModel.",
+      "reference": "LightGBM supports multiple validation sets, so please leave `sets`."
+    },
+    {
+      "prediction": "nit: remove trailing whitespaces",
+      "reference": "you don't need to create a enum for this, java allows you to switch on a string ;) (since java 7?)"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "What is this obscure syntax anyway? o.O"
+    },
+    {
+      "prediction": "I think this should be `self._sfdr.get_parameter(parameterName, index=-1)`",
+      "reference": "I think we should have one for the CLA Classifier, and KNNClassifier as well. And how about RecordSensor?"
+    },
+    {
+      "prediction": "I don't think you need this interface. You can just override the getCursorMenu() method in each of your Fragment classes.",
+      "reference": "Missing new line."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the base model.",
+      "reference": "`model.schema.base.options.strictPopulate != null`. Why hardcoding false? Also, `model.base.options` should work and be shorter. Finally, any way to move this up to the `populate()` function in `model.js` that calls `getModelsMapForPopulate()` and shallow clone the object before modifying? A bit random to modify this option here."
+    },
+    {
+      "prediction": "I think this should be `Builders.Build.A.Block.WithOmmers(Builders.Build.A.BlockHeader.TestObject).TestObject);`",
+      "reference": "We now have Prepare.A/Build.A/Builders.Build.A - what are the differences?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Could you reduce this to one line and change the first argument to a single quoted backslash string? Perhaps we can add this as an option named `:preserve_backslash` to the manify method. That way, we don't need an extra method in the first two cases. Obviously, we still need it for the inline case."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `test/testutil/testutil.go` since it's only used by `test/testutil/testutil.go` and `test/testutil/testutil.go` should be the home of unit tests that are not related to each other.",
+      "reference": "`has(default)` i think is the preferred way of doing this"
+    },
+    {
+      "prediction": "I'm not sure if this is the right approach. I think we should be able to use the same `ManifestFiles` instance for both `read` and `select`.",
+      "reference": "Does this mean we can only do rewrites if there are no `DeleteFile`'s?"
+    },
+    {
+      "prediction": "I think this should be `''` instead of an empty string.",
+      "reference": "Actually, these are fine; we're using 110 as the max line length. (it's also set in the .pylintrc)"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "No issue here, but what tool are you running to put these imports into the preferred organization? I have configuration to run `goimports` whenever I save a file, but it appears that that is not quite sufficient."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it's the only thing I can come up with.",
+      "reference": "guessing this method could/should close it but I didn't see where it was getting called."
+    },
+    {
+      "prediction": "I think this should be called `executionSourceId`",
+      "reference": "Could you please use `camelCase` in the property name to follow the naming convention used in the rest of the properties?"
+    },
+    {
+      "prediction": "I think we should use `fmt.Errorf` instead of `fmt.Errorf(fmt.Fprintln(os.Stderr, fmt.Sprintf(fmtErrPublicSubnetsFromEnv, fmt.Fprintln(os.Stderr, fmt.Fprintln(os.Stderr, \"get public subnet IDs from environment %s: %w\", envSecurityGroupCFNLogicalIDTagKey, err)))`",
+      "reference": "Can we revert this change? looks like an extra space was added at the end"
+    },
+    {
+      "prediction": "I think this should be `new InterfaceView().getNamespace(model)`",
+      "reference": "This just grabs any one of the proto services defined in the API, right? What happens when there are multiple (like with PubSub)?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a function that takes a model name and returns the table name. That way you don't have to do so many magic strings and you don't have to do so many magic substitutions in the code.",
+      "reference": "Remove the pylint disable and remove the .format() from the string. The tablename is just 'violations'"
+    },
+    {
+      "prediction": "Do we need to update the `finders` setter as well?",
+      "reference": "Will this implementation allow you to add new exercises to the trail?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to the event hook manager, since that's where the hooks are actually registered.",
+      "reference": "Actually there is race condition here, because pointer assignment isn't atomic on some platform. But this is a rare case we should hit, we can just let it go for now. Maybe this is one of the TODO list."
+    },
+    {
+      "prediction": "I think this should be `if (partitioned) {` instead of `if (partitioned && partitioned) {`",
+      "reference": "We should check somewhere here and in the partitioned_io_buffer that the max_mini_batch_size is properly computed for each model and that it is not aggregate (sum) for all models. The global/max_mini_batch_size should be as specified by the user in the prototext/cmd line."
+    },
+    {
+      "prediction": "I think we should keep the name as `unexpectedAlertBehaviour` for consistency with the rest of the platform.",
+      "reference": "you can change the string value, but not the code constant"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this test.",
+      "reference": "File is not `goimports`-ed"
+    },
+    {
+      "prediction": "Can you add a TODO here?",
+      "reference": "Is this Todo necessary?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to handle the case where `message.get('message', None)` returns `None` instead of an empty string.",
+      "reference": "I would recommend to leave out `None` because None is already the default."
+    },
+    {
+      "prediction": "I think this should be called `inputFileFlag` to be consistent with the rest of the CLI.",
+      "reference": "I may have missed previous discussion on this; apologies if that's the case. What do you think about just `input-yaml`, without the `cli-`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `OnStart` method of the `IConnectionHandler` interface. The `OnStart` method is the place where the `IConnectionHandler` interface is implemented by all the transport handlers.",
+      "reference": "Nit: rename to sendError."
+    },
+    {
+      "prediction": "I think this should be `import numpy as np` instead of `import warnings`.",
+      "reference": "`pylint` wants to know why this `import` isn't at the top of the file; and it's kinda right"
+    },
+    {
+      "prediction": "I don't think this is a `FeedUrlNotFoundException`. I think it should be called `FeedUrlNotFoundException` or something like that.",
+      "reference": "To me, using a `RuntimeException` for this feels a bit weird. I would just extend `Exception`, or maybe even `IoException`. Could you also please overwrite the `getMessage` function to return something like `\"Result does not specify a feed url\"`?"
+    },
+    {
+      "prediction": "I think we can remove the `SetDataContentType` call here, since the `SetData` call already sets the content type to `application/json`.",
+      "reference": "what was the issue here? why did you have to directly set?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think it would be better to check if the country is set in the `ServiceDefinition` before calling `ServiceDefinition.LocationOriginate.Country`.",
+      "reference": "I'm confusied - Isn't `countryString` left empty if `len(country)` != 0?"
+    },
+    {
+      "prediction": "The `QOverload` is not needed here.",
+      "reference": "Would you please update this by using `#if (QT_VERSION >= QT_VERSION_CHECK(5,15,0))` and `QSignalMapper::mappedInt` for Qt >= 5.15?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to stop the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent. I'm not sure if it's possible to do this in a way that doesn't interfere with the gpg-agent",
+      "reference": "This type of change opens a question on how we can cleanly shutdown or prevent all running agents so we do not run the risk of an agent holding kernel mountpoints, which can cause unsafe unmount/chroot exit. Will open a github issue to track this. This specific change itself is fine. It is a pointed change specifically for stopping the offending agent."
+    },
+    {
+      "prediction": "I think this should be `import io.protostuff.runtime.model.protobuf.ModelProtobuf.CodedOutputStream;`",
+      "reference": "this file is model generated by protobuf, just for compatible test it's better not change it."
+    },
+    {
+      "prediction": "I think we should keep the `reason` as part of the `track` method, so that we don't have to pass it in every time we want to track something.",
+      "reference": "It turns out we can send the reason as the property to Segment on the event. Then we can do whatever we want with it. In this case, I think we'll fire a web hook to Zapier which will add a note or message on the user in Intercom."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to go about it, but it's the only way I could come up with that didn't involve a ton of boilerplate.",
+      "reference": "This class is very similar to `Catalog`. Is there a way to have this class compose a `Catalog` instance or extract a common class which they can both compose?"
+    },
+    {
+      "prediction": "`la::matrix<float>::wrap(q, data_arr_device).to_host();` should be `la::matrix<float>::wrap(q, data_arr_host).to_host();`",
+      "reference": "Why did you decide to drop the `queue` from parameters of `wrap` function? The agreement was that all functions for DPC++ case that accept non-host-accessible memory shall accept the queue directly."
+    },
+    {
+      "prediction": "I think this should be `add_setting :hyperlink` and not `add_setting :link`.",
+      "reference": "I think this is a bad name, as its not clear what it does by name alone. If we merge this it should be something like `use_links_to_files_in_output` and should take either a custom scheme like `rspec` or default to `file`"
+    },
+    {
+      "prediction": "I think this should be `CheckSeverity.HARD` instead of `OFF`.",
+      "reference": "Arguably this is actually still valuable for perf, maybe it should just be a warning?"
+    },
+    {
+      "prediction": "I think it would be better to add the error to the response.",
+      "reference": "didn't way say that the topology driver will actually insert the entries to the address book?"
+    },
+    {
+      "prediction": "I think this should be `EMPTY_OBJ._children = undefined` instead of `EMPTY_ARR`.",
+      "reference": "This seems safe to me. My guess is the original checks were a combination of before + after `oldParentVNode` became reliably a `(vnode | null)`."
+    },
+    {
+      "prediction": "I think it would be better to have this as a separate option, e.g. `--filename-extension` or `--file-extension` or something like that.",
+      "reference": "also in the `.md` please."
+    },
+    {
+      "prediction": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "RHEL is an acronym and is usually all caps. Why are you adding a redhat regex matcher?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it should be in the `VectorizedColumnIterator` itself.",
+      "reference": "Maybe I am daft but it looks like you removed fixed width readers but I don't see where you added any readers?"
+    },
+    {
+      "prediction": "I think this should be `WebsiteErrorDocumentPathKey` and `EntryMetadataContentTypeKey` and `EntryMetadataFilenameKey`",
+      "reference": "@acud The reason I moved these here is because I need them in the traversal tests. The traversal package doesn't need this. I don't see any particular reason why any package apart from api would need this. api package imports traversal, so I cannot import api in traversal pkg. Also if these change in future, the traversal tests need to change. So maybe enough reason to move these here? That way both packages can access these as they both depend on manifest"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `AbstractTab` class, since it's the only one that uses the `pin` property.",
+      "reference": "I'd prefer this to be called `pinned` - also it should probably be in the `TabData` class instead, which is accessible as the `.data` attribute of the `BrowserTab` object - otherwise, `BrowserTab` would have a lot of quite \"random\" attributes :wink:"
+    },
+    {
+      "prediction": "I don't think we need to add a new command here, since it's the same as `SEND_COMMANDS_FOR_DOWNLOAD_CHROME`.",
+      "reference": "Nit: `Headless` is one word, not two, and so doesn't need camelcasing in this way."
+    },
+    {
+      "prediction": "I think we can remove the cleanupPlugins import now.",
+      "reference": "Why not import audit as well?"
+    },
+    {
+      "prediction": "Why do we need this change?",
+      "reference": "OK, i think there will be more of those..."
+    },
+    {
+      "prediction": "I don't think we should change the wording of the error messages.",
+      "reference": "Can you add : 'password' => 'The provided password is incorrect.', ?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Extraneous blank line."
+    },
+    {
+      "prediction": "I think this should be `nodes[index]` instead of `axe._tree[0]`",
+      "reference": "nitpick: use `const virtualTree`, I know it was `let` earlier."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the journal itself.",
+      "reference": "Actually, I think it might be easier to do it here. Rough outline: 1. Save a list of the md cache keys and new brmd while looping. 2. After the loop, for each entry: 2a. Look up the old cache entry. If it doesn't exist, continue. 2b. Otherwise, make a deep copy of it, and replace its brmd with the newly-generated one. 2c. Delete the old cache entry, put the new cache entry. What do you think? The problem with the current approach is that the gap between dropping the old version and putting the new version is large enough that when we put the new version, we may be kicking out hotter cache entries. In this case, we'd be doing so to a lesser extent. Ideally, we'd be able to replace the entry in the MDCache LRU without changing the LRU info."
+    },
+    {
+      "prediction": "I think this should be `if (!FileHelper.IsRootedPath(source)) { source = FileHelper.CombinePath(source); }` and not `if (!FileHelper.IsRootedPath(source)) { source = FileHelper.CombinePath(source); }`",
+      "reference": "We need tests for this"
+    },
+    {
+      "prediction": "I think this should be `otlp.WithGRPCDialOption(grpc.WithBlock(), otlp.WithInsecure())`",
+      "reference": "NewExporter takes in two arguments: a configuration for the metrics connection and one for the traces"
+    },
+    {
+      "prediction": "I don't think we need this anymore, right?",
+      "reference": "Why not use currentPlayer instead of playbackManager.getCurrentPlayer()?"
+    },
+    {
+      "prediction": "I'm not sure if we want to include this in the `params.require(:user).permit` above, but I think it's fine to include it here as well.",
+      "reference": "Avoid comma after the last parameter of a method call, unless each item is on its own line."
+    },
+    {
+      "prediction": "I think this should be DataDirOnHost, not DataDir.",
+      "reference": "You should remove the outdated sentence."
+    },
+    {
+      "prediction": "I think this should be called `STATE_FIRST_TIME_NO_SUBJECT` to be consistent with `STATE_FIRST_TIME_EMPTY_SUBJECT` and `STATE_KEY_NUM_ATTACHMENTS_LOADING`.",
+      "reference": "Please restore the empty line after the group of `STATE_*` constants. This visual separation makes the code easier to read."
+    },
+    {
+      "prediction": "I think you can remove the `fileIDs` variable here.",
+      "reference": "Unsure why this is `RestoreFiles` now vs `GoldenRetriever`, thought we deprecated the first?"
+    },
+    {
+      "prediction": "I think this should be `width = flowFileStatus.Width - 2 - (flowFileStatus.VerticalScroll.Visible ? SystemInformation.VerticalScrollBarWidth : 0);`",
+      "reference": "Not sure what this is. Bad merge?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be using `netlink.Attach` instead of `netlink.LinkDel` and `netlink.LinkCreate` instead of `netlink.LinkAdd`.",
+      "reference": "I'm confused, what did this actually fix? Is there a reason we are setting `err` in the parent scope to `nil` in this round-about way?"
+    },
+    {
+      "prediction": "I think this test should be moved to the next PR.",
+      "reference": "Can we remove this   ?"
+    },
+    {
+      "prediction": "I think this should be `confirmation: false` since we don't want to show the \"signing up\" message to non-confirmed users.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": " create a block\n```\nReview comment?",
+      "reference": "line is 195 characters (from `lll`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right package to be using here. I would expect this to be in the `extensions` package, but I don't see it being used there.",
+      "reference": "could not import k8s.io/kubernetes/pkg/util/slice (invalid package name: \"\") (from `typecheck`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in the `componentDidMount` callback of the `useEffect` hook.",
+      "reference": "@disordinary would you be able to configure your editor to use the `.editorconfig` file so that we can avoid final newline changes?"
+    },
+    {
+      "prediction": "I think this should be `Object.keys(Object.getPrototypeOf(this)).forEach(p => result[p] = this[p]);`",
+      "reference": "I really think we should not use this style where unused arguments are named with underscore. While this eases the person writing the function it does it at an expense of readability and debugging. So next person who needs to actually use this argument will need to check the spec and rename this to `obj` and then use it or if he is debugging it he will is not able to watch for a meaningful variable name and will need to inspect `_` which is weird and uncomfortable. We have not used this style anywhere in the code while I agree its very tempting to use it :)"
+    },
+    {
+      "prediction": "I think this should be `new ExLobGlobals(new LobGlobals(NULL));`",
+      "reference": "Thanks for avoiding the default value for the param. But, I think sending heap as NULL might make this program to fail. Have you tested this program? If you have tested this program already, then this PR is ready to be merged."
+    },
+    {
+      "prediction": "I'm not sure if this is the best name for this attribute. I think it should be something like `MvxModalViewPlacementAttribute` since it's specific to the `MvxModalView` control.",
+      "reference": "Don't make it sealed"
+    },
+    {
+      "prediction": "I think this should be `na.serverLimits` instead of `na.jsLimits`",
+      "reference": "Did we miss that for jwt before?"
+    },
+    {
+      "prediction": "I think this should be `Mode: PinSERCOM` to match the other UARTs.",
+      "reference": "It appears that using SERCOM1 was a bug: the only SERCOM that is supported on the TX/RX pins is SERCOM0. Previous UART support wouldn't have worked on the Trinket M0 TX/RX pins anyway (`PA06`/`PA07`). This PR fixes that, and should thereby add support for UART on the Trinket M0."
+    },
+    {
+      "prediction": "I think it would be better to use `Lang::get('system::lang.locale.be', 'flag-by')` instead of `Lang::get('system::lang.locale.be', 'flag-sa')`",
+      "reference": "Spaces for indentation, not tabs"
+    },
+    {
+      "prediction": "I think this should be `global_index_range_` instead of `global_index_range_map_`.",
+      "reference": "doing this as a binding map was insane, but with the ConstBindingIterator to hide the binding->index gunk and avoid the hashes during access it sticks out."
+    },
+    {
+      "prediction": "I think this should be `TEST_EVENT_PATTERN_1 = {'detail': {'EventType': ['0', '1']}}`",
+      "reference": "nitpick: I'd rather move this into the test method directly (we should avoid variables at the root scope if they are not reused in multiple places). (not critical, though - shouldn't hold back the merge..)"
+    },
+    {
+      "prediction": "I think this is the only place where we need to set the current page to match_modifications_page. The rest of the code is already set up to do this.",
+      "reference": "You should do: AssertEx.AreEqual(importPeptideSearchDlg.CurrentPage, ImportPeptideSearchDlg.Pages.match_modifications_page);"
+    },
+    {
+      "prediction": "Shouldn't this be `!= null && !existSchema.getSummary() == null`?",
+      "reference": "SC support batch register schemas we should swtich to new api be careful that maybe will send too big request, need to split to multi request by max size of limit."
+    },
+    {
+      "prediction": "I think this should be `go version` instead of `1.11, 1.12, 1.13, 1.14, or 1.15`",
+      "reference": "At this point, maybe it would make more sense to write this as a range rather than listing specific versions?"
+    },
+    {
+      "prediction": "I don't think we need to add this constant here, since it's only used in proxy_test.go, which is in the same package as this file.",
+      "reference": "I guess we can start by putting the `const StatusContextCancelled = 499` at the bottom of this file"
+    },
+    {
+      "prediction": "I think this is the only test that needs to be removed.",
+      "reference": "Blocking: populate a non-empty value and check it below."
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think it would be better to have an enum with a constant for each API.",
+      "reference": "What is this number value for? How is it used?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "This needs an end keyword"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `Trace.Telemetry.Dependencies` but I'm not sure if there is a better place.",
+      "reference": "I think this is the first record type in Datadog.Trace!"
+    },
+    {
+      "prediction": "I don't think we need the `else` here, since the `return Task.CompletedTask;` is already in the `else` block.",
+      "reference": "Nit: missing newline above."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think we should do it in `proxy.Start()`, but I'm not sure if there is a better place to do it.",
+      "reference": "Since it's a one-off, worth putting an info log before and after."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to pass the calendar to the controller instead of passing the token.",
+      "reference": "I'm sorry, but the indentation is still wrong. This should be indented by one tab."
+    },
+    {
+      "prediction": "I think this should be `:class:`~pandas.core.series.Series`, not `:class:`.",
+      "reference": "can we add an example?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do this in the `getExtendedIdentify` function.",
+      "reference": "We usually avoid blank lines at the start of functions."
+    },
+    {
+      "prediction": "This should be done in a separate PR as it's not related to the other changes.",
+      "reference": "I dont think this line is needed"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `server/http/server.go` since it's only used by that file.",
+      "reference": "We should not panic here !!! We return error i.e. 500 http code"
+    },
+    {
+      "prediction": "I don't think this is the right place to add this check. I think it should be in the `if self.linktype != conf.l2types.get(type(p), None):` block.",
+      "reference": "To avoid useless calls to `conf.l2types.get()` we could have a `set()` of types that are OK in this PCAP files. We could check if `p.__class__ not in self.l2_types_seen` first, only run this code block in that case, and add `self.l2_types_seen.add(p.__class__)`. I'm not sure if I'm clear here... let me know what you think."
+    },
+    {
+      "prediction": "I think this should be `EXECUTION_SOURCE_FLOW_TRIGGER = \"flow_trigger\";`",
+      "reference": "Shouldn't \"source\" be a noun? How about? -EXECUTION_SOURCE_USER -EXECUTION_SOURCE_SCHEDULE -EXECUTION_SOURCE_EVENT (because it's an event based trigger mechanism)"
+    },
+    {
+      "prediction": "I think you can remove `lvis_v1_instance.py` and `lvis_instance.py` now.",
+      "reference": "Should not switch to `lvis_v1_instance` here because that base config uses ClassBalancedDataset to oversample the data."
+    },
+    {
+      "prediction": "I'm not sure this is the right test. I think you want to test that `th` and `td` are rendered in the same order as they appear in the DOM.",
+      "reference": "Last time I checked, this technique did not work on all major screen reader - browser combinations. What screen readers have you tested?"
+    },
+    {
+      "prediction": "I think this should be `AspnetRouteTemplateResourceNamesEnabled` and not `FeatureFlagEnabled`",
+      "reference": "Most tracer settings can be set in several ways: environment variables, `app.config`/`web.config` file, or in a json file (`~/datadog.json` by default). Most exceptions to this rule are settings in native code (where we use env vars only). Currently, `DD_TRACE_ASPNET_ROUTE_TEMPLATE_RESOURCE_NAMES_ENABLED` will only work as an env var. Was this limitation intentional?"
+    },
+    {
+      "prediction": "I think this should be `NoInstanceID bool`",
+      "reference": "Would \"DisableIID\" be a better name? For example, we used DisableMetrics, ImageCleanupDisabled, ... etc. for other similar fields."
+    },
+    {
+      "prediction": "I'm not sure if this is correct. It seems to me that `total_instrs > DYNAMO_OPTION(max_bb_instrs) - 1` would be correct.",
+      "reference": "So this is a swing of 2: but I think we only need to reduce by 1? The artificial jump added on truncation is not passed to clients: it's like other mangling added later. So if `-max_bb_instrs 4` means 4 app instrs passed to clients, we'd want ==, not ==-1, right? Plus, if we did want -1 and to stop at 3 instead of 4, this is not sufficient: the inner loop above currently has `total_instrs <= cur_max_bb_instrs`."
+    },
+    {
+      "prediction": "I don't think we need to add BOOLEAN to the list of types that don't need to be converted to/from C strings. It seems like it should be converted to/from C strings just like INT, SMALLINT, LARGEINT, FLOAT, REAL, and DOUBLE PRECISION.",
+      "reference": "Does Tinyint belong to this set of types? I suppose the answer is No, but thought it safer to check."
+    },
+    {
+      "prediction": "I think this should be called `isMouseOver` instead of `isOutside`.",
+      "reference": "I'm a bit leery of this implementation, because it assumes that the only \"outside\" is to the right side. Note that Blockly assumes that there's nothing useful to the \"outside\" of the toolbox, based on the configuration of the workspace, so it's not like we're doing the right thing either. I think this will break in RTL. If you keep the editor in the same place but inject with RTL, everything you're looking for will be in the delete area. If the editor and stage swap places, this will just be looking on the wrong side."
+    },
+    {
+      "prediction": "I think we should rename this to `NatPortRange` to be consistent with the other configs.",
+      "reference": "Just spotted the `local` on here; that shouldn't be needed - no reason to limit this config to env vars only"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the extra configurations for writers. I think it should be in the `from_json` function of the `DiscoveryDataBase` class.",
+      "reference": "That's a tricky one, as the variable is indeed used in line 2346, which is a `logInfo` message that is not compiled unless `INTERNAL_DEBUG` is set or `CMAKE_BUILD_TYPE` is set to `Debug`. Our CI builds with `INTERNAL_DEBUG` to verify that all `logInfo` are correct, and thus we don't get the warning. The best is probably to remove the iterator (as you did), as well as the part of the log that uses it."
+    },
+    {
+      "prediction": "I think this should be `network_name(self.iface)`, not `network_name(conf.iface)`.",
+      "reference": "I suspect L2Socket needs a similar change (line 502) for the case where `iface` is None (line 479)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `app/models/admin/activation/activation_admin.rb` but I'm not sure.",
+      "reference": "Was this all that was needed to add CRUD operations for the LicensePermission model? :-)"
+    },
+    {
+      "prediction": "I think this should be `addon.manifest['name'].lower()` instead of `addon.isPendingRemove and bundle.name.lower()==addon.manifest['name'].lower()`",
+      "reference": "Did you look for other locations that compare name? I expected there to be more."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `org.sonar.plugins.java.api.tree.TreeTestTask`",
+      "reference": "There were some requests internally to allow users to opt out of this behaviour. We could expose a property that allows people to toggle ti"
+    },
+    {
+      "prediction": "I think you can remove the `final` from the `MockRandomMergePolicy` constructor.",
+      "reference": "Nice -- this forced `merge-on-getReader/commit` to be used more often in this test?"
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "Instead of a `bool`, can we just call this `MaxAssetDecimals` and have it be a `uint8`/`uint16`/`uint32` whose value is 0 before the upgrade and 19 after (2**64 - 1 is 20 decimal digits)? We'll need to set a maximum value anyway, and that way there's only one new proto variable instead of two."
+    },
+    {
+      "prediction": "Why do we need this check? I don't think it should be necessary.",
+      "reference": "where `$aliases` is defined ? @javiereguiluz"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add telemetry. I think it should be in the `if len(identities) > 0 {` block.",
+      "reference": "what is the justification for the removal of the registered label? it doesn't seem to have high cardinality nor is it redundant. Seems useful to shed insight into understand situations where workloads aren't registered...."
+    },
+    {
+      "prediction": "I don't think this is the right place to put this. I think we should move this to `web/web.go` since it's only used in that file.",
+      "reference": "`ctx` is unused in ListEvents"
+    },
+    {
+      "prediction": "I think this test should be moved to `SmartStoreTest` since it is not specific to SQLCipher.",
+      "reference": "We've had that test on iOS for years."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "This is for creating binlog file for building bdn generated template project. I feel that it's may be helpful in general for bdn diagnostic purpose."
+    },
+    {
+      "prediction": "I think this should be `self.send_cmd(':open -w %s' % url)` instead of `self.send_cmd(':open -w ' + url)`",
+      "reference": "Can you adjust `open_path` to simply call `path_to_url` and then `open_url` instead of duplicating the code?"
+    },
+    {
+      "prediction": "I think this should be 600 seconds, or 10 minutes.",
+      "reference": "I'm curious -- do we have information on the upper bound of the delay we would expect with high probability from a peer with no network issues? My intuition is that we want to set this as low as we can reasonably get away with before we start killing productive connections. My uninformed intuition is also that 10 seconds is probably higher than we need and I'd love to know if this is wrong and 10 seconds is already pushing the limit."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `shiftAmountInt < 0` when `shiftAmountInt > 256` and `shiftAmountInt > 256` is not a valid value for `UInt256`. I think you need to check that `shiftAmountInt <= 256` instead.",
+      "reference": "what happens if it's == 0"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I'm pretty sure it's because we're now using `Bolt::Interpreter::Powershell` instead of `Bolt::Util::WindowsPowershell`.",
+      "reference": "We could potentially log a message here instead that indicates the transport does not support `run-as` on Windows and will be ignored, just in case users expect it to and are surprised. Since it would _always_ be logged when using `--run-as` on Windows, even when the transport isn't being used, it would probably want to be either at debug or trace level."
+    },
+    {
+      "prediction": "I think this should be `github.com/bonitoo/go-sql-bigquery` instead of `github.com/bonitoo-io/go-sql-bigquery`",
+      "reference": "Did we need this alias? Or is it just a holdout from development?"
+    },
+    {
+      "prediction": "I think this should be `org.id.to_s` instead of `org.id`",
+      "reference": "This should have failed before."
+    },
+    {
+      "prediction": "I think this should be `RegexOptions.Compiled | RegexOptions.Compiled | RegexOptions.IgnoreCase | RegexOptions.CultureInvariant | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.IgnoreAlphabetic | RegexOptions.IgnoreCase | RegexOptions.Ignore",
+      "reference": "Slower construction, faster matching. I wonder if this will be noticeable in the relenv?"
+    },
+    {
+      "prediction": "This should be `std::cout << \"Pub: DATA Pub Matched \"`",
+      "reference": "I think that if `data_discovery_count_ > static_cast<int>(throughput_publisher_.subscribers_)`, then we should not proceed, since we have discovered some unexpected subscriber that can affect the test results. I'd change the comparison to `==` and have and `else if` contemplating the `>` case"
+    },
+    {
+      "prediction": "I think this should be `#include <Kokkos/Cuda/Kokkos_Cuda_Error.hpp>`.",
+      "reference": "Remove that include"
+    },
+    {
+      "prediction": "I think this should be `sprintf('%s/%s-form', $options['view'], strtolower($options['entity']))`",
+      "reference": "IMO this should be defined in the form theme instead. This class should always be there. If the user configured additional css classes, it should be appended instead of replacing the `{view}-form` css class."
+    },
+    {
+      "prediction": "Please remove this import.",
+      "reference": "These tests actually go to the server."
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere in the tests.",
+      "reference": "This test modification isn't correct. This task could run immediately and you wouldn't be able to tell if OnCompleted was called or not."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it's the only way I could come up with.",
+      "reference": "I'd be open to loosening this check if it can help offset the size."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It's not about applying a function, it's about applying a function to every element of the DataFrame.",
+      "reference": "I'm wondering which we should use, stable or latest?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this TODO. It should be in the `BftExtraDataCodec` base class.",
+      "reference": "possibly extending EncodingType enum?"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "I tested it only in Firefox (`./go //rb:firefox-test`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the Jaeger exporter.",
+      "reference": "can we do this in the baseexporter itself? So that exporters just access this.Provider.Resource, and baseexporters take care of populating Provider?"
+    },
+    {
+      "prediction": "I think we should move this to `frontend/express/public/core/events/javascripts/countly-events.models.js` and `frontend/express/public/core/events/javascripts/countly-events.views.js`",
+      "reference": "what is the difference between `/core/events/javascripts/countly.views.js` and `/core/events/javascripts/countly.events.views.js`"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to do something like: `if (!options.iframes) { options.iframes = true; }` and then do: `if (options.iframes) { options.iframes = true; }`",
+      "reference": "We should make sure this is also supported by `axe.run`. Perhaps moving it into run-rules would cover both API methods more easily?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to bump the timestamp. I think it should be in the same place as the `last_modified` field.",
+      "reference": "If the `MemoryBasedStorage` relies on a `self._bump_and_store_timestamp()`, then every child class will have to implement it. So it should not be prefixed with `_`. And should raise `NotImplementedError` etc. :)"
+    },
+    {
+      "prediction": "Why remove this? I think it's still useful to see how much memory is being used.",
+      "reference": "why removed the MemoryDiagnoser attribute?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `model/_utils.js` since it's only used by `model/_utils.js` and `model/_utils.js` should be in alphabetical order.",
+      "reference": "Why do this instead of `localFieldPath.applyGetters(doc[localField], doc)` ?"
+    },
+    {
+      "prediction": "This test should be moved to the `Isomorphism` subgraph test suite.",
+      "reference": "Just for a check. Do you really want to delete this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `consensus` since it's related to the consensus logic.",
+      "reference": "github.com/golang/protobuf/proto group in next group"
+    },
+    {
+      "prediction": "I think we should move this to the `daemon` subcommand.",
+      "reference": "Nit: what about just `pay`? Is there some other subcommand that would conflict with?"
+    },
+    {
+      "prediction": "I think this should be `0x` instead of `0x0`.",
+      "reference": "Just for the sake of keeping things logically co-located, I'd like to see this functionality in `Quantity.java` which has a lot of utility methods related to this. Maybe a `static` method like `Quantity.isValid(String string)`?"
+    },
+    {
+      "prediction": "Why do we need to make this public?",
+      "reference": "How about use `protected` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `StoreAccount.Root` function, since that is the place where the prefixAsNibbleBytes is calculated.",
+      "reference": "Should it not be `common.HashLength*2 + common.IncarnationLength` ?"
+    },
+    {
+      "prediction": "I think it would be better to move the `node.isKilledBySLA()` check inside the `if (node.getStatus() == Status.FAILED || (node.getStatus() == Status.KILLED && node.isKilledBySLA()))` block.",
+      "reference": "those change are done by save plugin."
+    },
+    {
+      "prediction": "I think it would be better to use `Charset.forName(\"UTF-8\")` instead of `\"UTF-8\"`.",
+      "reference": "There were problems with NullPointerException and with german Umlauts in message titles (I know they're not allowed in the specifiation but they caused my K-9-Inbox to not work at all) I don't know if this is related to PGP/Mime"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this interface. I think it should go in `Trace.Abstractions`, but I'm not sure if that's the best place to put it. What do you think?",
+      "reference": "This is an appsec concept right? Should it live in the Appsec namespace?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I'm not sure what the \"o\" is for, but it seems like we're trying to tell the blinkm that we're done, when really we're not.",
+      "reference": "I'd personally recommend having the demo check for error, unless it's worthless."
+    },
+    {
+      "prediction": "nit: `LOG.info(String.format(\"Firefox Session %d assigned to %s\", i, chromeSession.getUri()));` `LOG.finer(String.format(\"Firefox Session %d assigned to %s\", i, chromeSession.getUri()));`",
+      "reference": "Since this is in a test, I imagine that the choice of `info` level was deliberate."
+    },
+    {
+      "prediction": "I think we can remove the `Dashboard` and `Informer` modules now, since they are no longer used.",
+      "reference": "good call swapping these names! makes more sense this way i think"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. We need to make sure that the `replaceNode` is not `null` when we try to hydrate it.",
+      "reference": "Maybe leaving this out will save some bytes as in let `let isHydrating = replaceNode === null`"
+    },
+    {
+      "prediction": "I think this should be done in `GenerateNamespaceImports` instead of `GenNamespaceImports` here.",
+      "reference": "Let's manually create one extra whitespace line here (code_ += \"\")"
+    },
+    {
+      "prediction": "I think we can remove this function and just use `bp.total.count`",
+      "reference": "nit: copy or clone"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Document` struct, since that's where the `DocumentRevision` field is defined.",
+      "reference": "I think this can be called `HasField`, because that's what it's doing. It really is independent of revision."
+    },
+    {
+      "prediction": "I don't think you need to import `*`. You can just do `import static javaslang.API.*;`",
+      "reference": "Is this to avoid collisions?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `process_block_stats.go` since it's related to the block stats.",
+      "reference": "nit: a better name for this variable would be enableLogStats, but it's beyond the scope of your change."
+    },
+    {
+      "prediction": "I think this should be `linter = PyLinter(pylint.config.Configuration())` and then `linter.set_reporter(test_reporter)`",
+      "reference": "Because linters don't handle reports now, this was breaking the setup for _all_ tests. I deleted it so I could run my tests, but I didn't check the impact on other tests as many tests are failing at the moment."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work. I'm not sure if there is a better way to do this.",
+      "reference": "I'm not really a fan of this -- why not set up logging for `init` here (or setting a global flag) rather than doing it this way?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this, but it's the only way I could get it to work.",
+      "reference": "This is my favorite part of this PR. Being able to `import` 3rd party tools will hugely improve our workflows. :heart:"
+    },
+    {
+      "prediction": "I don't think this is the right rule to be used for this purpose. I think we need a rule that checks for the use of a static `MessageDigest` in an unsynchronized manner. I'm not sure what the best way to do this is, though.",
+      "reference": "having this extend `UnsynchronizedStaticFormatterRule` seems semantically incorrect even if it works  we should probably refactor `UnsynchronizedStaticFormatterRule` into a `UnsynchronizedStaticAccessRule` which can be configured through properties to track unsynchronized static access to any given types (with proper defaults to include `Formatter` and `MessageDigest`). This will also scale better as new types are suggested to have a thread-safe access enforced. The `UnsynchronizedStaticFormatterRule` can be marked as deprecated at the ruleset level, point to the new implementation, overriding the property defaults to keep just looking for `Formatter`. Moreover, the `THREAD_SAFE_FORMATTER` static should probably be replaced with a property of thread-safe classes to ignore, once again with proper defaults."
+    },
+    {
+      "prediction": "I think this should be `Keccak.Zero` and not `Keccak.FindBestHeaderHash()`",
+      "reference": "Keccak.Zero should not be used to mean null"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I can't think of a better way to test this.",
+      "reference": "perhaps add a couple more edge cases."
+    },
+    {
+      "prediction": "I think it's better to use `uint8` instead of `uint16` here.",
+      "reference": "Is this blank line needed, or typo?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to `prefetcher.go` since it's only used there.",
+      "reference": "This was a major PitA to find: it was causing some huge goroutine leaks and unfinished prefetches."
+    },
+    {
+      "prediction": "Line is too long. [85/80]",
+      "reference": "you can use Enum#flat_map here."
+    },
+    {
+      "prediction": "I think this should be removed as it is not used anymore.",
+      "reference": "Same potential FP: The stream is provided from outside (here as a method parameter), so we should not be responsible here to close it, should we?"
+    },
+    {
+      "prediction": "Do we need to set `force_ssl: true` here?",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should be moved to the `VerbsManager` class itself, since it's only relevant to that class.",
+      "reference": "It seems like this is the only way to access javascript `static` methods from Java."
+    },
+    {
+      "prediction": "Can we remove this file?",
+      "reference": "Why was this removed here?"
+    },
+    {
+      "prediction": "I think we can remove this function as well.",
+      "reference": "i am aware that you did not add these, but might i ask why we need these one-liner funcs instead of just exporting the struct fields in the first place?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it. I think we need to make sure that we don't project more than we need. For example, if we have a table with 10 columns, and we only need 5 of them, then we should only project those 5 columns. If we have a table with 10 columns, and we need all 10 columns, then we should project all 10 columns. I think we need to make sure that we don't project more than we need. For example, if we have a table with 10 columns, and we only need 5 of them, then we should only project those 5 columns. If we have a table with 10 columns, and we need all 10 columns, then we should project all 10 columns. I think we need to make sure that we don't project more than we need. For example, if we have a table with 10 columns, and we only need 5 of them, then we should only project those 5 columns. If we have a table with 10 columns, and we need all 10 columns, then we should project all 10 columns. I think",
+      "reference": "I agree with this because it is the opposite of `GetProjectedIds` used above."
+    },
+    {
+      "prediction": "I think this should be `ebsOptimized`, not `ebsOptimized`, `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or `ebsOptimized`, or",
+      "reference": "add 'sriovNetSupport' also to this list"
+    },
+    {
+      "prediction": "I think this should be `if (accountsInOrder.contains(newAccount)) {` instead of `if (!accountsInOrder.contains(newAccount)) {`",
+      "reference": "Code style issue: `if` body is not wrapped in braces."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we should add all the podcasts to the adapter and then call `adapter.notifyDataSetChanged()` to update the UI.",
+      "reference": "Call requires API level 11 (we are on 10)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way I could get it to work.",
+      "reference": "do we need to pass the DB if it is the same"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it's better to do it in `makeChainForOp`.",
+      "reference": "Might be a good idea to rename the variable so it doesn't shadow the type."
+    },
+    {
+      "prediction": "I think this should be `UNION ALL SELECT ...` instead of `WITH RECURSIVE`",
+      "reference": "Again I'd prefer if this is done in JavaScript rather than in SQL."
+    },
+    {
+      "prediction": "I think it would be better to say \"Refer to the `./promise` module documentation for a detailed explanation of how the control flow coordinates task execution.\"",
+      "reference": "there is still `2` spaces after `a` here.. :P"
+    },
+    {
+      "prediction": "I think this should be called `needsSetupRootfs` instead of `setupRootfs` to be consistent with the other functions in this file.",
+      "reference": "What we did in docker before, when working directly with libcontainer, is check if the source is `/dev` and the destination has a `/dev/` prefix. This implementation is probably ok. If someone is mounting their own `/dev` I would expect it to be pre-configured and not need the extra check for if it's the host's `/dev`. (that said, I've not tested yet, just reading through the code)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `config/config.go` since it's specific to the `config` package.",
+      "reference": "you should add some help on the struct too so that `plz halp featureflags` has a bit of explanation. Might be worth mentioning that these generally won't remain around very long relative to other config options (i.e. they're typically enabling \"preview\" features for the next major)."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be in `services/identity.go`",
+      "reference": "Maybe `consumerID, receiverID, issuerID`, at least it's a convension in overal repo"
+    },
+    {
+      "prediction": "I think it would be better to have a `s3_open` and a `s3_open_with_region` function. The `s3_open_with_region` function could take a `s3_region` parameter. The `s3_open` and `s3_open_with_region` functions could have the same signature as the `s3.Open` function. The `s3_open` and `s3_open_with_region` functions could have the same signature as the `s3.Object.Open` function.",
+      "reference": "Where does the auth go?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this check. It seems like it should be in the `calculateIntrinsicGas` function, but I'm not sure if that's the right place to put it.",
+      "reference": "0 < minimum gas price, `ErrUnderpriced` is more proper"
+    },
+    {
+      "prediction": "Shouldn't this be `TOKEN_EXPIRE_MINUTES`?",
+      "reference": "these should be alphabetized"
+    },
+    {
+      "prediction": "nit: I think it would be better to pass the type width as a parameter to `readBatchOfDictionaryEncodedLongs` instead of hardcoding it.",
+      "reference": "These changes look concerning. It looks like the old offset (only index) must not have been correct. If so, there are places where `getDataBuffer().setLong(...)` and similar methods are called but aren't updated like these. Are those cases bugs as well?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to flush the counters. I think it should be done in the `Flush()` function in `pkg/monitoring/prometheus/counter.go`",
+      "reference": "the CAS call here is now wrong since dirty will never be 3 anymore. it might be useful to sweep through the entire usage of this field to see that everything is correct"
+    },
+    {
+      "prediction": "Why do you need to use BaseModule and BaseCountry? Why not just use BaseModuleQuery and BaseCountryQuery?",
+      "reference": "Base model is imported here"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "import but not used"
+    },
+    {
+      "prediction": "I think this is the only place where `typeWidth` is used. I think it should be removed.",
+      "reference": "What about the call in `case PACKED` just below? Does that also need to use the `typeWidth`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right size. I think it should be 1/3 of the size of the serialized data.",
+      "reference": "not sure i like this change. The encryption package does not need to know about references"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this check. I think it should be in the `else ifeq ($(llvm_version),7.0.1)` block.",
+      "reference": "This shouldn't be added. It's not supported. It's not event the default. If we are going to support this, we need to have CI for it."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "omg only four left!!!!"
+    },
+    {
+      "prediction": "I think this should be `err = fmt.Errorf(\"Unknown consensus protocol %s\", proto.ConsensusCurrentVersion)`",
+      "reference": "I'm not really sure how this would happen. once the binary is already compiled, the config.Consensus should already have the entry for protocol.ConsensusCurrentVersion."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to make sure that `--local-scheduler` is passed to `luigi.run` even if `--no-lock` is passed to `--local-scheduler`.",
+      "reference": "Is there a reason for the order swap here?"
+    },
+    {
+      "prediction": "I think this should be `throw new FileNotFoundException(Resources.NoDotnetDotExeFileExist, dotnetExeName);`",
+      "reference": "Where will this exception get caught ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right name for this. It's not a backend, it's an identity registry. The backend is the thing that actually connects to the ethereum network.",
+      "reference": "Why this field is made public?"
+    },
+    {
+      "prediction": "(style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vmultiplier'? (style) 'vmultiplier' may be misspelled - perhaps 'vm",
+      "reference": "Is the `vmultiplier` only used here for checking if a grid value is nodata?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this change. I think it should be in `pkg/test` since it's a test.",
+      "reference": "Is this deliberate? If so, why?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good place to put this, but I think it's a good place to put it.",
+      "reference": "both URL works. I corrected it anyway."
+    },
+    {
+      "prediction": "I think it would be better to have a separate class for the reactive advice. It would make it easier to see what is reactive and what is non-reactive. Also, it would make it easier to see what is reactive and what is non-reactive in the code.",
+      "reference": "Rather than have another member and and if/else statement, if the user passes in `PreInvocationAuthorizationAdvice` it could be adapted to match `PreInvocationAuthorizationReactiveAdvice`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work. I'm not sure if there's a better way to do this.",
+      "reference": "The callback get the log level as a number."
+    },
+    {
+      "prediction": "I think we can remove this now.",
+      "reference": "Is there any issue with 2.2.3? If not this should be `<2.3`."
+    },
+    {
+      "prediction": "I think you can remove this line too.",
+      "reference": "Shall we keep this test?"
+    },
+    {
+      "prediction": "I think this should be `if (\\class_exists(' TYPO3\\\\CMS\\\\Backend\\\\Form\\\\Element\\\\SelectElement'))`",
+      "reference": "When would this be the case? / When would SelectElement not be available? Is this change compatible with 6.2? (I'd like to roll a release that works with 6.2+ first, before moving on to 7.x)"
+    },
+    {
+      "prediction": "I think this should be `public boolean isSameCatalog() { return false; }`",
+      "reference": "how about rename to shouldNarrowDataTypeTo?"
+    },
+    {
+      "prediction": "I'm not sure this is the right solution. If there's only one renderer, then we don't need to do anything. If there's more than one, then we need to set the current backend.",
+      "reference": "Why not check that the one available renderer is 'bokeh'?"
+    },
+    {
+      "prediction": "Should this be `bundler exceeded buffered byte limit` or `bundler reached buffered byte limit`?",
+      "reference": "Can you import `bundler.ErrOverflow` rather than redefine it here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we need to keep the `resourceNameType` as a field in `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and `ResourceType` and",
+      "reference": "This is just a renaming of the function getResourceType() to getResourceReference()"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Seems like this might need to allow for `\"`s to be used here?"
+    },
+    {
+      "prediction": "`Combat* combat = g_luaEnvironment.getCombatObject(g_luaEnvironment.lastCombatId); return combat != nullptr;`",
+      "reference": "here you assigning it to a local variable, in the original code it assigned to combat(CombatSpell member). Not sure if CombatSpell needs to use a shared_ptr since currently it does not and it works \"good\"."
+    },
+    {
+      "prediction": "I think we should keep the `p` tag here, since it's a part of the content.",
+      "reference": "Should we also remove this? This applies to subscriptions in general, but it seems sort of weird in the context of products now."
+    },
+    {
+      "prediction": "I think this should be `from .alert import make_email_on_pipeline_failure_sensor, make_email_on_run_failure_sensor`",
+      "reference": "I think we should keep these, o.w. our examples will be wrong (where we're importing from `dagster.utils`). Also, we might break folks who've imported following our docs."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in the SDK, but I'm not sure if that's the right place to put it.",
+      "reference": "This is good and something that I think opens a lot of possibilities. I think what would be even better is to change up `TracerProviderSdk` to take factories for the processors just like it already does for the instrumentations. Should probably use use a factory for the sampler. If we made those 2 changes then we could actually split this all up into the config and the build phase and we could use the factory methods in the hosting version to call into the IServiceProvider to get those instances."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think we should be able to just do `t.KeyValues[j] = values.New(t.KeyValues[j])` instead of `v = values.New(t.KeyValues[j])`",
+      "reference": "Is there a way to make it so `values.New(nil)` works instead of adding a new function?"
+    },
+    {
+      "prediction": "I think this should be `method_exists('Symfony\\Component\\Form\\AbstractType', 'getBlockPrefix')`",
+      "reference": "> when requirement of Symfony **will be** >= 2.8 Same for others."
+    },
+    {
+      "prediction": "I think this should be `roles/logging.logWriter,roles/iam.serviceAccountTokenCreator`",
+      "reference": "I'd like to see if this can just be set on the service account instead of the project. Using the SVC_ACCT_ROLES (which should otherwise be deleted as I don't think anything else is using it.)"
+    },
+    {
+      "prediction": "Please revert this change.",
+      "reference": "Please remove the unused imports :)"
+    },
+    {
+      "prediction": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Were these local edits? There doesn't appear to be a change here. If so, let's remove this from the commit."
+    },
+    {
+      "prediction": "Why is this needed? I don't think it is.",
+      "reference": "nit: this should be further down w/ the rest of the github imports"
+    },
+    {
+      "prediction": "I think it would be better to use `number_to_currency(100)` instead of `100`",
+      "reference": "I don't know if this is something we should address now, but we have a cents_to_dollars private method in both `SubscriptionCoupon` and `Invoice`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this check. I think it would be better to do it in the `hipSyncAndResolveStream` function.",
+      "reference": "Formatting here and all below: missing space before open bracket for the compound statement."
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "Avoid using old mortbay utilities for converting JSON. Uses existing Solr Utils to convert from JSON string."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. It should be in `command_v2.js` I think.",
+      "reference": "I actually think this wrapping is great, but let's call it `RunCommandOperation`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `errorsParse` since it's only used there.",
+      "reference": "Maybe it is better to add checking for status code duplications? It can prevent some stupid copy-paste user errors."
+    },
+    {
+      "prediction": "Metrics/BlockLength: Block has too many lines. [20/25]",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure what this is supposed to do, but it seems like it should be changed to `instr_get_src(instr, 0)`.",
+      "reference": "This lea should match instr_check_xsp_mangling and be undone. I don't think i#3307 applies here."
+    },
+    {
+      "prediction": "nit: I think it would be nice to have a `Put` method that takes an `ipldstate.Thing` and returns an `ipldstate.Thing` and a `error`.",
+      "reference": "I like the idea of this being an interface as it makes the test setup a bit easier, thoughts?"
+    },
+    {
+      "prediction": "`pubkey_hexes` -> `pubkeys`",
+      "reference": "Nit: I believe the electrum convention is to use `bfh` instead of `bytes.fromhex`. Great catch on this bug!"
+    },
+    {
+      "prediction": "I think this should be `raise Exception(\"Predicted field '%s' does not exist in included fields.\" % predictedField)`",
+      "reference": "Use a more granular error type. I would recommend `ValueError` in this case."
+    },
+    {
+      "prediction": "This should be `if (g_events->eventPlayerOnTradeRequest(player, tradePartner, tradeItem))`",
+      "reference": "A space after each argument, and an empty line above this if-statement."
+    },
+    {
+      "prediction": "I think we can remove `displayName` and `name` from `Memoed.prototype` since they're no longer used.",
+      "reference": "Things like this make me wonder whether we should ship a \"development\" and a \"production\" version. I remember the discussions about making things more complex bundling wise but couldn't we omit this line in a \"production\" or \"minimal\" variant of our build? That would save us some bytes here and there. Or are we somehow able to move this line to debug? I though about monkey patching the exported function but that only works in CJS and not ESM AFAIK."
+    },
+    {
+      "prediction": "I think this should be `ISpanListener` instead of `IScopeListener`.",
+      "reference": "Maybe rename to `RegisterListener` or `RegisterSubscriber`?"
+    },
+    {
+      "prediction": "I think we can remove the `|| EMPTY_ARR` from the `oldParentVNode!=null &&` and `oldParentVNode != EMPTY_OBJ` checks below.",
+      "reference": "My one concern here is that we are adding a new array allocation for each depth of the tree which could add some non-trivial GC churn. Perhaps this would be okay if we think that refs are used through a tree commonly enough such that they'll get allocated anyway but I'm not sure refs are *that* often used. Every app has a couple, but probably not at every level of their app :/ So maybe the extra bytes are worth it in this case? Though I have no data to back up my hypothesis so I could way off lol."
+    },
+    {
+      "prediction": "Line is too long. [191/180]",
+      "reference": "We should get rid of this big array and think about a better way to check this."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in the `OnRequestNavigate` method of the `FragmentHost`",
+      "reference": "Should we cache the result of `Mvx.Resolve<IMvxChildViewModelCache>()` to avoid a lookup each time?"
+    },
+    {
+      "prediction": "I think it would be better to move this logic to the `init_rabbitmq_channels` method.",
+      "reference": "If you get to this line avg_size_of_message is an undefined value, yet you use it below. You you should set this value to something in the exception block."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `server/mux.go`",
+      "reference": "let's just fold this inline in mux.go, no need for the indirection to this new file"
+    },
+    {
+      "prediction": "I think this import is unused.",
+      "reference": "This bit confused me just now as I was reading through -- turns out this is due to the `{@link}` on line 316. Reading up on this, the alternative seems to be to state the fully qualified name instead. Happy to edit if requested as I don't have a strong view either way (perhaps a small leaning towards keeping the import)."
+    },
+    {
+      "prediction": "I think you can remove the `PMD.EOL` and `PMD.LF` constants.",
+      "reference": "There's a unnecessary System.out."
+    },
+    {
+      "prediction": "I think this should be `if c.enableProxy && c.encapMode.SupportsEncap() {`",
+      "reference": "Should we add the flows only when NodePort is enabled?"
+    },
+    {
+      "prediction": "I think this should be `release_version` and not `release_version_no_prs`.",
+      "reference": "ITYM to flip this one..."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "What's the difference between typescript-eslint/eslint-recommended and typescript-eslint/recommended? I really can't get it."
+    },
+    {
+      "prediction": "Shouldn't this default to 30s?",
+      "reference": "Thinking: Maybe it would be more useful to define this in terms of RTTs. That would also make it easier to switch this to a bool, since we could then pick a reasonable default value. Unless of course an attacker could influence our RTT estimate. Is that the case?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Why is this line deleted?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that we don't know how many blocks we need to confirm, because we don't know how many blocks we have confirmed already.",
+      "reference": "minor type `boostrap`"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Nit: please don't separate imports into groups."
+    },
+    {
+      "prediction": "Shouldn't this be `getControl().getSelectedElement().ifPresent(getControl().getOnAllCategorySelection::run)`?",
+      "reference": "Thinking if it would be nice to have something like `setNothingSelected`."
+    },
+    {
+      "prediction": "I think this should be `f.Log.V(logf.DebugLevel).Debug(\"detected private key or certificate data on disk has changed. reloading certificate\")`",
+      "reference": "I think this is a useful InfoLevel message. In production, if webhook client connections suddenly start failing due to change of serving cert, you'd want to see this information in the logs."
+    },
+    {
+      "prediction": "I think this should be `std::string` instead of `folly::to<std::string>(tagId)`",
+      "reference": "Generally, if cast from int to char*, you could cast directly. Because no errors should happen."
+    },
+    {
+      "prediction": "I think this should be 1000*60*60*10, not 1000*60*10.",
+      "reference": "Is 10 min too long? What do you think about 1 minute? Unless there are many jobs in this state, I don't expect the CPU overhead to be too high."
+    },
+    {
+      "prediction": "I think this should be renamed to `LoggingConfig` or `LoggingConfigMap`",
+      "reference": "can you remind me why we keep the pointer if it's a `map[string]<PStruct>`? are there other scenarios where the pointer is kept?"
+    },
+    {
+      "prediction": "Shouldn't this be `RewriteDataFilesAction maxParallelism(int parallelism) { ... }`?",
+      "reference": "Would it make sense to also add `setMaxParallelism` in addition to this to match the Flink API?"
+    },
+    {
+      "prediction": "I think we can move this to the `BillingModal` component, since it's only used by that component.",
+      "reference": "Not needed for actions. `_super(...arguments)` is only needed when you're extending from a base class and want to run the logic in the base class before your own logic"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "I still need to run this code to verify my understanding, but my first reading through makes me think this may need to be `suspendingComponent._vnode` since we set `vnode._hydrating` on the vnode that threw the error (diff/index.js:275), not on `Suspense._vnode` (though we could in Suspense's `_catchError` implementation at the top of this file)."
+    },
+    {
+      "prediction": "I think this should be `if path is not None and os.path.exists(path):`",
+      "reference": "I think it's a good idea to log this as debug, the same way the other branch gets logged."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `add_annotation` returns an `Annotation` object, not a `AnnotationNote` object, so we can't compare the `id` of the two.",
+      "reference": "1:D100: Docstring missing 35:D101: Docstring missing 37:D102: Docstring missing 42:D101: Docstring missing 44:D102: Docstring missing 57:D102: Docstring missing 66:D101: Docstring missing 69:D102: Docstring missing"
+    },
+    {
+      "prediction": "Why is this constant used here? I don't see it being used anywhere else.",
+      "reference": "keeping this as an unexported constant, as importing the role pkg creates a cyclic dependency."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `encodeValue` function.",
+      "reference": "Why `c.RevisionField()` here but `c.opts.RevisionField` just below?"
+    },
+    {
+      "prediction": "I think we should remove the `OvmExecutionManager` and `OvmStateManager` fields from the `Context` struct, since they are no longer used.",
+      "reference": "nit: this file diff seems unrelated and probably should have been a separate PR"
+    },
+    {
+      "prediction": "I think this should be `bool *` instead of `bool*`.",
+      "reference": "@sjplimp just checked this API in the latte repository master branch (which is what Install.py downloads) and this still does not provide the 18th argument. We cannot merge this pull request until this is available. i would also suggest to implement a second binding, a function called latte_abiversion() returning an int with an ABI version number (starting with 1). This way, it can be tested, whether the ABI expected by the LAMMPS interface and provided by the library is compatible *before* calling the `latte()` function and risking a segmentation fault."
+    },
+    {
+      "prediction": "I don't think we should log the nonce here, since it's not related to the transaction itself.",
+      "reference": "Is this from debugging?"
+    },
+    {
+      "prediction": "I think it would be better to use `assertNotFalse` instead of `assertFalse`",
+      "reference": "Its betet use MockedObject"
+    },
+    {
+      "prediction": "nit: I think we can remove the `DeleteByTLF` method and just call `DeleteByTLF(ctx, tlfID, blockIDs)` directly.",
+      "reference": "\"including\"? Since it doesn't take any parameters, I'm not sure how it can update anything else..."
+    },
+    {
+      "prediction": "\"Azure Resource Management only supports HTTPS scheme.\" -> \"Azure Resource Management only supports HTTP/HTTPS scheme.\"",
+      "reference": "nit: I'm assuming you have verified this is what the error message should read like"
+    },
+    {
+      "prediction": "This should be `sys.exit(0)` instead of `exit(0)`.",
+      "reference": "Why do you need that? (real question)"
+    },
+    {
+      "prediction": "I think this should be replaced with `vStringPut (Signature, x);`",
+      "reference": "If I understand the patch correctly the `else` is no longer required."
+    },
+    {
+      "prediction": "I think it would be better to have a `BlacklistType` enum instead of a string.",
+      "reference": "Properly describe valid flag values here: `neither` should be replaced with `none` as they parsed below."
+    },
+    {
+      "prediction": "I think it would be better to use `logger.error(String.format(\"Exception when removing executor(host: %s, port: %s) entry from database...\", host, port), ex);` instead of `logger.error(String.format(\"Exception when removing executor(host: %s, port: %s)\", host, port), ex);`",
+      "reference": "Can we escalate this to `warn`. It is a major event."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it's the only place I could think of that would work.",
+      "reference": "Please move these up to line 86 so the numbers are in order."
+    },
+    {
+      "prediction": "`train` and `test` are not used in NAS-COS, please remove them.",
+      "reference": "Add a docstring to contain the paper link."
+    },
+    {
+      "prediction": "I think this should be `targ` instead of `targ!=\"\"`",
+      "reference": "Instead of adding the `--target` flag here, the `Target` struct should be set up correctly. Assuming this is for the Raspberry Pi 3, adding it to the `cflags` key of the JSON file should be enough (if not, you can print `config.CFlags()` here to check whether `--target` is already included)."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should be doing this in the `createFederatedBundle` function.",
+      "reference": "The changes in this file look unnecessary since `rpccontext.AuditRPCWithTypesStatus()` already does this check. Was there a reason you needed to add them?"
+    },
+    {
+      "prediction": "I'm not sure this is a good name for this. It's not a key, it's a value. I think it should be something like `BatchStateKey` or `BatchKey` or something like that.",
+      "reference": "you could probably use `const` for both?"
+    },
+    {
+      "prediction": "I'm not sure this is the best place to put this. I think it would be better to move it to `dom.js` since it's only used in that file.",
+      "reference": "This looks like an \"ES Module\" to me. I do not understand this TODO."
+    },
+    {
+      "prediction": "I think this should be `if (spec.isUnpartitioned()) {` instead of `if (spec.fields().isEmpty()) {`",
+      "reference": "Iceberg doesn't use `final` because it is unlikely that this actually helps. In Java 8, final detection is quite good, which is why non-final variables can be used in closures and lambdas. And final doesn't produce different byte code so it can't do much to help at runtime."
+    },
+    {
+      "prediction": "`to_string()` is not a function of `Player`, so you can't call it here. You need to call `player->getIP().asString()` instead.",
+      "reference": "There's `pushString` to use with C++ strings."
+    },
+    {
+      "prediction": "I think this should be `0.5s` instead of `8s`.",
+      "reference": "qq: Should this include the iptables6-save sim. to the inverse checks above?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think this should be done in the `create_writer` function. The `async_thread` should wake up the appropriate writers.",
+      "reference": "What about async user writers? We could be checking `param.mode == ASYNCHRONOUS_WRITER`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this rule. I think it would be better to move it to `org.infinispan.commons.util` or `org.infinispan.util` or `org.infinispan.util` or `org.infinispan.util.concurrent` or `org.infinispan.util.concurrent.atomic` or `org.infinispan.util.concurrent.locks` or `org.infinispan.util.logging` or `org.infinispan.util.logging.slf4j` or `org.infinispan.util.logging.slf4j.api` or `org.infinispan.util.logging.slf4j.api.trace` or `org.infinispan.util.logging.slf4j.api.trace.warn` or `org.infinispan.util.logging.slf4j.api.trace.warns` or `org.infinispan.util.logging.slf4j.api.trace.warns.warn` or `org.infinispan.util.logging.slf4j.api.trace.warns.warns` or `org.infinispan.util.logging.slf4j.api.trace.warns.warns.warn` or `org.infinispan.util.logging.slf4j.api.trace.warns.warns.warns.warns` or `org.infinispan.util",
+      "reference": "this double check for lombok annotation + specific lombok annotation seems pointless, just keep the second (specific) check. Moreover, since we have a property with ignored annotations... why don't we use it? a user may setup the property and it will be ignored here. Also, this class should probably not extend `AbstractLombokAwareRule` anymore. We have a configurable set of annotations that suppress the warning, defaulting to ALL lombok annotations, yet we only check for `lombok.NoArgsConstructor`... Maybe directly extending `AbstractIgnoredAnnotRule` and settig the default property to `lombok.NoArgsConstructor`? Or, to avoid a breaking API change, keep extending it, but override the default. For 7.0.0 we can change the class hierarchy."
+    },
+    {
+      "prediction": "I think it would be better to have a separate function for adding interceptors, so that we don't have to pass in an empty array if we want to add multiple interceptors.",
+      "reference": "nit: the ordering may be misunderstood, as \"invoked in the same order\" could suggest the passed-in functions are called in-order, but the actual wrapping is LIFO. Some other ways to describe it (don't think any of these are ideal, but maybe it will help you come up with something better), * the handled returned by the first interceptor is executed first * interceptors are applied outside-in (don't like the terms outside/in though) * interceptors are applied in LIFO order, leading to an earlier interceptor's handler being executed before latter interceptor handlers"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. It seems like it should be in the same package as the other components.",
+      "reference": "We should not use a static import to be consistent with other existing use of the QueryElevationParams."
+    },
+    {
+      "prediction": "(style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (style) 'config' may be misspelled - perhaps 'configuration'? (",
+      "reference": "No need to check return value?"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "~~Feel free to ignore that one, I'll fix things up when regenerating the `requirements.txt`.~~ As for the one below, this smells like a pylint bug..."
+    },
+    {
+      "prediction": "I think this should be `len(antreaSourceStatsElementList) + len(antreaDestinationStatsElementList) + len(antreaLabelsElementList) + 1`",
+      "reference": "it's a bit strange that this doesn't match `aggregationElements` which is what I would expect. I guess I am not familiar enough with go-ipfix."
+    },
+    {
+      "prediction": "(style) code indent should use tabs where possible (style) please, no space before tabs",
+      "reference": "Same thing: `num_simd_registers`. Ditto below."
+    },
+    {
+      "prediction": "I think this should be `version='2.2.1.dev0'` instead of `version='2.2.1'`",
+      "reference": "Should be 2.1.1"
+    },
+    {
+      "prediction": "Should this be `required_sqs_topic_name` instead of `required_subject`?",
+      "reference": "nitpick: `required_subject` doesn't seem to be used here"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "We can revert this to reduce the diff of the PR."
+    },
+    {
+      "prediction": "I think this should be `self._get_presentationType()`",
+      "reference": "Extraneous blank line."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Is this the same import as in the line 20?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like we should be able to do it in the same way we do for the other published templates, which is to set the `published` attribute to `true` and then call `reconcile_published` on the `template` object.",
+      "reference": "why those additional checks after published? template.version should always be present and greater than zero"
+    },
+    {
+      "prediction": "I think we can remove the Pod and ExternalEntity fields from the EntityReference struct and just use the Pod and ExternalEntity fields from the PodReference and ExternalEntityReference structs.",
+      "reference": "I did not see this is used in the controlplane API? Is it for internal use only? Then no need to define it here?"
+    },
+    {
+      "prediction": "I think this should be `gcnArchName` instead of `gcnArch`.",
+      "reference": "Should we also add a line to print gcnArch?"
+    },
+    {
+      "prediction": "`Automatically sends usage statistics and crash reports to Brave` -> `Automatically sends usage statistics and crash reports to Brave Software Inc.`",
+      "reference": "I merged already so need another PR, but I think this needs to be at the bottom."
+    },
+    {
+      "prediction": "I think this should be renamed to `accountOwnedByHouse`.",
+      "reference": "space in key string?"
+    },
+    {
+      "prediction": "I think this should be `cmd = \"nsenter\" + b.localMnt`",
+      "reference": "The commends of this function should be updated"
+    },
+    {
+      "prediction": "Why do we need to set `reg_decoded_bbox` to `False` here?",
+      "reference": "The docstring is outdated."
+    },
+    {
+      "prediction": "I'm not sure if we want to expose these in the public API. I think it would be better if we only expose them in the `TraceListener` and `SpanListener` classes.",
+      "reference": "Could rename to `ServiceVersionKey` for consistency with the suggestion to rename the TracerSetting."
+    },
+    {
+      "prediction": "I think this should be `print(\"Overwriting {owner}/{pkg} ({version})\".format(owner=owner, pkg=pkg, version=version))`",
+      "reference": "I seem to recall a UI issue with this... @akarve I think you didn't like this for some reason..."
+    },
+    {
+      "prediction": "I think this should be `setDefault` instead of `setDefault`",
+      "reference": "The `MediaBlockService` commit must be remove"
+    },
+    {
+      "prediction": "I think this test is a bit too specific to this PR. Can you move it to a separate PR?",
+      "reference": "Use a literal constant here, and use `assertEqual`, e.g. `self.assertEqual(2, len(...))` You should also check that the number of rules is correct, i.e. 1 rule for buckets, 0 rules for tables."
+    },
+    {
+      "prediction": "I don't think this is a breaking change.",
+      "reference": "@Murderlon Should it just say `some future version`?"
+    },
+    {
+      "prediction": "I think we can remove the `cspcGot` and `cspcObj` variables now.",
+      "reference": "change looks good.. but, better to make populateVersion to return same object in the case of error.. that avoids lot of complex logic and probable issues"
+    },
+    {
+      "prediction": "I think it would be better to return an error in this case.",
+      "reference": "invalid operation: exp (variable of type *Service) has no field or method getAddressDetails (from `typecheck`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think we should move this to the background flusher itself.",
+      "reference": "Probably overkill, but maybe we should be using a fake clock here. I won't push on whatever you decide though."
+    },
+    {
+      "prediction": "I'm not sure why we need to do this in the main goroutine. It seems like we can just do this in the same goroutine as `t`, right?",
+      "reference": "If you feel like it, may as well apply all the suggestions below to this test too."
+    },
+    {
+      "prediction": "I think this should be `newPointQuery(field, queryRelation, minLatitude, maxLatitude, minLongitude, maxLongitude)`",
+      "reference": "do we really need a relation or can we just assume INTERSECTS?"
+    },
+    {
+      "prediction": "I think this class is not needed anymore.",
+      "reference": "Is this a `MetricKey` ?"
+    },
+    {
+      "prediction": "I think it would be better to use `java.time` instead of `java.util` for this.",
+      "reference": "what's your rationale to pick up this date?"
+    },
+    {
+      "prediction": "I think this should be `fscommon` instead of `cgroups`?",
+      "reference": "NIT: I'd have preferred that the new package be called \"utils\", but that's not a big deal."
+    },
+    {
+      "prediction": "Nit: I think it's better to use `AWS::KMS` here instead of `Amazon::S3` to be consistent with the rest of the code.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be done in the `deleteByQuery` method of the `IndexQueueItem` class.",
+      "reference": "I think, retrieving the setting could also be done outside the loop. What do you think?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do it. I think it would be better to do it in the `TransactionPool` itself.",
+      "reference": "This TODO isn't related to this change. We should remove it."
+    },
+    {
+      "prediction": "I think this should be `logger.info(\"Using delegate supplier: \" + clazz.getName());`",
+      "reference": "We chose `info` in the test code for obvious reasons. Changing to `finest` makes debugging harder and noisier."
+    },
+    {
+      "prediction": "I think this should be `if subscript.as_string() == subscript.parent.value.as_string()`.",
+      "reference": "I'm not really sure this is worth it. Keep in mind that every special case we add has the potential to introduce new errors and complicates the code further. For common cases that is acceptable, but in this instance I don't think it's beneficial."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in the `agent` package.",
+      "reference": "nit: this `if` isn't necessary since we immediately follow it up with a range over the slice, which works fine with a `nil` slice."
+    },
+    {
+      "prediction": "This should be `if (player->getSex() == PLAYER_SEX_MALE || player->getSex() == PLAYER_SEX_FEMALE) {`",
+      "reference": "Remove this line please."
+    },
+    {
+      "prediction": "I think this map should be a `ConcurrentHashMap` instead of a `Map` to avoid concurrent modification issues.",
+      "reference": "why is this is class field? it should be something local within the auto apply template method since we only need this once to process templates and never use again."
+    },
+    {
+      "prediction": "I think this should be `getStandards().ariaAttrs` instead of `axe.utils.getStandards().ariaRoles`",
+      "reference": "I agree with Stephen it's better to invoke getStandards() only once."
+    },
+    {
+      "prediction": "I think this should be `irregular_dims=None` instead of `irregular_dims=()`.",
+      "reference": "After discussing what ``irregular_dims`` really is, we agreed that we need a better name that makes it clearer that this is more of an xarray concept of irregular dimensions than a holoviews one."
+    },
+    {
+      "prediction": "This should be `LOG.debug`",
+      "reference": "This code is new and not tested well. While we may drop the log level before we ship 4.0, right now this is extremely helpful to users."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "this makes sure there is some padding space around the size column"
+    },
+    {
+      "prediction": "I think this should be moved to the top of the file, as it is the only place where it is used.",
+      "reference": "Put these includes in alphabetical order"
+    },
+    {
+      "prediction": "I think this should be `isinstance(parent, JAB) and parent.role == controlTypes.ROLE_MENUBAR` instead of `if not binding.modifiers and self.role != controlTypes.ROLE_MENUITEM:`",
+      "reference": "Can this just be an else?"
+    },
+    {
+      "prediction": "`logger = get_root_logger()` should be moved to the top of the file.",
+      "reference": "Merge this line with Line7"
+    },
+    {
+      "prediction": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "`clearfix` is needed to stop the `float:right` content overflowing the list item container."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "Can this be put on the assembly?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `HealthCheckResult` class that has a boolean value for `isAlwaysOk` and a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has a `HealthCheckResult` object that has",
+      "reference": "Could we rename the parameter to `overrideStatusCode`? The name `alwaysOk` implies that the health check result would be always healthy."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it would be better to have a `getSchedPolicy` and a `putSchedPolicy` and to have a `deleteSchedPolicy` that takes the name of the policy as an argument.",
+      "reference": "\"/schedpolicy\" is repeated, i would make it a constant and use it here and cluster client.go"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `RSpec::Core::Ordering::BuiltIn`.",
+      "reference": "Having a second `configuration` arg seems kinda odd to me as an interface. It looks like you're just using it as a form of dependency injection for the tests, right? Part of what makes it seem weird is that it's leaked into all the other orderers where they don't use `configuration` at all. Instead, what do you think about accepting the configuration in `initialize` (and having it default to `RSpec.configuration` in the same way)? To me, the config feels like a bit of state the instance should hold onto rather than being given each time `order` is called. It also means the other orderers wouldn't need to take the odd second argument. Thoughts?"
+    },
+    {
+      "prediction": "I think this should be `if not globalVars.appArgs.secure or globalVars.appArgs.forceInstall:` instead of `if not globalVars.appArgs.secure and not globalVars.appArgs.forceInstall:`",
+      "reference": "This button should also be disabled when in secure mode since it opens a web browser from which you can easily do a lot of insecure stuff."
+    },
+    {
+      "prediction": "I think this should be `ExpireTableMetadata` and not `ExpireTableMetadataAPI`.",
+      "reference": "Should this be done as part of `expireSnapshots`? I'd like to avoid adding a lot of operations to `Table` because it is already a large API."
+    },
+    {
+      "prediction": "I think this should be `NewFullClose` to match the rest of the libp2p package.",
+      "reference": "I do not think that this type (and its constructor) have to be exported."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in the same place as the `Build` function.",
+      "reference": "why add this? don't think it's correct? for instance, it returns ErrCandidateNotExist (not critical), in this case we should return a receipt with corresponding status"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but it seems like we should be able to get rid of the `acme` package and just use `github.com/xenolf/lego/acme`.",
+      "reference": "Instead of creating a new file, put this in tls.go, like right after or before DNSProviderConstructor is defined."
+    },
+    {
+      "prediction": "I think this should be `oc.getLanguage()` instead of `OC.getLocale()`",
+      "reference": "Instead of OC.getLocale() for the default value the default should be the value of the user value 'core' 'lang'. When loading the page this can be added to the parameters by retrieving $this->config->getUserValue($user->getUID(), 'core', 'lang'); in viewcontroller.php."
+    },
+    {
+      "prediction": "This is not a good change. You are changing the interface of an object that is used by many other objects.",
+      "reference": "Why interface1 ? It should be in inner so."
+    },
+    {
+      "prediction": "I don't think this is necessary. The `to_dict` method already returns a dictionary with all the fields we need.",
+      "reference": "Ideally, I'd have converted dicts to object but this way was less changes so this way for now."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the same place as the `return` statement.",
+      "reference": "Why is that test different to line 4022 (allocate is not tested there)? Shouldn't that be consistent?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this.",
+      "reference": "Should we consider move this expression evaluation to SETUP_INSERT step since it evaluates from queue entry. Also, this expression is not evaluated in ExHbaseAccessVsbbUpsertTcb."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `OpenTelemetrySpan` class, since that's where the `SpanKind` and `SpanId` enums are defined.",
+      "reference": "Don't see any more TODOs here.."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to detect the name of the image. I think we should be able to get this from the docker image name, but I'm not sure how to do that.",
+      "reference": "Can you extract this out to a constant?"
+    },
+    {
+      "prediction": "I'm not sure about this change. It seems to me that the `__update_if_key_exists` call is not needed anymore, since the `log` module is loaded in the `__init__` method.",
+      "reference": "any testcase for saargs, and tidyargs argument processing?"
+    },
+    {
+      "prediction": "I think this should be `private string _platformNotSupportedExceptionMessage;` instead of `private string _platformNotSupportedExceptionMessage; private bool _forCompilationThrowPlatformNotSupported;`",
+      "reference": "@ericstj do you think it is worth combining these two? With the presence of the message meaning it is enabled?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get it to work.",
+      "reference": "why not use instance.RootDeviceName ?"
+    },
+    {
+      "prediction": "`std::ifstream` and `std::ofstream` should be included before `std::ifstream` and `std::ofstream`",
+      "reference": "not needed, ADIOS2 also needs to check for subfiles. ADIOS2 tries to remove dependency on serial `fstream`."
+    },
+    {
+      "prediction": "I think we should throw an error here instead of just returning a promise.",
+      "reference": "no need to wrap in new Promise - you can simply throw the exception"
+    },
+    {
+      "prediction": "I think this test should be moved to `tests/ocsp/manual`",
+      "reference": "Isn't this testing an implicit config, not a manual config?"
+    },
+    {
+      "prediction": "I think we should rename this to `chainToFullRules` to be consistent with `chainToContents` and `chainToDataplaneHashes`.",
+      "reference": "I don't think this is used any more, please remove."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `mysteriumClient`'s `FindProposals` method.",
+      "reference": "shouldn't we name abbreviations up-cased? `myId` -> `myID`?"
+    },
+    {
+      "prediction": "I think this should be `if (node[ATTR_KEY]!==null) {`",
+      "reference": "This just always calls refs. I think we need to either hoist ref invocation back out of `setProperty()` (it used to happen during rendering), or wait for component recycling to go away."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to register the controller. I think it should be in the `builder` package.",
+      "reference": "Oof. I may have missed this while reviewing #3805"
+    },
+    {
+      "prediction": "I don't think you need to include boost on OS X. You can just include boost at the top of the file.",
+      "reference": "why the NOLINTs here? please just alphabetize the headers (should be easier than adding NOLINTs) unless there's a good reason not to..."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Should we maybe return `<svg />` instead?"
+    },
+    {
+      "prediction": "I think this should be `loadPluginLoader`, not `plugin-loader`",
+      "reference": "Please use camelCase in new code"
+    },
+    {
+      "prediction": "This should be `''` instead of `''`",
+      "reference": "Looks like the whitespace is off here. Not sure if there's a space or two too many or if there is an issue with tabs vs. spaces, but you probably want to fix this :)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this note. I think it should be in `vendor.js` and `vendor.css` files.",
+      "reference": "This was just my laziness and I didn't copy the first line. Just remove it together with one empty line. Thanks"
+    },
+    {
+      "prediction": "Why not use `ProductDataFactoryInterface` here?",
+      "reference": "Why did you decide to change CartItemTest but you didnt change QueryBuilderWithRowManipulatorDataSourceTest?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `LoadUserPlusKeys`.",
+      "reference": "This was sort of a pre-existing bug -- we should be busting our local cache if the key isn't present. However, `KBPKIClient.HasVerifyingKey` already took care of it on that path. This way is better though."
+    },
+    {
+      "prediction": "I think this should be `go:go1.16`",
+      "reference": "Should this tag addition for generated files be handled explicitly in `private/model/cli/gen-api/main.go`"
+    },
+    {
+      "prediction": "Shouldn't this be `if (K9.isQuietTime() && !K9.getQuietTimeNotificationEnabled()) {`?",
+      "reference": "You want to return if it's quiet time and notifications during quiet time are **not** enabled. So this is either a logic error or it's bad naming of the setting/field/method."
+    },
+    {
+      "prediction": "Do we need to set this to `https` if we're already using `hsts`?",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. It seems like it should be in `Travis::Methods`, but I can't find where it is used.",
+      "reference": "cool, here (L146 && L148) maybe (repo_private? || force_private?) && !installation? to prevent double call of installation?"
+    },
+    {
+      "prediction": "I think this should be `util.Failed(\"Failed to GetActiveApp\", err)`",
+      "reference": "this error carries a bit more meaning for us, but less meaning for users. we also lose the context of what command produced the failure. I'd prefer to keep the original error message. If we need better identification of GetActiveApp errors, maybe we could address that in the error messages it returns?"
+    },
+    {
+      "prediction": "I'm not sure if this should be a warning or an error. I think it should be an error, but I'm not sure if there is a better way to do it.",
+      "reference": "Same: DO NOT USE IN PRODUCTION. Yes we should shout ;)"
+    },
+    {
+      "prediction": "I'm not sure if we want to set the lease duration to 5000ms or 10000ms. I think 10000ms might be a bit too long.",
+      "reference": "This change may be hiding more errors, if we continue to see more."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way I could get it to work.",
+      "reference": "Beware of exposing the underlying features directly. Any mutable fields should be reset per request."
+    },
+    {
+      "prediction": "I think this should be `baseParent.replaceChild(base, initialBase);` and not `baseParent.insertBefore(base, initialBase);`",
+      "reference": "This is probably more nuanced than I can feasibly check in a PR review, haha. Was the issue here that `replaceChild()` removes `initialBase` from the DOM before `recollectNodeTree()` invokes `componentWillUnmount()` on the owning component?"
+    },
+    {
+      "prediction": "I don't think this test is necessary, it's already tested in `testClickingALinkMadeUpOfNumbersIsHandledCorrectly`",
+      "reference": "I believe it's misleading name for the condition. I prefer \"element_to_be_disable\" We can have a condition, when element is enabled but we can't click it, because another element overlays above it. So, If we use \"unclickable\" we might mislead people, who use that condition to verify if element can be clicked"
+    },
+    {
+      "prediction": "I think this should be `FileAppender<T> appender = new FileAppender<>(path);` instead of `FileAppender<T> appender = new FileAppender<>(path); return appender;`",
+      "reference": "Do we need to expose it? It looks like it is only used in tests and only to obtain the final metrics. I think you can get the same by using `DataFile#lowerBounds` and `DataFile#upperBounds`. It seems `DataWriter` already exposes `toDataFile` that you can use."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "the test was failing likely due to locale issue. had to increment the date by 1 to make it pass"
+    },
+    {
+      "prediction": "This should be removed.",
+      "reference": "The changes below here can also be undone."
+    },
+    {
+      "prediction": "I think we can remove the `context` import here.",
+      "reference": "Instead of importing context besides beehivecontext, I'd suggest include golang context functionalities into beehivecontext. To keep channelq only rely on beehive."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "I'm not sure about this one. Are you sure we don't need to initialize this fragment anywhere else?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this treatment. I think it should be in `TreatmentsPlugin.getPlugin()`.",
+      "reference": "You didn't set a source here, so it will probably fail."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `integration/transport_test.go`",
+      "reference": "Might be easier to just do `handler{Handler: rpcHandler}` since zero-value of `Deps` is now valid."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Could you please add docs for this method?"
+    },
+    {
+      "prediction": "I think it would be better to have this method return `this.getClass().getSimpleName().replaceFirst(\"^AST\", \"\")` instead of `getXPathNodeName()`.",
+      "reference": "we should probably `@Deprecated` this implementation to be removed in PMD 7.0.0"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Bolt::Shell::Bash` class, since that's where the `Bolt::Shell::Bash` instance is created.",
+      "reference": "I wonder if 'unix' makes more sense for this now? I was thinking of putting up a 'cleanup' PR - I think I missed a few \"remote\" var names in the first PR, and want to reassess unifying the transport classes that just define `with_connection`. Renaming this could be part of that."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this change. I think it should go in the `cppcheck` converter.",
+      "reference": "Do something similar to what was done in JsonTests (can you derive from JsonTests?)"
+    },
+    {
+      "prediction": "I think this should be `# type: Union[ignore, None, ignore]`",
+      "reference": "Yes, not much we can do here."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Why do we need this?"
+    },
+    {
+      "prediction": "I think this should be `console.log(hash);` instead of `console.log(addrStr);`",
+      "reference": "Would prefer to see an actual script here."
+    },
+    {
+      "prediction": "I think this should be a `FileAlreadyExistsException` instead of an `IOException`.",
+      "reference": "seems reasonable to also want to include the 'to' location?"
+    },
+    {
+      "prediction": "ExtraAugmentation should be the last one.",
+      "reference": "This change is unnecessary."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in the `onMouseOver` event handler.",
+      "reference": "Just as a note: Is it possible that the select css class is the same as the hover one? If this is true we should choose a different style for the selection css class, to make it more clear to the user I think."
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think it's better to have a `CopiedAnnotations` field in the `ControllerOptions` struct, which is a slice of `string`. This way, you don't need to have a separate slice in the `Certificate` and `CertificateRequest` structs, and you also don't need to have a separate slice in the `Order` struct.",
+      "reference": "What about Secrets, are we not also trying to cover those?"
+    },
+    {
+      "prediction": "This should be `src/common/libkvs/jansson_dirent.h`, not `src/common/libkvs/treeobj.h`.",
+      "reference": "(commit messge): how about > update internal commit API for RFC 11"
+    },
+    {
+      "prediction": "I think this should be 30 seconds, not 5 seconds.",
+      "reference": "This will now wait for 2.5 hours, not 90 minutes, if the waiter delay is 5s"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `src/event-helper.js`",
+      "reference": "See this error on console too so I don't think it's just test simulation. For some reason event doesn't show up sometimes"
+    },
+    {
+      "prediction": "I think it would be better to move this to `cmd/etcd/snapshot.go` so that we don't have to duplicate the code here and in `cmd/etcd.go`.",
+      "reference": "We're passing the same thing twice?"
+    },
+    {
+      "prediction": "I think this should be `base64.StringSlice(\"user-data\")`",
+      "reference": "Shouldn't the value in Bootstrap.Data be just a plain string?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the sampling priority. I think it should be done in the `AppSecSpanBuilder` class, which is responsible for setting the sampling priority.",
+      "reference": "@robertpi Isn't this changing the sampling priority when `KeepTraces==false`? The sampling priority could be `AutoReject` or `AutoKeep`, based on the sampling decision (or the user may have specified something else). Seems like we shouldn't be changing it in this case?"
+    },
+    {
+      "prediction": "I think this should be `DefaultCloudReadyCheckURL = \"/readyz\"` and not `DefaultCloudCoreReadyCheckURL = \"/readyz\"`",
+      "reference": "The const can also be used in cloud/pkg/cloudhub/servers/httpserver/server.go L46-47?"
+    },
+    {
+      "prediction": "This should be `if (blocks[i].type === 'procedures_prototype' || blocks[i].type === 'blocks_procedures_prototype') {`",
+      "reference": "Use a constant for `procedures_prototype`."
+    },
+    {
+      "prediction": "`arg` is a string, so `arg!r` is a string too.",
+      "reference": "This file should probably not be touched as the formatting of the string seems to be a test in itself ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to fix the issue. I think it should be fixed in `sigaction_common.c`",
+      "reference": "style nit: missing leading '*'"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `process` method of the `MessageFrame` class.",
+      "reference": "The operationTracer is passed into the processTransaction() method. We should probably pass the no tracing in one in from where it is called."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. It's not an SPI driver, it's an I2S driver.",
+      "reference": "I believe this line was an error, it should have been `sercomRXPad1` to be consistent with the pin numbers. Resolving this ambiguity by following the pin numbers."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a link to the proposal itself.",
+      "reference": "May be useful to have the reason here, though clearly not pressing."
+    },
+    {
+      "prediction": "I think this should be `dask_array_module()` if `is_dask(arr)` is `True`, otherwise `dask_array_module().stack` should be used.",
+      "reference": "These were inverted before"
+    },
+    {
+      "prediction": "This method is not implemented.",
+      "reference": "DB is implemented in PR 951"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `pkg/prometheus/gauge.go` or `pkg/prometheus/metricspkg.go`.",
+      "reference": "move all constants to `pkg/util/constants.go`, these constants had been used in `volume_stats.go` file too. So better to import them."
+    },
+    {
+      "prediction": "This file is covered by the GNU General Public License. See the file `COPYRIGHTS` for more details.",
+      "reference": "Could you revisit the full header and add appropriate spaces after the hashes?"
+    },
+    {
+      "prediction": "I think this should be `hint` instead of `codeHint`.",
+      "reference": "This should either be `hint` that takes a string to render inside of the partial container or probably better you should just render a separate hint field above the code field in the same tab."
+    },
+    {
+      "prediction": "I think we should be able to get away with 2 seconds here. I'm not sure how long it will take to get the stream info, but I think it should be less than 100ms.",
+      "reference": "this was flapping"
+    },
+    {
+      "prediction": "I think this should be done in `OnViewLoaded` instead of `OnViewCentered`",
+      "reference": "Please add a check to see if SafeAreaInsets is available. It isn't on all iOS versions."
+    },
+    {
+      "prediction": "Line is too long. [86/80]",
+      "reference": "interpolation here was causing the translation to not get picked up"
+    },
+    {
+      "prediction": "I think this should be `flow_controllers.erase(writer);`",
+      "reference": "We don't need to disable the flow controller here as in `StatefulWriter`s destructor?"
+    },
+    {
+      "prediction": "I think this should be `create_no_window=True`",
+      "reference": "I would rather no have this as a `kwarg` as it encourages \"growth\" which lead to an unweildy constructor in other classes. Let's add a method or property to take care of this instead as I think it's usage is going to be quite low."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to apply the policy. I think it would be better to apply the policy on the inbound traffic to the workloads.",
+      "reference": "I just noticed that we use Return here, when we have logically allowed a packet, whereas in the forward chain we use AcceptAction. Do you know why that is?"
+    },
+    {
+      "prediction": "I think this should be `func Max(x int, y int) int`",
+      "reference": "There's one for floats but not for ints. The float implementation is non-trivial though which I guess is the argument there."
+    },
+    {
+      "prediction": "I think we should rename this to `DCRedirectionPolicyAllAPIsForwarding`",
+      "reference": "If we have a redirection policy for Selected API forwarding, why isn't \"redirect all\" just a special case where they are all Selected?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this. I think it should be in `prometheus`, since it's a Prometheus-specific thing.",
+      "reference": "I wouldn't add this. Just completely ignore config if custom reporter is not `nil` in server options."
+    },
+    {
+      "prediction": "I think this should be `std::string clusterIdFile = `",
+      "reference": "What's the meaning about the code?"
+    },
+    {
+      "prediction": "I think this should be `loadingUrl.getHost().contains(bridge.getHost())` instead of `loadingUrl.getHost().contains(bridge.getServerUrl())`",
+      "reference": "Would this incorrectly trigger if `loadingUrl.getHost() = \"something.app\"` and `bridge.getHost() = \"app\"`?"
+    },
+    {
+      "prediction": "`reg_decoded_bbox` -> `reg_decoded_bbox`",
+      "reference": "Note generally it -> It"
+    },
+    {
+      "prediction": "I think this should be `force -> createIndexSchema(dcore, solrConfig, force)`",
+      "reference": "Does this mean that every call to `configSet.getIndexSchema()` will create a new schema object (if shared schema is not enabled)?"
+    },
+    {
+      "prediction": "I don't think it's a good idea to add copyright notices to the source code.",
+      "reference": "License header should use the Apache one."
+    },
+    {
+      "prediction": "I think this should be `if len(matches) > 0 {`",
+      "reference": "I think this makes more sense to be applied within the parser code where we evaluate the `glob` builtin? This function looks generic in the sense of where it can be used, and perhaps no matches is fine in other cases?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be in `aws/config.go` since that is the place where the `aws/config` package is defined.",
+      "reference": "should we validate that AWSRegion has also been set here so that we don't get into the if-block that relies on the ec2 metadata client below?"
+    },
+    {
+      "prediction": "I think this should be `self.__db_version_info = db_version_info` instead of `self.__db_version_info = version`",
+      "reference": "maybe naming `version` to `package_version` or something similar can be more descriptive"
+    },
+    {
+      "prediction": "I think this should be `protected` instead of `protected int getId(org.apache.parquet.schema.Type type)`",
+      "reference": "Why was this changed?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Should not have snuck in here right? This is a golang 1.17 feature, wii definitely want to update these"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I found that works.",
+      "reference": "Pundit is unaware of namespacing, however you can still create the class under policies/org_admin/plans_policy.rb. That means, the condition for the unless can be reused in other places too. You would call the policy as OrgAdmin::PlansPolicy.new(current_user).index? after that unless"
+    },
+    {
+      "prediction": "Shouldn't this be `pmd-userdocs_getting_started.html#incremental-analysis`?",
+      "reference": "this warning shouldn't be produced if we configured a cache either"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in the `change-stream` folder, but I don't know for sure.",
+      "reference": "what if the collection is already there?"
+    },
+    {
+      "prediction": "I don't think this is the right place to define this. This is for use by the `Screenshot` class, which is for use by the `screenshot` method of the `WebDriver`. The `screenshot` method of the `WebDriver` should be called from the `screenshot` method of the `Screenshot` class.",
+      "reference": "instead of removing can you reference org.openqa.selenium.TakesScreenshot ?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Please revert unrelated changes"
+    },
+    {
+      "prediction": "I think this should be `if (countlyGlobal.locale) {` instead of `if (countlyGlobal.company) {`",
+      "reference": "As far as I know, this will replace only first occurance of Countly, not others, if there are more than one Countly word in localized string"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think you should use a regular expression to replace all characters except for letters, numbers and underscores.",
+      "reference": "I suggest, that we move the regex (`[^a-zA-Z0-9_]`) to a separate constant field, because we're using it in multiple places and I think it's quite prone to misspellings."
+    },
+    {
+      "prediction": "I think this should be `Iterable[Union[Dict[str, Any], List[str]]]` instead of `Iterable[Optional[Dict[str, Any]]]`.",
+      "reference": "Feel free to disregard, but I've been trying to use `Mapping` and `Sequence` instead of `Dict` and `List` when possible, because they communicate that the type is immutable, and also are covariant."
+    },
+    {
+      "prediction": "I think this is the wrong place for this.",
+      "reference": "No need for blank line here. This block will then be reordered alphabatically."
+    },
+    {
+      "prediction": "I'm not sure if we want to make this a constant. I think it would be better if it was a function that returns a string based on the value of `os.Getenv(\"KUBERNETES_SERVICE_ACCOUNT_TOKEN\")`.",
+      "reference": "let's call it `in-cluster` instead of `local`. less chance of confusion."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "Please use camelCase instead of underscore_case"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could come up with that worked.",
+      "reference": "Ideally, we'd fold the schema into an expanded system_meta, but this is a good first step."
+    },
+    {
+      "prediction": "I think this should be renamed to `TABLE_FIELDS` or something like that.",
+      "reference": "Is this used?"
+    },
+    {
+      "prediction": "Why do we need this ifdef? I don't see it being used anywhere in the PR.",
+      "reference": "Should we use `#if NETFRAMEWORK` for consistency with the other projects?"
+    },
+    {
+      "prediction": "I think this should be `pod.Name` instead of `containerName`",
+      "reference": "If it parses failed, both of the `containerName` and `podId` should be the empty strings, it's non-sense to define this error, I suggest just return it."
+    },
+    {
+      "prediction": "I don't think we need this method anymore, right?",
+      "reference": "line 26 can be removed i think > applicationId: or('config.unsplashAPI.applicationId', 'settings.unsplash.applicationId'),"
+    },
+    {
+      "prediction": "I think this should be `if (device_type == \"cpu\")`",
+      "reference": "To generalize for possible future new enum options."
+    },
+    {
+      "prediction": "This should be `os.environ.setdefault('SCAPY_ROOT_DIR', os.path.abspath(os.getcwd()))`",
+      "reference": "Shall we remove this variable at the end of UTScapy execution?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. I think this should be done in the `getInterfaces` method of the `SymbolTable`.",
+      "reference": "What is this checking for? Why is returning the list of interfaces the correct behaviour in this case?"
+    },
+    {
+      "prediction": "I think this should be `HmbSync` instead of `Hmb Sync`",
+      "reference": "Please name it correctly as Beam ;)"
+    },
+    {
+      "prediction": "I think this should be `return true` instead of `retry = true`",
+      "reference": "After chatting with Andrew, we think it's better to look for this particular error string that is causing issues (\"connection reset by peer\") instead of blindly retrying on any error we don't know about."
+    },
+    {
+      "prediction": "I don't think this is the right place for this. I think it would be better to put it in the `SwaggerCodeModelValidator` and `SwaggerDocumentationModelValidator` classes.",
+      "reference": "any reason to not have defaults here like you had before? (ARM)?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "This is a change. The old code was just wrong before, but now here (and one other machine-readable place) we expose the start index as 0 for the first item."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. The problem is that we're stripping whitespace from the start and end of the string, but we're not stripping whitespace from the middle of the string. I think we need to do something like: `std::stringstream ss(sln);` `boost::trim_if(ss,boost::is_any_of(\" \\t\\r\\n\"));` `res = boost::lexical_cast<RWMol *>(ss.str());`",
+      "reference": "This changes the behavior of the parser, right? Any particular reason to do this aside from the fact that you have to since the function takes a const?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best solution, but it's the only one I could come up with.",
+      "reference": "Given how simple the Portal implementation is, I wonder if we should move it core... Though we'd have to export it which would be more bytes we can't crawl back..."
+    },
+    {
+      "prediction": "I think this should be `s.s.hooks = utils.clone(this.s.hooks);`",
+      "reference": "The more correct way of doing this is `this.s.hooks.clone()` but either way works. Thanks for finding this :+1:"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now. I'm not sure if there is a better way to do this.",
+      "reference": "Can we update bool to an enum instead of a string? This helps future extensibility"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Should we share, since this is also being used with gsa18f?"
+    },
+    {
+      "prediction": "I think this should be `<div id=\"crash-data-subrow\" class=\"datatablesubrow\">'` instead of `<div id=\"crash-data-subrow\" class=\"datatablesubrow\">'`",
+      "reference": "There can be multiple subrows, so it is not a good idea to use `id` there, as id should identify single unique element. Instead you could just add the id of whole table or of whole crash plugin view"
+    },
+    {
+      "prediction": "I think it would be better to use `os.path.join(failed_dir, \"failure.c\")` instead of `os.path.abspath(failed_dir) + \"failure.c\"`.",
+      "reference": "Why was this assert removed? Shouldn't we check if the file is in the zip?"
+    },
+    {
+      "prediction": "I think this should be a debug message.",
+      "reference": "This logic should be handled at informer handle functions i.e. AddFunc, UpdateFunc, DeleteFunc"
+    },
+    {
+      "prediction": "I think we can remove this function and use `apisv1alpha1.PoolType(poolSpec.RaidGroups[0].Type) == apisv1alpha1.PoolStriped` instead.",
+      "reference": "S1008: should use 'return <expr>' instead of 'if <expr> { return <bool> }; return <bool>' (from `gosimple`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it should be in `tests/unit/issue/test_issue_edits.go` but I'm not sure if that's the right place to do it.",
+      "reference": "Just to remove branching, use `defaultTitle` explicitly in the test cases. (Is this gofmt'd?)"
+    },
+    {
+      "prediction": "I think this should be `SampleTransformer.create(SampleType.IN_CODE, SampleType.OUT_CODE)`",
+      "reference": "I don't think you use this variable anywhere. (You call `SampleTransformer.newBuilder()` below)"
+    },
+    {
+      "prediction": "I think this function should be called `chirp_wrap_hash_whoami` to match the rest of the function names.",
+      "reference": "The digest is in binary and variable size (the digest size is result if > 0). So we can't use xxstrdup."
+    },
+    {
+      "prediction": "I don't think we should add new methods to an existing interface. I think we should just remove the old methods and add the new ones.",
+      "reference": "We don't need this additional method."
+    },
+    {
+      "prediction": "Should this be `GetStreamOutbound() StreamOutbound`?",
+      "reference": "We can't do this. Adding a method to an interface is a breaking change. This was an oversight on our part when we converted ClientConfig from a struct to an interface. OutboundConfig was introduced to fix this, the idea being that we should use OutboundConfig everywhere instead of ClientConfig. In case of Dispatcher, we want to use the OutboundConfig and MustOutboundConfig methods instead of ClientConfig, and for streaming, the client needs to attempt to upcast a ClientConfig to OutboundConfig and error out if that's not possible."
+    },
+    {
+      "prediction": "I think this should be `eprosima::fastdds::dds::DomainParticipantFactory::get_instance()->get_default_participant_qos()`",
+      "reference": "Why not using the constant `PARTICIPANT_QOS_DEFAULT` here?"
+    },
+    {
+      "prediction": "I think this should be `tmImplementation = \"cpp\"`.",
+      "reference": "Shouldn't this be `tm_cpp` to match `temporalImp`? (There are multiple CPP implementations, so 'cpp' is ambiguous.)"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Need to remove this empty line for rubocop to be happy."
+    },
+    {
+      "prediction": "Why do we need to do this?",
+      "reference": "just checking: do we need to remove this because otherwise the `List` operations may not return the entire set of Pods / Namespaces? do you know why we used `alwaysReady` in the first place, I can't remember?"
+    },
+    {
+      "prediction": "I think we can remove the `serverSelectionTimeoutMS=1000` from `mongodb://%2Ftmp%2Fmongocryptd.sock/?serverSelectionTimeoutMS=1000`",
+      "reference": "maybe we can replace `mongoClient.s.options.autoEncryption.cryptdConnectionString` with `options.autoEncryption.cryptdConnectionString`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to call `os_clone_post()`. I think it would be better to call `os_clone_pre()` and then `os_clone_post()`.",
+      "reference": "But we're already doing os_switch_lib_tls to app a few lines above, so there is now redundancy we should alleviate."
+    },
+    {
+      "prediction": "I don't think we need to pass in the parent trace context here. We can just use the parent trace context.",
+      "reference": "Under what circumstance would we have a span id already? Is this for testing purposes?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "The original text can be read to say \"not waited on by queue ...\" meaing that the second queue was supposed to have *waited* for the signal... which is especially confusing when it's the same queue that has signalled the semaphore *twice* without a wait."
+    },
+    {
+      "prediction": "I think this should be \"timed out waiting for response\"",
+      "reference": "just clarifying the error messages a bit."
+    },
+    {
+      "prediction": "I'm not sure if we want these to be public, but I don't think they're used outside of this file, so it's probably fine.",
+      "reference": "Have those (and other data like conf...) shared across tracers?"
+    },
+    {
+      "prediction": "I think we can remove the `!windows` part of the `+build !windows`",
+      "reference": "could we add this file only when we need one?"
+    },
+    {
+      "prediction": "I think this should be `ACMEHTTP01ResolvConfFile` instead of `acmeDNS01ResolvConfFile`.",
+      "reference": "This may not even be needed."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be in `runtime` since it's related to the runtime, but I'm not sure if it should be in `debug` since it's not debug related.",
+      "reference": "Capitalisation seems inconsistent here?"
+    },
+    {
+      "prediction": "Should this be `Debug` and not `Error`?",
+      "reference": "maybe better Error(string text) and Error(Exception ex) separately?"
+    },
+    {
+      "prediction": "I think this should be `langserver` instead of `lps`.",
+      "reference": "can you leave the old one in too please? at least for now, otherwise anyone using it now will break."
+    },
+    {
+      "prediction": "Shouldn't this be `if (target && target->getZone() != ZONE_PVP) {`?",
+      "reference": "Trailing tab, remove it in another PR."
+    },
+    {
+      "prediction": "I think this should be done in the `tlMgr.signalFatalProblem` function.",
+      "reference": "what about merging the error checking logic to within `errShouldUnload`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this action. I think it should go in `articles_controller.rb`",
+      "reference": "Probably want Topic.find_by_slug! to handle bogus topics. The next line will try to load articles off of nil otherwise."
+    },
+    {
+      "prediction": "I think this should be `return !get_policy().is_gpu();`",
+      "reference": "Should it really be done on test side?"
+    },
+    {
+      "prediction": "nit: I think this should be `cleanupTask` instead of `cleanup`",
+      "reference": "Can we elaborate on why? I'm guessing because we need the execution role to call FE stopTask during `cleanupTask`, but would be good if we make it clear here."
+    },
+    {
+      "prediction": "I think this should be `if err := opts.validateFlowExporterConfig(); err != nil {`",
+      "reference": "is there a reason why this is not called from inside `validate`?"
+    },
+    {
+      "prediction": "nit: `rwTimeout` -> `readWriteTimeout`",
+      "reference": ":+1: I much prefer this style for functions with more than a few arguments."
+    },
+    {
+      "prediction": "I think this should be `else if (hasQualifiers) {`",
+      "reference": "Are these mutually exclusive options (parameters vs qualifiers)? If parameters aren't supported, may want to consider throwing an exception for \"hasParameters\". If parameters and qualifiers are both legit options, then maybe change this to `if (hasQualifiers) { ... } if(!hasQualifiers && !hasParameters)`"
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "reference": "Why did you remove this line?"
+    },
+    {
+      "prediction": "I think this should be `invalid_range_of_rows` instead of `invalid_indices`.",
+      "reference": "Very unclear error message. Remember that these are messages that your users see. Please be more specific"
+    },
+    {
+      "prediction": "I think we can remove this altogether. We don't need to support both windows and windows server 2019.",
+      "reference": "Can we add these when we need them"
+    },
+    {
+      "prediction": "I think this should be `array_filter` instead of `filter`",
+      "reference": "I believe you can simplify this to simply `return array_unique(array_filter($authors));` -- the default behavior of array_filter is to filter out empty elements, so the callback should not be required in this situation. Do you mind giving it a try and updating the PR if it works?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it would be better to do it in the `UIAElement` class itself.",
+      "reference": "Would it make sense to log a debug warning here?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "\"in the keystore\", but nevermind"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems to be the right place to do it.",
+      "reference": "This should be done in a separate PR."
+    },
+    {
+      "prediction": "I think this should be `player.getName().equals(playerName)`",
+      "reference": "Rather than this maybe `Text.sanitize` would be better here"
+    },
+    {
+      "prediction": "Why is this needed? I don't see any other changes in this file that would require it.",
+      "reference": "lot of places setting default 'false' to static value - noise in PR"
+    },
+    {
+      "prediction": "I think this should be `getimagesize($event->getCacheFilepath())` instead of `getimagesize($sourceFilePath)`",
+      "reference": "I think you should use `$event->getOriginalFileUrl()` instead of `$sourceFilePath` here."
+    },
+    {
+      "prediction": "I think this should be `api_quota_config` instead of `api_quota` to match the rest of the code.",
+      "reference": "nit: It's a matter of taste, but it would be tighter if this is called `api_quota`."
+    },
+    {
+      "prediction": "I think this should be renamed to `GasCalculatorPrecompiledContract` and `GasCalculatorAltBN128PairingPrecompiledContract` instead of `GasCalculator` and `GasCalculatorAltBN128PairingPrecompiledContract`",
+      "reference": "I noticed in all of these we now have to import the PrecompiledContract under testing. Could also rename the package these are in s/precompiles/precompile to keep them in the same package as they are elsewhere."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I'm not sure we should be modifying the config at all here. I think we should just use the value of `effective_user` from `hdfs.properties` instead of `hdfs.client_version`.",
+      "reference": "doubt this matters..."
+    },
+    {
+      "prediction": "I think this should be `scapy.constants`, not `scapy.consts`.",
+      "reference": "Gets updated versions of `LOOPBACK_INTERFACE`, `LOOPBACK_NAME` when importing scapy."
+    },
+    {
+      "prediction": "I think this should be `if (spec.equals(PartitionSpec.unpartitioned()))`",
+      "reference": "checking for ref. equality is probably fine here, but it takes a reader longer to navigate the code and figure out whether ref equality is really wanted here vs just using `equals()`"
+    },
+    {
+      "prediction": "I think this should be called `deleteManager` instead of `workflowDeleteManager`.",
+      "reference": "looks like it's not used? Do we plan to use it in the future?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should go in `statsd` package.",
+      "reference": "I think this is an unrelated remark. I thought we had moved toward the practice of using \"Config\" as the structure name, and Option as a functional argument (`func(*Config)`), and Options as a `[]Option`. See api/trace `StartConfig` and `StartOption`, for example. That's why I prefer this struct be called Config."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `kinto.core.errors`.",
+      "reference": "nitpick: superfluous parenthesis"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think we should move it to `Datadog.Trace.ClrProfiler.Integrations.GraphQL` since it's only used there.",
+      "reference": "If, for a given integration, the tag/measure is always set, it should be added to the strongly-typed tags (here for instance, in GraphQLTags). This way, the underlying dictionary is allocated only in the rare case where users add custom tags"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think it's better to do it in the `if receipt.Status == uint64(iotextypes.ReceiptStatus_Success)` block.",
+      "reference": "The position change of this line may be a hard fork."
+    },
+    {
+      "prediction": "This should be `if (id >= 0 && id < 100) {`",
+      "reference": "id is unsigned"
+    },
+    {
+      "prediction": "I think this should be `@param options A hash of additional options. Each option is a string key followed by a list of arguments.`",
+      "reference": "It looks like Puppet doesn't provide type validation for options and we don't provide it ourselves as part of the function. Is that something we'd want to do at some point?"
+    },
+    {
+      "prediction": "I think we can remove the `buildTask` argument from the constructor and just pass in the `ops`, `table`, `schema` in the `BaseTableScan` constructor.",
+      "reference": "Looks like this didn't need to change."
+    },
+    {
+      "prediction": "I think this should be `self.disable()` instead of `self.global_set_option()`.",
+      "reference": "I don't understand why this is here and what's the intention with it? Can you elaborate what's the reasoning behind this change?"
+    },
+    {
+      "prediction": "I think it would be better to have `ReadLockShards` and `WriteLockShards` as `int64` instead of `int`",
+      "reference": "I think @sergeybykov converted (wanted to convert) all `shardID` to `int32` from `int64`. Please check with him."
+    },
+    {
+      "prediction": "This should be `std::min(creature->getMaxHealth(), maxHealth)`",
+      "reference": "This will only fix client side (so showing it in-game) server still gonna hold invalid values I guess?"
+    },
+    {
+      "prediction": "I think this should be `cls, repr_kdims, repr_vdims, repr_kwargs, extras, class_name, apply_args`",
+      "reference": "Shouldn't this be `cls.param.warning`?"
+    },
+    {
+      "prediction": "I'm not sure why this test is here, but it seems to be testing something I don't understand.",
+      "reference": "This spec can be modified, giving you extra strength (Check this fetch key doesn't work and therefore returns `nil`)"
+    },
+    {
+      "prediction": "I think this should be `self.config.options.listen_host or \"localhost\"`",
+      "reference": "Could we not just always include `self.config.options.listen_host`? If that is empty, the `address[0]` check should also never be true."
+    },
+    {
+      "prediction": "I think we should make this `TableProperties.METADATA_SPLIT_SIZE_DEFAULT` instead of `TableProperties.METADATA_SPLIT_SIZE`.",
+      "reference": "Why not just access `ops` directly like before?"
+    },
+    {
+      "prediction": "I think you can remove the `else if (isLogin) {` as it's the same as the one in the `if (isMapLoaded) {`",
+      "reference": "Is there even a need to keep the bool?"
+    },
+    {
+      "prediction": "I think this should be `os.path.dirname(os.path.dirname(__file__).partition(\"tests/integration/nupic/opf\")[0])`",
+      "reference": "wrap to stay under 80 characters"
+    },
+    {
+      "prediction": "Why is this needed? I don't see anything in the code that would throw an exception if the layer does not have the correct engine.",
+      "reference": "`return NULL` is more clear than `throw` at least in my reading, and consistent with #3362."
+    },
+    {
+      "prediction": "I think this should be `self.env.add_path({\"PYTHONPATH\": os.path.dirname(self.engine.artifacts_dir)})`",
+      "reference": "Both of these are important. Without it, locust won't find my libraries in the cloud."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it. I think it would be better to use `o.store.ListServices(o.appName)`.",
+      "reference": "This should remain as `ListServices`"
+    },
+    {
+      "prediction": "I think we can remove this package.",
+      "reference": "If no tests, remove this file."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea. I think it's fine to leave it as-is for now, but I'm not sure if it's a good idea to change it in a future version.",
+      "reference": "This will require changes to the docs. I suggest searching the gh-pages branch for the previous variable name."
+    },
+    {
+      "prediction": "I think this should be `os.path.join(globalVars.appDir, \"lib\", buildVersion.version, u\"nvdaHelperRemote.dll\")`",
+      "reference": "`RemoteHandler` is used in nvda_slave and slave does not have access to `globalVars` currently. Perhaps `globalVars.appDir` should be set to `None` in `globalVarrs by default and set to NVDA directory when they're imported for the first time."
+    },
+    {
+      "prediction": "I think this should be `true` by default.",
+      "reference": "In case there is more configuration to be added to the \"inmem\" metrics later (however unlikely that is), we may want to play it safe and create an `InMemConfig` struct and include the disabled flag there. This also gives the config symmetry across the plugins."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "I'll need to pull this out of here since we're sharing this in other tests. Same for `procurement_spec.rb`"
+    },
+    {
+      "prediction": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "what about `observer` as a var name here (and below) rather than `obs` ? -- would be clearer, imo!"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The error message is a dictionary, not a string.",
+      "reference": "could you change this to instead of being a `try.. except` be `message = message.get('message')`"
+    },
+    {
+      "prediction": "I think this should be `MIT License`.",
+      "reference": "Add space between License and nameapace"
+    },
+    {
+      "prediction": "I'm not sure if we should be saving the user object here, or just the id. It seems like the id would be more appropriate, but I'm not sure if there is a better way to do this.",
+      "reference": "should we have a `before_filter` for auth here?"
+    },
+    {
+      "prediction": "I think we can remove the `err != nil` check here, since the `err` will be non-nil if there was an error.",
+      "reference": "@vincepri I know we discussed moving this up and only calling it once, but I didn't (quickly) see an easy way to generate the right event based on the appropriate action (attach vs detach). Happy to move it around if you have suggestions."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Can we use record.update here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do it in `writeHeaders` so that we don't have to do anything if we don't want to.",
+      "reference": "Shouldn't the exact case option matter here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right change. `Derived` is supposed to be used for derived tasks, not derived util tasks.",
+      "reference": "Errr, why haven't we noticed this until now?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it should be in the Lucene50PostingsFormat constructor.",
+      "reference": "It's a little spooky that this method throws exception if you try to set the attribute to a different value than it was set before, but then does leave the new value set in the attributes?"
+    },
+    {
+      "prediction": "I think this should be 1092000",
+      "reference": "Please only increment by 1."
+    },
+    {
+      "prediction": "I think we can remove the `org.joda.time` namespace now.",
+      "reference": "Sorry I might have missed some context, but what's the reason you choose `Period` over `Duration` or `Interval`? (Thumbs up for using `org.joda.time` instead!)"
+    },
+    {
+      "prediction": "I don't think we need to add this to the default settings. I think it should only be added to the default settings if it is enabled by the user. If it is not enabled by the user, then it should not be added to the default settings.",
+      "reference": "Please check if is it possible to mark it as private. What's more please add this to the TypeScript definition file."
+    },
+    {
+      "prediction": "This should be `static_cast<char>(std::toupper(c));`",
+      "reference": "Maybe move this to `util.h` ?"
+    },
+    {
+      "prediction": "Is this needed? I don't see `grpc.h` included anywhere else in this file.",
+      "reference": "oops. this won't work will it"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "@strib I assumed this is what you meant; but let me know if I'm wrong!"
+    },
+    {
+      "prediction": "I think this should be `WorkflowExecutionRunningError` instead of `WorkflowExecutionAlreadyStartedError`",
+      "reference": "I believe the error has changed"
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this method. It's a predicate, but it's not a predicate that returns true or false. It's a predicate that evaluates to true or false based on the presence of constraints in the query.",
+      "reference": "Lint/UnneededDisable: Unnecessary disabling of Naming/PredicateName."
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "Should this be removed now?"
+    },
+    {
+      "prediction": "I think this should be called `responseStreamHandle` to be consistent with the other methods in this class.",
+      "reference": "streamHandle sounds vague to me. Can you be more specific?"
+    },
+    {
+      "prediction": "I think this should be `Object.assign(cursor.cursorState, cursor.options)`",
+      "reference": "Does everything on cursorState belong in the command options?"
+    },
+    {
+      "prediction": "I think this should be `d.NextArg() || d.NextBlock(0)` instead of `d.Next() || d.NextBlock(0)`.",
+      "reference": "This is clever, but I do admit I think it's kinda weird. We can go with it for now and fix it later if people complain."
+    },
+    {
+      "prediction": "I think we can remove bucketRegex now. It's not used anymore.",
+      "reference": "the variable names are not entirely clear. \"bucket\" is essentially the regex that follows gcs naming pattern, right? if so, it should be changed to bucketregex. and then bucketregex, gsregex should be changed to something more specific to what pattern the regex is supposed to match."
+    },
+    {
+      "prediction": "I think this should be `implements` instead of `implements ModelToViewTransformer`",
+      "reference": "I think `ApiModelT` would be clearer than `T`"
+    },
+    {
+      "prediction": "I don't think this copyright is correct. It should be `Copyright (c) 2004-2016 Greg Landrum and Rational Discovery LLC, all rights reserved.`",
+      "reference": "Shouldn't this be 2019?"
+    },
+    {
+      "prediction": "I think this should be 64 instead of 65.",
+      "reference": "We should consider giving this state a more generic name which could be used for flows in similar state in non-containerized Azkban. For instance in the event of a bare-metal executor crashing we could the switch any flows assigned to that executor to this state. Something like `EXECUTE_INFRA_FAILED`, better alternatives should be possible though."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we need to check the `role` attribute instead of the `nodeName` here.",
+      "reference": "This should allow `role=definition` and `role=term`, possibly also `role=list`?"
+    },
+    {
+      "prediction": "I think it would be better to use `Boolean.parseBoolean(value)` instead of `return Boolean.parseBoolean(value);`",
+      "reference": "Should this use `boolean` instead of `Boolean`?"
+    },
+    {
+      "prediction": "I'm not sure if we want to keep this test case. I think it's fine to remove it, but I don't know if it makes sense to remove it.",
+      "reference": "Currently in our Zipkin tests, only one passed parameter would instantiate a Resource and checks for its tags populating. I left the If(UseTestResource) clause in the code in case we would like to return to the old resource tag checking, but if it makes more sense I can remove that parameter entirely."
+    },
+    {
+      "prediction": "I think this method should be called `isNodeAlive` instead of `isNodeLive`, since the latter implies that the node is alive.",
+      "reference": "While this is potentially convenient it seems off topic for the PR/Issue. Also if it is kept, in the realm of taste/style I tend to not use get/set/is for things that are not properties of the object. maybe hasLiveNode(String node) thus someone using it might write `if (zkReader.hasLiveNode(\"foo\")) ...` which reads quite nicely."
+    },
+    {
+      "prediction": "I think this should be `blogCsrPath` and `blogSvidPath` instead of `svidCsrPath` and `svidSvidPath`",
+      "reference": "What's the difference between blogSvid and the SVID above? Looks like the functions that use these aren't being called currently - can they be removed?"
+    },
+    {
+      "prediction": "Shouldn't `reader.Value` always be non-null here?",
+      "reference": "can you add numbers with this change? the CLI is a javascript engine and it can worh number - java adds strings vi concatenation"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `OrderBy` class.",
+      "reference": "Shouldn't this `TEXT` also be dependent on the platform, then?"
+    },
+    {
+      "prediction": "I think this is a bit of a nitpick, but I think it would be better to have a `TableObject` type in the `sem` package, and have this type implement `TableObject` there. This way we don't have to import `TableObject` from `sem` in the `sem` package.",
+      "reference": "Can we name this `TableObject`? I'm not a big fan of the `I` prefix for interfaces and that's not really used in Go. Is it also possible to have the method be unexported? I don't know if that's possible. If it's not, this is good. If it is possible, I'd prefer this method to be unexported."
+    },
+    {
+      "prediction": "This should be `#include <tbb/task.h>` instead of `#include <tbb/task_arena.h>`",
+      "reference": "If tbb.h is included, there is no big sense to include other TBB headers."
+    },
+    {
+      "prediction": "I think we should use `re.match(r'^(.+)\\.s3[.\\-][a-z]{2}-[a-z]+-[0-9]{1,}.*', path)` instead of `uses_path_addressing(headers)`",
+      "reference": "Let's use the same `S3_HOSTNAME_PATTERN` as above here."
+    },
+    {
+      "prediction": "I think this should be `console.log(\"R: \", r)` instead of `console.log(\"R: \", r.path.split('/').length > 1 && r.path.indexOf('::') === -1)`.",
+      "reference": "This shouldn't be here."
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "reference": "A `case` statement may be simpler here."
+    },
+    {
+      "prediction": "I think it would be better to have a method `calculateTagCount` that takes `tags` and returns `newTagCount` and `remainingTagsToInsert`",
+      "reference": "why are we adding the current tag count to new tag count ?"
+    },
+    {
+      "prediction": "I think it is better to use `OptionFile(\"dh\", \"none\", \"none\")` instead of `c.AddOptions(OptionFile(\"dh\", \"none\", \"none\"))`",
+      "reference": "Are you writing \"none\" to file content? no good"
+    },
+    {
+      "prediction": "Why do we need to set the template, name, target and zone here?",
+      "reference": "Should we allow `string|string[]` here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this. I think we should log the error once, and then log it again if the executedDropErrorOnce variable is still false. This way, we don't log the same error multiple times.",
+      "reference": "Is there a reason for logging `error.message` specifically, maybe log the whole error object?"
+    },
+    {
+      "prediction": "I think we can remove the `AddDiagnosticSourceInstrumentation` method and just use `builder.AddSource(typeof(AspNetInstrumentation).Assembly.GetName().Name);`",
+      "reference": "Why is this needed?"
+    },
+    {
+      "prediction": "I'm not sure what is needed here.",
+      "reference": "Nothing. We only expect container-connected events when going via the plugin, not the proxy."
+    },
+    {
+      "prediction": "I think this should be `const rootId = 0;` instead of `let rootId = 0;`",
+      "reference": "We might want to introduce some randomness here in case there are multiple completely separate Preact installations running, i.e. 2 widgets as those will be two roots with a resetting `_domDepth`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this test. I think it should go in `integration_test.go`",
+      "reference": "this seems a bit wrong no? why is this change needed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Transaction` class.",
+      "reference": "Still confusing. Why don't we return something like `'Fee too low: expected X but found Y'`?"
+    },
+    {
+      "prediction": "I don't think we need to add `Ports` field here.",
+      "reference": "Should be from lowercase `json:\"ports\"` the same is defined in MarshalJSON"
+    },
+    {
+      "prediction": "I think this should be `Amazon::Sdk::ApiVersion::V3` or `Amazon::Sdk::ApiVersion::CURRENT` instead of `v3`.",
+      "reference": "Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "I think it would be better to have a separate method for this, so that we don't have to repeat the code in `eval` and `evalLocal` if we decide to change the engine.",
+      "reference": "This code should be executed in NashornEngineFactory"
+    },
+    {
+      "prediction": "I think this should be `time.Now().Nanosecond()`",
+      "reference": "this is not so great as it sets the global rand seed. in tests that invoke this method multiple times for whatever reason, the global seed would be overridden every time the constructor would be called and the instances would have a different seed after every constructor call. it would be better to just pass a new source here as a parameter and do the same within the test. it should simplify things and clean this up"
+    },
+    {
+      "prediction": "I think this should be `--rpc-protect-chainid-in-txs`",
+      "reference": "This looks inverted to me: if `--rpc-require-chainid-in-txs=true`, then shouldn't `unprotectedTransactionsAllowed == false`? Suggest changing this variable to `requireTxReplayProtection` or similar. I think the flag name could be clearer as well, maybe: `--require-tx-replay-protection`."
+    }
+  ]
+}
\ No newline at end of file