diff --git "a/qwen25_lora_official_rank_4_1e-05/complete_results.json" "b/qwen25_lora_official_rank_4_1e-05/complete_results.json"
new file mode 100644--- /dev/null
+++ "b/qwen25_lora_official_rank_4_1e-05/complete_results.json"
@@ -0,0 +1,40694 @@
+{
+  "test_bleu_4": 0.026722299341944283,
+  "test_exact_match": 0.00019667617268167962,
+  "test_rouge": {
+    "rouge1": 0.14729236978584792,
+    "rouge2": 0.025702118262249128,
+    "rougeL": 0.12220512653363477,
+    "rougeLsum": 0.12222192011639979
+  },
+  "test_bertscore": {
+    "precision": 0.8427995183918255,
+    "recall": 0.8462318667257438,
+    "f1": 0.8440955590329605
+  },
+  "test_meteor": 0.11341991042079051,
+  "raw_data": [
+    {
+      "prediction": "I think this should be `language=java;transport=grpc`",
+      "reference": "can we also test for `transport=rest`?"
+    },
+    {
+      "prediction": "I think this should be `$this->config->get($this->mainConfig)->Index->record_batch_size`",
+      "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 think this should be `window.analytics.load(\"<%= ENV['SEGMENT_KEY'] %>\");`",
+      "reference": "I didn't realize we were hardcoding this, thanks for moving it to an env value."
+    },
+    {
+      "prediction": "I think this should be `file.data.size === undefined` instead of `isNaN(file.data.size)`. `file.data.size` can be `null` or `undefined` and `isNaN` will return `false` in those cases.",
+      "reference": "We are trying to support IE 10-11, so we'll need a polyfill for this one, I think."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "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 AbsDistMat& im` and `AbsDistMat& col`.",
+      "reference": "I think im2col should only accommodate CPUMat."
+    },
+    {
+      "prediction": "I think we should move this to `google.cloud.forseti.notifier.notifiers.email_violations`",
+      "reference": "alpha sort the imports"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "Can you explain why is this necessary? Is `snprintf()` with four integer arguments unsafe?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure if this is the right thing to do, but it seems to work for me. I'm not sure",
+      "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": "nit: remove the empty line",
+      "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 join the keys here. The server will do that for us.",
+      "reference": "Also update line 2205 below"
+    },
+    {
+      "prediction": "I think this should be `shared := volumeConfig.GetSharedMountValue() == \"true\"`",
+      "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 don't think this is correct. `freezeColumn()` does re-render the table.",
+      "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 think this should be `string.IsNullOrEmpty(activity.ParentId) || activity.ParentSpanId == ActivityTraceId.Invalid`",
+      "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 don't think we need to log the name of the image provider here.",
+      "reference": "`zap.String(\"image-provider\", provider.Name())` should be in a same line."
+    },
+    {
+      "prediction": "Please remove this line. It's not needed.",
+      "reference": "Why this deletion?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the tests to pass on Python 3.",
+      "reference": "This change looks unrelated to the CL description?"
+    },
+    {
+      "prediction": "`DEDUP_RESUTL_CHECK` -> `DEDUP_RESULT_CHECK`",
+      "reference": "I'm not certain that `res`+`utl` isn't a thing."
+    },
+    {
+      "prediction": "I think we should use `Status` enum instead of `Status.getNumVal()`.",
+      "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": "Please remove this empty line.",
+      "reference": "Merge with next condition"
+    },
+    {
+      "prediction": "I think this should be `DeleteTask(task deploy.TaskStackInfo, opts ...cloudformation.StackOption) error` so that we can pass in the stack options.",
+      "reference": "Maybe add it when it is used."
+    },
+    {
+      "prediction": "I think this should be `{'name': 'non_exist_scanner', 'enabled': False}`",
+      "reference": "More clear naming: NONEXISTENT_ENABLED"
+    },
+    {
+      "prediction": "I don't think we need to do this for `from` and `to` since they are not nullable.",
+      "reference": "why is this change necessary?"
+    },
+    {
+      "prediction": "I don't think this is the right thing to do. If we have a null response type, we should return `null` as the full name.",
+      "reference": "make a SchemaTypeNameConverter.getTypeNameForEmptyType() and call that here."
+    },
+    {
+      "prediction": "This is not the correct way to check if a session exists. You should use `if ($request->hasSession()) { ... }` instead.",
+      "reference": "Use braces on your conditional structures please"
+    },
+    {
+      "prediction": "I don't think we need to force saving the command history. We can just save the command history 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": "nit: extra space after `if`",
+      "reference": "Isn't this the same check as two lines above?"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(options.CA_DIR, 'mitmproxy', 'config.yaml')`",
+      "reference": "Don't redefine, just import the existing one in `cmdline.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 not call `respond_to?(\"#{name}=\")` from the `if` line?"
+    },
+    {
+      "prediction": "I think this should be `commandLine.setWorkDirectory(workingDir);` instead of `commandLine.withWorkDirectory(workingDir);`. `withWorkDirectory` returns a new `GeneralCommandLine` instance, but `setWorkDirectory` modifies the current instance.",
+      "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 think this should be `MvxBasePresentationAttribute` instead of `MvxPresentationAttribute`",
+      "reference": "Are all these namespaces required for this attribute?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to log the output. I think it should be logged in the `Beaker::Test::Result#to_s` method.",
+      "reference": "Given this may be used and printed in other ways, isn't `debug` more appropriate?"
+    },
+    {
+      "prediction": "I think this should return `nil` if the directory doesn't exist.",
+      "reference": "Looks like this is unused? Did you mean to call it when clearing the block journal?"
+    },
+    {
+      "prediction": "I think we can remove the `alerters` field and just use `this.alerters` instead.",
+      "reference": "why named alters? how about this.alerts = alerts?"
+    },
+    {
+      "prediction": "I think this should be `mset.sourcesInfo()` and `mset.mirrorInfo()` instead of `mset.sourcesInfo()` and `mset.mirrorInfo()`.",
+      "reference": "not sure if this has to be here or not tbh"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the topic name is too long, it will fail to create the topic. I think we should just return an error if the topic name is too long.",
+      "reference": "There's a better fix for this in #1741, which should replace this."
+    },
+    {
+      "prediction": "I think this should be `[]byte(\"node1\")` instead of `[]byte{110, 111, 100, 101, 53}`",
+      "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 `os.RemoveAll(dir)` instead of `require.NoError(t, os.RemoveAll(dir))`",
+      "reference": "It would be worth factoring this out to a function, but you don't have to do that here."
+    },
+    {
+      "prediction": "I'm not sure about the name of this interface. I think it should be `UriHandler` or something like that.",
+      "reference": "There's no need for `final` in interfaces."
+    },
+    {
+      "prediction": "I think this should be `Span.SetTag(Tags.Version, value);` instead of `Span.SetTag(DatadogTags.ServiceVersion, value);`",
+      "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 we should keep the empty line here, as it separates the docstring from the function body.",
+      "reference": "This shouldn't be changed."
+    },
+    {
+      "prediction": "I think this should be 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024, 2025, 2026, 2027, 2028, 2029, 2030, 2031, 2032, 2033, 2034, 2035, 2036, 2037, 2038, 2039, 2040, 2041, 2042, 2043, 2044, 2045, 2046, 2047, 2048, 2049, 2050, 2051, 2052, 2053, 2054, 2055",
+      "reference": "This one is important `2: I102 copyright year is outdated, expected 2014 but got 2012`"
+    },
+    {
+      "prediction": "I don't think these imports are needed.",
+      "reference": "Are these imports needed (BiConsumer and Consumer)? If they are unused imports, then precommit will fail."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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": "`$cartItems = $cart === null ? [] : $cart->getItems();` should be `$cartItems = $cart === null ? [] : $cart->getItems()->toArray();`",
+      "reference": "If there are `@param` tags in docblock, there should be `@return` tag also. (applies for a whole file)"
+    },
+    {
+      "prediction": "I think this should be \"validator\" not \"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 the import of `org.hyperledger.besu.crypto.SECP256K1.*`",
+      "reference": "Spotless is configured to reject star imports. Please replace with explicit imports."
+    },
+    {
+      "prediction": "This should be `sycl::info::device::local_mem_size` instead of `sycl::info::device::local_memory_size`.",
+      "reference": "Remind me why you are using CUDA vocabulary when naming these variables."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in `DBReader.getFeeds()`. The reason is that `DBReader.getFeeds()` is used by `DBReader.getAllFeeds()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup()` which is used by `DBReader.getFeedsByGroup",
+      "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 `not solid_subset is None` or `solid_subset is not None`",
+      "reference": "this is a behavior change. solid_subset=[] represents an empty pipeline where as solid_subset=None is the full pipeline"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "reference": "this is a breaking change, any specific reason you want to do this?"
+    },
+    {
+      "prediction": "I don't think we need to add these methods to the interface. They are only used in the `nodeChainReader` implementation.",
+      "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": "I think this should be `if (creature && !creature->isRemoved() && creature->getHealth() <= 0) {`",
+      "reference": "checking health again? I think it is not necessary."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it's better to fail fast and let the pod be restarted by k8s. If we want to retry, I think we should do it in the `Start` method of `typhaConnection` and not here.",
+      "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 we need to check for `typeof lib === 'function'` here. If `lib` is not a function, `new lib()` will throw an error anyway.",
+      "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'm not sure this is the right thing to do. I think we should only queue the target if it's not already queued.",
+      "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 don't think this is the right way to do this. I think you want to do `if not isinstance(pks, StreamSocket) and (arch.FREEBSD or arch.DARWIN):`",
+      "reference": "You should use `isinstance()` instead of comparing the class name to a string."
+    },
+    {
+      "prediction": "I don't think we need to check for `AUDIO` as well, since `isHidden` is only called for `VIDEO` elements.",
+      "reference": "Why should audio and video elements always return `false` for `isHidden`?"
+    },
+    {
+      "prediction": "Why is this commented out?",
+      "reference": "This obviously won't work for all the other platforms."
+    },
+    {
+      "prediction": "I don't think you need to change this.",
+      "reference": "Isn't he diamond operator Java7+?"
+    },
+    {
+      "prediction": "I think this should be `> 0` instead of `>= 0`.",
+      "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 why we need to pass in a pointer to an int here. I think we can just pass in an `int` and return an `int` from the `Rewrite` function.",
+      "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": "I think this should be `0700` instead of `0755`.",
+      "reference": "This seems like an improved pattern :)"
+    },
+    {
+      "prediction": "Align the parameters of a method call if they span more than one line.",
+      "reference": "`open_id_authentication` is no longer the name of the function, and it's not OpenID specific"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `Wallet balance (XRB): ` and `Wallet pending (XRB): `",
+      "reference": "This didn't require corresponding changes to the test case(s)?"
+    },
+    {
+      "prediction": "I think this should be `map[string]string` instead of `string` so that we can support multiple env files.",
+      "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.x-k8s.io/infrastructure-ready` and `aws.cluster.x-k8s.io/control-plane-ready`",
+      "reference": "Should this be: `aws.infrastructure.cluster.sigs.k8s.io` instead?"
+    },
+    {
+      "prediction": "I think this should be `zprd` and not `xprd`",
+      "reference": "Not clear on why this change is correct."
+    },
+    {
+      "prediction": "nit: please add a blank line between the `if` and the `return`",
+      "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 `padding: 20` to match the other stories.",
+      "reference": "All stories in this file also need to have the default padding."
+    },
+    {
+      "prediction": "I think we should assert that sibling->parent == subsurface->parent here.",
+      "reference": "I don't think these asserts are necessary, because `subsurface_find_sibling` already searches in the parent. Or am I missing something?"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<",
+      "reference": "Should this be `number` instead of `id`?"
+    },
+    {
+      "prediction": "Shouldn't this be `table->isNativeExternalTable()`?",
+      "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 think this should be `app2sim_alloc_tx` and `sim2app_alloc_rx`",
+      "reference": "Should most of these non-globals be static?"
+    },
+    {
+      "prediction": "Metrics/BlockLength: Block has too many lines. [31/25]",
+      "reference": "Please do not approve PRs without tests!!!!"
+    },
+    {
+      "prediction": "I don't think we need to pass `win_id` to `WebEnginePage` anymore.",
+      "reference": "Now that you handle this in `webenginetab.py` you can undo all the changes in this file."
+    },
+    {
+      "prediction": "I don't think it's a good idea to catch all exceptions here. It would be better to catch only `RuntimeException` and rethrow all other exceptions.",
+      "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 think this should be an error, not a warning.",
+      "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 don't understand why this change is needed. Can you explain please ?",
+      "reference": "do you really want to change the behavior on foreign key ?"
+    },
+    {
+      "prediction": "I think this should be `if (json_array_size (batch->state_trans) > 0) { event_publish_state (batch->event, batch->state_trans); json_decref (batch->state_trans); batch->state_trans = NULL; } else { json_decref (batch->state_trans); batch->state_trans = NULL; }`",
+      "reference": "Looks like my bad. Thanks for fixing!"
+    },
+    {
+      "prediction": "I think this should be called `replace_variables` instead of `no_replace_variables` to be consistent with the rest of the codebase.",
+      "reference": "I think \"Don't replace variables ...\" would be cleaner."
+    },
+    {
+      "prediction": "I think this should be a `param.Selector` instead of a `param.Parameter` since it has a fixed set of possible values.",
+      "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 don't think we need the `w.Vars[\"instance_name\"].Value` here anymore.",
+      "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 don't think we need to expose this to the public API. It's only used internally.",
+      "reference": "qq: do we want to include anything in the name to indicate that it is a position in the manifest?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be good to add a `Log.d(LOGTAG, \"Compositor pause begin\");` before the `mWindows.pauseCompositor();` call in the `onPause()` method.",
+      "reference": "These logs should probably use the `LOGTAG`"
+    },
+    {
+      "prediction": "I'm not sure if we want to use `golang.zx2c4.com/wireguard` instead of `github.com/mysteriumnetwork/wireguard-go`",
+      "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 don't think it's a good idea to show the exception message to the user. I think it's better to show a generic error message like \"Internal server error\" or something like that.",
+      "reference": "Any particular reason for this? This potentially will make harder to investigate users issues"
+    },
+    {
+      "prediction": "I think this should be `Blockly.highlightedConnection_ != closestConnection && Blockly.localConnection_ != localConnection`",
+      "reference": "Do you also need to check if Blockly.localConnection_ is non-null?"
+    },
+    {
+      "prediction": "I'm not sure I like the name `World` for this class. I think it would be better to rename it to `GlobalData` or something like that.",
+      "reference": "Not really a container, and it's not just about non-config data, not sure of a better description, @myronmarston ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think it's better to use `require(\"@aws-sdk/client-s3\").GetObjectCommand` instead of `require(\"@aws-sdk/client-s3\").default.GetObjectCommand`",
+      "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": "Why do we need to make a copy here?",
+      "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": "nit: I think this should be in the `init` function of the package, not the `init` function of this test file.",
+      "reference": "wut. We can do this at the beginning of tests if we want right? Why are we depending on init?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put the lock, but I think we need to make sure that we don't have a race condition where two clients connect at the same time and both try to set `c.acc` to `srv.gacc` at the same time. I'm not sure if this is the right place to put the lock, but I think we need to make sure that we don't have a race condition where two clients connect at the same time and both try to set `c.acc` to `srv.gacc` at the same time. I'm not sure if this is the right place to put the lock, but I think we need to make sure that we don't have a race condition where two clients connect at the same time and both try to set `c.acc` to `srv.gacc` at the same time. I'm not sure if this is the right place to put the lock, but I think we need to make sure that we don't have a race condition where two clients connect at the same time and both try to set `c.acc` to `srv.gacc` at the same time. I'm not sure if this is the right place to put the lock, but I think we need",
+      "reference": "Same trick here IMO."
+    },
+    {
+      "prediction": "I think this should be `POINTER` instead of `nil`",
+      "reference": "Should be `@origin = element || origin || POINTER`"
+    },
+    {
+      "prediction": "I think we need to check if `options.cstorSPC.spcName != \"\"` and `len(args) != 0` and throw an error.",
+      "reference": "this example needs a fix?"
+    },
+    {
+      "prediction": "I think this should be `wlr_xdg_toplevel_send_close(surface);`",
+      "reference": "Is it safe to assume this surface is a toplevel?"
+    },
+    {
+      "prediction": "This should be `projCtx pj_ctx_clone( projCtx ctx )`",
+      "reference": "we don't need that function"
+    },
+    {
+      "prediction": "I don't think we need to add a new platform for this. We should be able to use the existing `WINDOWS` platform.",
+      "reference": "we'll also have to think about adding \"Windows Server 2014\" or whatever they come up with later."
+    },
+    {
+      "prediction": "This method is not used in this PR. Can we remove it?",
+      "reference": "curious do we have an API to fetch an inactive flow?"
+    },
+    {
+      "prediction": "I'm not sure why this was here, but it seems like it should be removed.",
+      "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'm not sure if this is the best way to do this, but it seems to work for the cases I've tested.",
+      "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 you need to import `pyqtSignal` anymore.",
+      "reference": "Please import modules and not classes (except for Qt stuff)."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Hm... why do we have a \"DummyJavaNode\" in src/main? Looks like, this should have gone into src/test...."
+    },
+    {
+      "prediction": "nit: I think this should be `// Package v1alpha1` instead of `// Package v1alpha1`",
+      "reference": "This seems meaningless. If we cannot add any info let us remove this."
+    },
+    {
+      "prediction": "I think this should be `GetPortBindings` to be consistent with the other getters.",
+      "reference": "Can you rename `Ports` to `PortsUnsafe`?"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "This feels like it could be moved out to your first gem!"
+    },
+    {
+      "prediction": "Why do we need the `requestId` here?",
+      "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 if this is the best way to do this, but it seems to work.",
+      "reference": "This is super minor, but it'd be nice if this iterated over an arbitrary number of array inputs."
+    },
+    {
+      "prediction": "I think this should be `MAX_PATH` instead of `MAXIMUM_PATH`.",
+      "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": "This import is not used.",
+      "reference": "Why are we not wrapping here?"
+    },
+    {
+      "prediction": "`ParameterFilterData::parameter` can be `null` only if `ParameterFilterData::value` is `null`.",
+      "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": "46508508008000000\",\n \t\"libflux/flux-core/src/ast/visitor.rs\":                                                        \"8888888888888888888888888888888888888888888888888888888888888888\",\n \t\"libflux/flux-core/src/ast/walk.rs\":                                                           \"8888888888888888888888888888888888888888888888888888888888888888\",\n \t\"libflux/flux-core/src/ast/walk_mut.rs\":                                                       \"88888888888888888888888888888888888888888888888888888888",
+      "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 if this is the right thing to do, but I think we should remove the `GOOGLE_APPLICATION_CREDENTIALS` environment variable from the list of required environment variables in the error message below.",
+      "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 be `default:\"/var/lib/kubelet/volumes\"`",
+      "reference": "No need to speficy attribute here. Moreover in yaml it is default to use lowercase names."
+    },
+    {
+      "prediction": "nit: I think we can remove this empty line.",
+      "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 think we should use `urllib2.Request` instead of `urllib2.build_opener` and `urllib2.urlopen`.",
+      "reference": "Can rename `req` to be `opener`, since that is the object, and it wraps `req` internally?"
+    },
+    {
+      "prediction": "This should be `if \"breakpoint\" in attrs:`",
+      "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 don't think this is correct. `self._execute(Command.GET_ELEMENT_RECT)['value']` returns a dictionary with `x`, `y`, `width` and `height` keys. `self.size` and `self.location` return dictionaries with `x` and `y` keys.",
+      "reference": "This would return a tuple of two dictionaries. You need to combine them and return a dictionary"
+    },
+    {
+      "prediction": "nit: can we move this 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 think this should be `1593493828_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000",
+      "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 `std::distance(first, last) >= 9`",
+      "reference": "Nice use of std::distance. I'm a bit worried about first += 9 though."
+    },
+    {
+      "prediction": "I don't think we need to add this label.",
+      "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 right thing to do, but I think it's the only way to make this work. The problem is that `parent.GetTag(Tags.InstrumentationName)` will return `null` if the parent span was not created by the profiler. In that case, we don't know if the parent span was created by the profiler or not, so we don't know if we should ignore it or not. If we don't ignore it, we will create a new span that will be a child of a span that was not created by the profiler, which is not what we want. If we ignore it, we will not create a new span that will be a child of a span that was created by the profiler, which is also not what we want. So, I think the only way to make this work is to ignore the parent span if it was not created by the profiler. I'm not sure if this is the right thing to do, but I think it's the only way to make this work.",
+      "reference": "As long as we're being extra-vigilant about NREs, should we check that `httpMethod` and `requestUri` are not null?"
+    },
+    {
+      "prediction": "I don't think we need to add xrc20 here. xrc20 is a subcommand of transfer.",
+      "reference": "No need to have `xrc20` alias"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `SystemNamedDict` class itself.",
+      "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 `private final Boolean gpuMining = false;` so that it can't be changed after construction.",
+      "reference": "Don't call it GPU mining, call it `externalMining`, here and throughout."
+    },
+    {
+      "prediction": "I think this should be `imageMaxHeight` and `imageMaxWidth` instead of `imageMaxWidth` and `imageMaxHeight`.",
+      "reference": "`within this with` typo, should be `within this width`"
+    },
+    {
+      "prediction": "I don't think we need to disable openapi-gen for this field.",
+      "reference": "how come this is false?"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "How about `Object.prototype.toString`? Or using the `typeof` operator?"
+    },
+    {
+      "prediction": "I think this should be `ctx context.Context` and `cancel context.CancelFunc`.",
+      "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 don't think this is correct. `PlatformApis.IsDarwin` will be true for both Mono and CoreCLR on Darwin.",
+      "reference": "This check isn't right though."
+    },
+    {
+      "prediction": "I think this should be `public abstract String getCodeGenVersion();`",
+      "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": "`uploadResp` should be renamed to `response`",
+      "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 about this change. `time.Until(deadline)` will return a negative value if `deadline` is in the past, and `time.After` will block forever in that case.",
+      "reference": "This isn't really easy to read."
+    },
+    {
+      "prediction": "nit: can we use `fmt.Errorf` instead of `glog.Errorf` here?",
+      "reference": "This line changed due to go formatting. Format was not there earlier."
+    },
+    {
+      "prediction": "I think this should be `upstream.Without` instead of `staticUpstream.Without`.",
+      "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": "Shouldn't this be a 409 Conflict?",
+      "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/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "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": "I think it would be better to use `self.connection.add_timeout(5, self.connect_to_rabbitmq)` instead of `connection.add_timeout(5, self.connect_to_rabbitmq)`.",
+      "reference": "there is no static method as a go between -- how does this work?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Can you please change the name of this class to be something else?"
+    },
+    {
+      "prediction": "nitpick: I think it would be better to rename `h2o_next_token` to `h2o_next_comma_separated_token` or `h2o_next_token_with_comma_separator`.",
+      "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": "Why do we need this change?",
+      "reference": "why not recycling? I Lru cache now recycling?"
+    },
+    {
+      "prediction": "I think this should be `PaymentVersionV2`",
+      "reference": "Why it is now called `PaymentVersionV3`?"
+    },
+    {
+      "prediction": "I don't think we need to log all the nodes here.",
+      "reference": "changed for debug purpose?"
+    },
+    {
+      "prediction": "Why is this needed? `BaseProvider` already has a `SetParentProvider` method, which is called by `OpenTelemetrySdk.SetParentProvider`.",
+      "reference": "@cijothomas I needed a way to set the Exporter.ParentProvider when the Processor's gets set. Internal to SDK though."
+    },
+    {
+      "prediction": "I don't think this is correct. If you are on a motorway and you enter a ramp, you are still on a motorway.",
+      "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": "I think this should be `:q_puppetagent_server => master_certname, :q_puppetagent_certname => host_certname,`",
+      "reference": "where does `master_certname` come from?"
+    },
+    {
+      "prediction": "Please add a blank line between `SSDVGG` and `HRNet`.",
+      "reference": "The line breaking is unnecessary."
+    },
+    {
+      "prediction": "I think we should keep the old behavior, i.e. `\"\\t ,\"`",
+      "reference": "why not enable split by space here ?"
+    },
+    {
+      "prediction": "(style) 'value' may be misspelled - perhaps 'val' ?",
+      "reference": "Shouldn't the operator be && ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put these methods. I think they should be part of the `Vagrant` class, not the `Beaker::Vagrant` class. The `Beaker::Vagrant` class is just a wrapper around the `Vagrant` class, so it shouldn't have any of its own methods.",
+      "reference": "My best guess is that you want to use host.name in these parts to get the name of the host ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do. We don't need to know the nodeType when we're unmarshalling the manifest. We only need to know the nodeType when we're marshalling the manifest, and we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when we're marshalling the manifest because we don't need to know the nodeType when",
+      "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 || UserPreferences.showDownloadReport()`",
+      "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": "I think this should be `c.ic.req.Features.Clone()`.",
+      "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 think this should be `status.Errorf(codes.Unimplemented, \"GenerateAPIKey is not implemented\")`",
+      "reference": "`ctx` is unused in GenerateAPIKey"
+    },
+    {
+      "prediction": "I don't think this is the right place to log fatal error. I think it should be in the caller of `NewBlockDAO`.",
+      "reference": "Can you judge the type of error?"
+    },
+    {
+      "prediction": "I don't think this is correct. `np.convolve(data, [0.5, 0.5], 'valid')` is equivalent to `np.convolve(data, [0.5, 0.5], 'same')[:-1]`",
+      "reference": "Much simpler than a weird and confusing ``convolve`` call!"
+    },
+    {
+      "prediction": "I'm not sure if this is a bug or not, but it looks like the Suspender1 and Suspender2 render methods are being called 3 times instead of 2 times.",
+      "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": "Trailing whitespace detected.",
+      "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 should be changed.",
+      "reference": "Will this mess up parsing the messages by other editors? (vscode, vim ...)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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 place to return the block to the pool, but it seems to be the only place where we can guarantee that the block is no longer in use.",
+      "reference": "Why can the pool be null?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to add a `readonly` attribute to the `List` class, and then override `_get_states` to add `controlTypes.STATE_READONLY` if `readonly` is `True`.",
+      "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": "This should be `Set<DigraphNode<E>>`",
+      "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 think this should be `pull.frontier.is_zero ()` instead of `pull.head.is_zero ()`",
+      "reference": "There doesn't seem to be a reason to copy this here."
+    },
+    {
+      "prediction": "I don't think we need to include `h2o/ebpf.h` here.",
+      "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 this should be `AlertQueryURL` to be consistent with the rest of the struct.",
+      "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": "nit: I think we can remove the `m.externalNodeCIDRs` field from the `ipipManager` struct, since it's only used here.",
+      "reference": "Same change just above?"
+    },
+    {
+      "prediction": "I'm not sure about this test. I think it should be `expect((await engine.search('n')).length).toBe(2);` and `expect((await engine.search('\u043b')).length).toBe(2);`",
+      "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": "nit: I think it would be nice to have a constant for the `example.org` domain. It's used in a few places in this file.",
+      "reference": "nit: I think that `workloadID` should suffice for this test... that's what we'd be issuing anyways"
+    },
+    {
+      "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 don't think we should catch `NoClassDefFoundError` here. It's a subclass of `LinkageError` which is a subclass of `Error` which is a subclass of `Throwable` which is a subclass of `Exception` which is a subclass of `Throwable` which is a subclass of `Object` which is a subclass of `Serializable` which is a subclass of `Cloneable` which is a subclass of `Runnable` which is a subclass of `Thread` which is a subclass of `Object` which is a subclass of `Serializable` which is a subclass of `Cloneable` which is a subclass of `Runnable` which is a subclass of `Thread` which is a subclass of `Object` which is a subclass of `Serializable` which is a subclass of `Cloneable` which is a subclass of `Runnable` which is a subclass of `Thread` which is a subclass of `Object` which is a subclass of `Serializable` which is a subclass of `Cloneable` which is a subclass of `Runnable` which is a subclass of `Thread` which is a subclass of `Object` which is a subclass of `Serializable` which is a subclass of `Cloneable` which is a subclass of `Runnable` which",
+      "reference": "We should have a rule to detect identical catch branches"
+    },
+    {
+      "prediction": "Why are these removed?",
+      "reference": "remove these from the targets file."
+    },
+    {
+      "prediction": "why is this removed?",
+      "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": "Can you update the version of `google-auth-httplib2` to `0.0.4` as well?",
+      "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": "Please revert this change.",
+      "reference": "I usually set idea to never use wildcard import such as `import java.util.*;` Now i use setting from javaslang standard."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Where is host['ips'] coming from?"
+    },
+    {
+      "prediction": "I don't think we need to return anything here.",
+      "reference": "oh! missed a `:` here; let me fix that; also can get rid of the intermediate variable"
+    },
+    {
+      "prediction": "Line is too long. [84/80]",
+      "reference": "Is this necessary?"
+    },
+    {
+      "prediction": "I don't think we need to import `makeLogin` and `isTokenExpire` here.",
+      "reference": "I'd create a different method for each render section."
+    },
+    {
+      "prediction": "I think it would be better to make this a field of the `OperationTracer` interface, and have all implementations set it to `false` by default, and then override it in `EstimateGasOperationTracer` to be `true`.",
+      "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 `readOrcSchema` parameter from 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 think this should be `fromProjectionPos[i] = 13;`",
+      "reference": "These modifications allow BaseFile to translate into a SparkRow with the specID as a column"
+    },
+    {
+      "prediction": "I think we can remove the `typeof` check and just do `return new ReadPreference(readPreference);`",
+      "reference": "is this something we've been missing this whole time?"
+    },
+    {
+      "prediction": "I think this should be `allEvents` 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_thrift = os.path.join(thrift_files_dir, 'authentication.thrift')`",
+      "reference": "There seems to be some repetition. Does a local function make this code shorter overall?"
+    },
+    {
+      "prediction": "I think this should be `isHydrating = true;`",
+      "reference": "Might be cheaper to reuse the `null` assignment of line 313 and set `isHydrating` to null instead WDYT?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Revert this blank line"
+    },
+    {
+      "prediction": "I don't think we need to clear the properties here.",
+      "reference": "This was introduced recently for the Hadoop 3 upgrade. Copy/paste error but definitely causing some of the new test failures."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [118/100]",
+      "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 a good test. If you want to test that the response has a `Content-Length` header, you should use `resp.headers.get('Content-Length')` and assert that it is not `None`.",
+      "reference": "I wonder if we should create new tests for header checks; here for instance we're mixing querystring and headers. Thoughts?"
+    },
+    {
+      "prediction": "I think this should be `apply::helpers::custom_facts`",
+      "reference": "We should be able to add sensitive by hard-coding the parameters, same as you put into the metadata in apply_helpers."
+    },
+    {
+      "prediction": "This file is empty, so it can be removed.",
+      "reference": "Why do you think that it is necessary to create this class?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we should ignore errors here.",
+      "reference": "replacing the `1` with `io.SeekCurrent` may be good here"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it would be better to have it in the `jwks` package and have the `jwks` package implement this interface.",
+      "reference": "nit: perhaps this file would be better named `jwks_source.go` ?"
+    },
+    {
+      "prediction": "I think this should be moved to `AzkabanCommonConfigurationKeys`.",
+      "reference": "It will be helpful to specify the intended priority as well for the properties within this file."
+    },
+    {
+      "prediction": "nit: I think we can just use `func(*exec.Cmd) ([]byte, error)` instead of defining a new type.",
+      "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 \"/user/trafodion/hive/tpcds/customer/customer.dat\" instead of \"/user/trafodion/hive/tpcds/customer/customer.dat\"",
+      "reference": "Do you now why is it a specific table name is used as a default?"
+    },
+    {
+      "prediction": "I think this should be `Path.Combine(Environment.GetEnvironmentVariable(\"HOME\"), \"myagent/_work/_temp\")`",
+      "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 think we can remove the `v2` import here and just use `generated` everywhere.",
+      "reference": "nit- would be cleaner to import it once"
+    },
+    {
+      "prediction": "I think this should be `x-amz-user-agent, x-amz-date, x-amz-security-token, x-amz-content-sha256, authorization`",
+      "reference": "Looks like there is a closing parenthesis `)` missing here."
+    },
+    {
+      "prediction": "I don't think you need the `128` here. The `str` array is already declared as `char str[128]`.",
+      "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 don't think this is necessary.",
+      "reference": "You can put this in `kbfsOpsInit()` in you want, so we don't have to call it in every test."
+    },
+    {
+      "prediction": "I don't think we need the `#ifdef ARM_32` here. The `#ifdef ARM_32` is already in the `#if defined(AARCHXX)` block.",
+      "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 we can remove this field. It's only used in the `toString()` method, which is only used for debugging, so we can just use `schema` instead.",
+      "reference": "s/for use/to use for/"
+    },
+    {
+      "prediction": "Shouldn't this be `avatar_id` instead of `avatar_filename`?",
+      "reference": "` $event->getAvatarFilename()` this function return `string` or `null`. In this place ` $event->getAvatarFilename()->getValue()` return `Fatal error `"
+    },
+    {
+      "prediction": "I don't think we should change this to unavailable. This is not a transient error, it's a permanent one.",
+      "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.Value`",
+      "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 this line is not needed anymore.",
+      "reference": "Unrelated file. Please remove from PR."
+    },
+    {
+      "prediction": "I think this should be `send_worker_msg(q, w, \"Access-Control-Allow-Origin: *\\n\\n\");`",
+      "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 would be better to do this in `mitmproxy/proxy/protocol/http.py` in the `handle_request` method.",
+      "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": "I think this should be `requiredOwned: ['textbox', 'listbox', 'tree', 'grid', 'dialog'],`",
+      "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": "I don't think you need to create a new CostMethodHbaseDelete for each HbaseDelete. I think you can just create it once and reuse it.",
+      "reference": "maybe say \"returns the default cost method that returns an object of cost 1\"."
+    },
+    {
+      "prediction": "@danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship",
+      "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 `response.asString()` instead of `response.fullResponseAsJSONObject()`.",
+      "reference": "Use `response.asJsonObject()` instead. Also, use `put()` instead of `putOpt()`, `null` as a value is fine."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Are you sure? We've changed it during an introduction of the IME support."
+    },
+    {
+      "prediction": "I think this should be `!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 think we should keep the `err` variable here, and return `err` at the end of the function.",
+      "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 this should be `(nullable)` instead of `(allow-none)`.",
+      "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": "I think it should be `(labels >= 1) & (labels < label_channels)`",
+      "reference": "If the label is not binary, should we rename this function?"
+    },
+    {
+      "prediction": "I don't think you need to use BeanShell for this. You can just use `${__BeanShell(File.separator, )}`.",
+      "reference": "This is very-very bad idea because of its performance impact"
+    },
+    {
+      "prediction": "Please remove empty line.",
+      "reference": "WHY U NOT SORT IMPORTS?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this helper method, but I think it makes sense to have it in the `SignInRequestHelpers` module since it's related to the sign in request.",
+      "reference": "Put a comma after the last item of a multiline hash."
+    },
+    {
+      "prediction": "I think this should be `from .core.dimension import Dimension, Dimensioned`",
+      "reference": "How come we need ``Dimensioned`` in the top-level namespace?"
+    },
+    {
+      "prediction": "Please revert this change. It's unrelated to the PR.",
+      "reference": "excess whitespace through plugin."
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "What is the problem here? I feel we used this in a lot of places."
+    },
+    {
+      "prediction": "I think it is better to use `torch.stack(proposal_list).cpu().numpy()` instead of `[proposal.cpu().numpy() for proposal in proposal_list]`",
+      "reference": "Update the docstring."
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "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() : 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 alias was added in the first place. I don't see it being used anywhere.",
+      "reference": "Is this a breaking change, or is the API private?"
+    },
+    {
+      "prediction": "I don't think you need to do this. You can just do `test_data = pd.read_json(\"./dumps/courseData.json\").groupby('model').apply(lambda x: x.groupby('pk')['fields'].apply(list).apply(lambda x: x[0]).to_dict())`",
+      "reference": "I don't think we need this dependency in this project right now"
+    },
+    {
+      "prediction": "I think this should be `knative-eventing` instead of `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 don't think we need to add the `-r puppet` to the example.",
+      "reference": "The `_example` field should be updated to include an interpreter with an array value."
+    },
+    {
+      "prediction": "I don't think we need to do this.",
+      "reference": "Put this in the godoc of SupportedProtocols instead. Preferably we would just use one map though. Why not just use this one?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "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": "This import is not used.",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "You can safely revert changes to this file."
+    },
+    {
+      "prediction": "This is a syntax error.",
+      "reference": "This deletion is unnecessary."
+    },
+    {
+      "prediction": "I think this should be `smallImage && typeof smallImage === 'string'`",
+      "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": "I think we should avoid importing `java.util.*` and instead import only the classes we need.",
+      "reference": "I've had this happen recently with IntelliJ, was this intentional?"
+    },
+    {
+      "prediction": "I don't think you need to include hash_util.h here. You can just use std::hash.",
+      "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 it's better to use `github.com/mysteriumnetwork/node/core/location/factory` instead of `github.com/mysteriumnetwork/node/core/location/factory/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": "I think this should be `engineType + \" child class is not implementing this function\\n\");`",
+      "reference": "Use `UpperCamelCase` for function names"
+    },
+    {
+      "prediction": "I think this should be `Number(value) <= 275761`",
+      "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 don't think we need to do this. The `ZkStateReader.CLUSTER_STATE` constant is only used in `ZkStateReader` and `SolrZkClient`, both of which have been removed from the codebase.",
+      "reference": "Remember to close SOLR-11877 after this"
+    },
+    {
+      "prediction": "I'm not sure about this change. I don't think we want to remove the `.` and `,` from the list of ignored characters.",
+      "reference": "this change should be reverted"
+    },
+    {
+      "prediction": "I don't think we need this method at all. I think we can just remove it from the interface.",
+      "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 don't think we should add this here. This is only used for the e2e tests, and we don't want to include this file in the e2e tests.",
+      "reference": "`.wav`?! cannot we use something slightly more modern and compressed instead? :) I dunno, like `.mp3` or `.aac` or `.ogg`..."
+    },
+    {
+      "prediction": "This should be `luaA_object_push(L, swap);`",
+      "reference": "Wouldn't a single signal call be enough? (without the boolean even)"
+    },
+    {
+      "prediction": "I don't think we need the `&nbsp;` here.",
+      "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": "Style/TrailingWhitespace: Trailing whitespace detected.<br>Metrics/LineLength: Line is too long. [114/100]",
+      "reference": "Should we expose this attribute in the Asciidoctor.js API ?"
+    },
+    {
+      "prediction": "I think this should be `Objects.equals(v, element._2)`.",
+      "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 the default value should be 10.",
+      "reference": "delete this line"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [85/80]",
+      "reference": "Presumably we no longer need `force: true` here."
+    },
+    {
+      "prediction": "I don't think you need to check for `NULL` here. `ptr` is a pointer to a pointer, so it will never be `NULL` or `0`.",
+      "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 don't think we need to pass in `help_url('')` here, it will default to `help_url` without an argument.",
+      "reference": "these urls are still in `a` tags so we should probably include these links separately for a non-HTML version"
+    },
+    {
+      "prediction": "I don't think we need to release the lock before calling `addSystemAccountExports` and `sys.resetCh <- struct{}{}`.",
+      "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": "This should be `CHROMIUM_ENABLE_DOWNLOADING`",
+      "reference": "The name `enableDownloading` implies this is a boolean capability. How about `downloadDir`?"
+    },
+    {
+      "prediction": "I don't think we should deprecate this method.",
+      "reference": "Found one internal team is using this method. Will coordinate with them about the migration solution."
+    },
+    {
+      "prediction": "I think this should be `DD_TRACE_HTTP_SERVER_ERROR_STATUSES` and `DD_TRACE_HTTP_CLIENT_ERROR_STATUSES`",
+      "reference": "The field `HttpServerErrorCodes` should be called `HttpServerErrorStatuses`"
+    },
+    {
+      "prediction": "I think this should be `FORSETI_V2_RULE_FILES` and `FORSETI_V1_RULE_FILES` should be renamed to `FORSETI_RULE_FILES`.",
+      "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 don't think this is correct. The cleanup flag should be passed to the `createBaseZnode` method, not to the `createConfigSet` method.",
+      "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 `archive.File[0].FileInfo().Name()` instead of `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'm not sure if this is the right place to put this, but I think it makes sense to show the profile warning on all pages except the user's own profile page.",
+      "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 don't think you need to cast to `Long` and then back to `Integer`.",
+      "reference": "This isn't necessarily a year. It may be months, days, or hours. Can we return `intValue()` directly instead?"
+    },
+    {
+      "prediction": "Why remove this log statement?",
+      "reference": "this is redundant because it gets logged immediately on entering the handleContainerChange function"
+    },
+    {
+      "prediction": "I think this should be `e == model.NotificationEventType_EVENT_DEPLOYMENT_WAIT_APPROVAL.String()`",
+      "reference": "`ds.GenericDeploymentConfig.DeploymentNotification` in L128 is nullable."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "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'm not sure if we want to do this, but I think it's a good idea to make sure we don't accidentally pass `null` to `installExtension` or `uninstallExtension`.",
+      "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": "Style/EachWithObject: Use `each_with_object` instead of `each` with a `result = {}` initializer.",
+      "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'm not sure why this is necessary, but it seems to be required for the test to pass.",
+      "reference": "Please check the return on these."
+    },
+    {
+      "prediction": "I think this should be `if (uplink && uplink.config && uplink.config.cache) {`",
+      "reference": "Do we need `==` for true?"
+    },
+    {
+      "prediction": "I think we should keep the default value as 0.0002.",
+      "reference": "Changing the default value may cause BC-breaking. Suggest changing this value in config."
+    },
+    {
+      "prediction": "I don't think this is correct. `slider_pos` is a `float`, so `int(slider_pos)` will always be `0`.",
+      "reference": "how does that happen?"
+    },
+    {
+      "prediction": "I don't think you need to import `doctest` here.",
+      "reference": "This is gross/clever. :-)"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "Take a look at the name of this file. Definitely copy and"
+    },
+    {
+      "prediction": "nit: can we use `return err` instead of `return` here?",
+      "reference": "Can you check with @mahebbar how to work this error. Should be different from 404."
+    },
+    {
+      "prediction": "I don't think we need to remove the upper bound here. I think it should be `>=3.5,<3.9`",
+      "reference": "Do we still need the upper bound `<3.9`?"
+    },
+    {
+      "prediction": "I don't think we need to check for `psalter` here.",
+      "reference": "Does this mean `vendor/bin/psalm` is not a symlink (or whatever equivalent Windows has for symlinks) on Windows?"
+    },
+    {
+      "prediction": "I think we can remove this file.",
+      "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": "Why are these removed?",
+      "reference": "The exporter should still be able to test these histogram instrument kinds, right? Is there another reason to remove these?"
+    },
+    {
+      "prediction": "I don't think we need the `validateTxnBytes` parameter here.",
+      "reference": "should this increment eval.Round somehow?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Although this change looks good. Maybe there are other inconsistent cases too?"
+    },
+    {
+      "prediction": "Style/AccessorSyntacticSugar: Use attr_reader instead of a class method with only a getter.",
+      "reference": "This is a way better method name."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be nice to have a `t(\"marketing.show.language_flash\")` in the `marketing.en.yml` file.",
+      "reference": "Prefer single-quoted strings when you don't need string interpolation or special symbols."
+    },
+    {
+      "prediction": "I don't think this is necessary. `Array.isArray` will return `false` if `options` is `null` or `undefined`.",
+      "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 don't think we need to set this to FALSE here. It will be set to FALSE in the constructor of the base class.",
+      "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 we should be importing `org.openqa.selenium.remote.tracing.*` here.",
+      "reference": "Could you please leave the explicit imports?"
+    },
+    {
+      "prediction": "why change the import name?",
+      "reference": "Why need alias?"
+    },
+    {
+      "prediction": "I think this should be `return c.tlfIDGetter`",
+      "reference": "Shouldn't this be a `libfs.nullIDGetter` (exported, of course)?"
+    },
+    {
+      "prediction": "I don't think we should be showing the size of the program in the error message. The user doesn't need to know the size of the program, they just need to know that the program is too large.",
+      "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": "why remove these two lines?",
+      "reference": "Seems we don't need it anymore."
+    },
+    {
+      "prediction": "I think this should be `if: :published?` instead of `unless: :published?` since we want to validate the presence of `themes` only when `published` is true.",
+      "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 don't think this is a good default. I think it should be 10s or 30s.",
+      "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 this is the right place to do this. I think it would be better to do this in `caddy/caddy.go` in the `caddy.NewReplacer()` function.",
+      "reference": "I figure to make `{env.*}` work, right? I can't think of any other placeholder that makes sense here"
+    },
+    {
+      "prediction": "Why do you need to change the port?",
+      "reference": "Why change the port from 0 to getAvailablePort()?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think we need to add a `timeout` parameter to the `get_locator` method and use that instead of hardcoding 30.0.",
+      "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": "I think we can remove the `projectName` field from the `GlobalOpts` struct. We don't use it anywhere.",
+      "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. If `bytes` is `null` then `bytes.Length` will throw `NullReferenceException`.",
+      "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'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "nit: can remove this `else` since we return from the condition above."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "You missed `Write<T>(...)`! I'm kidding ofc."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's better to be explicit about the fact that we don't want to do anything if `subscription.stripe_customer_id` is `nil` or an empty string.",
+      "reference": "Was there a customer who had this set to an empty string?"
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "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": "I don't think we should add this method to this interface. I think it should be added to the `RestClientRequestImpl` class.",
+      "reference": "it's better to be : void attach(String name, Part part);"
+    },
+    {
+      "prediction": "Please add throws IOException to this method as well.",
+      "reference": "I didn't see where the IOException can get thrown, is this just to match the V1 reader?"
+    },
+    {
+      "prediction": "Why is this parameter not used anymore?",
+      "reference": "We normally suppress this warning with `RDUSED_PARAM(forceRDKit)`"
+    },
+    {
+      "prediction": "I think 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 \"Requires Patch\" instead of \"Patch Required\".",
+      "reference": "Why upper case \"R\"?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `required` variable and just return `mf.BuildRequired()`.",
+      "reference": "Does this need to return an error or could it return `false, nil`?"
+    },
+    {
+      "prediction": "I don't think we need to make this protected.",
+      "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": "I don't think we should remove the default value here. If the parameter is not set in the URL, it should default to 1.",
+      "reference": "this parameter is only used in POST, please use `getRequest()->request->get('...');` Thanks"
+    },
+    {
+      "prediction": "I don't think this is correct. `validatePackage` expects a package name, not a filename.",
+      "reference": "Problem number 1: Scoped packages would have a `/` character here. Changing this to `validatePackage` resolves the 403."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the key is not valid, it should raise an error, not silently ignore it.",
+      "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 this should be `if req.ChainID != 0 && _, ok := registry.Chains()[req.ChainID]; !ok {`",
+      "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 think this should be `return ctx` instead of `return ContextWithRemoteSpanContext(ctx, sc)`.",
+      "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": "I think this should be `paginated = false` by default.",
+      "reference": "here shouldn't we use `depotFreeLimit`?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Could we use the Rails application's secret token instead? Do we actually need our own here?"
+    },
+    {
+      "prediction": "`lambda: kdf.a.groupby(kdf.a, kdf.b))` -> `lambda: kdf.a.groupby(kdf.a, kdf.b)`",
+      "reference": "so should be fixed here also"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it would be better to just return false and let the caller know that the mcontext isn't fully filled in.",
+      "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/` instead of `/var/lib/ecs`",
+      "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 `ipsecESPHeaderOverhead` and `ipsecESPTrailerOverhead`.",
+      "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 think this should be `where T : unmanaged` instead of `where T : struct`.",
+      "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 we can remove this empty line.",
+      "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 don't think this is correct. If you configure a request logger, it will log all requests, including health checks. If you don't want to log health checks, you need to configure a request logger that doesn't log health checks.",
+      "reference": "This seems like a more appropriate message under `ListenAndServe`. WDYT?"
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "reference": "Is this not required?"
+    },
+    {
+      "prediction": "Missing backtick at the end of the line.",
+      "reference": "We should start thinking of an object that encapsulates this information :) we might need more arguments in the future."
+    },
+    {
+      "prediction": "This should be `01173` and `01174`, not `01174` and `01173`.",
+      "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": "`isMining` is not the same as `isRunning`. `isMining` means that the node is actively mining a block, while `isRunning` means that the node is configured to mine blocks. I think we should rename `isMining` to `isMiningActive` and `isRunning` to `isMiningEnabled` or something like that.",
+      "reference": "rename? MiningCoordinator always exists"
+    },
+    {
+      "prediction": "This is not the correct way to get the config. You need to use the `config` object that is passed to the `init` function of the plugin.",
+      "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 don't think we need to lock here, since we already have a lock on the folder.",
+      "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` argument from `sharedSetup` now that it's not used.",
+      "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, but I think we should also add a redirect for `/videos/vim-for-rails-developers` to `https://www.youtube.com/watch?v=9J2OjH8Ao_A`",
+      "reference": "Line is too long. [97/80]"
+    },
+    {
+      "prediction": "I think we can remove the `ANY` import now that we're not using it.",
+      "reference": "This seems unused."
+    },
+    {
+      "prediction": "I don't think we should panic here. We should return an error and let the caller handle it.",
+      "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'm not sure this is the right place for this function.",
+      "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": "This should be `VUID-vkCmdBindDescriptorSets-pipelineBindPoint-00362`",
+      "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 `:title` instead of `:name`.",
+      "reference": "This will need to be updated in the upcase-exercises repo as well."
+    },
+    {
+      "prediction": "I think this should be `Status::SUBMITTED` instead of `Status::COMPLETE`.",
+      "reference": "Think it's worth extracting this to `Exercise#complete?`?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "This is still a potentially useful public method - should we remove such methods?"
+    },
+    {
+      "prediction": "I don't think this is a good default. I think it should be `UNSET_RELEASE_LEVEL`.",
+      "reference": "I don't understand this change, what is happening here?"
+    },
+    {
+      "prediction": "I don't think you need to split the version here. You can just do `version = int(self.obj.excelCellObject.Application.Version)`.",
+      "reference": "I think its worth stating this is `versionMajor`"
+    },
+    {
+      "prediction": "I don't think we need this. We can just call `config.PreloadConfigurableConsensusProtocols(nc.algodDataDir)` in the `NodeController` constructor.",
+      "reference": "rebuild: say loads and merges"
+    },
+    {
+      "prediction": "I think we can remove this import as well.",
+      "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 to `local[4]`.",
+      "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 if this is the right thing to do, but I think we should be using `mock.HELLO` instead of `mock.DEFAULT_ISMASTER_36` here. `mock.DEFAULT_ISMASTER_36` is not a valid `ismaster` response.",
+      "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 this is the right thing to do. If we fail to suspend a thread, should we just give up and not detach at all?",
+      "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": "I think this should be `--mac_installer_signing_identifier <id>`, not `--mac_installer_signing_identifier <id>`.",
+      "reference": "think it should only be in create_dist"
+    },
+    {
+      "prediction": "nit: extra space after `if int(cport.HostPort) > 0 {`",
+      "reference": "Do you plan to support Pod spec change? Like hostPort is added/removed later after Pod creation?"
+    },
+    {
+      "prediction": "I think this should be `dokan.ExistingFile` instead of `dokan.ExistingDir`.",
+      "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\" << std::endl;`",
+      "reference": "What about just printing an empty list in this case? That would make the output more predictable."
+    },
+    {
+      "prediction": "I think we can remove this line as well.",
+      "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": "I think this should be `return locate_template('/templates/searchform.php', true, false);` instead of `locate_template('/templates/searchform.php', true, false); $form = ''; return $form;`",
+      "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": "Please remove this unused use statement.",
+      "reference": "Fixes test failed in some cases"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I think it should be added back in.",
+      "reference": "what's the prod url for this? just curious. @amirsaber"
+    },
+    {
+      "prediction": "I think this should be `azkaban.job.ignore.jvm.args` instead of `ignore.jvm.args`.",
+      "reference": "Can you please change this to azkaban.jobs.java.opts?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's better to keep the `sprintf` as it is, and just add the `Source:` string before it.",
+      "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": "I don't think we need the `cancellationToken` parameter here.",
+      "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": "Why is this check needed?",
+      "reference": "Would it be equivalent to check if `chain != nullptr` below? Not suggesting a change, just curious."
+    },
+    {
+      "prediction": "I think this should be `this->wallets.queue_work_regeneration (std::chrono::steady_clock::now () + std::chrono::seconds (1), block);`",
+      "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": "I'm not sure about this change. I don't think we need to escape the single quote here.",
+      "reference": "And here. Please, use `'` for apostrophes in changed strings in this file."
+    },
+    {
+      "prediction": "I don't think we should use `SimulateExecution` here. We should use `ExecuteContractRead` instead. `SimulateExecution` is used to simulate the execution of a transaction on the blockchain. `ExecuteContractRead` is used to simulate the execution of a read-only function on a smart contract.",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "I think this should be `DefaultNewBlockDataVersion(c Config, holes bool) DataVer` instead of `DefaultNewBlockDataVersion(c Config, holes bool) DataVer`.",
+      "reference": "Please move this to `block_types.go`."
+    },
+    {
+      "prediction": "I don't think these attributes are needed anymore.",
+      "reference": "So this is not needed anymore for all attributes?"
+    },
+    {
+      "prediction": "Shouldn't this be `Failed to get OpenTelemetry Tracer from Service Provider: '{0}'. OpenTelemetry tracing will not work.`?",
+      "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": "I think this should be `update_listener=update_listener` instead of `update_listener=update_listener`",
+      "reference": "I think we can remove this function, right? (duplicate with `ses_starter.py`)"
+    },
+    {
+      "prediction": "`alias` is not a pandas method, it's a spark method. I think it should be renamed to `spark_alias` or something like that.",
+      "reference": "With this fix, I am not even sure we need to overwrite this function."
+    },
+    {
+      "prediction": "I think this should be `os.getenv('QUILT_USE_TQDM', 'false').lower() == 'true'` so that it defaults to `false` if the environment variable isn't set.",
+      "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_port=>8140, :puppet_agent_port=>8140, :puppet_server_certname=>\"pe-ubuntu-lucid\", :puppet_agent_certname=>\"pe-centos6\"}}\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",
+      "reference": "You've got an absolute path here referencing your home dir :-)."
+    },
+    {
+      "prediction": "I don't understand why this change is needed.",
+      "reference": "Maybe I should do some sorting so that these properties don't keep moving around..."
+    },
+    {
+      "prediction": "I think this test should be kept, but it should check that the error returned is the one we expect.",
+      "reference": "Can we keep this test anyway, as it's part of the spec?"
+    },
+    {
+      "prediction": "Please remove empty line at the end of the file.",
+      "reference": "Can be removed"
+    },
+    {
+      "prediction": "This should be `use Config\\Config;` and `use Request\\Request;`",
+      "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 place to set the charset. The charset should be set in the Content-Type header of the response, not in the Content-Type header of the request.",
+      "reference": "It's better to get the charset from context or other setting to let the user override it."
+    },
+    {
+      "prediction": "Why is this change required?",
+      "reference": "this filename as well needs change"
+    },
+    {
+      "prediction": "Shouldn't this be `AccessKeyIDSecretRef` instead of `AccessKeyIDRef`?",
+      "reference": "I think we want to remove `omitempty` here"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think we need to make sure that the signal handler is closed before we return from this function.",
+      "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 we should remove the implicit role here. I think we should keep it, but I'm not 100% sure. @elrumordelaluz, do you have any thoughts on this?",
+      "reference": "I updated only the implicit roles who needed a update for this new rule to validate."
+    },
+    {
+      "prediction": "I think we can remove these two lines.",
+      "reference": "> // Code generated by protoc-gen-go. DO NOT EDIT."
+    },
+    {
+      "prediction": "I think this should be `skipOSConfig = map[string]string{\"osconfig-not-supported\": \"true\"}`",
+      "reference": "minor: skipOSConfigMetadata, otherwise, the name sounds like a boolean"
+    },
+    {
+      "prediction": "I don't think this is correct. `SyndTypeUtils.enclosureTypeValid` returns `false` if the type is `null`.",
+      "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 think it would be better to return an empty slice if there are no sessions, instead of returning an error.",
+      "reference": "`sessions := endpoints.SessionsDTO{}` and using only `sessions` should be simpler."
+    },
+    {
+      "prediction": "I don't think we need the `|| dcontext->is_exiting` here. I think we can just assert that `dcontext != NULL`.",
+      "reference": "These need to be in the other order to avoid crashing when dcontext == GLOBAL_DCONTEXT (==-1)"
+    },
+    {
+      "prediction": "Why is this change necessary? I don't see it in the PR description.",
+      "reference": "Why not change the other samples?"
+    },
+    {
+      "prediction": "I'm not sure if we need to escape the `id` here, since it's not being used in a `querySelector` or `querySelectorAll` call.",
+      "reference": "Indentation is mixed up here due to spaces/tabs, I'm guessing."
+    },
+    {
+      "prediction": "I think this should be `if (bytes_read != sizeof(count))`",
+      "reference": "Should we provide a little more context in the error message? Maybe throw in something about what operation was being performed?"
+    },
+    {
+      "prediction": "This should be `if (isset($configuration['settings'])) {`",
+      "reference": "Project standard is spaces, not tabs. Please fix and then we can merge this."
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "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 think we should keep the `partFilter` argument here. It is used in `PartialManifest` to filter out partitions that don't match the partition spec.",
+      "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/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>",
+      "reference": "I think BSON::Decimal128 should always be serialized as Decimal128 regardless of config option."
+    },
+    {
+      "prediction": "I think this should be `require_relative 'bolt/plan_creator'` since it's in the same directory as this file.",
+      "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'm not sure if this is the right place to do this, but I think it would be nice to have a way to specify the number of warmup iterations and the number of target iterations in the `Options` class.",
+      "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 this function should be renamed to `FinishMigrateRepositoryWithUpdateHook` or something like that.",
+      "reference": "Better call it `CleanUpMirrorInfo`?"
+    },
+    {
+      "prediction": "I think this should be `node.RetrievalProtocol.RetrievalProvider.Start(ctx)`",
+      "reference": "@shannonwells is there still something missing here?"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "This could be more indicative of what it is... e.g... `stresstest_time_limit`"
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "reference": "Shouldn't we do a `reader_payload_pool_.reset()` here?"
+    },
+    {
+      "prediction": "I think this should be `BeaconEntries` instead of `DrandEntries`",
+      "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 think this should be `self._current_key` instead of `getattr(self, '_current_key', None)`",
+      "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 don't think these need to be constants. They are only used in this file.",
+      "reference": "I guess I like the name `mode` okay. I can't think of anything better."
+    },
+    {
+      "prediction": "I don't think this is correct. If timeout is less than or equal to zero, we should not use the default timeout.",
+      "reference": "I would create a new method (NewStepWithDefaultTimeout?) for this purpose because timeout<=0 looks like a hidden logic"
+    },
+    {
+      "prediction": "`limit` should be `int` instead of `string`.",
+      "reference": "Are these limit changes related?"
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "reference": "Don't you mean `import scapy.consts`?"
+    },
+    {
+      "prediction": "I think this should be `model.StageStatus_STAGE_FAILURE`?",
+      "reference": "`ctx` is unused in rollbackPrimary"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "This is not what title says."
+    },
+    {
+      "prediction": "I'm not sure about the name of this interface. Maybe `RuleViolationFixer` or `RuleViolationOperation` would be better?",
+      "reference": "a fix applies several fixes? maybe this should simply be `applyToNode`"
+    },
+    {
+      "prediction": "nit: `fakeListOk` -> `fakeListFnOk` or `fakeListOkFn`",
+      "reference": "U1000: func `fakeGetOk` is unused (from `unused`)"
+    },
+    {
+      "prediction": "nit: I think we can remove the empty line here.",
+      "reference": "Because there is a \"k8s.io/client-go/tools/cache\" below, we can sort out all the \"imports\" here."
+    },
+    {
+      "prediction": "nit: please remove the extra blank line",
+      "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 `if o.PipelineSecret != \"\" && o.PipelineStack != \"\"`",
+      "reference": "Do we not have tests for pipeline delete"
+    },
+    {
+      "prediction": "Should this be `call_check` instead of `self.call_check`?",
+      "reference": "Good that you mention this constraint in the docstring :)"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "remove this blank?"
+    },
+    {
+      "prediction": "I think this should be set to 0, not -1. -1 is not a valid number of active partitions.",
+      "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.qutescheme import QuteScheme`",
+      "reference": "Please remove this blank line - those are only used to group Python/third-party/qutebrowser imports."
+    },
+    {
+      "prediction": "I think this should be `var insecure bool = false` to make it explicit that the default value is `false`.",
+      "reference": "`insecure` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "Shouldn't this be `if setting.Repository.DisableHTTPGit`?",
+      "reference": "Same as above (HTTP in upper-case)"
+    },
+    {
+      "prediction": "I think this should be `https://www.youtube.com/watch?v=123456`",
+      "reference": "This is getting repetitive  maybe you could introduce an abstract test case with a `getExpectedUrl($providerReference)` method ?"
+    },
+    {
+      "prediction": "I think this should return a `reflect.Value` instead of an empty `reflect.Value`.",
+      "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": "I don't think this is correct. `nameLower` is not guaranteed to be in `window`. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower` is the lowercased version of `name`. `name` is the original name. `nameLower`",
+      "reference": "Does this work for Custom Elements? iirc we're lacking test cases for them. /cc @andrewiggins @developit"
+    },
+    {
+      "prediction": "nit: I think we can remove the `m.externalNodeCIDRs` field from the `ipipManager` struct, since it's only used here.",
+      "reference": "Same change just above?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to be the only way to get this to work in IE11.",
+      "reference": "I suppose `apploader.js` isn't used by WebWorkers. So `self` will always be `window` here."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I think it's better than what we had before.",
+      "reference": "Since this var is no longer defined we should remove the `if skip_targets.any?` bit below."
+    },
+    {
+      "prediction": "I think this should be `ARN_TO_LAMBDA.get(arn).qualifier_exists(qualifier)`",
+      "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 this is the right thing to do here. I think we should be able to remove the NATProxy altogether if we don't need it, but I'm not sure how to do that.",
+      "reference": "Why session is started at all, if you need to shut it down e.g. DI should launch noopSession"
+    },
+    {
+      "prediction": "I think this should be `DeleteWorkspaceFile() error` instead of `DeleteAll() error`",
+      "reference": "nit: can we rename the interface to `wsFileDeleter`"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea to change the type of `FromAddresses` and `ToAddresses` from `v1beta1.GroupMemberPodSet` to `v1beta1.GroupMemberSet`. I think we should keep the type of `FromAddresses` and `ToAddresses` as `v1beta1.GroupMemberPodSet` and only change the type of `Pods` from `v1beta1.GroupMemberPodSet` to `v1beta1.GroupMemberSet`.",
+      "reference": "Why target cannot be external endpoints?"
+    },
+    {
+      "prediction": "I think this should be in `CollectionConfig` instead of `CollectionAdminParams`.",
+      "reference": "We use a suffix \".AUTOCREATED\" for configsets, maybe we can use the same here?"
+    },
+    {
+      "prediction": "I think we should keep the `Default` in the name of the method to make it clear that this is the default instance of `OpenTelemetrySdk`.",
+      "reference": "This one won't be disposed. Should be (something like) `using var openTelemetry = OpenTelemetrySdk.EnableOpenTelemetry(` no?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I don't think it's necessary and it makes the code harder to read.",
+      "reference": "Why this is in the diff?"
+    },
+    {
+      "prediction": "I don't think we need to add this to the driver interface. We can just call `CloudBackupCreate` in a loop for each volume in the group.",
+      "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 would be better to do this in the `diff` function itself. The reason is that the `excessDomChildren` array is only used in the `diff` function, so it doesn't make sense to pass it to the `options.diffed` function. Also, the `excessDomChildren` array is only used in the `diff` function, so it doesn't make sense to pass it to the `options._catchError` function. Also, the `excessDomChildren` array is only used in the `diff` function, so it doesn't make sense to pass it to the `options._catchError` function. Also, the `excessDomChildren` array is only used in the `diff` function, so it doesn't make sense to pass it to the `options._catchError` function. Also, the `excessDomChildren` array is only used in the `diff` function, so it doesn't make sense to pass it to the `options._catchError` function. Also, the `excessDomChildren` array is only used in the `diff` function, so it doesn't make sense to pass it to",
+      "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": "`array_merge(...array_values($this->getFilters()));` should be `call_user_func_array('array_merge', array_values($this->getFilters()));` or `call_user_func_array('array_merge', $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 think this should be `if !sc.IsValid() && !sc.IsSampled() { return ctx }`",
+      "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": "I think it should be \"weitere\" instead of \"weitere\"",
+      "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": "Shouldn't this be `status=200`?",
+      "reference": "nit: 200 is superfluous"
+    },
+    {
+      "prediction": "why change EMAIL to email?",
+      "reference": "This is testing `unreserved keyword`"
+    },
+    {
+      "prediction": "Please remove the extra blank line.",
+      "reference": "paramsNatOrigin doesn't include sUVtoST. I would just remove that definition for now"
+    },
+    {
+      "prediction": "I think this should be `t.Fatal` instead of `t.Log`.",
+      "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": "`AWS IAM` -> `AWS Identity and Access Management`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think this should be `RESOURCE_NAME` instead of `RESOURCE_NAME_INSTANCES`.",
+      "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.getAttribute('headers')) {` instead of `if (cell.getAttribute('headers')) {` because `getAttribute` returns `null` if the attribute doesn't exist.",
+      "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": "I don't think this is correct. `getProject()` returns the project where the task is defined, not the project where the task is executed.",
+      "reference": "I am pretty sure that this is infinite recursion as getAllProjects returns the project itself."
+    },
+    {
+      "prediction": "I think this should be `dataLists` instead of `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 don't think this is the right way to do it. I think it would be better to have a `scapy.arch.windows.disable_sendrecv` module that is imported by `scapy/arch/windows.py` and `scapy/arch/unix.py`. This way, you don't have to change anything in `scapy/arch/windows.py` and `scapy/arch/unix.py`.",
+      "reference": "Can you write the test as `if not (conf.use_winpcapy or conf.use_pcap or conf.use_dnet):`"
+    },
+    {
+      "prediction": "I think this import should be moved to the top of the file.",
+      "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": "I don't think this change is correct. `strncpy` will not null-terminate the string if the source string is longer than `len`. `strcpy` will always null-terminate the string.",
+      "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": "I'm not sure if this is a breaking change or not, but I think it should be.",
+      "reference": "Can we port this error msg to v2 too? This one is better, as it explicitly states what sources are allowed."
+    },
+    {
+      "prediction": "I think this should be `const { ReadPreference } = require('mongodb');`",
+      "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` can be `color`, `grayscale` or `unchanged`.",
+      "reference": "I suggest expanding dims here to simplify the formatting."
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "Why do you need to specify the kinto prefix here?"
+    },
+    {
+      "prediction": "Please use `!s` instead of `!r`",
+      "reference": "The `!r` is unnecessary."
+    },
+    {
+      "prediction": "I think this should be `if accountKey != \"\" || sasToken != \"\" || sasURL != \"\"`",
+      "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 think this should be `public bool IsPreparerNeeded { get; private set; }`",
+      "reference": "Should we default to `true` ?"
+    },
+    {
+      "prediction": "I think we don't need to import `fsm` here.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place for this, but I think we should also add `@babel/preset-typescript` here.",
+      "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'm not sure this is the right thing to do. I think we should log the error that caused the disconnection.",
+      "reference": "Nit: We can use `%s` since `disconnectReason` is a string."
+    },
+    {
+      "prediction": "`boosting_type` should be `boosting_type='rf'` or `boosting_type='gbdt'`",
+      "reference": "I think it's better to create a new example"
+    },
+    {
+      "prediction": "I think this should be `kDepthStencilAttachmentRasterOrder` instead of `kAttachmentRasterOrder`.",
+      "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": "Is this a typo? Shouldn't it be `--SO3H`?",
+      "reference": "To be consistent, I think it should actually just be a single `-`"
+    },
+    {
+      "prediction": "This import is not used anywhere in this file.",
+      "reference": "this seems extraneous."
+    },
+    {
+      "prediction": "I don't think we need to run this test with the long domain feature set.",
+      "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 don't think we need to use `gdebi` here. We can just use `dpkg -i` and `apt-get install -f`.",
+      "reference": "I think you need `-y` here to prevent a user confirmation prompt"
+    },
+    {
+      "prediction": "I think this should be `dynamo_avx512_code_in_use` instead of `dynamo_avx512_code_in_use`.",
+      "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 make this a list of tuples, where the first element is the name of the plugin and the second element is the version of the plugin.",
+      "reference": "right URL is like in cmdrunner, using search.maven .org"
+    },
+    {
+      "prediction": "I don't think we should change the type of this property. It should be `Uri` and we should validate that it starts with `http` or `https` in the constructor.",
+      "reference": "The link to valid syntax is not really applicable for NET2_1, right?"
+    },
+    {
+      "prediction": "I don't think we need to say anything about escaping here.",
+      "reference": "Shouldn't it be more specific, like \"gcppubsub supports all UTF-8 strings\"?"
+    },
+    {
+      "prediction": "I think this test should be renamed to `testFullReplicationFromSecondary`.",
+      "reference": "Everything in this class is SolrCloud-related, not legacy replication"
+    },
+    {
+      "prediction": "I think you can remove the `if (!manager->is_scratch_reg_dead) {` check here as well.",
+      "reference": "Could just assign directly and eliminate the `newval` var."
+    },
+    {
+      "prediction": "I think this should be `^[a-z0-9/]+: [a-z0-9/]+: .*$`",
+      "reference": "... with the name of the affected package, or \"all\", followed by a colon,..."
+    },
+    {
+      "prediction": "I think we can remove this line, it's not used anymore.",
+      "reference": "We'll want to remove this one if we make the assembly version constant."
+    },
+    {
+      "prediction": "I think this should be `setup_single_post_and_page_style()` to be consistent with `setup_single_post_style()`.",
+      "reference": "Can we have a single function here that uses the same subscribers and just changes meta based on context?"
+    },
+    {
+      "prediction": "I don't think we need to change the name of this class. I think it's fine to keep it as `startup_kernel_loader`.",
+      "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 it, but it seems 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": "Why is this removed?",
+      "reference": "Migrate the spec.Labels to locator.Labels ?"
+    },
+    {
+      "prediction": "I think this should be `if gotErr != tc.wantErr {`",
+      "reference": "The test case already existed but it never tested properly because of the conditional in the test."
+    },
+    {
+      "prediction": "nit: add a space after `return`",
+      "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 don't think we need to initialize this to an empty array.",
+      "reference": "@sarowe why was this volatile? It's fishy to see this as the only volatile field."
+    },
+    {
+      "prediction": "please remove this empty class.",
+      "reference": "template is not correct?"
+    },
+    {
+      "prediction": "I think we can merge these two registries into one.",
+      "reference": "Rename the registies to `BBOX_ASSIGNERS` and `BBOX_SAMPLERS` to avoid ambiguity. There is also a registry for dataset sampler."
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "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 think this should be `Puma.stats_hash` instead of `Puma.stats` since `Puma.stats` returns a string.",
+      "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 don't think we need to do anything for single step.",
+      "reference": "Convention is \"FIXME: i#2144\" or \"XXX: i#2144\""
+    },
+    {
+      "prediction": "I think this should be `IHttpMethod` instead of `IHasHttpMethod`",
+      "reference": "This should probably be a property instead of a method."
+    },
+    {
+      "prediction": "I am not sure if this is the right place to set the state to `RECONCILED`. I think this should be set to `RECONCILED` only when the upgrade is successful. If the upgrade fails, then the state should be set to `RECONCILING_FAILED`. What do you think?",
+      "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": "Metrics/LineLength: Line is too long. [114/100]",
+      "reference": "Line is too long. [82/80]"
+    },
+    {
+      "prediction": "This file is auto-generated. Please don't make changes to this file.",
+      "reference": "Add blank line below license header."
+    },
+    {
+      "prediction": "I think we can remove the `if (ops.current() != null) { ... } else { ... }` and just do `String baseLocation = location != null ? location : ops.current().location();` and `TableMetadata metadata = ops.current().buildReplacement(schema, spec, baseLocation, tableProperties);`",
+      "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": "nit: remove empty line",
+      "reference": "Just run a global replacement for all `executionproto`."
+    },
+    {
+      "prediction": "I don't think we need to add `options` to the `__init__` method. We can add it to the `start_session` method instead.",
+      "reference": "@AutomatedTester @davehunt thoughts on a new keyword argument?"
+    },
+    {
+      "prediction": "I think this should be `FPGA_NOT_FOUND`",
+      "reference": "Should line 47 be OPAE_ERR?"
+    },
+    {
+      "prediction": "nit: `var zeroTime = time.Time{}`",
+      "reference": "Not necessary, just write `time.Time{}`"
+    },
+    {
+      "prediction": "I think this should be `SettingValue(typ.PercOrInt(minperc=0, maxperc=100, minint=1, maxint=1000), '12')`",
+      "reference": "As these two strings simply get concatenated for the docs, there's a space missing after the dot here."
+    },
+    {
+      "prediction": "I think this should be moved to the `BeginMethodHandler<TIntegration, TTarget>.Invoke` method.",
+      "reference": "What about integrations where we don't plug ourselved on OnMethodBegin?"
+    },
+    {
+      "prediction": "`visibility_of_any_element_located`?",
+      "reference": "shouldn't **call** return a boolean?"
+    },
+    {
+      "prediction": "I think this should be `com.google.api.tools.framework.model.Method` instead of `com.google.api.tools.framework.model.Interface`.",
+      "reference": "ViewModel classes should not expose any classes from framework.model."
+    },
+    {
+      "prediction": "I think we should keep this test, but change the expected output to `{'string_field': None}`",
+      "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 `--targets` instead of `-t`",
+      "reference": "Why do we use the `winrm_nodes` variable here and not the string `\"winrm_nodes\"` as we were using before?"
+    },
+    {
+      "prediction": "copyright year should be 2018-2019",
+      "reference": "why update Copyright ?"
+    },
+    {
+      "prediction": "nit: extra space after `isSampleScan()`",
+      "reference": "Does this mean we cannot update statistics with sampling on a table with LOB columns?"
+    },
+    {
+      "prediction": "I think it would be better to have a `ClearCacheOnInitialize` property instead of `ClearCacheOption` enum.",
+      "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 `row.treeNode.state === 'expanded'` and `row.treeNode.state === 'collapsed'` instead of `row.treeNode.state === 'expanded'` and `row.treeNode.state === 'collapsed'`",
+      "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 don't think we need to remove this.",
+      "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 correct. I think it should be `\\'` instead of `\\\\\\'`",
+      "reference": "Why is the escaping of @ and $ removed?"
+    },
+    {
+      "prediction": "I think this should be `IHasHttpUrl<T>` where `T` is the type of the `HttpUrl` property.",
+      "reference": "Nit: `Http` is redundant in this interface's name."
+    },
+    {
+      "prediction": "I think this should be a warning, not an error. If the prefix doesn't exist, that's fine, we just need to create it.",
+      "reference": "what if I set SynchronizableEntitiesPrefix to \"a/b/c/d\" in config?"
+    },
+    {
+      "prediction": "I think this should be `!isDir` instead of `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": "Please remove this line, it's not needed.",
+      "reference": "This is done automatically, you can skip it next time ;)"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. If the tab is deleted, then the element is gone too, so I don't think it makes sense to raise an error here.",
+      "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 think this should be a `bool` rather than a `string`.",
+      "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": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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": "Why is this needed? I don't see it being used anywhere.",
+      "reference": "We could avoid passing proto since threshold isn't needed."
+    },
+    {
+      "prediction": "Why is this disabled?",
+      "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": "why move this line here?",
+      "reference": "why can this modification avoid crash?"
+    },
+    {
+      "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's a better way to check if the current user is the same as the account we're redirecting to.",
+      "reference": "This looks good -- signing out the user if the current user is disabled"
+    },
+    {
+      "prediction": "I think it would be better to use the term \"corner\" instead of \"corner header\".",
+      "reference": "The line exceeds 120 characters."
+    },
+    {
+      "prediction": "I think this should be `desc[\"php_version\"].(string)`",
+      "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": "Why remove this?",
+      "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 we can use `maxImageRampup.getImageVersion()` instead of `firstImageRampup.getImageVersion()`.",
+      "reference": "I thought we decided on using the latest active version and not the one which is max ramped up."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to have the user interrupt and rerun `copilot env init` instead of failing here.",
+      "reference": "can we update this to a `log.Warningf`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to make this work.",
+      "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": "`schemaMan` is not used in `AddVerticesProcessor::instance`?",
+      "reference": "Do we need `schema manager` at here ?"
+    },
+    {
+      "prediction": "I think this should be `#cgo LDFLAGS: -L${SRCDIR}/target/release -ldl -Wl,-Bstatic -lflux_parser -Wl,-Bdynamic`",
+      "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, but it seems to work.",
+      "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 think this should be `config.Labels.Merge(p.Spec.ServiceMetadata.Labels).Merge(map[string]string{\"operated-alertmanager\": \"true\"})`",
+      "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 a nullable bool, so that we don't have to change the schema if we want to change the default value in the future.",
+      "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": "Metrics/CyclomaticComplexity: Cyclomatic complexity for download_plan_page_title is too high. [7/6]",
+      "reference": "thanks for moving these over. makes more sense for them to be in the exports_helper"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "That import now isn't needed anymore"
+    },
+    {
+      "prediction": "I don't think we need to change the docstring here.",
+      "reference": "Can you update the `Usage` to be an absolute path as well? Something like `/path/to/firebug.xpi`"
+    },
+    {
+      "prediction": "I think this should be `Mvx.RegisterSingleton<IMvxAccelerometer>(new MvxWindowsUapAccelerometer());`",
+      "reference": "This class is missing the `MvxPlugin` attribute"
+    },
+    {
+      "prediction": "This should be `wxLang.LayoutDirection`, not `wxLang.LayoutDirection`.",
+      "reference": "What if `wxLang` is returned `None` from `getWxLang`"
+    },
+    {
+      "prediction": "why change this?",
+      "reference": "Please don't do changing like this unless it _**really**_ cares. BTW. Please look around to infer our naming conventions."
+    },
+    {
+      "prediction": "I don't think we should be using `finest` here. `finer` seems more appropriate.",
+      "reference": "This is in test code: understanding what we're doing is important in this context."
+    },
+    {
+      "prediction": "I think this should be `--miner-enabled should not be true`",
+      "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/SpaceInsideHashLiteralBraces: Space inside { missing.<br>Layout/SpaceInsideHashLiteralBraces: Space inside } missing.",
+      "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 don't think this will work. You need to use `util.Failed(\"Failed to stop containers for %s. Run `ddev list` to ensure your site exists. error=%v\", app.ContainerName(), err)`",
+      "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 <fastdds/rtps/flowcontrol/FlowController.h>`",
+      "reference": "We could put this implementation in `RTPSWriter::create_change_pool` and avoid an extra function."
+    },
+    {
+      "prediction": "nit: please remove this empty line",
+      "reference": "should this file ends with _test.go?"
+    },
+    {
+      "prediction": "I think this should be `ss = inet_pton(socket.AF_INET6, self.src)`",
+      "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": "nit: `let flattenIndex = index;`",
+      "reference": "\"Flatten\" is a verb, so I'd probably go with `flattenedIndex` as a variable name here."
+    },
+    {
+      "prediction": "I don't think we need to freeze this.",
+      "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 I think it's the right thing to do for now.",
+      "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 to put this, but I think it should be in `appModules\\microsoftEdge.py` instead of `appModules\\microsoftEdge\\__init__.py`.",
+      "reference": "Could you please end the file with an empty line?"
+    },
+    {
+      "prediction": "I don't think we need to log this at `finest` level. I think `info` is fine.",
+      "reference": "This is an incorrect change. The dialect spoken is an important part of the handshake and should be communicated to users."
+    },
+    {
+      "prediction": "nit: I think we can remove the `isCanceledMx` and just do `isCanceled := v.isCanceled` in the `IsCanceled` method.",
+      "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 this should be `return gcerrors.Code(err)`",
+      "reference": "Nit: this is just `return gcerrors.Code(err)`, isn't it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what else to do either.",
+      "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'm not sure about this change. I don't think we should be using `cPickle` anymore, but I don't think we should be using `pickle` either. I think we should be using `json` instead.",
+      "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 think we need to pass `params.qstring.method` to `fetchMergedEventGroups` as well.",
+      "reference": "did you remove **params.qstring.method** intentionally? if so why?"
+    },
+    {
+      "prediction": "I think we should have a default value for this property.",
+      "reference": "What is the current default? Is that inconsistent across uses and that's why this is null?"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "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 `dataverse.rserve.password` instead of `dataverse.rserve.pwrd`.",
+      "reference": "Nice to see this `dataverse.rserve.password` fix rolled in."
+    },
+    {
+      "prediction": "(style) code indent should use tabs where possible",
+      "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'm not sure if this is the right thing to do, but I think it makes sense to cc the user who submitted the feedback so they can follow up with us if they need to.",
+      "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'm not sure why we need to wrap this in an `act` block. I don't think it should be necessary.",
+      "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": "Why is this needed? `election->blocks[block_a->hash ()] = block_a;` should be enough.",
+      "reference": "Could use election from `info_a.election`, or is this deliberate?"
+    },
+    {
+      "prediction": "I don't understand why this is needed. Can you explain?",
+      "reference": "looks like this should be `return array(LegacyFormHelper::getType('form'));`"
+    },
+    {
+      "prediction": "I think this should be `if (typeof opt_port === 'function')` instead of `if (typeof opt_port !== 'function')`",
+      "reference": "Maybe it would simpler to ignore opt_port if type !== 'number'?"
+    },
+    {
+      "prediction": "I think this should be `for (var i = 0; i < categoryNames.length; i++) { name = categoryNames[i];`",
+      "reference": "Where was `name` being declared before?"
+    },
+    {
+      "prediction": "I think this should be `False` instead of `Unknown`.",
+      "reference": "nit: use `corev1.ConditionUnknown`"
+    },
+    {
+      "prediction": "I don't think we need to check for `v.length === 12` here. If `v.length === 12` then `v.length === 24` must be `false`.",
+      "reference": "You also need to add a similar check on line 992, there's another place where we check `length === 24`"
+    },
+    {
+      "prediction": "nit: I think it would be better to have a method `getVoteProviderAfterBlock` instead of `getVoteProviderAfterBlock` and `getVoteProviderAfterBlock`",
+      "reference": "nit: can extract the voteProvider as a local variable"
+    },
+    {
+      "prediction": "nit: `expected error, functions are not equal` -> `expected error, functions are not equal: eq(foo_i,2) != lt(foo_i,2)`",
+      "reference": "[0] Not suggesting you change it here, but....kindof weird that there's just not an `assertFuncNotEquals`"
+    },
+    {
+      "prediction": "I don't think we need to wrap the error here. The caller of this function will handle the error.",
+      "reference": "why are we passing 'registryAddress' here? We probably should construct it together with bc. Registry is not something that change?"
+    },
+    {
+      "prediction": "It's like `RA.spreadPath|spreadPath`, but preserves object under property path",
+      "reference": "`.` at the end of the sentence"
+    },
+    {
+      "prediction": "I think it's better to measure the time from the start of the function, not from the start of the ping.",
+      "reference": "wouldn't it be useful to split this into an error metric for the timing?"
+    },
+    {
+      "prediction": "Use 2 spaces for indenting an expression spanning multiple lines.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 think this should be `AnyTimes()` as well.",
+      "reference": "line is 138 characters"
+    },
+    {
+      "prediction": "I'm not sure I like the name \"Wire\" for this function. It's not really about the wire at all, it's about cleaning up the database connection. I'd prefer to call it `Cleanup` or `Close` or something like that. What do you think?",
+      "reference": "Nit: I would leave `Wire` out of the description; if you use this without wire you can still use it."
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "Nit: each field can be on a separate line to make it visually easier to read."
+    },
+    {
+      "prediction": "I don't think we need to add this file to the list of files to be sent to the server.",
+      "reference": "Do we still have these files? Shouldn't we remove these?"
+    },
+    {
+      "prediction": "This should be `ErrorDurationInvalidReason`",
+      "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 it's better to set it to 10000.",
+      "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 it should be done in `Bolt::Puppetfile::Plan`.",
+      "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 in `machine_nrf52840_dk.go`.",
+      "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 think this should be `final`.",
+      "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": "I'm not sure if this is the right way to do this, but it seems to work.",
+      "reference": "why is this needed?"
+    },
+    {
+      "prediction": "`key` is not used in this class.",
+      "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_I32` 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": "`strlen > PLAYER_NAME_LENGHT` should be `strlen > PLAYER_NAME_LENGTH`",
+      "reference": "spelling (variable name, all three changes)"
+    },
+    {
+      "prediction": "`desiredRole` should be `desiredRole keybase1.TeamRole`",
+      "reference": "I believe you meant \"specify `desiredUID` and `desiredRole`\"?"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "Please follow the import style, move it to its similar group"
+    },
+    {
+      "prediction": "I think this should be `errors.UnrecognizedEncodingError` instead of `errors.RouterUnrecognizedEncodingError`.",
+      "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": "I don't think we should remove the `passwordModificationTime` from the `validate` method.",
+      "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 think this should be `evt.CommandLine.Trim().Equals(commandLine.Trim(), StringComparison.OrdinalIgnoreCase)` instead of `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": "I'm not sure if this is the right thing to do, but it seems to be the only way to make the test pass.",
+      "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 why we need to add the margin here, but it seems to be necessary for the flyout button to line up with the flyout.",
+      "reference": "Hm, this looks like a change that should go upstream as well."
+    },
+    {
+      "prediction": "why remove this?",
+      "reference": "let's keep ActionByActionHash and delete getActionByActionHash"
+    },
+    {
+      "prediction": "I think it would be better to return `PricingGroupInterface` instead of `PricingGroup`.",
+      "reference": "please use return type"
+    },
+    {
+      "prediction": "I don't think we need to log this at `FINE` level. I think `INFO` is fine.",
+      "reference": "This is deliberately at this level."
+    },
+    {
+      "prediction": "I think this should be `msg: %s, exception: %s`",
+      "reference": "I'm going to go ahead and undo this change, I don't think it was intentional."
+    },
+    {
+      "prediction": "I don't think this is a typo, it should be `current` instead of `currnent`",
+      "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/target/storageTarget`",
+      "reference": "will change to `storage_target`"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "This would be better as `cls.securitycenter_client`, to match what is being instantiated."
+    },
+    {
+      "prediction": "I think this should be `nativeSys32 != Syswow64Sys32 and Syswow64Sys32 is not None`",
+      "reference": "Could we have this initialization as part of the initialize method?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "This line should be removed as editController is not used from this context anymore."
+    },
+    {
+      "prediction": "I don't think this is accurate. The `Process` method is called for each table, not for the entire transformation.",
+      "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": "`init_cfg is not allowed to be set` -> `init_cfg is not allowed to be set when pretrained is not None`",
+      "reference": "For insurance, it's best not to take it directly"
+    },
+    {
+      "prediction": "I don't think we need to pass `CharSequence[] value` to `beforeChangeValueOf` and `afterChangeValueOf`.",
+      "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 done in the `provisionContainerResources` method, not in the `startContainer` method.",
+      "reference": "is the plan to fully migrate to boltdb and then remove the state save here?"
+    },
+    {
+      "prediction": "This should be `rgb(255, 255, 255)`",
+      "reference": "This should be `output.colorBright`. Unless I'm missing some reason for adding a new theme variable."
+    },
+    {
+      "prediction": "This should be `2017` and `3.1.0`",
+      "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 javax.annotation.Nonnull;` instead of `import javax.annotation.*;`",
+      "reference": "Is this supported by every implementation of Java (and Android)? Should it be conditional upon `gen_nullable`?"
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "@briri Thanks! I just caught this too updating my feature branch"
+    },
+    {
+      "prediction": "I think this should be `if len(configKey) == 0 && envpw != \"\" { fs.Debugf(nil, \"Using RCLONE_CONFIG_PASS password.\") err := setConfigPassword(envpw) if err != nil { fmt.Println(\"Using RCLONE_CONFIG_PASS returned:\", err) } return nil, err }`",
+      "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 should be `@Ignore` for now. We don't have a good way to test this yet.",
+      "reference": "Is this only to set the node to draining?   I think we can do that inside `Before` or something?"
+    },
+    {
+      "prediction": "I think this should be `if (AzureAppServices.Metadata.IsRelevant && !string.IsNullOrEmpty(AzureAppServices.Metadata.SiteName)) { return AzureAppServices.Metadata.SiteName; }`",
+      "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 think this should be `type PlanResult = Bolt::PlanResult`",
+      "reference": "Do we expect people to use PlanResult directly?"
+    },
+    {
+      "prediction": "nit: `actualErr` -> `err`?",
+      "reference": "Maybe we should call this something like `errToReturn`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but I think it should be `epochNum <= fairBankEpochNum + a.foundationBonusLastEpoch` instead of `epochNum >= fairBankEpochNum && 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'm not sure if this is the right thing to do, but I think it makes the most sense. If we don't count `img` as phrasing, then we don't have to worry about the space between the `img`'s accessible name and other elements' accessible names.",
+      "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 think we can remove this line.",
+      "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 `except FileNotFoundError as e:` so that we can log the error.",
+      "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": "I think this should be `maxQueryLength = 1000000000;` instead of `maxQueryLength = 0;`",
+      "reference": "Do we want validation somewhere that max >= min? Or simply treat the max < min case as ignoring max?"
+    },
+    {
+      "prediction": "This change is not related to this PR, but I think it should be reverted.",
+      "reference": ":+1: that's a easier example :)"
+    },
+    {
+      "prediction": "I think this should be `findProductIdsWithBoundAttributeByAttributeId(AggregateId $attributeId): array;`",
+      "reference": "If an external module decorates this interface, such a change will cause it to generate an error"
+    },
+    {
+      "prediction": "I think this should be `FNM_PATHNAME` instead of `FNM_PERIOD`.",
+      "reference": "Oh ha hah, FNM_PERIOD worked out nicely there. Points for co-opting a file system convention."
+    },
+    {
+      "prediction": "remove empty line",
+      "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 { return err }`",
+      "reference": "Is `o.name` always set here?"
+    },
+    {
+      "prediction": "I think this should be `steadyStateDependencies` instead of `steadyStateDependencies`.",
+      "reference": "Can this line be removed, as `SteadyStateDependencies` isn't updated anywhere? Also do we need to check the `TransitionDependencySet` here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to put this logic, but I think it makes sense to have it in the `User` model since it's related to the `User`'s `plan`.",
+      "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 don't think we need to disable this rule here.",
+      "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 `c.config.Namespaces.AlertmanagerConfigAllowList`",
+      "reference": "L171 needs to be modified too?"
+    },
+    {
+      "prediction": "I think this should return -1 if d->nodeid < e->nodeid and 1 if d->nodeid > e->nodeid.",
+      "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": "Please remove `_base_` and `model` in this file.",
+      "reference": "configs/pvt/retinanet_pvt_t_fpn_1x_coco.py --> configs/pvt/retinanet_pvt-t_fpn_1x_coco.py"
+    },
+    {
+      "prediction": "I think this should be `json:\"desiredStatus,omitempty\"`",
+      "reference": "Nit, this could be json:\",omitempty\""
+    },
+    {
+      "prediction": "I think this should be `BuildRequired() (bool, error) { return buildRequired(s.Image.Image) }`",
+      "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 do we need both `integration` and `integration,deploy`?",
+      "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": "`ground2` should be `ground3`",
+      "reference": "Copy and paste error here. Edit: Also missing commas here and in other Perl croaks between some parameters and not others."
+    },
+    {
+      "prediction": "Why do we need `stateRoot` here?",
+      "reference": "stateRoot or maybe ParentHeader would make more sense from API point of view?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get rid of the notification.",
+      "reference": "Are you sure that this is no longer needed?"
+    },
+    {
+      "prediction": "nit: I think we can remove `SNAPSHOT_COLUMNS` and use `SCAN_COLUMNS` instead.",
+      "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 think we should remove this method from the proto file as well.",
+      "reference": "`ctx` is unused in ReportStat"
+    },
+    {
+      "prediction": "This is not needed.",
+      "reference": "We don't want to add boost as a dependency when building the Thrift compiler. Sorry."
+    },
+    {
+      "prediction": "I think this should be `int32_t speedChange = 0;` instead of `int32_t minSpeedChange = 0; int32_t maxSpeedChange = 0;`",
+      "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": "Why do we need this function? I think we can use `SchedPolicyList` to get the policy by name.",
+      "reference": "if name is empty will this become enumerate ? (and cause the unmarshal to fail ?)"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "How about including a flash message that explains the exercises are only available to subscribers of X plan?"
+    },
+    {
+      "prediction": "Shouldn't this be `if text in modeparsers.STARTCHARS and e.key() == Qt.Key_Backspace:`?",
+      "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'm not sure if this is the right place to put this, but I think it should be in `config/environments/production.rb` instead of `config/environments/development.rb`.",
+      "reference": "Freeze mutable objects assigned to constants."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 think this should be `activity.SetTag(SemanticConventions.AttributeHttpServerIP, xForwardedFor.Split(',').First().Trim());`",
+      "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 we should keep `/graph` as the path for the graph service.",
+      "reference": "`status` is just one of the features, named as status is not suitable."
+    },
+    {
+      "prediction": "I don't think this is the right place to add the icon. It should be done in the layout file.",
+      "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 default to `true`",
+      "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": "I think this should be reverted.",
+      "reference": "move the constructor to down the `namespace`"
+    },
+    {
+      "prediction": "I think this should be a `corev1.ObjectReference` instead of a `corev1.LocalObjectReference`.",
+      "reference": "Nit: ending with a '.' looks consistent with the fields around it."
+    },
+    {
+      "prediction": "nit: I think we can remove the `ctx` parameter from `s.fieldsFromJWTSvidParams` now.",
+      "reference": "Audit log will not have a warning about they are using a deprecated path, is it something we must care about?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "I think you also want to update the condition to add ` || impl.Visibility == TypeMemberVisibility.FamilyOrAssembly`."
+    },
+    {
+      "prediction": "I don't think we need to change the base gas for this. We can just increase the `DepositToRewardingFundGasPerByte` to `1000` or `10000` and keep the base gas as `10000`.",
+      "reference": "`DepositToRewardingFundBaseGas` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think this should be \"Getting Started \ud83c\udf31\"",
+      "reference": "What do you think of \"operations\"?? Also what do these emojis look like on Linux??"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be nice to have a link to the blog post that explains what a team is, and how it works.",
+      "reference": "I like the word \"give\" here. Feels like I'm giving a gift to my team."
+    },
+    {
+      "prediction": "I don't think we need to import `make_multilabel_classification` from `sklearn.datasets`.",
+      "reference": "Is it possible to use `np.random` module instead?"
+    },
+    {
+      "prediction": "I think this should be `allow_None=False` and `default='white'`",
+      "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 think this should be `system:k3s-controller` and not `system:k3s-controller-<version>`",
+      "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": "This should be `import \"github.com/ncw/go-acd\"` not `import \"github.com/ncw/go-acd\"`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "Should this be `OperatingSystem` instead of `Os`?",
+      "reference": "nit: all other types in this project have full names, so maybe a better name would be `OperatingSystem`?"
+    },
+    {
+      "prediction": "This should be `#include <GraphMol/MolTransforms/MolTransforms.h>`",
+      "reference": "Why is this needed here?"
+    },
+    {
+      "prediction": "I don't think this is the right thing to do. If the method is not a type variable resolution method, then we should return an empty optional. If it is a type variable resolution method, then we should return an empty optional if the type variables cannot be resolved.",
+      "reference": "mmm, why a method declaration should not have the TypeVariableResolutionCapability? Is this ok?"
+    },
+    {
+      "prediction": "I don't think we need to change the order of the parameters here.",
+      "reference": "nit: any reason you swapped the ordering?"
+    },
+    {
+      "prediction": "\").dispatchEvent(new CustomEvent(\"change\", {\n             bubbles: !0\n         }));\n     }\n```\nReview comment?",
+      "reference": "can you de-uglify at least this line?.. hard to tell what changed..."
+    },
+    {
+      "prediction": "nit: I think we can remove the `GetBPFCalicoDir` method as well.",
+      "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 we need this 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 don't think this is the right place to do this. I think it should be done in the `search` view, not in the `search.html` template.",
+      "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 think it would be better to do this in `NewWriter` rather than here.",
+      "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 it would be better to do this in the `doPreGainFocus` function, rather than in the `executeEvent` function. That way, we don't have to worry about `focusRedirect` being set to `None`, and we don't have to worry about `doPreGainFocus` returning `False` if `focusRedirect` is set to something other than `None`.",
+      "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": "This should be `github.com/GoogleCloudPlatform/compute-image-tools/osconfig_tests/config` and `github.com/GoogleCloudPlatform/compute-image-tools/osconfig_tests/gcp_clients`",
+      "reference": "You imports are out of order"
+    },
+    {
+      "prediction": "I don't think we need to make this public. It's only used in this class.",
+      "reference": "This should not be called Flow Trigger. FlowTrigger should has its own flowTrigger job name, for instance, \"flowtrigger\""
+    },
+    {
+      "prediction": "I don't think we need to create a new package for sqlite3. We can just use `db.NewSQLBase(\"sqlite3\", cfg.SQLite3File)` directly.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be `#if DEBUG && !NETCOREAPP2_1`",
+      "reference": "Why do we need `\\n` if we already use `WriteLine`? (and `\\n` is not cross platform)"
+    },
+    {
+      "prediction": "I think this should be `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `private final String protocol;` instead of `private static final String DEFAULT_METRICS_PROTOCOL = \"prometheus\";` and `",
+      "reference": "What is instrumentation name used for? I'm not seeing other classes use it, only a getter, constructor, and builder."
+    },
+    {
+      "prediction": "Why do we need to disable this warning?",
+      "reference": "@kristofgazso could you review these warnings?"
+    },
+    {
+      "prediction": "This file should be deleted.",
+      "reference": "will delete this."
+    },
+    {
+      "prediction": "I think we can remove the `.html_safe` here, since we're using `I18n.t` with the `html: true` option.",
+      "reference": "~~Indent 2 lines above~~ Disregard. My fault"
+    },
+    {
+      "prediction": "Why is this change necessary? I don't think it's related to this PR.",
+      "reference": "What was the bug that this is fixing?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `base64.b64encode` function returns a `bytes` object, not a `str` object, so you need to decode it to a `str` before you can concatenate it with the `Basic ` string.",
+      "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 `bazel-bin/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": "`date_utc` added so we can be sure to get the actual date and time based upon UTC and not the client date and time based on the client machine. `date_utc` added so we can be sure to get the actual date and time based upon UTC and not the client date and time based on the client machine.",
+      "reference": "Not an expert of ES, so not sure of the impact here."
+    },
+    {
+      "prediction": "I don't think you need to declare granteeList here. You can just pass NULL to getRoleList.",
+      "reference": "Who deletes the memory for the granteeList? (or the roleList for that matter)"
+    },
+    {
+      "prediction": "Shouldn't this be done in the destructor of `listener`?",
+      "reference": "This loop is missing a lock on subscriptions_mutex."
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "shouldn't clearance be setting these up for us?"
+    },
+    {
+      "prediction": "I'm not sure I like the name \"noOp\" here. I think it would be better to call it something like \"disabled\" or \"noop\" or \"null\" or something like that. \"noOp\" makes me think of a function that does nothing, but that's not what this is.",
+      "reference": "Need to update the doc for `TelemetrySDK` and `Host` structs deleting references from removed functions."
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "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'm not sure this is the right place to do this. I think it would be better to do this in the `scrollIfNecessary` function.",
+      "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": "I don't think this import should be here.",
+      "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": "Why do we need to update `workflow_id` here?",
+      "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": "Why is this disabled?",
+      "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 a good idea, but I think it would be nice to have a single `sds` package that contains both the v2 and v3 implementations. That way, we don't have to worry about which version of the SDS API we're using, and we don't have to worry about which version of the SDS implementation we're using. I'm not sure if this is a good idea, but I think it would be nice to have a single `sds` package that contains both the v2 and v3 implementations. That way, we don't have to worry about which version of the SDS API we're using, and we don't have to worry about which version of the SDS implementation we're using. I'm not sure if this is a good idea, but I think it would be nice to have a single `sds` package that contains both the v2 and v3 implementations. That way, we don't have to worry about which version of the SDS API we're using, and we don't have to worry about which version of the SDS implementation we're using. I'm not sure if this is a good idea, but I think it would be nice to have a single `sds` package that",
+      "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": "I don't think we need to check `beforeChangeResult[0] === null` here. `beforeChangeResult.length === 0` should be enough.",
+      "reference": "Please check also whether `null` occurs in the rest of `beforeChangeResult` array."
+    },
+    {
+      "prediction": "I don't think we need to mention that this is client-side.",
+      "reference": "Outdated docs. There is no response, there's an ack."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do.",
+      "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": "Why do we need to check `tx == nil` here?",
+      "reference": "can you check if there's other similar cases to add nil-check like this?"
+    },
+    {
+      "prediction": "I think this should 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": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.<br>Line is too long. [81/80]",
+      "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 should be called `getTransactionLocationByHash` to be consistent with the other methods in this class.",
+      "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 think this should be `value.endsWith(\"\\\\\") || value.endsWith(\"/\") || value.endsWith(\"-\") || value.endsWith(\".\") || value.endsWith(\":\") || value.endsWith(\"*\") || value.endsWith(\"?\") || value.endsWith(\"\\\"\") || value.endsWith(\"<\") || value.endsWith(\">\") || value.endsWith(\"|\") || value.endsWith(\" \") || value.endsWith(\"\\t\") || value.endsWith(\"\\n\") || value.endsWith(\"\\r\") || value.endsWith(\"\\b\") || value.endsWith(\"\\f\") || value.endsWith(\"\\u000B\") || value.endsWith(\"\\u000C\") || value.endsWith(\"\\u000E\") || value.endsWith(\"\\u000F\") || value.endsWith(\"\\u0010\") || value.endsWith(\"\\u0011\") || value.endsWith(\"\\u0012\") || value.endsWith(\"\\u0013\") || value.endsWith(\"\\u0014\") || value.endsWith(\"\\u0015\") || value.endsWith(\"\\u0016\") || value.endsWith(\"\\u0017\") || value.endsWith(\"\\u0018\") || value.endsWith(\"\\u0019\") || value.endsWith(\"\\u001A\") || value.endsWith(\"\\u001B\") || value.endsWith",
+      "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": "This should be `pid_t` instead of `int64_t`.",
+      "reference": "Looks like the X86 inline assembly is not happy with this type. I will update that"
+    },
+    {
+      "prediction": "I don't think we need the `&nbsp;` here.",
+      "reference": "Tabs and a space, again."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `QuestManager::GetTaskName()` function that returns a `std::optional<std::string>`. Then you can do something like `if (auto task_name = quest_manager->GetTaskName(task_id)) { ... }`",
+      "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 about this change. I think it should be `@proposal.cart_id` instead of `@proposal.cart.id` because `@proposal.cart` is a `Cart` object and not a `CartDecorator` object.",
+      "reference": "`version` isn't need anymore because it's built into `approve_proposal_url`, yes?"
+    },
+    {
+      "prediction": "I don't think we need to import `_dbmField` from `dot11`.",
+      "reference": "Moved to the bottom of the file. `CtrlPDU` sits on `BTLE_DATA` so let's put it after"
+    },
+    {
+      "prediction": "Trailing whitespace detected.",
+      "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'm not sure this is the right thing to do. I think we should return an error if we read 0 bytes and the error is not EOF.",
+      "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": "nit: I think it would be better to use `Optional` instead of `null` here.",
+      "reference": "If there is a reference to other fields or methods, please do the initialization in the constructor."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "These should be importing from `pill-container`"
+    },
+    {
+      "prediction": "I think this should be `RETURNVALUE_NOTPOSSIBLE` as well.",
+      "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 if this is the best way to handle this, but it seems to work.",
+      "reference": "Does this need to allocate a string or would a nil value (indicating no header) be a better?"
+    },
+    {
+      "prediction": "I don't think this is needed. `this.constructor` is already set to `Component` in the constructor.",
+      "reference": "Nit: Past tense of `reset` is also `reset`."
+    },
+    {
+      "prediction": "I think it would be better to have this in the `wallet.py` file, not in the `commands.py` file.",
+      "reference": "Shouldn't this use `'n'` instead?"
+    },
+    {
+      "prediction": "I don't think we need to change the timer task type here. I think we can keep it as `enumsspb.TASK_TYPE_ACTIVITY_TIMEOUT` and `enumsspb.TASK_TYPE_DECISION_TIMEOUT` and just change the timer task type in `timerTaskBase` to `enumsspb.TASK_TYPE_ACTIVITY_TASK_TIMEOUT` and `enumsspb.TASK_TYPE_WORKFLOW_TASK_TIMEOUT`",
+      "reference": "revert back to 'TASK_TYPE_ACTIVITY_TIMEOUT'"
+    },
+    {
+      "prediction": "I think this should be `template void Engine::GetBlock<T>(const std::string &, T **, const Mode);` instead of `template void Engine::GetBlock<T>(Variable<T>, T **, const Mode);`",
+      "reference": "Shouldn't we prefer passing a pointer by reference T*&, since these are C++ bindings?"
+    },
+    {
+      "prediction": "I think we can remove the `cgExists` variable and just use `len(ingressRule.SourceGroups) > 0`.",
+      "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": "Should this be `apis.ReconcilePending` or `apis.ReconcileInProgress`?",
+      "reference": "Pending has been set only for SPC.. why? for other CRs as well, this need to be done.. right?"
+    },
+    {
+      "prediction": "I don't think we need to set this to `false` by default, it should be `true` by default.",
+      "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 `methodInfo.responsePageTokenName()` instead of `methodInfo.responsePageTokenName().toUpperCamel()`.",
+      "reference": "How come we need case manipulation for the nextPageToken and not for pageToken?"
+    },
+    {
+      "prediction": "I think this should be `put` instead of `post`.",
+      "reference": "why use these rather than regular named paths?"
+    },
+    {
+      "prediction": "Should this be `OPTIONS[:global] + %w[format]`?",
+      "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'm not sure if this is the right place to put this docstring. I think it would be better to put it in the docstring of the `create_candidate_sets_for_all_users` function.",
+      "reference": "We should make this a docstring, so that editors are able to pick it up."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it seems to be the only way I could get it to work.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "App cache size is now managed dynamically by the `WebView`. This statement has no effect in the new framework."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "VirtualNode has a `hasAttr` function, any reason why you're not using it?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "This now throws when running tests (maybe the move to java 11??). Do we want to keep it?"
+    },
+    {
+      "prediction": "I think this should be `if 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": "`Cannot add the duplicate file %{fileName}, it already exists` -> `Cannot add the duplicate file %{fileName}. It already exists.`",
+      "reference": "this one should also have quotes i guess :)"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we should be using `b.unstruct.object.GetName()` instead of `b.unstruct.Object.GetName()`.",
+      "reference": "Using the variable on range scope `mock` in function literal (from `scopelint`)"
+    },
+    {
+      "prediction": "I think this should be `res.isCompleted() ? ok(json(res.getDataset())) : accepted(json(res.getDataset()));`",
+      "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 `ORDER_BY` instead of `ORDER_PATH`.",
+      "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": "nit: I think we can remove the extra backticks here.",
+      "reference": "VIM didn't like your whitespace."
+    },
+    {
+      "prediction": "I think we can remove the `using Datadog.Trace.ClrProfiler;` as well.",
+      "reference": "Are you sure the namespace should change?"
+    },
+    {
+      "prediction": "I don't think we should change the return type of this method. I think it should remain `void` and the `object` parameter should be renamed to `startupHint` or something like that.",
+      "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's better to use `new TimeType(MySQLType.TypeDuration)` instead of `new TimeType(MySQLType.TypeDuration)`.",
+      "reference": "why do we create a time type here?"
+    },
+    {
+      "prediction": "I don't think you need to include the css file here. The css file is already included in the html file.",
+      "reference": "Does this style actually need to get loaded in all of these components?"
+    },
+    {
+      "prediction": "remove unused import",
+      "reference": "import but not used so you did not resolve compile warnings?"
+    },
+    {
+      "prediction": "I don't think this is correct. If armor is 1, then the damage should be reduced by 1. If armor is 2, then the damage should be reduced by 1 or 2. If armor is 3, then the damage should be reduced by 1, 2, or 3. If armor is 4, then the damage should be reduced by 2, 3, or 4. If armor is 5, then the damage should be reduced by 2, 3, 4, or 5. If armor is 6, then the damage should be reduced by 3, 4, 5, or 6. If armor is 7, then the damage should be reduced by 3, 4, 5, 6, or 7. If armor is 8, then the damage should be reduced by 4, 5, 6, 7, or 8. If armor is 9, then the damage should be reduced by 4, 5, 6, 7, 8, or 9. If armor is 10, then the damage should be reduced by 5, 6, 7, 8",
+      "reference": "This is so much more concise and beautiful than the previous formula"
+    },
+    {
+      "prediction": "This should be `private final Logger LOG = LogManager.getLogger(AcceptanceTestBase.class);`",
+      "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 don't think we need this interface. We can just use `map[string]interface{}`.",
+      "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 `overrideEnableDebug bool` instead of `overrideEnableDebug *bool`.",
+      "reference": "Why is there an \"enable\" in the param name? Couldn't it just be `overrideDebug`? Here are everywhere else."
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "`published.recently_published_first` reads a little oddly to me. Is this the same thing as `Video.published.ordered`?"
+    },
+    {
+      "prediction": "`throw new Error('Unable to parse content of ' + file.originalPath)` -> `throw new Error('Unable to parse content of ' + file.originalPath + ': ' + e.message)`",
+      "reference": "Encountered this because I had a stray comma. Figured I'd tweak it a bit."
+    },
+    {
+      "prediction": "I think this should be `Eventually(getCIDRBlockRules(felix, \"ip6tables-save\"), \"8s\", \"0.5s\").Should(BeEmpty())`",
+      "reference": "qq: Should this include the iptables6-save sim. to the inverse checks above?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "im open to explicitly checking NotSet value here but json logic already handles it so kept it this way."
+    },
+    {
+      "prediction": "This should be `#include \"Util/Lua.h\"`",
+      "reference": "The naming of the include file appears to be unlucky."
+    },
+    {
+      "prediction": "I think this should be `self.legend_opts.update(leg_spec)`",
+      "reference": "Bit worried about this. At minimum you should make a copy of the dict here to avoid modifying a user supplied variable."
+    },
+    {
+      "prediction": "nit: I think it would be good to add a log message here to indicate that we are pruning the launch template versions.",
+      "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": "I don't think this is a good idea. The `ContainerInformationTab` should be able to handle any kind of container, not just `WinePrefixContainerDTO`.",
+      "reference": "Do we require the specific implementation information here?"
+    },
+    {
+      "prediction": "Style/PercentLiteralDelimiters: Use %q unless you need interpolation or special jobbing (such as newlines).",
+      "reference": "How about `exists?(['domain ~* ?', domain.downcase])` ?"
+    },
+    {
+      "prediction": "I think this should be `c.IsSet(FlagContextTimeout)` instead of `c.GlobalIsSet(FlagContextTimeout)`.",
+      "reference": "Is this getting used in other places? Can we switch all the places to use the new API you added?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "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 don't think we need to disable indent here.",
+      "reference": "Shouldn't we import `default`? I can't get here to test."
+    },
+    {
+      "prediction": "`ParseResult` should be backticks, not backticks.",
+      "reference": "!!!!!!!!!!!!! @graingert, does this mean that tests haven't been running???"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "I've already tried the null-coalescing operator here and got a strange error, so did this the old-fashioned way."
+    },
+    {
+      "prediction": "`parentRole === 'list'` should be `parentRole === 'list' || parentRole === 'listitem'`",
+      "reference": "This doesn't follow our spacing convention. It's also a little hard to read. Can you reformat?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `internal/pkg/deploy` import and just use `deploy` here.",
+      "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 thing to do here. I think we should keep the `DropAggregationConfig` as the default, and just add the `MetricStreamConfiguration` to the `AddConsoleExporter` method.",
+      "reference": "Consider making a constant (e.g. `MetricStreamConfiguration.Drop`)."
+    },
+    {
+      "prediction": "I think this should be `from rdkit.six import xrange`",
+      "reference": "same question: why not just switch this to range too?"
+    },
+    {
+      "prediction": "I think this should be `if (type == Type::Unknown)`",
+      "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": "Why do we need to do this?",
+      "reference": "should we assert for the type for `FeatureGroupData` ? I think it should be 1-Byte type."
+    },
+    {
+      "prediction": "I think this should be `np.flipud(...)`.",
+      "reference": "As long as you are sure this is definitely correct now... :-)"
+    },
+    {
+      "prediction": "I think this should be `tempBasalCancel.setText(MainApp.instance().getString(R.string.cancel) + \": \" + activeTemp.toStringShort());`",
+      "reference": "... so that all action buttons have the same height :-)"
+    },
+    {
+      "prediction": "This file is not used in this PR. Please remove it.",
+      "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'm not sure this is the right place for this. I think it should be in the `http` package.",
+      "reference": "I hate this name, open to suggestions."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just return an error and let the caller handle it.",
+      "reference": "Does this log line work? We need Warningf or just warning without the format string."
+    },
+    {
+      "prediction": "I think this should be `c.MetricsProcessors` instead of `c.MetricsProcessors...`",
+      "reference": "The pull controller would need similar updates."
+    },
+    {
+      "prediction": "I think this should be 2016-2018",
+      "reference": "ignore this.. just didn't want to litter with a commit update to satisfy license plugin"
+    },
+    {
+      "prediction": "I'm not sure why this change was made, but I think it should be reverted.",
+      "reference": "Is this definition needed since it doesn't look like it's being used anywhere?"
+    },
+    {
+      "prediction": "Trailing whitespace detected.",
+      "reference": "Can this condition be reworded using `if` please?"
+    },
+    {
+      "prediction": "`inputs != nullptr && !inputs->hasData()` -> `!inputs->hasData()`",
+      "reference": "after ||, no need inputs != nullptr"
+    },
+    {
+      "prediction": "I think this should be `IntegrationTestBuildspecPath` instead of `IntegTestBuildspecPath` to be consistent with the rest of the codebase.",
+      "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 thing to do. I think it should be `newVNode._dom = oldDom!=null ? oldDom!==oldVNode._dom ? oldDom : oldVNode._dom : null;`",
+      "reference": "This line always confused me   anyway. Good catch!"
+    },
+    {
+      "prediction": "I think this should be `j.jServer.shutdown(context.Background())`",
+      "reference": "Should we add `ctx` to `BlockServer.Shutdown()` for this purpose? Would be nice, but I don't care too much."
+    },
+    {
+      "prediction": "I think this should be `period.nanosec / 1000000000`",
+      "reference": "Duration_t at RTPS level must be serialized using fractions."
+    },
+    {
+      "prediction": "I think this should be `yarpcerrors.InvalidArgumentErrorf(msgInboundDispatcherNotRunning, service)`",
+      "reference": "\"peer\" has its own meaning within YARPC with its own class of objects. This should probably be \"dispatcher\" too?"
+    },
+    {
+      "prediction": "I don't think we need to do this for clang 5.0, right? It should work out of the box with clang 5.0, right?",
+      "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": "why remove these flags?",
+      "reference": "_plugins should be removed too"
+    },
+    {
+      "prediction": "I think this should be `!validAttrValue`",
+      "reference": "The rule selector will not include `<input>` and `<area>` elements. Why not make this a new, separate rule altogether?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense.",
+      "reference": "@The-Compiler Is this style okay or would you prefer a common userscript exception base?"
+    },
+    {
+      "prediction": "I don't think we need to pass `sub_type` to `toObject`",
+      "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": "Please remove this empty line.",
+      "reference": "Let's remove this change."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get \"not selected\" to be spoken for table cells.",
+      "reference": "Could you split this into multiple lines?"
+    },
+    {
+      "prediction": "I think this should be `currentEntityURL || global.googlesitekit.permaLink` instead of `currentEntityURL ? ... : global.googlesitekit.permaLink` so we don't fall back to `global.googlesitekit.permaLink` if `currentEntityURL` is falsy.",
+      "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'm not sure about this change. I don't think we should change the case of the first letter of the word \"response\".",
+      "reference": "It does not recommend capitalization"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it works for now.",
+      "reference": "Minor thing, but can you use 1.9 hash syntax for this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems to be the only place where `lambda_cwd_on_host` is used.",
+      "reference": "I tried to follow the code structure already present."
+    },
+    {
+      "prediction": "I think this should be `from.toPath().toString()`.",
+      "reference": "seems reasonable to also want to include the 'to' location?"
+    },
+    {
+      "prediction": "I think this should be `resource_filename(\"nupic.core\",\"\")`",
+      "reference": "Please leave 1 space char between commas here and other places in this PR."
+    },
+    {
+      "prediction": "I think this should be `||` instead of `&&`.",
+      "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 `cli.hasOption(\"collections\")` instead of `cli.hasOption(\"cluster\")`",
+      "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 `strcmp(p->data.compressed_string, p1->data.compressed_string) == 0`",
+      "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": "This should be `const bool confirmed_frontiers = request.get<bool> (\"confirmed_frontiers\", false);`",
+      "reference": "I think `bypass_frontier_confirmation` conveys the intention better (default false as well)"
+    },
+    {
+      "prediction": "I think we can remove the `require` import here as well.",
+      "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 don't think we need to pass the `tab` to the `AbstractAudio` class. The `AbstractAudio` class should be able to get the `tab` from the `self._widget`.",
+      "reference": "You'll also need to adjust `FakeWebTabAudio` in `tests/helpers/stubs.py`."
+    },
+    {
+      "prediction": "I think we should keep the license header.",
+      "reference": "@csrwng what should we do here? This one is from origin."
+    },
+    {
+      "prediction": "I don't think we need to include LocaleContext here.",
+      "reference": "Follow alphabetical order."
+    },
+    {
+      "prediction": "I don't think we need this interface. We can just use `IAutomaticTracer` directly.",
+      "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 don't think this change is correct. The `<b>` tag is used to highlight the matching text. Without it, the matching text will not be highlighted.",
+      "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`",
+      "reference": "Where would this be configured? In azkaban.properties?"
+    },
+    {
+      "prediction": "Shouldn't this be `checkForAtomicOperators(update.$set || {})`?",
+      "reference": "1. Use `const` instead of `var` here. 2. Move the check to after the `if (update == null || typeof update !== 'object')`."
+    },
+    {
+      "prediction": "nitpick: I think it would be better to use `#if defined(__linux__) && LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27)` instead of `#if defined(__linux__) && LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27)`",
+      "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 it's better to initialize it in the constructor.",
+      "reference": "This field could be `final`."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [137/120]",
+      "reference": "would be good to have a test for the case when `shutdown!` raises"
+    },
+    {
+      "prediction": "I think this should be `if (CmpCommon::getDefault(HBASE_ASYNC_OPERATIONS) == DF_ON && (t == ComTdbHbaseAccess::INSERT_ || t == ComTdbHbaseAccess::UPDATE_ || t == ComTdbHbaseAccess::DELETE_)) { hbasescan_tdb->setAsyncOperations(TRUE); } else { hbasescan_tdb->setAsyncOperations(FALSE); }`",
+      "reference": "Does upsert pass this check also?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to log the output of the patch command. I think it would be better to log the output of the `kubectl patch` command.",
+      "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 `kbfsConfigName + \".DS_Store\"`",
+      "reference": "I'm not sure this naming is a good idea; that's the format for macOS xattr metadata files on unsupported filesystems."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [83/80]",
+      "reference": "Metrics/LineLength: Line is too long. [85/80]"
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "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": "I think this should be `FLB_ERROR` instead of `FLB_RETRY`.",
+      "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 think this should be `self.config_state_before_fullscreen` instead of `config.val.content.desktop_fullscreen`",
+      "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 why this doesn't work, but it's not a big deal for now.",
+      "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": "This should be removed before merging.",
+      "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 don't think we need to expose this to the user. I think we should just use `CONTAINER` as the default and not allow the user to change it.",
+      "reference": "where is the default being set?"
+    },
+    {
+      "prediction": "I think this should be `matches[1].start` and `matches[1].length` instead of `matches[2].start` and `matches[2].length`.",
+      "reference": "these changes should likely be in the next commit instead"
+    },
+    {
+      "prediction": "I think we can remove this now that the table formatter is implemented.",
+      "reference": "Since you have defined this struct, I would suggest to replace all exist literal structs by this."
+    },
+    {
+      "prediction": "I think this should default to `false` instead of `true`.",
+      "reference": "The field name doesn't really capture what this option does. I think `mConfirmDiscardMessage` would be a better choice."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. I think it should be done in the `setBinaryContent` method of the `Media` class.",
+      "reference": "what if `$media->getBinaryContent() == Symfony\\Component\\HttpFoundation\\File\\File` does `is_string()` return `true`? ping @greg0ire"
+    },
+    {
+      "prediction": "I think this should be `if (field.nested_flatbuffer && field.nested_flatbuffer->type.struct_def) {`",
+      "reference": "maybe rather than `bool` make this field a `StructDef *` ?"
+    },
+    {
+      "prediction": "I don't think we need to add to the waitgroup here, since we're not waiting for the blocks to be deleted.",
+      "reference": "Shall we maybe put this in a `const` like `backgroundTaskTimeout`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the tests to pass.",
+      "reference": "CI is complaining about this."
+    },
+    {
+      "prediction": "I think this method should be moved to the `HistoryRemovalTimeProvider` interface, so that it can be used by the `DefaultHistoryRemovalTimeProvider` as well as by any custom implementation of the `HistoryRemovalTimeProvider` interface.",
+      "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": "`sprintf('%s%s', $useLazyLoading ? 'lazy' : '', $isAttributeClassExistsAndNotEmpty ? ' ' . $attributes['class'] : '')` can be simplified to `sprintf('%slazy%s', $useLazyLoading ? '' : ' ', $isAttributeClassExistsAndNotEmpty ? ' ' . $attributes['class'] : '')`",
+      "reference": "I would prefer to set up space between classes in format pattern, `%s %s` and then use `trim()` to remove unnecessary whitespaces. This will also solve stripping whitespaces from the beginning and end of a string `$attributes['class']`"
+    },
+    {
+      "prediction": "`LOG(ERROR) << \"unknown setence type[\" << sentence_->getType() <<\"]\";` -> `LOG(ERROR) << \"unknown sentence type[\" << sentence_->getType() <<\"]\";`",
+      "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 don't think we need the `&&=` here. We can just do `options[:page_ranges] = Array(options[:page_ranges])`",
+      "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 `if err != nil { s.IPSetLocker.Unlock(ipset.Name) return nil, err }`",
+      "reference": "This is intended. Pls revert this."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the least bad option.",
+      "reference": "Something wrong with cast to `(Comparator<> & Serializable)`, need to be investigated"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "I think we need this. Not sure, I'll test."
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but it seems like it should still be there.",
+      "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": "I don't think we need this anymore.",
+      "reference": "Did go change their tags with the new version?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the best we can do for now. We need to be able to get the trust domain config from the data store, but we also need to be able to get the federates with config from the config file. I think this is the best we can do for now, but I'm open to suggestions.",
+      "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 think this should be `len(dep.Outputs()) > 0` instead of `len(dep.Outputs()) > 1`.",
+      "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 `new MetricsConfig.MetricsConfigBuilder().build()`",
+      "reference": "*NULL_DEREFERENCE:* object `null` is dereferenced by call to `meterSupplier(...)` at line 122."
+    },
+    {
+      "prediction": "I think this should be `readPreference.mode !== 'primaryPreferred'`",
+      "reference": "switch over to .mode"
+    },
+    {
+      "prediction": "Please revert this change. It's not needed.",
+      "reference": "I guess your IDE did some auto-fixing here"
+    },
+    {
+      "prediction": "nit: remove the empty line",
+      "reference": "This is an unused import failing recommit still."
+    },
+    {
+      "prediction": "Line is too long. [84/80]",
+      "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": "I don't think we need to add `aria-label` here.",
+      "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 don't think we need to mention GhostTable and SamplesGenerator here.",
+      "reference": "Should these link be removed?"
+    },
+    {
+      "prediction": "I don't think we need to pass `add_undo` to this method. We can just call `self.undo_stack.push(CloseTabCommand(self, tab))` in the `close_tab` method of `TabbedBrowser`.",
+      "reference": "Please make this a keyword-only argument by adding a `*` argument before `add_undo`."
+    },
+    {
+      "prediction": "I think we can get rid of this now.",
+      "reference": "didn't quite follow what this code block is doing."
+    },
+    {
+      "prediction": "I'm not sure if this is the best solution, but it seems to work for now.",
+      "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": "Shouldn't this be `if (dom._isControlled && ('value' in newProps || 'checked' in newProps))`?",
+      "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 `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 `None` instead of an empty string. `%` is not a valid directory name.",
+      "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 `RowKeySchema` instead of `RowKey`.",
+      "reference": "Nit: this table's row key map -> this table's row keys map"
+    },
+    {
+      "prediction": "I think this should be `PJ_LOG_WARN` or `PJ_LOG_DEBUG` instead of `PJ_LOG_NONE`.",
+      "reference": "this change should be reverted"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(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 the line number changed, but I think it's correct. The error is now on line 5, column 19, which is the start of the `accounts` block.",
+      "reference": "Revert this change and see below why..."
+    },
+    {
+      "prediction": "Why not use `coco_instance.py`?",
+      "reference": "Should not switch to `lvis_v1_instance` here because that base config uses ClassBalancedDataset to oversample the data."
+    },
+    {
+      "prediction": "I think this should be `info.getMaximumSimultaneousSessions()` instead of `builder.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 don't think we need this method. We can just call `Raise` with the `singleton` flag set to `true`.",
+      "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": "This is not used in this class.",
+      "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 \"Check metadata objects and their associated blocks for errors.\"",
+      "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 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": "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 `KindAwsEcsApp` to be consistent with the other kinds.",
+      "reference": "should be `ECSApp`"
+    },
+    {
+      "prediction": "I don't think this is correct. `self.size` and `self.location` are not defined for non-W3C elements.",
+      "reference": "flake8 is going to fail on this having 2 lines"
+    },
+    {
+      "prediction": "Why was this removed?",
+      "reference": "should we make a dir named controllers/scheduler/podchaos"
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "there seems to be one KeyNode in kubernetes.go of PV.. would it make sense to use it?"
+    },
+    {
+      "prediction": "I think it's better to keep `clicontext` as it is used in other commands.",
+      "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": "This should be `case MODPRIVATECHATOUT:`",
+      "reference": "Don't think this belongs in this pr"
+    },
+    {
+      "prediction": "I think this should return a copy of the map, not the map itself.",
+      "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 about this change. I think it should be `TaskCompletionSource<TResult> CloseCompletionSource { get; set; }` instead of `TaskCompletionSource<object> CloseCompletionSource { get; set; }`",
+      "reference": "@martijn00 so this PR introduces a new ViewModel lifecyle method? It isn't in the PR description/any new docs"
+    },
+    {
+      "prediction": "I think it is better to add this information to `img_metas[0]` instead of all `img_metas`, because `img_metas[0]` is the meta information of the first image in the batch.",
+      "reference": "Are these modification duplicate? Or should we move it into base detector."
+    },
+    {
+      "prediction": "I think this should be `public static byte[] dagSeed(final long block) {`",
+      "reference": "Since this is private and single use it should be un-wrapped inside of the two-arg dagSeed method."
+    },
+    {
+      "prediction": "I think this should be `$GLOBALS['TSFE']->tmpl->setup['plugin.']['tx_solr.'], $configuration` instead of `$configuration, $GLOBALS['TSFE']->tmpl->setup['plugin.']['tx_solr.']`",
+      "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 a `map[string]struct{}` so that we don't have to iterate over it to check if a return type is allowed.",
+      "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 we can remove this method and use `getInputFile(FileScanTask task)` instead.",
+      "reference": "I don't think this needs to change. Can you go back to the previous implementation?"
+    },
+    {
+      "prediction": "This should be `from molecule import Molecule`",
+      "reference": "Bump it 2 lines higher, 'c' > 'u'."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Please do not add author here"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "Similarly, we should be able to persist action uniformly"
+    },
+    {
+      "prediction": "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 think this should be `0755` instead of `0777`.",
+      "reference": "This would be better with context added via wrapping or fmt.errorf()"
+    },
+    {
+      "prediction": "I don't think we need this package. We can just use the `assert` package from `github.com/google/go-cmp/cmp` directly.",
+      "reference": "What's the purpose of this file?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix, but it seems to be the only way to get the passcodeLengthKnown flag to be set to true when the passcode length is changed.",
+      "reference": "@bhariharan Why was passcode length requirement allowed to be lowered?"
+    },
+    {
+      "prediction": "I think we should keep these imports.",
+      "reference": "This file needs a license header, too. Sorry I didn't catch that earlier."
+    },
+    {
+      "prediction": "I think this should be `string(genesisText)` instead of `string(genesisText[:])`",
+      "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": "Why do we need to use `homogen_table_builder` instead of `homogen_table`?",
+      "reference": "Why reference is converted to table?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the token is not 0, we should return an error.",
+      "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'm not sure if this is the right place to do this, but I think it makes sense to do it here. The reason is that we don't want to vectorize the function if it's already vectorized, and we don't want to vectorize the function if it's not vectorized, but we do want to vectorize the function if it's not vectorized but has a vectorized version.",
+      "reference": "What mechanism will be exposed so the caller knows they're using the vectorized version?"
+    },
+    {
+      "prediction": "I think this should be `flash: { error: t('.error') }` instead of `redirect_to forum_path(@forum), notice: t('.success')`",
+      "reference": "Can we remove respond_to block its not required here"
+    },
+    {
+      "prediction": "I think it would be better to use `dal::io::read` instead of `dal::io::csv::read`.",
+      "reference": "Do not do that even in example, we demonstrate bad practice"
+    },
+    {
+      "prediction": "Line is too long. [85/80]",
+      "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": "Why is this needed? `writer.Commit()` should be enough.",
+      "reference": "We have to `.Complete` now because of empty/null string test cases."
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "Is there still a way to specify no recursion?"
+    },
+    {
+      "prediction": "I don't think we need this struct. I think we can just pass in the `HumanReadableName` and `URISafeName` to the functions that need them.",
+      "reference": "How is this name being URI safe and being used programmatically related?"
+    },
+    {
+      "prediction": "I don't think we should be using `this.plugin.uppy.i18n` here. We should be using `this.uppy.i18n` instead.",
+      "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 what this test is testing, but it seems to be testing that `OpDocExtra` returns an empty string for `-`?",
+      "reference": "nit: shouldn't be part of this PR."
+    },
+    {
+      "prediction": "I think this should be `expect(e).to.be.an.instanceOf(MongoError);`",
+      "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": "Shouldn't this be `if (!_layout.IsThreadAgnostic && _cachedPrevRawFileName != null && _cachedPrevRawFileName.Length == reusableBuilder.Length)` ?",
+      "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": "Please use `LoggerFactory.getLogger(LeftListWidgetChooser.class)` instead of `org.slf4j.LoggerFactory.getLogger(LeftListWidgetChooser.class)`",
+      "reference": "Please use only `Logger` instead of `org.slf4j.Logger`"
+    },
+    {
+      "prediction": "I think this should be `if (!node.flags.disable_block_processor_republishing && !node.flags.disable_republishing)`",
+      "reference": "Should this be \"no limiter drop\", since this isn't an absolutely essential activity for the stability of the network?"
+    },
+    {
+      "prediction": "This should be `process.argv[2]` instead of `process.argv[1]`.",
+      "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: return False`",
+      "reference": "Could you simply add it to the previous clause ?"
+    },
+    {
+      "prediction": "`localeCompare` is not supported in IE11. We should use `Blockly.utils.string.numericalCompare` instead.",
+      "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": "Shouldn't this be `if (slaOptions != null && !slaOptions.isEmpty())`?",
+      "reference": "@chengren311 : where does this logic move to?"
+    },
+    {
+      "prediction": "I think this should be `if len(s.scope.ManagedMachinePool.Spec.RoleAdditionalPolicies) > 0 && !s.scope.AllowAdditionalRoles() { return ErrCannotUseAdditionalRoles }`",
+      "reference": "Can we have a `nil` exception check at `s.scope.ManagedMachinePool`"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but it seems like it should still be here.",
+      "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": "I don't think this is the right way to go about this. You should use `t()` on the property name, not `t()` on the result of `str_replace(\"_\", \" \", $property)` and `ucfirst()` on the result of `t()` on the result of `str_replace(\"_\", \" \", $property)` and `ucfirst()` on the result of `t()` on the result of `str_replace(\"_\", \" \", $property)` and `ucfirst()` on the result of `t()` on the result of `str_replace(\"_\", \" \", $property)` and `ucfirst()` on the result of `t()` on the result of `str_replace(\"_\", \" \", $property)` and `ucfirst()` on the result of `t()` on the result of `str_replace(\"_\", \" \", $property)` and `ucfirst()` on the result of `t()` on the result of `str_replace(\"_\", \" \", $property)` and `ucfirst()` on the result of `t()` on the result of `str_replace(\"_\", \" \", $property)` and `ucfirst()` on the result of `t()` on the result of `str_replace(\"_\", \" \", $property)` and `ucfirst()`",
+      "reference": "wrap, single line"
+    },
+    {
+      "prediction": "I don't think this is correct. If the return type is `void`, then the generic aware return type should also be `void`, not `Void`.",
+      "reference": "always use brackets for if statements"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do it, but I think we should move the `CIEnvironmentValues.DecorateSpan(null);` call to the `Initialize` method of the `NUnitIntegration` class. The reason is that the `Initialize` method is called when the profiler attaches to the process, so it's a good place to set the environment variables.",
+      "reference": "`NUnitIntegration` doesn't need to cache this anymore."
+    },
+    {
+      "prediction": "I think this should be `WHEN cardinality(parts) > 1 AND lower(element_at(parts, -1)) = 'gz'`",
+      "reference": "Why did you change it to `>=` here?"
+    },
+    {
+      "prediction": "I think we can remove the `else if (tableName.startsWith(\"hive.\"))` part.",
+      "reference": "Looks like this needs to be updated. There is no need to remove `hadoop.` if Hadoop tables don't use this code path."
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "reference": "This require isn't need anymore, right?"
+    },
+    {
+      "prediction": "I think it's better to move this import to the top of the file.",
+      "reference": "a bit weird to see payments crypto package inside connection manager. There probably will be services not having any relation with payments that will use connection manager. Thats why probably ping-ping by itself can be useful without any relation to payments.."
+    },
+    {
+      "prediction": "Shouldn't this be `$xml->record->addAttribute('type', $this->xmlType);`?",
+      "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 100, not 50.",
+      "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 to do this. I think it would be better to do this in the `kinto` package.",
+      "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": "I think we can remove the `hashivault` import here.",
+      "reference": "? The package is named `hashivault`."
+    },
+    {
+      "prediction": "I think this should be `config->xwayland = false;`",
+      "reference": "I don't think this should enable xwayland, because it's convenient to toggle xwayland just by setting `xwayland` to `false`"
+    },
+    {
+      "prediction": "nit: please add a blank line here",
+      "reference": "Remove this line by accident?"
+    },
+    {
+      "prediction": "I think we can remove the `import org.gradle.api.plugins.JavaPluginConvention;` as well.",
+      "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": "This should be `if 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 `ip_dscp` instead of `nw_tos`.",
+      "reference": "What is the different between nw_tos and ip_dscp? Only high 6 bits vs low 6 bits and supported version?"
+    },
+    {
+      "prediction": "Shouldn't this be `StringData = MolToCXSmiles(*mol);`?",
+      "reference": "I'm assuming that asSmarts & cxSmiles == asSmarts"
+    },
+    {
+      "prediction": "I'm not sure what's going on here, but it seems like the `param` attribute is not being restored by the `__setstate__` method of `LabelledData`.",
+      "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 `internal` access modifier here.",
+      "reference": "Curious, do we consider this as pedantic for example code?"
+    },
+    {
+      "prediction": "I don't know if it's a good idea to use the entity name as the form name. What if there are two entities with the same name?",
+      "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": "Shouldn't this return `client.outboundIp` instead of `client.ip`?",
+      "reference": "Should be `client.outboundIp`"
+    },
+    {
+      "prediction": "Please remove unused usings.",
+      "reference": "Must these `using` statements not go inside the `namespace` block?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be able to get rid of the `fboMutexLevel` type and just use `mutexLevel` instead.",
+      "reference": "Why? I thought the previous way was idiomatic. (Same with the `MetadataVer` stuff above.)"
+    },
+    {
+      "prediction": "I'm not sure why this was removed. It looks like it should still be here.",
+      "reference": "shouldn't we `reject` here instead?"
+    },
+    {
+      "prediction": "`gasPriceUnitIOTX, err := util.StringToIOTX(action.Core.GasPrice)` -> `gasPriceUnitIOTX := util.StringToIOTX(action.Core.GasPrice)`",
+      "reference": "check err or use gasPriceUnitIOTX, _ := if we are sure action.Core.GasPrice is correct"
+    },
+    {
+      "prediction": "nit: I think we can remove the `wait` parameter from this method.",
+      "reference": "Just curious, was wait not used at all before?"
+    },
+    {
+      "prediction": "Why did you remove this?",
+      "reference": "nit: can we remove this?"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "is it not a problem when one synchronized function is called by other?"
+    },
+    {
+      "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 (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": "`Este :attribute j\u00e1 est\u00e1 ligado.` -> `Este :attribute j\u00e1 foi ligado.`",
+      "reference": "\"Este :attribute j est anexado.\" sounds better."
+    },
+    {
+      "prediction": "I don't think we need to import `frame` and `series` here.",
+      "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": "I think this should be `var errNoCert = errors.New(\"no certificate available\")`",
+      "reference": "This error was unused"
+    },
+    {
+      "prediction": "I don't think we need this struct. We can just use `CloseShardRequest{ShardID: shardID}`.",
+      "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": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Does this make more sense as `trails/_trail` now?"
+    },
+    {
+      "prediction": "I think this should be `outpostARN, err := agent.ec2MetadataClient.OutpostARN()` instead of `outpostARN, err := agent.ec2MetadataClient.GetOutpostARN()`",
+      "reference": "synced offline. let's move this to `agent_unix.go` to make the IMDS call, since this is not supported in Windows."
+    },
+    {
+      "prediction": "Please remove these empty lines.",
+      "reference": "Too much indentation here. Should match the lines above, which have four spaces."
+    },
+    {
+      "prediction": "I think this test should be kept.",
+      "reference": "Same here we still need this test to make sure the fallback works."
+    },
+    {
+      "prediction": "I'm not sure about the name of this interface. I think it should be `MvxOverridePresentationAttribute` instead of `IMvxOverridePresentationAttribute`.",
+      "reference": "I'm wondering if we can actually base this on a `IMvxPresentationAttribute` instead of the base one."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to fix the issue for me.",
+      "reference": "could we switch to `undefined` here?"
+    },
+    {
+      "prediction": "I think this should be `SyncState.setupSyncsSoupIfNeeded(smartStore);` instead of `SyncState.cleanupSyncsSoupIfNeeded(smartStore);`",
+      "reference": "The cleanup call"
+    },
+    {
+      "prediction": "I don't think this is a good idea. Albums can have multiple artists, and we want to be able to view all of them. If you want to view a specific artist, you should be able to do so by clicking on the artist's name in the album view.",
+      "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": "I think this should be `defined?(Puppet::Pops::Types::PSensitiveType::Sensitive) && 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": "This is not needed. You can just run `python test_temporal_memory.py` from the command line.",
+      "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 `body.append(self._get_selenium_options())` instead of `body.extend(self._get_selenium_options())` because `self._get_selenium_options()` returns a list.",
+      "reference": "1 - create browser specific options class: - chromeoptions - firefoxoptions - argsoptions"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to fix this issue, but it seems to work for me.",
+      "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 method. I think it would be better to have it in the `Account` model.",
+      "reference": "We must have a `fail ParamNotFound` here for cases where `@account.nil?`."
+    },
+    {
+      "prediction": "I think we can remove this `#if` now that we're targeting .NET Standard 2.0.",
+      "reference": "How come this change was needed?"
+    },
+    {
+      "prediction": "I don't think we need to expose these. They are only used internally by the ADC driver.",
+      "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 think this should be `gsf.ssb.Matcher()` instead of `gsf.ssb.ExploreFields(func(messagesSelector selectorbuilder.ExploreFieldsSpecBuilder) { messagesSelector.Insert(\"secpRoot\", gsf.ssb.Matcher()) })`",
+      "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 `config.action_mailer.asset_host = \"http://localhost:5000\" || \"http://2c429f18.ngrok.io\"`",
+      "reference": "what is this default about? I am not using ngrok atm so would prefer an env var."
+    },
+    {
+      "prediction": "I think this should be `false` by default.",
+      "reference": "@Samuell1 Might be better to say \"Automatically check for plugin updates on login\"."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "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 we can remove the `DisableDockerHealthCheck` from this test.",
+      "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 we should be able to just use `dir` instead of `srcDir` here. I don't think we need to create a `src` directory.",
+      "reference": "We should fail the test if this returns `err != nil`."
+    },
+    {
+      "prediction": "nit: I think it would be clearer to call this `shouldIncrementProvisionFailedTerminal` or something like that.",
+      "reference": "What is the drawback of not having this flag as a gating condition to report the metric?"
+    },
+    {
+      "prediction": "I don't think we need this. We can just use `os.Args[0]` directly.",
+      "reference": "`getProcessName()` will be less ambiguous, since Caddy has \"apps\" that it runs internally."
+    },
+    {
+      "prediction": "Is there a way to get the size of a pointer from the debugger?",
+      "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": "Why do we need a lambda here? Why can't we just call `buildFilter(req.filter_ref())`?",
+      "reference": "How about just override the `buildFilter`? Passing another function here is a little weird."
+    },
+    {
+      "prediction": "I think this should be a boolean flag, not a string.",
+      "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 think this should be `case swift::MetadataKind::Function:`",
+      "reference": "Is this needed?"
+    },
+    {
+      "prediction": "I think this should be `name: value` instead of `name: value.concat(' value')`",
+      "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 we should move this package to `chaos-mesh/pkg/selector/generic`",
+      "reference": "how about moving it under L26"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `utils.js` and not `utils/memoize.js`",
+      "reference": "I think this needs to be tested"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place for this constant, but it's the only place I could think of.",
+      "reference": "Style/MutableConstant: Freeze mutable objects assigned to constants."
+    },
+    {
+      "prediction": "I don't think we need this field anymore.",
+      "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 don't think this should be public.",
+      "reference": "Looks like this should be publicly accessible for tests in any modules?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. If the `.env.local.php` file is not found, we should not try to load the `.env` file.",
+      "reference": "Allow putenv is necessary to be able to get environment variables with `getenv`?"
+    },
+    {
+      "prediction": "Please add a newline at the end of the file.",
+      "reference": "use function to check it. - don't do everything in main"
+    },
+    {
+      "prediction": "I think we can remove the `global_configs` parameter from this method. It's only used in `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `BaseRulesEngine.build_rule_book`, which is only called by `Base",
+      "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   }, { /* vex_W_ext 69 */\n-    {OP_vgatherqps,0x66389318,\"vgatherqps\",Vvs,Hx,MVd,Hx,xx, mrm|vex|reqp,x,END_LIST},\n+    {OP_vgatherqps,0x66389318,\"vgatherqps\",Vvs,Hx,MVq,Hx,xx, mrm|vex|reqp,x,END_LIST},\n     {OP_vgatherqpd,0x66389358,\"vgatherqpd\",Vvd,Hx,MVq,Hx,xx, mrm|vex|reqp,x,END_LIST},\n   }, { /* vex_W_ext 70 */\n     {OP_vfmadd132ps,0x6638C1C0,\"",
+      "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": "Should this be `self.build_tracking_url(logs_output)`?",
+      "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 `!canSeeInvisibility() && creature->isInvisible()`",
+      "reference": "You introduced a bug here."
+    },
+    {
+      "prediction": "Shouldn't this be `string.IsNullOrEmpty(oldValue) && !string.IsNullOrEmpty(newValue)`?",
+      "reference": "Should we log a warning/error if only one is specified and not the other?"
+    },
+    {
+      "prediction": "I don't think we need to implement this method in the fake client.",
+      "reference": "`ctx` is unused in GetLatestEvent"
+    },
+    {
+      "prediction": "I think this should be `if (config.node.peering_port == 0)`",
+      "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": "remove empty line",
+      "reference": "The blank line between the summary and detailed description is better to be kept."
+    },
+    {
+      "prediction": "I think this should be `if a.RemoveUnusedShapes`",
+      "reference": "Curiously, are these indents intended or should they be aligned?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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 I understand why we can't use `safe_read_tls_magic` here. If the thread has already lost its TLS, shouldn't `safe_read_tls_magic` return `NULL`?",
+      "reference": "This will still result in a safe_read_tls_magic on AMD in tls_thread_preinit()."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Can you remove this change?"
+    },
+    {
+      "prediction": "I think this should be `field.Invalid(specPath.Child(\"isCA\"), crt.IsCA, \"ACME does not support CA certificates\")`",
+      "reference": "Oh, was that a bug in the validation?"
+    },
+    {
+      "prediction": "I think this should be `startsWith` instead of `contains` so that we don't accidentally remove the wrong part of the URL.",
+      "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 don't think you need to delete `val1` and `val2` here.",
+      "reference": "Actually, I don't think we need to create object on heap."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the generated code to compile.",
+      "reference": "this is a file we copy from gRPC, sp ideally it be fixed upstream as well.."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "changes like this, just strangle guava usages to be replaced by DelayLimiter into one place"
+    },
+    {
+      "prediction": "I'm not sure about this change. The `Header` and `Footer` props are used to render the header and footer of the widget. If you remove the `Header` and `Footer` props, the header and footer won't be rendered anymore.",
+      "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 think this should be `strncmp(resource_str, \"id:\", 3) != 0`",
+      "reference": "maybe extra paren around `strncmp(...) != 0`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to call this, but it seems like a good place to me.",
+      "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 don't think we need to make this generic. It's only used in `BftConsensus` where `C` is `BftConfigOptions`, so we can just make it `BftForksSchedule<BftConfigOptions>`.",
+      "reference": "Rename to ForksSchedule? Maybe in separate PR."
+    },
+    {
+      "prediction": "I think we can remove the `from kinto.core.cache import heartbeat` line.",
+      "reference": "Why a blank line here?"
+    },
+    {
+      "prediction": "I think this should be `deactivated_at.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 if this is the right thing to do, but I think it makes sense to skip the default value for positional parameters.",
+      "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 `msg.addByte(player->getVocation()->getId() != VOCATION_NONE ? 0x01 : 0x00);` instead of `msg.addByte(player->getVocation()->getId() != VOCATION_NONE ? 0x01 : 0x00);`",
+      "reference": "Why would you add those if client version min is set to 1100?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "You can delete this, it will now need to live in `gointerfaces` package"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the title to render as HTML.",
+      "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 think this should be `mActionBarSubject.setText(subject);` instead of `mActionBarSubject.showSubjectInMessageHeader();`",
+      "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 think we should keep the default sampler as `ProbabilitySampler(defaultSamplingProbability)`.",
+      "reference": "Could you also remove the `defaultSamplingProbability` constant from `sampling.go`? It seems to became unused with this change."
+    },
+    {
+      "prediction": "I think we can remove the `appsv1.Deployment{}` argument here and in the `GenerateStrategicPatch` function.",
+      "reference": "can you delete the `generateDeploymentStrategicPatch` function as well?"
+    },
+    {
+      "prediction": "I don't think we need to log this at `finest` level.",
+      "reference": "I'd keep this at `info` level..."
+    },
+    {
+      "prediction": "Why do we need this check?",
+      "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 thing to do. I think we should keep the `getImage` and `setImage` methods, and have them throw an `UnsupportedOperationException` if the node is not an `AntlrNode`.",
+      "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 done in `invokeCleanup` and `invokeEffect`.",
+      "reference": "react bails as well?"
+    },
+    {
+      "prediction": "This constant is not used anymore, so it can be removed.",
+      "reference": "Can you please tell me why you did this?"
+    },
+    {
+      "prediction": "I think this should be `videoConstraints.facingMode = this.opts.facingMode || videoConstraints.facingMode`",
+      "reference": "Reading this again it should prob prefer the `videoConstraints.facingMode` value over `facingMode` if the former was already set"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "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.Target.WasmAbi == \"js\" && strings.HasPrefix(config.Triple(), \"wasm\") {`",
+      "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 don't think we need to defer the stop here. We can just call it at the end of the function.",
+      "reference": "The nil check is not necessary for these. Append checks both sides for nil."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should only set the attributes if they are not already set.",
+      "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 we should return an error here.",
+      "reference": "nit: rename this in proto and update impl to `Search` to avoid stutter"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "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 don't think we need to make these constants public. They are only used in this file.",
+      "reference": "`ClaimFromRewardingFundBaseGas` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I don't think you need the NOGIL here.",
+      "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 don't think we need this interface. We can just pass in the `BlockTimer` directly to the `Sender`.",
+      "reference": "blockheighter would be more clear"
+    },
+    {
+      "prediction": "I think it's better to use `std::min` and `std::max` instead of `std::swap`.",
+      "reference": "This is so weird as a utility function."
+    },
+    {
+      "prediction": "I think we should be able to use `cmpopts.IgnoreUnexported` instead of `cmp.AllowUnexported` here.",
+      "reference": "Why? What would happen if you did not?"
+    },
+    {
+      "prediction": "I don't think this is necessary. If `o` is `null` then `((JSONObject) o).opt(pathElement)` will return `null`.",
+      "reference": "Unrelated bugfix for NPE."
+    },
+    {
+      "prediction": "I don't think we should change the exception message here.",
+      "reference": "What if some extended the `AdminController` and had a check for the old option name? Could that be a valid use case?"
+    },
+    {
+      "prediction": "I think we should use `countlyCommon.API_PARTS.data.r + '/data-manager/event-segment'` instead of `countlyCommon.API_PARTS.data.r + '/data-manager/event-segment'`",
+      "reference": "This will be only available when data-manager is enabled, is there a fallback in case data manager is disabled?"
+    },
+    {
+      "prediction": "I think this should be `In JavaScript you can write %s.`",
+      "reference": "Do you think we want to move the call to `generateWorkerServiceRecommendedActions` inside `RecommandedActions()`?"
+    },
+    {
+      "prediction": "This should be `VUID-vkGetQueryPoolResults-flags-00814` and `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 think it's better to have `ToDbKey` as an extension method for `long` and `Keccak`.",
+      "reference": "Iguess you should use ToDbKey here"
+    },
+    {
+      "prediction": "Why do we need to subtract 1 here?",
+      "reference": "x.tbk is the \"index\" below, and is done in Start()"
+    },
+    {
+      "prediction": "I think this should be `DD_APPINSIGHTS_INSTRUMENTATIONKEY` instead of `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 it would be better to check if `ctx->log_key` and `ctx->time_key` are not `NULL` instead of checking if they are not `NULL` and not empty.",
+      "reference": "this assignment is not necessary if the offsetof() is used in the configmap"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "I'd rather leave a TODO here at least."
+    },
+    {
+      "prediction": "I think this should be `\"\"`, not ````",
+      "reference": "Shouldn't this be `resource`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we can remove the `req` argument from this function. It's only used in the GPU implementation.",
+      "reference": "Why does this become blocking for the CPU path? Shouldn't it remain independent of the GPU path?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I think it would be better to rename this method to `configure_rvm` or something like that. `setup_rvm` sounds like it's setting up RVM, but it's actually just exporting the `TRAVIS_RUBY_VERSION` environment variable.",
+      "reference": "Do we need to change occurences where `setup` was called before?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this, but it's the only place I could think of.",
+      "reference": "Is this TODO still valid? Or are we going to put this somewhere else later?"
+    },
+    {
+      "prediction": "I think this should be `if from_ts is None:`",
+      "reference": "We generally don't do parantheses in if conditions in Python. :) this could be better written as `if from_ts is None`"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "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 don't think we need a `ScoreFunc` interface. We can just pass a `func(swarm.Address) float32` to the `ForEachPeer` method.",
+      "reference": "ScoreFunc is not used in this PR. Maybe to add it later when needed?"
+    },
+    {
+      "prediction": "I think this should be `if len(redirects) > 0 && !isXHR(copyRequestHeadersFromResponseWriter(w))`",
+      "reference": "super nit: I think this might be easier to read as `requestHeadersFromResponseWriter`"
+    },
+    {
+      "prediction": "Please remove the empty line.",
+      "reference": "Need to include \"common/base/Base.h\" in front of this line"
+    },
+    {
+      "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's a better way to do this, or if this is the best way to do this.",
+      "reference": "Changes in this file should go into a separate PR."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should be using double quotes here.",
+      "reference": "Please ensure that this will not start putting single quotes where double quotes are expected."
+    },
+    {
+      "prediction": "I'm not sure why we need to change the signature of this function. I think we can keep it as it is.",
+      "reference": "`ProvideConfig` was changed, looks like it will not compile for windows, and should be changed too."
+    },
+    {
+      "prediction": "I think this should be `models.GetUserRepositories(u.ID, true, 1, u.NumRepos, models.ListOptions{Page: ctx.QueryInt(\"page\"), PageSize: ctx.QueryInt(\"limit\")})`",
+      "reference": "This does not look right, you're listing all private repositories.."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "An OPTIONS request should respond from _any_ path? Seems weird to me..."
+    },
+    {
+      "prediction": "I don't think we should add these fields to the media admin. The media admin is only used to manage media, not to edit them. If you want to add these fields, you should add them to the media edit form in the media bundle.",
+      "reference": "you need to add the content type"
+    },
+    {
+      "prediction": "I think it would be better to have a `StopSubProcesses` method in the `ProcessHelper` class and call it from here.",
+      "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 think this should be `#if defined(AARCH64) || defined(X64) || defined(IA32)`",
+      "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 think this should be `ZIPKIN_SPANS.TRACE_ID_HIGH.eq(ZIPKIN_ANNOTATIONS.TRACE_ID_HIGH).and(ZIPKIN_SPANS.TRACE_ID.eq(ZIPKIN_ANNOTATIONS.TRACE_ID))`",
+      "reference": "guess I'm wondering if this needs to be refactored to use Schema.joinCondition() or similar?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 if this is the right thing to do, but it seems to be the only way to get this to work.",
+      "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": "I don't think this is correct. If you're in deadman, you should be able to attack people in the wilderness.",
+      "reference": "This should be `WorldType.isDeadmanWorld(client.getWorldType())` to be inline with the other WorldType calls."
+    },
+    {
+      "prediction": "`mlvl_confids` -> `mlvl_confidences`",
+      "reference": "do we also need to change `mlvl_confid` -> `mlvl_confidences`>"
+    },
+    {
+      "prediction": "I think this should be `config:\"oneof(kube-ipvs0,none);none;non-zero,die-on-fail\"`",
+      "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 don't think we need `re.fullmatch` here. `re.match` should be enough.",
+      "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'm not sure if this is the right place to do this, but I think it would be good to add a check for `obj == nil` here and return an error if it is.",
+      "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": "Why is this change needed?",
+      "reference": "Does codecov not run this test? Not sure how else it would not be covered."
+    },
+    {
+      "prediction": "`multisend 'JSON_DATA'` -> `multisend JSON_DATA`",
+      "reference": "can we add some sample of JSON_DATA format in the usage?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should keep the `scopeManager` and use it to get the `DistributedSpanContext` from the `Active` `IScope`.",
+      "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": "I think this should be `projectIdVariableName` instead of `projectIdToken`.",
+      "reference": "make all of these `static final` fields UPPER_SNAKE - they are constants."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but I think it should be `total_dotshielding = itembonuses.DoTShielding + itembonuses.MitigateDotRune[SBIndex::MITIGATION_RUNE_PERCENT] + aabonuses.MitigateDotRune[SBIndex::MITIGATION_RUNE_PERCENT] + spellbonuses.MitigateDotRune[SBIndex::MITIGATION_RUNE_PERCENT];`",
+      "reference": "No possible chance for weird overflows here, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this constant. I think it should be in the `TwoPhaseIterator` class.",
+      "reference": "Or we could use a Float object to more clearly show as user-settable via non-null?"
+    },
+    {
+      "prediction": "Why do we need to pass `finalizedHeader` to `ForkchoiceUpdated`?",
+      "reference": "finalizedHeader should be saved in FinalizationManager when we have FinalizationBlockHash != Keccak.Zero"
+    },
+    {
+      "prediction": "I think we should use `dataDiskIndex` instead of `dataDiskIndex` in the scratch disk name.",
+      "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": "Why do we need to select all columns here?",
+      "reference": "If we support multi-index column later, we need to rename to fit the pandas' requirement."
+    },
+    {
+      "prediction": "I think this should be `map[string]interface{}` instead of `map[string]interface{}`.",
+      "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 `Container::startDecaying()` instead of `Item::startDecaying()`.",
+      "reference": "sorry for nitpicking but I would love a new line under this line"
+    },
+    {
+      "prediction": "I think this should be renamed to `ChannelType` or `ChannelImplementationType` or something like that.",
+      "reference": "ChannelImplementation field naming is not clear, is it some kind of standard? Maybe it can be named ChannelID?"
+    },
+    {
+      "prediction": "I think this should be `default RemoveFiles removeFiles(String metadataLocation) { throw new UnsupportedOperationException(this.getClass().getName() + \" does not implement removeFiles\"); }`",
+      "reference": "Looks like there is a typo: `expireSnapshots` -> `removeFiles` or whatever name we go with."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the current sort order is null, then we should not be showing the sort dialog at all.",
+      "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": "I don't think we need to add new constants for these. We can just use the existing ones.",
+      "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 if this is a good idea, but I think it is better to catch the exception and print the error message than to let the program crash.",
+      "reference": "Should this be boost logged?"
+    },
+    {
+      "prediction": "session is not used in this function.",
+      "reference": "What is the reason for doing expunge here? This method is purely for converting the timestamp."
+    },
+    {
+      "prediction": "I think it would be better to add the original field name as a property on the schema itself, rather than on each field. That way, we don't have to iterate over all the fields to get the original field name.",
+      "reference": "This calls sanitize twice if the name isn't valid."
+    },
+    {
+      "prediction": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.",
+      "reference": "Is there an equivalent that'd allow zero or one?"
+    },
+    {
+      "prediction": "I think this should be `rwExec.IsTaskObjectNameRepeat()` instead of `rwExec.isTaskObjectNameRepeat()`. This is because `rwExec.IsTaskObjectNameRepeat()` is a public method, while `rwExec.isTaskObjectNameRepeat()` is a private method.",
+      "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": "Why is this change needed?",
+      "reference": "unnecessary conversion (from `unconvert`)"
+    },
+    {
+      "prediction": "I don't think you need to remove this.",
+      "reference": "typo in commit message `openening`"
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "User-code doesn't see the internal ADIOS headers anymore. This whole include block should just be `#include <adios2.h>`"
+    },
+    {
+      "prediction": "I don't understand why this change is needed.",
+      "reference": "nit: I guess we can omit this (unless you had a reason to specify it)"
+    },
+    {
+      "prediction": "I think this should be `c.leaf.remote != nil && !c.leaf.remote.Hub`",
+      "reference": "Why is Hub public?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "why is this deleted?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this, but I think it should be here.",
+      "reference": "Maybe move that into a partial"
+    },
+    {
+      "prediction": "I don't think we need to check the error here.",
+      "reference": "Error return value of `dao.Stop` is not checked (from `errcheck`)"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "reference": "util.Warning()? Easier to say."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `UnmarshalTrieNode` instead.",
+      "reference": "this default is kinda superfluous (although i see it was already there before PR )"
+    },
+    {
+      "prediction": "I don't think this change is necessary. `parent` is not used in the constructor.",
+      "reference": "How is this related?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but I think it makes sense to have all the plans on the same page.",
+      "reference": "I'm breaking one of the rules here, it feels like the right thing to do. Open to alternative suggestions."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 think this should be `if (chit != m_changesForReader.end() && chit->isValid())`",
+      "reference": "Why are you setting the CacheChange as REQUESTED when it is not valid (it was erased from history)?"
+    },
+    {
+      "prediction": "This should be `import { isEmpty } from 'lodash-es';`",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "Why is this constructor public?",
+      "reference": "This needs to be public so that Avro can call it."
+    },
+    {
+      "prediction": "I think we can remove the `map` and `isNull` imports here.",
+      "reference": "`Google*` shouldn't be under External dependencies - seems like ESlint is not properly recognizing that these are aliases to internal dependencies."
+    },
+    {
+      "prediction": "Why do we need `assert index is None` here?",
+      "reference": "I think he meant the docstring in this constructor. yea we should fix"
+    },
+    {
+      "prediction": "I don't think we need this method. It's only used in one place, and it's only used in one place in a way that can be replaced with `Files.newInputStream(Paths.get(path))`.",
+      "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'm not sure if this is the right thing to do, but it seems to be the only way to get this to work on Windows 10.",
+      "reference": "It would probably be better to do: const SecurityProtocolType Tls13 = (SecurityProtocolType)12288"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Similar to the above, put this setting back but keep the value as `'{}'`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to log this. I think it should be logged in the `refresh_continuous_aggregates` function.",
+      "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 this is the best way to do this, but it works for now.",
+      "reference": "I'm curious about why this needed to be removed."
+    },
+    {
+      "prediction": "I don't think we should import `plt` from `scapy.extlib` here. We should import it from `scapy.extlib` only when `scapy.extlib.MATPLOTLIB_INLINED` is `True`. Otherwise, we should import it from `scapy.extlib` only when `scapy.extlib.MATPLOTLIB_INLINED` is `False`.",
+      "reference": "why did you remove `zip`? It is used!"
+    },
+    {
+      "prediction": "Why remove the reason?",
+      "reference": "Is there any reason that we don't add a similar condition like `ChannelConditionTopic` to the channel?"
+    },
+    {
+      "prediction": "This should be `namespace TestVendor\\Goto;`",
+      "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 think this should be `NewSingleOutboundTLS` or `NewSingleOutboundTLSConfig` or something like that.",
+      "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'm not sure this is the right thing to do. If the response body is empty, we don't need to close it. If the response body is not empty, we should be able to read from it and then close it.",
+      "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 why this change was made. I don't think it should be removed.",
+      "reference": "We're already matching +. Not sure why this changed."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. If you want to use your own category manager, you should be able to do it without having to change the configuration.",
+      "reference": "Please add line breaks, so the line doesn't exceed 80 chars."
+    },
+    {
+      "prediction": "I think we should keep the `ActivitySourceAdapter` in the constructor and pass it to the `HttpInListener` instead of creating a new `ActivitySourceAdapter` in the `HttpInListener` constructor.",
+      "reference": "I initially thought (inccoreclty) this is a breaking change! The public api analyzer is a gift!"
+    },
+    {
+      "prediction": "I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the `GPtrArray` here, do we? I don't think we need to free the",
+      "reference": "I'm not sure how this one didn't segfault before."
+    },
+    {
+      "prediction": "I don't think these are used anywhere.",
+      "reference": "Same points as in other test file."
+    },
+    {
+      "prediction": "Please use `import java.nio.charset.*;` instead of `import java.nio.charset.*;` and `import java.io.*;` instead of `import java.io.*;`",
+      "reference": "Did you run gradlew tidy? Wildcard imports shouldn't be there, hence the question."
+    },
+    {
+      "prediction": "I'm not sure why we need to pass in a `StatusFinder` here. I think we should be able to use the `StatusFinder` from the `TrailWithProgress` instance itself.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I think this should be `FPGA_IFC_OPAE` instead of `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. I think it should be done in the VM.",
+      "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": "`EMRFS` -> `EMRFS`",
+      "reference": "electronic medical record file system (EMRFS)"
+    },
+    {
+      "prediction": "I think this should be `maxrequestbody` instead of `maxrequestbody`",
+      "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 don't think we need to remove these errors. They are used in other places.",
+      "reference": "oh come on :("
+    },
+    {
+      "prediction": "I don't think we need to log this at `finest` level. I think `info` is fine.",
+      "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 authenticated user is a service account`",
+      "reference": "Remove as this is not needed anymore."
+    },
+    {
+      "prediction": "This should be `int` not `Optional[int]`.",
+      "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 to add this. I think it should be added to the `Certificate` struct in `pkg/apis/certmanager/v1alpha1/certificate_types.go` and then used in the `buildCertificate` function in `pkg/controller/certificates.go`.",
+      "reference": "Will this cause all certificates to be re-issued?"
+    },
+    {
+      "prediction": "(style) trailing whitespace",
+      "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 `keyProjection != null && keyProjection.schema() != keyField.schema() || valueProjection.schema() != valueField.schema()`",
+      "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": "`spaceId < 0` -> `spaceId == -1`",
+      "reference": "Why not use StatusOr ? We can't ensure spaceId is greater than zero, especially when AdHocSchemaManager is used."
+    },
+    {
+      "prediction": "why we need to import `github.com/chaos-mesh/chaos-mesh/pkg/chaosdaemon/pb` here?",
+      "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 formatConfig[\"reportTables\"])\n \t\t\tor (role in (controlTypes.ROLE_LIST, controlTypes.ROLE_LISTITEM) and not controlTypes.STATE_READONLY in states and formatConfig[\"reportLists\"])\n \t\t\tor (role in (controlTypes.ROLE_FRAME, controlTypes.ROLE_INTERNALFRAME) and formatConfig[\"reportFrames\"])\n```\nReview comment?",
+      "reference": "Could you split this into multiple lines?"
+    },
+    {
+      "prediction": "nit: remove unused imports",
+      "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 why this change is necessary, but it seems like it should be.",
+      "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 don't think we need to create a new state view here. We already have one in `c.state.StateView(parentStateRoot)`.",
+      "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 we need to check if the hash is empty or not. If it is empty then only we need to calculate the hash and update it.",
+      "reference": "need to fix the error message here"
+    },
+    {
+      "prediction": "Why is this not fixed?",
+      "reference": "Unsure of how to handle this one."
+    },
+    {
+      "prediction": "I think this should be `if (this.isKilled() || this.flowFinished) {` instead of `if (isKilled() || this.flowFinished) {`",
+      "reference": "Will this.flowFinished verify with every finished state stated in Status.isStatusFinished method? If so, no need to verify isKilled."
+    },
+    {
+      "prediction": "I don't think we need this struct. We can just use the `RetryPolicy` struct directly.",
+      "reference": "replace \"Child Workflow\" with \"any workflow\""
+    },
+    {
+      "prediction": "I think this should be `{{- if or .IsSymlink .IsDir}}`",
+      "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 don't think we need to keep the `@internal` tag here.",
+      "reference": "Please add back the description of the parameter"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Superfluous space inserted."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it would be better to use `getAttribute` in all cases, and only use `node[attribute]` if `getAttribute` returns `null` or `undefined`. That way we don't have to worry about `ownerSVGElement` at all, and we don't have to worry about `getAttribute` returning `null` or `undefined` when `ownerSVGElement` is `false`.",
+      "reference": "Didn't fix the problem. `href=\"\"` for SVG will still result in `undefined` getting returned by this function."
+    },
+    {
+      "prediction": "I'm not sure why the bearing changed, but I think this is correct.",
+      "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 `isCaret=True`.",
+      "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": "I think we can remove the `.freeze` here as well.",
+      "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'm not sure this is the right place to sort the subjects. I think we should sort them when we create the `AutomationPolicy` in the first place. That way, we don't have to sort them every time we consolidate the automation policies. What do you think?",
+      "reference": "Why is this needed?"
+    },
+    {
+      "prediction": "I think this should be `/users/:username/repos` and `/orgs/:org/repos`",
+      "reference": "Why not put two other routes after this line (221)?"
+    },
+    {
+      "prediction": "`addrs.value().empty()` should be `addrs.value().size() == 0`",
+      "reference": "If FLAGS_meta_server_addrs is not empty, in which case \"toHosts\" return empty array?"
+    },
+    {
+      "prediction": "This should be `Settings.JSONBoolWithQuotes(DEFAULT_JSON_USE_QUOTES);` instead of `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 think we need to update `copy` method in `InternalFrame` to take `scol` as an argument.",
+      "reference": "oops, it should be `data._scol` so that it respects the expression stored in Series. Let me fix it quick."
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "Thanks for moving this up with the rest of the variables. Much tidier :)"
+    },
+    {
+      "prediction": "I think this should be `0.001 * money.Myst` instead of `money.NewMoney(0.001, money.CurrencyMyst)`",
+      "reference": "I think `Rate` is a more precise and concise term for `PaymentPerTime`."
+    },
+    {
+      "prediction": "I think this should be `if (limit < 0 || limit < posNext)`. If `limit < 0` then `limit < posNext` will always be true.",
+      "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": "This should be `if (true === ConfigQuery::checkAvailableStock()) {`",
+      "reference": "if `checkStockNotEmpty` returns a reponse, you must return it or your script will continue its execution."
+    },
+    {
+      "prediction": "I think we can remove the `v2` from the import here.",
+      "reference": "Any specific reason for this change? klog should work correctly here."
+    },
+    {
+      "prediction": "I think this should be `ON CONFLICT (id, parent_id, collection_id) DO UPDATE SET data = EXCLUDED.data, last_modified = from_epoch(:last_modified) RETURNING id, as_epoch(last_modified) AS last_modified;`",
+      "reference": "we don't mention `last_modified` here?"
+    },
+    {
+      "prediction": "I don't think we need to use `hyperx` here. `preact-compat` should be enough.",
+      "reference": "same deal about the global state maybe interfering as in ActionBrowseTagline"
+    },
+    {
+      "prediction": "I think it would be better to use `Assert.Throws<ArgumentOutOfRangeException>(() => new LruCache<int, int>(maxCapacity, \"test\"));` instead of `Assert.Throws<ArgumentOutOfRangeException>(() => { new LruCache<int, int>(maxCapacity, \"test\"); });`",
+      "reference": "so sweet to see a test as the first thing"
+    },
+    {
+      "prediction": "I think this should be `if (config->http_proxy != NULL && strcmp(config->http_proxy, \"\") == 0) {`",
+      "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 the right way to do it. If the width is not set, the height should not be used to calculate the width.",
+      "reference": "Can you throw an exception if width or height are both null"
+    },
+    {
+      "prediction": "I think these should be `httpSchemeHTTP` and `httpSchemeHTTPS` instead of `HTTPSchemeHTTP` and `HTTPSchemeHTTPS`.",
+      "reference": "This is a breaking change. I guess these might have been intentionally exported."
+    },
+    {
+      "prediction": "I think this should be `else if (options.containsKey(TableProperties.SPLIT_SIZE) || options.containsKey(TableProperties.METADATA_SPLIT_SIZE))`",
+      "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 this should be `*` instead of `\u00d7` and `//` instead of `\u00f7`",
+      "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 this should be `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(\"Unsupported encoding: \" + dataEncoding + \" for column \" + desc);`",
+      "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 this should be `positional=False, default=None, description=\"parameter name used to pass in parameterized value. Defaults to None, meaning use first positional parameter\"`",
+      "reference": "Maybe add `positional=False`. It does not need it more than the other parameters, but one must start somewhere."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the tests to pass.",
+      "reference": "Since we're here, can we use a template?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it works for now.",
+      "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 think this should be `Selector: \"default == ''\"` instead of `Tag: \"default\"`.",
+      "reference": "`has(default)` i think is the preferred way of doing this"
+    },
+    {
+      "prediction": "nitpick: I think it would be better to use the same variable names as in the docstring, i.e. `w`, `minval`, `maxval`, `n`, `radius`, `resolution`, `name`, `verbosity`, `clipInput`, `space`",
+      "reference": "always put a space after a comma"
+    },
+    {
+      "prediction": "I don't think this is the right place to call this method. It should be called in `selectAllPhotosUpToFav(int position)` method.",
+      "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 don't think we need to add `options` to the `__init__` method. We can add it to the `start_session` method instead.",
+      "reference": "@AutomatedTester @davehunt thoughts on a new keyword argument?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "This change shouldn't be necessary for this PR. Please remove it."
+    },
+    {
+      "prediction": "nit: `testCtx` -> `ctx`",
+      "reference": "This variable is mostly unused, how about we make `NewTestCtx` return a pointer?"
+    },
+    {
+      "prediction": "I think this should be 16 MiB. 32 MiB is a lot of data to send over the wire.",
+      "reference": "What is the basis of this value?"
+    },
+    {
+      "prediction": "This should be `#include \"gce_metadata/gce_metadata.h\"`.",
+      "reference": "Nit: was this extra blank line added intentionally?"
+    },
+    {
+      "prediction": "I think this should be `STATUS_GUARD_PAGE_VIOLATION`, not `EXCEPTION_GUARD_PAGE_VIOLATION`.",
+      "reference": "nit: inconsistent spacing around ="
+    },
+    {
+      "prediction": "I don't think we should remove the `LegacyExecutionContextInterface` import here.",
+      "reference": "We still need this import"
+    },
+    {
+      "prediction": "I think we can remove the `groupId` parameter from the `rewriteFiles` method in the `BinPackStrategy` class.",
+      "reference": "We are pulling this because we removed \"groupID\" state and put it into the strategy implementations"
+    },
+    {
+      "prediction": "I think it's better to use `oneapi::dal::kmeans_init::compute` instead of `oneapi::dal::kmeans_init::compute`",
+      "reference": "Is this file actually related to PCA?"
+    },
+    {
+      "prediction": "I think we should keep `isFirstAdmin` as well. It's used in `createEnableTracking` and `createTrackEvent`.",
+      "reference": "See above, this should probably remain `isFirstAdmin`."
+    },
+    {
+      "prediction": "I think this should be `Type` instead of `MonoType`.",
+      "reference": "Why the change to make it public? The expression interface has the `TypeOf` method?"
+    },
+    {
+      "prediction": "I think this should be `Single_Archive_File_Rolls_Correctly`",
+      "reference": "don't mind the snake cases in the test names. If there are in the test cases, it's fine IMO"
+    },
+    {
+      "prediction": "I think it would be better to have a separate DTO for unlocking, instead of reusing `identityRegistrationDto`. This way we can have different validation rules for registration and unlocking.",
+      "reference": "`Id` defines REST resource address and should not be in payload"
+    },
+    {
+      "prediction": "I think it's better to use `--class_wise` instead of `--class_wise`",
+      "reference": "We can omit the underscore and just use `classwise`."
+    },
+    {
+      "prediction": "I think you can remove the `!empty($rootCategory) ? $rootCategory->getId() : null;` and just use `$rootCategory->getId();`",
+      "reference": "Why not throw an exception instead then?"
+    },
+    {
+      "prediction": "I think we can remove this method now.",
+      "reference": "Looks like this doesn't need to change. Can you revert this?"
+    },
+    {
+      "prediction": "I don't think we need to import `template` here.",
+      "reference": "please import it like `from kinto.config import template`"
+    },
+    {
+      "prediction": "I don't think we should change the default value here.",
+      "reference": "There are lots of default value changed, not sure if it break the old behavior."
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I don't think we need to add `addToStaticNodes` to `RemovePeer` method.",
+      "reference": "parameter should be called removeFromStaticNodes"
+    },
+    {
+      "prediction": "I think this should be `override: Optional[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": "I think this should be `deactivated_at` instead of `deactivated_on` to be consistent with the rest of the codebase.",
+      "reference": "Is this method actually being used anywhere? If not, I think we should remove it."
+    },
+    {
+      "prediction": "I think this should be `!p.IsConstant && !p.IsReadOnly && !p.IsRequired`",
+      "reference": "Thanks! I somehow lost this between my far too many branchs."
+    },
+    {
+      "prediction": "I don't think we need to import `FirefoxWebElement` here. It's only used in the `find_element` and `find_elements` methods, which are defined in the `RemoteWebDriver` class.",
+      "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-amd64:3.0` or `gcr.io/google_containers/pause-amd64:3.1`",
+      "reference": "Can you add a TODO for loading the tarball of the pause image into Docker?"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "Is there any value in setting the process to null like this?"
+    },
+    {
+      "prediction": "`softmax_axis_ = this->layer_param_.softmax_param().axis(); outer_num_ = bottom[0]->count(0, softmax_axis_); inner_num_ = bottom[0]->count(softmax_axis_ + 1); CHECK_EQ(outer_num_ * inner_num_, bottom[1]->count()) << \"Number of labels must match number of predictions; \" << \"e.g., if softmax axis == 1 and prediction shape is (N, C, H, W), \" << \"label count (number of labels) must be N*H*W, \" << \"with integer values in {0, 1, ..., C-1}.\");`",
+      "reference": "This could be a good time to add a check that the prediction and target dimensions (except channel) agree?"
+    },
+    {
+      "prediction": "I think this should be `else if (instr_get_opcode(bb->instr) == OP_isb || instr_get_opcode(bb->instr) == OP_misb || instr_get_opcode(bb->instr) == OP_mrsb)`",
+      "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 don't think we need to replace {0} and {1} here. We can just do `Globalize.translate(\"ListPaging\", (totalRecordCount ? startIndex + 1 : 0) + \"-\" + recordsEnd, totalRecordCount);`",
+      "reference": "The translate library has a method to replace the variables."
+    },
+    {
+      "prediction": "Can you add a test for `inplace=True`?",
+      "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`",
+      "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": "Why do we need to flush here?",
+      "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'm not sure if this is the right place to do this, but I think it should be done in `getUrlFromRouteCache` instead.",
+      "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 don't think we need to do this. The `getEditedCell` method is only used in the `prepare` method, which is only called when `editorClass` is not null.",
+      "reference": "Maybe we can use `this.instance.getCell` with `topMost` flag to get `TD` element?"
+    },
+    {
+      "prediction": "Why do we need to log the error here?",
+      "reference": "Change this to Debug Level"
+    },
+    {
+      "prediction": "I think this should be `StackTraceUsage = _allLayouts.DefaultIfEmpty().Aggregate(StackTraceUsage.None, (seed,layout) => seed | (layout?.StackTraceUsage ?? StackTraceUsage.None));` instead of `StackTraceUsage |= _allLayouts.DefaultIfEmpty().Aggregate(StackTraceUsage.None, (seed,layout) => seed | (layout?.StackTraceUsage ?? StackTraceUsage.None));`",
+      "reference": "I refactored the Aggregate, those are hard to read"
+    },
+    {
+      "prediction": "I think we should keep this test, but change the iframe to a div with role=\"main\".",
+      "reference": "This is now done in the after method, so this test won't pass any more."
+    },
+    {
+      "prediction": "Trailing whitespace detected.",
+      "reference": "What was the point of creating `user_block` if you're then not going to use it ;-)"
+    },
+    {
+      "prediction": "I don't think we need to remove these imports.",
+      "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.value)[1:-1]`",
+      "reference": "Just `return tokens.string[1:-1]` is enough"
+    },
+    {
+      "prediction": "Style/SymbolArray: Use %i or %I for an array of symbols.",
+      "reference": "1 trailing blank lines detected."
+    },
+    {
+      "prediction": "I'm not sure why this was set to `Double.MAX_VALUE` in the first place, but I think it should be `Double.MAX_VALUE - 100` or something like that.",
+      "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 work. I'm not sure if there's a more idiomatic way to do this in Go.",
+      "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 move `Ftrl` to the top of the list.",
+      "reference": "Looks like this file has tabs instead of spaces."
+    },
+    {
+      "prediction": "I think this should be `if err := resource.Cleanup(); err != nil {`",
+      "reference": "Hmm should we clean up container resources before task resources here? or the order does not really matter here?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Please undo this :wink:"
+    },
+    {
+      "prediction": "Align the elements of a hash literal if they span more than one line.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "thanks for adding this missing association"
+    },
+    {
+      "prediction": "I think this should be `UseHttps` instead of `UseKestrelHttps` since it's not specific to Kestrel.",
+      "reference": "Should I remove this in favor of the HttpsConnectionFilterOptions overload?"
+    },
+    {
+      "prediction": "`max(index)` should be `MAX(index)`",
+      "reference": "This shode be nullable, there alway sholud be at least one status in system , if not exist that mean data problme"
+    },
+    {
+      "prediction": "This should be added in the `__construct` method of the `BackendController` class.",
+      "reference": "Spaces not tabs"
+    },
+    {
+      "prediction": "I think this should be `fs.readdirSync(path, {withFileTypes: true}).map(e => e.name);`",
+      "reference": "To get the files inside a directory, please use `readDirStats()`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these constants. I think they should be in the `container` package.",
+      "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": "this file is empty",
+      "reference": "File is empty, consider removing?"
+    },
+    {
+      "prediction": "I think this should be `return url is not None`",
+      "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 `L1ERC20GatewayAddress` and `L1ERC721GatewayAddress`",
+      "reference": "Was the addition of `GasPriceOracleAddress` here erroneous?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 best way to do this, but it's the only way I could think of to get the `SessionStore` and `BrowserStore` logs to be in the same order.",
+      "reference": "How was this causing the exception?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "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 think this should be `new MultiTermIntervalsQuery(field, this)`.",
+      "reference": "Isn't it better to stub AtomatonQuery and yield it here. IIRC it resolves simplest MTQ highlighting cases as nobrainer."
+    },
+    {
+      "prediction": "I'm not sure why we're multiplying by 12.0 here, but it seems like we should be using `price_in_dollars` instead of `price`.",
+      "reference": "Line is too long. [96/80]"
+    },
+    {
+      "prediction": "I think this should be `kubernetes.io/arch: amd64`",
+      "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": "I don't think we need to change this to `InnerObjectWithSelector`. We can just keep it as `InnerObject` and add the `selector` field to the `Object` struct.",
+      "reference": "Why do we need interface `InnerObjectWithSelector`, I searched the usage with this field, it seems nowhere use `GetSelectorSpecs()` methods provided by `InnerObjectWithSelector`"
+    },
+    {
+      "prediction": "nit: `InitContext(MsgCtxTypeChannel)` -> `InitContext(MsgCtxTypeChannel)`",
+      "reference": "Do we need this `init` here? We have already called the `InitContext` in `StartModule` directly."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think we should also add a `minimize_window` method to the `Remote` class in `webdriver/remote/remote_connection.py`.",
+      "reference": "This `if` is not necessary"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it seems to be the only place where we have access to `dom.value` and `newProps.value` at the same time.",
+      "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": "I think this should be `opts[:modulepath].join(File::PATH_SEPARATOR) if opts[:modulepath]`",
+      "reference": "Please update the yard docs to indicate this new option."
+    },
+    {
+      "prediction": "I think this should be `+kubebuilder:validation:MinItems=1`",
+      "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 we want to store the code version in the migration table or not. I think it's fine to store it, but I'm not sure if we want to store it in the same table as the schema version.",
+      "reference": "It would be great if this could be captured by a test."
+    },
+    {
+      "prediction": "10*time.Second),\n+\t\tESProcessorAckTimeout:    dc.GetDurationProperty(dynamicconfig.WorkerESProcessorAckTimeout, 1*time.Minute),\n \t\tESProcessorMaxGoroutines:          dc.GetIntProperty(dynamicconfig.WorkerESProcessorMaxGoroutines, 1000),\n \t\tESProcessorMaxQueueSize:           dc.GetIntProperty(dynamicconfig.WorkerESProcessorMaxQueueSize, 10000),\n \t\tESProcessorEnableMetrics:          dc.GetBoolProperty(dynamicconfig.WorkerESProcessorEnableMetrics, false),\n```\nReview comment?",
+      "reference": "1000 -> 100 maybe too much, 200 ish to 500ish maybe a good option"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "Any reason we need to keep this package's `ChecksumAddress` type or can we get rid of it too?"
+    },
+    {
+      "prediction": "I don't think this is the right translation. I think it should be `globalize.translate('ButtonRestartServer')`",
+      "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": "Pending hash is not the same as head hash. Pending hash is the hash of the best suggested block, while head hash is the hash of the last block in the chain.",
+      "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 don't think we need to preserve unknown fields here. We should be able to remove the `// +kubebuilder:pruning:PreserveUnknownFields` line.",
+      "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.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I don't think you need to include xkbcommon.h here.",
+      "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) !== '/') { $localPath = rtrim($localPath, '/') . '/'; }`",
+      "reference": "Add a space between if and opening parenthesis please (i.e. `if (`)"
+    },
+    {
+      "prediction": "I think it's better to use `if index is not None` instead of `if index`.",
+      "reference": "nit: `rename` instead of `alias`?"
+    },
+    {
+      "prediction": "I think this should be `if devName == \"/dev/sda1\" || devName == \"/dev/nvme0n1p1\"`",
+      "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 don't think this is correct. If `value_size` is `None` then `p.len - p.value_size - 6` will be `p.len - None - 6` which is `None - 6` which is `None`.",
+      "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 think this should be `type:\"integer\" minimum:\"0\" maximum:\"2147483647\"`",
+      "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 class.",
+      "reference": "Please follow the naming convention for variables, i.e. `windowText`."
+    },
+    {
+      "prediction": "This import is not used. Please remove it.",
+      "reference": "alias can be `cast`"
+    },
+    {
+      "prediction": "nitpick: I think it would be better to have `mmap_errors` as a `uint64_t` instead of a `size_t` so that we don't have to cast it to `uint64_t` in the format string.",
+      "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 think this should be `ERROR` instead of `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 `it->first` instead of `it->second`.",
+      "reference": "This should now be made into `if (it->second.empty())` ?"
+    },
+    {
+      "prediction": "I don't think you need the backticks around `QWebEnginePage.FindFlags` here.",
+      "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 think this should be `sc := trace.SpanContextFromContext(ctx)`",
+      "reference": "Should these assert that the extracted `SpanContext` is remote?"
+    },
+    {
+      "prediction": "`match_low_quality` -> `allow_low_quality_matches`",
+      "reference": "typo: allow low quality matches."
+    },
+    {
+      "prediction": "nit: I think we can remove the `cfg *cmds.Server` argument and just use `config.ControlConfig` instead.",
+      "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 think this should be `aria_label` instead of `aria-label`",
+      "reference": "Can we add the `if label` conditional to this please?"
+    },
+    {
+      "prediction": "I think this should be `mlflow>=0.2.0,<1.0.0`",
+      "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": "Nit: I think it would be better to call this `isFillDataChanged` or something similar. `hasFillDataChanged` sounds like it's a boolean, but it's not.",
+      "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": "nit: I think we can remove the `IntermediatesManager` import here and just use `FileSystemIntermediateManager` and `InMemoryIntermediatesManager` directly.",
+      "reference": "`Intermediates` or `Intermediate`"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. I think we should be raising an exception if the element is stale.",
+      "reference": "This is not the right exception class. There is an InvalidSelectorException class that covers bad locators."
+    },
+    {
+      "prediction": "Why is this line removed?",
+      "reference": "If this code isn't needed, it should be removed."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. If the connection is disconnected, we don't want to try to reconnect, we want to return an error and let the caller decide what to do.",
+      "reference": "Should the `doRequest` func be a method on the `connection` type instead of passing state from the type?"
+    },
+    {
+      "prediction": "I think this should be `COPILOT_ENVIRONMENT_NAME` and `COPILOT_LB_DNS` instead of `ECS_CLI_ENVIRONMENT_NAME` and `ECS_CLI_LB_DNS`",
+      "reference": "we'll probably tackle these e2e tests at the end"
+    },
+    {
+      "prediction": "Shouldn't this be `Dns.GetHostEntryAsync(\"34.206.253.53\").GetAwaiter().GetResult()`?",
+      "reference": "what is this address pointing to? what are we measuring here? I want to have a better understanding."
+    },
+    {
+      "prediction": "Shouldn't this be `dep_status = 3 or dep_status = 4` instead of `dep_status = 0 or dep_status = 3`?",
+      "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": "I think this should be `// +build !linux,!darwin,!freebsd,!netbsd,!openbsd,!plan9,!solaris`",
+      "reference": "Hah I had `gorename` failing without this too."
+    },
+    {
+      "prediction": "I think it would be better to use `string.Empty` instead of `\"(none)\"`",
+      "reference": "maybe prefix these with `appsec_` for consistency?"
+    },
+    {
+      "prediction": "I think we should remove this. We don't want to expose the pprof endpoint by default.",
+      "reference": "nit: this should be down further with the other github imports"
+    },
+    {
+      "prediction": "I think this should be moved to `__init__.py` so that it is done as soon as NVDA starts.",
+      "reference": "Why is this change necessary or related to the rest of the PR?"
+    },
+    {
+      "prediction": "I think this should be `'- see: ' . $issue_data->link` instead of `' - see: ' . $issue_data->link`",
+      "reference": "Hm, I wonder if the `see:` prefix is even necessary?"
+    },
+    {
+      "prediction": "Shouldn't this be `remainingMinutes = remainingMinutes || 0;`?",
+      "reference": "This should be translated."
+    },
+    {
+      "prediction": "I don't think we need to check `num_class` and `boosting` and `metric` here.",
+      "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": "I'm not sure about this change. I don't think we should be using a regex here, and I don't think we should be escaping the newlines in the string.",
+      "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 this should be `constants.HiveNamespace` instead of `controllerutils.GetHiveNamespace()`.",
+      "reference": "Shouldn't the global pull secret by in the namespace of the hive-operator? It is part of the operand."
+    },
+    {
+      "prediction": "I think we can remove these sleeps now that we're using `wait_until_stream_exists` and `wait_until_stream_has_1_shard`.",
+      "reference": "kinesis-mock applies KINESIS_LATENCY to RegisterStreamConsumer actions so I added a sleep here."
+    },
+    {
+      "prediction": "Why can't we upgrade from V1 to V3?",
+      "reference": "Only V1 or V3 is return from `getMetaVersionFromKV`. So where do we call `updateMetaV2ToV3`?"
+    },
+    {
+      "prediction": "I think this should be `return []catalog.Version{nodeAttestorV0{}, nodeAttestorV1{}}`",
+      "reference": "I think that V0 is missing here, which will prevent plugins that haven't been converted to work."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "Should we remove this?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Put on top of the import."
+    },
+    {
+      "prediction": "I think this should be \"Google Analytics\".",
+      "reference": "See above, this shouldn't be added."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `ITEM_AMULET_OF_FURY` and `ITEM_FURY_ORNAMENT_KIT`.",
+      "reference": "This'll likely get fixed upstream, we should let them deal with it."
+    },
+    {
+      "prediction": "I think this should be a `libp2p.Host` instead of a `host.Host`.",
+      "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 think it would be better to use `moment().startOf(timeDuration).subtract(n, timeDuration).valueOf()` instead of `moment().startOf(timeDuration).subtract(n, timeDuration).format('x')` because `valueOf()` returns the number of milliseconds since 1970-01-01T00:00:00.000Z and `format('x')` returns the number of milliseconds since 1970-01-01T00:00:00.000Z as a string.",
+      "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 this is the right place to set the language. It should be done in the `__init__` method of the app module.",
+      "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": "Why change the port number from 10255 to 10256?",
+      "reference": "why is this changed ?"
+    },
+    {
+      "prediction": "I don't think we need to test this.",
+      "reference": "Should this be \"the new revision value\"?"
+    },
+    {
+      "prediction": "`Dette feltet m\u00e5 aksepteres n\u00e5r :other er :value.` -> `Dette feltet m\u00e5 aksepteres n\u00e5r :other er lik :value.`",
+      "reference": "You have deleted :attribute"
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "The license header should be updated to the Apache one."
+    },
+    {
+      "prediction": "I think we can remove the `expectedOperationName` and `expectedServiceName` as they are not used.",
+      "reference": "@zacharycmontoya Is there any way to distinguish a `SocketHttpHandler` from another `HttpMessageHandler` request?"
+    },
+    {
+      "prediction": "I don't think we need to implement these.",
+      "reference": "Not sure about this one -- did you check somehow?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Is there a good reason to keep this at the `pe_ver_win` name now that it's per-host?"
+    },
+    {
+      "prediction": "I don't think you need to create a new instance here. You can just use `new SummaryProductFragment()`.",
+      "reference": "Not in the order of display : Front, Ingredient, Nutrition here (which is the right thing), Actually displayed: Front, Nutrition, Ingredients"
+    },
+    {
+      "prediction": "why remove this field?",
+      "reference": "Why delete `StartTime`? This `StartTime` was used to avoid the PID was reused."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `teardown` method of the `Host` class. The `teardown` method of the `Host` class is called by the `teardown` method of the `HostGroup` class. The `teardown` method of the `HostGroup` class is called by the `teardown` method of the `Beaker::Test` class. The `teardown` method of the `Beaker::Test` class is called by the `teardown` method of the `Beaker::TestCase` class. The `teardown` method of the `Beaker::TestCase` class is called by the `teardown` method of the `Beaker::TestSuite` class. The `teardown` method of the `Beaker::TestSuite` class is called by the `teardown` method of the `Beaker::TestRunner` class. The `teardown` method of the `Beaker::TestRunner` class is called by the `run` method of the `Beaker::TestRunner` class. The `run` method of the `Beaker::TestRunner` class is called by the `run`",
+      "reference": "My concern here, is that if the dump_puppet_log also throws then we will lose the data about the teardown_exception."
+    },
+    {
+      "prediction": "Why do we need to cast `size_t` to `int` here?",
+      "reference": "Should it be `int64_t`?"
+    },
+    {
+      "prediction": "I think this should be 1000 * 60 * 60 * 24 * 365 * 10",
+      "reference": "is this value related to something ?"
+    },
+    {
+      "prediction": "Why do we need to return `Keccak?` and `AddTxResult?`? Why not just `Keccak` and `AddTxResult`?",
+      "reference": "Add names to tuple elements ValueTask<(Keccak? Hash, AddTxResult? AddResult)>, should they both be nullable?"
+    },
+    {
+      "prediction": "I think it would be better to do this replacement in the constructor, so that we don't have to do it every time `toUnderscore` is called.",
+      "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 don't understand why this is needed.",
+      "reference": "Huh? The point is that a cast isn't necessary"
+    },
+    {
+      "prediction": "I think this should be `OPENEBS_IO_INSTALL_CRDS` to be consistent with `OPENEBS_IO_CREATE_DEFAULT_STORAGE_CONFIG` and `OPENEBS_IO_CREATE_DEFAULT_STORAGE_CLASS`.",
+      "reference": "can we name it like `InstallV1Alpha1CRDs` ?"
+    },
+    {
+      "prediction": "I think this should be `hive.openshift.io/hiveutil-created` instead of `hive.openshift.io/hiveutil-created`",
+      "reference": "I need something to match when creating selectorsyncsets so added this label."
+    },
+    {
+      "prediction": "I think we can remove the `# :nocov:` here as well.",
+      "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 this should be \"This code example demonstrates how to add a cross-origin resource sharing (CORS) configuration containing a single rule to an Amazon Simple Storage Service (Amazon S3) bucket.\"",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "Shouldn't this be `r.idx.Obsolete().Sub(validIndex)`?",
+      "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": "Line is too long. [81/80]",
+      "reference": "We might want to move this logic to a Plain Old Ruby Object down the road. Not a blocker."
+    },
+    {
+      "prediction": "I think we can remove the `if (this.toolbox_)` check here.",
+      "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": "I don't see how this can ever be false.",
+      "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": "why do we need this?",
+      "reference": "Let's not create new connection, but rather user `common.db` one"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Can this use asset_path rather than defining an image_url helper?"
+    },
+    {
+      "prediction": "I think it would be better to use `constant` instead of `always` here.",
+      "reference": "Hindley-Milner signatures don't use literal types. They explusively use types. It's deal with that"
+    },
+    {
+      "prediction": "I don't think we need to store the name and namespace of the network policy here. We can just store the UID of the network policy.",
+      "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 the comma.",
+      "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 I follow this. If `is_complete` is called before `get_steps_to_execute`, shouldn't `is_complete` return `False` because `get_steps_to_execute` hasn't been called yet?",
+      "reference": "Could also consider modifying the ActiveExecution..."
+    },
+    {
+      "prediction": "I think it would be better to raise a `TypeError` here, since this function is supposed to return an `astroid.Const`, not an `astroid.Const` or `None`.",
+      "reference": "Does it make sense to change that to a custom error instead of `ValueError`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the best we can do for now.",
+      "reference": "I guess this will need to change soon."
+    },
+    {
+      "prediction": "I'm not sure why this is here. I don't think it should be.",
+      "reference": "Restore this spacing - not sure why it has to be removed?"
+    },
+    {
+      "prediction": "I don't think we need to define `_XOPEN_SOURCE` if we're already defining `_POSIX_C_SOURCE`.",
+      "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": "`loopBody != getDst` -> `loopBody != dst`",
+      "reference": "Why not delete these codes?"
+    },
+    {
+      "prediction": "`FLAGS_meta_client_retry_times` is not used in `waitForMetadReady`.",
+      "reference": "If pull message failed after some reties, we should ensure the storaged exit, otherwise it has risk to loss data."
+    },
+    {
+      "prediction": "`first` and `count` are not used in this file.",
+      "reference": "looks like you don't need to import first and count here anymore"
+    },
+    {
+      "prediction": "I think we can just remove this method for now. We can add it back later if we need it.",
+      "reference": "FYI, visibility task processing is local only, meaning pushing something to local ES cluster / local archival endpoint"
+    },
+    {
+      "prediction": "This import should be removed.",
+      "reference": "This import can be removed."
+    },
+    {
+      "prediction": "I think this should be `strings.HasPrefix(a.Region, \"fips-\") || strings.HasSuffix(a.Region, \"-fips\") || strings.Contains(a.Region, \"-fips-\")`",
+      "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 we can remove this exception now.",
+      "reference": "is this phrase completed in a next line? Or forgot to add... what?"
+    },
+    {
+      "prediction": "Please remove the `total_epochs` and `workflow` in this file.",
+      "reference": "`('train', 5)` and `('train', 1)` are actually the same."
+    },
+    {
+      "prediction": "I think this should be `limit float64` instead of `limit int64`.",
+      "reference": "Would it be too complicated to refactor `currLimit` to be an `int64`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Would it be more performant to just `include ::RSpec::Core::Pending` here? /cc @myronmarston"
+    },
+    {
+      "prediction": "`+optional` is not needed here.",
+      "reference": "Should be podiochaos?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it seems to work.",
+      "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 think it would be better to throw an `ArgumentException` here. This will make it easier for users to understand why their instrument is not being exported.",
+      "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 think it would be better to use `this.driver_.getExecutor().w3c ? 'handle' : 'name'` instead of `this.driver_.getExecutor().w3c ? 'handle' : 'name'`.",
+      "reference": "I'd rather just send the parameter twice than break encapsulation here. There's already precedence with webelement IDs"
+    },
+    {
+      "prediction": "Please remove the `# NOQA` and `# NOQA`",
+      "reference": "You may put 'open-mmlab://res2net101_v1d_26w_4s' here. MMCV will be updated later."
+    },
+    {
+      "prediction": "nit: I think it would be better to have these constants in a separate file (e.g. `constants.js`) so that they can be used in other files as well (e.g. `pagespeed-insights.js` and `pagespeed-insights-reducer.js`).",
+      "reference": "May be worth making this more specific, to avoid potential conflicts, maybe `pagespeedDashboardWidget`? Or `webVitalsDashboardWidget`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get rid of the `Failed because:` part of the error message.",
+      "reference": "I guess this assignment is now redundant?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `Forward` method of each backend.",
+      "reference": "See above note re: blockNum synchronization."
+    },
+    {
+      "prediction": "nit: I think it's better to use `github.com/spiffe/spire/pkg/agent/common/backoff` instead of `github.com/spiffe/spire/pkg/agent/common/backoff/backoff`",
+      "reference": "wonder if we could avoid doing named imports if we type aliased `backoff.Backoff` in the `.../agent/common/backoff` package?"
+    },
+    {
+      "prediction": "nit: please remove this empty line",
+      "reference": "Or maybe put into alphabetical order with the other #includes."
+    },
+    {
+      "prediction": "Why not use `time.Sleep` here?",
+      "reference": "This runs the entire duration of the program; this change is not needed."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "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 don't think this is a clang-format bug. The `clang-format on` line is indented by 4 spaces, and the `clang-format off` line is indented by 8 spaces. That's a difference of 4 spaces, not a tab.",
+      "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 we should be importing `org.openqa.selenium.remote.tracing.*` here.",
+      "reference": "Could you please leave the explicit imports?"
+    },
+    {
+      "prediction": "Why do we need this? If the method config is not in the map, it means that the method does not exist in the interface.",
+      "reference": "Can we check for containsKey here?"
+    },
+    {
+      "prediction": "I think this should be `os.Args[1] == \"help\" || os.Args[1] == \"-h\" || os.Args[1] == \"-help\" || os.Args[1] == \"--help\"`",
+      "reference": "Exit successfully. We served the help the user wanted."
+    },
+    {
+      "prediction": "I think this should be `InitialWorkingDirPath` to be consistent with `InitialPackagePath`.",
+      "reference": "Had to rename this because `InitialPackage()` already existed."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Can we put this in a different helper?"
+    },
+    {
+      "prediction": "I don't think we need to do this. We should be able to just use `dstack_offs += XSP_SZ * 2;` instead of `dstack_offs += 15 * XSP_SZ;` and `dstack_offs += XSP_SZ;` at the end.",
+      "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 think this should be `@Path(\"/apiTermsOfUse\")` and `@Path(\"/api/termsOfUse\")`",
+      "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": "Why is this needed?",
+      "reference": "Why do we need to override this if all we're doing is passing it up?"
+    },
+    {
+      "prediction": "I don't think this is the right place to send the controls. The controls should be sent after the table is created and before any data is inserted into the table.",
+      "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 = packaging.version.parse(\"0.12.4\")` so that we can use `DAGSTER_EARLIEST_RELEASE > packaging.version.parse(\"0.12.4\")` later on.",
+      "reference": "EARLIEST_TESTED_RELEASE might be slightly more clear"
+    },
+    {
+      "prediction": "I don't think this is correct. The `.*` at the start of the regex will match the entire line, and the `.*` at the end of the regex will match the entire line again. I think you want `.*#.*pylint:\\s*([^;#]+)[;#]?.*` instead.",
+      "reference": "I think replacing `.*` with `\\s*` from after the hash makes sense, but not the rest of the changes."
+    },
+    {
+      "prediction": "I think this should be `omitempty` as well.",
+      "reference": "interesting .. did not know varint tag covered bool as well"
+    },
+    {
+      "prediction": "nit: I think it would be better to use `uint64_t` instead of `int64_t` here.",
+      "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 don't think we need to change this method. It's only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only used in `getTransactionHashes` which is only",
+      "reference": "getHash should be a public and without an `_` if used here."
+    },
+    {
+      "prediction": "I don't think this is a good idea. The genesis block is a special case and should be handled separately.",
+      "reference": "Unrelated to these changes -- we will probably want some special handling for genesis block validation."
+    },
+    {
+      "prediction": "nit: I think it would be better to call this `DiskBlockCacheFractionSetter` to make it more obvious that it's a setter.",
+      "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 add the peer to the `receivedPeers` map. I think it should be done in the `AddPeer` method of the `p2pService`.",
+      "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 do this, but it seems to work for now.",
+      "reference": "Minor: can we put the conditional outside of the partial?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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 task which can be programmatically run, either on a schedule or via service code in response to events.`",
+      "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 revert this change.",
+      "reference": "Please move unrelated formatting changes to their own PR (same with the resources.py file)."
+    },
+    {
+      "prediction": "I think this should be `window._state_before_fullscreen = window.windowState() & ~Qt.WindowFullScreen` instead of `window._state_before_fullscreen = window.windowState() & ~Qt.WindowFullScreen`",
+      "reference": "Make this a public attribute (remove the leading `_`) if you want to access it from the outside."
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "this conditional is useless now so it should be removed"
+    },
+    {
+      "prediction": "I'm not sure why this is needed, but it seems like a good idea to be explicit about the type of `i` and `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 think this should be `if (configProto == null || configProto.getEnableProtoAnnotations() == null) {`",
+      "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'm not sure if this is the right thing to do, but it seems to be the only way to get the tests to pass.",
+      "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 right thing to do, but I think we want to use `data-turbolinks=\"false\"` here so that the skip link doesn't trigger a turbolink page load.",
+      "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 think this should be `if (result == VK_SUCCESS && pProperties) {`",
+      "reference": "This was already in vkGetPhysicalDeviceDisplayProperties2KHR..."
+    },
+    {
+      "prediction": "nit: I think this should be `DockerEngineValidator` to be consistent with the rest of the codebase.",
+      "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 think we can keep this at 512. 700 seems like a lot for a type name.",
+      "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 think this should be `GetBufferIdentifier` instead of `BufferHasIdentifier`.",
+      "reference": "When calling functions with keyword arguments: please provide the arguments as keyword arguments, not positional arguments."
+    },
+    {
+      "prediction": "Why not use `act_cfg` here?",
+      "reference": "We may follow the argument order: conv_cfg, norm_cfg, act_cfg."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "`options['binary'] = WebDriver::Chrome.path` if set?"
+    },
+    {
+      "prediction": "I think this should be `#include \"folly/ssl/Init.h\"`",
+      "reference": "folly library should be put on the top."
+    },
+    {
+      "prediction": "I don't think we need the `isSvg` check here. `ns` will always be `false` if `isSvg` is `false`.",
+      "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 it would be better to use `row` instead of `this.hot.toPhysicalRow(row)` here.",
+      "reference": "Please add it to the `REMOVED_HOOKS` constant."
+    },
+    {
+      "prediction": "I think we can remove this line, it's not being used.",
+      "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 think it would be better to have this method in the `Account::Access` class.",
+      "reference": "Is this access protected so that only admins can access this route?"
+    },
+    {
+      "prediction": "I think this should be `FLT_MAX` instead of `DBL_MAX`.",
+      "reference": "What happens if the minimum x coord is > 0? This will incorrectly set it as zero, right?"
+    },
+    {
+      "prediction": "Shouldn't this be `o.maxAckPending`?",
+      "reference": "I think you want to look at current pending + batchSize and if that is > o.maxp no?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "Should this be `codes.Error`? `codes.Internal` was an error status before."
+    },
+    {
+      "prediction": "Align the parameters of a method call if they span more than one line.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "I don't think you need these lines."
+    },
+    {
+      "prediction": "I think this should be `DiskLimiter` instead of `DiskLimiter`.",
+      "reference": "You didn't mean to export this, right?"
+    },
+    {
+      "prediction": "Nitpick: I think this should be \"the AWS SDK for Ruby (v3)\".",
+      "reference": "Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "I think this should be `egressConfig *config.EgressConfig` instead of `egressConfig *config.EgressConfig`.",
+      "reference": "Do we still need this?"
+    },
+    {
+      "prediction": "Is this debug code? If so, please remove it.",
+      "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 don't think this is the right translation key. I think it should be something like `MessageImageEditorLoadError`.",
+      "reference": "This is a bit inconsistent with the `ErrorStartHourGreaterThanEnd` in the previous file. Wouldn't `ErrorDefault``fit better here?"
+    },
+    {
+      "prediction": "I don't think we need to skip this test for IPv6 cluster.",
+      "reference": "I think it will revert the change @antoninbas did on purpose 05eee251c9c53400277def576f92d614ca234898"
+    },
+    {
+      "prediction": "I think this should be `if ifaceName == \"wireguard.cali\" || ifaceName == \"wg-quick.cali\"`",
+      "reference": "nit: a switch perhaps?"
+    },
+    {
+      "prediction": "Line is too long. [117/80]",
+      "reference": "Would it make sense for `opensearch_description_tag` to have default values for the title and href attributes?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to return an empty list of ImmutableRootMetadata if the TLF journal is disabled.",
+      "reference": "Just noticed this while debugging -- it's unlikely to be causing issues in practice."
+    },
+    {
+      "prediction": "I'm not sure why this needs to be longer than the default 500ms.",
+      "reference": "That seems unrelated - did that fix a random test failure you noticed while working on this issue?"
+    },
+    {
+      "prediction": "nit: I think we can simplify this to `metadata = base != null && base.file() != null ? metadata.addPreviousMetadata(base.file().location(), base.lastUpdatedMillis()) : metadata;`",
+      "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_tensor_list = [mlvl_tensors[i][batch_id].detach() if detach else mlvl_tensors[i][batch_id] for i in range(num_levels)]`",
+      "reference": "We need to know why all the tensors are detached in `get_bboxed`."
+    },
+    {
+      "prediction": "I think this should be `type(self).__a = ''`",
+      "reference": "Unrelated but shouldn't this raise `attribute-defined-outside-init`?"
+    },
+    {
+      "prediction": "I think this should be `create_no_window=True` by default.",
+      "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": "Metrics/LineLength: Line is too long. [123/120]",
+      "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 is the right thing to do, but I added a `clear-fix` class to the submit button and the cancel button so that they don't float next to each other on the page.",
+      "reference": "do you mean 'clearfix' class?"
+    },
+    {
+      "prediction": "I don't think we need to mention causal consistency here.",
+      "reference": "`causalConsistency` is already defined on `SessionOptions`. We do not need to duplicate it here."
+    },
+    {
+      "prediction": "I don't think we should use `static import` for `any`, `anyString` and `eq`",
+      "reference": "This should not be included since there are no code changes."
+    },
+    {
+      "prediction": "Shouldn't we also set `Access-Control-Allow-Credentials` to `true`?",
+      "reference": "If all we need just response 200, then we could use `c.Status(http.StatusOK)` here."
+    },
+    {
+      "prediction": "I think we can remove the `if _, ok := err.(errors2.ErrorResourceDoesNotExist); err != nil && !ok {` check and just do `if err != nil { return }`",
+      "reference": "Should be checking for error here"
+    },
+    {
+      "prediction": "I think this should be `FormTabs::SECTION_OUTSIDE, $this->config`",
+      "reference": "Put a space between `(array)` and `$this` and I'll merge it"
+    },
+    {
+      "prediction": "I think this should return `runtime.Stack(buf, all)`.",
+      "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": "This class is not needed. The `notificationId` and `content` can be accessed directly from the `NotificationContent` object.",
+      "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 should be `import java.util.*;` instead of `import java.util.*;`",
+      "reference": "Avoid wildcard imports"
+    },
+    {
+      "prediction": "I think this should be `CandidatesByHeight(height uint64) ([]*Candidate, error)` instead of `CandidatesByHeight(uint64) ([]*Candidate, error)`",
+      "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": "Shouldn't this be `Object.keys(items).map(k => k.toLowerCase()).indexOf('loadbalanced') !== -1`?",
+      "reference": "same question here as I have on the other file regarding doing this check here vs L112 vs not at all"
+    },
+    {
+      "prediction": "I think this should be `!linux` instead of `!openbsd`.",
+      "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 don't think we need this anymore.",
+      "reference": "this is used as important way of mocking for unit tests, FYI"
+    },
+    {
+      "prediction": "I think this should be `muted_alpha=0.5`",
+      "reference": "Sorry, should have been clearer, I meant adding this to the ``Histogram`` options one line above."
+    },
+    {
+      "prediction": "I'm not sure why this change is necessary, but I don't think it's related to this PR, so I'm okay with it.",
+      "reference": "Just ignorance on my part, but what is the significance of the change from C23NHe2 to C23NHx2?"
+    },
+    {
+      "prediction": "I think this should be `param.Array` instead of `param.ClassSelector`.",
+      "reference": "What about tuples or pandas series? Do we want to support lots of different types or force a single type?"
+    },
+    {
+      "prediction": "I don't think we should be throwing an exception here. I think we should just log a warning and return an empty map.",
+      "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": "Line is too long. [101/80]",
+      "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'm not sure if this is the right thing to do, but I don't think we want to warn about this anymore.",
+      "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 think this should be `self.filter_options(unfiltered) must be implemented by the transport class` instead of `self.filter_options(unfiltered) or self.filter_options(unfiltered) must be implemented by the transport class`",
+      "reference": "This is kind of confusing with the other `filter_options`."
+    },
+    {
+      "prediction": "I think this should be `if err == nil || _, ok := err.(OpsCantHandleFavorite) {`",
+      "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 is this changed to Int64?",
+      "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": "I think this should be `if status == 204 || status == 205 {`",
+      "reference": "Hmm, probably we should instead write whatever `status` is returned instead of coding a special case."
+    },
+    {
+      "prediction": "nit: I think it would be better to use `in.HTTPMatch.Methods == nil` instead of `in.HTTPMatch != nil`.",
+      "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": "Line is too long. [101/80]",
+      "reference": "why this recursive relationship definition?"
+    },
+    {
+      "prediction": "I'm not sure I understand the difference between `Encoding` and `Signature` here.",
+      "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 think it would be good to mention that these metrics are only collected if the `trace.Exporter` and `stats.Exporter` are set.",
+      "reference": "Not part of this PR, but should the section above say what the name of the metric is for each method?"
+    },
+    {
+      "prediction": "I think this should be `valueDisplay = (metric as IGaugeMetric).LastValue.Value.ToString(CultureInfo.InvariantCulture);`",
+      "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 thing to do. If we are in the background process group, we should not be reading from the terminal at all.",
+      "reference": "This is equivalent to stdout. Why not just replace stdin with stdout in `IsProcessBackground`?"
+    },
+    {
+      "prediction": "Shouldn't this be `if (p_jmsg[\"compress_method\"] != nullptr && p_jmsg[\"compress_method\"] == \"zfp\")`?",
+      "reference": "Rather than test for nullptr, just treat the pointer as a bool, i.e. `if(p_jmsg[\"compress_method\"])`"
+    },
+    {
+      "prediction": "I don't think this is the right place to show the error dialog. I think it should be done in the `createShortcut` method of the `LibraryFeaturePanel` class.",
+      "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 listen to every resources. I think we only need to listen to `bucket` and `collection` resources.",
+      "reference": "nitpick: _and collections_"
+    },
+    {
+      "prediction": "nitpick: I think it would be better to use `unsigned int` instead of `unsigned`.",
+      "reference": "I believe these need to be explicitly initialized in `proxy_send_prepare`"
+    },
+    {
+      "prediction": "I think this should be `t0 = time.time()` instead of `t0 = time.time() # Failed task start time`",
+      "reference": "This is unnecessary. The declaration of `t0` on line 179 is still in scope inside the `except` block."
+    },
+    {
+      "prediction": "I don't think we need to mention that only solid colors are supported.",
+      "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": "Do we need this? I don't think we do.",
+      "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 `if (isEmptyLine(path)) { continue; }`",
+      "reference": "FYI: inlined this method in the internal review."
+    },
+    {
+      "prediction": "I don't think this import is needed anymore.",
+      "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": "I'm not sure this is the best way to do this. I think it would be better to have a method in the `ILS` class that checks if the auth method is `Shibboleth` and if the `auth_based_library_cards` config is enabled.",
+      "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 && num_segments > 0)`",
+      "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": "Line is too long. [101/80]",
+      "reference": "I'm surprised rubocop isn't picking up singe quotes?"
+    },
+    {
+      "prediction": "I think we can remove this now that we don't use `parseTime` anymore.",
+      "reference": "Is this line still needed?"
+    },
+    {
+      "prediction": "I don't think we should change the default heartbeat period for the SEDP endpoints.",
+      "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 `brailleTables` or `braille_tables` to be consistent with the rest of the codebase.",
+      "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": "I don't think we need this anymore.",
+      "reference": "Why are we removing the option to skip machine pool generation?"
+    },
+    {
+      "prediction": "Should this be `IHttpStartLineHandler<T>` where `T` is the type of the `HttpStartLineContext`?",
+      "reference": "\"Request line\" here too."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have the zone controller check if the loot is allowed, and if it isn't, send the message to the client.",
+      "reference": "Please remove this."
+    },
+    {
+      "prediction": "I'm not sure why we need this.",
+      "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. [101/80]",
+      "reference": "This can be simplified as discussed before."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `execute` package, but I'm not sure if it should be in the `execute/table` package or the `execute` package.",
+      "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": "Why is this change necessary?",
+      "reference": "does this boolean indicate unsafe serialization?"
+    },
+    {
+      "prediction": "I don't think we need to throw an error here. We can just return an empty array.",
+      "reference": "Maybe rewrite to `Is only implemented for React Native`? (I assume that is the case)."
+    },
+    {
+      "prediction": "I think we can remove the `if (language == TargetLanguage.PHP)` check here, since the `resourcesField` will always be non-null in that case.",
+      "reference": "Do we not need this in Java because Java *is* handling map responses?"
+    },
+    {
+      "prediction": "I think this should be `True` by default.",
+      "reference": "Maybe \"For configuring which scheduler messages can be received.\"?"
+    },
+    {
+      "prediction": "`$this->getMediaManager()->getEntityManager()->clear();` should be `$this->getMediaManager()->getObjectManager()->clear();`",
+      "reference": "It will stop working with orm, so this is not a good fix, Try using some method in a common interface"
+    },
+    {
+      "prediction": "I think we should have a single `Azure.Sdk` listener that listens to both `Azure.Clients` and `Azure.Pipeline`.",
+      "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": "I don't think we need to set it to null here. It will be set to `null` in the constructor of `ParameterGenerator`.",
+      "reference": "This code is a bit confusing. Should HttpParameterType set to the original one or it is always BODY?"
+    },
+    {
+      "prediction": "I think this should be `!isTest && c.state.Config.Remote.VerifyOutputs` since we don't want to verify outputs for tests.",
+      "reference": "Why did this change?"
+    },
+    {
+      "prediction": "I think this should be `newProtoForTableStatus(result.Table.TableStatus)`",
+      "reference": "nit: `status` collides with imported package named `status`"
+    },
+    {
+      "prediction": "nit: `IsWritingCatchpointFile returns true when a catchpoint file is being generated. The function is used by the catchup service so that the memory pressure could be decreased until the catchpoint file writing is complete.`",
+      "reference": "\"memory pressure could be decreased\" --> \"to avoid memory pressure\""
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way to create a user that doesn't already exist in the database.",
+      "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'm not sure if this is the right place to do this, but it seems like a good place to me. We want to make sure that the collapsible state is set to `false` before we merge the new state with the existing state.",
+      "reference": "I'm thinking about covering this change with the test. Can you do that?"
+    },
+    {
+      "prediction": "I don't think we need to change the signature of this method. I think we can keep it as it is and just add a new method `PingProviderWithPorts` with the new signature. This way we don't have to change the implementation of `NATProviderPinger` in `nat_pinger.go`.",
+      "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 we can remove the `sdkAuthConfig` variable and just use `authConfig` directly.",
+      "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": "Why is this a task item array and not just a string array?",
+      "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 best way to do this, but it works for now.",
+      "reference": "Can you leave a note about why we are skipping regex tests?"
+    },
+    {
+      "prediction": "I think this should be `repo.OpenRepo(repoDir, newConfig)` instead of `repo.OpenFSRepo(repoDir)`.",
+      "reference": "This was the only caller of CreateRepo, so I inlined it."
+    },
+    {
+      "prediction": "I think this should be `AuthMountPath` to be consistent with the other fields in this struct.",
+      "reference": "As this is already a child of `auth`, I think we can shorten this to `path`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the first visible range.",
+      "reference": "This call should use element, not 0 I think."
+    },
+    {
+      "prediction": "I think it would be better to call this `PartialPageFetchThresholdBytes` to make it clear that it's in bytes.",
+      "reference": "This is missing the definition of `fetch_until_empty_page` - it needs to go in the config section above."
+    },
+    {
+      "prediction": "`File.expand_path?(path)` should be `File.expand_path(path)`",
+      "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 `table.locationProvider()` here and just pass `null` to the `newStaticTable` method.",
+      "reference": "Since we kept the original constructor we might not need this change now."
+    },
+    {
+      "prediction": "I think this should be `LOG.debug(\"Refreshed table [{}] from metadata location [{}]\", fullName, metadataLocation);`",
+      "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 package to `pkg/filestore/minio` instead of `pkg/filestore/minio`.",
+      "reference": "This should be in the last group."
+    },
+    {
+      "prediction": "I don't think this is the right place to send the statsd. I think it should be in the `SendTracesAsync` method, right after the `await _tracesSender.SendAsync(traces);` line.",
+      "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 `certmanager.k8s.io/v1alpha1`",
+      "reference": "I've gone with this group name, however I'm open to suggestions on alternatives!"
+    },
+    {
+      "prediction": "I think this should be `public static final RpcApi QUORUM_RPC = new RpcApi(\"QUORUM_RPC\");`",
+      "reference": "should this be GOQUORUM"
+    },
+    {
+      "prediction": "Why do we need to pass in `checkCoinbase` here?",
+      "reference": "I think checkCoinbase is not skipping checking coinbase. Instead, true-> checking num(coinbase tx) = 1, false -> checking num(coinbase tx) = 0."
+    },
+    {
+      "prediction": "`protocol.StateManager` should be `factory.WorkingSet`",
+      "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": "Metrics/LineLength: Line is too long. [123/120]",
+      "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 it's better to use `context.Context` instead of `golang.org/x/net/context.Context`.",
+      "reference": "why not context from stdlib?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I don't know what else to do.",
+      "reference": "Just saw this and wasn't able to hold myself from renaming :-)."
+    },
+    {
+      "prediction": "`Object.keys(core.plugins)[0]` should be `Object.keys(core.plugins)[0]` or `Object.keys(core.plugins)[0]`",
+      "reference": "Can we keep the other assertion too? I think it's helpful to ensure that the uninstall function was called too"
+    },
+    {
+      "prediction": "I think we should use `Kernel.warn` instead of `RSpec.warn_with` here.",
+      "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": "I think this should be `MessageTimeout` to be consistent with the rest of the codebase.",
+      "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 don't think this is correct. If the bucket doesn't exist, we should return a 404.",
+      "reference": "Can we remove the `TODO` statement here? (as this is actually fixed in this PR)"
+    },
+    {
+      "prediction": "I think this should be `private DisplayPreference mDisplayPreference = DisplayPreference.DEFAULT;`",
+      "reference": "m prefix no longer in code style"
+    },
+    {
+      "prediction": "I think this should be `return kvb.store.Filter(ns, c, kvb.buffer)`",
+      "reference": "need to filter the entities in buffer as well"
+    },
+    {
+      "prediction": "This should be `ID = \"de.danoeh.antennapod.core.glide.FastBlurTransformation\";`",
+      "reference": "Please add spaces before and after the equals sign"
+    },
+    {
+      "prediction": "Why is this line commented out?",
+      "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 about this change. I think it's better to keep the output the same as it was before.",
+      "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 don't think we need to add this to the `__init__` method. We can just add it to the `start` method.",
+      "reference": "We shouldn't add this as a new `kwarg` here. This should all be done on the options class"
+    },
+    {
+      "prediction": "Should this be `ENV['OS_VOL_SUPPORT'] || 'true'`?",
+      "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 = NULL;`",
+      "reference": "In C, file scope variables are already initialized to 0 so the \"= NULL\" is redundant. Not a big deal."
+    },
+    {
+      "prediction": "`getRootNode` returns `null` if `node` is not attached to the DOM. This will cause `doc` to be `null` in this case, which will cause a `TypeError` on the next line.",
+      "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 is a bug. It should be `nameof(value)`, not `nameof(milliseconds)`. `nameof(milliseconds)` is the name of the parameter, not the name of the value being validated.",
+      "reference": "nit: just to be similar to others, can you change to interpolation?"
+    },
+    {
+      "prediction": "I think this should be `IncludeRootCert` or something like that. `FullCertChain` is a bit ambiguous.",
+      "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": "Is this used anywhere? If not, can we remove it?",
+      "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 think this is the only place where we need to handle the case where the resource is not found. All the other places where we call `Get` should return an error if the resource is not found, so we don't need to handle that case here.",
+      "reference": "expected statement, found 'else' (and 1 more errors)"
+    },
+    {
+      "prediction": "I think this should be called `FakeIDRandomOrPanic` or something like that.",
+      "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 don't think we need to have these constants in a separate file. We can just have them in the `controller.go` file.",
+      "reference": "What's the difference from ModuleName in `pkg/apis/meta/v1alpha1/types.go`?"
+    },
+    {
+      "prediction": "I think we can remove these two using statements now.",
+      "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'm not sure this is the right thing to do here. I think we want to be able to skip XSP and XBP even if we don't skip any other registers. I'm not sure what happens if we skip XSP and XBP but not any other registers, but I think we should be able to do that.",
+      "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": "nit: I think we can remove this empty line.",
+      "reference": "Should the API have a dependency on the SDK?"
+    },
+    {
+      "prediction": "Why is this skipped?",
+      "reference": "Is this a sometimes leaky test?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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) or not locale.IsAvailable(wxLang.Language + '_' + wxLang.Country):`",
+      "reference": "It might make sense to log this."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Line is too long. [84/80]"
+    },
+    {
+      "prediction": "I think this should be `goog:chromeOptions` instead of `goog:chromeoptions`.",
+      "reference": "nice touch since Google likes to change things every now and then."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's better than having a bunch of `if`/`elsif`/`else` statements.",
+      "reference": "Thank you this is so much nicer than the old `Rails.configuration.branding['blah']['blah']['blah']` :)"
+    },
+    {
+      "prediction": "I think this should be `BACKEND_SERVICES` (plural).",
+      "reference": "nit: Would you mind alpha-sorting this?"
+    },
+    {
+      "prediction": "I think this should be `ManifestContent.PARTITION` instead of `ManifestContent.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": "Shouldn't this be `Rails.configuration.branding[:organisation][:contact_us_url] || contact_us`?",
+      "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": "I think this should be `Erstellen und weiteres Element hinzuf\u00fcgen`",
+      "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 `List<GrpcStreamingDetailView> grpcStreamingDescriptors();` instead of `List<GrpcStreamingDetailView> grpcStreamingDescriptors();`",
+      "reference": "Is PHP the first dynamic MVVM language with grpc streaming support?"
+    },
+    {
+      "prediction": "I don't think this is necessary. If `tmpStream` is `nullptr`, then `tmpStream->bad()` will throw an exception.",
+      "reference": "the `if` isn't necessary here; it's safe to call `delete` on `nullptr`"
+    },
+    {
+      "prediction": "I think this should return a `ReturnCode_t` instead of a `bool`.",
+      "reference": "Change if the method's name finally changes to `clear_history` as proposed elsewhere"
+    },
+    {
+      "prediction": "I think this should be `import docUtils` and `helpFile = docUtils.getDocFilePath(\"userGuide.html\")`",
+      "reference": "can this one also be moved to the top of the file?"
+    },
+    {
+      "prediction": "I think we should call this `Resources` instead of `GroupMembers` to avoid confusion with the `GroupMember` type.",
+      "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 don't think we need to change the names of these constants.",
+      "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', 'bigquery-credentials.json')`",
+      "reference": "will this fail if the env variable doesn't exist? Is this the preferred behaviour?"
+    },
+    {
+      "prediction": "I think this should be `self._random.getReal64()` instead of `0.01 * self._random.getReal64()`.",
+      "reference": "bad line wrap"
+    },
+    {
+      "prediction": "This should be `sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (bulk_pull_blocks_mode) + sizeof (uint32_t) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint256_union) + sizeof (rai::uint",
+      "reference": "What does the `8` here represent? Can you give it a name?"
+    },
+    {
+      "prediction": "I think this should be `os.environ.get(\"MP_API_KEY\")` so that it doesn't raise an error if the environment variable is not set.",
+      "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. [91/80]",
+      "reference": "Looks like there may be more changes to this file than intended?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. I think we need to make sure that `accessInfo.GetContent()` returns a `map[string]interface{}` and not a `map[interface{}]interface{}`.",
+      "reference": "when message type is not expected, contine to next loop instead of return to finish infinite loop"
+    },
+    {
+      "prediction": "nit: extra blank line",
+      "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 don't think we should allow changes to `instanceID`. I think we should only allow changes to `providerID` and `additionalTags`. `instanceID` should be immutable.",
+      "reference": "instanceID should stay the same for a AWSMachine?"
+    },
+    {
+      "prediction": "I think this should be `NO_DOWNLOAD=\"true\"`",
+      "reference": "Since this is a dict don't you want `False` or do we clean that up elsewhere?"
+    },
+    {
+      "prediction": "I think this should be `127.0.0.1` instead of `localhost`.",
+      "reference": "what's the reason you had to add the `host` argument to all `get_work` calls?"
+    },
+    {
+      "prediction": "I don't think this will ever be an instance of `Dataset`, will it?",
+      "reference": "This will never be a dataset so we can simplify the code here"
+    },
+    {
+      "prediction": "I think it would be better to have a `copy()` method on `AbstractBlockProcessor` that returns a new instance of `AbstractBlockProcessor` with the same parameters.",
+      "reference": "This should be moved into the PrivacyBlockProcessor instead of adding the constructor here."
+    },
+    {
+      "prediction": "I think we should rename this package to `insightstore` instead of `insightfilestore`.",
+      "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 think this should be `@round_count = 3` instead of `@round_count == 3`.",
+      "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 don't think this is necessary. The `throttle_set` is called in the `__init__` of the `Percentage` class, so it should already be set to `-1` at this point.",
+      "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 don't think we need to change this to a string.",
+      "reference": "why remove `Symbol`?"
+    },
+    {
+      "prediction": "Shouldn't this be `\\u00A3` instead of `\u00a3`?",
+      "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 don't think it's a good idea to add maintainers here.",
+      "reference": "You have forgotten a coma `,`"
+    },
+    {
+      "prediction": "`osp.join(out_dir, osp.basename(img_meta['filename']))` -> `osp.join(out_dir, img_meta['filename'])`",
+      "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'm not sure I understand the logic here. If `c1 == c2` and `bb.get(ia) == '\\0'` then `c1 - c2 == 0` and `bb.get(ia) - '\\0' == 0` so `cmp == 0` and `cmp != 0 || bb.get(ia) == '\\0'` should be `0 || 0` which is `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 don't think we need to make this configurable. I think it should always be `true`.",
+      "reference": "nit: can this be a primitive value?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just return `http.StatusNotFound` if the path doesn't end in `/`.",
+      "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'm not sure about this change. I think it's better to keep the order of the modules as they are defined in the config file.",
+      "reference": "Why change this?"
+    },
+    {
+      "prediction": "`ErrorListener` is deprecated in Symfony 4.4 and removed in Symfony 5.0. We should use `ExceptionListener` instead.",
+      "reference": "_nitpick_ Isn't it now actually NotLogFakeHttpExceptions**Error**Listener ?"
+    },
+    {
+      "prediction": "I think this should be `resourceDefinition.ResourceType = resourceType;` instead of `resourceDefinition.AddProperty(\"type\", JsonSchema.CreateStringEnum(resourceType), true);`",
+      "reference": "Why not just look for the \"type\" property instead of creating a ResourceType property?"
+    },
+    {
+      "prediction": "I don't think we need to pass the chain ID here.",
+      "reference": "should use `bc.config.Blockchain.ChainID`"
+    },
+    {
+      "prediction": "I don't think we need to change the `@see` here.",
+      "reference": "this doesn't need to know about NativeClient"
+    },
+    {
+      "prediction": "I don't think we need the backticks here.",
+      "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 / surface->current.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 escape these.",
+      "reference": "Why are you using these tags to carry these msids?"
+    },
+    {
+      "prediction": "I don't think we need the +1 here. `strncpy` will null-terminate the string for us.",
+      "reference": "flb_strdup() should do the work here. note: use Fluent Bit memory wrappers"
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "This line should be separated from the package declaration by one line as it is not the package description."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "shadow: declaration of \"err\" shadows declaration at line 410 (from `govet`)"
+    },
+    {
+      "prediction": "I think it's better to use `github.com/sonm-io/core/proto` instead of `github.com/sonm-io/core/proto`",
+      "reference": "Maybe next time better to make separate PR?"
+    },
+    {
+      "prediction": "I don't think we need to raise an error here. We can just return `segm_results` directly.",
+      "reference": "line 316 also handles this case, need to simplify the logic."
+    },
+    {
+      "prediction": "I think this should be `errors.New(\"gcsblob.OpenBucket: bucketName is required\")`",
+      "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 don't think we need to calculate the timeout here. We can just pass `std::chrono::microseconds::max()`.",
+      "reference": "Line too long"
+    },
+    {
+      "prediction": "I don't think these should be added to the magic effect list.",
+      "reference": "This isn't aligned though."
+    },
+    {
+      "prediction": "`PAAHead does not support test-time augmentation` -> `PAA does not support test-time augmentation`",
+      "reference": "it is mean -> it means PAAHead does not support test-time augmentation."
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "@lnielsen-cern thanks. I missed it :("
+    },
+    {
+      "prediction": "I think this should be `content-pages` instead of `content-drilldown`.",
+      "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": "Please remove this 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 think this should be a `bool`, not a `string`.",
+      "reference": "This should be a bool"
+    },
+    {
+      "prediction": "I think this should be `if (swift_typebase && llvm::isa<swift::WeakStorageType>(swift_typebase))`",
+      "reference": "`if (llvm::dyn_cast_or_null<swift::WeakStorageType>(GetSwiftType(compiler_type).getPointer())` ?"
+    },
+    {
+      "prediction": "nit: `canaryManifests = append(canaryManifests, canaryConfigMaps, canarySecrets)`",
+      "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": "Why do you need this exception here?",
+      "reference": "are this docs neccesary ?"
+    },
+    {
+      "prediction": "I don't think you need to make these final. They are only set in the constructor.",
+      "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 don't think you need to do this. If you don't set the loss weight, it will default to 1.0, which is what you want.",
+      "reference": "This change suppose to resolve issue #2968. `loss_weight` should be of length 2, the second entry is ignored."
+    },
+    {
+      "prediction": "I think this should be `strings.HasSuffix(base, prefix)` instead of `strings.HasPrefix(base, prefix)`.",
+      "reference": "Is there an easy way to check for conflict suffix to avoid false positives here?"
+    },
+    {
+      "prediction": "I think this should be `@context.get_validated_task`, not `@context.get_validated_task`.",
+      "reference": "`apply_prep.run_task` also seems like it can just use `@context.run_task`"
+    },
+    {
+      "prediction": "I don't think this is the right fix. I think we should be able to call `autoFocus` without waiting for all sections to load.",
+      "reference": "do you have any idea on how to fix it?"
+    },
+    {
+      "prediction": "I think it would be good to add a test for `GCSTarget` as well, to make sure that the `delete_on_close` flag is respected.",
+      "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 think this should be `time.Now().Before(eni.ExpiresAt)`.",
+      "reference": "When does 'expiresAt' change? Do you need this?"
+    },
+    {
+      "prediction": "This should be a `const Outfit* const`",
+      "reference": "There are enums for genders + `PLAYERSEX_LAST`."
+    },
+    {
+      "prediction": "I think this should be `int64(dataCost) * int64(tt.dataPrice)`",
+      "reference": "Should this use the constants as well? Or intentionally covering the fact that the constants may change"
+    },
+    {
+      "prediction": "I think this should be `return self._size if self._size >= 0 else 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.enableInfiniteTimeout` instead of `worker.enableInfiniteTimeout`",
+      "reference": "I think we should keep this dynamic config knob for operations purpose."
+    },
+    {
+      "prediction": "I'm not sure if this is a bug or not, but I think it should be `(3, 5)` instead of `(2, 4)`.",
+      "reference": "Was this intentional? Perhaps this could be replaced with another way to check the offsets?"
+    },
+    {
+      "prediction": "Should this be `ArtifactBuckets`?",
+      "reference": "the buckets are created via the stackset, customers don't need to provide them"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way to ensure that the `qt` parameter is always included in the request.",
+      "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 don't think we need to export the path here, since we're already doing it in the `before_install` section.",
+      "reference": "This should be ` /$(ghc_find #{config[:ghc]})/ ` instead of ` /#{ghc_version}/ `, I think."
+    },
+    {
+      "prediction": "I think this should be `more_vert` for both Android and Chrome.",
+      "reference": "We should deprecate this at some point."
+    },
+    {
+      "prediction": "I think the default value of `center_sampling` should be `True`.",
+      "reference": "Move these two arguments before `loss_cls`."
+    },
+    {
+      "prediction": "I think this should be `i = serialNumber + mathrand.Int63n(100000000000000000)`",
+      "reference": "[Super NIT]: i := serialNumber if i == 0 { i = mathrand.Int63n(...) }"
+    },
+    {
+      "prediction": "I think this should be `--tool-format` instead of `--tool` to be consistent with the rest of the options.",
+      "reference": "-f short form here collided with --force. :)"
+    },
+    {
+      "prediction": "Why did you remove this?",
+      "reference": "I think it is better to describe in the release note that `rerender()` was removed."
+    },
+    {
+      "prediction": "I don't think we need to cover this.",
+      "reference": "We could also test this easily with `runpy.run_module`, what do you think?"
+    },
+    {
+      "prediction": "I think this should be `v.Type().ConvertibleTo(reflect.TypeOf(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 think we can remove `porcelainAPI` and `retrievalAPI` from this struct.",
+      "reference": "I know this is after the fact but imo blockAPI should be blockMiningAPI."
+    },
+    {
+      "prediction": "I don't think we need this constant. We can use `self::SYSTEM_ATTRIBUTE_PREFIX` directly.",
+      "reference": "it's just a pattern as it was. SystemAttributeCode should have it's own definition and validate by it"
+    },
+    {
+      "prediction": "remove this file?",
+      "reference": "Similar to the code block when we use prefix, maybe we can unify them to base?"
+    },
+    {
+      "prediction": "I think it's better to use `schedulerField.Child(\"duration\")` instead of `schedulerField.Child(\"duration\").Child(\"duration\")`",
+      "reference": "I think `Duration` is useful for this action and we shouldn't ignore it."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to return an error if the stream has already been closed.",
+      "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 && KOKKOS_ENABLE_HIP_ATOMICS`",
+      "reference": "Again, we can also guard `g_host_hip_lock_arrays.atomic`, can't we?"
+    },
+    {
+      "prediction": "I don't think we need to pass `true` here.",
+      "reference": "Variable name should start from lowercase latter"
+    },
+    {
+      "prediction": "I'm not sure why this needs to be 40 seconds. I think 20 seconds should be enough.",
+      "reference": "Why has this timeout doubled?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the default timeout of 30s.",
+      "reference": "Should IptablesPostWriteCheckIntervalSecs be set back to its previous smaller value, if use of the iptables lock is disabled?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be `self._tab.key_press(Qt.Key_Enter, modifier=Qt.NoModifier)` or `self._tab.key_press(Qt.Key_Enter, modifier=Qt.ControlModifier)` depending on whether the focused element is an input field or not.",
+      "reference": "You're missing a `return` here, but I'll add it :smile:"
+    },
+    {
+      "prediction": "I think this should be `gles2_get_texture_in_context(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 why this is necessary. The `format` method should be able to handle unicode strings just fine.",
+      "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 done in `Mongoid::Criteria::Queryable::Extensions::ClassMethods#where`.",
+      "reference": "Why is the respond_to check here?"
+    },
+    {
+      "prediction": "`$char === '\\\\'` should be `$char === '\\\\' || ctype_alnum($char)`",
+      "reference": "@orklah are you sure it's the same as `$chars[$i]`? It looks like `$i` is changed above, after `$char` assignment."
+    },
+    {
+      "prediction": "why nullptr here?",
+      "reference": "I have a question? How do I construct a indexManager for testcase?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. If the plugin is not loaded, we should return an error.",
+      "reference": "These checks aren't needed. See pluginmgr.c : opae_plugin_mgr_load_dflt_plugins( )."
+    },
+    {
+      "prediction": "I don't think this change is correct. If the `q` parameter is not set, it should return `true`.",
+      "reference": "Please take a care about prefixing."
+    },
+    {
+      "prediction": "I think it would be better to rename this method to `getOrCreateAccountForSender` or something like that.",
+      "reference": "this line doesn't match what the rest says"
+    },
+    {
+      "prediction": "Why do we need `linux,android` here?",
+      "reference": "Since it's not working for linux, maybe leave the !linux in there as well?"
+    },
+    {
+      "prediction": "I think this should be `if len(v) == 0 {` instead of `if v == \"\" {` because `v` is a `string`.",
+      "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 think this should be `getModificationDateFieldName()` instead of `modificationDateFieldName`",
+      "reference": "Unrelated bug fix"
+    },
+    {
+      "prediction": "I don't think we need to concatenate the name and version here. We can just pass it as an array of strings.",
+      "reference": "We probably want to let the translation library handle the string replacement for these as well."
+    },
+    {
+      "prediction": "I think we can remove the `getDBFlags` function now.",
+      "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": "Metrics/MethodLength: Method has too many lines. [31/25]",
+      "reference": "What do you think about prefixing the Policy method names with `can_`?"
+    },
+    {
+      "prediction": "I'm not sure if this change is intentional, but it seems like it should be `AWS_CREDENTIAL_RESPONSE_EXPIRATION` instead of `AWS_HTTP_RESPONSE_EXPIRATION`.",
+      "reference": "Based on the docs, credential process can return a key `SessionToken`"
+    },
+    {
+      "prediction": "I think this should be `today.setHours(0, 0, 0, 0);`",
+      "reference": "I think this if should be removed (see issue)"
+    },
+    {
+      "prediction": "nit: I think this should be `github.com/projectcalico/libcalico-go/lib/backend/api`",
+      "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 `Tuple[str, str]`.",
+      "reference": "nitpick: Please move this down to the other imports, as it's a Python stdlib import."
+    },
+    {
+      "prediction": "I don't think we need to link to the how-to guide here.",
+      "reference": "\"for getting started guides\" reads a bit weird, like it's missing punctuation. Maybe just \"for a tutorial\"?"
+    },
+    {
+      "prediction": "I think this should be `$useCaptcha = false` instead of `$uc = false`",
+      "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 we should use `int64` instead of `int` and `uint64` instead of `uint`.",
+      "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'm not sure about this change. I think it's better to keep the class final.",
+      "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": "This class is not needed. You can just use `new Object()` as the event object.",
+      "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": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "I think we have to manually write a down for this migration."
+    },
+    {
+      "prediction": "I don't think we need to check this here. We can check this in `_InternalFrame.attach_default_index` and raise an error there.",
+      "reference": "I don't think we still need this check, but I'd leave it as an assertion for now."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it works for now.",
+      "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 `appExe = Path.Combine(workingDir, appExe);` instead of `appExe = Path.Join(workingDir, appExe);`",
+      "reference": "wondering why we need to join the paths here; seems evt.commandLine only takes whatever appExe is"
+    },
+    {
+      "prediction": "`encodeServiceName` is not used, remove it.",
+      "reference": "1.not format code 2.failed but still continue? 3.\"UTF-8\" can changed to java.nio.charset.StandardCharsets.UTF_8.name()"
+    },
+    {
+      "prediction": "This import is unused.",
+      "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 think this should be `if (_corruptedRequest && !SocketInput.RemoteIntakeFin)` so that we don't call `ProduceEnd` if the remote side has already closed the connection.",
+      "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('instanceGroup')`",
+      "reference": "Please make the change in the CAI inventory, rather than changing the scanner."
+    },
+    {
+      "prediction": "I don't think this should be abstract.",
+      "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 think we can remove the `it_behaves_like 'a Plan with countable subscriptions'` and `it_behaves_like 'a Plan for public listing'` from this file.",
+      "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'm not sure if this is the right thing to do, but I think it's the right thing to do for now. We don't want to limit the number of connections from IPv4 addresses, because the address space isn't as big as IPv6 /64, and we don't want to limit the number of connections from IPv6 /64 addresses, because the address space is too big.",
+      "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": "\"Remove one cookie from remote\" -> \"Remove a cookie from a remote\"",
+      "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 don't think we need to make this public.",
+      "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 don't think we need to check for `type0.getType() == java.lang.Enum.class` because `type0.getType().isEnum()` will return `false` if `type0.getType()` is `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 think it would be better to have a `QuteSchemeHandler.register_handler(name, function)` method, so that we don't have to use a decorator at all.",
+      "reference": "Please use `{}` instead of `dict()`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "I have concerns about using this in more places before #3278 - I'll take a look at that again shortly."
+    },
+    {
+      "prediction": "Why do we need to set `TxIndex` to 1 here?",
+      "reference": "new field needs to be asserted"
+    },
+    {
+      "prediction": "I think this should be `self->logIdToSend_ + 1`",
+      "reference": "when send log failed, why update the last sent log id?"
+    },
+    {
+      "prediction": "I think this should be `prepend` instead of `append`.",
+      "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{}` so that we can store any type of metadata.",
+      "reference": "Why is this necessary? Could you provide some context?"
+    },
+    {
+      "prediction": "I don't think this is necessary. The default charset is already UTF-8.",
+      "reference": "Is there a recommended quote style for attributes? I see single and double here, double further down."
+    },
+    {
+      "prediction": "I think this should be `Eventually(returned).Should(BeClosed())` instead of `Eventually(returned).ShouldNot(BeClosed())`",
+      "reference": "Please use an atomic bool."
+    },
+    {
+      "prediction": "I think this should be in `WebHelpers.cs` in the `Datadog.Trace.ClrProfiler.Managed` project.",
+      "reference": "Consider naming this class `SpanExtensions` to follow C# conventions."
+    },
+    {
+      "prediction": "I don't think we should remove this. It's useful to know which app domain we're running in.",
+      "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": "I don't think we need to log the creation of the container.",
+      "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": "Redundant `self` detected.",
+      "reference": "used named path (`proposals_path`) instead?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 think this should be `yarpcerrors.InvalidArgumentErrorf` instead of `yarpcerrors.InternalErrorf`.",
+      "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 `kwargs[\"protocol\"] = kwargs.get(\"protocol\", \"https\")`",
+      "reference": "if using dict.get, this could just be `kwargs.get('protocol', 'https')`"
+    },
+    {
+      "prediction": "I think this should be `getGrpcStreamingApiMethodExampleName` instead of `getGrpcStreamingApiMethodName`.",
+      "reference": "Add docs to the new methods here"
+    },
+    {
+      "prediction": "Why is this always true?",
+      "reference": "what does it mean enabled?"
+    },
+    {
+      "prediction": "I think this should be `db.execSQL(\"PRAGMA rekey = '\" + newKey + \"'\");`",
+      "reference": "Getting an error when calling with db.execSQL (in sqlcipher 4.3.0, pragma returns ok)."
+    },
+    {
+      "prediction": "Please add a space between `ctxUser.DisplayName()` and `ctx.Tr(\"dashboard\")`",
+      "reference": "Maybe we could remove `\" \" + ctx.Tr(\"dashboard\")` completely?"
+    },
+    {
+      "prediction": "Please remove the extra blank line here.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be nice to have a `jsonHandler3` that takes a `yarpc.Request` instead of a `context.Context` and a `reqBody`.",
+      "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 `TimeSpan.FromMinutes(2)` to match the default http.sys connectionTimeout.",
+      "reference": "Where did we take this default from?"
+    },
+    {
+      "prediction": "I think it would be better to use `compareStartDate` and `compareEndDate` instead of `startDate` and `endDate`.",
+      "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.Count > 0) { _sharedStringMap.Clear(); }`",
+      "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 think we can remove this line.",
+      "reference": "did you mean to delete this line?"
+    },
+    {
+      "prediction": "This should be removed.",
+      "reference": "Should we use logger instead?"
+    },
+    {
+      "prediction": "nitpick: `expect(analytics?).to be true` is the same as `expect(analytics?).to be_truthy`",
+      "reference": "I think it would be preferred to do `expect(helper).to be_analytics`"
+    },
+    {
+      "prediction": "I don't think we need to add `pd.Index` here. `loc` should only accept `str`, `tuple`, `list` and `slice`. `pd.Index` can be converted to `list` or `tuple` before passing to `loc`.",
+      "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 do we need to set this to empty string?",
+      "reference": "If there any other tests, I would make it optional there also."
+    },
+    {
+      "prediction": "This should be `match '/my-accounts' => 'high_voltage/pages#show', as: :my_accounts, id: 'my-accounts'`",
+      "reference": "This can be removed too, right?"
+    },
+    {
+      "prediction": "I think this should be `account.getTransportUris().get(0)`",
+      "reference": "If you're going to allow people to specify multiple SMTP servers, you should also properly export/import them."
+    },
+    {
+      "prediction": "Shouldn't this be `runtime.dotnet.connections.queue_length`?",
+      "reference": "nit: Can we prefix these variables with `AspNetCore`?"
+    },
+    {
+      "prediction": "I think this should be `yaml::conversion.yaml` instead of `yaml::conversion`",
+      "reference": "Can this also include a quick test for transpiling by name?"
+    },
+    {
+      "prediction": "I'm not sure about this. I think we should keep the `SessionStatsDeprecated` struct for now, and add a new `SessionStats` struct for the new format. This way we can still support the old format for a while.",
+      "reference": "This TODO can be removed"
+    },
+    {
+      "prediction": "I think this should be `string tabAccessibilityIdentifier = null`",
+      "reference": "can we make tabAccessibilityIdentifier = null as default?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should return an error from `initProvider` and let the caller decide what to do with it.",
+      "reference": "Should we print this error?"
+    },
+    {
+      "prediction": "`int(multi_scores.size(1) - 1)` -> `multi_scores.size(1) - 1`",
+      "reference": "For which reason do we need to convert this to int here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it is. If the freezer cgroup isn't supported, we don't want to return an error, because that will cause the container to fail to start. Instead, we want to return `nil` so that the container starts normally.",
+      "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 about this change. I think it's better to keep the test plan info.",
+      "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 think this is the only thing that needs to be updated.",
+      "reference": "seems like `RegistryURL` is missing? Same for a few other places that create `archer.Environment` below."
+    },
+    {
+      "prediction": "Please remove unused import.",
+      "reference": "Are these imports needed?"
+    },
+    {
+      "prediction": "I think it would be better to use `if (target.actualNode.type === 'color') { return; }` instead of `if (target.actualNode.type === 'text') { return; }`",
+      "reference": "I think you can remove the `(isIE11 ? it.skip : it)(` above, if you're going to skip the test this way."
+    },
+    {
+      "prediction": "I think this should be `Long.reverseBytes(actualChecksumIn)`.",
+      "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": "Shouldn't this be `DeleteFiles(gopts, repo, removePacksFirst, restic.IndexFile)`?",
+      "reference": "I think we also need something similar for `rebuild-index`?"
+    },
+    {
+      "prediction": "I think this should be 50051.",
+      "reference": "I think convention is snake case for import naming?"
+    },
+    {
+      "prediction": "I think it's better to use `tableInfo.getPrimaryKey().getColumns()` instead of `tableInfo.getPrimaryKey().getIndexColumns()`.",
+      "reference": "i think haspk is false?"
+    },
+    {
+      "prediction": "I think this should be `surefire_dir` to be consistent with the other flags.",
+      "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 think we can remove the `&metrics.ClientConfig{}` here and just pass `nil`.",
+      "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'm not sure this is the right thing to do, but I think it's better than before.",
+      "reference": "A fair bit faster, although still not great, hence also adding a hashkey."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be nice to have a link to the user's profile page in the top right corner of the trail page.",
+      "reference": "If we're going to leave this in here, how about we at least pull it into partial?"
+    },
+    {
+      "prediction": "Why is this timeout increased?",
+      "reference": "This `context` is used correctly, there is nothing left TO DO."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should be passing the `result` object to the callback and let the user decide what they want to do with it.",
+      "reference": "If `docs` is an empty array there will be an error"
+    },
+    {
+      "prediction": "I think this should be `!strings.HasPrefix(host, \"quic://\")` instead of `!strings.HasPrefix(host, \"quic:\")`",
+      "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 commented out?",
+      "reference": "just remove the entire variable."
+    },
+    {
+      "prediction": "I think this should be `File.expand_path(File.join(pwd, moduledir))`",
+      "reference": "Maybe we should put that in the option description?"
+    },
+    {
+      "prediction": "I don't think we need to change this to an `Optional`. The `keyFilePath` field is already an `Optional`, so we don't need to wrap it in another `Optional`.",
+      "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 this should be `MetricsConfig.forTable(table.schema(), table.properties())`",
+      "reference": "Shall we do the same for ORC and Avro?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Should be a space between line"
+    },
+    {
+      "prediction": "I think this should be `SecureRandom.GetBytes(bytes, 0, bytes.Length);`",
+      "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 this should be `video_tutorial.tile` instead of `video_tutorial.card`",
+      "reference": "Probably will undo this change for now, as I'm just targeting Weekly Iteration in this PR."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "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` -> `group_of`",
+      "reference": "I recommend move all openimage related logic to a new function"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `CgroupManager` struct.",
+      "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 about this, but I think it would be better to use `t` instead of `raw` here.",
+      "reference": "I like how you used the count on the translate call to handle this."
+    },
+    {
+      "prediction": "Shouldn't this be `if ($handler->hasDismax() || $handler->hasExtendedDismax())`?",
+      "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": "Please remove this file.",
+      "reference": "[Checkstyle] INFO: Using the '._' form of import should be avoided - java.util._."
+    },
+    {
+      "prediction": "I don't think we need to override the default cache/storage/permission backends here.",
+      "reference": "So, why everything is in memory but the permission backend?"
+    },
+    {
+      "prediction": "I don't think we need to copy the list here. The `allowedExtensions` parameter is only used in the `showBrowseStep` method, which doesn't modify the list.",
+      "reference": "Are you sure it should be null and not an empty List?"
+    },
+    {
+      "prediction": "I think this should be an `ITaskItem[]` instead of a `string[]`.",
+      "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 think these should be `not-localhost` and `not-value` instead of `localhost` and `value`.",
+      "reference": "This test was verifying the wrong thing."
+    },
+    {
+      "prediction": "I think this should be `util.buildMuon('brave', buildConfig, options)` so that we can pass in the buildConfig and options.",
+      "reference": "pls pr a name change fro muon here too separately."
+    },
+    {
+      "prediction": "I think this should be `opts.spinner.Stop(\"Error!\")`",
+      "reference": "nit: we can use `defer opts.spinner.Stop(\"Error!\")` after l.180 to make it slightly neater"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set this flag. I think it should be set in `make_executable()`, which is called by `make_executable_from_file()` and `make_executable_from_memory()`. I think it should be set in `make_executable_from_memory()` because that's the only place that calls `make_executable()` with a non-`NULL` `vmlist`.",
+      "reference": "This is declared in the wrong block"
+    },
+    {
+      "prediction": "`msg.Header:{%v},msg.Route:{%v}` -> `msg.Header:{%+v},msg.Route:{%+v}`",
+      "reference": "In fact it would be useful for debug, i prefer combine them into one debug level log."
+    },
+    {
+      "prediction": "I don't think we need to add \"per session\" here.",
+      "reference": "We would need to localize them eventually too, just FYI"
+    },
+    {
+      "prediction": "I think this should be `proto->set_type(type());`",
+      "reference": "It looks like the proto code needs to change the function name from set_name to set_type to be consistent."
+    },
+    {
+      "prediction": "I don't think we need to generate mocks for this package.",
+      "reference": "I think we can avoid using this pattern for this use case."
+    },
+    {
+      "prediction": "I think this should be `#if !defined(__ANDROID__) && !defined(_WIN32)`",
+      "reference": "Could you please change this to `#ifndef __linux__`? That's when we use our own implementation instead of `posix_spawnp`."
+    },
+    {
+      "prediction": "I don't think we need to filter out the slots that already have a session. We can just check if the count is less than `maxSessionCount` and then check if there is a free slot that supports the capability.",
+      "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 `string(libkbfs.TLFPathType(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": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Should there be a template or something associated with this email? Right now it's identical to `notification_for_approver`"
+    },
+    {
+      "prediction": "I think we should use `EntityManagerInterface` instead of `EntityManager` here.",
+      "reference": "i hope there is some extension in phpstan for this"
+    },
+    {
+      "prediction": "I think this should be `if (sep <= 0) return 0;`",
+      "reference": "We don't store empty dictionary entries anymore: they bring no benefits, only trouble."
+    },
+    {
+      "prediction": "I think this should be `opts_.scoped_enums && opts_.g_cpp_std >= cpp::CPP_STD_17 && IsEnum(type);`",
+      "reference": "Optional: This function is C++ specific and could be declared as `static`."
+    },
+    {
+      "prediction": "I think this should be `getRoles(String domainName, Boolean members, Optional<String> tagKey, Optional<String> tagValue)`",
+      "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 think this should be `warn_deprecation` instead of `warn`.",
+      "reference": "I'd recommend keeping this in `warn_deprecation` and not adding an override of `warn`."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "I think we should use a `blitz` bin instead"
+    },
+    {
+      "prediction": "I think this should be `if dataset._virtual_kdims:`",
+      "reference": "Why not make the ``derived_vdims`` flag (or similar, ``validate_vdims`` maybe?) an explicit argument to ``validate``?"
+    },
+    {
+      "prediction": "I think this should be `if spec.IsPureVolume() && spec.IsSnapshot() {` so that we don't try to create snapshots for non-Pure volumes.",
+      "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": "Please remove the extra space between `*` and `Copyright`.",
+      "reference": "please update the year. it can be setup in intellij template."
+    },
+    {
+      "prediction": "Shouldn't this be `passSlf4j(\"log.debug(null);\");`?",
+      "reference": "what's the use-case for this?"
+    },
+    {
+      "prediction": "I think this should be `isinstance(as_index, bool)`, not `not isinstance(as_index, bool)`.",
+      "reference": "nit: `as_index must be an boolean` -> `as_index must be a boolean`"
+    },
+    {
+      "prediction": "I think this should be `public function upgrade($currentVersion, $newVersion, ConnectionInterface $con = null);`",
+      "reference": "What is the goal of that method ? And what are `$currentVersion` and `$newVersion` ? Are they Thelia or Module version ? :)"
+    },
+    {
+      "prediction": "I don't think we need this.",
+      "reference": "Is there an advantage to this vs setting `this.version` in the constructor? Cleaner this way, at the top?"
+    },
+    {
+      "prediction": "I think this should be `trim()` instead of `trim()`.",
+      "reference": "Fix for URISyntaxException."
+    },
+    {
+      "prediction": "I think this should be `if routeUpd.Dst == nil || routeUpd.Gw == nil {`",
+      "reference": "Would be good to UT this case"
+    },
+    {
+      "prediction": "Why do we need to disable these rules?",
+      "reference": "Why we need to add this?"
+    },
+    {
+      "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.<br>",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `int64` to be consistent with the rest of the API.",
+      "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 `if (_run.ConfigurationNotifications != null && _run.ConfigurationNotifications.Count > 0) { _issueLogJsonWriter.WriteConfigurationNotifications(_run.ConfigurationNotifications); }`",
+      "reference": "_jaw drops_ I would have sworn I wrote those lines. Good catch."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to clear the memoized fns. I think it should be done in the axe._cache.clear() method.",
+      "reference": "This needs to be tested."
+    },
+    {
+      "prediction": "Line is too long. [85/80]<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.",
+      "reference": "I believe you're missing the comparison after 'be'"
+    },
+    {
+      "prediction": "I don't think this is correct. If the key is deleted, we should return ErrNotExist. If the key is not found, we should return ErrNotExist. If the key is found, we should return the value. If the key is found, but the value is deleted, we should return ErrNotExist. If the key is found, but the value is deleted, but the key is not found in the cache, we should return ErrNotExist. If the key is found, but the value is deleted, but the key is not found in the cache, but the key is found in the dao, we should return the value. If the key is found, but the value is deleted, but the key is not found in the cache, but the key is found in the dao, but the value is deleted, we should return ErrNotExist. If the key is found, but the value is deleted, but the key is not found in the cache, but the key is found in the dao, but the value is deleted, but the key is not found in the dao, we should return ErrNotExist. If the key is found, but the value is deleted, but the key is not found in the cache, but the key is found in the dao, but",
+      "reference": "move 103~105 to in front of 97?"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "Lots of unnecessary imports left in this file."
+    },
+    {
+      "prediction": "I don't think we need to do this. We can just use `descriptor->GetClass() == DescriptorClass::InlineUniform` here.",
+      "reference": "Does this actually produce different code?"
+    },
+    {
+      "prediction": "I think this should be `node.getAttribute('onfocus')` instead of `node.getAttribute('onFocus')`.",
+      "reference": "This should account for whitespace. Simply putting `.trim()` on the attribute value should do."
+    },
+    {
+      "prediction": "nit: I think it would be nice to have a `routetable.BlackholeTarget` and `routetable.NoEncapTarget` instead of creating them every time.",
+      "reference": "I don't think we need these. Simpler just to put `routetable.TargetType...` inline below."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Nested if statement. You can merge both statements nested together to create one"
+    },
+    {
+      "prediction": "Shouldn't this be `Integer.parseInt(jdkVersion.substring(0,jdkVersion.indexOf('.'))) >= 10` ?",
+      "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": "`EXPECT_EQ(ResultCode::SUCCESSED, engine->get(\"key\", &result));` -> `ASSERT_EQ(ResultCode::SUCCESSED, engine->get(\"key\", &result));`",
+      "reference": "For your reference in future, I sugguest to use the `ASSERT_*` family."
+    },
+    {
+      "prediction": "This should be `#include <fmt/core.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": "This import is not used. Please remove it.",
+      "reference": "Can we have prefix of `OPENEBS_IO` e.g.: `OPENEBS_IO_INSTANCE_NAME`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we need to implement this method for `folderBranchOps`.",
+      "reference": "Might as well fix these bare returns by making them `return errors.New(...` directly."
+    },
+    {
+      "prediction": "I think this should be `org.hyperledger.besu.datatypes.Hash.EMPTY`",
+      "reference": "we can use import here"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I think it should be added back in.",
+      "reference": "thx for removing these"
+    },
+    {
+      "prediction": "I think this should be an `IOException` instead of a `RuntimeException`.",
+      "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": "Is this a typo? Should it be `testMachineType` instead of `testMachineType2`?",
+      "reference": "How is this different than \"good case\"?"
+    },
+    {
+      "prediction": "I don't think we should change the API here. I think it's better to keep it as `services []string` so that we can support multiple services in the future.",
+      "reference": "the last parameter name should be `serviceName` as is in the implementation"
+    },
+    {
+      "prediction": "I don't understand why this change is needed.",
+      "reference": "@ftomassetti - okay, this is a little meh."
+    },
+    {
+      "prediction": "I think this should be `errors.New(\"os: not implemented\")`",
+      "reference": "Note, changed initial capitalisation of this for consistency. Can do it the other way around too if that'd be better. :wink:"
+    },
+    {
+      "prediction": "I think we can remove the `typeof window === \"object\" ? window : this` and just use `window` here.",
+      "reference": "hey, aren't we supposed to be passing in two parameters here?"
+    },
+    {
+      "prediction": "I think this should be `bitcore.Unit = require('./lib/unit').Unit;`",
+      "reference": "please keep alphabetical ordering :)"
+    },
+    {
+      "prediction": "I think we should use `StandardCharsets.UTF_8` instead of `Charset.forName(\"UTF-8\")`",
+      "reference": "Selenium must compile against Java 6. Revert this line."
+    },
+    {
+      "prediction": "I think we can remove the `resourceGroup` parameter from the `MakePullSubscription` function as well.",
+      "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": "I think this should be `pType !== 'boolean' && pType !== 'undefined'`",
+      "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'm not sure if this is the best way to do this, but it seems to work.",
+      "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'm not sure this is the right place to do this. I think it would be better to do this in `BoltServer::Transport::SSH#connect` and pass the `host-key-check` option to `Net::SSH.start`.",
+      "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": "I think this should be `null` instead of `true` by default.",
+      "reference": "Why is that? Should be redundant as the default value exists."
+    },
+    {
+      "prediction": "I think this should be `DebugSimulateDataRace` instead of `DebugSimulateDataRace`.",
+      "reference": "Are all fields beginning with \"Debug\" automatically `;local` ? (I guess so, but just checking.)"
+    },
+    {
+      "prediction": "I think this should be `UpdateNameMapping updateNameMapping()` instead of `UpdateNameMapping updateNameMapping()`.",
+      "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": "Why is this change necessary?",
+      "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 think we can remove this now that we have `getLLVMRuntimeType`.",
+      "reference": "Is this really necessary? I would expect these types to be included with the loop below."
+    },
+    {
+      "prediction": "I think we should use `BitConverter.ToInt64(bytes.Reverse().ToArray(), 0)` instead of `Array.Reverse(bytes); BitConverter.ToInt64(bytes, 0);`",
+      "reference": "You could do span<byte> and it's magic of typecast to int for better efficiency"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we should use `cssRule.style.getPropertyValue('transform')` instead of `cssRule.style.transform` and `cssRule.style.webkitTransform`. This way we can be sure that we're getting the value of the `transform` property and not the value of the `transform` property prefixed with `-webkit-`.",
+      "reference": "Looks like you've covered `-webkit-transform`, but what about `-ms-transform`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think it should be a warning, not an error.",
+      "reference": "Is there something missing from the message here?"
+    },
+    {
+      "prediction": "I don't think we need to expose `rpc` and `RemoteScheduler` in the public API.",
+      "reference": "Hmm, what does this syntax mean?"
+    },
+    {
+      "prediction": "nit: I think it's better to put `callerTypeAPI` as the first argument.",
+      "reference": "should this be callerTypeTask?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [83/80]",
+      "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 it's better to make it configurable.",
+      "reference": "Maybe make it configurable? Or store the actual level index in DB, making this binary search obsolete?"
+    },
+    {
+      "prediction": "nit: please remove this empty line",
+      "reference": "Nit: this file doesn't need to change. Can you revert this to avoid git conflicts?"
+    },
+    {
+      "prediction": "I think we should add `handleMenuSelected` to the dependency array here.",
+      "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 don't think we need a new type for this. We can just use the `Message` type.",
+      "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 `tearDownClass` instead of `tearDown` so that it only runs once per test class instead of once per test method.",
+      "reference": "nit: you could have named it just \"TestCase\" (if django's TestCase was imported differently)"
+    },
+    {
+      "prediction": "I don't think we need the `&&=` here. We can just do `options[:page_ranges] = Array(options[:page_ranges])`",
+      "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 don't think this change is correct. `dynamic_value` should be `lldb::eNoDynamicValues` by default, and `synthetic_value` should be `true` by default.",
+      "reference": "All the callers of `ExtractSomeIfAny`, always pass `true` to `synthetic_value`. Can we get rid of the extra argument?"
+    },
+    {
+      "prediction": "I don't think we need to log the config file path.",
+      "reference": "Better to use the absolute path."
+    },
+    {
+      "prediction": "I think this should be `ChannelExpiryBuffer` instead of `ChannelExpiryInterval`",
+      "reference": "this is 16 hours, is that enough?"
+    },
+    {
+      "prediction": "I don't think we need to pass in a pointer to `err` here. I think we can just pass in `err`.",
+      "reference": "removing boilerplate. You could also add a `StartSpan` with varargs to inline string attributes."
+    },
+    {
+      "prediction": "I think this should be `uint64` instead of `uint32`",
+      "reference": "Can you replace these single letter variables with full names?"
+    },
+    {
+      "prediction": "This should be `public Action<IFeatureCollection> PrepareRequest { get; set; }`",
+      "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": "I think 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 `getZkController().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": "nit: I think we can use `Iterables.empty()` instead of `CloseableIterable.empty()`.",
+      "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 we should have at least one unit test here.",
+      "reference": "this class should be named `ArrayInitializationVerbosenessTest` to work"
+    },
+    {
+      "prediction": "I think this should be an `else` to the `if val == 'disable'` above.",
+      "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": "Can we remove this? It looks like it was left in by accident.",
+      "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 don't think we need to pass `handleDataError` and `handleDataSuccess` to the `useEffect` hook.",
+      "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": "I think this should be `world_comm_ptr{new lbann_comm(0), lbann::finalize }`.",
+      "reference": "Should we be doing this with `make_unique` or something?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [121/120]",
+      "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": "This import is not needed anymore.",
+      "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 think we should keep `StrictUnusedVariable` as the primary name, and add `UnusedVariable` as an alias.",
+      "reference": "It's unnecessary to duplicate the `name`."
+    },
+    {
+      "prediction": "I don't think this change is necessary. The `unlink` field is a checkbox, so its value will always be `true` or `false`, not `null`.",
+      "reference": "this was a mistake when adding phpstan strict plugin"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this. I think it would be better to have a flag on the `oldVNode` to indicate that it is a placeholder. Then we can just check for that flag instead of having to check for `oldDom` and `oldVNode._dom` and `oldDom.parentNode != parentDom` all at the same time.",
+      "reference": "Nit: I think it should be `to handle` here."
+    },
+    {
+      "prediction": "I don't think this change is correct. It should be `|| message.getId() <= 0`",
+      "reference": "Please don't make unrelated changes or let your tools make unrelated changes."
+    },
+    {
+      "prediction": "I think we should keep the `numAccounts` as `int`.",
+      "reference": "`numAccounts` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think this should be `unset($this->plugins[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($this->loaded[$id]); unset($",
+      "reference": "Is the ID correctly normalized at this point?"
+    },
+    {
+      "prediction": "I don't think you need to pass `0` to `NodeInfo` here.",
+      "reference": "if 0 is a special constant meaning \"no limit\" let's make it a constant and name it correctly :)"
+    },
+    {
+      "prediction": "Why is this moved here? It seems like it should still be after the `setPatchPolicies` call.",
+      "reference": "Curious about the reason behind this change"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be using the `cborStore` instead of the `onlineStore` here. The `onlineStore` is only used for blocks that have been fetched from the network, while the `cborStore` is used for blocks that have been created locally.",
+      "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 done after the `pkg.build` call, but before the `pkg.save` call. That way, if the build fails, we don't have to worry about cleaning up the temp files.",
+      "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 we need this method at all. We can just remove it from the interface.",
+      "reference": "How about name it with NullResponseTypeProcessor"
+    },
+    {
+      "prediction": "I don't think this check is necessary.",
+      "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 think this should be `_process` instead of `_apply`.",
+      "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 don't think we need this constant. We can just use the string `\"node\"` directly.",
+      "reference": "Let's use this constant when ASG is created as well (currently hardcoded)."
+    },
+    {
+      "prediction": "I'm not sure what you mean by \"most unique values to least unique\". Can you give an example?",
+      "reference": "Oh is `bucket` actually higher cardinality than `eventname`?"
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "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 we need to list all of these. I think it's enough to just list `NVDAObjects`.",
+      "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 way to do it, but it works for now.",
+      "reference": "What is this page going to do that is different than the products index we already have?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to return `nil` if we can't find the extension id, rather than raising an exception.",
+      "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 think this should be `attr_accessor :script, :hash_config, :config, :allow_failure, :versioned_runtimes`",
+      "reference": "Is this intended to be here? Seems it belongs to a different PR."
+    },
+    {
+      "prediction": "I think this should be `Console.ResetColor();` instead of `Console.ForegroundColor = ConsoleColor.White;`",
+      "reference": "These changes, while harmless, were no longer needed with newer versions of System.Console thanks to @ianhays"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "reference": "Umm... this isn't relevant to this PR"
+    },
+    {
+      "prediction": "I think this should be `0x0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_0000_000",
+      "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 we can remove this file.",
+      "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 think we should keep the `resetNamespace: true` here and in the `users` route.",
+      "reference": "This change is causing tests to fail because users sits at `/app/users`, not `/app/admin/users`"
+    },
+    {
+      "prediction": "This should be `null` by default.",
+      "reference": "Could you provide more information on what exactly each of those three options do?"
+    },
+    {
+      "prediction": "nit: extra space after `get {`",
+      "reference": "Why this line break?"
+    },
+    {
+      "prediction": "I think we can remove the `apiextensionsv1` import and just use `v1` everywhere.",
+      "reference": "We're importing this twice with different aliases."
+    },
+    {
+      "prediction": "I'm not sure if we want to pin this to a specific version. I think it's fine to have it as `pandera>=0.7.0,<0.8.0` or even `pandera>=0.7.0,<0.9.0`.",
+      "reference": "We should avoid adding a pandera dependency to Dagster itself."
+    },
+    {
+      "prediction": "I don't think you need to create a new array here. You can just do `this.permissions = Arrays.copyOf(this.permissions, this.permissions.length + 1);` and then add the new permission to the end of the array.",
+      "reference": "Can be replaced with `org.jkiss.utils.ArrayUtils#add`."
+    },
+    {
+      "prediction": "I don't think we need to do this. `electionContexts` is a `ConcurrentHashMap`, so it's already thread-safe.",
+      "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 this should be `if len(p.AzureDNS.ManagedIdentityClientID) > 0 || len(p.AzureDNS.ManagedIdentityResourceID) > 0 {`",
+      "reference": "Should these restrictions also be reflected in the API docs?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems to be the only place where we have access to the original request.",
+      "reference": "What about the request headers if someone uses `request_header`? :thinking:"
+    },
+    {
+      "prediction": "I think this should be `--domain-filter=%s` instead of `--domain-filter=%v`",
+      "reference": "Verify that external-dns allows setting this parameter multiple times (that it's not \"last one wins\")."
+    },
+    {
+      "prediction": "I think this should be `etc_hosts: {'host1.example.com': '10.3.1.5'}`",
+      "reference": "Maybe 2 host/ip values could be used in the example?"
+    },
+    {
+      "prediction": "I don't think we need to create a new `Tracer` instance here. We can just use `Tracer.Instance`.",
+      "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": "I think this should be `trainer->get_name()` instead of `pb_trainer->get_name()`. `pb_trainer` is a pointer to a `lbann_data::Trainer`, not a pointer to a `lbann::Trainer`.",
+      "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 `block_a.type () == rai::block_type::state` instead of `block_a.type ()`",
+      "reference": "Wondering if we can directly put types here & for other similar ledger_processor items like rai::block_type::state"
+    },
+    {
+      "prediction": "I think it would be nice to have a progress bar here, too.",
+      "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 think this should be `:attribute \u5df2\u7ecf\u88ab\u4f7f\u7528\u3002`",
+      "reference": "It should be simplified character. `` -> ``"
+    },
+    {
+      "prediction": "I don't think we need to have this here. We can just have `ParallelLevels` in `TrieVisitContext` and set it to `-1` by default.",
+      "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": "I think this should be `TreeMap` instead of `TreeMap`.",
+      "reference": "'A' instead of 'An' here"
+    },
+    {
+      "prediction": "I think this should be `time.Second * 10`",
+      "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 think it would be better to use the same data as in `test_regressor`.",
+      "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 if this is the right thing to do, but I think it's better to use `link_to` instead of `link_to raw(@purchaseable.github_url)`.",
+      "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 used.",
+      "reference": "Please revert the lines above. Not touching unrelated code keeps the git history clean."
+    },
+    {
+      "prediction": "Please update the version to 2.4.0rc0.",
+      "reference": "`short_version` and `version_info`"
+    },
+    {
+      "prediction": "`stats` -> `status`",
+      "reference": "This is supposed to be `status`. Why did you change this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work for me.",
+      "reference": "I will change all of these at some point next week."
+    },
+    {
+      "prediction": "I don't think we need to throw a `NotSupportedException` here. We can just return an empty string.",
+      "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 think this should be `model.KubernetesResourceState_ERROR` instead of `model.KubernetesResourceState_OTHER`.",
+      "reference": "nit: `\"Unexpected error while calculating: unable to convert %T to %T: %v\"`"
+    },
+    {
+      "prediction": "I think this should be `applicationName` instead of `appName` to be consistent with the rest of the codebase.",
+      "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 don't think we need to export this. It's only used in one place.",
+      "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": "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 getUpperCamelMethodName();`",
+      "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 don't think we need to check for '/' here. I think we should only check for '.' as that is not a valid character in a file name.",
+      "reference": "What if .. appears, but not as the first character, eg \"errors/../../../../../../../../../../../\"?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [82/80]",
+      "reference": "Thanks, this should make things a bit easier for people who pick up the codebase but aren't a `curation center`"
+    },
+    {
+      "prediction": "Can we remove this?",
+      "reference": "why not removing it?"
+    },
+    {
+      "prediction": "I think this should be `disable` instead of `disableHostVerification`",
+      "reference": "same nit here: maybe word as - \"disables validation of the temporal cluster's server certificate\""
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "Nit: other statements use all caps for SQL reserved words. Should this be `SHOW TABLES`?"
+    },
+    {
+      "prediction": "I think this should be `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 think this should be `if c.Wireguard.EncryptHostTraffic && c.KubernetesProvider == config.ProviderAKS && c.RouteSource == \"WorkloadIPs\"`",
+      "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 \"Name of the service.\"",
+      "reference": "Do we need to change this flag as well?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 `nameOrHandle` instead of `handle`.",
+      "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'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Only one line of code? What a simple issue!"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we want to use the `NewDefaultIncomingDynamicRateLimiter` for the outgoing rate limiter, and the `NewDefaultOutgoingDynamicRateLimiter` for the incoming rate limiter.",
+      "reference": "Conceptual question: why do history and matching need throttler at all? Shouldn't throttling to be handled on FE only?"
+    },
+    {
+      "prediction": "nit: I think it would be clearer to say \"The internal NetworkPolicy may not have been created yet. It's fine to skip processing this CNP as addCNP will create it eventually.\"",
+      "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/MethodLength: Method has too many lines. [31/25]",
+      "reference": "Minor: you can access `current_user` in here directly - don't need to pass it in. Unless you prefer passing it explicitly?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "What is the purpose of `&` on this line?"
+    },
+    {
+      "prediction": "Shouldn't this be `^[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 we should close the connection here.",
+      "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 right place to do this. I think it would be better to do this in `getSpeechTextForProperties`.",
+      "reference": "Could you split these conditions up over multiple lines please."
+    },
+    {
+      "prediction": "I'm not sure why we need 3 here. I think 2 should be enough.",
+      "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 this should be `{{ range $port := .ports }} - \"{{ $port }}:{{ $port }}\" {{ end }}`",
+      "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": "nit: I think we can remove the `if s.shardInfo != nil` check here.",
+      "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 `LastWriter keybase1.UID` instead of `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": "I think this should be `c->maximized` instead of `c->maximized_horizontal || c->maximized`.",
+      "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'm not sure why these were removed, but I think they should be added back in.",
+      "reference": "Why did you remove these packages from the prelude, will your new functions not work without this change?"
+    },
+    {
+      "prediction": "`kvKeys` should be passed by value, not by reference.",
+      "reference": "We'd better reserve enough space before using kvKeys to avoid extra malloc."
+    },
+    {
+      "prediction": "I think this should be called `GetNodeName` since it returns the name of the node.",
+      "reference": "would it make sense to return empty or error if label doesn't exists?"
+    },
+    {
+      "prediction": "I think this should be `numAttempts < maxAttempts` instead of `numAttempts <= maxAttempts`.",
+      "reference": "Don't penalize everyone with 250ms delay b/c some machines have problems."
+    },
+    {
+      "prediction": "I think this should be `out = fmt.Sprintf(\"%s, %s\", out, it.Value().String())`",
+      "reference": "Probably want a space after the second %s?"
+    },
+    {
+      "prediction": "I think this should be `serviceerror.NewUnavailable(archiver.ErrHistoryNotExist.Error())`",
+      "reference": "This seems like an internal error still?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "reference": "Line is too long. [94/80]"
+    },
+    {
+      "prediction": "Nit: I think it would be better to use `time.Time` instead of `int64` for the creation time.",
+      "reference": "out of curiosity, why creating another struct to wrap it?"
+    },
+    {
+      "prediction": "I don't think you need to use `fmt::format` here, you can just use `fmt::format_to` to write to a `std::string` and then call `GetOwner()->Message` with that string.",
+      "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 `setProtoFile(protoFile)` instead of `setAssignedProtoFile(protoFile)`.",
+      "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 `0` instead of `\"0\"`.",
+      "reference": "How about `SPARK_DEFAULT_SERIES_NAME`?"
+    },
+    {
+      "prediction": "I think this should be `COMMIT_STATUS_CHECKS_TOTAL_WAIT_MS`",
+      "reference": "The other properties are in `commit.status-check`, not `commit.status-checks`. Could you remove the extra `s`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it should be done in `setupDatabase` rather than here.",
+      "reference": "Side note: I'd love to see more of these fields exposed on configuration."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we can remove the `@resource.is_a?(User)` check and just use `@resource.invited_by.name` in both cases.",
+      "reference": "In the case of accounts generated by API clients, what name gets put for the `inviter_name` or `invited_by` record?"
+    },
+    {
+      "prediction": "I don't think you need to implement it here. The `getStartBlock` method in `MutableBftConfigOptions` will return `Optional.empty()` if `startBlock` is not set.",
+      "reference": "I think this class is only relevant for updating the QbftConfigOptions based on the transitions config, so probably not."
+    },
+    {
+      "prediction": "I think it should be `TracerTimeoutExceeded` or `TracerTimeoutExceededError`",
+      "reference": "what is the number thrown by Geth?"
+    },
+    {
+      "prediction": "Why do we need to split on `?` here?",
+      "reference": "This would still crash on malformed database connection string like \"test\""
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `plugin` package.",
+      "reference": "it is not an interface, maybe we can move to another package?"
+    },
+    {
+      "prediction": "`tableMeta` -> `meta`",
+      "reference": "Same here. Can we use `this.instance.getSettings()` to get `tableMeta`?"
+    },
+    {
+      "prediction": "I think this should be `cgroup.Resources.CpuWeight != 0 || cgroup.Resources.CpuShares != 0 || cgroup.Resources.CpuQuota != 0 || cgroup.Resources.CpuPeriod != 0 || cgroup.Resources.CpuRealtimePeriod != 0 || cgroup.Resources.CpuRealtimeRuntime != 0 || cgroup.Resources.CpuCfsQuotaMicros != 0 || cgroup.Resources.CpuCfsPeriodMicros != 0 || cgroup.Resources.CpuShares != 0 || cgroup.Resources.CpuQuota != 0 || cgroup.Resources.CpuPeriod != 0 || cgroup.Resources.CpuRealtimePeriod != 0 || cgroup.Resources.CpuRealtimeRuntime != 0 || cgroup.Resources.CpuCfsQuotaMicros != 0 || cgroup.Resources.CpuCfsPeriodMicros != 0 || cgroup.Resources.CpuShares != 0 || cgroup.Resources.CpuQuota != 0 || cgroup.Resources.CpuPeriod != 0 || cgroup.Resources.CpuRealtimePeriod != 0 || cgroup.Resources.CpuRealtimeRuntime != 0 || cgroup.Resources.CpuCfsQuotaMicros != 0 || cgroup",
+      "reference": "Not sure whether this validation is useful."
+    },
+    {
+      "prediction": "I think it would be better to use `message.info(\"Yanked download URL to clipboard: {}\".format(question.yank_text))` instead of `message.info(\"Yanked download URL to clipboard: {}\".format(s))`.",
+      "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": "This should be removed.",
+      "reference": "If this is needed for testing, you should use `pytest --log-cli-level=INFO` instead."
+    },
+    {
+      "prediction": "I think this should be `https://testnet2-location.mysterium.network/api/v1/location`",
+      "reference": "We should drop ipify in all places i guess"
+    },
+    {
+      "prediction": "I don't think we should change the names of the consensus params. I think it's better to keep them as they are.",
+      "reference": "Add explanations around these variables - what do they mean, how they should be configured, etc."
+    },
+    {
+      "prediction": "Metrics/BlockLength: Block has too many lines. [31/25]",
+      "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": "I think this should be `displayGatheredResources` instead of `displayResources`",
+      "reference": "To much tabs here (1 tab)"
+    },
+    {
+      "prediction": "nit: I think we can use `String.format` instead of `+` here.",
+      "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 don't think we need to override the retry policy for this test. We can just use the default one.",
+      "reference": "I don't like this helper func :-). Yeah, definitely don't like."
+    },
+    {
+      "prediction": "I think we can remove the `else if` here as well.",
+      "reference": "Same here. This should remain unchanged. Please, add it back."
+    },
+    {
+      "prediction": "Shouldn't this be `options.ignoreUndefined = this.s.options.ignoreUndefined;`?",
+      "reference": "@kvwalker should this method have `ignoreUndefined` supported as an `option`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it would be nice to have a `NewLogger` function that takes a `zap.Logger` and returns a `*zap.SugaredLogger`. That way we don't have to do this everywhere.",
+      "reference": "we dont use a `sugaredlogger` anywhere else in clutch, just curious why were using it here?"
+    },
+    {
+      "prediction": "Why is this line removed?",
+      "reference": "Not evident from looking at this diff but the queue is already being cleared four lines below."
+    },
+    {
+      "prediction": "I think this should be `t.Sender == transactions.CompactCertSender` instead of `t.Sender == transactions.CompactCertSender.String()`.",
+      "reference": "Why is this a 'transaction' and not in the block header?"
+    },
+    {
+      "prediction": "I think this should be `X509Certificate` instead of `X509Certificate2`. `X509Certificate2` is a subclass of `X509Certificate`.",
+      "reference": "This doesn't fit the abstraction level. Should we have a property bag for extra stuff?"
+    },
+    {
+      "prediction": "I don't think we need to include `ci/gendoc.py` in the sdist.",
+      "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 don't think we need to defer this. The p2p service will be stopped when the bee is stopped, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p service, which will stop the p2p",
+      "reference": "do we really need the `Ready` call if startup fails midway?"
+    },
+    {
+      "prediction": "Why do we need to specify the port here?",
+      "reference": "You can avoid changing the code if there are parameter defaults. right ?"
+    },
+    {
+      "prediction": "I'm not sure if this is a change in behavior or not, but it seems like it should be a change in behavior.",
+      "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 `atoi(sep->arg[1]) == 0`",
+      "reference": "I know this is legal, but I prefer we put brackets on our calls"
+    },
+    {
+      "prediction": "`AWS IAM` -> `AWS Identity and Access Management`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I don't think we need to create a new exception for this. Let's just use `InvalidElementStateException`.",
+      "reference": "We keep who wrote the code anonymous."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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'm not sure this is the right place to put this. I think it should be in the `Transport` interface.",
+      "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": "I don't think it's necessary to mention the author here.",
+      "reference": "Feel free to use the full notation with my email address: `Maxime Steinhausser <maxime.steinhausser@gmail.com>` :smile:"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "This was unused across the codebase"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "reference": "Should we put in which account we were trying to look up?"
+    },
+    {
+      "prediction": "I think this should be `InventoryTypeClass.Supported_TypeClass`",
+      "reference": "either SUPPORTED_TYPECLASS or supported_typeclass. Camelcase only for class names."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add this column. I think it would be better to add it to the `notes` table in the `notes` schema.",
+      "reference": "This is not going to work. You need to add a migration to the database."
+    },
+    {
+      "prediction": "I think this method should be renamed to something like `resumePlayback` or `resumeGroupPlayback`.",
+      "reference": "Should these methods be renamed in this PR?"
+    },
+    {
+      "prediction": "I think it would be better to throw an `ArgumentException` here.",
+      "reference": "we need to include the `metricStreamName` which is invalid, so users know whats causing the issue."
+    },
+    {
+      "prediction": "I think we need to use `AssemblyLoadContext.Default.LoadFromAssemblyPath(binaryPath);` instead of `Assembly.Load(AssemblyLoadContext.GetAssemblyName(binaryPath));`",
+      "reference": "Please run Platform tests `DiaSessionTests`."
+    },
+    {
+      "prediction": "I think this should be `MSG_AVX_512_SUPPORT_INCOMPLETE` instead of `MSG_AVX_512_SUPPORT_INCOMPLETE_STRING`.",
+      "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": "Shouldn't this be `if (!options.directConnection && !options.directconnection)`?",
+      "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 don't think we should use the same name for all benchmarks.",
+      "reference": "Does the name suffix hardcode something? Not clear how changing the name fixes this."
+    },
+    {
+      "prediction": "Should this be a `List<String>` instead of a `String`?",
+      "reference": "Is undispatched same as queued? getQueuedFlows?"
+    },
+    {
+      "prediction": "I think this should be `Core Site data store: connection info tests`.",
+      "reference": "See above, same for all similar cases below."
+    },
+    {
+      "prediction": "I'm not sure why we need to append the trace logs twice here, but it seems to be the only way to get the test to pass.",
+      "reference": "Why is this done twice?"
+    },
+    {
+      "prediction": "I don't think we need to propagate the deadline to the children nodes. If the deadline is already exceeded, we should not propagate it to the children nodes.",
+      "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=true`.",
+      "reference": "This was renamed to avoid any confusion with the more general case of debugging via `plz debug`"
+    },
+    {
+      "prediction": "I don't think this is correct. If the length of the string is less than `prefixLength`, this will throw `StringIndexOutOfBoundsException`.",
+      "reference": "String valStr = (String)val; valStr.substring(0, Math.min(valStr.length(), prefixLength)) Make it clean."
+    },
+    {
+      "prediction": "I think this should be `evb.CreateEnumerator(\"NONE\", BASE_TYPE_NONE);` instead of `evb.CreateEnumerator(\"NONE\", 0);`",
+      "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 can remove this method from the Alerter interface. It is only used by the email alerter.",
+      "reference": "Looks like getAzkabanURL() is added after concrete Alerter (Emailer)::getAzkabanURL(), so that Override annotation is needed."
+    },
+    {
+      "prediction": "Why do we need this? `segms[i]` is already a boolean mask.",
+      "reference": "The above 3 lines can be written as: `mask = segms[i].astype(bool)`"
+    },
+    {
+      "prediction": "I think we don't need to import `address` here.",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I don't think we need to log this every time the node starts.",
+      "reference": "Thought it's good to have this logged down so that we can check the used value in the logs."
+    },
+    {
+      "prediction": "I don't think we need to make this optional.",
+      "reference": "I don't think you can call this optional unless it's a pointer - remove this?"
+    },
+    {
+      "prediction": "I think it would be better to show a toast when there are no products to send, not when there are no products saved.",
+      "reference": "@naofum please use `isEmpty`method instead"
+    },
+    {
+      "prediction": "I think you can remove the `auto` here.",
+      "reference": "i think this is unrelated and needs to be reverted. need to discuss at sync."
+    },
+    {
+      "prediction": "I think this should be `org.hyperledger.besu.plugin.data.BlockHeader` instead of `org.hyperledger.besu.plugin.data.BlockHeader`",
+      "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 don't think we need to return an array of results here. We can just return the result of the last one.",
+      "reference": "Love how we can finally use async/await in the codebase."
+    },
+    {
+      "prediction": "I think this should default to `true`",
+      "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": "Why do we need this field?",
+      "reference": "when is this consumed? I'm probably missing it but I can't find it right now"
+    },
+    {
+      "prediction": "I think this should be `if (isRollup() || isGroupingSet())`",
+      "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 don't think you need to import `com.google.api.codegen.viewmodel.SimpleParamDocView` here.",
+      "reference": "can use `getParamTypeName` instead"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this code. It should be in the `toggleBooleanProperty` method.",
+      "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 think this should be `if selected_file is None:` or `if not selected_file:` or `if len(selected_file) == 0:` or `if not selected_file:`",
+      "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": "I think it would be better to return an error here and not remove the name file. The name file should only be removed when the server is started for the first time.",
+      "reference": "How will this code behave with hostnames that contain hyphens?"
+    },
+    {
+      "prediction": "I think this should be `if (kegg != null && !string.IsNullOrWhiteSpace(kegg))`",
+      "reference": "I think this is redundant since \"NullForEmpty\" already calls \"Trim()\"."
+    },
+    {
+      "prediction": "I think this should be `CloudBackupGroupID` instead of `GroupCloudBackupID`.",
+      "reference": "Change this to IDs too?"
+    },
+    {
+      "prediction": "This should be a `Set` not an `ImmutableSet`.",
+      "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'm not sure why we need to limit the size of the buffer to 127 bytes here. I don't see anything in the man page for fscanf that would limit the size of the buffer.",
+      "reference": "attr_value is 128 bytes?"
+    },
+    {
+      "prediction": "shouldn't it be `self.method = self.method.upper() if not has_variable_pattern(self.method) else self.method`?",
+      "reference": "if it's unpredictable - why uppercase it at all?"
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "Please remove all these debug prints. It doesn't look good."
+    },
+    {
+      "prediction": "I think this should be `SampleChainRandomness(height *types.BlockHeight) ([]byte, error)`",
+      "reference": "Consumers of this interface should not be required to provide `sampleHeight`. This should be an expected consensus parameter."
+    },
+    {
+      "prediction": "I think this should be `if closeTime.Before(startTime) || closeTime.Equal(startTime) {`",
+      "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 think this should be a `map[string]string` instead of a `string` so that we can filter on multiple tags.",
+      "reference": "Not sure if we need tag here, it is an implementation detail in portworx that isn't exposed in openstorage"
+    },
+    {
+      "prediction": "I don't think we need the backticks here.",
+      "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 we should use `searcher.getCore().getCoreContainer().getSolrClientCache()` instead of `new SolrClientCache()`.",
+      "reference": "Need the `solrClientCache.close();` further down in the method be removed since a shared cache is now used?"
+    },
+    {
+      "prediction": "nit: I think we should call this `MutableData` and `MutableDataEncoding` to make it clear that it's not immutable.",
+      "reference": "according to the PR (schema changes?) these 2 field should be removed?"
+    },
+    {
+      "prediction": "I think we can remove the `ContainerNames` field from `Target.StressChaos` and use `ContainerName` instead.",
+      "reference": "Panic here if `exp.Target.StressChaos.CintainerName` is nil."
+    },
+    {
+      "prediction": "Shouldn't this be `DocIdSetIterator filterIterator(DocIdSetIterator scorerIterator)`?",
+      "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": "I think we can remove this debug log.",
+      "reference": "Aren't there already logs for this from the underlying file system implementation?"
+    },
+    {
+      "prediction": "I think this should be `['DismaxMunge' => []]` instead of `['DismaxMunge' => null]`.",
+      "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`",
+      "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 it would be better to add this help text to the `AttributeProductMetaTitleType` instead of the `AttributeIdType`. This way the help text is only shown when the `AttributeProductMetaTitleType` is used and not when the `AttributeIdType` is used.",
+      "reference": "Value in product should contain 255 characters or less."
+    },
+    {
+      "prediction": "I don't think we need to do this. We can just use the `ProposerInterval` from the FSM config.",
+      "reference": "line is 133 characters (from `lll`)"
+    },
+    {
+      "prediction": "I think this should be `personal_options` instead of `personal_options_file`.",
+      "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'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Will we want to refactor this when we drop rsolr 1.x support?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what else to do either. We don't have a way to get the IP address of the node, and we don't have a way to get the name of the node, so I'm not sure what else to do. I think this is the best we can do for now.",
+      "reference": "should this really be multiple IPs?"
+    },
+    {
+      "prediction": "Extra empty line at the end of a file.",
+      "reference": "Hmm, is that partial used in multiple places? Maybe we can just move that file in here."
+    },
+    {
+      "prediction": "I don't think we need to generate mocks for this package.",
+      "reference": "Should be 2014-2016"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like a good place to me.",
+      "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'm not sure if this is necessary, but it doesn't hurt to close the cursor.",
+      "reference": "similarly here, are the these cursors not being closed when the client is closed?"
+    },
+    {
+      "prediction": "I think we can remove the `string` parameter from the `Policy` interface.",
+      "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'm not sure this is the right place to call this. I think it should be called in `serviceLoggedIn` after `config.MakeDiskBlockCacheIfNotExists()` is called. This is because `MakeFBOsForExistingJournals` will create FBOs for existing journals, and `MakeDiskBlockCacheIfNotExists` will create FBOs for new journals. If we call `MakeFBOsForExistingJournals` before `MakeDiskBlockCacheIfNotExists`, then we will create FBOs for new journals twice.",
+      "reference": "Should this happen even if the above failed?"
+    },
+    {
+      "prediction": "copyright year should be 2017-2018",
+      "reference": "Please dont change this ."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to log this. I think it should be logged in the `erase` method of `nano::active_transactions`.",
+      "reference": "There is a strange $b that looks wrong."
+    },
+    {
+      "prediction": "I think this should be `containerContainsText = _(\"with %s items\") % childControlCount`",
+      "reference": "Is it still important to have `speakEntry` here? It is checked before this value `containerContainsText` is used in the \"General\" section."
+    },
+    {
+      "prediction": "I think this should be `sh.export 'PATH', '$JAVA_HOME/bin:$PATH'` instead of `sh.export 'PATH', '$JAVA_HOME:$PATH'`",
+      "reference": "It is `$JAVA_HOME/bin`, not `$JAVA_HOME`, which should be added."
+    },
+    {
+      "prediction": "This should be `#include <assert.h>`.",
+      "reference": "Should move this to within the `#else` block as it's only used there."
+    },
+    {
+      "prediction": "Why do you need to wait for AJAX and then wait for 1 second?",
+      "reference": "is this a common rule to wait one extra second after ajax? I'm thinking about moving this extra wait into waitForAjax method."
+    },
+    {
+      "prediction": "I think this should be `unset($properties['site'], $properties['logger'], $properties['solr']);`",
+      "reference": "Maybe you can correct the typo \"if\" too :)"
+    },
+    {
+      "prediction": "I think this should be `IdentifyChat` instead of `IdentifyForChat`",
+      "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": "This should be `if (PyMPIComm_Get == nullptr) {`",
+      "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 correct. The `defer resp.Body.Close()` is in the `else` clause, so it won't be called if the `resp.StatusCode == http.StatusNotFound` condition is true.",
+      "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 to add these features. I think they should be added to the `IConnection` interface instead.",
+      "reference": "Nit: implement the methods for these interfaces explicitly in this file for consistency. It can be passthrough."
+    },
+    {
+      "prediction": "I don't think you need to add `getCallName(\"\")` here.",
+      "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 this should be `NABoolean isFirstBatch = true;`",
+      "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's better to use `folly::toAppend` instead of `std::string::reserve` and `std::string::append`.",
+      "reference": "You could reserve a const space to save the extra computation."
+    },
+    {
+      "prediction": "I think it would be good to have a test case with a path that has a mix of redundant and non-redundant segments. Something like `/home/user/runtime/src/coreclr/runtime/src/libraries/System.Private.CoreLib/src/../System/IO/..//./Path.cs`",
+      "reference": "Did you consider adding rooted paths like `C:\\ProgramData` and paths with flipped separators?"
+    },
+    {
+      "prediction": "I don't think this is correct. The defer should be after the `if err != nil {` block.",
+      "reference": "does this line cause a panic if it is above the if condition ?"
+    },
+    {
+      "prediction": "I think it would be better to name this enum `LibraryDetailsPanel` instead of `LibraryDetailsPanels` as it is not a collection of panels but a single panel.",
+      "reference": "Maybe `LibraryDetailsPanelType` would be clearer."
+    },
+    {
+      "prediction": "I think this function is unused now.",
+      "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 removed?",
+      "reference": "I verified offline with @jeremyg-lunarg this should be removed."
+    },
+    {
+      "prediction": "This should be removed.",
+      "reference": "This should be removed? If not, typo in incldued_wf"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this. I think it's a bit of a hack, but I couldn't think of a better way to do it.",
+      "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 think this should be `SplitPaths` to be consistent with the rest of the codebase.",
+      "reference": "Do you think the godoc should mention that all delimiters will be suffixed with `/`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it should be done here. If the user is in private browsing mode and they try to navigate to a `data:` URL, we should redirect them to the private browsing page.",
+      "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 `private long minSequenceNumber = Long.MAX_VALUE;`",
+      "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 don't think we should change this to `integer` as it will not work for versions like 1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0",
+      "reference": "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_HISTORY_LENGTH` or `MAX_HISTORY_SIZE` or something like that. `MAX_HISTORY` is too vague.",
+      "reference": "Are there tests for this? What happens when you get to limit?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be able to set the transport to `nil` and have it default to `local` if the target is `localhost`.",
+      "reference": "Whole thing could be collapsed to `data['config']['transport'] ||= 'local' if target.name == 'localhost'`."
+    },
+    {
+      "prediction": "I think you can remove this method and call `getSignKeyId(data)` directly from `onClick()`.",
+      "reference": "maybe throw in `&& mServiceConnection.isBound()`, there are cases where the connection might die"
+    },
+    {
+      "prediction": "I don't think this is correct. EIP-1344 should be enabled in MuirGlacier.",
+      "reference": "Can we add Istanbul with this first EIP please? (or as a separate issue)"
+    },
+    {
+      "prediction": "I think this should be `return errors.Wrap(err, \"failed to get current user\")`",
+      "reference": "I don't understand this change"
+    },
+    {
+      "prediction": "\"Invalid Public key\" should be \"Invalid public key\"",
+      "reference": "Let's allow translators for this string to be translated: `self.show_warning(_('Invalid Public key'))`"
+    },
+    {
+      "prediction": "I think this should be `if (!role && (parent.tagName.toLowerCase() !== 'form' && parent.tagName.toLowerCase() !== 'body'))`",
+      "reference": "This needs to work with `role=form` too."
+    },
+    {
+      "prediction": "I think this should be `if (localConnection && localConnection.type == Blockly.OUTPUT_VALUE) {` instead of `if (localConnection && localConnection.type == Blockly.OUTPUT_VALUE) {`",
+      "reference": "This should be before the if on line 1028, which should turn into an else if"
+    },
+    {
+      "prediction": "nit: I think we should use `[]metav1.LabelSelectorRequirement` instead of `*[]metav1.LabelSelectorRequirement`",
+      "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 think this should be `sample-r.minRTT > ackDelay` instead of `sample-r.minRTT >= ackDelay`.",
+      "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": "Why is this `internal`?",
+      "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 if this is the best way to do this, but it works for now.",
+      "reference": "nice. we should do this elsewhere too. We have a lot of repeated stuff in the policies"
+    },
+    {
+      "prediction": "I think this should be `TEST_ERRORS_1(src, \"can't find declaration of 'y'\")`",
+      "reference": "Small formatting thing, but can you remove the space before the semicolon?"
+    },
+    {
+      "prediction": "I think this should be `long remaining = InnerLoopCount * (long)_writeLenght;`",
+      "reference": "Not new, but nit: _writeLeng*th*."
+    },
+    {
+      "prediction": "I think this should be `assert.NoError(checkDockerCompose())` instead of `assert.NoError(checkDockerCompose)`.",
+      "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": "nit: please remove the empty line",
+      "reference": "Remove this file from the PR"
+    },
+    {
+      "prediction": "I think this should be `Blockly.Variables.nameValidator_.bind(null, varType, variable.name)`.",
+      "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 `libinput_device_config_tap_get_finger_count(libinput_dev) > 1`",
+      "reference": "I think I would rather fish this interface through than use it as the basis for heuristics."
+    },
+    {
+      "prediction": "Remove the extra space at the end of this line.",
+      "reference": "Probably not worth making this line-ending change, since it is the only change to this file."
+    },
+    {
+      "prediction": "I don't think we need to assert anything here. If the size is not 64, then we will fail to decode the instruction, which is the right thing to do.",
+      "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 it's a good idea to use `proto` as a package name. It's too generic.",
+      "reference": "No pb please"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "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'm not sure if this is the right place to put this function. I think it should be in the slackbot package, but I'm not sure if it should be in the slackbot/slackbot.go or slackbot/slackbot_test.go file.",
+      "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'm not sure about this change. I think it's better to keep the docstring as it was.",
+      "reference": "\"not to be used elsewhere other than\""
+    },
+    {
+      "prediction": "I think we need to set the `spark.sql.parquet.output.committer.class` property here as well.",
+      "reference": "Thanks for the contribution, @coolderli ! I also think the newPosDeleteWriter need the properties setting ..."
+    },
+    {
+      "prediction": "I'm not sure if we need to add an index on `plan_type` as well, but I think it doesn't hurt to have it.",
+      "reference": "These additions seem unrelated to this change?"
+    },
+    {
+      "prediction": "I think this should be `iceberg.engine.hive.enabled` instead of `engine.hive.enabled`",
+      "reference": "Could we move this to a class for Hadoop configuration properties, like `org.apache.iceberg.hadoop.ConfigProperties`?"
+    },
+    {
+      "prediction": "why remove this?",
+      "reference": "The related statements should also be deleted."
+    },
+    {
+      "prediction": "Shouldn't this be `params[\"PolicyNames\"] = None`?",
+      "reference": "Shouldn't this line be removed?"
+    },
+    {
+      "prediction": "I don't think we need this function. We can just call `username in bad_names` directly.",
+      "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 why this is needed, but it seems like it should be removed.",
+      "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 change this value?",
+      "reference": "Need change it back?"
+    },
+    {
+      "prediction": "Trailing whitespace detected.",
+      "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 we can remove this line.",
+      "reference": "nit: Looks like this can be removed now"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it is. If the user doesn't specify any CAs, then we should create a default one for them. If they do specify CAs, then we don't want to create a default one.",
+      "reference": "Can probably remove the lines above it too (L60-62), since ProvisionDefaultCA() makes sure the map isn't nil."
+    },
+    {
+      "prediction": "nitpick: `static int handler_registered = 0;` should be `static int handler_registered = 1;`",
+      "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 don't think this is the right fix. I think it should be `# +1: [fixme]`",
+      "reference": "I think the functional test should not change here, this is probably a test in itself :)"
+    },
+    {
+      "prediction": "I think it's better to use `var Sanitizer = bluemonday.UGCPolicy().AllowAttrs(\"class\").Matching(regexp.MustCompile(`[\\p{L}\\p{N}\\s\\-_',:\\[\\]!\\./\\\\\\(\\)&]*`)).OnElements(\"code\").AllowAttrs(\"type\").Matching(regexp.MustCompile(`^checkbox$`)).OnElements(\"input\").AllowAttrs(\"checked\", \"disabled\").OnElements(\"input\")`",
+      "reference": "Why this to be a public function?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to not log the timing of the `export` command.",
+      "reference": "I think we can remove the timing for export env vars"
+    },
+    {
+      "prediction": "I think it's better to keep `config.EndpointSliceHandler` and `config.NodeHandler` in the `Provider` interface, and add `Run()` and `GetServiceByIP()` to the `Provider` interface.",
+      "reference": "wrong import grouping"
+    },
+    {
+      "prediction": "I don't think this is the right fix. I think it should be `UserPreferences.getBackButtonGoToPage()` instead of `context.getResources().getStringArray(R.array.back_button_go_to_pages_tags)`.",
+      "reference": "There is a doubled semicolon ;)"
+    },
+    {
+      "prediction": "This should be `#include <openssl/md5.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": "`visibility_of_any_element_located`?",
+      "reference": "shouldn't **call** return a boolean?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the `CastSpell` function of the `NPC` class.",
+      "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": "This should be `String(256)`",
+      "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 think this should be called `GetDestinationAddress` or `GetRecipientAddress` or something like that. `GetReceiverAddress` sounds like it's the receiver of the transaction, not the destination of the transaction.",
+      "reference": "should the 'Receiver' be the clawback address in case of clawback transaction?"
+    },
+    {
+      "prediction": "I don't think we need these constants. They are not used anywhere.",
+      "reference": "@ashishranjan738 -- Where are these consts used? in which pkg?"
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "The only change I would suggest is moving these lines back below the `templates_path` variable to keep the diffs smaller."
+    },
+    {
+      "prediction": "I'm not sure why this is necessary. `changes` is a pointer, so `changes.Info = info` should modify the original `changes` object.",
+      "reference": "Having it above kept the block info in the cached changes, which could end up confusing things quite a bit."
+    },
+    {
+      "prediction": "I think we can remove the `task_name = options[:object]` line here as well.",
+      "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": "Why do we need to pass a channel here?",
+      "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 don't think we need to do this. If the wal has been rolled back, we will get E_RAFT_NO_WAL_FOUND anyway.",
+      "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": "I don't think this is correct. The form type name should be used here, not the form type class.",
+      "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": "I'm not sure if this is the right thing to do, but I think we need to add the `rpm-package-key.gpg` key to the yum repo for EL6 as well.",
+      "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 these here. They are already defined in `org.openqa.selenium.remote.CapabilityType` and `org.openqa.selenium.remote.Timeouts`.",
+      "reference": "These are really meant to be the keys in the capabilities, not the keys of values within the capabilities"
+    },
+    {
+      "prediction": "I'm not sure why this is needed, but it seems to be required for the tests to pass.",
+      "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 `1.0 / 4`",
+      "reference": "Maybe move all the Loss Recovery constants to a separate file."
+    },
+    {
+      "prediction": "I think we should rename this to `name` to be consistent with the rest of the codebase.",
+      "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 { continue }`",
+      "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'm not sure about this change. I think it's better to keep the same structure as the other pages.",
+      "reference": "What changed in the styles that means we don't need this?"
+    },
+    {
+      "prediction": "I think this should be `projectJsonRoot[\"frameworks\"][framework][\"dependencies\"] as JObject;`",
+      "reference": "This pattern should be applied to the other instance where NewtonsoftEscapeJProperty is used and you can remove the NewtonsoftEscapeJProperty method."
+    },
+    {
+      "prediction": "This should be a deprecation warning, not a warning.",
+      "reference": "Something to mention in the next changelog/release notes. It will be good to get histogram working consistently with everything else."
+    },
+    {
+      "prediction": "I don't think we need to add this here. This is only used in the `initContainers` section, and we don't have any `initContainers` in this template.",
+      "reference": "What does this contain? Is it PVC namespace or is it openebs?"
+    },
+    {
+      "prediction": "I think this should be `IHttpBodyMinimumDataRateFeature`",
+      "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": "I think this should be `R_AS_INT(data_idx) - 1` since R is 1-indexed and C++ is 0-indexed.",
+      "reference": "this will break R package, you should update the R files accordingly. BTW, the call state should be in the last argument."
+    },
+    {
+      "prediction": "`$mediaContext` should be `$media->getContext()`",
+      "reference": "Shouldn't we throw an exception if the context is null instead ?"
+    },
+    {
+      "prediction": "This should be `>= 1.12.0`",
+      "reference": "These should both be const, not var right?"
+    },
+    {
+      "prediction": "`build_match_cost` -> `build_match_cost_calculator`",
+      "reference": "Rename the directory, match_costers -> match_costs"
+    },
+    {
+      "prediction": "I think this should be `self.assertEqual(int(count), 2)`",
+      "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 don't think this is the right message. It should be something like \"Session ID is null. Did you call quit()?\" or \"Session ID is null. Did you call close()?\" or \"Session ID is null. Did you call shutdown()?\" or \"Session ID is null. Did you call stop()?\" or \"Session ID is null. Did you call destroy()?\" or \"Session ID is null. Did you call terminate()?\" or \"Session ID is null. Did you call exit()?\" or \"Session ID is null. Did you call kill()?\" or \"Session ID is null. Did you call stopServer()?\" or \"Session ID is null. Did you call stopWebDriver()?\" or \"Session ID is null. Did you call stopDriver()?\" or \"Session ID is null. Did you call stopBrowser()?\" or \"Session ID is null. Did you call stopApplication()?\" or \"Session ID is null. Did you call stopProcess()?\" or \"Session ID is null. Did you call stopService()?\" or \"Session ID is null. Did you call stopApplicationServer()?\" or \"Session ID is null. Did you call stopApplicationDriver()?\" or \"Session ID is null. Did you call",
+      "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 think this should be `exporter.Controller().Stop(context.Background())` instead of `exporter.Controller().Stop(context.TODO())`.",
+      "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 should keep this test, but change it to use `expect(STDOUT).to receive(:print).with('Please enter your password: ')` instead of `allow(STDOUT).to receive(:print).with('Please enter your password: ')`",
+      "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 think this should be `MaxAllowedMaxFrameSize` instead of `_maxFrameSize`. `_maxFrameSize` can be less than `MaxAllowedMaxFrameSize`.",
+      "reference": "Remove this since it always has to be set afterwards."
+    },
+    {
+      "prediction": "I think this should be `blockChangeSize(20*1024*1024)`?",
+      "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 think we can remove the `now` and `requestContext` parameters from the `fetch` method in the `LeafNode` class. The `now` parameter is not used in the `LeafNode` class, and the `requestContext` parameter is not used in the `fetch` method of the `LeafNode` class.",
+      "reference": "Is that `try..except` block really needed? I mean, when it could fail?"
+    },
+    {
+      "prediction": "I think this should be `**/suspense.test.js` so that we don't have to update this file every time we add a new test file.",
+      "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 don't think we need to hardcode these values here. We can get them from the `boto3` SDK.",
+      "reference": "nit: I'd probably rename this to `BATCH_SIZE_RANGES`, to use a slightly more descriptive name."
+    },
+    {
+      "prediction": "This should be `90` and `100`",
+      "reference": "aren't these the wrong way around?"
+    },
+    {
+      "prediction": "This is not used in this PR. Can you please remove it?",
+      "reference": "Lets call this getConfigSetService and maybe rename the field now (or later)"
+    },
+    {
+      "prediction": "I think we can remove the `klogr` import now?",
+      "reference": "Could / should we make these constants and give them a type?"
+    },
+    {
+      "prediction": "I don't think we need to hardcode the platform names here. We can get them from the `Platform` field of the `ClusterDeployment` object.",
+      "reference": "lets stick with lowercase for consistency."
+    },
+    {
+      "prediction": "I don't think we need to change the order of the imports here.",
+      "reference": "ordering of imports went wrong"
+    },
+    {
+      "prediction": "I think this should be `return MetadataTableType.ALL_DATA_FILES.name();`",
+      "reference": "I think these can all be .name() to get the value we are looking for"
+    },
+    {
+      "prediction": "I think this should be `(print \"test\") === 1;` instead of `(print \"test\") === 2;`",
+      "reference": "Is this a good place for these test cases?"
+    },
+    {
+      "prediction": "Please remove this print statement.",
+      "reference": "Did you mean to leave the print statement?"
+    },
+    {
+      "prediction": "This should be `new BN(500000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "I would do `new BN(Interpreter.LOCKTIME_THRESHOLD)`"
+    },
+    {
+      "prediction": "I think we should keep the imports as they were before.",
+      "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 this should be `TimeoutType%v` instead of `TimeoutTimeoutType%v`",
+      "reference": "Need to add type name here to be compatible with proto string representation."
+    },
+    {
+      "prediction": "I think this should be `log.info`",
+      "reference": "This is to let people know that the tooling is doing something during a build. Please leave."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I think it's better to use a real user principal here rather than a fake one.",
+      "reference": "Why is the principal changing here?"
+    },
+    {
+      "prediction": "I think this should be `AZURE_STORAGE_DOMAIN` instead of `AZURE_BLOB_DOMAIN`.",
+      "reference": "Naming nit: let's use `AZURE_STORAGE_` prefix for consistency with the other ones."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "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 don't think we need to import `errors` in this file.",
+      "reference": "can we use `github.com/pkg/errors`?"
+    },
+    {
+      "prediction": "I think this should be `actualType` instead of `inferredType`.",
+      "reference": "I changed the name of the arguments here just to help make the code clearer."
+    },
+    {
+      "prediction": "I think this should be \"unnecessarily\" instead of \"unncessarily\".",
+      "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 auto &attachment : shared_attachments_` instead of `const auto attachment : shared_attachments_`.",
+      "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 don't think we need to use `Nethermind.Monitoring.Metrics` namespace here. We can use `Nethermind.Monitoring` instead.",
+      "reference": "minor: typo (knowMetricsTypes -> knownMetricsTypes)"
+    },
+    {
+      "prediction": "I think this should be `((CharType*)getType())->setCaseinsensitive(((CharType *)(((ConstValue*)(right))->getType()))->isCaseinsensitive() || ((CharType *)(((ConstValue*)(right))->getType()))->isCaseinsensitive());`",
+      "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 don't think this is the right place for this method. It should be in `TypeSolver` or `TypeSolverFactory`.",
+      "reference": "I would throw an exception if the type is not solved"
+    },
+    {
+      "prediction": "I don't think we need to skip these anymore.",
+      "reference": "if remove `dt.skipLoad` - then this tests will run. You probably mean opposite?"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "I feel like we do this in a lot of places should we move that out into its own PR?"
+    },
+    {
+      "prediction": "Why is this async now?",
+      "reference": "Already asked you in Matrix, but I'll do it here again so it can be discussed publicly. Why?"
+    },
+    {
+      "prediction": "I think it would be better to say \"de-indent all the code inside it\" instead of \"de-indent all the code inside it and de-indent all the code inside it\".",
+      "reference": "I think here should be a comma before remove, in all the cases"
+    },
+    {
+      "prediction": "why do we need `stages` here?",
+      "reference": "Single quote is used by default in this project."
+    },
+    {
+      "prediction": "Why don't we want to reconcile on spc?",
+      "reference": "S1023: redundant `return` statement (from `gosimple`)"
+    },
+    {
+      "prediction": "Why do we need to implement `Close`?",
+      "reference": "`ctx` is unused in Close"
+    },
+    {
+      "prediction": "I think this class is not needed. You can throw `NotFoundHttpException` directly.",
+      "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": "Please add `type='DCNv2'`",
+      "reference": "Is it necessary to add the suffix `on_head`, because it belongs to` bbox_head`?"
+    },
+    {
+      "prediction": "This should be `AS_IF([test x$with_modern_grub = xyes], [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 `DeleteInstance` instead of `Delete`.",
+      "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": "Line is too long. [101/80]",
+      "reference": "Line is too long. [104/80]"
+    },
+    {
+      "prediction": "I think this should be `expectedAssignedCDs` instead of `expectedAssignedCDS`",
+      "reference": "Since CD updates (assignment & power state) are now done in this controller..."
+    },
+    {
+      "prediction": "I think this should be `opts && opts.conflictStack && opts.conflictStack.join('\\n\\n')`",
+      "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": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "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 this should be a separate test case.",
+      "reference": "This line is too long"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just log the error and keep going.",
+      "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 why this change is necessary, but it seems like it should be.",
+      "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'm not sure about the `:host => SERVER_URL` here. I think it should be `:host => request.host_with_port` or something like that.",
+      "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 think this should be `EnableLogging *bool` to be consistent with the other fields.",
+      "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": "why we need to set the data version to 3.0 here?",
+      "reference": "use `NebulaKeyUtilsV3::dataVersionValue()` to replace the literal value?"
+    },
+    {
+      "prediction": "I don't think this is the right thing to do. I think we should be raising an exception here.",
+      "reference": "No need for the `\\` here, inside parentheses whitespace is ignored by Python."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to return an empty string if the certificate is not present.",
+      "reference": "This doesn't seem like the right expected result... it should be the same as the `clientCert` var defined above."
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I think it should be restored.",
+      "reference": "need to rename this now."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it would be better to add the missing fields to the `PrivateTransactionReceipt` class in the `org.web3j.protocol.besu.response.privacy` package.",
+      "reference": "NIT: This shouldn't be a TODO. It is just a note isn't it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this warning. I think it should be in the `set_value` method of the `webelem` module.",
+      "reference": "I changed this to an error, that seems more appropriate than a warning here."
+    },
+    {
+      "prediction": "I don't think we need to change this part, since we still need to check if the log is a system log or not.",
+      "reference": "This change could mute unit test failure, but it is better to update unit tests"
+    },
+    {
+      "prediction": "I think this should be `notificationIdsInUse.remove(notificationId);` instead of `notificationIdsInUse.delete(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 this should be `NotFoundErr` instead of `ErrNotFound`",
+      "reference": "keep as `ErrNotFound`, it is the same convention as in other packages"
+    },
+    {
+      "prediction": "I think this should be `if (offlineMode === true) {` instead of `if (offlineMode) {`",
+      "reference": "Same here, we need to call `callback` in else branch"
+    },
+    {
+      "prediction": "I think this is a bug fix. If the host name is null, it should be an empty string, not null.",
+      "reference": "At this point we already check that the record is valid and so it could never be nulll"
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.<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.<br>Layout/Space",
+      "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() should be called in the end of heartBeatThreadFunc()",
+      "reference": "why not do it in `reclaimExpiredSessions`? and `pushSessionToCache` can delete"
+    },
+    {
+      "prediction": "Please remove `_base_` and `model` in this file.",
+      "reference": "out_channels/num_outs/type are unnecessary"
+    },
+    {
+      "prediction": "I'm not sure what you're trying to do here. `Object.keys(dict).length` is always going to be 0 because `dict` is an empty object.",
+      "reference": "You can create an issue for it and put it on the backlog so we don't forget it."
+    },
+    {
+      "prediction": "Shouldn't this be `if validNetworkMode(container.HostConfig)`?",
+      "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 thing to do here. I think we should keep the contents of the package in the database even if we fail to install it, so that we don't have to download it again.",
+      "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": "I don't think we need this struct. We can just pass the fields directly to the `ProposeBlock` method.",
+      "reference": "Let's add fee recipient as well."
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "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 `APPLICATION_NAME` instead of `applicationName`.",
+      "reference": "I think there's another spot for this in DefaultCapabilityMatcher"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "this name seems pretty vague -- thoughts on a more descriptive method name?"
+    },
+    {
+      "prediction": "I think we should keep the copyright year as 2020.",
+      "reference": "what changed in this line?"
+    },
+    {
+      "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 get the `reportArgs` object into the `hasData` function.",
+      "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": "I don't think this is needed anymore.",
+      "reference": "This line was useful for auto-completion, I think it should be re-added with specifying the var name (`$exception`) and simplifying the FQCN."
+    },
+    {
+      "prediction": "Why do we need a channel here? Why can't we just return the error from the goroutine?",
+      "reference": "This looks like you're not collecting everything from the error channel."
+    },
+    {
+      "prediction": "I'm not sure if we want to use the plan name here. I think we should use the product name instead.",
+      "reference": "This currently violates the Law of Demeter. Using `checkout.plan_name` is the quick resolution."
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "reference": "is there any scenario when `$temporaryFilename` can be null?"
+    },
+    {
+      "prediction": "Why do we need `@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": "I think this should be done in `roots_drag_icon_update_position` instead of duplicating the code.",
+      "reference": "No need to damage after `roots_drag_icon_update_position`, this is already done in `roots_drag_icon_update_position`"
+    },
+    {
+      "prediction": "`device` is not used in this function, please remove it.",
+      "reference": "Have you tested exporting to ONNX with `device=cuda`?"
+    },
+    {
+      "prediction": "I think this should be `if o.name != \"\" && o.since != 0 && o.humanStartTime != \"\" {`",
+      "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": "Shouldn't this be `Request::setTrustedHosts(['^(.+\\.)?' . preg_quote(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": "Shouldn't this 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 the right thing to do, but I think it's better to have the test fail than to have it pass with the wrong value.",
+      "reference": "Are we marking what the browser's accessibility tree says or what screen readers with those browsers say?"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "This method should behave as intended, to block until it reads an EOF from the other side."
+    },
+    {
+      "prediction": "`Fail to health check repository '%s': %v` -> `Health check failed on repository '%s': %v`",
+      "reference": "I prefer starts with `Failed to xxx`, how about `Failed to perform health check on xxx`?"
+    },
+    {
+      "prediction": "I think it would be better to have a single `Tags.AzureAppServicesMetadata` tag that contains all the metadata as a JSON string. That way we don't have to add a new tag every time we add a new metadata field.",
+      "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": "I'm not sure if this is a bug or not, but I think this should be `v1` instead of `v1_test`.",
+      "reference": "This file shouldn't be in this PR"
+    },
+    {
+      "prediction": "I don't think this should be 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 debug level.",
+      "reference": "Are you going to change that back?"
+    },
+    {
+      "prediction": "I think this should be `ifeq (true,$(staticbuild))`",
+      "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 think this should be `delay` instead of `scale`",
+      "reference": "Note this should be `delay` instead of `scale` here. I'll fix this in my branch."
+    },
+    {
+      "prediction": "I think we should keep `this.on` and `this.off` here.",
+      "reference": "`on` likely still needs a bind, similar to `this.resetProgress = this.resetProgress.bind(this)` correct?"
+    },
+    {
+      "prediction": "Please remove this line.",
+      "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: please remove this extra blank line",
+      "reference": "Can you move this up along with the other builtin libraries?"
+    },
+    {
+      "prediction": "I think it would be better to put this checkbox in the update settings panel, as it is related to the update process.",
+      "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'm not sure if this is the right place to put this, but I think it would be useful to have a `--help` for the `--checkout_command` flag that explains that the `%s` will be replaced with the revision to check out.",
+      "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 `f.server_conn.ip_address and 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 it should be `UtcDateTime` instead of `LocalDateTime`",
+      "reference": "Why local and not UTC?"
+    },
+    {
+      "prediction": "I don't think this is the right thing to do. If the error code is -2013 or -2012, then the error message should not be generated. If the error code is not -2013 or -2012, then the error message should be generated.",
+      "reference": "2012 is a retryable error. Will avoiding rgenerating it here cause a difference in behavior in createServer() ?"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "move to line 14 below"
+    },
+    {
+      "prediction": "I think \"crosses\" is not the right word here. I think \"intersects\" is the right word here.",
+      "reference": "the method name should match teh docs"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do for now. We don't have any tests for MDv3 yet, and we don't have any tests that use `createNewRMD`, so I don't think this will break anything.",
+      "reference": "Won't this happen automatically when we flip the default version in the config?"
+    },
+    {
+      "prediction": "I'm not sure why this changed, but I think it should be reverted.",
+      "reference": "This would also change back to /var/www/html/docroot if we go that way."
+    },
+    {
+      "prediction": "I think this should be `hasImplicitRole` instead of `hasImplicitChromiumRoleMatches`",
+      "reference": "This still needs to use the new option. We'll need tests for it too."
+    },
+    {
+      "prediction": "I don't think you need to set `shouldAllowIAccessibleFocusEvent` to `True` here. It's already `True` by default.",
+      "reference": "We are no longer including file name in the source files."
+    },
+    {
+      "prediction": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.",
+      "reference": "no need for this. Devise provides us with `current_user` and `user_signed_in?` helpers."
+    },
+    {
+      "prediction": "I don't think we need this method. We can just use `capability.getVersion() >= ETH66.getVersion()` directly.",
+      "reference": "Would `Object.equals(capability.getName(), NAME)` would be safer, as you wouldn't have to do null checks?"
+    },
+    {
+      "prediction": "Line is too long. [84/80]",
+      "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' %q 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 this test should be updated to use `normalizeToPartialURL` instead of `extendPartialURL`.",
+      "reference": "TestNormalize... I would also split them"
+    },
+    {
+      "prediction": "`next()` is a built-in function, not a method of `generator` objects.",
+      "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 fix. I think we should be using `rand.Intn` instead of `r.random.Intn`.",
+      "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 don't think you need to import `astroid_version` and `pylint_version` here.",
+      "reference": "Can we grab it directly from `__pkginfo__` as that is the source of truth for the version?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "We can revert this to reduce the diff of the PR."
+    },
+    {
+      "prediction": "I think this should be `if (isCloud && !protoPkg.contains(\"cloud\")) {`",
+      "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": "nit: I think it would be better to use `==` instead of `== Lob_Append_Or_Error`.",
+      "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 `finest` instead of `info`.",
+      "reference": "This change prevents a user understanding how their server is configured. Best to leave at `info` level."
+    },
+    {
+      "prediction": "I think this should be `#if defined(SSL_set_tlsext_host_name) && OPENSSL_VERSION_NUMBER < 0x009080FFL`",
+      "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 think we should rename this variable to `readActiveBlockProducersByHeightTests`",
+      "reference": "`readActiveBlockProducersByHeightTests` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I don't think we need to pass `idField` to `saveRecordsToSmartStore`",
+      "reference": "Falls back on `Constants.ID` if there's no custom field set."
+    },
+    {
+      "prediction": "I think this should be `in.skipNBytes(this.getOffset());`",
+      "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 `getReturnType().isPrimitiveType()`",
+      "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 think this should be `yarpc.Interceptors{yarpc.InboundInterceptor{requestLogInterceptor{}}}`",
+      "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'm not sure if this is the right place to do this, but it seems like a good place to me.",
+      "reference": "Is there anything we should do in the event we're unable to renaming the directory?"
+    },
+    {
+      "prediction": "nit: I think it would be good to add a `TODO` here to remove the `TODO` in `pkg/broker/ingress/ingress.go` as well.",
+      "reference": "How is this env var populated? Can this be retrieved from cluster metadata?"
+    },
+    {
+      "prediction": "I think this should be `SuccessfulTagInternetGateways` (plural) since we're tagging multiple gateways.",
+      "reference": "Should probably skip the success event here, since it could be a noop."
+    },
+    {
+      "prediction": "This should be `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 why this was removed, but I think it should be restored.",
+      "reference": "Let's keep this line in the docstring, to let end users know they shouldn't be using this directly."
+    },
+    {
+      "prediction": "why remove this check?",
+      "reference": "can do the same for other validateXXX()"
+    },
+    {
+      "prediction": "I don't think you need to cast away the const here. `participant_` is already a pointer to a `const DomainParticipantImpl`.",
+      "reference": "Why is this necessary?"
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "reference": "Please move this import upper to other similar imports"
+    },
+    {
+      "prediction": "I think this should be `IF_ARM(ASSERT_NOT_TESTED(dcontext));` so we know which dcontext is not tested.",
+      "reference": "Include the key word \"translated\". Otherwise the reader will find this confusing."
+    },
+    {
+      "prediction": "I'm not sure I understand why this is needed. The `ElasticsearchSpanStore` is not a singleton, it's created for each request.",
+      "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": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the selected option. The `selected` attribute doesn't seem to be set on the `option` element.",
+      "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 don't think it's a good idea to have a `Migrator` as a base class for the storage. It's not a storage, it's a migration tool. I'd rather have a `Migrator` as a separate class that can be used by the storage, but not as a base class.",
+      "reference": "ditto about use of `abspath`"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it makes the code harder to understand, and I don't think it's worth it.",
+      "reference": "this reduces the runtime by 50 times"
+    },
+    {
+      "prediction": "I don't think we need a test for this class.",
+      "reference": "Probably best not to have an empty test...."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think it would be better to throw an `IllegalStateException` with a message that includes the stack trace of the exception that was thrown when trying to instantiate the exception type.",
+      "reference": "direct `newInstance` call is also deprecated now"
+    },
+    {
+      "prediction": "nit: please remove empty line",
+      "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 this should be `final` to prevent subclasses from overriding it.",
+      "reference": "Why was this change needed?"
+    },
+    {
+      "prediction": "I don't think this should be a warning, it should be an exception.",
+      "reference": "can you use this opportunity to remove `.readlines()` useless list creation? (`addresses = [l for l in fd if l.find(\"netmask\") >= 0]`)"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [123/120]",
+      "reference": "Why does this check differ from `plan_ssh.rb`?"
+    },
+    {
+      "prediction": "why remove this import?",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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 we should have debug options in the command line.",
+      "reference": "We have cli --online_weight_clear in cli.cpp"
+    },
+    {
+      "prediction": "I think this should be `strings.SplitN(values[0], \",\", 2)[0]` in case there are multiple addresses in the `X-Forwarded-For` header.",
+      "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": "I think this should be \"Execution stopped, crashed executor/container\" instead of \"Execution stopped, crashed executor/container\"",
+      "reference": "How is this message used? Can we remove the \"crashed executor/container\" part which is an implementation detail?"
+    },
+    {
+      "prediction": "I don't think we need to add this to the `nameToUrl` map. It's only used in the `execute` method, which doesn't seem to use the `nameToUrl` map at all.",
+      "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 this is the right place for this link. I think it should be in the `watch_video` partial.",
+      "reference": "Space after `=`."
+    },
+    {
+      "prediction": "I don't think this is the right place to put the invocation context. The context should be set when the invocation is created, not when it is invoked.",
+      "reference": "if just need to do something after createInvocation just override and call super first is enough?"
+    },
+    {
+      "prediction": "I think this should be `it_behaves_like 'a jvm build'`",
+      "reference": "By the way, I fixed `announce` method (missing `super` call to announce JDK version)"
+    },
+    {
+      "prediction": "I don't think we need to suppress the warning here. The `contains` method of `SortedSet` takes an `Object` as parameter.",
+      "reference": "I'm not sure about that... This line can produce <code>ClassCastException</code> if <code>Comparator</code> do not check this."
+    },
+    {
+      "prediction": "I think this should be `*BareTlfHandle`",
+      "reference": "There wasn't any special reason for this to return a pointer, right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best solution, but it seems to be the only way to make this thread-safe.",
+      "reference": "Why the need for a `ConcurrentHashMap`?"
+    },
+    {
+      "prediction": "I think this should be `if (order.by === 'title' && Setting.value('titleNaturalSort')) {`",
+      "reference": "New code should use strict equality `===`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it seems to work.",
+      "reference": "`expect(resources['Notify[local pid]']['events'][0]['desired_value']).to match(/(\\d+)/)` seems clearer."
+    },
+    {
+      "prediction": "I don't think we need to pass in `changeMd` anymore.",
+      "reference": "Removed now-unneeded param."
+    },
+    {
+      "prediction": "I think this should be `go-filecoin bitswap --help` or `go-filecoin bitswap help`",
+      "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 `Google Analytics` instead of just `Analytics`.",
+      "reference": "The `_x` function needs to be imported at the top of the file (in addition to `__`)"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "This is functionally equivalent to the code it replaces."
+    },
+    {
+      "prediction": "nit: `ENIType` -> `Type`",
+      "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": "I think this should be `Widgets data store: widget tests`.",
+      "reference": "See above, same for the other cases."
+    },
+    {
+      "prediction": "I don't think we need to suppress the unchecked warning here.",
+      "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 don't think we need to disable this.",
+      "reference": "I would prefer to just fix the occurences of this new check rather than disable them"
+    },
+    {
+      "prediction": "I think this should be `https://news.google.com/rss?hl=en-US&gl=US&ceid=US:en` instead of `https://news.google.com/news/rss?`",
+      "reference": "The tests should be fixed in #4841, so this is no longer needed"
+    },
+    {
+      "prediction": "Why do you need this class?",
+      "reference": "I think it should been in `SharedKernel` module."
+    },
+    {
+      "prediction": "I don't think we need to add a new field for this. We can just use the `nginx.ingress.kubernetes.io/whitelist-source-range` annotation.",
+      "reference": "Can we add some description for this field?"
+    },
+    {
+      "prediction": "I think this should be `DurationType.class` instead of `TimeType.class`.",
+      "reference": "Should its name be Time or Duration? I can see them both."
+    },
+    {
+      "prediction": "nit: I think it would be better to call this `toGRPCOptions` or something like that. `asGRPCOptions` sounds like it's converting something to GRPC options, but it's not. It's creating a slice of `otlpconfig.GRPCOption`s from a slice of `Option`s.",
+      "reference": "Is this used anywhere?"
+    },
+    {
+      "prediction": "I think it would be good to add a test for `resource.Empty()` as well.",
+      "reference": "I believe that you can use `assert` instead of `require` here and in the following lines"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do. We don't want to return an error if we find the channel, we want to return an error if we don't find the channel.",
+      "reference": "In `storagemarket.go` you used `nil` for an unwanted out parameter. Do something consistent (nil seems fine if supported)."
+    },
+    {
+      "prediction": "Why do we need to export these functions?",
+      "reference": "is this a new pattern we use in bee? interesting"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the bucket already exists, we should probably return an error.",
+      "reference": "If you think this PR works, i'll clean this up before resubmitting."
+    },
+    {
+      "prediction": "what is the meaning of GROUPID and ARTIFACTID?",
+      "reference": "are you sure you can read version by this artifactid?"
+    },
+    {
+      "prediction": "I don't think we need this import.",
+      "reference": "We can get rid of this import then."
+    },
+    {
+      "prediction": "I don't think we need the `case ROUTER` anymore.",
+      "reference": "Should we do our own vtable?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [115/100]",
+      "reference": "This is the actual fix."
+    },
+    {
+      "prediction": "I think this should be 30s",
+      "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 way to do this. I think it would be better to have a `DecodeMapExact` and `DecodeMapCaseInsensitive` method on the `Decoder` interface, and then have the `DecodeMap` method just call `DecodeMapCaseInsensitive` by default.",
+      "reference": "should be called => will be called"
+    },
+    {
+      "prediction": "`lastBlockTime.Add(lastBlock.Timestamp().Sub(lastBlockTime) / c.blockInterval * c.blockInterval)` can be simplified to `lastBlockTime.Add(lastBlock.Timestamp().Sub(lastBlockTime) % c.blockInterval)`",
+      "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'm not sure this is the right thing to do. If the exception is a `SolrException`, then it's already wrapped in a `SolrException`, so we don't need to wrap it in another `SolrException`.",
+      "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 don't think it's a good idea to hardcode the name of the module here. I think it would be better to use `logger.msg.startswith(\"datatable.rbind(...) {\")` or something like that.",
+      "reference": "`dt` won't work for some reason?"
+    },
+    {
+      "prediction": "I don't think we need this change. `os.getenv(\"KOALAS_USAGE_LOGGER\", \"\") != \"\"` is equivalent to `os.getenv(\"KOALAS_USAGE_LOGGER\", None) is not None`.",
+      "reference": "Is this because \"KOALAS_USAGE_LOGGER\" can be `None` ??"
+    },
+    {
+      "prediction": "I think it would be better to have these constants in the `EthGasPriceOracle` class, as they are related to it.",
+      "reference": "I like it but maybe the better name will be EthGasPriceConstants or EthGasPriceEstimatorConstants? but Constants not Config"
+    },
+    {
+      "prediction": "nit: I think it would be nice to have a `NewController` function that takes a `Constructor` as an argument, so that we don't have to create a `Constructor` just to pass it to `NewController`.",
+      "reference": "I would inline the constructor as well"
+    },
+    {
+      "prediction": "`dummyyylval` -> `dummyyylval`",
+      "reference": "I'm not confident about this..."
+    },
+    {
+      "prediction": "I think this should be `v1alpha2.SchemeGroupVersion.WithKind(\"CertificateRequest\")`?",
+      "reference": "There is also `v1alpha1.CertificateRequestKind` I think?"
+    },
+    {
+      "prediction": "I think this should be `gcpbroker_sender` instead of `sender_gcpbroker`",
+      "reference": "Can we rename it to make it less confusing? Instead of \"sender-gcpbroker\", maybe name it to \"retryable-sender\""
+    },
+    {
+      "prediction": "nit: I think this should be `literal().value().getClass().getSimpleName()` 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 this should be `abs(nrmse.getMetric()[\"value\"] - target) < OPFMetricsTest.DELTA`",
+      "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 it would be better to raise an error here, so that the user knows that they need to install `datatable` in order to use it.",
+      "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 think we can remove the `bold` call here as well.",
+      "reference": "We're fine with double quotes here. Just for future reference."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to return `unknown_internal_yarpc` if the error code is `OK` than to return `OK`.",
+      "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 `if (graphicsPipelineCI.pInputAssemblyState && graphicsPipelineCI.pInputAssemblyState->topology != VK_PRIMITIVE_TOPOLOGY_MAX_ENUM)`",
+      "reference": "`RecordPipelineShaderStage()` might change `topology_at_rasterizer `, according to shader code, so we should check `pInputAssemblyState `first."
+    },
+    {
+      "prediction": "I think this should be `const Assets = require('./assets').default;`",
+      "reference": "Should we just inline the assets here? I can't think of an advantage to having them in a separate file."
+    },
+    {
+      "prediction": "I'm not sure about the name of this function. `detectStale` sounds like it's going to do something more than just detect stale endpoints. Maybe `detectStaleEndpoints` or `detectStaleEndpointsAndServices`?",
+      "reference": "detectStaleEndpoints to be more specific?"
+    },
+    {
+      "prediction": "I'm not sure why this is commented out. Is it because it's not implemented in C++ yet?",
+      "reference": "Please create a new issue for addressing this (if there isn't one already), so it doesn't get lost."
+    },
+    {
+      "prediction": "I'm not sure why this is needed, but it seems like it should be fixed in a separate PR.",
+      "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": "`AWS IAM` -> `AWS Identity and Access Management`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I don't think we need to log this at `FINE` level. I think `INFO` is fine.",
+      "reference": "This is deliberately at this level."
+    },
+    {
+      "prediction": "I don't think we need to do this here. We can just pass `args[2]` to the `hex.DecodeString` function.",
+      "reference": "payload entered on command line is in hex-encoded format should use same processing as in ioctl/cmd/action/actiontransfer.go"
+    },
+    {
+      "prediction": "I think it would be better to create a new method `getPricingGroupByDomainIdAndName` in `PricingGroupFacade` and use it instead of `getByDomainId` and `reset`.",
+      "reference": "should this be kept in the `else` branch?"
+    },
+    {
+      "prediction": "Can you move this to the top of the file?",
+      "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 we need this anymore.",
+      "reference": "What is triggering this message in this file?"
+    },
+    {
+      "prediction": "Why is this change needed? I don't think it's related to this PR.",
+      "reference": "Are you sure this should be removed?"
+    },
+    {
+      "prediction": "I don't think this is a good idea to have a default value here. I think it should be `null` and we should check for `null` in the `createSocket` method.",
+      "reference": "What's that good for? The field is initialized in the constructor."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to fix the issue for me. I'm not sure if there's a better way to do this.",
+      "reference": "This shouldn't be necessary (same below). Do you have a traceback for me?"
+    },
+    {
+      "prediction": "nit: please remove the empty line.",
+      "reference": "No newline needed here."
+    },
+    {
+      "prediction": "nit: remove extra blank line",
+      "reference": "Why a separate block?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "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 don't think we need to pass `segwit` to this function, it's always going to be `False` for us.",
+      "reference": "`purpose` and `coin_type` and `account` would mimic the BIP-0044 wording, but it's up to you."
+    },
+    {
+      "prediction": "I think this should be `rbQValue01.Checked = true;` instead of `rbQValue01.Select();`",
+      "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 this should be 4, 2, and 16 respectively.",
+      "reference": "Seems a little low, but is this an installer default?"
+    },
+    {
+      "prediction": "I don't understand why this is needed. Can you explain a bit more?",
+      "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'm not sure if this is a good idea, but I think we should be able to get rid of `ParentWeightNum` and `ParentWeightDenom` and just use `ParentWeight`.",
+      "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'm not sure why this was changed. I don't think it should be removed.",
+      "reference": "Fixed the test; @strib want to validate that this is okay? Seems to be consistent new GC behavior."
+    },
+    {
+      "prediction": "Shouldn't this be `subdom` instead of `subdoom`?",
+      "reference": "Afaict, nothing cares"
+    },
+    {
+      "prediction": "I'm not sure this is true. I think we still need to re-embed the block changes in minimal mode, because we don't have a node cache, so we don't know if the block changes are already embedded or not.",
+      "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 `readonly` as well.",
+      "reference": "Are these changes from `class` to `struct` breaking if called from an older version of `Datadog.Trace.ClrProfiler.Managed`?"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "These are just aliases now"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do.",
+      "reference": "Does this mean we can remove `raw` from `_head_contents` partial?"
+    },
+    {
+      "prediction": "I think you can just use `!this.splashWindow.isDestroyed`",
+      "reference": "`isDestroyed` is a function, not a property. This condition will always return false."
+    },
+    {
+      "prediction": "I think this should be `Version_2_0_0` and `Version_2_1_0`",
+      "reference": "Why are we adding these versions?"
+    },
+    {
+      "prediction": "I think this should be `str(user.user_id)` instead of `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 if this is the best way to do this, but it seems to work.",
+      "reference": "Would this work, as a one liner? `setTimeout(window.location.reload, 5000);`"
+    },
+    {
+      "prediction": "I think we can remove the empty line here.",
+      "reference": "Can you move `crypto/tls` into the block at the top with all the other stdlib imports?"
+    },
+    {
+      "prediction": "I think this should be `inferenceElement='multiStepBestPredictions'`",
+      "reference": "What is the motivation for this change?"
+    },
+    {
+      "prediction": "I'm not sure I understand the difference between \"transient failure\" and \"service returns no messages for some other reason\". Can you give an example of each?",
+      "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 think this should be `if (matchClause->path()->pathSize() > 0) { return Status::SemanticError(\"Multi paths not supported.\"); }`",
+      "reference": "The data structure of `MatchClauseContext` may need to be redesigned."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it seems to work.",
+      "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": "Metrics/LineLength: Line is too long. [123/120]",
+      "reference": "how do we know that the step in question is an approval type step?"
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "JMX offers normal APIs for this. I don't think you want the glassfish one."
+    },
+    {
+      "prediction": "Why do we need to export `io.vavr.control`?",
+      "reference": "Is this still Java 8 compatible? @danieldietrich do we need a separate, modularized Java 9 release also?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the texture is not normalized, then the address mode should be CLAMP_TO_EDGE. If the texture is normalized, then the address mode should be CLAMP_TO_BORDER.",
+      "reference": "Should update the user input address and filter modes and pass those in fillSamplerDescriptor ?"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "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 `if task.supports_noop && executor.noop` so that we don't set `_noop` if the task doesn't support it.",
+      "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": "remove empty line",
+      "reference": "File is not `goimports`-ed"
+    },
+    {
+      "prediction": "Shouldn't this be a 403 as well?",
+      "reference": "why should unknown raise a 403?"
+    },
+    {
+      "prediction": "`c._nextState == c.state` should be `c._nextState === c.state`",
+      "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": "nit: missing space after `+`",
+      "reference": "nit: align the leftmost quotes"
+    },
+    {
+      "prediction": "I think this should be `ClusterMembershipPort` instead of `ClusterPort`.",
+      "reference": "Probably \"server port\" -> \"TCP port\""
+    },
+    {
+      "prediction": "I don't think we need to change the name of the codec here.",
+      "reference": "can you use FilterCodec instead?"
+    },
+    {
+      "prediction": "I think this should be `true` instead of `false`.",
+      "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": "This migration is no-op, but it forces the migration mechanism to apply it and thus write the DB schema version info.",
+      "reference": "I think this migration should fail for MDBX if it uses old table names"
+    },
+    {
+      "prediction": "Shouldn't this be `@options[:hosts_file].nil? || File.exist?(@options[:hosts_file])`?",
+      "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 think this should be `public static ISetupBuilder LoadConfiguration(this ISetupBuilder setupBuilder, bool applyOnReload = true, Action<ISetupLoadConfigurationBuilder> configBuilder)`",
+      "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 don't think we need to make this public.",
+      "reference": "This is used in o.a.l.sandbox.search.PhraseWildcardQuery."
+    },
+    {
+      "prediction": "I don't think we need to pass the protocol view to the stateDB. The stateDB should be able to get the protocol view from the blockchain.",
+      "reference": "view should be loaded on start for both statedb and factory"
+    },
+    {
+      "prediction": "I think this should be `private Task _readInputTask;` instead of `private Task _readInputTask;`",
+      "reference": "Why rename? Just curious..."
+    },
+    {
+      "prediction": "I don't think we need to do this here. The `onStopTrackingTouch` is called when the user releases the seekbar, so the controls should already be visible at that point.",
+      "reference": "Is this needed? I think it calls the super function that already does that."
+    },
+    {
+      "prediction": "Why do we need to repeat bboxes here?",
+      "reference": "This might slow down the inference time, are we sure we need to do that?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "reference": "This would read better if you add \"Skip code/lines\" or (exclude/hide code if you follow my earlier suggestion\")"
+    },
+    {
+      "prediction": "Looks like this is a copy-paste error.",
+      "reference": "Looks like glassfish went a bit nuts doing a refactor"
+    },
+    {
+      "prediction": "I think this should be `public override Task ForceFlushAsync(CancellationToken cancellationToken) => Task.CompletedTask;`",
+      "reference": "nit: allocate the list using the number of processors."
+    },
+    {
+      "prediction": "nit: `initializedClasses` -> `javaClasses`",
+      "reference": "Was this change triggered by baseline?"
+    },
+    {
+      "prediction": "`arel_table[:thirty_day_commit_count].gt(0).and(arel_table[:affiliate_count].gt(0))`",
+      "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": "Why is this needed? I don't think it should be enabled by default.",
+      "reference": "Setting browser based login as the default for `RestAPIExplorer`."
+    },
+    {
+      "prediction": "I don't think we need to resize the buffer every time. We can resize it once at the beginning of `FindBestSplitsFromHistograms` and reuse it.",
+      "reference": "maybe we can do this in Init function?"
+    },
+    {
+      "prediction": "I think this should be `errRequestIDTooLong` instead of `errRequestIDNotSet`.",
+      "reference": "How about if request ID not set, service should generate a random uuid?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `nil` arguments here.",
+      "reference": "minor: this is an opportunity for you to split these long lines into multiple lines :)"
+    },
+    {
+      "prediction": "why not `except Exception as e: raise TaurusConfigError(\"Wrong request:\\n %s\" % req) from e`?",
+      "reference": "It would be nice to have exception dumped somewhere (debug logs, info logs)."
+    },
+    {
+      "prediction": "I think this should be `MessageAs(m *Message, i interface{}) bool` instead of `MessageAs(m Message, i interface{}) bool`.",
+      "reference": "So I don't think this is right, as discussed on #657."
+    },
+    {
+      "prediction": "I think this should be `WHERE group_id = \"{0}\" and member_status = \"ACTIVE\"`",
+      "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 don't think it's a good idea to add this check here. I think it should be done in `get_full_path` instead.",
+      "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": "This should be `staticMethodOptions`",
+      "reference": "@bennothommo can you revert this section? It shouldn't be required."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Minor: maybe move the above to a `#fiscal_year` method?"
+    },
+    {
+      "prediction": "I think it would be good to have a test that checks the contents of the `build.yml` file to make sure it's what we expect it to be.",
+      "reference": "You should use the current directory instead - it's a temporary directory that will get removed when the test is done."
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I don't think it should have been.",
+      "reference": "I think we don't need synchronization here."
+    },
+    {
+      "prediction": "I don't think this is correct. If the user is not authenticated, then the `registrationId` will be `null` and the error message should reflect that.",
+      "reference": "I don't think this should be changed since on the reactive side we don't support anonymous users."
+    },
+    {
+      "prediction": "Style/UnlessElse: Favor unless over negated if with a negating operator.",
+      "reference": "Maybe this `if` should be on line 118?"
+    },
+    {
+      "prediction": "Why is this a float64 and not an int64?",
+      "reference": "Cool. (I thought I had fixed this already in #812.)"
+    },
+    {
+      "prediction": "I think this should be `interpreters: []` instead of `interpreters: nil`.",
+      "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 if this is the right thing to do, but it seems to work for me.",
+      "reference": "Did you file a bug to follow up on this?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `!batchReadsSparkSessionConf ? false :` part and just do `batchReadsEnabledTableProp`.",
+      "reference": "Looks like these 4 lines didn't actually change. Can you revert the whitespace changes here?"
+    },
+    {
+      "prediction": "Trailing whitespace detected.",
+      "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": "I think this should be a separate test, not part of this one.",
+      "reference": "It would be nice to fix the indentation above."
+    },
+    {
+      "prediction": "`short_names` is optional, so it should be `bool|null`.",
+      "reference": "`short_names` is still optional, so `=` that indicates it has to stay there."
+    },
+    {
+      "prediction": "I think this should be `appendNameOnlyAttribute(capabilities, attributePrefix+capabilityEFS)` instead of `appendNameOnlyAttribute(capabilities, attributePrefix+capabilityEFS)`",
+      "reference": "missing calling of this method"
+    },
+    {
+      "prediction": "nit: can we rename this to `archer` to be consistent with the rest of the package?",
+      "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 don't think we need to add `action` and `duration` here.",
+      "reference": "`StressChaos` does not contains a field .spec.action"
+    },
+    {
+      "prediction": "I think it would be good to add a test case for this as well.",
+      "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": "This macro is already defined in layer_validation_tests.h.",
+      "reference": "I this used anyplace?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Why was this change required?"
+    },
+    {
+      "prediction": "`missingAdminRole` -> `isMissingAdminRole`",
+      "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": "I think this should be `getSizePrefixedRootAsMonster(ByteBuffer _psbb, Monster obj) { ByteBuffer _bb = _psbb.slice(); _bb.order(ByteOrder.LITTLE_ENDIAN); _bb.position(4); return getRootAsMonster(_bb, obj); }`",
+      "reference": "Not sure why we're creating a new `ByteBuffer` here, ideally this refers to the existing one?"
+    },
+    {
+      "prediction": "I think we can remove `DATE_RANGE_OFFSET` from `constants.js` as it's only used here.",
+      "reference": "Same here re: `STORE_NAME` to `MODULES_SEARCH_CONSOLE`."
+    },
+    {
+      "prediction": "I think this should be `999` instead of `9999`.",
+      "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 this should be `KeyStoreManager` instead of `KeystoreManager`",
+      "reference": "This attribute was intentionally private. - lets force usage of factory `NewIdentityManager()` - lets ramake to `NewIdentityManager(keydir string)` -> `NewIdentityManager(keystore keystoreManager)`"
+    },
+    {
+      "prediction": "I think we should make this configurable.",
+      "reference": "Is this too frequent?"
+    },
+    {
+      "prediction": "I think this should be `['csv', 'zip']` to be consistent with the rest of the code.",
+      "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": "nit: please remove the extra blank line here.",
+      "reference": "Needs to be mentioned in `__all__` below."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be in signal.c, but I'm not sure if it should be public or not.",
+      "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": "It is important that the returned list is sorted in ascending order. -> It is important that 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 think we can remove this import.",
+      "reference": "Can you remove this blank line and put the added import in sorted order (likely your editor will do this)"
+    },
+    {
+      "prediction": "nitpick: `pthread_create` returns `0` on success, so this should be `if (err != 0) { ... } else { ... }`",
+      "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 it would be better to move these constants to `constants.js` and export them from there. That way they can be used in other files as well.",
+      "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": "I think this should be called `localeCompare` instead of `compareStrings` since that's what it is doing.",
+      "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 `strings.SplitN(DbCfg.Host, \":\", 2)` instead of `strings.Contains(DbCfg.Host, \":\") && !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 this is the right place to put this flag. I think it would be better to have it in the `ServerOptions` struct.",
+      "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 `failed to unmarshal session config`",
+      "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 we should keep `asset` as the flag name, and just update the help text to say \"Unit name of the asset to look up\"",
+      "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'm not sure this is the right way to do this, but it seems to work for me.",
+      "reference": "Which windows gets activated? A random other window?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `Send` method of the transport.",
+      "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 this should be `long WATERMARK_VALUE_DEFAULT = -1L;`",
+      "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 think this should be `||` instead of `!= null`",
+      "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.encode('utf-8'))`",
+      "reference": "Do we still support handshakes without SNI present?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we should be using `warn` here. I think we should be using `logger.warn` instead.",
+      "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 think this should be \"it has no sense in python\"",
+      "reference": "it have no sense in python -> it makes no sense in python"
+    },
+    {
+      "prediction": "I don't think we need to add this method to the interface. It's only used in one place, so we can just add it to the implementation of `ServicePortSupplier` in this file.",
+      "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 right thing to do. I think we should be able to call `topology.close()` even if `topology` is `null` or `undefined`.",
+      "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": "Use the new Ruby 1.9 hash syntax.",
+      "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 think we should keep the `DriverName` and `AliasName` in the `DataBase` struct.",
+      "reference": "We need to do the same for edgecore"
+    },
+    {
+      "prediction": "Can we use `Task.Delay` instead of `AutoResetEvent`?",
+      "reference": "This fix has nothing to do with the product change. Just a fix I am making for the test issue."
+    },
+    {
+      "prediction": "I don't think we should do this yet. We need to make sure that the BatchingActivityProcessor is ready for production use before we enable it by default.",
+      "reference": "Don't know what should be the default. Lets keep BatchingProcessor for now. And revisit the area after beta."
+    },
+    {
+      "prediction": "I think we can remove the `require 'rspec/support/reentrant_mutex'` from `lib/rspec/core/example_group.rb` and `lib/rspec/core/example.rb` as well.",
+      "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 should be done in the `RegisterNibForCellReuse` method of the `MvxSimpleTableViewSource` class.",
+      "reference": "What happens if there is no NIB/XIB with that name?"
+    },
+    {
+      "prediction": "I think we should rename `cfg` to `apiCfg` to make it clear that this is the config for the API server.",
+      "reference": "this is not necessary?"
+    },
+    {
+      "prediction": "nit: I think it's better to put asmsecretres before ssmsecretres.",
+      "reference": "typo: `asmsecrets \"github...`"
+    },
+    {
+      "prediction": "Shouldn't this be `wp_json_encode( $decoded )`?",
+      "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": "I don't think we need this method.",
+      "reference": "Should we put a TODO here to route the http.Transport through here?"
+    },
+    {
+      "prediction": "Why do we need to specify the computer name here?",
+      "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 this should be `if err != nil && !allowError {` instead of `if err != nil && allowError {`",
+      "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 don't think this is the right place to put this error. I think it should be in the `xrc20Contract` function.",
+      "reference": "return statements should not be cuddled if block has more than two lines (from `wsl`)"
+    },
+    {
+      "prediction": "I don't think we need to implement `Metric` for this class.",
+      "reference": "Why does VirtualMachineMetrics have to implement Metric?"
+    },
+    {
+      "prediction": "I think this test should be updated to use the new `SignVote` method.",
+      "reference": "line is 121 characters"
+    },
+    {
+      "prediction": "I think this should be `name[0] === 'o' && name[1] === 'n' && name[2] !== 'c'`",
+      "reference": "How's that linked esbench looking with slice?"
+    },
+    {
+      "prediction": "I'm not sure this is correct. `nlst` returns the list of files in the given directory, not the list of directories in the given directory. If you want to get the list of directories in the given directory, you should use `listdir` instead of `nlst`.",
+      "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 think this should be `return sessions.size();`",
+      "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 `s.maybeResetTimer(time.Time{})` instead of `s.maybeResetTimer(pacingDeadline)`.",
+      "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 about this change. I think it should be `std::unique_ptr<ROMol> res{te.canonicalize(std::move(*mol))};` or `std::unique_ptr<ROMol> res{te.canonicalize(std::move(mol))};`",
+      "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 think this should be `PLUGINS` instead of `PLUGIN`",
+      "reference": "why change the variable name at all?"
+    },
+    {
+      "prediction": "I think this should be `if (vNode.actualNode === node) { return vNode; }` instead of `if (vNode.actualNode === node) { return vNode; }`",
+      "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 we need this try/catch. If `mAccount` is null, then `mAccount.getReportSpamRecipient()` will throw an NPE.",
+      "reference": "Prefer TextUtils.isEmpty() which handles getReportSpamRecipient() being null"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to return `nil` if we can't find the extension id, rather than raising an exception.",
+      "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 think this should be `d.GetValues(\"Include\")` and `d.GetValues(\"Exclude\")` instead of `d.GetValueList(\"Include\")` and `d.GetValueList(\"Exclude\")`.",
+      "reference": "Do we actually use Include anywhere yet or is this just for completion?"
+    },
+    {
+      "prediction": "I think this should be `SettingValue(typ.Referer(none_ok=True), 'same-domain')`",
+      "reference": "It still bugs me this was misspelled in the standard and now the wrong spelling is the commonly used one :wink:"
+    },
+    {
+      "prediction": "I'm not sure why this change is necessary.",
+      "reference": "seems we missed de-uglifying this one"
+    },
+    {
+      "prediction": "I think this should be `NewWorkflow` instead of `NewWorkflowType`.",
+      "reference": "i guess these types & cron / retry specific belong to a dedicated util / struct"
+    },
+    {
+      "prediction": "I don't think we need to join the keys here. The server will do that for us.",
+      "reference": "Also update line 2205 below"
+    },
+    {
+      "prediction": "I think this should be `default: \"horizontal\"`",
+      "reference": "`direction` is referenced nowhere. Do I miss something?"
+    },
+    {
+      "prediction": "I think we can remove this now that we're not using `deparam` anymore.",
+      "reference": "The two places this was used have been removed so we can fully remove this dependency."
+    },
+    {
+      "prediction": "I don't think we need to generate this file anymore.",
+      "reference": "This is a duplicate generation of the same file as below."
+    },
+    {
+      "prediction": "`cfg.gpus` should be `len(gpu_ids)`",
+      "reference": "We may deprecate `gpus` if `gpu_ids` is specified."
+    },
+    {
+      "prediction": "I think this should be `finer` instead of `finest`.",
+      "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 what this is doing. Can you explain?",
+      "reference": "unnecessary spaces here are doubling the size of this otherwise extremely useful addition"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think we want to overwrite if newer, but I don't think we want to overwrite if newer if it's the same file. I think we want to overwrite if newer if it's the same file, but not overwrite if newer if it's a different file. I think we want to overwrite if newer if it's a different file, but not overwrite if newer if it's the same file. I think we want to overwrite if newer if it's the same file, but not overwrite if newer if it's a different file. I think we want to overwrite if newer if it's a different file, but not overwrite if newer if it's the same file. I think we want to overwrite if newer if it's a different file, but not overwrite if newer if it's the same file. I think we want to overwrite if newer if it's a different file, but not overwrite if newer if it's the same file. I think we want to overwrite if newer if it's a different file, but not overwrite if newer if it's the same file. I think we want to overwrite if newer if it's a different file, but not overwrite if newer if it's the same file.",
+      "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 `test3_6`",
+      "reference": "this should probably be removed"
+    },
+    {
+      "prediction": "I'm not sure why this change is necessary. I don't see any changes to the output of `json.MarshalIndent` in the diff.",
+      "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 think this should be a `map[SecurityGroupRole]map[string]IngressRules` so that we can have multiple ingress rules for each role.",
+      "reference": "Will need to think about this one. `additionalIngressRules` feels a bit opaque in terms of eventual outcome."
+    },
+    {
+      "prediction": "I think this should be `private static int fieldId(TypeDescription orcType) {`",
+      "reference": "Can we avoid making this public with the other updates?"
+    },
+    {
+      "prediction": "I think this should be `CHECK_GE` instead of `CHECK_EQ`.",
+      "reference": "Why did you change this equal instead of less or equal?"
+    },
+    {
+      "prediction": "I don't think we need to pass `blockHash` to `isWorldStateAvailable` and `getMutable`",
+      "reference": "missing final for the blockHash field"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [121/120]",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "`expand_as` should be FP32 as well.",
+      "reference": "`x_float` instead of `float_x`."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Is this for `No Babel config ...` from ESLint? If so, it will be fixed in my ES6 PR."
+    },
+    {
+      "prediction": "I think this should be `t.Fatal(err)`, not `t.Fatal(err.Error())`.",
+      "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": "nit: I think this should be `import \"github.com/docker/docker/api/types\"` instead of `import \"github.com/docker/docker/api/types\"`",
+      "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": "nit: I think this should be `LOG(THREAD, LOG_SYSCALLS|LOG_THREADS, IF_DGCDIAG_ELSE(1, 2), \"syscall: NtSetContextThread handle=\"PFX\" tid=%d cxt->Xip=\"PFX\" flags=\"PFX\"\\n\", thread_handle, tid, cxt->CXT_XIP, cxt->ContextFlags);`",
+      "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 endpoint. We already have the total count of products in the `Search` object returned by the `getProducts` endpoint.",
+      "reference": "@huzaifaiftikhar Changed the endpoint as suggested by Stephane in the latest commit."
+    },
+    {
+      "prediction": "I don't think this is a good idea. We want to keep the size of `Value` as small as possible, so that we can store lots of them in memory and send them over the network. If we add a `Number` to `Value`, we'll double the size of `Value`.",
+      "reference": "There's no core.Number yet. ;)"
+    },
+    {
+      "prediction": "I think this should be `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 lost. This command should be used cautiously and only when there is an issue with the current state of the data.`",
+      "reference": "..., ...will be posted."
+    },
+    {
+      "prediction": "I think this should be `go-filecoin show block --header`",
+      "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 `os.getenv('QUILT_USE_TQDM', 'false').lower() == 'true'`",
+      "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": "nit: `time.Millisecond * waitTimeoutMillis` can be simplified to `time.Duration(waitTimeoutMillis) * time.Millisecond`",
+      "reference": "unrelated but just changing for consistency with code below it"
+    },
+    {
+      "prediction": "I'm not sure about this change. If `shouldComponentUpdate` returns `false` then we should skip the render, right?",
+      "reference": "This will call `componentWillUpdate()` for mounts because the else clause is removed. It should only be called for updates."
+    },
+    {
+      "prediction": "Shouldn't this be `read` instead of `write`?",
+      "reference": "We should add write inheritance too."
+    },
+    {
+      "prediction": "I think this should be `JAVA_OPTS = os.environ.get('JAVA_OPTS', '').strip().split(' ')`",
+      "reference": "Please rename this to `LAMBDA_JAVA_OPTS`, and add a short description to the README."
+    },
+    {
+      "prediction": "I think we should use `github.com/google/uuid` instead of `github.com/satori/go.uuid`.",
+      "reference": "Should be in the same import group."
+    },
+    {
+      "prediction": "I don't think we need the `or 0x0303` here. I think we can just use `pkt.tls_session.tls_version > 0x0200`.",
+      "reference": "Operator precedence is very confusing here. Care to add parentheses? ` ((version or 0x0303) >= 0x0200)` Same below."
+    },
+    {
+      "prediction": "Please remove this line.",
+      "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 think it would be better to use a `Set` instead of an array here.",
+      "reference": "Why not just retest all? it's not expensive (I think)."
+    },
+    {
+      "prediction": "I don't think we need to change the import order here.",
+      "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 don't think we need to use `gdebi-core` here. `gdebi` should be enough.",
+      "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 think we should throw if traceparentCollection is null. I don't think it's possible for it to be null.",
+      "reference": "I don't understand why this change is in this PR?"
+    },
+    {
+      "prediction": "I think this should be `blacklight 7.0` since we're deprecating this in 7.0 and removing it in 8.0.",
+      "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 think you can just use `PersonObject.__proto__ = Realm.Object.__proto__` instead of `PersonObject.__proto__ = Realm.Object.__proto__.__proto__`.",
+      "reference": "The right-hand part shouldn't use `__proto__`. Should be just `PersonObject.__proto__ = Realm.Object`."
+    },
+    {
+      "prediction": "I think we should return an error here instead of panicking.",
+      "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 `memcpy(new_p, p, old_size < new_size ? old_size : new_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 don't think this is necessary.",
+      "reference": "This is the only one that worries me a bit. It would be good if Brendan signed off on it."
+    },
+    {
+      "prediction": "I think this should be called `isLongRunningOperation` to be consistent with `isLongrunningOperation` above.",
+      "reference": "`LongRunningOperationDetailView` already has `isEmptyOperation`."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `base64.b64encode` function returns a `bytes` object, not a `str` object, so you need to decode it to a `str` before you can concatenate it with the `Basic ` string.",
+      "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 double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I believe `Base.` is unnecessary in this case, as the class already extends `Base`."
+    },
+    {
+      "prediction": "`fillWidth` and `fillHeight` should be `maxWidth` and `maxHeight`.",
+      "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 it would be better to use `plugins.getConfig(\"api\", \"offline_mode\")` instead of `plugins.getConfig(\"api\").offline_mode`",
+      "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 `isset($config->Catalog->shibboleth_library_cards) && $config->Catalog->shibboleth_library_cards && ($this->getAuthManager()->getAuthMethod() == 'Shibboleth')`",
+      "reference": "You can combine the two checks into a single `!empty()` check."
+    },
+    {
+      "prediction": "I think this should be `>=` instead of `>`.",
+      "reference": "Should we add a warning for anything over say 8M? Just suggesting that is not recommended?"
+    },
+    {
+      "prediction": "I think you can remove the `currentUID` arg now.",
+      "reference": "Looks like there are no callers left of this function."
+    },
+    {
+      "prediction": "I think this should be `attachStdin bool` instead of `attachStdin, _ bool`",
+      "reference": "Should there still be a ` bool` trailing `showOutput`?"
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "Did we can change to v1"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think we need to set `_ancestorComponent` to `null` at the end of the `outer` block. Otherwise, if we have a `Fragment` with a `Suspense` inside of it, the `Suspense`'s `_ancestorComponent` will be the `Fragment`'s `_ancestorComponent`, which is not what we want.",
+      "reference": "Can you please double check that this won't break anything?"
+    },
+    {
+      "prediction": "I don't think it's a good idea to pass `null` here. I think it's better to create a new instance of `JsonRpcLocalStats` and pass it to the `JsonRpcService` and `JsonRpcProcessor`.",
+      "reference": "optionals, shouldn't need to be stated explicitly"
+    },
+    {
+      "prediction": "I don't think this is correct. 0x51303030 is 0x51303030, 0x51303439 is 0x51303439, and 0x51303439 - 0x51303030 = 0x00000400.",
+      "reference": "That seems quite low - why not just 0x5130ffff?"
+    },
+    {
+      "prediction": "I think this should be `__repr__` instead of `__str__`. `__str__` is supposed to return a user-friendly string representation of the object, while `__repr__` is supposed to return a string representation that can be used to recreate the object.",
+      "reference": "Why is this needed ?"
+    },
+    {
+      "prediction": "I think this should be `File.exist?(Bolt::Config.user_path) && !File.exist?(first_runs_free)`",
+      "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.CompoundMessage.Proposal.value().String()`",
+      "reference": "nit: use %s for strings and %v for objects."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "Same as before, the properties already exists in parent"
+    },
+    {
+      "prediction": "`datetime.datetime.now().year` -> `datetime.datetime.now().strftime('%Y')`",
+      "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'm not sure if this is the right thing to do, but I think it makes sense to ignore the \"gone\" event for the DRM device.",
+      "reference": "Maybe we should only set active = false if `strcmp(type, \"pause\") == 0`?"
+    },
+    {
+      "prediction": "I think this should be `options[:page_ranges] = Array(options[:page_ranges])` instead of `options[:page_ranges] &&= Array(options[:page_ranges])`",
+      "reference": "Same here. and all others."
+    },
+    {
+      "prediction": "I think this should be in the `libkbfs` package, not here.",
+      "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 don't think this is correct. If `PParam.prefix` is `c_GuidPrefix_Unknown`, then `guidP` will be `c_GuidPrefix_Unknown` as well, and the `if (guidP == c_GuidPrefix_Unknown)` condition will be true.",
+      "reference": "I think this blank line may be removed"
+    },
+    {
+      "prediction": "I think it would be better to raise an error if `toggle` is True and the bookmark 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": "This should be `UINT32 DevStatus;` instead of `UINT32      DevStatus;`",
+      "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": "Style/RedundantSelf: Redundant self detected.",
+      "reference": "How about YAML.load_file()"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "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": "I don't think we need to unregister the module unload event.",
+      "reference": "Check the return value of the drmgr ones."
+    },
+    {
+      "prediction": "I don't think we need to add `data-cfasync` here.",
+      "reference": "This should be `'cache'`"
+    },
+    {
+      "prediction": "I think this should be `(*os.File, error)`",
+      "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": "This should be `command.readPreference = resolveReadPreference(options, { db: this.s.db, collection: this });`",
+      "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 `to-1` instead of `(to + s.BlockNumber) / 2`",
+      "reference": "During genesis sync it can unwind 5M blocks?"
+    },
+    {
+      "prediction": "I think we should keep this test, but just remove the `Skip`",
+      "reference": "was it caused by the removal of removeTaskQueueManager() from this test?"
+    },
+    {
+      "prediction": "Line is too long. [114/80]",
+      "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 think this should be `if (!rp->rr_sim_misses_deadline) return false;`",
+      "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 `if (id == 0 || !Item::items[id].blockPickup) {`",
+      "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 throw an exception in this case. If the item doesn't have the `clientCert` attribute, then it's not a valid item and shouldn't be processed.",
+      "reference": "If clientCert attribute doesn't exist for some reason I set it to false."
+    },
+    {
+      "prediction": "I think this should be `app_host_flags += ' -k'`",
+      "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": "I don't think this should be here. It's already done in `onWalkComplete()`",
+      "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 don't think this is the right place to do this. This is a joiner, not a decryptor. The decryptor should be responsible for decrypting the root chunk. The joiner should only be responsible for joining the chunks together.",
+      "reference": "typo in `weather` (should be `whether`)"
+    },
+    {
+      "prediction": "I think this should be `--use-staging-install-media` instead of `--use-staging-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": "Why not just remove this line?",
+      "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": "Why do we need to specify the port here?",
+      "reference": "Could we set a default value for `partitions` and `{0, network::NetworkUtils::getAvailablePort()}` ?"
+    },
+    {
+      "prediction": "I think this should be `public TimeSpan MinimumTime { get; set; } = TimeSpan.FromSeconds(30);`",
+      "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'm not sure if this is the best way to do this, but it works for now.",
+      "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 think this should be `code += \"}\";` instead of `code += \"}\";`",
+      "reference": "This change is due to `clang_format` and is not related to this PR."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in `sigprocmask_syscall` itself, so that we don't have to remember to do it in every place we call `sigprocmask_syscall`, and so that we don't have to remember to do it in every place we call `sigprocmask_syscall`, and so that we don't have to remember to do it in every place we call `sigprocmask_syscall`, and so that we don't have to remember to do it in every place we call `sigprocmask_syscall`, and so that we don't have to remember to do it in every place we call `sigprocmask_syscall`, and so that we don't have to remember to do it in every place we call `sigprocmask_syscall`, and so that we don't have to remember to do it in every place we call `sigprocmask_syscall`, and so that we don't have to remember to do it in every place we call `sigprocmask_syscall`, and so that we don't have to remember to do it in every place we call `sigprocmask_syscall",
+      "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 think this should be `FeedFile` instead of `FeedFileDownloadStatusRequesterInterface`.",
+      "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": "nit: `content` -> `contentFile`",
+      "reference": "why not just use `ManifestContent` instead of `Integer`?"
+    },
+    {
+      "prediction": "I think it would be better to move this check to the `IsWellFormed` method.",
+      "reference": "can we do that in TxValidator?"
+    },
+    {
+      "prediction": "Why do we need this field?",
+      "reference": "We definitely want some detailed and thoughtful docs being added to this type."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be able to enable XDP even if BPF is not enabled.",
+      "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": "`if (spaceRet.left() == ResultCode::ERR_SPACE_NOT_FOUND)` -> `if (spaceRet.left() == ResultCode::ERR_SPACE_NOT_FOUND || spaceRet.left() == ResultCode::SUCCEEDED)`",
+      "reference": "If the status is `ERR_SPACE_NOT_FOUND` why return `SUCCEEDED` ?"
+    },
+    {
+      "prediction": "`reverse (bool): Reverse the order of planes in layer sequence. False for ResNet, True for Houglass. Default: False` -> `reverse (bool): Reverse the order of planes in layer sequence. False for ResNet, True for Hourglass. Default: False`",
+      "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": "nit: remove the empty line",
+      "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 it would be better to say \"Instead of using `exit()` or `quit()`, consider using the `sys.exit()` function.\"",
+      "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'm not sure if this is the right thing to do, but I think we should be using `crt.Spec.CommonName` instead of `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 this is necessary. `cmdhistory.init()` should be called in `conftest.py`.",
+      "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 `getint(self.spark_version, \"pickle-protocol\", pickle.HIGHEST_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'm not sure if this is the best way to test this, but I think it will do for now.",
+      "reference": "nit: Think it can be one line.."
+    },
+    {
+      "prediction": "I don't think we should use `import *` in this case.",
+      "reference": "please inline imports"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be able to join cgroups even if we're in a rootless container.",
+      "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 think this should be `aggregates changed to operate on just a single column.`",
+      "reference": "Should we update integral to operate on a single column as well?"
+    },
+    {
+      "prediction": "I think this should be `cuda_get_deep_copy_stream(instance)`.",
+      "reference": "Doesn't belong to this PR"
+    },
+    {
+      "prediction": "I think this should be `registrationToken *string` so that we can pass in `nil` instead of an empty string.",
+      "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 a constant in the `UserAccount` class.",
+      "reference": "Could we shorten this constant to maybe `FEATURE_USER_AUTH`?"
+    },
+    {
+      "prediction": "I think this should be `numRekeyQueueWorkers` or something like that.",
+      "reference": "Any particular reason for 8? I feel like we could probably handle more..."
+    },
+    {
+      "prediction": "Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.",
+      "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 this is the best way to do this, but it's the only way I could think of to add a class to the `status-header-container`",
+      "reference": "Should the \"cancelled\" class be added regardless of the proposal's state?"
+    },
+    {
+      "prediction": "I think we should keep the empty line here.",
+      "reference": "nit: random whitespace change, let's try to keep these out of commits unless there are other useful changes."
+    },
+    {
+      "prediction": "I don't think we need to do this. If `model` is a string, then `models` will be `undefined` and `i < models.length` will always be `false` so `models[i]` will always be `undefined`.",
+      "reference": "Seems kinda dangerous - what if models length is different from paths length?"
+    },
+    {
+      "prediction": "I don't think we need to pass `prune.Mode{}` here.",
+      "reference": "set real one plz (get it from DB)."
+    },
+    {
+      "prediction": "I don't think you need to cast this to (struct fpga_port_uafu_irq_set *). The compiler should be able to figure this out.",
+      "reference": "Should this be initialized to zeroes?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do. If there are no base docs, then there are no facet results.",
+      "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": "This import is not used.",
+      "reference": "re-add the type in the annotation here"
+    },
+    {
+      "prediction": "`stderr.String()` should be `string(stderr.Bytes())` as `stderr` is a `bytes.Buffer` and not a `string`.",
+      "reference": "Could we simpify it as `cmd.Stderr = &bytes.Buffer{}`?"
+    },
+    {
+      "prediction": "I think this should be `return now.After(cert.NotAfter)`",
+      "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 think this should be `github.com/fortio/fortio/log`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be 1e-6",
+      "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'm not sure if this is the best way to test this, but it seems to work.",
+      "reference": "Be careful. This is not a preference but an option in a dialog."
+    },
+    {
+      "prediction": "I think this should be `if deployment.ClusterName == \"\" { deployment.ClusterName = cluster } else { deployment.ClusterName = deployment.ClusterName }`",
+      "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 not needed anymore?",
+      "reference": "Is this not required?"
+    },
+    {
+      "prediction": "I think this should be `if (input.type === Blockly.INPUT_VALUE || input.type === Blockly.INPUT_STATEMENT)`",
+      "reference": "You're using scare quotes on these terms instead of defining them."
+    },
+    {
+      "prediction": "This should be `>= 1.12.0`",
+      "reference": "These should both be const, not var right?"
+    },
+    {
+      "prediction": "I don't think you need the `try-with-resources` here. The `FileOutputStream` is closed by the `persister.store` method.",
+      "reference": "Can you catch a more specific exception here? Thanks :-)"
+    },
+    {
+      "prediction": "Why do we need to return `ctrl.Result{Requeue: true}` here?",
+      "reference": "This is a bug during handling the recover logic in `common chaos` controller, I fixed in this request when I found it."
+    },
+    {
+      "prediction": "I think it is better to add `score_thr` as an argument of `single_gpu_test` and `multi_gpu_test` instead of adding it as an argument of `inference_detector`.",
+      "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 we should keep the `role=\"navigation\"` on the `<nav>` element.",
+      "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 `replace(/[^a-zA-Z0-9-]+/g, '')` instead of `replace(/[^\\w\\-]+/g, '')`",
+      "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 right thing to do, but I don't have a better idea.",
+      "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 think this should be `(this.flyoutWidth_ - statusButtonWidth - marginX) / this.workspace_.scale - marginX`",
+      "reference": "Why are you dividing by scale in the LTR case but not the RTL case?"
+    },
+    {
+      "prediction": "This should be `Command.W3C_MINIMIZE_WINDOW`",
+      "reference": "Update after command rename"
+    },
+    {
+      "prediction": "I think this should be `googlesitekit.ModuleSettingsDetails-${ moduleSlug }` instead of `googlesitekit.SettingsLegacy`.",
+      "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 don't think we need this constant. We can just use `2.0` in the code.",
+      "reference": "isn't AZKABAN_FLOW_VERSION_2_0 more explicit?"
+    },
+    {
+      "prediction": "Why do we need to remove the default case?",
+      "reference": "Please don't leave an empty default"
+    },
+    {
+      "prediction": "Can you 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 think this should be `sh.cmd \"brew update >/dev/null\", retry: true`",
+      "reference": "You can also use `echo: false` instead. Either is fine; I'm just pointing it out."
+    },
+    {
+      "prediction": "I think this should be `base.InitializeLayout();` before `switch (Delimiter) { case CsvColumnDelimiterMode.Auto:`",
+      "reference": "what's the reason for this moved line?"
+    },
+    {
+      "prediction": "I think this should be `if (overlayText && showTitle && item.Name) {`",
+      "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": "Can you explain why this change is needed?",
+      "reference": "`s => s.IsHidden == false` What's the purpose of adding this?"
+    },
+    {
+      "prediction": "I think this should be `optionNameClefSignerAddress` instead of `optionNameClefSignerAddress`",
+      "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": "Why is this skipped?",
+      "reference": "you can just remove this"
+    },
+    {
+      "prediction": "I think we need to add a check here to see if the SPC is already upgraded or not. If it is already upgraded then we can skip this 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 if this is the right place to put this, but I think it would be nice to have a section for locked features, so that users know what they're missing out on.",
+      "reference": "Honestly not sure myself, but do you think it makes sense to move this conditional into the partial?"
+    },
+    {
+      "prediction": "I don't think this is necessary. `mkdtemp` already creates a unique directory.",
+      "reference": "As discussed in IRC (just so it doesn't get lost): This probably should be `-basedir-`, not `-prefix-`"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "basepath can be absoluta path and this needs to be supported"
+    },
+    {
+      "prediction": "This should be `ui->comboBox_position->addItem(name, QVariant::fromValue(ScreenPosition{screen, position}));`",
+      "reference": "Didn't get to the commit message"
+    },
+    {
+      "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": "What about raising `ActionView::MissingTemplate` instead? That's what HighVoltage does."
+    },
+    {
+      "prediction": "This should be `response_l.put (\"pending\", balance.second.convert_to<std::string> ());`",
+      "reference": "Is it kept for compatibility?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we should be using the worker address as the worker's signing address. I think we should be using the worker's account address as the worker's signing address. @whyrusleeping thoughts?",
+      "reference": "I would find it pretty reasonable to add a MinerSigner method on the state view that puts these together."
+    },
+    {
+      "prediction": "This should be 1090001",
+      "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 `error executing CQL statement: %v`",
+      "reference": "What about method name itself? Do we run this for SQL too?"
+    },
+    {
+      "prediction": "This should be `instanceUrl = \"https://mobilesdk.cs1.salesforce.com\";`",
+      "reference": "`instanceUrl` should be `cs1.salesforce.com`. `communityUrl` would be `mobilesdk.cs1.my.salesforce.com`."
+    },
+    {
+      "prediction": "Why is this needed? I don't see how `name` can be `nullptr` at this point.",
+      "reference": "Let's use `if(name != nullptr)` to remove ambiguity"
+    },
+    {
+      "prediction": "I don't think we need to implement these in this PR. We can do that in a follow-up.",
+      "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 a single constructor with a default value for `extraDataValidatorsAndVoteMustBeEmpty` instead of 2 constructors.",
+      "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 should be a default method. I think it should be an abstract method that is implemented in `LinkedSet` and `SortedSet` and throws an `UnsupportedOperationException` in `Set`.",
+      "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": "`response_headers` should be a `dict` of `str` to `str`",
+      "reference": "we can replace `for` with `params.update(response_headers)`"
+    },
+    {
+      "prediction": "I think we can remove this method entirely now.",
+      "reference": "Any chance this is available in .NET Core 2.0 now?"
+    },
+    {
+      "prediction": "remove empty line",
+      "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": "I think this should be `#include \"lbann/execution_contexts/execution_context.hpp\"` and `#include \"lbann/execution_contexts/sgd_execution_context.hpp\"`",
+      "reference": "Why do we need both includes here?"
+    },
+    {
+      "prediction": "I think this should be `SOUP_NAME_COL + \"=?\", new String[]{SOUP_NAME_COL}`",
+      "reference": "SmartStoreInspectorTest expected results in a certain order"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "I think that you need something like `m_using_random_node.emplace(m_io_thread_pool->get_local_thread_id());`"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. I think it should be done in the `isIgnorablePrefixIncrement` method. The reason is that the `isIgnorablePrefixIncrement` method is used in the `UnusedAssignmentRule` and in the `UnusedVariableRule`. So, if we move this check to the `isIgnorablePrefixIncrement` method, we will be able to reuse the same logic in the `UnusedVariableRule` as well.",
+      "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 think this should be `long targetSplitSizeInBytes()`.",
+      "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 think this should be `tm[:maxTrace+1]` instead of `tm[1:maxTrace+1]`",
+      "reference": "Not sure what is this change doing?"
+    },
+    {
+      "prediction": "90 days? That's a long time for a token to be valid.",
+      "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 don't think this is the right fix. The `key` should never be `null`. If the `key` is `null` then the `RememberMeServices` should be `null` as well. If the `key` is `null` then the `RememberMeServices` should be `null` as well. If the `key` is `null` then the `RememberMeServices` should be `null` as well. If the `key` is `null` then the `RememberMeServices` should be `null` as well. If the `key` is `null` then the `RememberMeServices` should be `null` as well. If the `key` is `null` then the `RememberMeServices` should be `null` as well. If the `key` is `null` then the `RememberMeServices` should be `null` as well. If the `key` is `null` then the `RememberMeServices` should be `null` as well. If the `key` is `null` then the `RememberMeServices` should be `null` as well. If the `key` is `null` then the `RememberMeServices` should be `null",
+      "reference": "Please use a tab for indentation instead of spaces."
+    },
+    {
+      "prediction": "`AbstractsQueue` should be renamed to `AbstractQueue`",
+      "reference": "I can't believe we didn't see this typo before :))"
+    },
+    {
+      "prediction": "This should be `from google.cloud.forseti.common.util import logger` and `LOGGER = logger.get_logger(__name__)`",
+      "reference": "If the logger isn't used, it probably doesn't need to be added."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "reference": "These files should in `/javascript/brython`"
+    },
+    {
+      "prediction": "I think this should be `if imageAlpha != nil && imageAlpha.Deprecated != nil && imageAlpha.Deprecated.StateOverride != nil && len(imageAlpha.Deprecated.StateOverride.DefaultRolloutTime) > 0 {`",
+      "reference": "minor, you can squash these into one `if`"
+    },
+    {
+      "prediction": "I think we should use `uuid.NewV4().String()` instead of `primitives.NewUUID().String()`.",
+      "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": "Why is this needed?",
+      "reference": "Add a TODO here to revisit if this was problematic."
+    },
+    {
+      "prediction": "I think it's better to put this in the docstring of `bbox_coder` instead of `reg_decoded_bbox`.",
+      "reference": "Note generally it -> It"
+    },
+    {
+      "prediction": "I think this should be `#if 0` instead of `#if 0`.",
+      "reference": "Is this relevant to this PR?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [123/120]",
+      "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 don't think these are used anywhere.",
+      "reference": "Is this just moved out of the other files? It doesn't seem like this was generated like the other files."
+    },
+    {
+      "prediction": "I think this should be `IHttpContent<TStream>` where `TStream` is `Stream` or `Task<Stream>`.",
+      "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 it's a good idea to pass `LogManager` to the `MevBlockProducerTransactionsExecutorFactory` constructor. I think it would be better to pass `LogManager` to the `MevBlockProducerTransactionsExecutorFactory` constructor and then pass `LogManager` to the `MevBlockProducerTransactionsExecutor` constructor.",
+      "reference": "this looks already too complicated..."
+    },
+    {
+      "prediction": "I think it's better to add a test case to check whether the compaction is successful or not. For example, you can get the number of files before and after the compaction, and then check whether the number of files is reduced after the compaction.",
+      "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 it's better to use `tableMap.values().stream().filter(TiTableInfo::isView).collect(Collectors.toList())` instead of `tableMap.values().removeIf(TiTableInfo::isView)`.",
+      "reference": "should we add a TODO here?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the `ErrorHandler` component around, and just move it into the `DashboardDetailsApp` component. That way, we don't have to wrap every single component in the app in an `ErrorHandler`. What do you think?",
+      "reference": "This can be inlined below as `GoogleSitekitDashboardDetails` is an unnecessary wrapper now."
+    },
+    {
+      "prediction": "I don't think we need to create a new `VersionHistories` here. We can just use `histories` from line 195.",
+      "reference": "NewVHS -> NewVersionHistories this one looks better"
+    },
+    {
+      "prediction": "I don't think we need to use `self` here. `attr_accessor :executable, :missing_text` should be enough.",
+      "reference": "Is this `@missing_text` ?"
+    },
+    {
+      "prediction": "I think this should be `^([\\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 don't think we should use `this.props.i18n` in the `title` attribute of the button. The `title` attribute should be a string, not a function call. I think you can just use `this.props.i18n('finishEditingFile')` instead of `this.props.i18n('finishEditingFile')()`",
+      "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 don't think this check is necessary. If `pglob.gl_pathv` is `NULL` then `pglob.gl_pathc` will also be `0` and the loop will not execute.",
+      "reference": "Probably want break instead of continue."
+    },
+    {
+      "prediction": "I think this should be `o.NoLog = true`",
+      "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 don't understand why this change is needed.",
+      "reference": "does adding a `Stream.<String>empty()` help here?"
+    },
+    {
+      "prediction": "I think this should be `{'name': 'non_exist_scanner', 'enabled': False}`",
+      "reference": "This is not being used anymore, so can be removed."
+    },
+    {
+      "prediction": "I think this should be `if len(p.ipStrings) == 0 && len(p1.ipStrings) == 0 { return true }`",
+      "reference": "Is it called some where?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Please remove this extra blank line."
+    },
+    {
+      "prediction": "I think this should be `@Input` instead of `@InputFiles` since it's a list of configurations, not a list of files.",
+      "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 `rtest.Assert(t, !r, \"Failed to match untagged snapshot\")`",
+      "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": "Can we move this file to `Microsoft.TestPlatform.VsTestConsole.TranslationLayer.Payloads` namespace?",
+      "reference": "nit: please move these inside namespace."
+    },
+    {
+      "prediction": "I don't think we need to hide this flag.",
+      "reference": "may users try to abuse it?"
+    },
+    {
+      "prediction": "I'm not sure if this is correct. I don't think we can assume that the new buffer will have the same capacity as the old one. I think we need to get the limit of the new buffer and use that as the new buffer size.",
+      "reference": "why is this space removed?"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "Can we remove this other line now then?"
+    },
+    {
+      "prediction": "I think this should be moved to `qutebrowser/misc/` as well.",
+      "reference": "There's a comma missing here at the end :wink:"
+    },
+    {
+      "prediction": "Why is this no longer required?",
+      "reference": "Why is this not required any longer?"
+    },
+    {
+      "prediction": "Please add a blank line after `import` statements.",
+      "reference": "Indentation seems a bit fucked up here. Run `go fmt`."
+    },
+    {
+      "prediction": "I think we should also check `in.GetByBlock()` and `in.GetByTransaction()` to make sure they are not nil.",
+      "reference": "any chance in == nil? same below"
+    },
+    {
+      "prediction": "`h2o_sysfn(listen, listen_fd, H2O_SOMAXCONN) != 0` should be `h2o_sysfn(listen, listen_fd, H2O_SOMAXCONN) == -1`",
+      "reference": "We would change the source code of h2o so that certain syscalls will be invoked through the `h2o_sysfn` macro."
+    },
+    {
+      "prediction": "I think this should be `SubmitError::kSemCannotBeSignalled` instead of `SubmitError::kTimelineCannotBeSignalled`.",
+      "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": "`percentiles` should be `percentiles: Optional[List[float]] = None`",
+      "reference": "Could you add a type hint? `percentiles: Optional[List[float]] = None`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to go about it, but I think it's the best we can do for now.",
+      "reference": "Maybe putting ones they've used before first would be good, but this is fine for now."
+    },
+    {
+      "prediction": "nit: I think it would be nice to have a `routetable.BlackholeTarget` and `routetable.NoEncapTarget` instead of creating them every time.",
+      "reference": "I don't think we need these. Simpler just to put `routetable.TargetType...` inline below."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it works for now.",
+      "reference": "Why this instead of the old way?"
+    },
+    {
+      "prediction": "I don't think we need to log this.",
+      "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.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": "I'm not sure this is the right place to clear the playback speed. I think it should be cleared in `onPlaybackEnded` in `PlaybackService`.",
+      "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 don't think this is the right place to call xcb_flush()",
+      "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 be deleted.",
+      "reference": "This class can be removed?"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be `Logger *zap.Logger` instead of `*Logger *zap.Logger`",
+      "reference": "To minimize the possibility of regret, perhaps we name this `ZapLogger`."
+    },
+    {
+      "prediction": "This should be `Proc.new { |current_user| current_user.has_logged_in_to_forum? }`",
+      "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 think this should be `from nupic.frameworks.opf.anomaly import Anomaly`",
+      "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 seems to work.",
+      "reference": "We needed this testutils file after all"
+    },
+    {
+      "prediction": "Why is this changed?",
+      "reference": "line is 161 characters (from `lll`)"
+    },
+    {
+      "prediction": "I think it's better to use `chaos-mesh/chaos-mesh/pkg/controllerutils` instead of `chaos-mesh/chaos-mesh/pkg/controllerutils`",
+      "reference": "Please re-format/groupimport this line. You might need a little manual work."
+    },
+    {
+      "prediction": "I think this should be `worker.env.ApplyWorkerCustomizations(worker.wf, worker.env.NoExternalIP)`",
+      "reference": "minor, you can just use `wf` instead of `worker.wf`"
+    },
+    {
+      "prediction": "nit: can you please remove the extra line at the end of the string?",
+      "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 this should be `pkg/upgrade/v1alpha1/templates`",
+      "reference": "could not import github.com/openebs/maya/pkg/upgrade/templates (invalid package name: \"\") (from `typecheck`)"
+    },
+    {
+      "prediction": "I don't think we need to make this change. We can keep the test as it is.",
+      "reference": "a official -> an official"
+    },
+    {
+      "prediction": "I think we can remove the `and config.KINESIS_PROVIDER == 'kinesalite'` check here. The `if action == 'RegisterStreamConsumer'` check should be enough.",
+      "reference": "I only want the proxy request for this to run for kinesalite."
+    },
+    {
+      "prediction": "I think this should be `dateFormat: this.props.dateFormat || 'YYYY-MM-DD'`",
+      "reference": "White space boya?"
+    },
+    {
+      "prediction": "I think this should be `StackSummary` instead of `StackSummaryDescription`.",
+      "reference": "Maybe ditch these since `StackSummary` is not used anymore."
+    },
+    {
+      "prediction": "why remove `label_bias`?",
+      "reference": "suggest to indicate the value range & meaning in docstring"
+    },
+    {
+      "prediction": "Why do we need to add the trailing slash?",
+      "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.CleanupEndPointMark(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(privRegistries.Registry())`",
+      "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 this should be `doTRegex = /{{.+?}}/g;` instead of `doTRegex = /\\{\\{.+?\\}\\}/g;`",
+      "reference": "IMO `dot` should be preferred. Remember `aXe`?"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "Not a huge deal but this will block tests, also, don't they stay open by default now?"
+    },
+    {
+      "prediction": "I think this should be `cfg.LogHistoryDepth`",
+      "reference": "Could you make it a local constant for now ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `validate` method of the `Filter` class.",
+      "reference": "What is this `collection_id` field here?"
+    },
+    {
+      "prediction": "I think this should be `if not conf.interactive:`",
+      "reference": "Is there a reason why we don't use logging for this?"
+    },
+    {
+      "prediction": "I think it should be `DenormalizationException`",
+      "reference": "Exceptions should be placed in application layer -> infrastructure is aware of application - not the other way around"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to include the `message-id` in the JSON output.",
+      "reference": "IIRC issue mentioned reporting _symbolic message_ - so in this case it would be `line-too-long`."
+    },
+    {
+      "prediction": "This should be `else if (id == \"feedbackable\") { return spells[spell_id].feedbackable; }`",
+      "reference": "Is this intended to be `spells[spell_id].reflectable` here? Edit: Should be `spells[spell_id].feedbackable`, right?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "How should we set this for existing teams?"
+    },
+    {
+      "prediction": "I'm not sure this is a fatal error. I think it should be an error, but not fatal.",
+      "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 import.",
+      "reference": "I hate then some internal process specific middleware leaks into connection manager :( todo later."
+    },
+    {
+      "prediction": "I'm not sure what this is doing. I don't see any `nupic.support.configuration_base` or `nupic.support.__init__` packages.",
+      "reference": "Reflects where these files were moved for `pkg_resources`."
+    },
+    {
+      "prediction": "I'm not sure why this is needed, but it seems to fix the issue.",
+      "reference": "This should be computed('i18n.locale'...."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `BeforeAll` of the test suite.",
+      "reference": "Can you explain to me what goes wrong here? Can the test be adjusted to set up working routing instead?"
+    },
+    {
+      "prediction": "I don't think we need this import.",
+      "reference": "We can get rid of this import then."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `mapEventToMetricsEvent` function that returns a `map[string]*metrics.Event` where the key is the metric ID and the value is the `metrics.Event`. Then you can just loop over the map and call `SendMetric` for each key/value pair.",
+      "reference": "Not sure how useful is unlock event."
+    },
+    {
+      "prediction": "I think it would be better to pass the `lowercaseFields` flag to the `encodeDoc` function, rather than storing it in the `encoder` struct. That way, we don't need to create a new `encoder` for each call to `encodeDoc`.",
+      "reference": "Consider make the second argument a `encoderOptions` which includes the `lowercaseFields` just like opening a collection."
+    },
+    {
+      "prediction": "nit: I think we can remove the `ctx` parameter and just use `TezUtil.enrichContextWithAttemptWrapper(context)`.",
+      "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 don't think this is correct. If the first tag is uturn, then the second tag must be uturn as well. If the second tag is uturn, then the first tag must be uturn as well. If the first tag is uturn, then the second tag must be uturn as well. If the second tag is uturn, then the first tag must be uturn as well. If the first tag is uturn, then the second tag must be uturn as well. If the second tag is uturn, then the first tag must be uturn as well. If the first tag is uturn, then the second tag must be uturn as well. If the second tag is uturn, then the first tag must be uturn as well. If the first tag is uturn, then the second tag must be uturn as well. If the second tag is uturn, then the first tag must be uturn as well. If the first tag is uturn, then the second tag must be uturn as well. If the second tag is uturn, then the first tag must be uturn as well. If the first tag is uturn, then the second tag must be uturn as well. If",
+      "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 rename `success` to `Value` to be consistent with the rest of the codebase.",
+      "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 don't think we need ReadOnlyTransactionProcessorFactory here. We can just pass ReadOnlyDbProvider to the AuRaAdditionalBlockProcessorFactory.",
+      "reference": "we can just use read env here, pretty sure"
+    },
+    {
+      "prediction": "I think it would be better to use `afterTransformEnd` instead of `afterTransformStart`.",
+      "reference": "Could you replace local hook with global hook `afterModifyTransformStart`?"
+    },
+    {
+      "prediction": "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 `if (typeof opt_port === 'function')` instead of `if (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>> Attributes { get; set; }` instead of `IDictionary<string, object> Attributes { get; set; }`.",
+      "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 `finest` instead of `info`.",
+      "reference": "This change prevents a user understanding how their server is configured. Best to leave at `info` level."
+    },
+    {
+      "prediction": "I don't think we need to add these to the deny list. We should be able to remove them from the spec.",
+      "reference": "I will add the following to the list: - api.SpecScale - api.SpecSticky"
+    },
+    {
+      "prediction": "Why is this line commented out?",
+      "reference": "As defaultSendPort is being removed, and I don't like TODOs on examples, please remove the whole line"
+    },
+    {
+      "prediction": "Why did you remove this?",
+      "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 think this should be `Register(reg *Registry)` instead of `Register(nil)`.",
+      "reference": "It would be cleaner if you were to pack the name&desc in a `MetricName` and pass it to `MakeCounter`"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "Do you know why do we need this?"
+    },
+    {
+      "prediction": "I think this should be `using MvvmCross.iOS.Views.Presenters.Attributes;`",
+      "reference": "I guess this using is what causes the build to fail"
+    },
+    {
+      "prediction": "I think this should be done in `ponyint_asio_backend_stop` instead of in `ponyint_asio_backend_dispatch`.",
+      "reference": "I like the explicit unregister approach."
+    },
+    {
+      "prediction": "I don't think we need to reserve these. We can add them in the future if we need to.",
+      "reference": "We should able to use 0x30 as summary, and make 0x80 reserved for future."
+    },
+    {
+      "prediction": "I think we can remove the `|| HttpBypassHelper.UriContainsAnyOf(requestUri, tracer.Settings.HttpClientExcludedUrlSubstrings)` here.",
+      "reference": "This allows strategic exclusion of http spans."
+    },
+    {
+      "prediction": "`?array` should be `?string[]`",
+      "reference": "why not just `string[]` instead of nullable"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get `go list` to work with `-tags` that have spaces in them. Without this change, `go list -tags=foo bar` works, but `go list -tags=foo bar baz` does not.",
+      "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 `AzureAppServicesInstanceId` instead of `AzureAppServicesResourceId`",
+      "reference": "This key is actually yet to be determined. Meeting with the backend team and Garner to discuss."
+    },
+    {
+      "prediction": "I think this assert should be `if (iv_state == nullptr) { ... }` instead of `assert(iv_state);`",
+      "reference": "Why remove the crash protection? We're just going to get a bug filed on it."
+    },
+    {
+      "prediction": "I think this should be `X-Content-Type-Options`",
+      "reference": "Maybe be consistent and name it `FRAME_OPTIONS`."
+    },
+    {
+      "prediction": "I think we need to keep this.",
+      "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": "Please remove this empty line.",
+      "reference": "Could probably move these constants too?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think we should stop the server if we can't ping the proposal. What do you think?",
+      "reference": "Maybe 'failed' instead of 'missing'. Also do we really need to write ticket numbers here?"
+    },
+    {
+      "prediction": "nit: remove unused import",
+      "reference": "Test used to use old bouncycastle dependency which isn't needed anymore from Hadoop. Switched to use builtin Java split."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `VoiceInfo` constructor, so that we don't have to remember to do it everywhere we create a `VoiceInfo` object.",
+      "reference": "nit: I think this would be more readable as \"Some language names contain Unicode characters\"."
+    },
+    {
+      "prediction": "I don't think you need to import `os`, `sys` and `_winreg` in this file.",
+      "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": "I think we can remove the `/** @jsx createElement */` line now.",
+      "reference": "Oh interesting - does web-test-runner not resolve package.json files?"
+    },
+    {
+      "prediction": "I think this should be `P\u00e4iv\u00e4m\u00e4\u00e4r\u00e4`",
+      "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": "Shouldn't this be `component.getDerivedStateFromProps(props, component.prevState)`?",
+      "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 you can use `SmartStore.SOUP_NAME_COL`, `SmartStore.PATH_COL` and `SmartStore.COLUMN_TYPE_COL` directly instead of using `SmartStore.SOUP_ATTRS_TABLE` and `SmartStore.SOUP_INDEX_MAP_TABLE`.",
+      "reference": "Use `String.format(Locale.US, ...) to avoid the `Lint` warning."
+    },
+    {
+      "prediction": "I don't think we need to add the namespace here.",
+      "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": "RuntimeTLP is a top level property supported by CAS template engine The policy specific properties are placed with RuntimeTLP as the top level property RuntimeTLP is a top level property supported by CAS template engine The policy specific properties are placed with RuntimeTLP as the top level property RuntimeTLP is a top level property supported by CAS template engine The policy specific properties are placed with RuntimeTLP as the top level property RuntimeTLP is a top level property supported by CAS template engine The policy specific properties are placed with RuntimeTLP as the top level property RuntimeTLP is a top level property supported by CAS template engine The policy specific properties are placed with RuntimeTLP as the top level property RuntimeTLP is a top level property supported by CAS template engine The policy specific properties are placed with RuntimeTLP as the top level property RuntimeTLP is a top level property supported by CAS template engine The policy specific properties are placed with RuntimeTLP as the top level property RuntimeTLP is a top level property supported by CAS template engine The policy specific properties are placed with RuntimeTLP as the top level property RuntimeTLP is a top level property supported by CAS template engine The policy specific properties are placed with RuntimeTLP as the top level property RuntimeTLP is",
+      "reference": "Can we avoid this. upgrade engine code should take care of this."
+    },
+    {
+      "prediction": "This should be `TOKEN_EXP_DEFAULT = {'days': 90}`",
+      "reference": "should also be alphabetized or at least grouped and alphabetized within group"
+    },
+    {
+      "prediction": "I think we can remove this now.",
+      "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 don't think this error is used anywhere.",
+      "reference": "nit: can we remove \"can't\" from the error message?"
+    },
+    {
+      "prediction": "I think this is not required. If the status is empty then the phase will be Pending.",
+      "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 don't think we need to change this to `yum` since we are using `apt` in this file.",
+      "reference": "Rename this as KubernetesGPGURL - >KubernetesUbuntuGPGURL , Modify it where ever it is used."
+    },
+    {
+      "prediction": "`DEFAULT_FEE` should be `0.0001 * 100000000`",
+      "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>const</small></td><td><small>linear</small></td><td><small>linear</small></td></tr>\n  * <tr><td>{@linkplain javaslang.collection.Map}</td><td><small>const</small></td><td><small>linear</small></td><td><small>const</small></td><td><small>const</small></td><td><small>linear</small></td><td><small>linear</small></td></tr>\n  * <tr><td>{@linkplain javaslang.collection.Queue}</td><td><small>const</small></td><td><small>linear</small></td><td><small>const</small></td><td><small>const</small></td><td><small>linear</small></td><td><small>linear</small></td></tr>\n  * <tr><td>{@linkplain javaslang.collection.Set}</td><td><small>const</small></td><td><small>linear</small></td><td><small>const</small></td><td><small>",
+      "reference": "oh yeah, `tail`, `update`, `prepend`, `append`, `subSequence`, `trim` just became effectively constant (without memory leak) :)"
+    },
+    {
+      "prediction": "I don't think this is needed. `$resource` is already an array.",
+      "reference": "Can you add back a deprecated `destroyLinks` method for backward compatibility? It can simply proxy `destroyResourceLinks`, and we can remove it in the next major release."
+    },
+    {
+      "prediction": "I think this should be `RegisterGlobalOption(\"server\", parseServerOptions)` instead of `RegisterGlobalOption(\"servers\", parseServerOptions)`.",
+      "reference": "A reminder that we should discuss whether to rename this to \"sockets\" or \"listeners\"."
+    },
+    {
+      "prediction": "This is not needed.",
+      "reference": "Why do you need to include this using declaration?"
+    },
+    {
+      "prediction": "I'm not sure why this change is needed. I don't think it's related to this PR.",
+      "reference": "@halter73 where is this file generated from?"
+    },
+    {
+      "prediction": "Why is this `TRUE` and not `true`?",
+      "reference": "What do these `#if TRUE` accomplish?"
+    },
+    {
+      "prediction": "I think this should be `if (this.handler.SupportsNullActivity || Sdk.SuppressInstrumentation) { return; }`",
+      "reference": "Probably a good optimization to have this, but I don't think it's strictly required. For \"legacy\" Activity flows we pass them through an ActivitySource for sampling. So the logic below should also catch these?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it works for now.",
+      "reference": "@jessieay Intentionally moving away from passing a param?"
+    },
+    {
+      "prediction": "`$firstname = trim(implode(' ', array_slice($nameArr, 0, count($nameArr) - 1)));`",
+      "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": "I think we can remove this line.",
+      "reference": "hmm, we actually should _add_ `'use strict'` everywhere"
+    },
+    {
+      "prediction": "I don't think this should be a debug message.",
+      "reference": "it's not just fetching override properties here, right?"
+    },
+    {
+      "prediction": "I think this should be `seelog.Infof(\"ACS Websocket connection closed for a valid reason: %v\", acsError)`",
+      "reference": "Is it worth logging the error?"
+    },
+    {
+      "prediction": "nit: `key not in self.ns` -> `key in self.ns`",
+      "reference": "What's the point of this class?"
+    },
+    {
+      "prediction": "I think this should be `contentType !== 'movies' && contentType !== 'mixed' && contentType !== 'tvshows'`",
+      "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 don't think you need to import `github.com/syndtr/goleveldb/leveldb` here.",
+      "reference": "Please put this down with the other \"github.com\" imports."
+    },
+    {
+      "prediction": "I don't think this is a good idea to add a new parameter to an existing function. I think it would be better to add a new function `void nano::block_processor::add (nano::unchecked_info const & info_a, bool push_front_preference)` and keep `void nano::block_processor::add (nano::unchecked_info const & info_a)` as it is.",
+      "reference": "Minor but probably want `push_front_preference` to have a trailing `_a` to be consistent with the other parameter"
+    },
+    {
+      "prediction": "nit: I think we can remove the `caseSensitive` parameter from the constructor and always pass `true` to `Projections.inclusive` and `Binder.bind`.",
+      "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 this import.",
+      "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 to add the API capability. I think it should be done in `kinto/__init__.py` where the `config.add_api_capability` call is done.",
+      "reference": "nitpick: I wonder if we should name it `flush` only (?)"
+    },
+    {
+      "prediction": "`dealsGauge` -> `dealsCurrentGauge`",
+      "reference": "Why gauges, not counters?"
+    },
+    {
+      "prediction": "I don't think we need to change this constant. We can just use `sysHashLen` instead of `gwHashLen`.",
+      "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": "Should this be \"Pod Failure\" or \"Pod Failed\"?",
+      "reference": "Shall we rename this status to CONTAINER_FAILED? ^^ cc: @sshardool"
+    },
+    {
+      "prediction": "I think this should be `Fetcher` instead of `FetcherInterface`",
+      "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": "nit: I think this should be \"The standard definitions of CEL functions and operators can be found [here](https://github.com/google/cel-spec/blob/master/doc/langdef.md#standard-definitions).\"",
+      "reference": "I know it has no effect but my eyes can't help. Is that line not too long?"
+    },
+    {
+      "prediction": "I'm not sure why this was added in the first place, but I don't think we need it anymore.",
+      "reference": "Does that mean we disable service loop prevention for packet generated by local host?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 think this should be `write(2, spec, outputFile, 0)`.",
+      "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": "nit: I think we can use `rewrittenFileScanTaskSetId` instead of `rewritten-file-scan-task-set-id`",
+      "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 `namedTools map[string][]BuildInput `name:\"named_debug_tools\"`",
+      "reference": "Nice! Good idea to move these out of the main struct."
+    },
+    {
+      "prediction": "This should be `if 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 `public enum Type { IMAP, SMTP, WEBDAV, POP3 }`",
+      "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": "I'm not sure if this is the right place to put this, but I think it would be nice to have a `tokenList` function in `commons/utils.js` so that we don't have to repeat this code in multiple places.",
+      "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 `if (label) { this.data('has-label'); }` instead of `if (label) { this.data('has-label'); } else { this.data('no-label'); }`",
+      "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 don't think this is the right place to do this check. I think it should be done in the `syncApplication` function.",
+      "reference": "nit: \"The current project does not have requested application\""
+    },
+    {
+      "prediction": "Why do we need this constant? It is not used anywhere.",
+      "reference": "Extend Unit test for this class"
+    },
+    {
+      "prediction": "I think this should be `DD_APPSEC_EXTRA_HEADERS` instead of `DD_APPSEC_EXTRA_HEADERS`",
+      "reference": "This is this a copy / paste error from above."
+    },
+    {
+      "prediction": "This file is not needed.",
+      "reference": "this package has to be renamed to `typeinference`"
+    },
+    {
+      "prediction": "Shouldn't this be `TestCase.assertNotEqual(user.providerType, \"anon-user\");`?",
+      "reference": "This cancels the above null-check I guess."
+    },
+    {
+      "prediction": "I think we should keep the old name, and add the new name as alias.",
+      "reference": "change to timerIntervalInMillis"
+    },
+    {
+      "prediction": "I don't think this is needed. If you don't assign a role to a user, they will not have any permissions.",
+      "reference": "`any rules` should be `any roles`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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": "This should be `conf = keyconf.KeyConfigParser(None, 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": "Why is this removed?",
+      "reference": "@tmetzke shouldn't we replace this library with the `2.12.1` instead of removing it?"
+    },
+    {
+      "prediction": "I'm not sure what you're trying to say here, but I don't think this is correct.",
+      "reference": "uint64 seems excessive here. Should we reduce to a uint32 @whyrusleeping ?"
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "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 think this should be `r.Host` instead of `r.RemoteAddr`. `r.RemoteAddr` is the IP address of the client making the request, while `r.Host` is the host that the client is trying to connect to.",
+      "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": "I think we can remove the `a.hasFoundationBonusExtension()` check and just call `a.grantFoundationBonus(epochNum)`.",
+      "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": "`f'Invalid flipping direction \"{direction}\"'`",
+      "reference": "Use single quote to wrap the str."
+    },
+    {
+      "prediction": "Shouldn't this be `perm_getNodesWhitelist`?",
+      "reference": "Is the plan to rename this later?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the strategy is not in the list of strategies, it should throw an exception.",
+      "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 an error, not a panic.",
+      "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 don't think this is needed. `m_hpx_initialized` should be sufficient.",
+      "reference": "Why do we need both `HPX::m_hpx_initialized` and `HPX:: m_was_initialized`?"
+    },
+    {
+      "prediction": "I don't think we need to panic here. We can just return an error.",
+      "reference": "also check RPCName?"
+    },
+    {
+      "prediction": "I don't think we need to import `WebSessionOAuth2ServerAuthorizationRequestRepository` here.",
+      "reference": "Remove unused import"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do for now.",
+      "reference": "I assume there will never be a scenario where justValidate is expected to be true on a restart"
+    },
+    {
+      "prediction": "I don't think this is the right place to add the listener. I think it should be done in `PmdExecutor.execute()`.",
+      "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 think this should be `nil` instead of an empty string.",
+      "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 don't think we need to do this. If `args` is `undefined`, `EJSON.stringify(args, { relaxed: false })` will return `\"null\"`, which is what we want.",
+      "reference": "Why do this rather than `args = []` in the function signature?"
+    },
+    {
+      "prediction": "I think this should be `['puppet-agent', 'bolt']`",
+      "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 way to do it.",
+      "reference": "@StrikerRUS here may need to concat two data by col."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it's better to keep the javadoc as is, and let the user add the leading space if they want it.",
+      "reference": "Why the space?"
+    },
+    {
+      "prediction": "Please remove the extra `}` at the end of this file.",
+      "reference": "please resolve the extra indentation"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "More proof that this check is needed always."
+    },
+    {
+      "prediction": "I think this should be `m_al_comms.clear()` instead of `m_al_comms = nullptr`.",
+      "reference": "`m_al_comms` now contains smart pointers."
+    },
+    {
+      "prediction": "Shouldn't this be `if (type is EnumType && (type as EnumType).ModelAsString && defaultValue != null)`?",
+      "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 to lock here. `c.nc` is only set in `initClient` which is called from `connect` which is protected by `c.mu`.",
+      "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": "Can you explain why the number of partitions has changed?",
+      "reference": "This change is suspicious. Why did the number of partitions increase?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it works for now.",
+      "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'm not sure why this was disabled in the first place, so I'm not sure if this is the right thing to do.",
+      "reference": "Should there be any changes to this file at all?"
+    },
+    {
+      "prediction": "I think this should be `var autoPolicy *caddytls.AutomationPolicy = caddytls.NewAutomationPolicy()`",
+      "reference": "The only field being used is the KeyType; Instead, we can probably just make a `keyType` variable here."
+    },
+    {
+      "prediction": "I think this should be `if (!options.credentials) { options.credentials = {}; }`",
+      "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 best way to do it, but it seems to work. I'm not sure if there's a better way to do it, or if this is the best way to do it.",
+      "reference": "Cleaned up when I originally added support for passing `screenContext` here (and later `viewContext`) but that was removed."
+    },
+    {
+      "prediction": "Please don't use tabs for indentation.",
+      "reference": "This looks like a bunch of whitespace fixing that's unrelated. Can you revert?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "should be removed instead"
+    },
+    {
+      "prediction": "I don't think this should be a global variable. It should be a parameter to `checkRandomExpression()`.",
+      "reference": "should use randomization ?"
+    },
+    {
+      "prediction": "I think this should be `return fmt.Errorf(\"failed to get volume info: %v\", err)`",
+      "reference": "returning nil because we want to mayactl to exit with 0 status code."
+    },
+    {
+      "prediction": "I think this should be `H2O_STRLIT(\"application/json; charset=utf-8\")`",
+      "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": "I think this should be `[Flags]` instead of `[Flags]`.",
+      "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 `log.WithField(\"useResourceUpdates\", configParams.UseResourceUpdates()).Info(\"using resource updates where applicable\")`",
+      "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 `programbytes` instead of `programstr`",
+      "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 the right way to do this. You should be able to get the value of an attribute directly from the `IAccessible2::attribute_current` key.",
+      "reference": "Extraneous blank line."
+    },
+    {
+      "prediction": "I think this should be \"given key is not a possible signer for either Sender or AuthAddr\"",
+      "reference": "nit: syntax : neither -> nor"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the only thing we can do. We don't have access to the `ActiveScope` anymore, so we can't get the `SpanId` and `TraceId` from the `ActiveScope`.",
+      "reference": "Is it safe to assume that these two keys are always present? `this[string]` will throw a `KeyNotFoundException` if they are not."
+    },
+    {
+      "prediction": "Shouldn't this be `exflow.getFlowId()` instead of `exflow.getFlowDefinitionId()`?",
+      "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` instead of `DateTime`",
+      "reference": "Shouldn't it be in the reverse order? `DateTime::createFromInterface()` returns `self` and accepts `DateTimeInterface`."
+    },
+    {
+      "prediction": "I don't think we should remove this check.",
+      "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 `alpha_reverse` instead of `alpha-reverse`",
+      "reference": "Please could you remove the useless space."
+    },
+    {
+      "prediction": "I don't think you need the `assert(arc.target < Integer.MAX_VALUE);` here.",
+      "reference": "What's the point of this block (and isn't it effectively dead code)?"
+    },
+    {
+      "prediction": "I think this should be `ENV[\"DISCOURSE_SSO_SECRET\"] || ENV.fetch(\"DISCOURSE_SSO_SECRET\")`",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I think this should be `@job(version_strategy=SourceHashVersionStrategy())` instead of `@job(version_strategy=SourceHashVersionStrategy())`",
+      "reference": "Newline at end of file plz"
+    },
+    {
+      "prediction": "nit: I think we can remove this empty line.",
+      "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": "I think this should be `protected` as it is only used in the `IndexQueue` class.",
+      "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 right way to do this. I think we should be raising an exception if the element is stale.",
+      "reference": "This is not the right exception class. There is an InvalidSelectorException class that covers bad locators."
+    },
+    {
+      "prediction": "I don't think you need to include `Kokkos_Core.hpp` and `Kokkos_Cuda.hpp` here. `Kokkos_CudaSpace.hpp` should suffice.",
+      "reference": "I assume this came from format?"
+    },
+    {
+      "prediction": "I think this should be a `protected $clearSearchIndex = false;` instead of a `protected $clearSearchIndex;`",
+      "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'm not sure if this is the right thing to do, but I think we need to log the keys that we're trying to decrypt with.",
+      "reference": "Do we really need this? Seems like it would pollute the logs pretty badly."
+    },
+    {
+      "prediction": "I think this should be `private const 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 don't think this is the right way to do it. I think you should use `unittest.skipIf(sqlalchemy is None, \"PostgreSQL not available\")` instead.",
+      "reference": "Can we use a skipIf decorator instead?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to be the only way to get the deprecation warning to show up.",
+      "reference": "Missing version tags"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `PluginResolver` class, so that we don't have to do it in multiple places.",
+      "reference": "I think this should also filter out `validate_resolve_reference` hooks."
+    },
+    {
+      "prediction": "I think we can remove the `using OpenTelemetry.Exporter.OpenTelemetryProtocol.Implementation;` line.",
+      "reference": "unrelated, but required fix as metrics branch got broken with the latest main->metrics merge and conflict resolution."
+    },
+    {
+      "prediction": "Please remove the empty line between the copyright header and the import statements.",
+      "reference": "Feel free to remove this line"
+    },
+    {
+      "prediction": "I think this should be `I18n.t(\"shared.subscriptions.upcase_membership\")`",
+      "reference": "~~Should this match the key updated above?~~ :+1:"
+    },
+    {
+      "prediction": "I don't think we should ignore the error here. If we can't close the file, we should at least log it.",
+      "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.Rat` instead of `*big.Int`.",
+      "reference": "collateral is in FIL"
+    },
+    {
+      "prediction": "Why do we need to create a new channel here?",
+      "reference": "We should not create channel of node which are not connected."
+    },
+    {
+      "prediction": "I don't think this is correct. `options.source` is not the same as `options.db`. `options.source` is the name of the database that the user authenticates against. `options.db` is the name of the database that you want to connect to.",
+      "reference": "I think the docstring should say something like \"Alias for the `source` option\" or something similar."
+    },
+    {
+      "prediction": "I don't think we should add new fields to this class. Instead, we should add new fields to the `ProductTemplate` class. This way, we can keep the `Product` class as simple as possible, and we can add new fields to the `ProductTemplate` class without having to change the `Product` class. Also, we can add new fields to the `ProductTemplate` class without having to change the `Product` class, and we can add new fields to the `ProductTemplate` class without having to change the `Product` class, and we can add new fields to the `ProductTemplate` class without having to change the `Product` class, and we can add new fields to the `ProductTemplate` class without having to change the `Product` class, and we can add new fields to the `ProductTemplate` class without having to change the `Product` class, and we can add new fields to the `ProductTemplate` class without having to change the `Product` class, and we can add new fields to the `ProductTemplate` class without having to change the `Product` class, and we can add new fields to the `ProductTemplate` class without having to change the `Product` class, and we can add",
+      "reference": "Please use properties without suffix `_fr` so they can work across different languages as @teolemon mentioned"
+    },
+    {
+      "prediction": "I think this should be `static void 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 think this should be `ContentMD5 []byte` instead of `ContentMD5 string` so that we don't have to do `base64.StdEncoding.EncodeToString(contentMD5)`.",
+      "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'm not sure if this is the right thing to do, but I think we need to assign the transport IP address to the gateway interface, not the node IP address. The node IP address is used for the node-to-node communication, while the transport IP address is used for the pod-to-pod communication. If we assign the node IP address to the gateway interface, then the pod-to-pod communication will not work.",
+      "reference": "Was it a bug?"
+    },
+    {
+      "prediction": "I don't think this is correct. If `attr_value['string']` is `None` then `attr_value['string'] == value` will be `False` even if `value` is `None`.",
+      "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 `tabbed_browser._last_hovered_link` instead of `_now_focused._last_hovered_link`. `_last_hovered_link` is an attribute of `TabbedBrowser`, not of `TabbedBrowser._now_focused`.",
+      "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 about the name of this mailer. `InvitationMailer` sounds like it's sending an email about an invitation, but this mailer is actually sending an email to someone who has been invited to sign up. Maybe `InvitedUserMailer` would be a better name?",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "This file should be deleted.",
+      "reference": "We should file a bug for this - we don't need the extensions class if there are no operations on the client"
+    },
+    {
+      "prediction": "nit: I think we can remove the empty string here.",
+      "reference": "Is it possible to import from k8s instead of copy in? :)"
+    },
+    {
+      "prediction": "Shouldn't this be `int`?",
+      "reference": "`@param float` can be dropped here."
+    },
+    {
+      "prediction": "why remove final?",
+      "reference": "maybe we can leave this change un-reverted."
+    },
+    {
+      "prediction": "I think this should be `AbstractJavaTypeNode` instead of `AbstractJavaAccessNode`.",
+      "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 think this should be `PlayerHeadIconEvent` instead of `PlayerHeadIconChanged`",
+      "reference": "delete this class"
+    },
+    {
+      "prediction": "I think this should be `else if (/^[1-9][0-9]+|0|[1-9]$/.test(s)) {`",
+      "reference": "Replacing `+` (1 or more) with `*` (0 or more) would've also done the trick."
+    },
+    {
+      "prediction": "I don't think we should remove the `throw()` here.",
+      "reference": "I've been meaning to fix this for a while. Thanks."
+    },
+    {
+      "prediction": "I'm not sure if this is related to this PR, but it seems like there's a bug here. I would expect `MyActivityProcessor(\"C\").OnEnd` to be called before `MyActivityProcessor(\"B\").OnEnd` and `MyActivityProcessor(\"A\").OnEnd`, but that doesn't seem to be the case.",
+      "reference": "This is adding multiple processor pipelines. I guess you wanted to add multiple processors to the same, single pipeline?"
+    },
+    {
+      "prediction": "`str(x[0]) in \"0123456789\"` should be `x[0] in \"0123456789\"`",
+      "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 this should be 3000, not 3600000. 3600000 is 1 hour. 3000 is 5 minutes.",
+      "reference": "3.6 seconds? what does it stand for?"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "If the command fails, is stdout nil or \"\"?"
+    },
+    {
+      "prediction": "I don't think we should remove the `apiVersion` query parameter.",
+      "reference": "Why is api-version now removed in the generated code?"
+    },
+    {
+      "prediction": "I think this file is not needed.",
+      "reference": "We may register all built-in optimizers of PyTorch here to simplify the builder."
+    },
+    {
+      "prediction": "I think this should be `const auto &components = ns->components;`",
+      "reference": "No need for this new variable."
+    },
+    {
+      "prediction": "I don't think this is correct. If `c._nextState` is `null`, then `c._nextState = c.state;` will set `c._nextState` to `c.state`, but `c._nextState == null` will still be true.",
+      "reference": "doesn't this enqueue a double render or is that safeguarded somehow"
+    },
+    {
+      "prediction": "I'm not sure why this is necessary, but it seems like it should be safe to remove.",
+      "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 you need to check for `ADIOS_HAVE_PHDF5` here. The `HDF5Reader` constructor will throw an exception if `ADIOS_HAVE_PHDF5` is not defined.",
+      "reference": "Use `ADIADIOS2_HAVE_HDF5`, not `ADIOS_HAVE_PHDF5`"
+    },
+    {
+      "prediction": "I'm not sure I like this change. I think it would be better to keep the existing behavior and add a new method called `group_value_for_string` or something like that. That way we don't have to change the behavior of `group_value_for` and we don't have to add a new method called `group_value_for_string_and_class`. WDYT?",
+      "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": "This change is not related to this PR. Can you please revert it?",
+      "reference": "Should MAX_PERTABLE_ENTRY here be 10 ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these. I think they should be part of the `Http11Response` class.",
+      "reference": "Should move these out of frame into a static class? Similar to reason phrases?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "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": "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 to put this test. I think it should be in a separate test program.",
+      "reference": "Looks like a clang-format style violation: should turn Travis red."
+    },
+    {
+      "prediction": "This should be `add_setting :colors` and then you can use `colors[:success]`, `colors[:pending]` and `colors[:failure]`.",
+      "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'm not sure if this is the best way to do this, but it's the only way I could think of to get a property from the `siteInfo` object.",
+      "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 about this one. It seems to be deprecated in GCC 9.0.0.",
+      "reference": "The change here is to address what exactly?"
+    },
+    {
+      "prediction": "I think this should be `IsMSIEnabled` or `IsMSIAvailable` or something like that. `IsMSIEnvironment` sounds like it's a property of the environment, but it's actually a property of the blob storage account.",
+      "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": "I think this should be `dp_bits = new unsigned char[size / 8 + 1];` instead of `dp_bits = new unsigned char[size / 8 + 1]();`",
+      "reference": "Was this done automatically or by hand?"
+    },
+    {
+      "prediction": "I think this should be `desired_group = child_group.name if current_group.name == child_group.name`",
+      "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": "I think this should be `VUID-NONE` as well.",
+      "reference": "Should this be a constant somewhere?"
+    },
+    {
+      "prediction": "Do we want to make these public?",
+      "reference": "nit: sort alphabetically."
+    },
+    {
+      "prediction": "I don't think we need to use the class name as the context key. We can just use `String contextKey = \"_tt\";` and then use `contextKey` instead of `JsonQueryConverter.class.getSimpleName()`.",
+      "reference": "Now, It's redundant, I suppose."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to deprecate this, but it's the only place I could find it.",
+      "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 an error, not an info.",
+      "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'm not sure this is the right way to do it, but it seems to work.",
+      "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'm not sure if this is the best way to do it, but it seems to work.",
+      "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 think this should be `static int json_output = 0;` instead of `static int json_input = 0;`",
+      "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": "Shouldn't this be `nameof(location)`?",
+      "reference": "This one was wrong."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what else to do either. The problem is that we don't have a good way to know if a change is visible to the user or not, and we don't want to send a notification if it's not visible to the user. I think this is the best we can do for now, but I'm open to suggestions.",
+      "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": "nit: remove empty line",
+      "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": "I think it would be better to call this `formatted_lines` to avoid confusion with `fully_formatted` which returns a single string.",
+      "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": "nit: I think we can remove the `/** Defines the SNIPPET_RESOURCE_ROOT. */` since it doesn't add any value.",
+      "reference": "Is it necessary to keep this class around just to define this constant? Or can we place it somewhere else?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `Policy` struct that has a `Name` and `Namespace` field. Then you can just pass a pointer to the `Policy` struct to the `rule` struct. That way, you don't have to worry about keeping the `Name` and `Namespace` fields in sync with the `Policy` struct. Also, you can just pass the `Policy` struct to the `rule` struct, instead of having to pass the `Name` and `Namespace` fields separately.",
+      "reference": "Question - do we cache NetworkPolicy itself? If so, here we can point to NetworkPolicy?"
+    },
+    {
+      "prediction": "why remove workqueue?",
+      "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 `telemetryEnabled` instead of `telemetryInitialized`.",
+      "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": "Shouldn't this be `this.Methods.FirstOrDefault(m => m.Parameters.Any(p => p.Type == PrimaryType.TimeSpan)) != null;`?",
+      "reference": "please use new line to maintain reasonable line width"
+    },
+    {
+      "prediction": "I think this should be `if err != nil { return }` instead of `if ta == nil { return }`",
+      "reference": "I would skip this check or add an internal server error response."
+    },
+    {
+      "prediction": "I think this should be `if (config.notarize) { notarize = config.notarize; notary_user = config.notary_user; notary_password = config.notary_password; } else { notarize = null; notary_user = null; notary_password = null; }`",
+      "reference": "missing `{` here (and then `}` after `notary_password = config.notary_password`); it's only going to do the first one"
+    },
+    {
+      "prediction": "I think this should be `app.SiteSettingsPath` instead of `app.SiteLocalSettingsPath`",
+      "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/go/extractors/config/default` 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 don't think we need to return an error here. If we can't start kube-proxy, we can just log the error and continue.",
+      "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 don't think we need to disable this rule here.",
+      "reference": "@btecu why is this override needed here? I'm not seeing a `new` being used here."
+    },
+    {
+      "prediction": "I think this should be `c.Errorf(\"Registering leaf with account %q resulted in error: %v\", acc.Name, err)`",
+      "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\").ComputeServiceAccount)`",
+      "reference": "minor: use a non-default CE service account for testing (since default is handled in a specific way in the code)"
+    },
+    {
+      "prediction": "I don't think we need to keep this in the PR.",
+      "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": "I'm not sure this is the right place to do this. I think it should be done in the `Beaker::HostManager` class where the `@options[:hosts_file]` is set.",
+      "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 we should keep this as a `var` and not expose it as a `const`. I don't think we want users to be able to mutate it.",
+      "reference": "Agree that exporting these maps is the best tradeoff, just registering my ongoing discontent with the lack of `const` collections."
+    },
+    {
+      "prediction": "I think this should be `self._tab.run_js_async` instead of `self._tab.run_js_async`.",
+      "reference": "This could probably use `const` as well?"
+    },
+    {
+      "prediction": "This should be `Orig` and `Tmp`.",
+      "reference": "it doesn't look like these need to be exported?"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "I think you should remove \"fmt\" here. And change `fmt.Errorf` to `errors.Errorf`."
+    },
+    {
+      "prediction": "I'm not sure about the name of this column. I think it should be `video_resources` or something like that. What do you think?",
+      "reference": "Here's the `resources` attribute @jferris. I'm not against renaming `Product` to `Resource`, and this one... something else."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we care about the error from `errgroup.WithContext` in this case.",
+      "reference": "If the context is unused, this is equivalent to `var wg errgroup.Group`."
+    },
+    {
+      "prediction": "I think this should be `HIGHLIGHT + \".fragsizeIsMinimum\"`, not `HIGHLIGHT + \".fragsizeIsMinimum\"`.",
+      "reference": "very minor: I'd prefer these two added rows are switched so that fragsizeIsMinimum directly follows fragsize"
+    },
+    {
+      "prediction": "I think this should be `if host in ('localhost', '127.0.0.1', '::1'):`",
+      "reference": "Given that 115 returns `True`, why would this be `False` here?"
+    },
+    {
+      "prediction": "I think this test should be moved to `OAuth2LoginConfigFormLoginTests` since it's testing the form login configuration.",
+      "reference": "Please move this test method just below `oauth2LoginWithOneClientConfiguredThenRedirectForAuthorization()`"
+    },
+    {
+      "prediction": "This should be `c_TimeInfinite` instead of `Duration_t(1, 0)`.",
+      "reference": "Why this new configuration?"
+    },
+    {
+      "prediction": "I think you can just do `dupe.array()` instead of `dupe.get(new byte[dupe.remaining()]).array()`",
+      "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": "Shouldn't this be `SpanReference`?",
+      "reference": "`spanReference` -> `baggage`"
+    },
+    {
+      "prediction": "I think this should be `/static/logo.png` instead of `/-/static/logo.png`",
+      "reference": "Why the `/-/verdaccio/` ?"
+    },
+    {
+      "prediction": "I think we can remove the `enableAddons` field from the `appDeployVars` struct.",
+      "reference": "Has been waiting for a million years"
+    },
+    {
+      "prediction": "`after than` should be `after than`",
+      "reference": "Minor grammar detail: the 'than' is unnecessary"
+    },
+    {
+      "prediction": "`item_top_two = self._internal._sdf.select(self.spark_column).head(2)`",
+      "reference": "I think you can just simply fix this line to `self[:2].to_pandas()`"
+    },
+    {
+      "prediction": "I think this should be `if (debuggerAddress.startsWith(\"ws://\") || debuggerAddress.startsWith(\"http://\")) {`",
+      "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 `import { registerStore } from 'assets/js/googlesitekit/datastore/site';`",
+      "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 `actualTimeSinceParent <= minimumSecondsSinceParent` instead of `actualTimeSinceParent < minimumSecondsSinceParent`",
+      "reference": "not sure \"actual\" adds value - secondsSinceParent is probably closer."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do this, but it seems to work.",
+      "reference": "Line is too long. [94/80]"
+    },
+    {
+      "prediction": "I think this should be `You cannot delete any items. You should modify the existing.`",
+      "reference": "The error should be `At least :number items are required`"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the least bad of the options we have.",
+      "reference": "singleCaseSwitch: should rewrite switch statement to if statement (from `gocritic`)"
+    },
+    {
+      "prediction": "nit: `mdWithKeys` -> `mdWithKey`",
+      "reference": "Could combine this with `encryptionKeyGetter` to have a single `keyGetter` interface. I'm not sure which way is better."
+    },
+    {
+      "prediction": "Why is this not deleted?",
+      "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'm not sure if this is the right thing to do, but I think we need to remove the `writeConcern` field from the `illegalCommandFields` array as well.",
+      "reference": "Does this mean we do not support writeConcern on `createCollection`?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "I guess `sid` was for `sessionID`, but now it looks strange: `sid.ID`."
+    },
+    {
+      "prediction": "I think we can remove the `anyIPv4CidrBlock` here.",
+      "reference": "We'll need `infrav1.SecurityGroupBastion` as well"
+    },
+    {
+      "prediction": "I think this should return an error if the feature is not supported.",
+      "reference": "This `fileName` should probably be `remote` and be relative to the Fs root as per normal rclone usage."
+    },
+    {
+      "prediction": "I think we should keep this test as it is, and add a new test for the case where we drop multiple indexes.",
+      "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": "This should be `private static final String specialCharacters = \"[^\\\\d\\\\s\\\\w!#\\\\$%&'\\\\(\\\\)\\\\-@\\\\^_`\\\\{\\\\}~.,]\";`",
+      "reference": "Expression that uses a negation pattern to exclude all characters that aren't in the expression."
+    },
+    {
+      "prediction": "I think this should be `Rpc-Respond-Service-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 || commandSet == targetLocalCommandPath {`",
+      "reference": "The serviceDirOnHost[0:1] was completely wrong. It wasn't just Windows, glad you got this fixed!"
+    },
+    {
+      "prediction": "I think this is too long for a unit test. Can we reduce it to 100ms or 200ms? I don't think we need to wait 15 seconds for a unit test.",
+      "reference": "Would membership take 10+ seconds to detect the failed node?"
+    },
+    {
+      "prediction": "I think this should be `except Exception as err: raise RuntimeError(f'Markdown file \"{path}\" could not be loaded: {err}')`",
+      "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 it's better to use `os.UserHomeDir()` instead of `homedir.Dir()`.",
+      "reference": "This function does not need to be public"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way to get rid of the bottom margin.",
+      "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 think this should be `if (compositeType.Extensions.ContainsKey(AzureExtensions.PageableExtension) && this.Extensions.ContainsKey(\"nextMethodName\") && (bool)this.Extensions[\"nextMethodName\"])`",
+      "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 to do this. I think it would be better to do this in the `forwardRef` implementation.",
+      "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 think this should be `#if !SILVERLIGHT && !NETSTANDARD1_3 && !NETSTANDARD2_0`",
+      "reference": "Please use \".NET Core 3\" - i'm trying hard to use one form, and this is the one MS advices"
+    },
+    {
+      "prediction": "I think this should be `case <-params.Ctx.Done(): return params.Ctx.Err()`",
+      "reference": "Unrelated to this change, but this is a fix for when TestDoStartCgroupInitHappyPath has a failure after the test goroutine has already exited."
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "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": "Why is this change needed?",
+      "reference": "Does this changes affect process building dynamic libraries anyhow?"
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "reference": "Could you limit what is imported here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it should be done here. If we don't do it here, then we'll have to do it in `do_get_function_code` as well.",
+      "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": "nit: I think it would be better to have a `icmpPkt_with_type_code(src, dst string, icmpType, icmpCode int) packet` helper function that calls `icmpPkt_with_type_code(src, dst string, icmpType, icmpCode int) packet` instead of duplicating the code.",
+      "reference": "Golang naming convention is to use camel case `icmpPktWithTypeCode` Often the linter will complain"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "We can do away with assigning the response here too."
+    },
+    {
+      "prediction": "I'm not sure why we need to add 3 instead of 2. Can you explain?",
+      "reference": "Bumping the newViewportHeight"
+    },
+    {
+      "prediction": "I think this should be `master` instead of `secondaryy`.",
+      "reference": "Is this change correct? Looks like a typo and not sure this should be changed?"
+    },
+    {
+      "prediction": "I think this should be `class_name = qname.rsplit('.', 1)[-1]`",
+      "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": "This file should not be changed.",
+      "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 don't think this is correct. If a model is only returned by a POST operation, then it is not a resource.",
+      "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 `Optional[Hash[String[1], Resource]]` instead of `Optional[Hash[String[1], ResourceInstance]]`.",
+      "reference": "It seems like most of these will never be nil, are they optional just in case?"
+    },
+    {
+      "prediction": "why -1? I think it should be `tblInfo.getColumns().size() - 1`?",
+      "reference": "`.addColumn(\"c1\", IntegerType.INT, true)` means `PkHandle=true`, maybe should add `.setPkHandle(true)` in line 31"
+    },
+    {
+      "prediction": "`obj instanceof Long || obj instanceof Integer`",
+      "reference": "how about `Cannot encode row key with non-integer type` directly?"
+    },
+    {
+      "prediction": "I think we should rename this method to `fetchUnfinishedExecutionsByExecId(int execId)` to be consistent with the other methods.",
+      "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 way to do it, but it seems to work for me.",
+      "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 file should be named `answer.go`.",
+      "reference": "ultra nit: missing Oxford comma :P"
+    },
+    {
+      "prediction": "Do we need to add `doctest: +SKIP` here?",
+      "reference": "This is not supported in Spark 3.0.0-rc2 yet. I'd skip this for now."
+    },
+    {
+      "prediction": "I think this should be called `Exec` instead of `Query` to avoid confusion with the `Query` method on the `ChainReader` interface.",
+      "reference": "BLOCKING: Why does `Query` return an `*exec.FunctionSignature`?"
+    },
+    {
+      "prediction": "I think this should be `c.dirtyBcache = NewBlockCacheStandard(c, 10000, MaxBlockSizeBytesDefault*1024)` instead of `c.dirtyBcache = NewBlockCacheStandard(c, 10000, MaxBlockSizeBytesDefault*1024)`",
+      "reference": "please move this down to immediately above the assignment to `c.dirtyBcache`"
+    },
+    {
+      "prediction": "I think we can remove the `if err != nil { log.Fatal(err) }` lines from all of these examples.",
+      "reference": "Don't need the `, err` part since you've already handled it."
+    },
+    {
+      "prediction": "I think this should be `CHECK(k == 1);`",
+      "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 this should be done in the `Serve` method of the `Cluster` struct. The `Serve` method is the one that actually starts the server, and it's the one that should be responsible for regenerating the certificates.",
+      "reference": "What happens if the certificate rotation fails and we are prematurely removing this file?"
+    },
+    {
+      "prediction": "I think this should be `if s.once.IsRunning() { return ... }`",
+      "reference": "does this change belong here?"
+    },
+    {
+      "prediction": "I think this should be `os.Getenv(\"PWD\")` instead of `os.Getwd()`. `os.Getwd()` returns the current working directory of the process, while `os.Getenv(\"PWD\")` returns the value of the `PWD` environment variable.",
+      "reference": "This returns an error"
+    },
+    {
+      "prediction": "I think this should be `config:\"oneof(legacy,nft);legacy;local;nft\"`",
+      "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'm not sure if this is the right thing to do, but I think we need to get the head tipset and state from the miner node, because the miner node is the only node that has the full chain.",
+      "reference": "Looks like that helper function is general enough that it should reside in testhelpers"
+    },
+    {
+      "prediction": "I think this should be `if app.DisableSettingsManagement || app.DisableDrushrcManagement { 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": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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'm not sure if this is the right wording, but I think it's better to say \"If you proceed without specifying public subnets, you will not be able to deploy Load Balanced Web Services in this environment.\" than \"If you proceed without public subnets, you will not be able to deploy Load Balanced Web Services in this environment.\"",
+      "reference": "Should we say \"specifying two public subnets\"?"
+    },
+    {
+      "prediction": "I think we can remove this TODO now.",
+      "reference": "Issue number? When? Why?"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "You should not rely on `PitchCommand` being imported into speech. Please import it from `speech.commands`."
+    },
+    {
+      "prediction": "I think this should be `auth.type` 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 should keep the `ReadLevel` and `MaxReadLevel` for backward compatibility.",
+      "reference": "maybe we should rename: MinTaskID -> MinTaskIDExclusive, MaxTaskID -> MaxTaskIDInclusive,"
+    },
+    {
+      "prediction": "I think this should be `nvidia` instead of `ecs-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'm not sure this is correct. I think we need to set both `Publish` and `Subscribe` to `&SubjectPermission{Deny: remote.DenyExports}` and `&SubjectPermission{Deny: remote.DenyImports}` respectively.",
+      "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 don't think we need this function. We can just use `opmask_with_dsts()` directly.",
+      "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 `dataDirStat.Mode() & 0777` instead of `dataDirStat.Mode() & 0077`",
+      "reference": "I believe that you meant to have the following on the second one, right ? `dataDirStat.Mode()^0077`"
+    },
+    {
+      "prediction": "I think this should be `protocol.ByteCount` instead of `uint64`.",
+      "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": "I think you can just return `IsEffectInSpell(spell_id, SE_Invisibility) || IsEffectInSpell(spell_id, SE_Invisibility2) || IsEffectInSpell(spell_id, SE_InvisVsUndead) || IsEffectInSpell(spell_id, SE_InvisVsUndead2) || IsEffectInSpell(spell_id, SE_InvisVsAnimals) || IsEffectInSpell(spell_id, SE_InvisVsUndead3) || IsEffectInSpell(spell_id, SE_InvisVsUndead4) || IsEffectInSpell(spell_id, SE_InvisVsUndead5) || IsEffectInSpell(spell_id, SE_InvisVsUndead6) || IsEffectInSpell(spell_id, SE_InvisVsUndead7) || IsEffectInSpell(spell_id, SE_InvisVsUndead8) || IsEffectInSpell(spell_id, SE_InvisVsUndead9) || IsEffectInSpell(spell_id, SE_InvisVsUndead10) || IsEffectInSpell(spell_id, SE_InvisVsUndead11) || IsEffectInSpell(spell_id, SE_InvisVsUndead12) || IsEffectInSpell(spell_id, SE_InvisVsUndead13",
+      "reference": "I dig the helper function!"
+    },
+    {
+      "prediction": "I think this should be `package samples contains sample programs using the pubsub API.`",
+      "reference": "Please add the Copyright header."
+    },
+    {
+      "prediction": "This import should be moved to the top of the file.",
+      "reference": "Is this the best way to import `FileSystemTarget`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. The problem is that `thread_id_t` is 64-bits on x64, but `int32_t` is 32-bits on x64 and 16-bits on x86. I think the right fix is to change `READ_TLS_INT_SLOT_IMM` and `WRITE_TLS_INT_SLOT_IMM` to use `int64_t` instead of `int32_t`.",
+      "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": "Shouldn't this be `IfFailRet`?",
+      "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.exists(dagster_home_path)`",
+      "reference": "print out what we got here in the error"
+    },
+    {
+      "prediction": "I think this should be `static struct jx *select_expr = NULL;`",
+      "reference": "Does this need to be global?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `iterator().map(Tuple2::_2).toStream()`.",
+      "reference": "Yep, could be simplified. Now looks like other *Map.values() impls"
+    },
+    {
+      "prediction": "I think this should be `@Then /^print the last api response$/`",
+      "reference": "This method will be for debug?"
+    },
+    {
+      "prediction": "I think it would be better to have `Key(name, password string) (k *ecdsa.PrivateKey, created bool, err error)` return `k *ecdsa.PrivateKey, err error` instead of `k *ecdsa.PrivateKey, created bool, err error` and have `Exists(name string) (bool, error)` return `created bool, err error` instead of `exists bool, err error`.",
+      "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 think it would be better to have `PackageEntity` and `ClassEntity` inherit from `DiagramEntity` instead of `DiagramEntity` inheriting from `PackageEntity` and `ClassEntity`.",
+      "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 don't think we need to deprecate `add_prefix` and `add_suffix`.",
+      "reference": "These two functions should be available now."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. If the file is being updated, we should probably return an error. If the file is not being updated, we should probably continue reading from the file.",
+      "reference": "`fi` is what the result of Stat is called elsewhere in this file not `finfo`"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but it seems like it should still be tested.",
+      "reference": "I'm confused; are we just dropping these methods without deprecation?"
+    },
+    {
+      "prediction": "I don't think we need to print the version here.",
+      "reference": "can we use PC's time to version in Distributed Systems?"
+    },
+    {
+      "prediction": "I think this should be `c.user.full_name` instead of `c.user_full_name`",
+      "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 think this should be `>= 30` and `>= 50`",
+      "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 think this should be `W3C_MINIMIZE_WINDOW` instead of `w3cMinimizeWindow`.",
+      "reference": "No need for this to be `W3C_` since there is no JWP equivalent"
+    },
+    {
+      "prediction": "I think it would be better to use `require 'rspec/core/backport_random' unless defined?(::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 think this should be called `ListenerAddr` to avoid confusion with the `Addr` method on the `net.Listener` interface.",
+      "reference": "Is this ever actually used?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we need to pass the tipset to `ProcessTipSet` in order to get the correct receipt.",
+      "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 don't think we need the `IF_AARCHXX` here. The `set_stolen_reg_val` is only defined for ARM and AArch64.",
+      "reference": "Do we need to save the existing value of the stolen reg somehow?"
+    },
+    {
+      "prediction": "I think this should be `maxResults=max_results` and `max_results=500` in the docstring.",
+      "reference": "maxResults should come from FLAGS once #244 is submitted."
+    },
+    {
+      "prediction": "I think we can remove this `#if NETCOREAPP` and the corresponding `#endif` at the end of the file.",
+      "reference": "is this wanted?"
+    },
+    {
+      "prediction": "I think this should be `linode-python` 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 don't think we need to set the `Content-Type` header here. It's only used when sending a POST request to the server, and we don't do that here.",
+      "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 don't think we need this option. We can just remove the `--ignore-chromium` check in `update-chromium.js` and call it a day.",
+      "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": "Why do we need to hash the transfer and execution here?",
+      "reference": "`testExecutionPb` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "This method is not used anymore, so we can remove it.",
+      "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 `log.Debug(\"%v:\\n%s\", err, debug.Stack())`",
+      "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": "This should be `--rust-keep-namespaces` to be consistent with the other options.",
+      "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 `return nil, fmt.Errorf(\"open bucket %q: invalid query parameter %q\", u.String(), param)`",
+      "reference": "Should unsupported query parameters just be ignored? I think that's more common than error out."
+    },
+    {
+      "prediction": "I think this should be `Register.UNSET_NAMESPACE` instead of `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'm not sure this is the right thing to do here. I think it would be better to raise the exception and let the user know that something went wrong.",
+      "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": "This should be `github.com/abiosoft/caddy-minify`",
+      "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 don't think we need to keep a reference to the provider here. The provider will keep a reference to the tracer, and the tracer will keep a reference to the provider.",
+      "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`",
+      "reference": "Nit: from the Internet."
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "this is internal package, move to bottom and run 'make fmt'"
+    },
+    {
+      "prediction": "Should this be `IConnectionLifetimeNotificationFeature` or `IConnectionLifetimeNotification`?",
+      "reference": "Why is this better than ApplicationStopping?"
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "Maybe worth expanding that these 2 are supported by cert-manager but other values exist"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `getHelperProperties` method should only return the `src` option. The other options should be passed to the `render` method.",
+      "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 this should be `if (node.jsonRpcConfiguration().getAuthenticationPublicKeyFile() != null && node.jsonRpcConfiguration().getAuthenticationCredentialsFile() != null) {`",
+      "reference": "What if `node.jsonRpcConfiguration().getAuthenticationPublicKeyFile()` is empty string, would that cause a problem here?"
+    },
+    {
+      "prediction": "I don't think we want to add this to the spec. I think we should just use the defaults in the controller.",
+      "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 about this change. I don't think we should be using `equals` here. I think we should be using `==` instead.",
+      "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 `namespace` and `namespaceUID` from the `ns` struct and just use `ns.name` and `ns.namespaceLabels`.",
+      "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 `aws/codebuild/amazonlinux2-x86_64-standard:3.0` instead of `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 don't understand why this is needed. `linter.set_current_module(\"toto\")` should be enough.",
+      "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 don't think we need to create a new instance of JsonPrinter every time we call this method. We can create a single instance in the constructor and reuse it.",
+      "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 don't think we need to add COMMAND here. It's not a modifier key, it's a command key.",
+      "reference": "Keys.COMMAND seems to be an alias to Keys.META. That isn't mentioned?"
+    },
+    {
+      "prediction": "I don't think we need this interface. We can just call `SetSyncBlockCacheFraction` directly.",
+      "reference": "This should go in `ConfigLocal` (and then re-generate the mocks)."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way to get the tests to pass.",
+      "reference": "Nitty but entry.get for the same thing"
+    },
+    {
+      "prediction": "nit: I think this should be `// This is the golang entry point for runc init, executed before main() but after libcontainer/nsenter's nsexec().`",
+      "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 an `if (Constants.ECONNRESET.HasValue && Constants.EADDRINUSE.HasValue) { ... } else { _logger.LogWarning(\"Unable to determine ECONNRESET or EADDRINUSE value on this platform.\"); }`",
+      "reference": "nit: add new line after this block"
+    },
+    {
+      "prediction": "I think this should be 3.25rem instead of 3.25em.",
+      "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": "This import should be moved to the top of the file.",
+      "reference": "retrying is an installed 3rd party library just like jinja and sendgrid; perhaps try grouping all of them together?"
+    },
+    {
+      "prediction": "I think this should be `elements instanceof java.util.List` instead of `elements instanceof ArrayList || elements instanceof Vector` because `ArrayList` is a subclass of `java.util.List` and `Vector` is a subclass of `java.util.Vector` which is a subclass of `java.util.List`.",
+      "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 if this is the best way to do it, but I think it's the best we can do for now.",
+      "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 should be 2021.",
+      "reference": "this is a existing file so let's keep 2019"
+    },
+    {
+      "prediction": "I think this should be `type.getBitMask()` instead of `type`",
+      "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 don't think we need to use ReadOnlyTrieStore here. We can just use TrieStore and wrap it with ReadOnlyTxProcessingEnv.",
+      "reference": "AsReadOnly() would be better here"
+    },
+    {
+      "prediction": "I think it would be clearer to say that `insertions` and `newModifications` are indices into the new version of the collection, and `deletions` and `oldModifications` are indices into the old version of the collection.",
+      "reference": "I suggest that we use `query_based_sync` instead of `partial_sync`."
+    },
+    {
+      "prediction": "I think this should be in the README, not the godoc.",
+      "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": "nit: `initCodeTransformer` should be `initCodeTransformerSupplier` to make it clear that it's a supplier and not the transformer itself.",
+      "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'm not sure this is true. I don't think you can open a collection with a zero-value key.",
+      "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 `self.__set_up_test_dir('ctu_on_demand_failure')`",
+      "reference": "Why do we have to change the test dir?"
+    },
+    {
+      "prediction": "I think this should be `!el.hasAttribute('open')` instead of `!el.querySelector('summary')`.",
+      "reference": "This should test the flattened tree instead. details > summary works across shadow tree boundaries."
+    },
+    {
+      "prediction": "This should be `Mvx.Resolve<IMvxLogProvider>().GetLogFor(typeof(MvxAndroidLog))`",
+      "reference": "Not sure I am a big fan of these duplicated Log classes."
+    },
+    {
+      "prediction": "I don't think we need these constants. They are only used in one place, and we can just use the strings directly.",
+      "reference": "It's long-winded, but the convention is to give constants a common prefix which, see `reasonDomainVerified` ."
+    },
+    {
+      "prediction": "I think this should be 30.0. 20.0 is too short.",
+      "reference": "Using lanes later down, this threshold could be reduced for similar effects. Otherwise we look a bit to far."
+    },
+    {
+      "prediction": "I think this should be `char *tarball; /* Conda environment as produced by conda-pack. */` and `char *expansion; /* Directory in cache/. Only set when environment has been expanded. */`",
+      "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": "nit: `lastPageInPaginatedResp` -> `isLastPageInPaginatedResp`",
+      "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": "nit: remove the empty line",
+      "reference": "What about using `dataManifests == null || deleteManifests == null`?"
+    },
+    {
+      "prediction": "I think it would be better to pass the `TOTAL_HITS_THRESHOLD` to the constructor of `GlobalHitsThresholdChecker` instead of creating a new instance of `GlobalHitsThresholdChecker` every time `newCollector` is called.",
+      "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": "nit: please remove the empty line here.",
+      "reference": "move this to below `antrea.io` import section"
+    },
+    {
+      "prediction": "I don't think we need to say \"predate ActivitySource\" here. We can just say \"predate OpenTelemetry.Trace\".",
+      "reference": "Seems likely to confuse people. What if we moved it into ActivitySourceAdapter and made it private?"
+    },
+    {
+      "prediction": "I think this should be `if len(signedTx.Data()) > 0 && signedTx.Value().Cmp(common.Big0) != 0 { return errors.New(\"Cannot send transactions with value and calldata\") }`",
+      "reference": "gas price don't need to be a multiple cuz we support allll gas prices now"
+    },
+    {
+      "prediction": "Can we rename this to OnClientProcessTerminated or something like that?",
+      "reference": "We are exposing implementation details in the interface. What if there is no processes involved in an implementation of `ITestRequestSender`?"
+    },
+    {
+      "prediction": "Please use `str(e)` instead of `e.message`.",
+      "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": "This should be `Customize theme`",
+      "reference": "This already exists under the `theme` lang key, please remove this."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "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": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I believe that you end up printing out the result of the assignment here instead of the netmask."
+    },
+    {
+      "prediction": "I think it would be better to have these constants in a separate file (e.g. `constants.js`) so that they can be used in other tests as well.",
+      "reference": "We have to update the filename here."
+    },
+    {
+      "prediction": "I think this should be `snake_case` instead of `snake_case`.",
+      "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 don't think we should do this. If the user wants to discard a hotfix, they should be able to do so. If they don't want to discard it, they should be able to keep it.",
+      "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 don't think we need to import `SECP256K1` here.",
+      "reference": "In my opinion I think it is possible to add `private final` here"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we want to jump to the `exit` label, but I don't think we want to jump to the `exit` function. I think we want to jump to the `exit` function, but I don't think we want to jump to the `exit` label. I think we want to jump to the `exit` label, but I don't think we want to jump to the `exit` function. I think we want to jump to the `exit` function, but I don't think we want to jump to the `exit` label. I think we want to jump to the `exit` label, but I don't think we want to jump to the `exit` function. I think we want to jump to the `exit` function, but I don't think we want to jump to the `exit` label. I think we want to jump to the `exit` label, but I don't think we want to jump to the `exit` function. I think we want to jump to the `exit` function, but I don't think we want to jump to the `exit` label. I think we want to jump to the `exit`",
+      "reference": "Feel like \"exit\" doesn't convey that the packet will be dropped. \"drop-and-exit\" or \"error-exit\" maybe?"
+    },
+    {
+      "prediction": "I'm not sure this is correct. The `create` constructor of a `Main` actor must take a single `Env` parameter. The `create` constructor of any other actor must not take a single `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 don't think we need to add a new permission for this. We can just use `READ_PERM` and `UPDATE_PERM` instead.",
+      "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 don't think we need to add `list` here. `np.ndarray` should be enough.",
+      "reference": "Is this reasonable?"
+    },
+    {
+      "prediction": "I think it's better to remove this line.",
+      "reference": "We may move this TODO to Line15."
+    },
+    {
+      "prediction": "This should be `if (true === ConfigQuery::checkAvailableStock()) {`",
+      "reference": "You have to verify the return type. If it's a reponse, return it. Otherwise do nothing."
+    },
+    {
+      "prediction": "I don't think we need to change this to `int32_t` and `int64_t`. I think we can just leave it as `int`.",
+      "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": "Shouldn't this be `max({})` and `min({})`?",
+      "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 should be updated to check that `long_r2` is not present in `projectedL1`.",
+      "reference": "Why did this need to change?"
+    },
+    {
+      "prediction": "why do we need this interface? why not just use `protocol.StateDB`?",
+      "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 `COLLECTION_CONFIGNAME_PROP`",
+      "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 think we should use `assertThat(check.body().string()).isEqualTo(\"{\\\"status\\\":\\\"UP\\\",\\\"zipkin\\\":{\\\"status\\\":\\\"UP\\\",\\\"details\\\":{\\\"InMemoryStorage{}\\\":{\\\"status\\\":\\\"UP\\\"}}}}\");` instead of `assertThat(check.body().string()).isEqualTo(\"{\\\"status\\\":\\\"UP\\\",\\\"zipkin\\\":{\\\"status\\\":\\\"UP\\\",\\\"details\\\":{\\\"InMemoryStorage{}\\\":{\\\"status\\\":\\\"UP\\\"}}}}\\n\");`",
+      "reference": "in preparation of directly writing the /health endpoint one day.. we should have content checks."
+    },
+    {
+      "prediction": "I think this should be `e.prefix` instead of `%q`. `%q` is for quoted strings.",
+      "reference": "question (non-blocking): Is there a benefit from using `%q` instead of `%s`?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "I would suggest to keep two empty lines between functions in cpp files"
+    },
+    {
+      "prediction": "I think this should be `true`?",
+      "reference": "This change causes the test to fail on both rocm-1.9.x and rocm-head."
+    },
+    {
+      "prediction": "I don't think we need this interface. We can just use `RuleViolationFix` as the return type of `getRuleViolationFixer()` in `AutoFixableRuleViolation`.",
+      "reference": "I'd consider moving all fix related stuff to a distinct package to avoid contaminating the base package"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(entry['directory'], inc_dir)` instead of `os.path.join(entry['directory'], entry['directory'], inc_dir)`",
+      "reference": "Use `os.path.join` for path concatenation."
+    },
+    {
+      "prediction": "I think it's better to put `YolactHead`, `YolactSegmHead` and `YolactProtonet` together.",
+      "reference": "Use upper case: YOLACTHead, YOLACTProtonet, YOLACTSegmHead"
+    },
+    {
+      "prediction": "I think this should be `&Crypter{...}` instead of `&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 don't think this is the right place for this message. I think it should be in the `Type::getTypeFromTree` method.",
+      "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 fix. I think we need to figure out why we're getting a `ClassCastException` in the first place.",
+      "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": "Style/RedundantArrayCoercion: Redundant array coercion detected.",
+      "reference": "The `[] +` seems weird to me. Why is it there?"
+    },
+    {
+      "prediction": "Line is too long. [85/80]<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.",
+      "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'm not sure this is the best way to do this, but it works for now.",
+      "reference": "What about a line with just spaces/tabs? Or a line with a couple spaces followed by a `#`?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [128/120]",
+      "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": "This should be `if (GetTarget() != nullptr && GetTarget() == this) {`",
+      "reference": "Don't need this-> here."
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "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 right thing to do, but it seems to be the only way to make the tests pass.",
+      "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 don't think this is the right place to do this. I think it should be done in the parser.",
+      "reference": "FYI, It won't have any benefit to move from trivial types."
+    },
+    {
+      "prediction": "I think this should be `capabilitySecretEnvSSM` instead of `capabilitySecretEnvSsm`",
+      "reference": "i missed these discussions- but ` \"secrets-ssm-environment-variables\"` is what was agreed upon with cp?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `gogoproto` import since we don't use it anymore.",
+      "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'm not sure this is the right way to do it, but it's the only way I could get it to work.",
+      "reference": "Just asking: What does this line do? Does it add the stylesheet from the parent to the dialog?"
+    },
+    {
+      "prediction": "I think this should be `scikit-learn<1.0.0; python_version < '3.7'`",
+      "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": "why remove this check?",
+      "reference": "Why remove this?"
+    },
+    {
+      "prediction": "I think we can remove this test as well.",
+      "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 `global._googlesitekitUserData?.capabilities || {}`.",
+      "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 `> 0`",
+      "reference": "Please add a space after the comma here (generally, with arguments there's always a space after commas)."
+    },
+    {
+      "prediction": "I think this should be \"cautiously\" rather than \"caution\".",
+      "reference": "From the context, should it say \"Flags defined in this class must be used with caution...\" ?"
+    },
+    {
+      "prediction": "Do we need this? I don't think we do.",
+      "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 `Exported key for account %s: \\\"%s\\\"`",
+      "reference": "\"Exported key for account\"?"
+    },
+    {
+      "prediction": "I think this should be `\"\\r\\n\"`, not `\"\\n\\r\"`.",
+      "reference": "Normally we use Environment.NewLine unless you really want it to always be \\r\\n"
+    },
+    {
+      "prediction": "Why is this removed? It's still used in `consensus/ethash/ethash.go` and `consensus/clique/clique.go`",
+      "reference": "why did you remove `verifyHeaders`?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [125/120]",
+      "reference": "Thanks for cleaning up these deprecated calls"
+    },
+    {
+      "prediction": "I think this should be \"Invalid username\" as well.",
+      "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 should keep `Service` here.",
+      "reference": "We support process remoteQuery for Service. Why are we removing it ?"
+    },
+    {
+      "prediction": "I think this should be a 500 error, not a 200.",
+      "reference": "This is specifically for testing?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "I think this is related to the social stats and probably does not need to be removed."
+    },
+    {
+      "prediction": "I don't think we need to disable this rule for the whole file.",
+      "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 this should be `if len(values) == 0 && !result.NotDone() { return \"\", errs.New(\"principal %q not found\", principalID) }`",
+      "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 don't think this is the right way to do it. If the hash is not a block or an execution, it should return an error.",
+      "reference": "unknown field AddressDetails in struct literal (from `typecheck`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to be the only way to set the region in the session.",
+      "reference": "Just noticed, are we specifically missing the addition of `SharedConfigState: session.SharedConfigEnable` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the old time intensities are empty, then we don't know what the old time was, so we can't find the nearest time in the new time intensities. In this case, I think we should just set `ScanIndex` to 0.",
+      "reference": "Not sure what this is. Bad merge?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should still return `ErrAccountResolverSameClaims` in this case.",
+      "reference": "I think this may be a RW lock."
+    },
+    {
+      "prediction": "I think this should be `helpers/pluginhelper` so that it's clear that this is a sub-package of `helpers`.",
+      "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": "I'm not sure this is the right place to do this check. I think it should be done in the `GetTradeRecipe` function.",
+      "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 we should add `copy=False` as the default value for `copy` argument.",
+      "reference": "too long here. (104 > 100)"
+    },
+    {
+      "prediction": "I think it would be better to throw an exception here and let the caller decide what to do with it.",
+      "reference": "Use TryGetValue instead of catching exception"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's better to keep it as `Add new` since that's what the button says.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "This should be `FASTRTPS_STATISTICS` instead of `FASTDDS_STATISTICS`.",
+      "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 think this should be `Http2ErrorCode.CANCEL`",
+      "reference": "You add a stream back reference for trailers, and all of the sudden we have tight coupling!"
+    },
+    {
+      "prediction": "nit: can you remove this empty line?",
+      "reference": "I don't think we use `AspNetAmbientContext` anymore since . We can probably delete this file."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be reverted.",
+      "reference": "Please avoid committing build artifacts, as they produce unnecessary conflicts and noise."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to work.",
+      "reference": "this is vendored code so i would hesitate to make random changes to it"
+    },
+    {
+      "prediction": "I don't think we need to add the `required: true` here, since it's already required in the `payoutInfo` struct.",
+      "reference": "We are using a different style for JSON tags. `ethAddress` - camalCase `referral_code` - snake_case"
+    },
+    {
+      "prediction": "I think this should be `System.Net.Http.HttpRequestMessage`",
+      "reference": "Should this be `\"Microsoft.AspNetCore.Http.DefaultHttpContext\"`?"
+    },
+    {
+      "prediction": "I don't understand why this assertion was removed.",
+      "reference": "removed asserts from `Vector` as it's stable enough and it may hinder inlining, even if turned off :/"
+    },
+    {
+      "prediction": "I don't think we need to disable the lru_cache here. We can just remove the `@functools.lru_cache()` decorator.",
+      "reference": "Do we want to raise on the function or on the decorator?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do for now. We don't have a `release_msid` field in ListenBrainz, but we do have an `album_msid` field in ListenBrainz. I think it's better to use the `album_msid` field in ListenBrainz than to create a new `release_msid` field in ListenBrainz. We can always add a new `release_msid` field in ListenBrainz in the future if we want to.",
+      "reference": "How complex is to to replace `album_*` with `release_*` everywhere in the `Listen` class and places that use it?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the namespace already exists, we should not add it again.",
+      "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 don't think we should be caching this value on the `vNode` object. We should be caching it on the `el` object.",
+      "reference": "@straker should this not push to `vnode._cache.isHiddenWithCSS`?"
+    },
+    {
+      "prediction": "nit: I think this should be \"the runtimevar package uses to interact with the underlying runtime configuration services.\"",
+      "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 about this, but I think it would be better to have `ioError` and `connectError` as separate conditions.",
+      "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'm not sure I understand the purpose of `doSync` here. It seems like it's only used in `checkCaches` and `checkCaches` is only called by `PutInCaches` which always passes `false` for `doSync`.",
+      "reference": "Can you make this accept an action instead?"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "Tab inconsistency here (3 spaces instead of 2 spaces)"
+    },
+    {
+      "prediction": "I think this should be `unit.ConvertIotxToRau(200000000).String()`",
+      "reference": "these change means we are changing epoch to 1 hour?"
+    },
+    {
+      "prediction": "This should be `PANEL_DESCRIPTION_WIDTH = 544 * self.scaleSize`",
+      "reference": "Why this is defined in the middle of imports?"
+    },
+    {
+      "prediction": "I don't think we need to import `importlib` here.",
+      "reference": "Could you move this to the original position?"
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it should be `oAndR.repo`.",
+      "reference": "Is there a test that'd have caught this?"
+    },
+    {
+      "prediction": "nit: can you remove the empty line here?",
+      "reference": "ListWorkflowExecutionsRequest <- this can be nil?"
+    },
+    {
+      "prediction": "I think this should be `forgeapi.puppet.com` instead of `forgeapi.puppetlabs.com`",
+      "reference": "Do we need to continue to support the old link, or is it dead dead dead?"
+    },
+    {
+      "prediction": "I think it would be better to use `Trim()` instead of `Trim()`.",
+      "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 string: %s\", config.GetOSFamilyType())`",
+      "reference": "this seems like more of a Debugf statement"
+    },
+    {
+      "prediction": "I think this should be `cb_state->queue_submit_functions.emplace_back` instead of `cb_state->eventUpdates.emplace_back`.",
+      "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.UseDualStackEndpoint != nil || req.Config.UseFIPSEndpoint != nil {`",
+      "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 `expect(fn).to.throw(/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": "nit: `nil` instead of `nil`",
+      "reference": "I think this is added to wrong file. You wanted to add to cmd/agent/, right?"
+    },
+    {
+      "prediction": "I think this should be `elementText != null && elementText.contains(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 a `string` instead of an `int`.",
+      "reference": "Could we consider a more explicit name, like `devicePciAddress`?"
+    },
+    {
+      "prediction": "I think this should be `cmdarray[i]` instead of `cmdarray[1]`.",
+      "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'm not sure this is the right thing to do. I think it should be `simplexml_load_string($response->getBody())`",
+      "reference": "Is this supposed to be part of this PR?"
+    },
+    {
+      "prediction": "I think it would be better to use `actualNode.style.display === 'none'` instead of `window.getComputedStyle(actualNode, null).getPropertyValue('display') === 'none'` because `window.getComputedStyle(actualNode, null).getPropertyValue('display') === 'none'` will always return `false` if the element is not attached to the DOM.",
+      "reference": "DRY, worth extracting this method to axe.utils"
+    },
+    {
+      "prediction": "I think this should be `APPLICATION_NAME` instead of `applicationName`.",
+      "reference": "I think there's another spot for this in DefaultCapabilityMatcher"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this error message. I think it should be in the `savefile_open` function itself, so that it can be used in other places as well.",
+      "reference": "`QtOSError` inherits `OSError`, so I don't think it's needed to list them both here."
+    },
+    {
+      "prediction": "I think we can remove this now.",
+      "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": "This should be `[]string{\"BUILD\", \"BUILD.bazel\"}`.",
+      "reference": "Don't think this should be here. The default is set somewhere else."
+    },
+    {
+      "prediction": "I don't think this is a good idea. If the test was stopped by the user, we should not terminate it.",
+      "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 think this should be `Optional<Integer> sortOrderId()`. I don't think it makes sense to have a default value of `null`.",
+      "reference": "nit: `<p>` after the line"
+    },
+    {
+      "prediction": "I think this should be `glob(\"fonts/*.ttf\")` instead of `glob(\"fonts/*.ttf\")`.",
+      "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": "why remove this check?",
+      "reference": "i think it's OK to leave this?"
+    },
+    {
+      "prediction": "Why do we need to make this class public?",
+      "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": "I don't think this is necessary.",
+      "reference": "No line at end of file warning"
+    },
+    {
+      "prediction": "I don't think it's a good idea to create a new instance of `CompositeBlockTracerFactory` here. I think it should be passed as a constructor parameter.",
+      "reference": "This looks unused, shouldn't it point to inner processor in any way?"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "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'm not sure this is the right place to do this. I think it should be done in `GrpcClientGenerator`.",
+      "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 we can remove this now, right?",
+      "reference": "All stories in this file also need the default padding."
+    },
+    {
+      "prediction": "This should be `nameof(PrometheusExporterMiddlewareTests) + \".\" + nameof(PrometheusExporterMiddlewareTests.PrometheusExporterMiddlewareIntegration)`",
+      "reference": "Curious - what would be the actual value?"
+    },
+    {
+      "prediction": "`datasets[0].PALETTE` should be `datasets[0].metainfo.get('palette', None)`",
+      "reference": "Is it necessary to record the palette in the checkpoint? Any reason?"
+    },
+    {
+      "prediction": "This import should be moved to the top of the file.",
+      "reference": "Not sure how to best do this without creating a new connection to the server everytime. Do this in another thread?"
+    },
+    {
+      "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.<br>",
+      "reference": "should we have a similar test for gsa18f procurements?"
+    },
+    {
+      "prediction": "Why is this change necessary? I don't see it in the PR description.",
+      "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 it would be better to have `TracerTimeoutEnabled` as `bool` and `TracerTimeout` as `TimeSpan`.",
+      "reference": "remove this and always enable tracer, just set the default timeout to something higher (20 seconds)"
+    },
+    {
+      "prediction": "I think it would be better to use `Optional<URI>` instead of `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 `command:\"filter\" short:\"f\" description:\"Filter the given set of targets according to some rules\"`",
+      "reference": "can't these just use the global `include` and `exclude` flags?"
+    },
+    {
+      "prediction": "Why do we need to move this to GPU?",
+      "reference": "Is this still necessary?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get it to work.",
+      "reference": "Interesting! Happen to catch an exception message or anything I can look into?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think it's better to have two separate methods, `addCallback` and `addErrback`.",
+      "reference": "did this get moved to mpromise? I don't recall if it's in that lib or not."
+    },
+    {
+      "prediction": "I think this should be `[]int` instead of `[]int32`.",
+      "reference": "json objects should be `camelCase`"
+    },
+    {
+      "prediction": "I don't think we need to have this as a separate type. We can just use `uint64` directly.",
+      "reference": "QuarterGiBSectorSize? SectorSize265MiB? Spelling out 256 seems overly verbose."
+    },
+    {
+      "prediction": "I think this should be `using OpenTelemetry.Tests;` instead of `using OpenTelemetry.Tests.Logs;`",
+      "reference": "Maybe with this change we can remove `using OpenTelemetry.Logs`."
+    },
+    {
+      "prediction": "`r.FormValue(\"ViewerRoleAsDefault\") == \"on\"` should be `r.FormValue(\"ViewerRoleAsDefault\") == \"on\" || r.FormValue(\"ViewerRoleAsDefault\") == \"\"`",
+      "reference": "Fix this name too."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `int()` function will raise a `ValueError` if the string is not a valid integer. You should use `int()` inside a `try`/`except` block.",
+      "reference": "why this is an issue?"
+    },
+    {
+      "prediction": "I don't think this is correct. If `$was_static` is true, then `$value === $this_class` is always false.",
+      "reference": "The condition should be inverse?"
+    },
+    {
+      "prediction": "I don't think we need this constant. We can just use `VERSION_2_0` directly.",
+      "reference": "this seems to be a property key name, so should it be a inside configurationkey.java?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get `zfs destroy` to work.",
+      "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' && dimensionFilters[ dimension ].length > 0 )`",
+      "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": "nit: I think we can get rid of the `unaryTransportHandler` type and just use `yarpc.UnaryTransportHandler` directly.",
+      "reference": "let's do a pointer cast of nil, similar to the `MapRouter` above"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place for this hook. I think it would make more sense to have it in the `SettingsManager::add_setting` method.",
+      "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 think this should be `if (sslOptions.ApplicationProtocols.Contains(SslApplicationProtocol.Http2)) { sslOptions.AllowRenegotiation = false; } else { sslOptions.AllowRenegotiation = true; }`",
+      "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": "`visibility_of_any_element_located`?",
+      "reference": "shouldn't **call** return a boolean?"
+    },
+    {
+      "prediction": "I don't think we need to add a blank import here.",
+      "reference": "I don't think you need the blank import here. This one uses awsdynamodb directly."
+    },
+    {
+      "prediction": "I think we can just return `t -> false` instead of `isDeleted` here.",
+      "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 don't think this is the right place to set `ld_lib_path_extra` and `path_env_extra`. I think they should be set in `__build_clangsa_config_handler` and passed to `__build_clangsa_config_handler`.",
+      "reference": "What is this and the next entry in config_handler used for?"
+    },
+    {
+      "prediction": "I think this should be `NewEmptyMemoryPlugin` to be consistent with the naming of the other functions.",
+      "reference": "will your editor integrate `gofmt`, `goimports`, etc...?"
+    },
+    {
+      "prediction": "I think you can use `Objects.requireNonNull(intent.getExtras().getSerializable(\"state\"))`",
+      "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 it would be better to have this as `--min-bug-path-length` and `--max-bug-path-length`.",
+      "reference": "Please add some valid examples to help message like `\"4:10\"`, `\"4:\"`, `\":10\"`"
+    },
+    {
+      "prediction": "Extra empty line detected at block body end.",
+      "reference": "Is it only Symbols we're worried about converting? Is it not safe to just call `to_s` anyway?"
+    },
+    {
+      "prediction": "I think we should keep this constant, as it's used in a lot of places.",
+      "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'm not sure about this change. I think it's better to keep the `incomplete_trails` method returning only the incomplete trails, and add a `in_progress_trails` method that returns only the in-progress trails.",
+      "reference": "What about `sort_by(&:in_progress?)`? Maybe with a `.reverse` thrown in?"
+    },
+    {
+      "prediction": "I think we should keep this flag around for backwards compatibility.",
+      "reference": "Flag is removed because it is only gRPC now."
+    },
+    {
+      "prediction": "I think this should be `.. program:: molecule init role --role-name foo --template 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 think this should be `position=1` for `write_text` and `position=2` for `write_bytes`.",
+      "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": "This import is not used in this file.",
+      "reference": "why depend on recovery???"
+    },
+    {
+      "prediction": "I'm not sure why this change is needed. I don't see anything in `dynamicconfig.NewFileBasedClient` that uses the type of `doneChan`.",
+      "reference": "the done channel only aims to be closed, so `chan struct{}` is better then `chan interface{}`"
+    },
+    {
+      "prediction": "nit: I think this should be `error creating gas price oracle contract binding`",
+      "reference": "Do you mean to have gas_oracle changes in this PR?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Is this effectively requiring browsers support javascript?"
+    },
+    {
+      "prediction": "I think you can just use `SalesforceSDKManager.getInstance().isDarkTheme()` instead of `SalesforceSDKManager.getInstance().isDarkTheme(getActivity())`",
+      "reference": "Lowercase `boolean` - use the primitive type."
+    },
+    {
+      "prediction": "I think this should be `Log.LogError($\"Item '{item}' already exists in {relativeBlobPath}.\" + Environment.NewLine + exc.Message);`",
+      "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 think this should be `else if` instead of `if`.",
+      "reference": "Can you replace these by `else if` please?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to go about it, but it seems to be working for now.",
+      "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": "`includes` is not supported in IE11. We should use `indexOf` instead.",
+      "reference": "Does this code fit on one line under 80 characters? It would be more consistent with our existing style."
+    },
+    {
+      "prediction": "I don't think we need to do this. The CASTs in 0.7.0 are the same as in 0.6.0, so we can just use the 0.6.0 CASTs for 0.7.0.",
+      "reference": "Not a requirement here. But i see a change in naming convention. We can rename the function to `CstorSparsePoolArtifactsFor070`."
+    },
+    {
+      "prediction": "I think this should be `machine.D10` instead of `machine.GP10`. GP10 doesn't exist on the Pico.",
+      "reference": "Why do you configure the UART like this? It's already configured by default."
+    },
+    {
+      "prediction": "I think this should be `#include \"opae/enum.h\"` and `#include \"opae/properties.h\"`",
+      "reference": "Wow. Good catch. \"This `#include` was brought to you by the department of redundancy department.\" ;)"
+    },
+    {
+      "prediction": "nit: I think we should use `logf.Logger` instead of `logr.Logger` here.",
+      "reference": "Well it's actually calling `AuthorizeOrder` and this seems like a DebugLevel message to me, maybe it should be dropped entirely."
+    },
+    {
+      "prediction": "I think this should be `if ((ptr == nullptr) || (bytes == 0) || (attr.type == cudaMemoryTypeUnregistered)) return;`",
+      "reference": "Should this be using the `CUDA_SAFE_CALL` macro here?"
+    },
+    {
+      "prediction": "I think this should be `Console.ForegroundColor = ConsoleColor.White;` instead of `Console.ForegroundColor = ConsoleColor.Gray;`",
+      "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 if this is the right thing to do, but I think it's the right thing to do. If the mirror is nil, then there's nothing to do.",
+      "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 think this should be `pixman_region32_copy(&state->opaque, &next->opaque);` instead of `pixman_region32_copy(&state->opaque, &next->opaque);`",
+      "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 done in the `processClientOrLeafAuthentication` function. The reason is that the `processClientOrLeafAuthentication` function is the one that checks if the account has the issuer. If the account doesn't have the issuer, then the `processClientOrLeafAuthentication` function should return false. If the account has the issuer, then the `processClientOrLeafAuthentication` function should return true. If the account has the issuer, then the `processClientOrLeafAuthentication` function should return true. If the account has the issuer, then the `processClientOrLeafAuthentication` function should return true. If the account has the issuer, then the `processClientOrLeafAuthentication` function should return true. If the account has the issuer, then the `processClientOrLeafAuthentication` function should return true. If the account has the issuer, then the `processClientOrLeafAuthentication` function should return true. If the account has the issuer, then the `processClientOrLeafAuthentication` function should return true. If the account has the issuer, then the `processClientOrLeafAuthentication` function should return true. If the account has the issuer, then the `",
+      "reference": "already what? I know what, but seems like smth. is missing in this sentence"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "Can you put spaces after the `{`and before the `}`"
+    },
+    {
+      "prediction": "I think this should be `%(version)s` instead of `%(__version__)s`",
+      "reference": "We said we would put `[ ]` around the version. `[forseti-security] [v2.3.0]`"
+    },
+    {
+      "prediction": "I think this should be `len(cnb) > 0 && n >= int64(cap(cnb[0]))`",
+      "reference": "Seems len(cnb[0]) will be the length after the write since of the *v thing we saw no?"
+    },
+    {
+      "prediction": "I think this should be `interface{}` instead of `interface{}`.",
+      "reference": "can we make use of storagepoolclaim type rather than the generic interface{} type?"
+    },
+    {
+      "prediction": "I don't think we need to do this here. The peers will be added in the `manage` loop.",
+      "reference": "this should happen before the `go k.manage()`, otherwise the node will always try the bootnodes first"
+    },
+    {
+      "prediction": "I think this should be `(?=[^\\\\]%[nbs])` instead of `(?=[^\\\\]\\%[nbs])`",
+      "reference": "Is this actually a lint related change?"
+    },
+    {
+      "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 don't think we need the `default` argument here.",
+      "reference": "what does this do that str_param does not?"
+    },
+    {
+      "prediction": "I think this should be `AzureUserAgentPrefix()` instead of `AzureUserAgentPrefix`.",
+      "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 think this should be `from botocore.exceptions import BotoCoreError, ClientError`",
+      "reference": "what would you think about doing this include on-demand within S3ObjectStore methods and then failing in a loud way?"
+    },
+    {
+      "prediction": "I think this should be outside of the loop, otherwise it will break out of the loop on the first iteration.",
+      "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 don't think this is correct. `future_addEdgesAtomic` and `future_chainAddEdges` are not the same. `future_addEdgesAtomic` is a single RPC call, while `future_chainAddEdges` is a chain of RPC calls.",
+      "reference": "In `src/storage/InternalStorageServiceHandler.h ` file, the client in `client->future_chainAddEdges ` should be InterStorageServiceAsyncClient?"
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "`go.sum` needs to be updated for this. Please run `go mod tidy`"
+    },
+    {
+      "prediction": "I think this should be `SecretBackendAWSParameterStore` and `SecretBackendAWSSecretsManager` to be consistent with the rest of the codebase.",
+      "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": "Why is this removed?",
+      "reference": "Same thing with the OtlpExporter"
+    },
+    {
+      "prediction": "I think this should be `case JX_OP_CALL: return \"(\";`",
+      "reference": "Should not be removed."
+    },
+    {
+      "prediction": "I think this should be `require 'beaker-puppet/install_utils'` instead of `require 'beaker-puppet'`",
+      "reference": "Not a blocker for anything, but should this be moved to beaker-pe? Is that ticketed anywhere?"
+    },
+    {
+      "prediction": "This should be `afterInteractingWith(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 this is the best way to do this, but I think it's the least bad option we have right now.",
+      "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 think this should be `@Deprecated` instead of `@InternalApi`.",
+      "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 `gridded = True` and `multi = False` for the `Gridded` interface.",
+      "reference": "Does it have to be arrays? Isn't it ragged 'data' (i.e multiple elements of different lengths)?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "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 `SparkStreamingWrite extends BaseBatchWrite implements StreamingWrite`",
+      "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": "I'm not sure about this, but I think it would be better to keep the old names and add the new ones as aliases.",
+      "reference": "Shouldn't this be called `hp` instead of `life`?"
+    },
+    {
+      "prediction": "I don't think this is the right change. The `allowedOrigins` parameter should be set to `*` and the `allowedMethods` parameter should be set to `GET,POST,PUT,DELETE,HEAD`.",
+      "reference": "Because the default value of allowedOrigins is * (all origins), so it isn't necessary to set again at all."
+    },
+    {
+      "prediction": "Please remove empty lines at the end of the file.",
+      "reference": "Undo this modification"
+    },
+    {
+      "prediction": ",\"\"boolKey\"\":true,\"\"library.name\"\":\"\"CreateTestActivity\"\"}}}}]\",\n                 json);\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 `else if (settings.getResultType() == CameraResultType.BASE64) { returnBase64(call, exif, bitmapOutputStream); } else if (settings.getResultType() == CameraResultType.URI) { returnFileURI(call, exif, bitmap, u, bitmapOutputStream); } else if (settings.getResultType() == CameraResultType.BASE64NOMETADATA) { returnBase64NoMetadata(call, exif, bitmapOutputStream); } else { call.reject(INVALID_RESULT_TYPE_ERROR); }`",
+      "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": "nit: I think we can remove the `project` parameter from the `ovfimportparams.OVFImportParams` struct.",
+      "reference": "Can you explain the history of the bug a bit, and why this fixes it?"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "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": "nit: I think we can use `metricPoint.Keys == null ? 0 : metricPoint.Keys.Length`",
+      "reference": "Looks like this will make the check in every loop. Consider extracting the null check."
+    },
+    {
+      "prediction": "I think it would be good to have a test that verifies that `Registry.clear` is idempotent, i.e. that it can be called multiple times in a row without raising an error.",
+      "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": "nit: I think we can remove the `ssmfactory` import and just use `factory` in the `ssm` package.",
+      "reference": "naming: please use `asmfactory`"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the output is a directory, this will break. I think we need to make sure that the output is a file before we do this.",
+      "reference": "FYI you could use a raw string for this which removes the need for escaping the inner quotes."
+    },
+    {
+      "prediction": "I think it would be good to have a few more tests here. For example: - `<Ctrl-Alt-x>` - `<Ctrl-Alt-Shift-x>` - `<Ctrl-Alt-Shift-Alt-x>` - `<Ctrl-Alt-Shift-Alt-Shift-x>` - `<Ctrl-Alt-Shift-Alt-Shift-Alt-x>` - `<Ctrl-Alt-Shift-Alt-Shift-Alt-Shift-x>` - `<Ctrl-Alt-Shift-Alt-Shift-Alt-Shift-Alt-x>` - `<Ctrl-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-x>` - `<Ctrl-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-Alt-x>` - `<Ctrl-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-x>` - `<Ctrl-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-Alt-x>` - `<Ctrl-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-x>` - `<Ctrl-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift-Alt-Shift",
+      "reference": "2 blank lines here (between functions)"
+    },
+    {
+      "prediction": "I think this should be `super(tableName, ops, ops.current().spec().withSortOrder(ops.current().sortOrder()));`",
+      "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, 2) === '::'`",
+      "reference": "As mentioned below, I'm not sure why this is only allowing refs that are tagged with `*` at the start!"
+    },
+    {
+      "prediction": "Shouldn't this be `delegatedRepository.equals(that.delegatedRepository)`?",
+      "reference": "Don't forget the {} symbols, and please use EqualsBuilder and HashcodeBuilder as possible"
+    },
+    {
+      "prediction": "I'm not sure why we need to create a new buffer for each read. I think we can just reuse `proxyBuf` here.",
+      "reference": "Allocate buffer once outside loop."
+    },
+    {
+      "prediction": "I don't think this test is necessary.",
+      "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 don't think this is the right place to put this message. I think it should be in `Access.java` where the exception is thrown.",
+      "reference": "Should this be in a bundle?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be `Activity.Current = null;`",
+      "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 if this is the right place to call `RegisterHandlers`, but it seems to be the only place where the handlers are registered.",
+      "reference": "what was the point in moving handlers registration into a separate method?"
+    },
+    {
+      "prediction": "I think this should be `if state.Config.Cache.RetrieveCommand != \"\" && state.Config.Cache.StoreCommand != \"\"`",
+      "reference": "At this point we probably want to ensure there's a store command set."
+    },
+    {
+      "prediction": "`step[step.length-1] == \"'\"` should be `step[step.length-1] === \"'\"`",
+      "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 it be `Vault message should not be in hex; message: {message.Substring(2)}`?",
+      "reference": "Don't throw base exception, specify more precise type"
+    },
+    {
+      "prediction": "`Inputs that to be casted` -> `Inputs that need to be casted`",
+      "reference": "Recursive -> Recursively converted -> convert"
+    },
+    {
+      "prediction": "I think this should be `opts.generate_object_based_api || has_no_struct_fields`",
+      "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 don't think we need to change this.",
+      "reference": "just to confirm, so backend will deal with both old format and new format?"
+    },
+    {
+      "prediction": "I'm not sure why this is needed, but it seems to be required for the compiler to know that `T` is an enum.",
+      "reference": "huh, look at that, I wasn't aware you could constrain a type by Enum"
+    },
+    {
+      "prediction": "I don't think we need the `const` here.",
+      "reference": "Even though the debug parameter is effectively ignored, should this default to `false` instead of `true`?"
+    },
+    {
+      "prediction": "I think we can remove this file.",
+      "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 should be `github.com/mholt/caddy/onevent/hook` instead of `github.com/mholt/caddy/onevent/hook`.",
+      "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 if this is the right thing to do, but I think we should be able to stop the pinger at any time, even if we're in the middle of a ping.",
+      "reference": "Shouldn't this be `i * Interval`? You're multiplying nanos with a Duration, I'm not sure what's the result :)"
+    },
+    {
+      "prediction": "I think this should be MEMPROT_READWRITE",
+      "reference": "I guess it's not no-access to match Windows where guard pages are readable."
+    },
+    {
+      "prediction": "Style/RedundantSelf: Redundant self detected.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br",
+      "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 `namedtuple` instead of `collections.namedtuple`",
+      "reference": "Add trailing comma"
+    },
+    {
+      "prediction": "I think we should keep SIGKILL here. It's the only signal that can't be caught or ignored, so it's the only one that can't be handled gracefully.",
+      "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 `Setting.THEME_OLED_DARK` instead of `Setting.THEME_OLED_DARK`.",
+      "reference": "As it is a mobile only theme, please make sure the option appears only on mobile"
+    },
+    {
+      "prediction": "I think it would be better to use `split(\", \")` instead of `split(\",\")`.",
+      "reference": "Would like a docstring and type annotations here"
+    },
+    {
+      "prediction": "Style/HashSyntax: Use the new Ruby 1.9 hash syntax.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one",
+      "reference": "This is an API change. Why was it made?"
+    },
+    {
+      "prediction": "Is this required?",
+      "reference": "Please move using inside namespace."
+    },
+    {
+      "prediction": "Why do we need to pass `poSEnabled` to `Suggest`?",
+      "reference": "shall we create some enum flags for setasmain shouldprocess and pos?"
+    },
+    {
+      "prediction": "I don't think we need to check the registry here. We can just check the service start mode.",
+      "reference": "Should this `PUPPET_AGENT_STARTUP_MODE` have a corresponding yardoc change?"
+    },
+    {
+      "prediction": "I think we can remove this now.",
+      "reference": "Why not simply remove the `pytest` import now that it's not needed anymore? :wink:"
+    },
+    {
+      "prediction": "This file is empty. Should it be deleted?",
+      "reference": "Yeah, not filled in yet."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "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": "Why do we need to reload cache here?",
+      "reference": "should we also do `reloadCache` when database is null?"
+    },
+    {
+      "prediction": "I think this should be a `std::unique_ptr<ROMol>`.",
+      "reference": "Why not just return QueryMol directly? Why require the copy?"
+    },
+    {
+      "prediction": "I don't think we need this function. We can just use `os.path.join(os.path.dirname(__file__), '../../', fname)` directly.",
+      "reference": "Could you add a docstring?"
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "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": "Please remove this empty line.",
+      "reference": "@jlerbsc 's fix in #2918 in action - thanks! :smiling_face_with_three_hearts:"
+    },
+    {
+      "prediction": "I think this should be `if zero_count and not count: raise TypeError(\"Zero_count Argument cannot exist without 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.Issuer == issuer` instead of `a.Name == issuer`",
+      "reference": "change the name of the function to match its functionality? account IsIssuing?"
+    },
+    {
+      "prediction": "I don't think we need to support `by` as a `str` or `Series` here. I think we should only support `by` as a `list` or `tuple` of `str` or `Series`.",
+      "reference": "for later: add docstring (can just copy paste Pandas')"
+    },
+    {
+      "prediction": "nit: I think we can remove the `archerMocks` alias and just use `mocks` instead.",
+      "reference": "nit: @sonofachamp pointed out to me that the idiomatic way is \"archermocks\" (lowercase for package names)"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "Ugh I'm stupid, can you change this to `yarpc.InternalErrorf`?"
+    },
+    {
+      "prediction": "I think this should be `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 `else if (queryOptions.perDocumentLimit != null && queryOptions.limit != null) { throw new Error('Can not use `limit` and `perDocumentLimit` at the same time. Model `' + mod.model.modelName + '`.' ); } else if (queryOptions.perDocumentLimit != null) { queryOptions.limit = queryOptions.perDocumentLimit; delete queryOptions.perDocumentLimit; } else if (queryOptions.limit != null) {`",
+      "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": "I'm not sure why we need to pass `proxyPort` to `PingProvider`. It seems to me that `proxyPort` is always the same as `consumerPort`, so we can just pass `consumerPort` to `PingProvider` and remove `proxyPort` from `PingProvider` and `NATProviderPinger`.",
+      "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": "I don't think we need a test for this class.",
+      "reference": "Probably best not to have an empty test...."
+    },
+    {
+      "prediction": "I'm not sure if we want to use the word \"Professional\" here. I think it might be better to just say \"Anthony Lee\".",
+      "reference": "`image_tag` in the `testimonials` files?"
+    },
+    {
+      "prediction": "I think you can use `JSONUtils.toJSON(flow)` instead of `JSONUtils.toJSON(flow.toObject())`",
+      "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 think this should be `this.siteSettings.favicon_url`",
+      "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 think this should be `catch (std::exception &e) { std::cerr << e.what() << std::endl; result = 1; }`",
+      "reference": "Why swallow the exception here rather than propagate it?"
+    },
+    {
+      "prediction": "nit: `column_names` -> `columns`",
+      "reference": "How about `column_index_names`? `column_names` sounds ambiguous."
+    },
+    {
+      "prediction": "I think this should be `write.metadata.metrics.column.` instead of `write.metadata.metrics.column.`",
+      "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 it's better to use `np.arange(1, 6)` instead of `[1, 2, 3, 4, 5]`.",
+      "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 import `string_types` here.",
+      "reference": "we should sort the headers like in spark: public packages, then pyspark, then internal"
+    },
+    {
+      "prediction": "I don't think this is correct. If `t.options.clientTLSConfig` is nil, then `t.options.clientTLS` should be false.",
+      "reference": "looks like we should drop this var from the transport options struct"
+    },
+    {
+      "prediction": "I'm not sure why this test was removed. It seems like it should still be valid.",
+      "reference": "Ah, yeah, this is the test for the validation that I mentioned in a remark above about relaxing the validation."
+    },
+    {
+      "prediction": "`inside_flags = valid_flags & (abs(flat_anchors[:, 0]) < img_h + allowed_border) & (abs(flat_anchors[:, 1]) < img_w + allowed_border) & (abs(flat_anchors[:, 2]) < img_h + allowed_border) & (abs(flat_anchors[:, 3]) < img_w + allowed_border)`",
+      "reference": "For tensors, it is better to illustrate the shape."
+    },
+    {
+      "prediction": "I think this should be `request.url().match( /google-site-kit\\/v1\\/data/ )`",
+      "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 think we can remove the `#if !NETCOREAPP2_1` and `#else` and just have the `Assert.Equal(\"Microsoft.AspNetCore.Hosting.HttpRequestIn\", activity.OperationName);`",
+      "reference": "nit: probably swap the if condition and `if/else` and check `NETCOREAPP2_1` which looks more natural."
+    },
+    {
+      "prediction": "Shouldn't this be `defer a.mu.RUnlock()`?",
+      "reference": "for the tag, should it be `requestor_rtt` since the other is `responder_rtt`?"
+    },
+    {
+      "prediction": "I think this should be `MutateUint8` instead of `MutateByte`.",
+      "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": "nit: I think we can remove the `files` prefix here.",
+      "reference": "nit: there should be a newline between `types` and `go-ipfs-files`."
+    },
+    {
+      "prediction": "I don't think we need to export this.",
+      "reference": "would be nice to have this as a parameter (Options), re: technical debt discussions"
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "complete ignorance here  but is it ok to import this from `com.sun`? can it break under non-oracle JREs?"
+    },
+    {
+      "prediction": "I think this should be `if span.IsRecordingEvents() && !span.IsSampled() {`",
+      "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 should return the error here and not just log it.",
+      "reference": "just calling UpdateSnapshotInfo() should be fine"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should keep the default rate limit, and only disable it in debug mode. What do you think?",
+      "reference": "Was `DefaultLogMessageRateLimit` not configurable anywhere?"
+    },
+    {
+      "prediction": "Shouldn't this be `docker.ErrInactivityTimeout`?",
+      "reference": "Please use the `assert` instead. Same below."
+    },
+    {
+      "prediction": "typo: \"suche\" should be \"such as\"",
+      "reference": "Owww, I think I found a typo. `suche`"
+    },
+    {
+      "prediction": "I'm not sure why these assertions were here, but they don't seem to make sense to me. The heaps can be empty if there are no phantom nodes in the graph.",
+      "reference": "This can happen because we could get `PhantomNodes` that are not admissible as source/target right?"
+    },
+    {
+      "prediction": "I think this should be `!structsEqual(&clientSessionState{}, &tls.ClientSessionState{})`",
+      "reference": "Should we add a note here to file a bug if this ever happens?"
+    },
+    {
+      "prediction": "I think this should be `errors.Errorf(\"benchmarks quantity overflows int64: %s\", num.String())`",
+      "reference": "overflow is still possible, isn't it?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [115/100]",
+      "reference": "This seems more like a personal preference, so maybe this should be in your `~/.rspec`?"
+    },
+    {
+      "prediction": "I don't think we should do this. We should keep `PcapTimeoutElapsed` in `scapy/arch/pcapdnet.py` and import it from there.",
+      "reference": "To avoid a circular import between pcapdnet and automaton"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Please remove this blank line."
+    },
+    {
+      "prediction": "nit: I think this should be `s.Status.WorkloadIdentityStatus.InitWorkloadIdentityStatus(s.Spec.GCPServiceAccount)`",
+      "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": "Shouldn't this be `provokingVertexMode == VK_PROVOKING_VERTEX_MODE_FIRST_VERTEX_EXT`?",
+      "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": "nit: please remove this empty 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/plugins/modules:/usr/share/ansible/plugins/modules:/etc/ansible/plugins/modules`",
+      "reference": "The library -> modules and filters -> filter path changes are breaking changes or?"
+    },
+    {
+      "prediction": "I think we can remove these unused imports.",
+      "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": "I'm not sure if this is the best way to reverse an array, but it's the only way I could get it to work. `child.reverse().forEach((item) => { stack.push(item); });` doesn't work because `reverse()` mutates the original array.",
+      "reference": "An `if` statement would be better here. Functions, concat and reverse are all very expensive."
+    },
+    {
+      "prediction": "I think this should be `0x334394E8c891E77b1449084aaDD659920BB25247` instead of `0x334394e8c891e77b1449084aaDD659920BB25247`",
+      "reference": "`Log here: connect to default factory for goerli network <address>`"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it. I think we should have a `switch` statement here, but I'm not sure how to get the `Event` value from the `model.NotificationEventType` enum.",
+      "reference": "nits, should add `break` to avoid redundant iterations."
+    },
+    {
+      "prediction": "I think this test should be moved to `StateProviderTests`",
+      "reference": "I think we want an exception here, there should never be a transfer to a nonexisting account."
+    },
+    {
+      "prediction": "I don't think we need to log the creation of the container.",
+      "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": "I don't think this is a good idea to use `max(0.01, 0.1 - 0.01 * x)` instead of `lambda x: 0.95 ** x * 0.1`",
+      "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 `interpreters[Pathname(executable).extname] if interpreters && interpreters[Pathname(executable).extname]`",
+      "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": "I don't think we need to assert that the sysroot is loaded here. I think we can just return NULL if it's not loaded.",
+      "reference": "In this and the one below, should we assert `self != NULL` first?"
+    },
+    {
+      "prediction": "Same here, `getStatusCode` is a method of the `Response` object, not the `View` object.",
+      "reference": "Why this is needed?"
+    },
+    {
+      "prediction": "I think this should be `return self._meta.get('user_meta', None)`",
+      "reference": "we already have .meta() so I don't think we want get_meta() for PackageEntry"
+    },
+    {
+      "prediction": "Why do we need to pass `startTs` to `CoprocessorIterator`?",
+      "reference": "question: can we always get the startts from dagrequest?"
+    },
+    {
+      "prediction": "I think we can remove these variables now that we're not using them.",
+      "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 `@unresolved_targets.delete(target.name)`",
+      "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'm not sure if this is the right thing to do. I think we should throw an exception if the contract doesn't exist.",
+      "reference": "So what happens if I try to get code on a EOA?"
+    },
+    {
+      "prediction": "I don't think we need to check for `isinstance(inferred, (nodes.Tuple, nodes.List))` here, because `self._get_values_to_unpack` will return `None` if `inferred` is not a tuple or list.",
+      "reference": "Can't we return an empty list and then just compare the lengths? Thereby removing this `if`."
+    },
+    {
+      "prediction": "I think this should be called `passesAllRestrictions` or something like that. `passesRestrictions` sounds like it only checks a single restriction.",
+      "reference": "Mb change it to 'doesPassRestrictions()'? Doesn't sound enough like a boolean-returning method to me."
+    },
+    {
+      "prediction": "`currentIndex` and `nextIndex` don't seem to be used.",
+      "reference": "TODO: in fact At when used with noncurrent time should not return a next index. only latest should.."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to fix the issue for me.",
+      "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 don't think this is correct. `List.Nil` is a subtype of `List<T>`, so it should be matched before `List<T>`.",
+      "reference": "@danieldietrich, this wasn't tested so I presumed it wasn't working before either :p"
+    },
+    {
+      "prediction": "This is not a typo, it's the correct class name.",
+      "reference": "Please leave two empty lines after the `use` statement"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Please move this import in the 3rd group which is for external dependencies"
+    },
+    {
+      "prediction": "I think this should be `isinstance(node.get(RESERVED['file']), str) or node.get(RESERVED['file']) is None`",
+      "reference": "So what exactly is this fixing? The unit test is passing with or without this change."
+    },
+    {
+      "prediction": "I don't think we need to make the parent dir here. The parent dir should have been created by the caller of this function.",
+      "reference": "auto targetFile = fs::FileUtils::joinPath(newPath, folly::stringPrintf(\"%019ld.wal\", it->first));"
+    },
+    {
+      "prediction": "I think this method should be called `snapshotIdsBetweenIncludingFrom` or `snapshotIdsBetweenIncludingFromAndTo` to make it clear that the from snapshot is included in the result.",
+      "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": "`?? ''` is the same as `?: ''`",
+      "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'm not sure this is the right thing to do. I think it should be `@configuration.output_stream ||= out`",
+      "reference": "Why the change? If `output_stream` is set to something non-nil, it seems odd (and potentially wrong) to overwrite it...."
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.",
+      "reference": "Thinking this should have an underscore."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we should be testing the private method `_get_message_state_scope` in this way. I think we should be testing the public method `get_message_state_scope` instead.",
+      "reference": "To accommodate the change to the method being private."
+    },
+    {
+      "prediction": "I think this should be `[CollectionRule(typeof(PropertiesNamesCamelCase))]` instead of `[CollectionRule(typeof(PropertiesNamesCamelCase))]`",
+      "reference": "Shouldn't we traverse all definitions and apply this rule over all model definitions in the doc"
+    },
+    {
+      "prediction": "I think we should add a unit test for this rule.",
+      "reference": "Please rename this class that it ends with the suffix \"Test\" - otherwise the tests won't be executed."
+    },
+    {
+      "prediction": "I don't think we need this file at all.",
+      "reference": "As a JS file, this should also receive our standard file header."
+    },
+    {
+      "prediction": "I think this should be `double bmoMemoryUsagePerNode = getEstimatedRunTimeMemoryUsage(TRUE, &numStreams).value();` instead of `double bmoMemoryUsagePerNode = getEstimatedRunTimeMemoryUsage(TRUE, &numStreams).value() / numBMOsInFrag;`",
+      "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": "Why is this no longer needed?",
+      "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 don't think we need to store this in the database. We can just calculate it from the number of digits in the total.",
+      "reference": "not dramatic, but why 32bits if we allow a range that is smaller than 8?"
+    },
+    {
+      "prediction": "I think this should be `BeneficiaryId` instead of `BenefiterId`.",
+      "reference": "- Use type from `service_discovery/dto/price.go` - Rename `Price` -> `Money`"
+    },
+    {
+      "prediction": "I think it would be better to use `AtomicInteger::get` instead of `Function1.identity()`.",
+      "reference": "one space missing at the beginning"
+    },
+    {
+      "prediction": "I think this should be `PlatformVersion` instead of `Platform`",
+      "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 `http://localhost:<%= connect.test.options.port %>/test/integration/api/external/`",
+      "reference": "I like the new directory `/api`, what is the thinking behind `/external`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do. We don't care about the order of the signing keys, we just want to know if they changed or not.",
+      "reference": "staticcheck flags this: replace with `for k := ..`"
+    },
+    {
+      "prediction": "I don't think we need to wait for a response here.",
+      "reference": "There's already a sleep right above"
+    },
+    {
+      "prediction": "I don't think this is correct. `$var_id` is the name of the variable, not the name of the method.",
+      "reference": "This feels hacky, is there a better way?"
+    },
+    {
+      "prediction": "I think we can remove this try/except block and just use `from ordereddict import OrderedDict`",
+      "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 `Gets or sets` instead of just `Gets`",
+      "reference": "\"Gets the JSON serialization settings.\" would be better?"
+    },
+    {
+      "prediction": "I think this should be `/file/{addr}/{path}` to be consistent with the rest of the API.",
+      "reference": "The endpoint should be in plural `/files`."
+    },
+    {
+      "prediction": "I don't think these are needed.",
+      "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": "This should be `GetContext() context.Context`",
+      "reference": "I think this line is gofmt issue? @daixiang0"
+    },
+    {
+      "prediction": "I think we should also test the case where `targetNewSyntax` doesn't start with `@`.",
+      "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": "Why is this changed?",
+      "reference": "line is 161 characters (from `lll`)"
+    },
+    {
+      "prediction": "I don't think we need to make this serializable.",
+      "reference": "Are these changes needed?"
+    },
+    {
+      "prediction": "I think this should be `SensitivityWatchEventModifier.HIGH`",
+      "reference": "It seems like reducing the sensitivity would make the test failure less likely to occur, but not fix the issue?"
+    },
+    {
+      "prediction": "I don't understand why you need to use this trait in this file.",
+      "reference": "missing use statement for importing the full namespace"
+    },
+    {
+      "prediction": "I think this should be `ErrInvalidAddressFormat` or something like that.",
+      "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 think this test should be updated to reflect the new behavior.",
+      "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 think it would be better to have `ZeroGasPrice` and `CommitAndRestoreWithZeroGasPrice` as separate flags.",
+      "reference": "I would prefer flag SkipGasPricingValidation"
+    },
+    {
+      "prediction": "I think it would be better to use `std::unordered_map::count` instead of `std::unordered_map::find` here. `count` will return `1` if the element is found and `0` otherwise.",
+      "reference": "could be `const &` ?"
+    },
+    {
+      "prediction": "I think this should be `hideProgressOnFinish` or `hideProgressAfterSuccess` or something like that. `hideProgressAfterFinish` is a bit confusing.",
+      "reference": "The docs should also be updated with the new option."
+    },
+    {
+      "prediction": "I think it would be better to use the `boto3` library instead of `boto`. `boto3` is the AWS SDK for Python (Boto) 2.0. `boto` is the AWS SDK for Python (Boto) 1.0.",
+      "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 think this should be `&cfg.Logging.Level` instead of `*cfg.Logging.Level`",
+      "reference": "Out of the scope."
+    },
+    {
+      "prediction": "This should be `thisArg = thisArg || null;`",
+      "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 think this should be `if (name && typeof name !== 'string') { editorClass = name; name = editorClass.EDITOR_TYPE; } else if (typeof name !== 'string') { editorClass = name; name = editorClass.EDITOR_TYPE; }`",
+      "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": "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 `Tags []string` so that we can keep track of all the tags that have been applied to the block.",
+      "reference": "Wouldn't it be more space efficient to make this a `byte` enum or bitfield?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `CREATE TABLE IF NOT EXISTS` instead of `CREATE TABLE IF NOT EXISTS IF NOT EXISTS`.",
+      "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 don't think we need this interface. We can just use `ASTAnyTypeBodyDeclaration.class.isAssignableFrom(node)`.",
+      "reference": "Maybe add short example list: .... type body declarations, such as AnnotationMembers, Methods, Fields"
+    },
+    {
+      "prediction": "I don't think we need to add these here. They are already defined in `org.openqa.selenium.remote.CapabilityType` and `org.openqa.selenium.remote.Timeouts`.",
+      "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 called `qualifiedName` to be consistent with the rest of the API.",
+      "reference": "It is debatable whether we have to default this. I did this to avoid breaking custom implementations."
+    },
+    {
+      "prediction": "I don't think this is correct. If `attr_value['string']` is `None` then `attr_value['string'] == value` will be `False` even if `value` is `None`.",
+      "reference": "# `attr_value['string'] is not None` probably not required as `attr_value['string'] == value` check is already being done"
+    },
+    {
+      "prediction": "nit: I think we can remove the `addon` prefix here.",
+      "reference": "we shouldn't need the \"addon\" rename"
+    },
+    {
+      "prediction": "Use 2 (not 1) spaces for indenting an expression spanning multiple lines.",
+      "reference": "These are covered by routing constraints."
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "Could we use just a normal string here?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to check for Symbol support, but it seems to work in all the browsers I've tested so far.",
+      "reference": "`isFunction(Symbol)` triggers a `ReferenceError` if Symbol is undefined. Using `typeof Symbol === 'function'` is OK however."
+    },
+    {
+      "prediction": "I don't think you need to unlock here.",
+      "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 if this is the right thing to do, but I think we need to pass a zero value for the nonce.",
+      "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": "Please remove this empty line.",
+      "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 works for now.",
+      "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": "`nodeIsHeader` should be `node.tagName.toLowerCase() === 'header' && node.getAttribute('role') !== 'banner'`",
+      "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": "Why is this check necessary? If `core` is `null` then `MDCLoggingContext.setCore(core)` is a no-op.",
+      "reference": "Why do we need this check? setCore already handles nulls."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. `__dir__` is supposed to return a list of strings, but `__dir__` in `inject_type` returns a list of objects. I think it would be better to define `__dir__` in `inject_type` to return a list of strings.",
+      "reference": "good catch, thanks. This is going to make the user experience much better."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "TODO: remove this unused import"
+    },
+    {
+      "prediction": "I think we should keep the original error message here.",
+      "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 think this should be `self.elif '-f build.gradle', 'gradle check'`",
+      "reference": "I think `-f gradlew` is better for code unity"
+    },
+    {
+      "prediction": "I think this should be `source.iceberg.reader.fetch-record-batch-size`",
+      "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 is not changed in this PR. Can you please revert it?",
+      "reference": "Can we avoid no-change changes being committed - they make it harder to distinguish actual changes from code editor changes"
+    },
+    {
+      "prediction": "`AWS IAM` -> `AWS Identity and Access Management`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I don't think we need to pass `di.SignerFactory` to `identity_selector.NewHandler` anymore.",
+      "reference": "You dont use this dependency anymore"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to check for these flags. I think it would be better to check for them in the `SmartStoreTest` class.",
+      "reference": "Here is the test that checks that the sqlcipher in use was compiled with the right flags"
+    },
+    {
+      "prediction": "I think you can remove the `}` here.",
+      "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": "RestClientException is not used in this file.",
+      "reference": "this test only covers the happy path. what about no acceptable rest template found?"
+    },
+    {
+      "prediction": "I don't think we should use `finer` here. `finest` would be more appropriate.",
+      "reference": "Since this is in a test, I imagine that the choice of `info` level was deliberate."
+    },
+    {
+      "prediction": "nit: I think we should keep the imports in alphabetical order.",
+      "reference": "this should probably be named somewhere if you want to reuse it."
+    },
+    {
+      "prediction": "I don't think we need this constant. We can just use `BoundServiceAccountSigningKeyFile` directly.",
+      "reference": "needs a rename to not include AWS"
+    },
+    {
+      "prediction": "I think it's better to use `oneapi::dal::backend::math::pow` instead of `oneapi::dal::backend::math::exp` and `oneapi::dal::backend::math::sqrt`.",
+      "reference": "Isn't sigma have `double` type?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "May be better use nano::keypair that generates ramdom value?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should be able to rely on the `checkKeys: false` option to do the right thing here. If the user has `ignoreUndefined` set to `true` in their BSON options, then we should be able to rely on that. If they have it set to `false` in their BSON options, then we should be able to rely on that. If they don't have it set at all in their BSON options, then we should be able to rely on that.",
+      "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 if this is the right thing to do. I think we should at least show a message to the user that they need to log in again.",
+      "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 we can remove the `AnyTimes()` here.",
+      "reference": "huh does this work without a `Return`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should keep the old behavior of not setting the exception if the example is already pending.",
+      "reference": "Don't think we want this line, right? (Plus \"failing\" is spelled wrong)."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in `FeedItemMenuHandler#onPrepareOptionsMenu` where you already check if the feed is a local feed.",
+      "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": "I'm not sure if this should be removed or not. I don't think it should be removed, but I'm not sure if it should be moved to a different place.",
+      "reference": "Why remove this one? It doesn't work yet, but we'll add it sooner or later."
+    },
+    {
+      "prediction": "I think we should keep the `docstore:\"...\"` tags.",
+      "reference": "Why did you remove the struct tags?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the engine.",
+      "reference": "Let's be neutral with tag names. Let's just have \"ReportURL\""
+    },
+    {
+      "prediction": "I think this should be `time.Date(2020, 3, 31, 5, 1, 26, 1234, time.UTC)`",
+      "reference": "Use an explicit time to ensure conversion is not copy-paste and wrong."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what else to do either.",
+      "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": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Use nested module/class definitions instead of compact style."
+    },
+    {
+      "prediction": "I think this should be `t.Logf(\"Init KBFSOps\")` so that it doesn't show up in the test summary.",
+      "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 `&` instead of `|`",
+      "reference": "showNPCs is not needed"
+    },
+    {
+      "prediction": "This should be `parameter %s already exists in region %s`",
+      "reference": "Docstring for public function pls"
+    },
+    {
+      "prediction": "I think this should be `expectedFilteredErr2 := \"unable to parse \\\"[ { \\\\\\\"op\\\\\\\": \\\\\\\"replace\\\\\\\", \\\\\\\"path\\\\\\\": \\\\\\\"/data/foo\\\\\\\", \\\\\\\"value\\\\\\\": \\\\\\\"baz-json\\\\\\\" ]\\\": yaml: did not find expected ',' or '}'\"`",
+      "reference": "This should use `t.Errorf` instead. You want the second part of the test to run even when the first part fails."
+    },
+    {
+      "prediction": "nit: `ctx.protocol_selector.ratio.http3 > 0` can be simplified to `ctx.protocol_selector.ratio.http3`",
+      "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'm not sure if this is the right thing to do, but I think it's the right thing to do. If the reducer returns the same value as the current state, we don't need to update the state.",
+      "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 don't think this is correct. If `options.roles` is a string, then `roles` will be an array with a single string in it. If `options.roles` is not a string, then `roles` will be an empty array. I think you want `roles = typeof options.roles === 'string' ? [options.roles] : options.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_COLOCATED_WITH` or `COLLECTION_WITH_COLLECTION` or something like that.",
+      "reference": "It would be helpful to explicit here what this really means and assumes."
+    },
+    {
+      "prediction": "`ExpectedBalances` should be `expectedBalances`",
+      "reference": "`ExpectedBalances` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think this should be `Extracts code snippets by looking at the backtrace of the passed error and applies syntax highlighting and line numbers using HTML.`",
+      "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 don't think we should be doing this here. I think we should be doing this in the tests that need it.",
+      "reference": "Feels like this should be left as is and add a new createClientWithIssuers or something like that. Avoid all the \"\""
+    },
+    {
+      "prediction": "Shouldn't we check if the callback is not null before calling it?",
+      "reference": "Minor unrelated bug in the IDP flow where the error wasn't getting displayed as a `Toast`."
+    },
+    {
+      "prediction": "This should be `require('../../lib/utils').mongoErrorContextSymbol`",
+      "reference": "I thought we got rid of this thing?"
+    },
+    {
+      "prediction": "I don't think we need to remove these fields. They are used in `TaintNodeData::serialize()` and `TaintNodeData::deserialize()`.",
+      "reference": "Please convert int params as well."
+    },
+    {
+      "prediction": "I'm not sure why this change was made, but I think it should be reverted.",
+      "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 think this should be `DD_TRACE_HTTP_SERVER_ERROR_STATUSES` and `DD_TRACE_HTTP_CLIENT_ERROR_STATUSES`",
+      "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 `strcasecmp(tmp, \"false\") == 0`",
+      "reference": "would you please use: flb_utils_bool(...) here ?, that function wraps the on/off/true/false stuff."
+    },
+    {
+      "prediction": "nit: I think we can remove the `null` here.",
+      "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 don't think we need to add `async` here.",
+      "reference": "maybe renamed to `async_simple_test` if we consider supporting aug test later on?"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "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 think we should use `DateTimeOffset` instead of `DateTime` here.",
+      "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 best way to test this, but it seems to be the only way I could get it to work.",
+      "reference": "While we're doing some translation work, could you extract this button text and the other static text from the view?"
+    },
+    {
+      "prediction": "I don't think this is the right place to replace the tokens. The tokens should be replaced when the headers are created, not when the headers are added to the message.",
+      "reference": "Nit: Please fix indentation (note: Azkaban uses 2 spaces and not tabs)."
+    },
+    {
+      "prediction": "I think this should be `new Realm({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": "Shouldn't this be `WINDOWS_VERSION_10_1809`?",
+      "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 think this should be `if (node && node.hasAttribute(aria) && node.getAttribute(aria) === 'false')`",
+      "reference": "Just do: `return node.hasAttribute('aria-hidden')`"
+    },
+    {
+      "prediction": "Extra empty line detected at block body end.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "nit: I think we can remove the `private static bool _processDataPoisoned;` and `private static bool _domainDataPoisoned;` as they are not used anymore.",
+      "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 if this is the right place to set the NOTIFY_SOCKET env var, but it seems to be the only place where it is unset.",
+      "reference": "Did it not work out to wait until after containerd and kubelet are started?"
+    },
+    {
+      "prediction": "I don't think these should be added to this list.",
+      "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 think this should be a `tinyint(1)` instead of a `bool`.",
+      "reference": "Why can't this be determined by 'PARTIAL_SUCCESS' in the `status` field?"
+    },
+    {
+      "prediction": "`$media->getContext() ?? ''` should be `$media->getContext() ?: ''`",
+      "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": "This is not thread-safe.",
+      "reference": "This should be made private."
+    },
+    {
+      "prediction": "I think this should be `a.baseSVIDPath` instead of `a.getBaseSVIDPath()`. `a.baseSVIDPath` is the path to the base SVID file, while `a.getBaseSVIDPath()` is the path to the directory containing the base SVID file.",
+      "reference": "perhaps this is better modeled as a pkg-level var?"
+    },
+    {
+      "prediction": "I think this should be MAX_NUM_LEN_ALLOWED or something like that. MAX_NUM_LEN is a bit confusing.",
+      "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 think these should return `status.Error(codes.NotFound, \"\")` instead of `status.Error(codes.Unimplemented, \"\")`",
+      "reference": "`ctx` is unused in UpdateProjectStaticUser"
+    },
+    {
+      "prediction": "I think this should be `https://www.googleapis.com/discovery/v1/apis/{apiName}/{apiVersion}/rest`",
+      "reference": "Why is this not the default, and why only for JS?"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.",
+      "reference": "A repository has many badges so we should also add type column in scope."
+    },
+    {
+      "prediction": "This should be `result.CodeFlows[0].Locations.ToArray().Should().Equal(new[]`",
+      "reference": "Now a hash set, so can't index into it."
+    },
+    {
+      "prediction": "I think this should be `!!ENV['PSModulePath']` instead of `!!ENV['PSModulePath'].nil?`",
+      "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 a good idea. If the content type is `application/json; charset=UTF-16` it will fail.",
+      "reference": "1.toLowerCase(Locale.US)? 2.where is accept bug fix?"
+    },
+    {
+      "prediction": "I think we should keep the `SystemExit` check here.",
+      "reference": "We should still be providing something to the user in the case of a `KeyError` instead of a stack trace!?"
+    },
+    {
+      "prediction": "Why do we need to get the nonce here?",
+      "reference": "check Lukasz's fix with NonceReserving - is that not better?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to set all 3 colours to the same value.",
+      "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 don't think we need to lock here, as we are only reading from the map.",
+      "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 don't think you need to override this method. The default implementation should work for you.",
+      "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 this was removed, but I think it should be added back.",
+      "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 `NSCD_SOCKET_VOLUME_NAME` and `NSCD_SOCKET_VOLUME_MOUNT_PATH`",
+      "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 don't think we need to do this. We should be able to create a stream with the same name as the MQTT stream.",
+      "reference": "Could be a non-public field. `allowNoSubject`"
+    },
+    {
+      "prediction": "nit: I think we can remove the `skipCacheLookup` parameter and just pass `c.state.ForceRerun && isTest || (!isTest && c.state.ForceRebuild)` as the `skipCacheLookup` argument to `reallyExecute`.",
+      "reference": "Don't think this is quite right - think the build one also needs an `IsOriginalTarget` (c.f. code in `src/build`)"
+    },
+    {
+      "prediction": "`$entities` and `$targetEntityConfig` should be typed as `array`",
+      "reference": "what about adding typehints instead? I don't think we need to add docblocks for every private methods."
+    },
+    {
+      "prediction": "Is this the default block size for cstor volumes? If not, can we make it configurable?",
+      "reference": "Is it fine to pass the \"-o compression=on\" in the middle instead of at the end ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Probably make them all keyword arguments."
+    },
+    {
+      "prediction": "Why is this being 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 an error, not a warning.",
+      "reference": "why are we turning this one into a Warning?"
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "reference": "Wouldn't it be better to use double quotes for the opening parenthesis ?"
+    },
+    {
+      "prediction": "This should be `static_cast<uint64_t>(player->blessings.to_ulong())`",
+      "reference": "Remove the cast"
+    },
+    {
+      "prediction": "I think this should be 2017-2018.",
+      "reference": "What criteria do we use to bump the copyright years? An update to the module?"
+    },
+    {
+      "prediction": "Why do we need this line?",
+      "reference": "should we rm line 269?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `__init__` method of the `NVDASpyLib` class. That way, you don't have to worry about whether or not NvDA's core loop has started running, and it has processed initial focus. You can just set the `_isNvdaStartupComplete` variable to `True` in the `__init__` method, and you can be sure that NvDA's core loop has started running, and it has processed initial focus.",
+      "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": "Use the new Ruby 1.9 hash syntax.",
+      "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 think this should be `AcceptResponseError` instead of `AcceptResponseErrorHeader`",
+      "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": "nit: I think it would be better to pass the library file as a parameter to the `readModel` method, instead of hardcoding it here.",
+      "reference": "Adding in the service yaml `library.yaml` here."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I don't have a better idea.",
+      "reference": "Why is this dereference needed?"
+    },
+    {
+      "prediction": "I don't think you need to say \"allow repo directories to be removed\" here.",
+      "reference": "Might want to add that this only happens if you have write permissions to the folder."
+    },
+    {
+      "prediction": "I think this should be `EnforceShouldRetryCheck *bool` instead of `EnforceShouldRetryCheck bool`.",
+      "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": "Why is this empty string?",
+      "reference": "assume the user will be able to register for a custom service name in future?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "The extra two empty lines seem unnecessary. Please review your diffs more carefully."
+    },
+    {
+      "prediction": "I think this should be `OracleAPI` instead of `OracleUSD`",
+      "reference": "What the reason to embed the Oracle?"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "Shouldn't this happen automatically with High Voltage?"
+    },
+    {
+      "prediction": "I don't think this is correct. `initValueConfig.getResourceNameBindingValues().get(entityName)` returns a `InitValue` object, so `initValue.getValue()` returns a `String`. The `varReference` method expects a `Name` object, not a `String`.",
+      "reference": "This is changed because we need `projectId` to be printed as `$projectId` in PHP."
+    },
+    {
+      "prediction": "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.<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.<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.<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.<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",
+      "reference": "This title change seems more accurate to me however not sure if this will confuse users."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Here (and on line 158 as well), would it be more efficient to use the String end_with() method? Not tested"
+    },
+    {
+      "prediction": "I don't think you need to create a new instance of `RAMHandler` here. You can just set `self._data` to a new `collections.deque` with the new capacity.",
+      "reference": "Why are you creating a new (second) `RAMHandler` here? Is this just an oversight from deleting the previous code?"
+    },
+    {
+      "prediction": "`resp_.set_code(cpp2::ErrorCode::E_EXISTED);` -> `resp_.set_code(cpp2::ErrorCode::SUCCEEDED);`",
+      "reference": "Move the line 27 to line 31 is better?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like the right place to me.",
+      "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'm not sure why this was removed, but I think it's a good idea to leave it in.",
+      "reference": "I would instead explain that random is the default, but you can switch back to defined, or another ordering scheme."
+    },
+    {
+      "prediction": "`elHeight <= 1` is not necessary. `elHeight === 0` is sufficient.",
+      "reference": "Is there a reason you left this at 0?"
+    },
+    {
+      "prediction": "I don't think we need to do this. RememberMeServices is a public API, so if someone wants to use it, they should be able to set the key.",
+      "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 this is the right place to do this. I think it should be done in the `GetDbType` method.",
+      "reference": "NIT: You should first check if AdoNetExcludedTypes.Count is greater than 0 before calling Contains (fast-path)"
+    },
+    {
+      "prediction": "Please remove the 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": "I don't think we need a test for this class.",
+      "reference": "Probably best not to have an empty test...."
+    },
+    {
+      "prediction": "I think this should be `#include <Kokkos_Tools.hpp>` instead of `#include <impl/Kokkos_Tools.hpp>`.",
+      "reference": "Missing the license"
+    },
+    {
+      "prediction": "I think we should keep the `if p.config.EnableCleanupReplicationTask()` check here.",
+      "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": "Why is this removed?",
+      "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 we need this function anymore. It's only used in `plot_topo` and `plot_evoked_topo` which can be replaced by `plot_topomap` and `plot_evoked_topomap` respectively.",
+      "reference": "I would just mention which tolerance - namely as reported by ``sys.float_info``"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [127/120]",
+      "reference": ":+1: Ideally this'll become something you could pass in to the config"
+    },
+    {
+      "prediction": "I don't think we need to use `via` here. `executor_` is a `std::shared_ptr<folly::CPUThreadPoolExecutor>`, which is a `std::shared_ptr<folly::ThreadPoolExecutor>`, which is a `std::shared_ptr<folly::Executor>`, which is a `std::shared_ptr<folly::AsyncExecutor>`, which is a `std::shared_ptr<folly::DrivableExecutor>`, which is a `std::shared_ptr<folly::ExecutorBase>`, which is a `std::shared_ptr<folly::Executor>`, which is a `std::shared_ptr<folly::AsyncExecutor>`, which is a `std::shared_ptr<folly::DrivableExecutor>`, which is a `std::shared_ptr<folly::ExecutorBase>`, which is a `std::shared_ptr<folly::Executor>`, which is a `std::shared_ptr<folly::AsyncExecutor>`, which is a `std::shared_ptr<folly::DrivableExecutor>`, which is a `std::shared_ptr<folly::ExecutorBase>`, which is a `std::shared_ptr<folly::Executor>`,",
+      "reference": "Should using `thenValue` to replace `then` ?"
+    },
+    {
+      "prediction": "I think this should be `if (Tracer.Instance.Settings.IsIntegrationEnabled(IntegrationName)) { return; }`",
+      "reference": "This follows the convention of each callback starting with the `IsIntegrationEnabled` check."
+    },
+    {
+      "prediction": "I think it would be better to have `isDeltaValue` and `isMonotonicValue` as parameters of `SumMetricAggregatorLong` constructor.",
+      "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": "Should this be `IConnectionHandler<TContext>` where `TContext` is `IConnectionContext`?",
+      "reference": "Add the PipeFactory here"
+    },
+    {
+      "prediction": "I think we should keep the `common.ChaosImpl` interface.",
+      "reference": "please add `var _ impltypes.ChaosImpl = (*Delegate)(nil)`"
+    },
+    {
+      "prediction": "I think this should be called `skipCleanup` or something like that. `cleanExpiredFiles` sounds like it's going to clean up expired files, but it doesn't.",
+      "reference": "nit: is this added deliberately?"
+    },
+    {
+      "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": "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 this should be `maybePromise && typeof maybePromise.then === 'function' && typeof maybePromise.catch === 'function'`",
+      "reference": "we don't use this anymore, so we can delete it."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to log this, but I think it's useful to know where the telemetry config is coming from.",
+      "reference": "nit: change to \"Telemetry configuration loaded from '%s'\""
+    },
+    {
+      "prediction": "00601024704010402040304\")\n \tif !bytes.Equal(b.Bytes(), expected) {\n \t\tt.Errorf(\"Serialised block witness does not match expected value\")\n \t}\n```\nReview comment?",
+      "reference": "Why did this value change?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `gcpduckv1alpha1` import and just use `duckv1alpha1` everywhere.",
+      "reference": "maybe `duckpubsubv1alpha1` ?"
+    },
+    {
+      "prediction": "I think this should be `manifest['name']` instead of `manifest['version']`.",
+      "reference": "The id is not mandatory for web extensions. Could you update this to support a web extension without an id?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we need these constants anymore. They are only used in the `CloudAuditLogsSource` function, which is only called in the `CloudAuditLogsSourceWithTestImpl` function, which is only called in the `TestCloudAuditLogsSource` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called in the `TestCloudAuditLogs` function, which is only called",
+      "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 `defer 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 this should be moved to `nano::node::start` and `nano::node::stop` so that the websocket server can be stopped when the node is stopped.",
+      "reference": "The bind address should be configurable."
+    },
+    {
+      "prediction": "Shouldn't this throw an exception?",
+      "reference": "Maybe the empty `delete()` should be implemented as default in the interface."
+    },
+    {
+      "prediction": "This should be `CMAKE_SOURCE_DIR \"/caffe/test/test_data/sample_data.h5\"`",
+      "reference": "How to ensure CMAKE_SOURCE_DIR is set correctly?"
+    },
+    {
+      "prediction": "Why do we need to do this?",
+      "reference": "safer to make a copy of incoming *big.Int"
+    },
+    {
+      "prediction": "I don't think we need to set this to 10 minutes. 10 minutes is a very long time to wait for a list of images.",
+      "reference": "how did we choose this value? does this hold good for Windows too?"
+    },
+    {
+      "prediction": "I think this should be `errors.New(\"checkpoint/restore not supported\")`",
+      "reference": "To follow the other build tags, maybe we can call it `criu`?"
+    },
+    {
+      "prediction": "I think this should be `connect_timeout: 30`",
+      "reference": "Should we load this from the config file too?"
+    },
+    {
+      "prediction": "I don't think we need to validate this in the webhook. The webhook is used to validate user input, and this is not user input. This is an implementation detail of the reconciler.",
+      "reference": "Can you file an issue for this?"
+    },
+    {
+      "prediction": "I think this should be `this[kIsMaster] != null`",
+      "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 to do this. I think it should be done in the `Bolt::Inventory` constructor.",
+      "reference": "This whole function feels messy. I don't have a better idea at the moment though."
+    },
+    {
+      "prediction": "I don't think you need to cast to `Long` here. `parameters.get(\"type\")` is already a `Long`.",
+      "reference": "should use Number instead of Long"
+    },
+    {
+      "prediction": "I think this should be `var _ influxdb.Provider = (*MockProvider)(nil)`",
+      "reference": "Can you rename this and the file `InfluxDBProvider` and `influxdb_provider.go` respectively?"
+    },
+    {
+      "prediction": "`len` is already defined as `originalChildren.length`, so you don't need to redefine it here.",
+      "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 we should rename this to `noDataString`",
+      "reference": "missing space before `TextView`"
+    },
+    {
+      "prediction": "This should be a `uint32_t`.",
+      "reference": "so this attribute is used as u8 in code, but saves as int32? Why?"
+    },
+    {
+      "prediction": "I think this should be `adaptiveScalarEncoder` instead of `adaptiveScalarEncoder`",
+      "reference": "can RDSE be added to the list? It has its capnp files."
+    },
+    {
+      "prediction": "`client.Reader` doesn't have a `List` method.",
+      "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 don't think we need to do anything here.",
+      "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 think this should be `from tests.namespace_test.NamespaceB.TableInB import TableInB, TableInBT`",
+      "reference": "this is generated code.. these changes will need to be made in the code generator to have them stick."
+    },
+    {
+      "prediction": "I think this should be `var shadowSupport = !!document.body && typeof document.body.attachShadow === 'function';` so that `shadowSupport` will be `false` if `document.body` is `null` or `undefined`.",
+      "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: I think we can remove the `require('./unified-spec-runner/runner').runUnifiedTest` and just do `const runUnifiedTest = require('./unified-spec-runner/runner').runUnifiedTest;`",
+      "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 don't think this is correct. `$options` is not a string, it's an object with a `value` property that is a string.",
+      "reference": "Why is this change necessary? `handleSingle()` will cast it to a string, no?"
+    },
+    {
+      "prediction": "I think this should be `default String name() { return spec().name(); }`",
+      "reference": "This newly added method should be well defined, should it return `TableIdentifier` or just `String`?"
+    },
+    {
+      "prediction": "I don't think we need this constant. We can use `sprintf` directly in the exception message.",
+      "reference": "have a relation with a product"
+    },
+    {
+      "prediction": "nit: I think it would be better to have a `DeletePod` function that does this check and then calls `Delete` if the pod exists. That way we don't have to remember to do this check in every `AfterEach` block.",
+      "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 right thing to do. I think we should be more explicit about what exceptions we want to catch and re-raise.",
+      "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": "I'm not sure if this is the right thing to do, but I think it's the right thing to do. We don't need to convert the action to a transfer/vote, we just need to convert the action to a nonce.",
+      "reference": "This switch statement can be removed. Just return q.items[act.Nonce] != nil"
+    },
+    {
+      "prediction": "I don't think you need to use `equals` here. `==` should be enough.",
+      "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": "I think it would be better to use `IEnumerable<string> Filters { get; set; }` instead of `IEnumerable<string> Filter { get; set; }`.",
+      "reference": "What's the scenario for passing multiple filters?"
+    },
+    {
+      "prediction": "nit: I think it would be more readable to use `auto [id, term, _, log] = iter->logEntry();` instead of `auto logEntry = iter->logEntry();` and `auto log = std::get<3>(logEntry);`",
+      "reference": "move to after check log.empty()"
+    },
+    {
+      "prediction": "This import is not used. Please remove it.",
+      "reference": "This line needs to be removed"
+    },
+    {
+      "prediction": "I think this should be `com.google.cloud.cloudbuild.build.v1.statusChanged` instead of `google.cloud.cloudbuild.build.v1.statusChanged`",
+      "reference": "Can we delete such const like `CloudBuildSourceEvent` and `CloudBuildSourceBuildId` since all those are contained under schemas/v1?"
+    },
+    {
+      "prediction": "I don't think we need this file.",
+      "reference": "This is unused."
+    },
+    {
+      "prediction": "I don't think we need to include `BaggageFormat` in the default propagator.",
+      "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 right thing to do, but it seems to be the only way to get this to work.",
+      "reference": "Could the name of this controller just be shortened to `TwitterCardsController`? Is the `Player` there adding information?"
+    },
+    {
+      "prediction": "I think this should be `if uh.Dynamic && uh.Unhealthy { return true }`",
+      "reference": "Backends with hostnames that depend on the request are always up?"
+    },
+    {
+      "prediction": "I think it's better to use `jsonpatch.JsonPatchException` and `jsonpointer.JsonPointerException` instead of `jsonpatch.JsonPatchException` and `jsonpointer.JsonPointerException`.",
+      "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": "I'm not sure if this is the right place to remove the default, but it seems to be the only place where it's defined.",
+      "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 this should be `AZURE_KEYVAULT_AUTH_VIA_AZURE_CLI`",
+      "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": "`sessionStorage` should be renamed to `sessionStore` or `sessionStore` should be renamed to `sessionStorage`.",
+      "reference": "Depend on interfaces not on structures"
+    },
+    {
+      "prediction": "I think it's better to call `gStorageServer->stop()` here.",
+      "reference": "This fix looks good. My concern, don't forget to fix metad too!"
+    },
+    {
+      "prediction": "I think this should be `earliest.BID() == NullBranchID && latest.BID() == LocalSquashBranchID`",
+      "reference": "Just noticed this nil context. Perhaps plumb through ctx too, or change to `Debug`?"
+    },
+    {
+      "prediction": "I don't think we need to set the `Content-Type` header here. It's only used when sending a POST request to the server, and we don't do that here.",
+      "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 `DEFAULT_ROS2_SERVER_GUIDPREFIX` instead of `DEFAULT_ROS2_SERVER_GUIDPREFIX`.",
+      "reference": "Again use of `constexpr` is recomended."
+    },
+    {
+      "prediction": "This should be `except ImportError:`",
+      "reference": "Minor point but I would call this ``redim_graph`` instead."
+    },
+    {
+      "prediction": "nit: I think we can remove this import.",
+      "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 don't think we should log this as a warning. If we can't enable IP forwarding, we should fail to start the service.",
+      "reference": "Now it's double logging, is not it? Because `service.ipForward.Enable()` logs extra `warning`"
+    },
+    {
+      "prediction": "Trailing whitespace detected.",
+      "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 think this should be `if (localConnection && localConnection.type == Blockly.OUTPUT_VALUE) { updatePreviews = true; // Always update previews for output connections. } else if (Blockly.localConnection_ && Blockly.highlightedConnection_) { var xDiff = Blockly.localConnection_.x_ + dxy.x - Blockly.highlightedConnection_.x_; var yDiff = Blockly.localConnection_.y_ + dxy.y -",
+      "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 this should be `find_policy_violations` to be consistent with the other scanners.",
+      "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 think this should be `union X { X, Y } table X {` instead of `union X { X, Y } table X { }`.",
+      "reference": "This seems like an incomplete schema definition. will that have any effect on the tests?"
+    },
+    {
+      "prediction": "I don't think we need to call `CheckCanLoadFromBin` here. We can just call `SetHeader` directly.",
+      "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 about this change. I don't think we want to change the type of `err` to `*exec.ExitError` if it's not an `*exec.ExitError` to begin with. I think we want to keep the type of `err` as `error` and just check if it's an `*exec.ExitError` in the `if` statement.",
+      "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": "why not use `std::move(name)` here?",
+      "reference": "`Status` has identical semantics with `StatusOr<bool>`."
+    },
+    {
+      "prediction": "I think this should be `/mutate/add-labels` instead of `/mutate/addlabels` to be consistent with the rest of the paths.",
+      "reference": "why is this named in a different style from other paths?"
+    },
+    {
+      "prediction": "I think we can remove this check now that we have `util.IsPaused`",
+      "reference": "Swapped these to ensure that we aren't passing a nil cluster in to util.IsPaused"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I think it's better than having a bunch of `if` statements.",
+      "reference": "I don't see pkgManager used anywhere, is there a reason we need this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be the caller's responsibility to provide an absolute path.",
+      "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 if this is the best way to do it, but it seems to work.",
+      "reference": "There's an awful lot of test failures here because `localFieldPathType.schema` may not contain a `getters` array."
+    },
+    {
+      "prediction": "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": "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 we can remove this import.",
+      "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": "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": "I'm curious why you are using `draw` in this route file?"
+    },
+    {
+      "prediction": "Shouldn't this be a 404 if the file doesn't exist?",
+      "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": "I'm not sure this is the right thing to do, but I don't know what else to do.",
+      "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 `oidcProviderName` or something like that. `clutchProviderName` sounds like it's the name of the provider, but it's actually the name of the service.",
+      "reference": "to fix the docs build failure, make this const var named without the suffix `Name`"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "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 think this interface should be in the `com.fsck.k9.ui.settings` package.",
+      "reference": "Same here. Feels like the wrong location."
+    },
+    {
+      "prediction": "I don't think we need to import this from `qutebrowser.browser.downloads` here. We can just import it from `qutebrowser.browser.downloads._path_suggestion`.",
+      "reference": "Please make `_path_suggestion` public (i.e. remove the `_`) in `downloads.py`."
+    },
+    {
+      "prediction": "I don't think this is the right place for these flags. They should be in the `caddytls` package.",
+      "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 `m_is_objc_enum = enum_decl->isObjC();`",
+      "reference": "Does this not apply to C enums on Linux?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the label. I think it should be added in the `handleRetryableWriteError` function.",
+      "reference": "Should this check if the error is `RetryableWriteError` before adding the label?"
+    },
+    {
+      "prediction": "I'm not sure why this is necessary. `rollup-plugin-commonjs` is supposed to automatically resolve `.default` for ES modules.",
+      "reference": "Oh wow, that's one complicated require :)"
+    },
+    {
+      "prediction": "`request_context.go shows how to usie context.Context with SDK requests.` -> `request_context.go shows how to use context.Context with SDK requests.`",
+      "reference": "do you mean how to \"use\""
+    },
+    {
+      "prediction": "I'm not sure if this is a bug or not, but it seems like `children.props.ref` should be `undefined` when `children.ref` is `undefined`.",
+      "reference": "We can't really test this anymore since ref and props.ref are at the backing node level now"
+    },
+    {
+      "prediction": "nit: `currentTask` -> `currentFileScanTask`",
+      "reference": "Did you intend to set this in the constructor?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what else to do either. We don't have a default AMI for Ubuntu 18.04, and we don't have a default AMI for Ubuntu 20.04, and we don't have a default AMI for Ubuntu 20.10, and we don't have a default AMI for Ubuntu 21.04, and we don't have a default AMI for Ubuntu 21.10, and we don't have a default AMI for Ubuntu 22.04, and we don't have a default AMI for Ubuntu 22.10, and we don't have a default AMI for Ubuntu 23.04, and we don't have a default AMI for Ubuntu 23.10, and we don't have a default AMI for Ubuntu 24.04, and we don't have a default AMI for Ubuntu 24.10, and we don't have a default AMI for Ubuntu 25.04, and we don't have a default AMI for Ubuntu 25.10,",
+      "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_VERSION_3)`",
+      "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: %v\", err)`",
+      "reference": "Should we take out this change? This was only in there to try to brute force the start of the apiserver yesterday."
+    },
+    {
+      "prediction": "Line is too long. [85/80]",
+      "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": "Why is this removed?",
+      "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 don't think this is the right place to set the default datasource. I think it should be done in `Propel::init()`.",
+      "reference": "@lunika Please see this. It allows us to do `Propel::getConnection()` easily, as we have only one database."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could find to get the `AspNetTags` from the `scope.Span.Tags`.",
+      "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 `const DRIVE_FILES_FIELDS = `kind,nextPageToken,incompleteSearch,files(${DRIVE_FILE_FIELDS})`",
+      "reference": "why do we need to explicitly declare these fields?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it works for now.",
+      "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": "Metrics/LineLength: Line is too long. [117/80]",
+      "reference": "This is no longer needed and should be removed."
+    },
+    {
+      "prediction": "I don't think we need to check for `nil` here. If `request` is `nil` then `json.Unmarshal` will return an error.",
+      "reference": "could use named return params here to avoid declaring the opts, and using naked returns instead."
+    },
+    {
+      "prediction": " <img src={item.images[0].url} />\n   }\n \n```\nReview comment?",
+      "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 `parentWeight uint64, parentStateID cid.Cid`",
+      "reference": "\"... after protocol version 1\"?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `re.M | re.DOTALL` instead of just `re.DOTALL`. The reason is that `re.M` makes `^` and `$` match the start and end of each line, while `re.DOTALL` makes `.` match any character, including newlines. So if we want to match the start and end of the whole string, we need both `re.M` and `re.DOTALL`. Otherwise, `^` and `$` will only match the start and end of the whole string, not the start and end of each line.",
+      "reference": "Is there a unit test that covers this regexp? I am afraid that changing it might break things =/"
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it should be `@` + `name.toLowerUnderscore()`.",
+      "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 don't know if this is related to this PR or not, but I noticed that the `{{ url }}` placeholder is not replaced with the actual URL. Is this expected?",
+      "reference": "Why this change was needed? Because in `po` files it's with curly braces?"
+    },
+    {
+      "prediction": " | (0x0007fff0 & off >> 2 << 5);\n+        ASSERT(off + 0x80000 < 0x100000);\n+        *pc_writable = (enc & 0xfff8001f) | (0x0007fff0 & (off >> 2 << 5));\n     } else if ((enc & 0xff000000) == 0x10000000) { /* BL */\n         ASSERT(off + 0x8000000 < 0x10000000);\n-        *pc_writable = (0x10000000 | (0x03ffffff & off >> 2));\n+        *pc_writable = (0x10000000 | (0x03ffffff & (off >> 2)));\n     }\n }\n```\nReview comment?",
+      "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 don't think we need this file.",
+      "reference": "seems to be redundant (mention `docusaurus`   )"
+    },
+    {
+      "prediction": "I think this should be `CherryPick 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": "manager_env_prefix -> master_env_prefix",
+      "reference": "Maybe just `env_prefix`, I don't think this is referring to the WQ manager."
+    },
+    {
+      "prediction": "Please remove empty lines at the end of the file.",
+      "reference": "Unrelated line change :)"
+    },
+    {
+      "prediction": "I don't think we need to log here. The caller will log the error.",
+      "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 `html = html.replace(\"Countly\", countlyGlobal.company);` instead of `html.replace(\"Countly\", 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 don't think we need to check for `error == null` here. If `error` is `null` then `error instanceof MongoParseError` will be `false`.",
+      "reference": "ticket for the 4.0 epic?"
+    },
+    {
+      "prediction": "I think it's better to move this line to the beginning of the `if cfg.model.get('neck', False):` block.",
+      "reference": "`cfg.model.get('neck')` will return None if neck does not exist, thus we can omit the default value."
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "Trailing whitespace detected."
+    },
+    {
+      "prediction": "I think this should be `CXXFlags` instead of `cxxflags` to be consistent with `CFlags`.",
+      "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": "nit: extra space before `PFX`",
+      "reference": "Update year range in Copyright notice, and elsewhere too."
+    },
+    {
+      "prediction": "This should be `AccessKeyIDSecretRef` and `SecretAccessKeySecretRef`",
+      "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 `!params[:read_tou] || params[:read_tou] == \"0\"` instead of `params[:decline] || !params[:read_tou] || params[:read_tou] == \"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": "I'm not sure if this is the best way to test this, but it's the only way I could get it to work.",
+      "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 \tFailsafeInboundIPs          []net.IP  `config:\"ip-list;127.0.0.1;die-on-fail\"`\n \tFailsafeOutboundIPs         []net.IP  `config:\"ip-list;127.0.0.1;die-on-fail\"`\n \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 don't think you need the `static_cast<unsigned int>(rootedAtAtom) < mol.getNumAtoms()` check here.",
+      "reference": "I'm fairly sure this precondition is always true due to the mod (%)"
+    },
+    {
+      "prediction": "nit: I think this should be `assert !seen.contains(info.info.name) : \"seen twice: \" + info.info.name + \" in \" + readerMap.keySet();`",
+      "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 don't think this is correct. If a config file does not have version, it is assumed to be zero, but that does not mean that all fields listed in migrate() might be changed if an actual value matches to default value from a previous version. It means that all fields listed in migrate() might be changed if an actual value does not match 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": "I don't think we need to use `sha1` here. `md5` should be enough.",
+      "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 don't think we need to import `ks.DatetimeIndex` here. We can just use `DatetimeIndex`.",
+      "reference": "Could you also update it in `at_time`?"
+    },
+    {
+      "prediction": "nit: I think we can use `Boolean.TRUE.equals(options.get(\"locality\").map(Boolean::parseBoolean).orElse(localityFallback))`",
+      "reference": "Because schema is non-final we can't just switch this to a lambda"
+    },
+    {
+      "prediction": "I think this should be `if (exception.MustBeRethrown(\"While resolving function '{0}' following exception occured: {1}\", functionName, exception))`",
+      "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 is better to add a new method `setInvocationMonitorType` in `InvocationMonitor` and call this method in `process` method of `InvocationStartProcessingEventListener` and `InvocationEndProcessingEventListener`.",
+      "reference": "why? event.getInvocationType is not you want?"
+    },
+    {
+      "prediction": "I think this should be `string` instead of `[]string`.",
+      "reference": "nit: how about `GitSHA`?"
+    },
+    {
+      "prediction": "I think this should be `static const char *FAIL_DIR = \"makeflow.failed.%d\";`",
+      "reference": "If you move the other functions, this should also be moved to `makeflow_gc.h`."
+    },
+    {
+      "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 listen and returns the duration in seconds. Then you can just call that function here and in the other places where you need to get the duration.",
+      "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 if this is the best way to do it, but it seems to work.",
+      "reference": "Can we add `_dom` as an argument to createVNode here? I think it might be shorter (could totally be wrong!)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done after the `popf` instruction is added to `ilist` but before the `popf` instruction is executed.",
+      "reference": "I think that the iret handling is not yet good."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's better to keep the `}` on the same line as the `return` statement.",
+      "reference": "Huh, that's kinda weird. I get it, but it's unexpected to me..."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "why is this removed? it is important when you upload from stream of unknown size"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [121/120]",
+      "reference": "Out of curiosity, why is this needed? Does it enforce what class can be assigned?"
+    },
+    {
+      "prediction": "I think this should be `0777` instead of `os.ModePerm` which is `0777 & ^umask`",
+      "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 don't think it's a good idea to have a public property for this. I think it would be better to have a method `GetLastExecutor` that would return the last executor or `null` if there is no executor yet.",
+      "reference": "factory should be stateless if possible, looks like much complexity added"
+    },
+    {
+      "prediction": "Should this be `self.output().table.project_id`?",
+      "reference": "Could you please add short description for the default value as well?"
+    },
+    {
+      "prediction": "I don't think we should use Fatal here. Fatal is used for fatal errors that we can't recover from. I think this is a recoverable error, so we should use Error here.",
+      "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` and `write.delete.parquet.row-group-check.max-record-count`",
+      "reference": "this can be integer"
+    },
+    {
+      "prediction": "I think this should be `font: {{ font['completion.category.font'] }};`",
+      "reference": "nitpick: Please remove the blank line here"
+    },
+    {
+      "prediction": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 this should be `if (authenticationPlugin != null && authenticationPlugin.getMetricRegistry() == null) {` instead of `if (authenticationPlugin != null && authenticationPlugin.plugin.getMetricRegistry() == null) {`",
+      "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 `if (it->second.getNeedAccess() && !player->getGroup()->access) { return TALKACTION_BREAK; }`",
+      "reference": "shouldn't this be `return TALKACTION_BREAK;` as the player does not meet the required group access?"
+    },
+    {
+      "prediction": "I think this should be `awaitBlockIsCreatedByProposer(minerNode1)`",
+      "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 to do this, but I don't know where else to do it.",
+      "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 think `pdf = kdf.to_pandas().groupby(level=1).apply(lambda gpdf: gpdf.drop('level_1', 1).set_index('level_0').transpose()).reset_index(level=1)` is better than `pdf = kdf.to_pandas().groupby(level=1).apply(lambda gpdf: gpdf.drop('level_1', 1).set_index('level_0').transpose()).reset_index(level=1).reset_index(level=1).drop(columns='level_1')`",
+      "reference": "We can just use `.stack()` here? Then I guess we can reuse when supporting multi-index columns."
+    },
+    {
+      "prediction": "I think we can just remove the `#if NETSTANDARD2_0` and `#else` and `#endif`",
+      "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 think this should be `in ReadOnlySpan<byte> context`.",
+      "reference": "should the parameters be called reference or context?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get a `String` from a `StringIO`.",
+      "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.toDisplayString()` instead of `url.toString()`.",
+      "reference": "This should be `toDisplayString()` to not contain e.g. passwords."
+    },
+    {
+      "prediction": "Line is too long. [84/80]",
+      "reference": "Can we just say that having a non-nil `autocomplete_path` implies that autocomplete is enabled?"
+    },
+    {
+      "prediction": "I'm not sure why this change is necessary. I don't see any references to `gogo/protobuf` in the code.",
+      "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'm not sure if we want to remove these routes. I think we should keep them around for now, and if we don't need them in the future, we can remove them in a separate PR.",
+      "reference": "Are these gone routes? Do we need to 301 them?"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "move them to internal group of imports"
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "can we rewrite this as: `ignored event '%s' for cstor pool '%s'`"
+    },
+    {
+      "prediction": "I think this should be `class_name: 'User'` instead of `class_name: User`",
+      "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 don't think this is correct. `findElements` should return `List<WebElement>`, not `List<T>` where `T` extends `WebElement`.",
+      "reference": "This change should also probably go into the corresponding method of the abstract By class?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "why this remove?"
+    },
+    {
+      "prediction": "I think this should be `io.ReadCloser` instead of `FileReader`.",
+      "reference": "As the test failures indicate, making this change is going to mean changing everything which currently implements this interface."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we have an office hours page anymore.",
+      "reference": "Do we want to redirect this to the forum or something in case people have it linked/bookmarked?"
+    },
+    {
+      "prediction": "I don't think this is the right thing to do. If we don't have a name or email, we shouldn't have a background at all.",
+      "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 don't think we need to store this in the database. We can just fetch it from the API.",
+      "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 right thing to do, but I think it's the least bad option. We don't want to log this, because it's not an error, and we don't want to raise an exception, because we don't want to prevent the user from using the rest of the library.",
+      "reference": "This is the common way of dealing with optional dependencies"
+    },
+    {
+      "prediction": "I think this should be `duckv1.KResource` instead of `duckv1.Source`.",
+      "reference": "nit: we can replace all `kr` in this file with like `src`."
+    },
+    {
+      "prediction": "I think this should be `readPreference.mode || readPreference.preference`",
+      "reference": "same concern here as above"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the `:read` option from the criteria.",
+      "reference": "Thank you for this, I gather this repairs failures that I've seen in another PR."
+    },
+    {
+      "prediction": "I don't think we need to do this here. The `distinct` will be done in the `buildManifestFileDF` method.",
+      "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 should return an error here and let the caller handle it.",
+      "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` and `value_name` should have default values.",
+      "reference": "Seems the default value of `melt`'s `var_name` at namespace.py should be changed as well."
+    },
+    {
+      "prediction": "I think it would be good to add a test for a service that doesn't use the `X-Amz-Target` header, e.g. `ec2`.",
+      "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 don't think we should use `singular` here. We should use `plural` because we are deleting multiple items.",
+      "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": "`/auth` should be `/authorize`",
+      "reference": "in case you want to rename one of the plugins when instantiating, from options?"
+    },
+    {
+      "prediction": "I think this should be `ctu-func-mapping` instead of `clang-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 don't think this is a good test. I think it should be something like `expect(visibleCalc.startColumn).toBe(0); expect(visibleCalc.endColumn).toBe(3); expect(visibleCalc.startPosition).toBe(0); expect(visibleCalc.endPosition).toBe(200);`",
+      "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 think this should be `KOKKOS_IMPL_ENABLE_CUDA_HOST_INIT_CHECK` instead of `KOKKOS_IMPL_TURN_OFF_CUDA_HOST_INIT_CHECK`.",
+      "reference": "So the intention is you configure with `-CMAKE_CXX_FLAGS=\"-D KOKKOS_IMPL_TURN_OFF_CUDA_HOST_INIT_CHECK\"`?"
+    },
+    {
+      "prediction": "I don't think we should be setting these values to 1.0 and 0.0.",
+      "reference": "This was a bug."
+    },
+    {
+      "prediction": "I think it would be better to use `sprintf('%s%s', $useLazyLoading ? 'lazy' : '', $isAttributeClassExistsAndNotEmpty ? ' ' . $attributes['class'] : '')` instead of `sprintf('%s%s', $useLazyLoading ? 'lazy' : '', $isAttributeClassExistsAndNotEmpty ? ' ' . $attributes['class'] : '')`",
+      "reference": "Is OK that we don't have some placeholder?"
+    },
+    {
+      "prediction": "I think this should be 1. We don't want to use more threads than we have cores.",
+      "reference": "Sectorbuilder behaves differently depending on whether it's given 1 or more threads. It won't seal if only given 1."
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "Sort of believe We should put MetricManager in constructor parameter given this case."
+    },
+    {
+      "prediction": "nit: I think it would be better to have `sourceId` and `fieldId` be the same.",
+      "reference": "Do we need to check `fieldId` is larger than 1000?"
+    },
+    {
+      "prediction": "I don't think this is the right place to call `check_acked_status()`. It should be called in `StatefulWriter::unsent_change_added_to_history()` after `unsent_change_added_to_history()` is called in `StatefulWriter::write()`.",
+      "reference": "I think this should be called after the if below (the one for disable positive acks)"
+    },
+    {
+      "prediction": "nit: can we use `org.apache.hadoop.fs.Path` instead of `java.net.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 we can remove the `os.Clearenv()` calls now.",
+      "reference": "Should these restore the stashed env after the test runs?"
+    },
+    {
+      "prediction": "I don't think this is correct. `scope.args.kw_defaults` is a list of `None`s, so `d is not None` will always be `False`.",
+      "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 don't understand why we need this struct. Can't we just use `accountDelta`?",
+      "reference": "Can ndelta differ depending on when the deltas are compacted? That is, when intermediate updates are dropped?"
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "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 don't think this is correct. If CRIU crashes, we don't want to close the connection to the client. If CRIU unexpectedly exits, we don't want to close the connection to the client. If CRIU is killed, we don't want to close the connection to the client. We only want to close the connection to the client if CRIU succeeds.",
+      "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 `npm run test` instead of `npm run test:js`.",
+      "reference": "This task can probably be deleted since we don't need gulp to run Jest."
+    },
+    {
+      "prediction": "I don't think this is necessary. `SocketError` is a subclass of `Errno::ECONNREFUSED` and `Errno::ENOTCONN`.",
+      "reference": "Doesn't this need to include `Errno::ECONNRESET` to fix the issue?"
+    },
+    {
+      "prediction": "I think it should be `Faker::Movie.director` instead of `Faker::Movie.qirector`",
+      "reference": "never heard of a qirector... what's their role again? (also should be 'for a movie')"
+    },
+    {
+      "prediction": "Why do we need to check `tagContexts_.empty()` here?",
+      "reference": "Implement the \"collect all props for one vertex\" logic inside QueryVertexPropsProcessor"
+    },
+    {
+      "prediction": "I think this should be `if (stream.fail ())`",
+      "reference": "This is to prevent parsing things like \"1.1\" as \"1\" until we implement it properly."
+    },
+    {
+      "prediction": "I think this should be `from nupic.frameworks.opf.experiment_runner import runExperiment, _getCheckpointParentDir as getCheckpointParentDir` so that we can use `getCheckpointParentDir` instead of `_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 it would be better to put this in the UIA settings panel, not the advanced settings panel.",
+      "reference": "Why is a restart necessary?"
+    },
+    {
+      "prediction": "I think we can remove the `populateGPUEnvironmentVariables` method from the `Task` struct. It's only used in the `addGPUResource` method, which is only called when `cfg.External.Enabled()` is false.",
+      "reference": "Should setting the task runtime also go under the above if condition?"
+    },
+    {
+      "prediction": "I think this should be `const defaultListEntriesPageSize = 50`",
+      "reference": "SPIRE uses the leadingLowerCase convention for unexported constants"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "The newest JSchema code generator orders `System` usings first."
+    },
+    {
+      "prediction": "I think we can remove the `caseSensitive` parameter from the `IcebergObjectInspector.create` method and always use `tableSchema.caseInsensitiveSelect(selectedColumns)`.",
+      "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 this should be moved to the `cStorPoolAddEventHandler` function.",
+      "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 thing to do, but it seems to work.",
+      "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 think these should be constants in the `iptables` package.",
+      "reference": "Why isn't this 4789 (which I think is the Felix default)?"
+    },
+    {
+      "prediction": "I don't think we need to have a separate reason for this. We can use `FailureImportOperations` for this.",
+      "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": "Line is too long. [81/80]<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing.<br>Space inside ) missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside ( missing",
+      "reference": "I suspect this isn't needed because I think `rb_intern` gets marked already."
+    },
+    {
+      "prediction": "I think this should be `log.fine`",
+      "reference": "This change is incorrect: the current log level is correct."
+    },
+    {
+      "prediction": "I think this should be `VERSION_CURRENT = VERSION_LOW_CARDINALITY_LEAVES + 1;`",
+      "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 don't think we need to remove the runningFlows cache from the web server. We can keep the cache in the web server and update the cache in the web server when we receive an update request from the executor. This way, we don't need to fetch the flow info from the DB every time we receive an update request from the executor. Also, we don't need to sync the updateTime between the web server and the executor. We can keep the updateTime in the web server and update the updateTime in the web server when we receive an update request from the executor. This way, we don't need to fetch the flow info from the DB every time we receive an update request from the executor. Also, we don't need to sync the updateTime between the web server and the executor. We can keep the updateTime in the web server and update the updateTime in the web server when we receive an update request from the executor. This way, we don't need to fetch the flow info from the DB every time we receive an update request from the executor. Also, we don't need to sync the updateTime between the web server and the executor. We can keep the updateTime in the web server and update the updateTime in the web server when we receive an update request from the",
+      "reference": "Please follow the todo naming convention."
+    },
+    {
+      "prediction": "I don't think we want to expose the `Ready` condition here. I think we should only expose the `Issuing` condition.",
+      "reference": "These are no longer used and changes should instead be made directly to the files in `deploy/chart/crds`   /assign @JoshVanL /unassign"
+    },
+    {
+      "prediction": "`torch.cat((xy, pred_map[..., 2:4]), axis=-1)` -> `torch.cat((xy, pred_map[..., 2:4]), dim=-1)`",
+      "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 `errors.New(\"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": "Metrics/LineLength: Line is too long. [113/80]",
+      "reference": "Is this a testing debug statement?"
+    },
+    {
+      "prediction": "I don't think we should be using `console.log` here.",
+      "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 don't think you need this lock here.",
+      "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": "`remote_config` -> `remoteConfig`",
+      "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 `if (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 don't think we need this anymore. The `beforeEach` and `afterEach` in `prepareDatabaseForSuite` should take care of it.",
+      "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 think it is better to put this argument in the `__init__` of `LoadAnnotations` instead of `__init__` of `OpenImagesCustomDataset`.",
+      "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 test should be updated to use the `trace.TracerProvider` interface instead of the `trace.Provider` interface.",
+      "reference": "Should we change the name of this test struct to be `testTracerProvider` as well?"
+    },
+    {
+      "prediction": "remove empty line",
+      "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 if this is the right place to inject the print script. I think it should be injected in `browsertab.AbstractTab._inject_early_js` so that it is injected for all tabs.",
+      "reference": "No need for a separate variable here - just call `utils.read_file` directly below."
+    },
+    {
+      "prediction": "Why do we need to stop the timer here?",
+      "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 place to put this, but I think it would be good to have a `Senders` stage in the `SyncStage` enum. This stage would be responsible for recovering the senders from the signatures and re-writing the block bodies with the recovered senders.",
+      "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 don't think we need to change the order of the fields here.",
+      "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 think this should be `Disconnected` instead of `NotConnected`.",
+      "reference": "Could we public events if `manager.serStatus()`?"
+    },
+    {
+      "prediction": "I think this should be a warning, not an error.",
+      "reference": "could be misread as `<body>`, maybe something redundant but very clear like \"before the closing `</body>` tag\"?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be able to use `WriteAsync` here.",
+      "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 do this, but it works for now.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `CstorVolumeSource` instead of `CstorVolumeSource`",
+      "reference": "nit: Should we have something like `volumeName@snapShotname` and this will be filled only if it is a clone creation"
+    },
+    {
+      "prediction": "I think this should be `0700` instead of `0755`.",
+      "reference": "This seems like an improved pattern :)"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. We should be able to detect if snakebite is installed or not, and if it is, then we should use it. If not, then we should fall back to the old behavior.",
+      "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": "I'm not sure if this is the best way to do this, but it seems to work. I'm not sure if there's a more idiomatic way to do this in Ruby.",
+      "reference": "Why are we filtering content twice? It seems like just filtering it in PAL should work."
+    },
+    {
+      "prediction": "I think this should be `ACCOUNT_STATUS_NO_CLIENT`",
+      "reference": "These should probably move to `constants.js`"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "Again, you don't want to depend on glassfish."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "it should be `Faker::Space.launch_vehicle`, otherwise it will crash"
+    },
+    {
+      "prediction": "Shouldn't this be `if (this.ShouldAutoArchive(fileName, logEvent, bytes.Length) && this.ArchiveNumbering == ArchiveNumberingMode.Date && this.ArchiveEvery != FileArchivePeriod.None)` ?",
+      "reference": "Isn't it quite expensive to check whether the file exists on every write?"
+    },
+    {
+      "prediction": "I think this should be `if (webdriver.http.CorsClient.isAvailable() || url[0] == '/') {`",
+      "reference": "Please hide this behind a flag and add a setter function to the builder."
+    },
+    {
+      "prediction": "I think this should be `verb: \"DELETE\", path: backupPath(\"/deleteAll\", volume.APIVersion), fn: vd.cloudBackupDeleteAll,`",
+      "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 it's a good idea to use `warn` here. If the port is occupied, it's not a warning, it's an error.",
+      "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": "nit: I think we can remove this constant and use `SELECT_COUNT + \" FROM (%s)\"` directly in the `getCountQuery` method.",
+      "reference": "Did you mean for this to be `FORM`, instead of `FROM`?"
+    },
+    {
+      "prediction": "I think this should be `+kubebuilder:validation:Enum=true;false`",
+      "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 don't think we need to check `verbose` here. If `verbose` is `false`, then `command` will be `nil` and the `puts` call will be a no-op.",
+      "reference": "There's a syntax error on this line."
+    },
+    {
+      "prediction": "I think this should be `return result;` instead of `goto out;`",
+      "reference": "You lose the value of `result` when you `goto out`"
+    },
+    {
+      "prediction": "I think this should be `this.ratio_ || 0`.",
+      "reference": "Just a style thing, why not `this.ratio_ || 0`?"
+    },
+    {
+      "prediction": "I think this should be `ipNet.IP.Mask(ipNet.Mask)` instead of `getMaskedIp(ipNet)`. `ipNet.IP.Mask(ipNet.Mask)` returns a `net.IP` and `getMaskedIp(ipNet)` returns a `net.IPNet`.",
+      "reference": "I think the stdlib's `net.IP` has a `Mask()` method, does that do what we need?"
+    },
+    {
+      "prediction": "I think this should be called `substitute` instead of `Substitute` to be consistent with the rest of the codebase.",
+      "reference": "Why does substitute need to be public? Are we allowing/encouraging variable usage when using Daisy as a library?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `webengineelem.py` instead.",
+      "reference": "We should probably use ctrl-enter if `tab` is given, just like in `follow_selected`?"
+    },
+    {
+      "prediction": "nit: I think we should rename this to `tchannelDispatcher` to make it clear that it's a TChannel dispatcher.",
+      "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 `long` instead of `Long`.",
+      "reference": "we can just have this method call the next, passing in the file metadatas (deduplication of logic)"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think it should be done in the `TabBar` class, not in the `TabBarStyle` class. The `TabBarStyle` class should not know anything about the `TabBar` class. The `TabBarStyle` class should only know about the `QTabBar` class. The `TabBarStyle` class should not know anything about the `TabbedBrowser` class. The `TabBarStyle` class should only know about the `TabBar` class. The `TabBarStyle` class should not know anything about the `Window` class. The `TabBarStyle` class should only know about the `TabBar` class. The `TabBarStyle` class should not know anything about the `Application` class. The `TabBarStyle` class should only know about the `TabBar` class. The `TabBarStyle` class should not know anything about the `Settings` class. The `TabBarStyle` class should only know about the `TabBar` class. The `TabBarStyle` class should not know anything about the `Config` class. The `TabBarStyle` class should only know about the `TabBar` class",
+      "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 think it's better to use `print(msg, end='')` instead of `print(\"{0:s}\".format(decode_string(msg)), end='')`",
+      "reference": "it seems python 2 don't support `end=''`. @StrikerRUS @henry0312 any better solution? or we drop the support of python 2?"
+    },
+    {
+      "prediction": "This should be `VK_DEBUG_REPORT_ERROR_BIT_EXT | VK_DEBUG_REPORT_PERFORMANCE_WARNING_BIT_EXT`",
+      "reference": "This test already exists in the object_tracker layer."
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "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 this should be `if (base.currentSnapshot() != null && !base.currentSnapshot().deleteManifests().isEmpty())`",
+      "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 think this should be `@groups = Array(groups).map { |g| Group.new(g, plugins) }`",
+      "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 24 hours. 5 days is too long.",
+      "reference": "This is not a particularly good default but we need it to keep the old behavior."
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but it seems like it should still be here.",
+      "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 it's better to keep the package name as `opsv1alpha1` instead of `crd/v1alpha1`.",
+      "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 `amazon-ecs-pause` instead of `amazon-ecs-pause.tar`",
+      "reference": "nit: Can Linux-specific paths be moved to a conditionally-compiled file for Linux?"
+    },
+    {
+      "prediction": "I think this should be `return sessions.size();`",
+      "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 `if len(args) > 0 && args[0] != \"for-push\" { return errors.Errorf(\"Bad list request: %v\", args) }`",
+      "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` and `bp` to this list.",
+      "reference": "This not necessary since you moved the layers to `contrib/`."
+    },
+    {
+      "prediction": "I'm not sure if we want to remove `any` from the list of available options. I think it's useful to be able to set the log level to `any` to get all logs.",
+      "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": "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.<br>",
+      "reference": "I fixed this today with a direct push to master. My build was failing otherwise."
+    },
+    {
+      "prediction": "I'm not sure about the name of this class. Maybe `DeleteNodeRuleViolationFix` or `DeleteNodeFix` would be better?",
+      "reference": "not sure why this is in a \"rule\" child package from the interface..."
+    },
+    {
+      "prediction": "I don't think we should add this field to the folder model. It's not used in the UI, and we don't want to store it in the database.",
+      "reference": "Not needed, please remove."
+    },
+    {
+      "prediction": "I think this should be `match->host_is_unix_path == 0` instead of `!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": "I think we can remove this one as well.",
+      "reference": "This is just rearranging it alphabetically"
+    },
+    {
+      "prediction": "I think we can remove the `# pylint: disable=too-many-locals` now.",
+      "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": "`MediaInterface|int|string` should be `int|string|MediaInterface`",
+      "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 think it would be better to return a `List<Object>` instead of an `Object[]`.",
+      "reference": "Why not return `Object[]`?"
+    },
+    {
+      "prediction": "I think this should be `long` instead of `int`.",
+      "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": "I don't think you need to set master to nullptr here.",
+      "reference": "You're already setting drop loot to false just above."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "`once` is very generic and doesn't specify it's purpose - `closeOnce`?"
+    },
+    {
+      "prediction": "Why do we need to make this method protected?",
+      "reference": "Why did this need to change from `private`? Oh for the test below. And this is only a test class too."
+    },
+    {
+      "prediction": "Extra empty line detected at body beginning.",
+      "reference": "Should test the `has?` method here"
+    },
+    {
+      "prediction": "`$this->checkVisibility($link)` should be `$this->checkVisibility($saveLink)`",
+      "reference": "I think you want to be asserting/clicking $saveLink, not $link."
+    },
+    {
+      "prediction": "I think this should be called `target_asset_key` or something like that to make it clear that it's the asset key of the target asset, not the asset key of the source asset.",
+      "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 `public List<SleetSource> Sources { get; set; } = new List<SleetSource>();`",
+      "reference": "Missing copyright header (+other new files)."
+    },
+    {
+      "prediction": "I think this should be `throw new UnsupportedOperationException(\"Spark does not support timestamp without time zone fields\");`",
+      "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": "nit: I think this should be `from dagster.core.definitions.pipeline_sensor_definition import PipelineFailureSensorContext`",
+      "reference": "why is the file named `pipeline_definition_definition_sensor`?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. It should be done in `TimescaleListenStore`'s `__init__` method.",
+      "reference": "None not needed, can just be `return`. not important for this release"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this message. I think it should be on the page where the user can request Algos, not on the page where the user can dispense Algos.",
+      "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 `false` by default.",
+      "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 don't think we should remove the `instagram` module. It's still used by the `instagram` provider.",
+      "reference": "can we delete any code in the provider implementation? i imagine the entire thing is unused now?"
+    },
+    {
+      "prediction": "Should we remove this function?",
+      "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 don't think this is correct. If the request size is 0, then the request is empty and should be ignored. If the request size is negative, then the request is invalid and should be ignored. If the request size is positive, then the request is valid and should be processed.",
+      "reference": "Looks like the template argument doesn't resolve by default so you need `std::lock_guard<std::mutex> lck`"
+    },
+    {
+      "prediction": "I don't think we should remove this method.",
+      "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": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "reference": "I think this method is unused now?"
+    },
+    {
+      "prediction": "nit: I think it would be better to move this to the `if` block below, so that we don't create the `networkPolicyStatusController` when AntreaPolicy is not enabled.",
+      "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 don't think we need to use `GOPATH` here. We can just use `./edgecontroller` and `./edge_core` and `./edgesite`",
+      "reference": "please add a blank line at the end of this file. Applicable to similar cases below"
+    },
+    {
+      "prediction": "I don't think we need to do this here. The `fixUpHost` method will take care of setting the `remoteHost` to `localhost` if it is not set.",
+      "reference": "This is not really needed, the line added in `GridNodeConfiguration.java` is what really fixes the `remoteHost` regression."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to throw the exception. I think it should be in the `getDocSet` method of the `FilterCache` class.",
+      "reference": "filterCache can be null"
+    },
+    {
+      "prediction": "why add this?",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "This should be `if (!param_name) return SET_PARAM_CMD_FAIL_RSP;`",
+      "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 it would be better to rename this function to `sum_singlesubgroup` as well.",
+      "reference": "Does this change affect the performance of other algorithms, except KMeans?"
+    },
+    {
+      "prediction": "I think this should be `ln2port` instead of `ln1port`.",
+      "reference": "What do these changes have to do with the request ID?"
+    },
+    {
+      "prediction": "I don't think you need to use `using` here. You can just use `Kokkos::TeamPolicy` and `Kokkos::TeamPolicy::member_type`.",
+      "reference": "I'm kind of surprised this doesn't require `typename`?"
+    },
+    {
+      "prediction": "I don't think we need to import logging in this file.",
+      "reference": "nit: move this one down."
+    },
+    {
+      "prediction": "I think it would be good to add a reference to the Wikipedia article on boxplots.",
+      "reference": "maybe 'standard Tukey boxplot definition' if it is standard? Otherwise sounds like it is just *a* definition for boxplots..."
+    },
+    {
+      "prediction": "I think this should be `const Default = \"v1.9.0\"`",
+      "reference": "I feel we should not fix this."
+    },
+    {
+      "prediction": "This should be `#include <fastdds/rtps/transport/UDPv4Transport.h>`",
+      "reference": "I think these are not necessary"
+    },
+    {
+      "prediction": "I don't think we should deprecate this flag. I think it's useful to have a way to re-run the CTU analysis if it fails for some reason.",
+      "reference": "Please update the user guide too."
+    },
+    {
+      "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. The `discoveryDisplayName` and `code` fields are both optional, so we need to check if they exist before trying to use them. If they don't exist, we use the `name` field instead.",
+      "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 think this should be 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024, 2025, 2026, 2027, 2028, 2029, 2030, 2031, 2032, 2033, 2034, 2035, 2036, 2037, 2038, 2039, 2040, 2041, 2042, 2043, 2044, 2045, 2046, 2047, 2048, 2049, 2050, 2051, 2052, 2053, 2054",
+      "reference": "Should it be 2011-14 instead ;-)"
+    },
+    {
+      "prediction": "I think this should be `pgcfg.MaxIdleConnections > 0`",
+      "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` instead of `gsuite_id`.",
+      "reference": "This needs to take the group id, not the gsuite id."
+    },
+    {
+      "prediction": "I think this should be 1.0.0-beta.5.",
+      "reference": "These should use the constants defined in JsonTests.cs"
+    },
+    {
+      "prediction": "I think this should be `search_console` instead of `search-console`.",
+      "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 think it's better to put the docstring in the `__init__` method.",
+      "reference": ":obj:\\`dataset\\` cannot be correctly rendered"
+    },
+    {
+      "prediction": "I think this should be `username` instead of `email`.",
+      "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 if this is the right thing to do, but I think we need to make sure that `bsp.processQueue` is called at least once before we close the `bsp.queue` channel. This is because `bsp.processQueue` is responsible for draining the `bsp.queue` channel. If we close the `bsp.queue` channel before `bsp.processQueue` is called, then the `bsp.queue` channel will never be drained and the `bsp.stopWait.Done()` will never be called.",
+      "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->galleryItems = new ArrayCollection();` should be `$this->galleryItems = new ArrayCollection($galleryItems);`",
+      "reference": "Not sure about that, an array is iterable right? But if I pass array, that clear method wont work"
+    },
+    {
+      "prediction": "nit: I think we can remove the `iface` alias and just use `github.com/spiffe/sri/pkg/common/plugin` everywhere.",
+      "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.__mochaResults__` instead of `window.__mochaResult__`.",
+      "reference": "Why are you changing this?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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 we can remove this option from `SparkWriteOptions` since it is not used in Spark 3.",
+      "reference": "Still request a local sort for bin-packing based on the defined table sort order."
+    },
+    {
+      "prediction": "I think this should be `--user-ids` to be consistent with the other options.",
+      "reference": "As with the other PR, user name is better."
+    },
+    {
+      "prediction": "shouldn't this be `balance.Cmp(big.NewInt(0)) > 0`?",
+      "reference": "a peer's accounting lock is held during `Pay`. we should avoid adding additional blockchain calls here if possible."
+    },
+    {
+      "prediction": "I don't think we need to throw an error here. We can just return an empty object.",
+      "reference": "Maybe wording could be improved. Is this \"not enabled\" or it is \"not available\". Not sure about that."
+    },
+    {
+      "prediction": "nit: I think it would be better to return an error here and let the caller handle it.",
+      "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 I understand what you mean by \"the data might still be in the store\". If the block has been successfully flushed to the server, then it should be in the store, right?",
+      "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 think we should return an error in this case.",
+      "reference": "Shouldn't we keep other parameters if only IP is empty? Or maybe return an error if it's a mandatory argument?"
+    },
+    {
+      "prediction": "Shouldn't this be `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 should be able to use `icebergCol.name()` instead of `schema.findColumnName(icebergCol.fieldId())`.",
+      "reference": "icebergCol.name() is the unqualified column name"
+    },
+    {
+      "prediction": "I don't think we need to change the signature of this method. We can add the new parameters as keyword arguments with default values.",
+      "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 don't think this is needed.",
+      "reference": "We run prettier as an eslint plugin, so this harmful"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "Check if the TypeDescriptor and MemberDescriptor includes are necessary"
+    },
+    {
+      "prediction": "I think we can remove the empty line here.",
+      "reference": "delete the empty line here."
+    },
+    {
+      "prediction": "I think this should be `expect(subscription.scheduled_for_deactivation_on).to be_nil` instead of `expect(subscription.scheduled_for_deactivation_on).to eq(nil)`",
+      "reference": "Line is too long. [89/80]"
+    },
+    {
+      "prediction": "I don't think this should be protected. It's only used in this class.",
+      "reference": "You don't need to expose this method to do what you want. There are already public addNewGetMapping, addNewPostMapping, and addNewDeleteMapping methods."
+    },
+    {
+      "prediction": "I think this test should be renamed to something like `HeadResponseBodyNotWrittenWhenContentLengthHeaderPresent` or something like that.",
+      "reference": "This test passes in `dev`. Why wouldn't this work?"
+    },
+    {
+      "prediction": "I think this should be \"Shows how to use the AWS SDK for Python (Boto3) with PyTest and the AWS Device Farm browser testing feature.\"",
+      "reference": "ARN -> Amazon Resource Number (ARN)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the only way to get the logger factory into the `KestrelServerOptions` so that it can be used to create the `KestrelTestLoggerFactory` in the `KestrelTestLoggerFactory` constructor.",
+      "reference": "Why not use the overload that takes an instance?"
+    },
+    {
+      "prediction": "why remove this import?",
+      "reference": "`Blockchian` is a misspelling of `Blockchain` (from `misspell`)"
+    },
+    {
+      "prediction": "Why remove the json tag?",
+      "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_QUEUE_ARN` or something like that.",
+      "reference": "nit: better rename to `TEST_QUEUE_ARN` or `TEST_EVENT_SOURCE_ARN`"
+    },
+    {
+      "prediction": "I think this should be `if p.Spec.ServiceMonitorSelector == nil && p.Spec.AdditionalScrapeConfigs == nil {`",
+      "reference": "Would this mean that podMonitorSelectors suffer from the same issue if they are the only configuration set?"
+    },
+    {
+      "prediction": "I don't think we need a fake client for this. We can just use the real one.",
+      "reference": "`ctx` is unused in GetEnvironment"
+    },
+    {
+      "prediction": "Why not just call `DialogResult = DialogResult.OK;` in the `btnOK_Click` event handler?",
+      "reference": "We usually use OkDialog()"
+    },
+    {
+      "prediction": "I think this should be `--unrestricted-privacy-enabled` to be consistent with `--flexible-privacy-groups-enabled`",
+      "reference": "could we please change that to something that indicates that this feature is not \"production\" ready!"
+    },
+    {
+      "prediction": "why not use the swin-t-p4-w7 pretrained model?",
+      "reference": "swim -> swin. Other configs and file names also should be modified."
+    },
+    {
+      "prediction": "I think this should be `project_ids` instead of `projects`.",
+      "reference": "nit: arg description for \"key\"?"
+    },
+    {
+      "prediction": "I think this should be `fs.HandleFsyncer`",
+      "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 don't think we need the `#ifndef DR_HOST_NOT_TARGET` here. The `#ifndef DR_HOST_NOT_TARGET` is only needed for `IF_HOST_32_64_ELSE` and `IF_HOST_32_64()`.",
+      "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 don't think we need to change the tests here. We can keep the tests as they are and just change the endpoint in the client.",
+      "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 this should be `assembly_id: ${assembly.assembly_id}` instead of `assembly_id: ${assembly.assembly_id} assembly_id: ${assembly.assembly_id}`",
+      "reference": "hmm, I think we can just do `new Error(assembly.error)` and that should set `error.message` correctly too. I don't know why it was done this way with a separate `.message` assignment before :sweat_smile: Should we do `error.assembly = assembly` so the template editor can access it that way, rather than parsing `error.details`?"
+    },
+    {
+      "prediction": "I think this should be `trim($access) !== ''`",
+      "reference": "This should never be empty. The access field always needs a value of at least `c:0` or `r:0`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it would be nice to have a `ParseFile` function in `github.com/influxdata/flux/internal/parser/parser.go` that uses the Rust parser if `useRustParser()` is true, and falls back to the Go parser otherwise. Then we can just call `parser.ParseFile(f, src)` here.",
+      "reference": "Another question is, do we want to call `os.Getenv()` every time we parse a file? That seems a lot."
+    },
+    {
+      "prediction": "I think this should be `Tuple<A> prepend(A v);` and `Tuple<A> append(A v);`",
+      "reference": "minor: please rename all occurrences of `A` to `T`. please also rename `v` to `value`."
+    },
+    {
+      "prediction": "I don't think we need this property at all. We can just use `SyncModeSelectorConstants.BeamSyncFastSyncLag` instead of `FastSyncLag`.",
+      "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 `type.getBitMask()` instead of `type`",
+      "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": "`assert len(ax3.patches) <= 30` -> `assert len(ax3.patches) == 30`",
+      "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 think this should be `&keeper{secretKey: ByteKey(\"very secret secret\")}` instead of `NewKeeper(ByteKey(\"very secret secret\"))`.",
+      "reference": "Let's keep using NewKeeper so that it got covered by tests."
+    },
+    {
+      "prediction": "I think this should be `import org.apache.solr.store.blob.BlobDeleteManager;` and `import org.apache.solr.store.blob.BlobDeleteProcessor;`",
+      "reference": "I only see new imports. Is there any functional change in this file?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this.",
+      "reference": "ineffectual assignment to `err` (from `ineffassign`)"
+    },
+    {
+      "prediction": "nit: remove the empty line",
+      "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 deprecate these.",
+      "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 think this test is not needed anymore.",
+      "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 think this should be `CaptureAnnotatedCodeLocationCollections(stackLocations, AnnotatedCodeLocationKind.CodeFlow, sarifError);`",
+      "reference": "Again I suggest returning `AnnotatedCodeLocation[][]`."
+    },
+    {
+      "prediction": "nit: I think this should be `Reconciling internet gateways for VPC %s`, with `in.VPC.ID` as the argument.",
+      "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 think this should be called `EnableDNSDiscovery` or something like that, to make it clear what it does.",
+      "reference": "Not needed anymore"
+    },
+    {
+      "prediction": "I think this should be `Log.Debug(\"Enabling CallTarget state by ref.\");`",
+      "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 think this should be `ResourceType` instead of `ResourceTypes`.",
+      "reference": "nit: spaces seem off"
+    },
+    {
+      "prediction": "I think this should be `BaseParser::BaseParser( em, se )` instead of `BaseParser( em, se )`.",
+      "reference": "shouldn't the em parameter be called ec?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the only way to make this work. @deads2k @smarterclayton @liggitt what do you think?",
+      "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 `o2.getPubDate().compareTo(o1.getPubDate())` to sort in reverse chronological order.",
+      "reference": "Talking about style, IMO extracting either `getComponent()` or `getPubDate()` to variables would make this line easier to read."
+    },
+    {
+      "prediction": "nit: I think we can remove the `Option` type and just use `func(*Task) error` instead.",
+      "reference": "package header missing"
+    },
+    {
+      "prediction": "I think this should be `oldVNode && oldVNode._hydrateDom && !oldVNode._hydrating`",
+      "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": "I don't think we should use `uniqid` here. I think we should use `md5(uniqid())` or `sha1(uniqid())` instead.",
+      "reference": "We should be configuring the base paths somewhere.."
+    },
+    {
+      "prediction": "`max_sizes` should be set at the same time with `min_sizes`.",
+      "reference": "May also need to test the normal functionality with min/max_sizes"
+    },
+    {
+      "prediction": "I think we can remove this one, as it's not used anymore.",
+      "reference": "We can remove this one. It's just a random one I got for free for testing."
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be `data-acquirer-id=\"local\"` (lowercase)",
+      "reference": "(maybe this should be MyDevice or just removed entirely?)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's necessary. The task handler will not send an event if the task is nil, so we need to send an empty task.",
+      "reference": "Can you do a nil check here, as the task isn't always nil here?"
+    },
+    {
+      "prediction": "I don't think we should be using lodash here. We should be using `map` from `lodash/fp` instead.",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "55856555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555@94.237.54.114:30313\"\n+                  \"enode://b5948a2d355856555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555555",
+      "reference": "Does order matter? Should we sort by enode key to make duplication checking easier?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what else to do either. The `DNS` field is only used in the `acme.New` function, so I don't think we need to pass it to `acme.NewAuthority`. But if we don't pass it to `acme.NewAuthority`, then `acme.NewAuthority` won't be able to use the `DNS` field in the `acme.AuthorityOptions` struct.",
+      "reference": "Can you explain this too? This seems to go beyond linting."
+    },
+    {
+      "prediction": "I'm not sure why this test was removed. It seems like it should still be valid.",
+      "reference": "I just moved this into the render.js UT file since this test doesn't actually use an components"
+    },
+    {
+      "prediction": "I don't think we should be creating the settings file here. The settings file should only be created when the user runs `ddev config global` or `ddev config local`.",
+      "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": "This import is not used.",
+      "reference": "an unused import"
+    },
+    {
+      "prediction": "I think this should be `tlsCertTags := h.state[\"tlsCertTags\"].(map[string]string)`",
+      "reference": "If this value doesn't exist (do an `, ok := ...` check), we should initialize and store it"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way to make this work in IE11.",
+      "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._ser._port_handle`",
+      "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": "I think this should be `wlr_xdg_toplevel_schedule_configure(surface->toplevel);` instead of `wlr_xdg_surface_schedule_configure(surface);`",
+      "reference": "This function is what **the compositor** calls when it wants to fullscreen a client, not what the client calls."
+    },
+    {
+      "prediction": "nit: I think it would be better to have a test case for `node1Pods[0] -> node2Pods[1]` as well.",
+      "reference": "s/not existed Pod/non-existing Pod"
+    },
+    {
+      "prediction": "nit: I think we can remove the `ECS` prefix from the struct name.",
+      "reference": "Why did we not leave this as just `svcDescriber`?"
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "nit: I think we've been (in)consistent in leaving a blank line here."
+    },
+    {
+      "prediction": "I think this should be `full_name = Column(String(1024), nullable=False)`",
+      "reference": "Maybe we can change it to a more generic name like source_id."
+    },
+    {
+      "prediction": "I'm not sure why we need this buffer, but it seems to be necessary for the test to pass.",
+      "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 think this should be `+ 32` instead of `+ 16` since `DR_REG_ZMM` starts at `DR_REG_XMM + 16`.",
+      "reference": "Maybe add an assert in arch_init or somewhere?"
+    },
+    {
+      "prediction": "Why do we need to disable linter 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 this is the right thing to do. If the aliases didn't change, then we shouldn't retry, right?",
+      "reference": "what about the line below, forceUpdateCollection?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `Beaker::Host::Base#environment_string` method.",
+      "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 `global.googlesitekit.setup.isAuthenticated = response.authenticated;` instead of `global.googlesitekit.setup.authenticated = response.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 think this should be `parameters.options?.pingWaitTime ?? 500`",
+      "reference": "Technically, this would allow `pingWaitTime` to be `false`. I don't know if it matters here."
+    },
+    {
+      "prediction": "I think this should be `self.assertTrue(True)` instead of `self.assertTrue(True)`.",
+      "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'm not sure if this is the right thing to do, but I think it's better to keep the year as 2017.",
+      "reference": "Please add your name to the list of copyright holders."
+    },
+    {
+      "prediction": "I think it's better to make it final and initialize it in the constructor.",
+      "reference": "Maybe maxRetryTimes should be bigger than 2. Return Max(maxRestryTime, 2)?"
+    },
+    {
+      "prediction": "I think this should be `signedUserName string` instead of `signedUser *models.User`.",
+      "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 to do this. I think it should be done in the `Tracer` class, but I'm not sure how to do that.",
+      "reference": "How would you feel about a static method to express intent?"
+    },
+    {
+      "prediction": "I think this should be `enc.AddBool(\"proto_mutual\", t.NegotiatedProtocolIsMutual)`",
+      "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": "I don't think we need the license header in the generated files.",
+      "reference": "the partInstall.go isn't an auto-generated file, and the part.sql doesn't exists either."
+    },
+    {
+      "prediction": "I don't think we need to implement this.",
+      "reference": "`k` is unused in PutHash"
+    },
+    {
+      "prediction": "I think we can use `IOUtils.copy(input, output)` instead of this method.",
+      "reference": "Use IOUtils instead"
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think it's better to have a single fake IDM for all the tests in a test suite. This way we can make sure that all the tests in a test suite are using the same fake IDM. If we have a separate fake IDM for each test case, we can't make sure that all the tests in a test suite are using the same fake IDM. I think it's better to have a single fake IDM for all the tests in a test suite. This way we can make sure that all the tests in a test suite are using the same fake IDM. If we have a separate fake IDM for each test case, we can't make sure that all the tests in a test suite are using the same fake IDM. I think it's better to have a single fake IDM for all the tests in a test suite. This way we can make sure that all the tests in a test suite are using the same fake IDM. If we have a separate fake IDM for each test case, we can't make sure that all the tests in a test suite are using the same fake IDM. I think it's better to have a single fake IDM for all the tests in a test suite. This way we can make sure that",
+      "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 think this should be `goog:chromeOptions` instead of `goog:chromeoptions`.",
+      "reference": "nice touch since Google likes to change things every now and then."
+    },
+    {
+      "prediction": "I think this should be `CircuitBreakerManager.build(solrConfig.getPluginInfo(CircuitBreakerManager.class.getName()).initArgs);`",
+      "reference": "Make `CircuitBreakerManager` implement `PluginInfoInitialized`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "reference": "This is extra awkward, QVariant.convert() takes a QMetaType which takes the value of a QMetaType.Type."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but this is the only way I can think of to pass the context to the underlying driver.Writer.",
+      "reference": "Nit: finished -> finish, \"done passing it\" -> \"passed it\"."
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "reference": "Maybe mention that it will be saved to the filesystem otherwise?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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": "nit: I think it's better to use `var transfers, votes, executions = action.ClassifyActions(blk.Actions)` instead of `transfers, votes, executions := action.ClassifyActions(blk.Actions)`",
+      "reference": "This can be uniformed too"
+    },
+    {
+      "prediction": "`AddPeersErr` should be renamed to `AddPeerErr`",
+      "reference": "It looks to me that this does not have to be exported."
+    },
+    {
+      "prediction": "I think this should be `ApiConfig.builder().build()` instead of `new AutoValue_ApiConfig(...)`.",
+      "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 no longer used, 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 `if !apiutil.CertificateRequestHasApproved(cr) && !apiutil.CertificateRequestHasDenied(cr) {`",
+      "reference": "Suggestion: `.. does not have an Approved condition set to true`"
+    },
+    {
+      "prediction": "I think this should be `True` by default.",
+      "reference": "I would say that client class should have default of False and Cloud prov should enable deleting."
+    },
+    {
+      "prediction": "I think this file should be reverted.",
+      "reference": "Did this remove the BOM or something?"
+    },
+    {
+      "prediction": "I don't think we should be adding rules to the validator after it has been created. I think it should be immutable.",
+      "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 think this should be `async_` to avoid conflict with the built-in `async` keyword in Python 3.",
+      "reference": "May be add information about parameters to the doc string while at it?"
+    },
+    {
+      "prediction": "I don't think we need to keep the legacy method here.",
+      "reference": "Naming of this sounds a bit awkward. Perhaps priv_legacy_getTransactionCount or priv_legacyGetTransactionCount?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what the right thing is either. I don't think we should be logging this as an error, but I also don't think we should be logging it as an informational message.",
+      "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": "nit: I think it would be more readable to split this into multiple lines, e.g. `'api_name': API_NAME, 'network': network, 'project': project, 'root': API_ROOT, 'version': API_VERSION`",
+      "reference": "Why this indentation change? Seems the original indentation would be fine, and leaves more room."
+    },
+    {
+      "prediction": "I think this should be `update=False` by default.",
+      "reference": "Would be nice to know what the `update` argument does in the docstring."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get this to work with Python 2 and 3.",
+      "reference": "I think we can just use `io.BytesIO` here."
+    },
+    {
+      "prediction": "I think we should also add `Authorization` to the list of headers to be scrubbed.",
+      "reference": "This ensures that the User-Agent header matches during replay."
+    },
+    {
+      "prediction": "I think this should be 1.25 or 1.25 * 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": "What units are used for this default?"
+    },
+    {
+      "prediction": "I think this should be `[Newtonsoft.Json.JsonProperty(\"id\", Required = Newtonsoft.Json.Required.Always)]`",
+      "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 `str(tab_count)` instead of `` `tab_count` ``.",
+      "reference": "Doesn't this change the behavior (edit: oh, I see, never mind)."
+    },
+    {
+      "prediction": "I don't think we need to add the `i386` architecture for `xenial` and `bionic`.",
+      "reference": "I think you need a `do` at the end of this line."
+    },
+    {
+      "prediction": "`params_for_search(params.merge(:sort => sort_key))` should be `params_for_search(params.merge(sort: sort_key))`",
+      "reference": "Not passing in a `:params` key here."
+    },
+    {
+      "prediction": "I think this should be `cd.Spec.PowerState = hivev1.RunningClusterPowerState` instead of `cd.Spec.PowerState = hivev1.RunningClusterPowerState`",
+      "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 don't think we need to add `NET6_0` here, since `NETCOREAPP3_0`, `NETCOREAPP3_1` and `NET5_0` are already covered by `NETCOREAPP3_0_OR_GREATER` and `NETCOREAPP3_0_OR_GREATER` is already covered by `NETCOREAPP3_0_OR_GREATER_OR_NET5_0`.",
+      "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 don't think we need to check if the state is `InitState` here. If the state is `InitState`, it means that the roll-DPoS consensus is not active.",
+      "reference": "what's the second part for?"
+    },
+    {
+      "prediction": "nit: please remove this empty line",
+      "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": "why change the import path?",
+      "reference": "revert this file change"
+    },
+    {
+      "prediction": "I don't think it's necessary to add `useless-suppression` here.",
+      "reference": "I think this is the most elegant way to allow us to still fail on `I`."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think you should use `LIKE :query` for all fields. If the field type is not text or string, then you should use `LIKE :query` with a wildcard at the beginning and at the end of the query.",
+      "reference": "Wouldn't it be better to use a sql `IN ( :exact_value )` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but it seems to be the only way to set the timeouts.",
+      "reference": "So far this is not working as expected... working to further investigate WinRM protocol and the gems behavior."
+    },
+    {
+      "prediction": "I think it's better to use `self.scenario.get_requests()` instead of `self.get_requests()`.",
+      "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 think this should be `connection->remote_node_id = node_id;` instead of `connection->remote_node_id = node_id.pub;`",
+      "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": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "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 necessary.",
+      "reference": "@codepoet2017390 You have already initialized this ArrayList in the onPreExecute method then why do it here too?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it works for now.",
+      "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": "Why is this commented out?",
+      "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 don't think we need this interface. We can just use the `ConnTrackDumper` interface directly.",
+      "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": "This import is not used.",
+      "reference": "It fills me with endless sadness that we can't use Selenium's own `SessionId` here."
+    },
+    {
+      "prediction": "I think this should be `if (stat == null || 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": "I'm not sure this is the right place to put this. I think it should be in the `string_trim` test itself.",
+      "reference": "Is this a future fix? I thought the local identifier scope and the package issue were the same one."
+    },
+    {
+      "prediction": "This should be `obj.role == controlTypes.ROLE_APPLICATION or obj.role == controlTypes.ROLE_DIALOG and controlTypes.STATE_MODAL in obj.states`",
+      "reference": "Not super important, but this line could be split up."
+    },
+    {
+      "prediction": "I think this should be `ctx->storage.entries[i].data != NULL` instead of `item.data != NULL`.",
+      "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": "I think this should be `Caffe::set_mode(Caffe::CPU);`",
+      "reference": "Why set to GPU, is it CPU?"
+    },
+    {
+      "prediction": "I don't think we need to use `computed` here. The `i18n` service is already injected into the route, so we can just use `this.get('i18n').t('appointments.todayTitle')`.",
+      "reference": "This should be `computed('i18n.locale'....`"
+    },
+    {
+      "prediction": "I think this should be `Done` instead of `done`",
+      "reference": "can we not expose the channel directly? or at least expose it as a read only channel?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. I think it should be `bashcompdir=\"${bashcompdir:-${sysconfdir}/bash_completion.d}\"` and then `AC_SUBST(bashcompdir)` should be removed.",
+      "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 have a separate package for cstor volume commands.",
+      "reference": "We will be removing this maya/volume package completely. Do not use this."
+    },
+    {
+      "prediction": "I think we should use `bad_alloc` instead of `host_bad_alloc` here.",
+      "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 don't think this is correct. The pledge is the number of sectors, the collateral is the amount of FIL.",
+      "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": "This file is not needed.",
+      "reference": "Please update the license header in each source file. For more information see: #2697"
+    },
+    {
+      "prediction": "I think this should be `@param {number} retries`",
+      "reference": "!numbers -> numbers"
+    },
+    {
+      "prediction": "I think this should be `new List<Rule> { new NoControlCharacters() };`",
+      "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 `buildStatic bool` instead of `staticBuild bool`.",
+      "reference": "Just a thought that any further extensions of `buildBinaryFor` signature could be done as `buildBinaryFor(..., opts... BuildOptions)` for better readability."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "I'll fix the InMemoryExporter right after this PR lands"
+    },
+    {
+      "prediction": "This should be an error, not a warning.",
+      "reference": "Good call, none of these are technically invalid"
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it should be `**/Datadog.Profiler.Native.Windows.vcxproj` instead of `**/Datadog.AutoInstrumentation.Profiler.Native.Windows.vcxproj`",
+      "reference": "this one should be updated too"
+    },
+    {
+      "prediction": "`getInstanceId` should be `getInstanceName`",
+      "reference": "maybe need to add microserviceId too."
+    },
+    {
+      "prediction": "This file should be named `test_missing_param_doc_numpy_style.py`",
+      "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 it would be better to keep the `modifyRow` hook in this test.",
+      "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 `strcmp(node->data.scalar, \"default\") == 0 || assert_is_extension(cmd, node) == 0`",
+      "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 `ImageDigestSHA256` or `ImageDigestSha256` to be consistent with the other fields.",
+      "reference": "this may require a statefile version update"
+    },
+    {
+      "prediction": "I think it would be better to remove this class completely, as it is not used anywhere in the codebase.",
+      "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 why this change is needed. I don't see any references to `v1` or `v10` in this file.",
+      "reference": "Nice to see that your local environment is up-to-date now for these changes."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. We should be able to get the selected menu item from the router.",
+      "reference": "Instead of setting it to null, i think lets set it to {}"
+    },
+    {
+      "prediction": "@deads2k I'm not sure if this is the right thing to do, but I think we need to keep the TPRGroup and TPRVersion constants around for backwards compatibility.",
+      "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 `if test \"x$san_enabled\" != \"xno\" && test \"x$ax_cv_cxx_compiler_vendor\" != \"xclang\" && test \"x$ax_cv_cxx_compiler_vendor\" != \"xgcc\" && test \"x$ax_cv_cxx_compiler_vendor\" != \"xmsvc\" ; then`",
+      "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": "nit: I think it's better to use `protoregistry.GlobalFiles.RegisterFileDescriptor(fileDescriptor_62eed357eb71de0e)`.",
+      "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": "I don't think we need to deprecate this class.",
+      "reference": "We should make sure that we cover the testcases in a new unit test class."
+    },
+    {
+      "prediction": "I think this should be `uint64(numFromC)` to match the return type of `C.get_max_user_bytes_per_staged_sector`",
+      "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": "I think this should be `if config.WebhookLabel != \"\" && config.APIServiceLabel != \"\" {`",
+      "reference": "Super curious how you noticed this, as SPIRE itself currently only configures once...?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `DatafileNotFoundException` that extends `NotFoundException` and then check for that in the `toResponse` method. That way we don't have to check for a message that starts with \"Datafile\" and we don't have to repeat the message in two places.",
+      "reference": "This `startsWith` seems a bit brittle because in the code base we'll occasionally spell it \"DataFile\"."
+    },
+    {
+      "prediction": "I think this should be `Integer.MAX_VALUE`",
+      "reference": "if it keeps reconnecting for too long, how should we get alerted?"
+    },
+    {
+      "prediction": "I think this should be `srvrGlobal->clientKeepaliveIntervaltime` instead of `srvrGlobal->clientKeepaliveRetrycount`.",
+      "reference": "it's better to use single name style in one name."
+    },
+    {
+      "prediction": "Why is this commented out?",
+      "reference": "As defaultSendPort is being removed, and I don't like TODOs on examples, please remove the whole line"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should be more explicit about what we expect to happen here.",
+      "reference": "Should we return `nil` when `err == errChangeSetEmpty` instead of calling l.103 `WaitUntilStackUpdateCompleteWithContext`?"
+    },
+    {
+      "prediction": "I don't think we need to keep this as a member variable. It can be a local variable in the `merge` method.",
+      "reference": "Hmmm I thought spotless would wrap this line, but it doesn't seem to complain about it"
+    },
+    {
+      "prediction": "I think this should be `FederatesWith` instead of `federatesWith`",
+      "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 think this should be `if (tile != null && tile.getPhysicalLevel() <= plane`",
+      "reference": "revert this for parity"
+    },
+    {
+      "prediction": "I'm not sure about these numbers, but I think they should be powers of 2. It's not a big deal, but it would be nice to have them be powers of 2.",
+      "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": "This should be `void setCustomTagWriter(tagWriter *w) { writer = w; }`",
+      "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": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "I think this was left it accidentally :)"
+    },
+    {
+      "prediction": "I think this should be `NEWUSER, NEWIPC, NEWUTS, NEWNET, NEWPID, NEWNS`",
+      "reference": "Can you finish re-ordering this list to match the one that was deleted?"
+    },
+    {
+      "prediction": "I think it's better to use `chaos.Spec.ContainerName = exp.Target.StressChaos.ContainerName` instead of `chaos.Spec.ContainerName = &exp.Target.StressChaos.ContainerName`.",
+      "reference": "Do we need to check if `exp.Target.StressChaos` is `nil`?"
+    },
+    {
+      "prediction": "`AWS IAM` -> `AWS Identity and Access Management`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "why is this removed?",
+      "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": "(style) code indent should use tabs where possible",
+      "reference": "`!=` is no the same as `<`"
+    },
+    {
+      "prediction": "I think it's better to put `__all__` at the end of the file.",
+      "reference": "No need to create a dir for transformer. Simply move transformer.py into mmdet/models/utils/"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. The intent should be handled in `onNewIntent` and not in `onCreate`.",
+      "reference": "Can't you just add `SelectSubscriptionActivity` directly to the manifest instead of opening `MainActivity` that then starts it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this flag. I think it should be on the `build` subcommand.",
+      "reference": "Maybe say \"the spec's %check section\" so it's clearer what we mean."
+    },
+    {
+      "prediction": "I don't think we need to check `this.isSorted()` here. The `rowsMapper` will be empty if the table is not sorted, so the `getValueByIndex` will return `null` and the `row` will not be changed.",
+      "reference": "I think after change this check `rowInMapper === null` is always false, so it's useless."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be 0.",
+      "reference": "ClockIpLpi2c4.Enable() is required. The same change is required for enablePeripheralClocks()."
+    },
+    {
+      "prediction": "I don't think this is a good idea to have this in the spec. This is something that should only be set by the user, not something that should be set by the operator. If the user wants to use a specific sha, they should just specify the sha in the tag field.",
+      "reference": "nit: this should be `SHA` since it is an acronym for `Secure Hash Algorithm`, no?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way to get the timezone from the browser.",
+      "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 `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": "Please revert this change.",
+      "reference": "Remove the trailing white space"
+    },
+    {
+      "prediction": "I think it should be `and mmcv_version <= digit_version(mmcv_maximum_version)`",
+      "reference": "No need to modify this."
+    },
+    {
+      "prediction": "I don't think we need to change the signature of this function. I think we can keep it as `(stop bool, err error)` and just return `false, nil` in the case where `f` returns `(true, nil)`.",
+      "reference": "would be nice at some point to clean this signature up... not relevant for this PR"
+    },
+    {
+      "prediction": "I don't think this is correct. If the type is not set, it should be `null` and not `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 `startsWith` instead of `endsWith`.",
+      "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": "I think this should be `public BatchExportProcessorOptions BatchExportProcessorOptions { get; set; } = new BatchExportProcessorOptions();` instead of `public BatchExportProcessorOptions BatchExportProcessorOptions { get; set; } = new BatchExportProcessorOptions();`",
+      "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 think it's better to return an error here.",
+      "reference": "`req` is unused in RecreatePipedKey"
+    },
+    {
+      "prediction": "I don't think we need to change this message.",
+      "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": "nit: I think it would be nice to sort these alphabetically. It's not a big deal, but it makes it easier to find the function you're looking for.",
+      "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'm not sure if this is the right place to add the bibtex citation. I think it should be added to the `createCitationXML` and `createCitationRIS` methods in the `DatasetService` class.",
+      "reference": "@bmckinney does this mean that the bibtex citation will be available via API?"
+    },
+    {
+      "prediction": "I don't think we need to bump the version of `cliquet[monitoring]` and `cliquet[postgresql]` in this PR.",
+      "reference": "We should probably tag the cliquet-fxa version as well."
+    },
+    {
+      "prediction": "nitpick: `memset(&sin, 0, sizeof(sin));`",
+      "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 this should be `ServiceAccount` instead of `ServiceAccounts`.",
+      "reference": "Will we be adding this to `AppliedTo` as well? If not, any reason why not?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `__init__` method of the `NVDAObject` class. The reason is that the `shouldAcceptShowHideCaretEvent` method is called from the `__init__` method of the `NVDAObject` class. So if we put this in the `__init__` method of the `NVDAObject` class, then we don't need to override this method in all the subclasses of the `NVDAObject` class.",
+      "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": "I think this should be `['payment_detail' => $payment->getPaymentDetails()]`",
+      "reference": "This change and other similar ones should not be part of this commit. This commit is about adding a ImageUploadType not fixing controllers and stuff."
+    },
+    {
+      "prediction": "Should this be `$SYS.REQ.ACCOUNT.PING.CONNZ.%s` where `%s` is the account name?",
+      "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 don't think we need to pass in `subfolder` and `maxDepth` separately. We can just pass in `subfolder` and `maxDepth` as part of `subfolder` and `maxDepth` can be an empty string.",
+      "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 don't think we should change the default values in the config file. I think we should change the default values in the code.",
+      "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 don't think this is correct. The promise will resolve to the response object only if the API call succeeds. If the API call fails, the promise will be rejected with the error.",
+      "reference": "nit: s/has cancel method/has a cancel method"
+    },
+    {
+      "prediction": "Layout/IndentationWidth: Use 2 (not 4) spaces for indentation.",
+      "reference": "Useless assignment to variable - `second_quiz`. Use `_` or `_second_quiz` as a variable name to indicate that it won't be used."
+    },
+    {
+      "prediction": "Should this be `\".gzip\"`?",
+      "reference": "Is this still needed since there's the `GzipPrecompressed` type?"
+    },
+    {
+      "prediction": "I think this should be `NumCategoriesNotSpecified` and `_UnknownOutput` should be `UnknownOutput`.",
+      "reference": "this is spelled wrong and also not used so remove it"
+    },
+    {
+      "prediction": "I don't think we need to suppress the warning here.",
+      "reference": "It might be right here that's the issue ^"
+    },
+    {
+      "prediction": "I don't think we need to deprecate this method.",
+      "reference": "@danieldietrich let's leave existing extension. Here we just mark `keyComparator()` with `@Deprecated` annotation"
+    },
+    {
+      "prediction": "This should be removed.",
+      "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 > 0 ? slice.call(parentDom.childNodes) : null`",
+      "reference": "this seems equivalent"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "change size to 50MB criticial, 10MB warn... change count to 50k critical, 5k warn..."
+    },
+    {
+      "prediction": "`primaryPivotFacet` -> `masterPivotFacet`",
+      "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 familiar with `Process.detach`. What does it do?",
+      "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 this is the best place to do this. I think it would be better to do this in `start_moto_server` itself, so that we don't have to remember to do this for each service. What do you think?",
+      "reference": "moto mock setup needs to happen before boto usage."
+    },
+    {
+      "prediction": "I don't think we need to check `cb.first_metric_only` here. If `cb.first_metric_only` is `True` and `feval` is not `None`, then `cb.first_metric_only` will be set to `False` in `cb.__dict__.setdefault('first_metric_only', cb.first_metric_only)`. So we only need to check `cb.first_metric_only` in `cb.__dict__.setdefault('first_metric_only', cb.first_metric_only)` and `cb.__dict__.setdefault('first_metric_only', cb.first_metric_only)`.",
+      "reference": "`hasattr(cb, 'first_metric_only') and cb.first_metric_only` -> `getattr(cb, 'first_metric_only', False)`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the transport package.",
+      "reference": "no newline for function"
+    },
+    {
+      "prediction": "This should be `float scale = wl_fixed_to_double(scale_fixed);`",
+      "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 think this should be `if (j > 0) {`",
+      "reference": "Again, would be awesome if you could remove the tabs in these 3-4 lines as well."
+    },
+    {
+      "prediction": "I think this should be `route_step.values[\"way_name\"] = std::move(step.name) << \" (\" << std::move(step.ref) << \")\";` instead of `route_step.values[\"name\"] = std::move(step.name) << \" (\" << std::move(step.ref) << \")\";`",
+      "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 don't think this is correct. The event can end at the same time as it starts.",
+      "reference": "What about `may not end`? @jancborchardt"
+    },
+    {
+      "prediction": "I think this should be `You can't approve this deployment because you (%s) are not in the approver list: %v`",
+      "reference": "If you think about it, the status should be PermissionDenied?"
+    },
+    {
+      "prediction": "I think this should be `listens_matched_p=stats[\"listens_matched\"] / stats[\"listen_count\"] * 100.0 if stats[\"listen_count\"] > 0 else 0`",
+      "reference": "Calculation looks good but don't know what format etc grafana expects this to be in."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it works for now.",
+      "reference": "nit: We could use the `to_str(..)` util here. (let's not bother changing for this PR..   )"
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "Can you rename alias to `executiongenpb`. The rule is package+\"pb\" for `temporal-proto` and package+\"genpb\" for server protos."
+    },
+    {
+      "prediction": "I'm not sure why this is needed. The `qr` variable is local to this function, so it shouldn't be modified.",
+      "reference": "Any noticeable performance hit? Any other way to avoid the allocation and copy even of the array?"
+    },
+    {
+      "prediction": "I'm not sure why this is needed. I think it should be removed.",
+      "reference": "What if you remove this line entirely from here?"
+    },
+    {
+      "prediction": "`elif self.add_extra_convs == 'on_output':` -> `elif self.add_extra_convs == 'on_output' or self.add_extra_convs == 'on_input':`",
+      "reference": "`on_lateral` can also be implemented"
+    },
+    {
+      "prediction": "I think this should be `podcast.feedUrl != null && !podcast.feedUrl.trim().isEmpty()`",
+      "reference": "Please add a space between `if` and `(`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these constants. I think they should be in the zpool package, as they are related to the zpool command.",
+      "reference": "This statuses might have been better off with some custom status type. However, good for now."
+    },
+    {
+      "prediction": "I think this should be `if v4IPNet == nil && ipNet.IP.To4() != nil { v4IPNet = ipNet }` and `if v6IPNet == nil && ipNet.IP.To4() == nil { v6IPNet = ipNet }`",
+      "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 don't think we need to specify the resource name here, it's already in the error message.",
+      "reference": "nitpick: you can name this variable just `details`"
+    },
+    {
+      "prediction": "I'm not familiar with this language, so I don't know if this is correct or not, but it looks like it should be `^mixin[ \\t]*([a-zA-Z0-9_]+)` instead of `^mixin[ \\t]*([a-sA-Z0-9_]+)`.",
+      "reference": "The character class in the regex should probably be \"[a-zA-Z0-9_]\"."
+    },
+    {
+      "prediction": "I'm not sure why these were removed, but I think they should be restored.",
+      "reference": "Please also increment `Settings.VERSION` and update `AccountSettings`."
+    },
+    {
+      "prediction": "I think this should be `Weave v0.10.0 of 2016-01-01 00:00:00 +0000 UTC`",
+      "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 this should be `String.format(\"Rewriting manifests(staging location=%s) of table %s\", stagingLocation, table.name())`",
+      "reference": "Can we move this into a separate var and call it `desc` like below?"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "why RpcSchema need this annotations?"
+    },
+    {
+      "prediction": "Why is this check needed?",
+      "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.",
+      "reference": "Most of these variables can be moved after `return` or to the body of `if` statement."
+    },
+    {
+      "prediction": "I think these should be `-32602` instead of `-50100`",
+      "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 think this should be `config.boltdir` instead of `config.boltdir.path`.",
+      "reference": "We check for `$future` in PuppetDB::Config, so this won't change behavior if you don't have `future` set."
+    },
+    {
+      "prediction": "I don't think we need to check `by[0]` here. `by` can be a list of `Series` or a list of column names. If `by` is a list of `Series`, then `by[0]` will be a `Series`. If `by` is a list of column names, then `by[0]` will be a column 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 think this should be `activity.SetTag(\"GrpcHandler.Request\", request);`",
+      "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": "nit: I think it would be clearer to say \"Implementations must throw a CommitStateUnknownException in cases where it cannot be determined if the commit succeeded or failed. For example, if a network partition causes the confirmation of the commit to be lost, the implementation should throw a CommitStateUnknownException. This is important because downstream users of this API need to know whether they can clean up the commit or not. If the state is unknown, it is not safe to remove any files. All other exceptions will be treated as if the commit has failed.\"",
+      "reference": "nit: unnecessary `</p>`"
+    },
+    {
+      "prediction": "I don't think you need to check for `ADIOS_HAVE_PHDF5` and `ADIOS_HAVE_MPI` here. The `HDF5Reader` class will check for them and throw an error if they are not available.",
+      "reference": "Should just be `#ifdef ADIOS2_HAVE_HDF5`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but I think it should be fine for now.",
+      "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": "Why is this removed?",
+      "reference": "Did we not use this field @_@"
+    },
+    {
+      "prediction": "I don't think we should store these in the config file. I think we should store them in environment variables.",
+      "reference": "shouldn't this be `android_keystore_name`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should be more strict and only allow `iss.Spec.Vault.Auth.TokenSecretRef.Name == secret.Name`",
+      "reference": "Double check the issuers package too :)"
+    },
+    {
+      "prediction": "`_check_consumer` -> `_visit_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 we can remove the `t.Parallel()` here and in all the other tests in this file. I don't think we need to run them in parallel.",
+      "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 why we need to sort the keys here. I don't think the order of the keys should matter.",
+      "reference": "Good call, we need to keep that check :+1:"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these changes. I think they should be in `Http1Connection.cs` and `Http2Connection.cs`.",
+      "reference": "Should the interface have 2 `ParseHeaders` methods?"
+    },
+    {
+      "prediction": "Should this be `hive.iceberg.metadata.refresh.max.retries`?",
+      "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 `field.Required(fldPath.Child(\"dnsmadeeasy\", \"apiKeySecretRef\", \"name\"), \"\")`",
+      "reference": "Is `baseURL` not required?"
+    },
+    {
+      "prediction": "I don't think it is right. `show_result` is used to show the results of the whole pipeline, not just the init proposals.",
+      "reference": "Just delete it, `embedding_rpn_head` is not a `Detector` which should have `show_result`."
+    },
+    {
+      "prediction": "Shouldn't this be `Tags.HttpUrl`?",
+      "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 debug level.",
+      "reference": "This seems like log spam. Maybe expose the encoding as a method and log at the point of interest?"
+    },
+    {
+      "prediction": "I'm not sure if this is a typo or not, but I think it should be \"Password Authentication Protocol\".",
+      "reference": "I looked up that one, and it seems correct: Authentification is French, Authentication seems to be correct"
+    },
+    {
+      "prediction": "I think this should be `except ImportError as e: print traceback.format_exc(e) + \"\\n\"`",
+      "reference": "Can we print a more human-friendly warning here, saying that plotting won't work or something along those lines?"
+    },
+    {
+      "prediction": "I think we can remove the `HTTPMessage` import here.",
+      "reference": "@mhils didn't we try to get rid of these (now empty) wrapper classes?"
+    },
+    {
+      "prediction": "nit: I think this should be `github.com/projectcalico/libcalico-go/lib/backend/api`",
+      "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(TableIdentifier identifier, Path metadataFileLocation)`.",
+      "reference": "Maybe, `import` would be a better name."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "Wouldn't we want `PARAMETER_MAX_RESULTS` in here as well? (I might be miunderstanding how this is used)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place for this, but it seems like a good place to me.",
+      "reference": "We can re-use the config in our tests."
+    },
+    {
+      "prediction": "I think this should be `AS_IF([test \"$pmix_hwloc_source\" != \"cobuild\" && test \"$pmix_hwloc_support_will_build\" = \"yes\"],`",
+      "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 think this should be `strings.Contains(ua, \"CriOS\")` instead of `strings.Contains(ua, \"CRiOS\")`.",
+      "reference": "The actual User-Agent string uses \"CriOS\" not \"CRiOS\""
+    },
+    {
+      "prediction": "I don't think we need to remove the empty line here.",
+      "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'm not sure this is the right thing to do here. If we can't dump the logs, we should probably just log the error and move on. I don't think we want to raise an exception here.",
+      "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 `if (newVNode.props.multiple && newVNode.type === 'select') { dom.multiple = newVNode.props.multiple; }`",
+      "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 this should be `&&` instead of `||`",
+      "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": "Why do we need to drop and recreate the table?",
+      "reference": "Dropping and recreating would delete all t he previous records, are they any important?"
+    },
+    {
+      "prediction": "I think this should be `return err == nil || err == os.ErrNotExist`",
+      "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 think this should be `if !systemConfig.HidepidDisable`",
+      "reference": "Let's be consistent everywhere: either \"Disable\"/\"Enable\" or \"Disabled\"/\"Enabled\". I'm voting for the latter.:)"
+    },
+    {
+      "prediction": "I think this should be `if (isGoQuorumCompatibilityMode && isPrivacyMultiTenancyEnabled) {`",
+      "reference": "this works and is the smallest possible change but we could definitely fail earlier"
+    },
+    {
+      "prediction": "I think this should be `!amd64` instead of `!arm64`",
+      "reference": "Should not have snuck in here right? This is a golang 1.17 feature, wii definitely want to update these"
+    },
+    {
+      "prediction": "I think this should be `this.prevFolderId_ = this.props.folderId;` instead of `this.prevFolderId_ = null;`",
+      "reference": "Could you explain the logic with prevFolderId?"
+    },
+    {
+      "prediction": "I think it's better to use `time.Now().Unix()` instead of `clock.New().Now().Unix()`.",
+      "reference": "Please add license header"
+    },
+    {
+      "prediction": "I think this should be `lbann_comm_ptr{new lbann_comm(0), lbann::finalize }`.",
+      "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 `autoInsertRow: true` by default.",
+      "reference": "Can you add missing asterisk?"
+    },
+    {
+      "prediction": "I think this method should be called `capitalizeDigitsAfterNumbersInCamelCase` to make it clear what it does.",
+      "reference": "late to the party, but I bet it meant to be `capitalizeLettersAfterNumbers` :)"
+    },
+    {
+      "prediction": "I think this should be `private static final boolean FIRST_TIME_EMPTY_OBJECT = true;`",
+      "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": "I'm not sure about this change. I think it's better to keep it as `curOpts` and `newOpts` to make it clear that we're dealing with the current and new options.",
+      "reference": "Looks like a \"find and replace\" unintended change here"
+    },
+    {
+      "prediction": "I think this should be `if not self.learningMode and self._sfdr.getIterationLearnNum() == 0:`",
+      "reference": "haven't we got rid off `randomSP` recently? (I think `not learn` implied that)"
+    },
+    {
+      "prediction": "I don't think this is the right place to set the status. The status should be set when the chaos instance is created.",
+      "reference": "Why we need use `in.GetStatus()` function here? Can we use `in.Status.xxxx` directly?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it would be better to return an error if the renewal is already in progress, so that the caller can decide what to do (e.g. wait for the renewal to complete).",
+      "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": "nit: I think we can remove the `#ifndef FEATURE_PAL` and `#else` and just have the `extern \"C\" HRESULT ExtQuery(PDEBUG_CLIENT client)` in the `#ifndef FEATURE_PAL` block.",
+      "reference": "We don't need this for the PInvoke?"
+    },
+    {
+      "prediction": "I think this should be `if ds.GenericDeploymentConfig.DeploymentNotification == nil || ds.GenericDeploymentConfig.DeploymentNotification.Mentions == nil { return nil, nil }`",
+      "reference": "nits, I think add a log (using s.logger) to show why does this return with no error is better."
+    },
+    {
+      "prediction": "I think it's better to return `status.Errorf(codes.PermissionDenied, \"Requested piped doesn't belong to the project you logged in\")` instead of `status.Error(codes.PermissionDenied, \"Requested piped doesn't belong to the project you logged in\")`.",
+      "reference": "Btw, It would be nice if we have some tests for those validation functions."
+    },
+    {
+      "prediction": "I think this should be `@Nullable String methodName();` so that we don't have to check for `null` in the `PathTemplateCheckView` implementation.",
+      "reference": "Should this be called something that indicates its function, rather than its content? validationMessagePrefix, or similar?"
+    },
+    {
+      "prediction": "I don't think this is correct. The copyright year should be 2010-2020, not 2018-2019.",
+      "reference": "Please remove this line"
+    },
+    {
+      "prediction": "I think this should be \"Don't run the daemon\".",
+      "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": "I'm not sure about this change. I think it's better to keep it as it was before.",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be an error, not a debug message.",
+      "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 think we can remove the `toString()` here and just check the error object directly.",
+      "reference": "You can remove these if they've been resolved in NODE-3405 and this isn't depending on it"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "don't need the class here if its empty"
+    },
+    {
+      "prediction": "I don't think this is necessary. `t` is a macro, so it will be evaluated at compile time.",
+      "reference": "This should be computed('i18n.locale'...."
+    },
+    {
+      "prediction": "I think this should be `raise BigQueryExecutionError(job_id, status['status']['errorResult'])` instead of `raise BigQueryExecutionError(job_id, job_id, status['status']['errorResult'])`",
+      "reference": "Please add return and return type description to docstring of this method."
+    },
+    {
+      "prediction": "Line is too long. [87/80]",
+      "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 think this should be `False` by default, and only set to `True` if `element` is an instance of `Streamable`.",
+      "reference": "This was very confusing until I realized this might refer to the *bokeh* use of the word 'streaming'."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be part of the `Task` struct.",
+      "reference": "Can you rename this as `tasksToContainerStates` ? If you do that, you can rename the lock as well"
+    },
+    {
+      "prediction": "I don't think this change is correct. If `accounts` is `null` then `accounts.length == 0` will throw a `NullPointerException`.",
+      "reference": "Fixing `lint` warnings that have existed for a while."
+    },
+    {
+      "prediction": "I think it would be better to have `fill(int n, T element)` and `fill(int n, Supplier<? extends T> elementSupplier)` instead of `fill(int n, T element)` and `fill(int n, Supplier<? extends T> elementSupplier)`",
+      "reference": "An iterator **of {\\@code n} sequential elements,** where each element ~are~ **is the** given {\\@code element}."
+    },
+    {
+      "prediction": "I don't think we should use the `nameof(IntegrationIds.AdoNet)` here. We should use `nameof(IntegrationIds.AdoNetMySql)` instead.",
+      "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 think we should be using `pkg/logs` instead of `github.com/go-logr/logr` here.",
+      "reference": "Maybe regroup this import with the other cert-manager packages."
+    },
+    {
+      "prediction": "This should be `else if (theme == R.style.Theme_AntennaPod_TrueBlack)`",
+      "reference": "There are spaces missing ;) Just have a look at the code style of the `if` statement above"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it should be done here. The reason is that we don't want to accept more than `block.BlockMessageLimit` messages in a single block, and we don't want to accept more than `block.BlockMessageLimit` messages in a single block even if we have more than `block.BlockMessageLimit` messages to choose from.",
+      "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 think this should be `len = (int)(nxt_pc - pc) + 2 * AARCH64_INSTR_SIZE;`",
+      "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": "Why is this change needed?",
+      "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": "Line is too long. [83/80]",
+      "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 `false` by default.",
+      "reference": "I think it's wise to keep the behavior backwards-compatible (so `false` as default)"
+    },
+    {
+      "prediction": "I think this should be `opts[:sameSite] = opts[:sameSite] || 'Lax'`",
+      "reference": "We don't need this line, do we?"
+    },
+    {
+      "prediction": "I think this should be `Disallow: /` instead of `Disallow: /``.",
+      "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 `0.09804f` and the same for the other values.",
+      "reference": "I'm not good at picking colors, if anyone else has a preference."
+    },
+    {
+      "prediction": "I don't think we need to cancel the timer here. It will be cancelled when the component unmounts.",
+      "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 should be `expect(diagnostics).to.have.lengthOf(0);`",
+      "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 don't think you need to change this.",
+      "reference": "this method is need for full screen video button to appear in players"
+    },
+    {
+      "prediction": "I think this should be `*uint32` instead of `*uint64`.",
+      "reference": "Why is this a pointer?"
+    },
+    {
+      "prediction": "I think it's better to use `Preconditions.checkArgument(width > 0, \"The width of truncate must larger than zero, but is %s\", width);` instead of `Preconditions.checkArgument(width > 0, \"The width of truncate must larger than zero, but is %s\", width);`",
+      "reference": "nit: Space after comma"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "A similar field is already define in `scapy/layers/bluetooth.py`. Can you merge both definitions ?"
+    },
+    {
+      "prediction": "I think it would be better to use `std::istringstream` and `std::getline` instead of `std::strtok`.",
+      "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": "I think this should be `public int $raw_line_number;` instead of `public $raw_line_number;`",
+      "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 `sys.stdout.flush()`",
+      "reference": "This is required, otherwise multiprocessing will (for some reason) crash"
+    },
+    {
+      "prediction": "nit: please remove this empty line",
+      "reference": "Please put in a separate line-separated block"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "reference": "Should error be eaten? If so, not clear why"
+    },
+    {
+      "prediction": "I think this should be `if self.on_collection and self.resource_name in ('bucket', 'container'):`",
+      "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 this should be called `WithInstalledAt` or `WithInstalledTimestampAt` or something like that. `InstalledTimestamp` sounds like it's the time the cluster was installed, not the time it was installed at.",
+      "reference": "Can we use the existing `InstalledTimestamp` function?"
+    },
+    {
+      "prediction": "I think you can remove the `isNutritionDataAvailable()` check here, as it's already checked in `saveEditedProduct()` and `saveProduct()`.",
+      "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 think this should be `/user/<user_name>/` instead of `/user/<user_name>/`.",
+      "reference": "Does this mean that `listenbrainz.org/user/iliekcomputers` will start getting 404s?"
+    },
+    {
+      "prediction": "Line is too long. [85/80]",
+      "reference": "Line is too long. [82/80]"
+    },
+    {
+      "prediction": "I think this should be `True` by default.",
+      "reference": "That's a strange name =)"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "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 think this should be `subscribeTo` or `subscribeToChanges` or something like that. `subscribe` sounds too generic.",
+      "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": "Why is this needed? I don't see how it could be possible for a link to not have a valid span context.",
+      "reference": "would it not be safer to move this condition to the `addLink` method?"
+    },
+    {
+      "prediction": "I don't think we need to check `req.ProtoMajor == 2` here, because `res.ContentLength == -1` implies `req.ProtoMajor == 2`.",
+      "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": "nit: `the partition data for the file of this task` -> `the partition data of the file of this task`",
+      "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 `RepoDoesNotExistError` instead of `RepoDoesntExistError`.",
+      "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 `localhost, .localhost, localdomain, .localdomain`",
+      "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 `$strictVariables = (Config::get('cms.enableTwigStrictVariables', null) !== null) ? Config::get('cms.enableTwigStrictVariables', false) : $isDebugMode;`",
+      "reference": "Wouldn't this set $strictVariables = true when Config::get returns false?"
+    },
+    {
+      "prediction": "I think we should remove the `@Ignore` and fix the test instead.",
+      "reference": "was this meant to be checked in?"
+    },
+    {
+      "prediction": "I think this should be 10s or 30s. 1h is way too long.",
+      "reference": "check spaces after const name..."
+    },
+    {
+      "prediction": "I think we should still append the attribute even if the driver version is empty.",
+      "reference": "If the GPU setting is enabled, resourceFields.NvidiaGPUManager will be nil, right?"
+    },
+    {
+      "prediction": "Should this be `File.exist?`?",
+      "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 think this should be `request.isInitiatedByUser()` instead of `true`.",
+      "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 an error, not a warning.",
+      "reference": "I think the wording on the left is better"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do.",
+      "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 right thing to do here. I think it should be `args.Name.StartsWith(AssemblyName) && !args.Name.StartsWith(\"Datadog.Trace, Version=\")` so that we don't try to load `Datadog.Trace, Version=1.0.0.0` with `Datadog.Trace, Version=2.0.0.0`.",
+      "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 int here?",
+      "reference": "Use instead `convert.Int32ToString` that was added in #762 for this purpose."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to avoid creating a new `HashSet` if the element is already in the set.",
+      "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 necessary. `Ellipsis` is already defined in `builtins`, so you can just do `from builtins import Ellipsis` and be done with it.",
+      "reference": "This is a regression test for code I found while working on this."
+    },
+    {
+      "prediction": "I think you can use `strncpy_s` 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 `if user_from < 0 or not isinstance(user_from, int):`",
+      "reference": "Does it come as `str`?"
+    },
+    {
+      "prediction": "I don't think you need this change.",
+      "reference": "This should actually read as the following `for (int i = 0; i < allergens.size(); i++)`"
+    },
+    {
+      "prediction": "I think this should be `if v.Type() == semantic.Null`",
+      "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 if this is the right thing to do, but it seems to fix the issue for me.",
+      "reference": "Really, really small nit   I believe there's 3x tabs in here? And should it be just 2x?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this warning. I think it should be in the description of the `leave_parts_on_error` option.",
+      "reference": "Perhaps note that rclone can't do this yet?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it would be better to have it in a separate package, like `github.com/keybase/client/go/net/http3` or something like that.",
+      "reference": "You probably don't need to set an empty config here."
+    },
+    {
+      "prediction": "I think this should be `except ImportError:`",
+      "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` instead of `openebs.io/bd-predecessor`",
+      "reference": "don't we need to set this label on pool pods? I don't see them being set"
+    },
+    {
+      "prediction": "I think it's better to keep the exclamation mark in the error message.",
+      "reference": "can remove, it isn't reachable."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [83/80]",
+      "reference": "`expect(cancellation).to be_can_downgrade_instead` would be preferred :rainbow: :rainbow:"
+    },
+    {
+      "prediction": "I think this should be `if (_socket.IsClosed || result.IsCancelled) { break; }`",
+      "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 don't think we need to pass the `io_thread_pool` to `fetch_datum` anymore. We can just call `io_thread_pool.get_local_thread_id()` in `fetch_datum`.",
+      "reference": "If this is not used, we should just delete the line."
+    },
+    {
+      "prediction": "I think we should keep the original error message here.",
+      "reference": "Are there other callers to this function. How will they be impacted."
+    },
+    {
+      "prediction": "I think this should be `from .plugin import Plugin`",
+      "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 this should be `--client-id` or `--client-id-string` to be consistent with the rest of the options.",
+      "reference": "Suggestion: specify `arity` for this option."
+    },
+    {
+      "prediction": "I think this should be `False` by default.",
+      "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": "I think we should also trim the `traceId` in the `validateHexAndReturnZeroPrefix` method.",
+      "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 think this should be `# pylint: disable=no-member` instead of `# pylint: enable=no-member`",
+      "reference": "Probably also needed for the `http.client.FOUND` below?"
+    },
+    {
+      "prediction": "I think this should be `--es6-js-exports` to be consistent with `--es6-js-imports` and `--es6-js-types`.",
+      "reference": "Can you also add this to Compiler.md ?"
+    },
+    {
+      "prediction": "I think this should be `@param {number} retries`",
+      "reference": "!numbers -> numbers"
+    },
+    {
+      "prediction": "I'm not sure what you mean by \"where 'artifact.name' could be empty\"",
+      "reference": "This description doesn't seem quite right for the longrunning case - `longrunning` isn't an orgname."
+    },
+    {
+      "prediction": "DocIdSetIterator is not thread safe, so it should not be returned from a method that is called from multiple threads.",
+      "reference": "maybe give it a more descriptive name, e.g. `competitiveFilter`"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's better to keep it as \"master list\" since it's a list of all the keys that can be used to verify the token.",
+      "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 should be `github.com/filecoin-project/go-filecoin/proofs/sectorproofs`",
+      "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 think this should be `if msg.MessageId != nil && err != nil {`",
+      "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": "I think this should be `propertyId` instead of `propertyID`.",
+      "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": "I think this should be added to `frontend/express/public/core/events/javascripts/countly.details.models.js` instead of creating a new file.",
+      "reference": "remove this in the new pr aswell when moving compare to plugins as discussed."
+    },
+    {
+      "prediction": "Why do we need to check nullptr here?",
+      "reference": "check varHolder is nullptr is redundant, ExecutionContext ensure"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it makes sense to have it in the helper since it's used in the view.",
+      "reference": "Make these frozen constants?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. If the message is not HTML, then it will be wrapped in a `<pre>` tag, which will make it appear as a single line in the dialog. If the message is HTML, then it will not be wrapped in a `<pre>` tag, which will make it appear as multiple lines in the dialog. I think it would be better to leave the message as is if it is not HTML, and only wrap it in a `<pre>` tag if it is HTML.",
+      "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 think this should be `maxSelects`",
+      "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": "Line is too long. [91/80]",
+      "reference": "There seem to be more changes in here than I'd expect (mostly `limit: 255` additions)."
+    },
+    {
+      "prediction": "I don't think we need to hardcode these values. We can get them from the spec.",
+      "reference": "I'm not sure, but I think it's `Tenant`"
+    },
+    {
+      "prediction": "I think it would be better to make this the default behavior and remove the `allowOptionalMappings` parameter.",
+      "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 think this should be `return o.chooser` instead of `return o.chooser.Chooser()`.",
+      "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": "This should be `DRM_MODE_OBJECT_CONNECTOR`",
+      "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 think this should be `watch.nonce <= nonce`",
+      "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 think this should be `const { ShareExtension } = NativeModules;` instead of `const { NativeModules, Platform } = require('react-native'); const ext = (Platform.OS === 'android' && ShareExtension) ? { data: () => ShareExtension.data(), close: () => ShareExtension.close(), } : { data: () => {}, close: () => {}, }; export default ext;`",
+      "reference": "Please name the file ShareExtension.js, as you import it under this name (also could you convert it to TypeScript please?)"
+    },
+    {
+      "prediction": "I think this should be `TLSInsecure` instead of `tls_insecure`.",
+      "reference": "Not sure if we need it, but if we set it, be sure to add `,omitempty`."
+    },
+    {
+      "prediction": "nit: I think we should remove the trailing dot here.",
+      "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 you need to cast here. `element` is of type `Object` and `back` is of type `IndexedSeq<Character>`, so `back.indexOf(element, from)` should be fine.",
+      "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 we should add this method to `ASTAnyTypeDeclaration`. I think it should be added to `ASTInterfaceDeclaration` and `ASTAnnotationTypeDeclaration` instead.",
+      "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 = user_terminate_after or 10000`",
+      "reference": "This will potentially skip package results. Do we really want that?"
+    },
+    {
+      "prediction": "I don't think we need to add this to the map.",
+      "reference": "Please remove `HIP_UNSUPPORTED`"
+    },
+    {
+      "prediction": "I think this should be `MutableStateStats` instead of `NewMutableStateStats`",
+      "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 don't think we need to hardcode this address. We can get it from the genesis block.",
+      "reference": "maybe will be better to put it into the default worker's config rather than hardcoding the address?"
+    },
+    {
+      "prediction": "I think it would be better to use `expect(collection.find({}, options)).to.throw...` instead of `expect(() => collection.find({}, options)).to.throw...`",
+      "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 think this should be `output->fullscreen_view->set_fullscreen(output->fullscreen_view, false);`",
+      "reference": "This is already done by `view_set_fullscreen`"
+    },
+    {
+      "prediction": "I think it's better to return a copy of the slice, so that the caller can't modify the original slice.",
+      "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": "I think it would be better to use `FlagInterface` instead of `Flag` here.",
+      "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 this should be `term_ != currTerm` instead of `term_ < currTerm`.",
+      "reference": "\"break\" is right? You skip the line 768."
+    },
+    {
+      "prediction": "Align the parameters of a method call if they span more than one line.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Can this use the route helper instead?"
+    },
+    {
+      "prediction": "I think this should be `Handler http.Handler` instead of `Handler *http.Handler`.",
+      "reference": "Why is this field exported and others aren't?"
+    },
+    {
+      "prediction": "I think we should add this to `test/functional/README.md` as well.",
+      "reference": "should we rename this like `mongodb_srv_tests.js`?"
+    },
+    {
+      "prediction": "I think this should be `googlesitekit.pageTitle && googlesitekit.pageTitle.length > 0 ? 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 don't think this is what you meant to link to.",
+      "reference": "Oops.. meant to link to SplitOp here. I'll clean up in my next commit"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "Remove these unnecessary imports that you've added."
+    },
+    {
+      "prediction": "I think this should be `success_http` instead of `success_https`.",
+      "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, self.local_port = addr, self.local_port`",
+      "reference": "It seems this is not the way to fix issue #505."
+    },
+    {
+      "prediction": "`class_weight` -> `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": "nit: I think it would be good to have a test case for each of the network tags we support, so that we can verify that each one works as expected. For example, we could have a test case for each of the following network tags: - `kubernetes.io/cluster/{cluster-name}` - `kubernetes.io/role/control-plane` - `kubernetes.io/role/etcd` - `kubernetes.io/role/node` - `kubernetes.io/role/master` - `kubernetes.io/role/control-plane` - `kubernetes.io/role/etcd` - `kubernetes.io/role/node` - `kubernetes.io/role/master` - `kubernetes.io/role/control-plane` - `kubernetes.io/role/etcd` - `kubernetes.io/role/node` - `kubernetes.io/role/master` - `kubernetes.io/role/control-plane` - `kubernetes.io/role/etcd` - `kubernetes.io/role/node` - `kubernetes.io/role/master` - `kubernetes.io/role/control-plane` - `kubernetes.io/role/etcd` - `kubernetes.io/role/node` - `kubernetes.io/role/master",
+      "reference": "Same here, merge with an existing test."
+    },
+    {
+      "prediction": "This change is not related to this patch. Please revert it.",
+      "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 we should use `org.apache.iceberg.types.Types` instead of `org.apache.iceberg.type`",
+      "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 you can remove the `Promise` import here.",
+      "reference": "Why do we need to pull bluebird in here?"
+    },
+    {
+      "prediction": "I think we should not delete the tempDir here. If there is an error in preparing the flow execution, we should keep the tempDir for debugging purpose.",
+      "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": "nitpick: I think it would be better to use `h2o_linklist_t pool_link` instead of `h2o_linklist_t target_link` and `h2o_linklist_t pool_link` in `h2o_socketpool_target_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'm not sure I understand what this test is testing. It seems to be testing that the utterance is split at the ConfigProfileTriggerCommand, but I don't see how it's testing that the utterance is split at the ConfigProfileTriggerCommand. Can you explain what this test is testing?",
+      "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'm not sure if this is the right thing to do, but I think it's better to use the user's email instead of the checkout's email.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "reference": "I don't know if we even need this parameter."
+    },
+    {
+      "prediction": "This should be `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 why this was changed, but I think it should be left as `ReadToEnd()`. `ReadToEndAsync()` is not supported in .NET 4.5.",
+      "reference": "This needs to be synchronous for the timeout to work."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I think we can remove the `purchase.starts_on.to_s(:simple)` and just say \"We will be in touch before the workshop with a reminder and any further instructions.\"",
+      "reference": "Can you break this onto multiple lines?"
+    },
+    {
+      "prediction": "I think this should be `0xf2800000` and `0xf2880000` instead of `0xf2800010` and `0xf2880010`.",
+      "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 `+optional` and `false` by default.",
+      "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 this test is not necessary.",
+      "reference": "Looks like loop with overhead. You can: - create SID - test private function `sessionIsUnique()`"
+    },
+    {
+      "prediction": "I think this should be `PREF_QUEUE_SORT_ORDER`",
+      "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": "nit: can you remove the extra blank line here?",
+      "reference": "/me can't wait to rename this repo :)"
+    },
+    {
+      "prediction": "I don't think we need to do this. We can just do `const Collection = require('./collection');` and `const Db = require('./db');`",
+      "reference": "nit: maybe call these `loadCollection()/loadDb()`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `ExternalTask` class, not in the `Task` class.",
+      "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 this is the best way to do this, but it works for now. I'm not sure if there's a better way to check if the BlacklightMarc module is loaded.",
+      "reference": "Some day, we really need to make these toolbars pluggable.."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `pygment` argument should be removed from the `show_source` method, and the `pygment` argument should be added to the `WebEngineAction` constructor. Then, the `show_source` method should call `self._show_source_pygment()` if `self._pygment` is not `None`, otherwise it should call `self._widget.triggerPageAction(QWebEnginePage.ViewSource)`.",
+      "reference": "This should be `pygment=False` like in `browsertab.py`"
+    },
+    {
+      "prediction": "I think this should be `c.msubs = opts.MaxSubs - 1`",
+      "reference": "Should this be moved down (or up) to under the client lock instead of server lock?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Why are we removing this?"
+    },
+    {
+      "prediction": "I think this should be `self.cart.id.to_s` so that we don't get a `NoMethodError: undefined method 'id' for nil:NilClass` if `self.cart` is `nil`.",
+      "reference": "What do you think about using `self.id` ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this. I think we should be able to do something like `healthAggregator.Report(healthName, &health.HealthReport{Live: false, Ready: false})` in the `typhaConnection.Start` function.",
+      "reference": "Please log once here at Error level \"Failed to connect to Typha, will retry...\""
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Can you revert this change? The `VERSION` value will be updated after maintainer determines the version at release."
+    },
+    {
+      "prediction": "`qulifiedServiceNameKey` -> `qualifiedServiceNameKey`",
+      "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 `if examples_count == 0 && @configuration.fail_if_no_examples` instead of `if @configuration.fail_if_no_examples && examples_count == 0`",
+      "reference": "This should be `@configuration.failure_exit_code`, we don't want to hard code 1."
+    },
+    {
+      "prediction": "I think this should be `@override` instead of `@override`.",
+      "reference": "Changed per review in Blockly."
+    },
+    {
+      "prediction": "nit: I think it would be better to have a constant for `macRewriteMark` and `macRewriteMarkRange`.",
+      "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 call this `exclude-too-few-public-methods-classes` or something like that.",
+      "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 `self.config['platforms'][platform]['instances']` instead of `self.config['vagrant']['instances']`.",
+      "reference": "Docstring for `platform`."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should just return the error and let the caller decide what to do with it.",
+      "reference": "This should be called before any `os.Exit` otherwise it could still mess up things."
+    },
+    {
+      "prediction": "I'm not sure why this was changed to an int, but I think it should be left as a byte.",
+      "reference": "shouldn't this be `short` ?"
+    },
+    {
+      "prediction": "I don't think we should print the capacity of the volume here. I think we should print the capacity of the volume in the `volume inspect` command.",
+      "reference": "Can you please paste the output of `volume list` command which will show capacity of a openebs volume."
+    },
+    {
+      "prediction": "I think this should be `IllegalArgumentException(\"metric variable must not be null\")`",
+      "reference": "I'd make this a `throw new NullPointerException(\"metric == null\");` instead"
+    },
+    {
+      "prediction": "This should be `#include \"Util/LuaUtil.h\"`",
+      "reference": "Include should be order lexicographically."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like a good place to me. I don't think we need to do anything with the return value of `setMinPasscodeLength`.",
+      "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": "Line is too long. [101/80]",
+      "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 `!shouldRepeat`?",
+      "reference": "Looks like the episode should also not be deleted when repeating"
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "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 if this is the best way to do this, but it seems to work.",
+      "reference": "`options['binary'] = WebDriver::Chrome.path` if set?"
+    },
+    {
+      "prediction": "I think this should be `explorer.db` instead of `./explorer.db`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think it would be better to move this to `HttpProtocols` so that it can be shared with `ListenOptions` and `KestrelServerOptions`.",
+      "reference": "Nit: Supported -> Enabled."
+    },
+    {
+      "prediction": "I'm not sure why this change is necessary, but it seems like a good idea to use `.at` instead of `.set_value`.",
+      "reference": "Yay! I was too lazy to fix this."
+    },
+    {
+      "prediction": "@aravindhp can you please revert this change? I think it is not required.",
+      "reference": "other declaration of errors (from `typecheck`)"
+    },
+    {
+      "prediction": "I think this should be `^https:\\/\\/github\\.com\\/(?P<owner>.+)\\/(?P<repo>.+)$`",
+      "reference": "nit: error starts with capital letter"
+    },
+    {
+      "prediction": "I don't think this is a good idea. It's better to keep the old code and add a new method `Flag.valueOfByRealName(String realName)` that returns the corresponding `Flag` enum value.",
+      "reference": "the original code ignores \\Draft and \\Recent, whereas your change will not. what this matters, i'm not sure."
+    },
+    {
+      "prediction": "I think this should be `defer config2.SetBlockServer(oldBServer)`",
+      "reference": "Why was this necessary?"
+    },
+    {
+      "prediction": "I think this should be called `--force-rerun` or something like that. `--rerun` sounds like it's going to rerun all tests, not just the failing ones.",
+      "reference": "Can you also do this for Cover?"
+    },
+    {
+      "prediction": "Shouldn't this be `VK_IMAGE_ASPECT_PLANE_0_BIT | VK_IMAGE_ASPECT_PLANE_1_BIT | VK_IMAGE_ASPECT_PLANE_2_BIT` instead of `(VK_IMAGE_ASPECT_PLANE_0_BIT | VK_IMAGE_ASPECT_PLANE_1_BIT | VK_IMAGE_ASPECT_PLANE_2_BIT)`?",
+      "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": "Style/TrailingCommaInArrayLiteral: Avoid comma after the last item of an array.",
+      "reference": "This method is not needed for postgresql, by default sum function omits null values so we can remove this file"
+    },
+    {
+      "prediction": "I don't think we need to do this here. The `systemdsystemgeneratordir` variable is only used in `src/Makefile.am`, which is only included when `BUILDOPT_LIBSYSTEMD` is true.",
+      "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": "nit: remove empty line",
+      "reference": "`classification` is misspelled"
+    },
+    {
+      "prediction": "nit: `methodViewBuilder.initCode(methodSampleViews.get(0).methodInitCode());`",
+      "reference": "Can you change this to `sampleInitCode`?"
+    },
+    {
+      "prediction": "`startTs` should be `endTs`?",
+      "reference": "I added a startTs information in `TiDagRequest` output. Not really sure if it is useful? @zhexuany"
+    },
+    {
+      "prediction": "I think we should use `frame.node.offsetWidth` and `frame.node.offsetHeight` instead of `frame.node.getBoundingClientRect().width` and `frame.node.getBoundingClientRect().height`.",
+      "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": "I'm not sure if this is the right thing to do, but it seems to be the only way to get this to work.",
+      "reference": "prefer to just add this to the `verify_authorized` exception list above since we're already making exceptions."
+    },
+    {
+      "prediction": "I don't think this is correct. `reset` returns the first element of the array, but we need to pass the name of the parent class, not the name of the trait.",
+      "reference": "was there a reason to not loop over parent classes here?"
+    },
+    {
+      "prediction": "I think we can remove the `const { computed } = Ember;` and just use `this.get('controller')` instead of `this.get('controller').get('userSession')`",
+      "reference": "@billybonks Why move `import Ember from 'ember';` to the top?"
+    },
+    {
+      "prediction": "Align the parameters of a method call if they span more than one line.",
+      "reference": "Everything's so simple (a `ThreeStateWorkflow`) until it isn't :smirk:"
+    },
+    {
+      "prediction": "I'm not sure why this is necessary, but it seems to be required for the test to pass.",
+      "reference": "I don't quite understand where the NaNs come from..."
+    },
+    {
+      "prediction": "I'm not sure if this is intentional, but it looks like you've removed the `Host` and `Port` fields from the `Network` struct in `config/config.go`.",
+      "reference": "File is not `goimports`-ed"
+    },
+    {
+      "prediction": "I think this should be `//_please: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": "I think it would be better to have `IsEip3238DifficultyBombDelayEnabled` and `IsEip3529ReductionInRefundsEnabled` instead of `IsEip3238Enabled` and `IsEip3529Enabled`.",
+      "reference": "I am not sure if we need this, we already have DifficultyBombDelays"
+    },
+    {
+      "prediction": "Why do we need to dispose them here?",
+      "reference": "move to finally - I can see at least one return that would omit it, also any exception and so on"
+    },
+    {
+      "prediction": "I don't think we need to check `stakingV2 == nil` here. If `electionCommittee` is nil, then `stakingV2` should be nil as well.",
+      "reference": "remove `genesisConfig.GravityChainStartHeight == 0`"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think we should just use `require_once __DIR__ . '/../../vendor/symfony/var-dumper/Resources/functions/dump.php';`",
+      "reference": "this might kill whole application if `var-dumper` will not be installed. What about two separate file_exists conditions (monorepo/project)?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `app.CreateSettingsFile()`, which is called from `app.Init()`, which is called from `app.New()`, which is called from `app.NewApp()`, which is called from `app.NewAppFromConfig()`, which is called from `app.NewAppFromConfigFile()`, which is called from `app.LoadApp()`, which is called from `app.LoadAppFromConfigFile()`, which is called from `app.LoadAppFromConfig()`, which is called from `app.LoadAppFromConfigFile()`, which is called from `app.LoadAppFromConfig()`, which is called from `app.LoadAppFromConfigFile()`, which is called from `app.LoadAppFromConfig()`, which is called from `app.LoadAppFromConfigFile()`, which is called from `app.LoadAppFromConfig()`, which is called from `app.LoadAppFromConfigFile()`, which is called from `app.LoadAppFromConfig()`, which is called from `app.LoadAppFromConfigFile()`, which is called from `app.LoadAppFromConfig()`, which is called from `app.LoadAppFromConfigFile()`, which is called",
+      "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": "I think this should be renamed to `pretty_table` or `pretty_table_list` or something like that. `pretty_list` is too generic.",
+      "reference": "Why are you changing this name ?"
+    },
+    {
+      "prediction": "I don't think this is enough. We need to check if the current user is `localstack` and if the current working directory is `/opt/code/localstack`.",
+      "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 `isinstance(node.parent, astroid.Call) and node is node.parent.func`",
+      "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'm not sure why this is necessary, but it seems to fix the issue.",
+      "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 just remove this test for now. We can add it back when we support Spark 3.",
+      "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 think this should be `check.file_param(notebook_path, \"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 think this should be `search_builder_class.new(processor_chain, self)` instead of `Blacklight::Solr::SearchBuilder.new(processor_chain, self)`",
+      "reference": "I wonder if we should push this and the search_params_logic into the blacklight config."
+    },
+    {
+      "prediction": "I think this should be `cfg.Genesis.KamchatkaBlockHeight = 10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "don't need this line, since it is not tested here?"
+    },
+    {
+      "prediction": "I think this should be `ami_version='1.0'` to make it clear that this is a keyword argument.",
+      "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|src/generated.*)/.*\", project.getBuildDir().getAbsolutePath())`",
+      "reference": "I wonder if this absolute path is gonna affect the build-cache relocatability of these compile tasks??"
+    },
+    {
+      "prediction": "I think we can remove the `if (this.inTransaction) this.writeConcern = undefined;` line as well.",
+      "reference": "possible bug if `fullResponse` is false"
+    },
+    {
+      "prediction": "This import is not used in this file. Please remove it.",
+      "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": "Please remove these debug statements.",
+      "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": "Shouldn't this be `System\\Classes\\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 a `bool` rather than a `string`.",
+      "reference": "this should be `EnabledAdminAPI` and the json string should then be `enableAdminAPI`"
+    },
+    {
+      "prediction": "This import is not used anywhere in this file.",
+      "reference": "Noticed some missing imports in manage.py so thought I'd add them"
+    },
+    {
+      "prediction": "I think we can remove the `DyingAndDoneExport` field from the `flowRecord` struct and just use the `conntrackConnStore.SetExportDone(key)` to do the deletion.",
+      "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 don't think this is a good example. It's not clear what it is trying to show.",
+      "reference": "I am not sure what GitHub wants to tell me with this icon here :)"
+    },
+    {
+      "prediction": "I think it would be good to have a test case with a zoom level of 0 as well.",
+      "reference": "can we also check for smaller zoom levels returning valid error codes?"
+    },
+    {
+      "prediction": "I think this should be `LogInfo` and `LogWarning` instead of `LogHeader1` and `LogHeader2`.",
+      "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 is this change needed?",
+      "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'm not sure if this is the right thing to do, but I think we need to pass `true` for both arguments here.",
+      "reference": "out of curiosity, with IPv6 introduced, why there's no change to the expected output of openflow in the test?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. I think we need to add a new flag to indicate that we don't have precise flag information.",
+      "reference": "nit: grammar: instruction"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "We can revert this to reduce the diff of the PR."
+    },
+    {
+      "prediction": "nit: can you add a blank line between the fields and the body of the function?",
+      "reference": "i think this missed the reconciler to callsite change"
+    },
+    {
+      "prediction": "I think this should be `getNotImplementedString(\"SurfaceNamer.getGrpcClientConfigImportModule\")`",
+      "reference": "Please reuse `getClientConfigPath`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the matcher to work.",
+      "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": "Why do we need to order by protocol code here?",
+      "reference": "Maybe move _agreedCapabilities and _availableCapabilities from List to SortedSet ? We could order them by code and then by version already."
+    },
+    {
+      "prediction": "I don't think this is correct. If there are no messages in the filtered state, then we should not update the starting sequence.",
+      "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'm not sure if this is the right place to put this, but I think it would be nice to have a button to download the preferred citation as a `.txt` file.",
+      "reference": "why is the `send` needed?"
+    },
+    {
+      "prediction": "`bboxes` and `labels` should be `torch.Tensor`.",
+      "reference": "Is this necessary? If so, we need to update the docstring."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Checking for actionable here instead of grabbing the first pending approval. I tested this with proposal ID 265 on staging."
+    },
+    {
+      "prediction": "I don't think we need this helper function. The `assert` statement is already in the `check_result` method.",
+      "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 think this should be `IHttpParser<T>` where `T` is `IHttpStartLineHandler` and `IHttpHeadersHandler`.",
+      "reference": "nit: the correct name is \"request line\", so let's use the right one here :)"
+    },
+    {
+      "prediction": "I think this should be `os.environ.get('ALLOW_INSECURE_CATALOG_ACCESS', False)` so that it defaults to `False` if the environment variable isn't set.",
+      "reference": "This should now use `[]`, so it crashes if the var is missing"
+    },
+    {
+      "prediction": "I think this should be `if (map_state == NULL) { ... }` instead of `if (!map_state) { ... }`.",
+      "reference": "you could use `bpf_map_update_elem` with `BPF_EXIST`"
+    },
+    {
+      "prediction": "I think this is a typo, it should be \"Filed\" instead of \"Filed\".",
+      "reference": "thanks for noticing this! But I think we should not have it in this PR."
+    },
+    {
+      "prediction": "This should be `this.inputs[i].nSequence = Input.DEFAULT_LOCKTIME_SEQNUMBER;`",
+      "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 `if data_format not in cls.supported_data_formats: raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError('GCS uploader', data_format) raise InvalidDataFormatError",
+      "reference": "If this is a base method that could be used by others, should you hardcode 'GCS uploader' here?"
+    },
+    {
+      "prediction": "I don't think we need to create a new `StringBuilder` here. We can just use `back` instead.",
+      "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 think this should be `POST` instead of `GET`",
+      "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": "I think this should be `0x4200000000000000000000000000000000000000` instead of `0x420000000000000000000000000000000000000F`",
+      "reference": "This is the default value?"
+    },
+    {
+      "prediction": "I don't think we need to import `strftime` here.",
+      "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 don't think we should do this. The `MimeTypeDetectionByFileExtension.properties` file is used to detect the MIME type of a file based on its extension. If we return `.md` for `text/markdown` then we'll have a circular dependency between `MimeTypeDetectionByFileExtension.properties` and `DownloadInstanceWriter`.",
+      "reference": "FWIW: We have FileUtil.generateOriginalExtension for tabular files but it uses a hardcoded list rather than the properties file."
+    },
+    {
+      "prediction": "I think this should be `configParams.KubernetesProvider() == config.ProviderAKS || configParams.KubernetesProvider() == config.ProviderEKS`",
+      "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 think this should be `DebugSimulateDataRace` instead of `DebugSimulateDataRace`.",
+      "reference": "Are all fields beginning with \"Debug\" automatically `;local` ? (I guess so, but just checking.)"
+    },
+    {
+      "prediction": "I think it's better to use `DeprecationWarning` instead of `DeprecationWarning`",
+      "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 you need to pass `false` and `true` here. `getQuotedIdentifier` should be enough.",
+      "reference": "Let's use prepared statements here instead"
+    },
+    {
+      "prediction": "I think this should be `import Button from 'GoogleComponents/button';`",
+      "reference": "Just leaving this in one place but thanks so much for re-organising all these imports."
+    },
+    {
+      "prediction": "I don't think you need to do this for x86.",
+      "reference": "Ditto, here and below"
+    },
+    {
+      "prediction": "I don't think we need this check. `vnode.type` is always a function.",
+      "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": "Shouldn't this be `default=[]`?",
+      "reference": "Seems like the docstring here should simply be \"Deprecated; use `hooks` instead.\"."
+    },
+    {
+      "prediction": "This should be `else if (UserPreferences.getTheme() == R.style.Theme_AntennaPod_TrueBlack)`",
+      "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": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "How about `null: false` on these two lines? Also `t.timestamps null: false`?"
+    },
+    {
+      "prediction": "I think this should be a warning, not an error.",
+      "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 called `isSymbolDefined` or `isSymbolDeclared` or something like that. `isSymbolUsed` sounds like it means that the symbol has been used in some way, but I don't think that's the case here.",
+      "reference": "The `SymbolTable` deals only with symbols, I think we can just call this `isUsed` here."
+    },
+    {
+      "prediction": "I think this should be `if (webdriver.http.CorsClient.isAvailable() || url[0] == '/') {`",
+      "reference": "Please hide this behind a flag and add a setter function to the builder."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Just a side-node, no need to change it here: The term \"recent\" might be confusing at a later point in time."
+    },
+    {
+      "prediction": "`val1.keys !== undefined` should be `val1.keys !== undefined && typeof val1.keys === 'function'`",
+      "reference": "When can it be undefined? Wouldn't that be a bug?"
+    },
+    {
+      "prediction": "I don't think this import is needed.",
+      "reference": "JMX offers normal APIs for this. I don't think you want the glassfish one."
+    },
+    {
+      "prediction": "Why do we need to add a trailing comma here?",
+      "reference": "Could you remove `,' at the end?"
+    },
+    {
+      "prediction": "I think this should be `if (typeof opt_port === 'function')` instead of `if (typeof opt_port !== 'function')`",
+      "reference": "Maybe it would simpler to ignore opt_port if type !== 'number'?"
+    },
+    {
+      "prediction": "SpendingKey -> EffectiveAddr",
+      "reference": "do you want to rename the codec tag here too?"
+    },
+    {
+      "prediction": "I think we should rename this to `CAST_TAB_MIRRORING` to be consistent with the rest of the commands.",
+      "reference": "These command names are specific to Chromium-based browsers. Please move to `ChromiumDriverCommand`"
+    },
+    {
+      "prediction": "I don't think we need to do this. `F` is already imported from `pyspark.sql.functions`, so we can just do `from databricks.koalas import functions as kF` and use `kF` instead of `F`.",
+      "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": "I think it would be better to use `Zalas\\Injector\\PHPUnit\\TestCase\\ServiceContainerTrait` instead of `ServiceContainerTestCase`.",
+      "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 don't think we need to maintain the old flink job's id in flink state backend to find the max-committed-checkpoint-id when traversing iceberg table's snapshots. The max-committed-checkpoint-id can be found by traversing iceberg table's snapshots without maintaining the old flink job's id in flink state backend.",
+      "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 don't think we need to pass `log_signal` to `_filter_signals`. We can just use `debug.signal_name(signal) not in self.BLACKLIST` inside `_filter_signals`.",
+      "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 think this should be `reject(error); return;`",
+      "reference": "Please use async/await and try/catch"
+    },
+    {
+      "prediction": "I think this should be `if (field.nested_flatbuffer != nullptr && field.nested_flatbuffer->type.struct_def != nullptr) {`",
+      "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": "This should be ASSERT_NO_FATAL_FAILURE(InitState(nullptr, nullptr, VK_COMMAND_POOL_CREATE_RESET_COMMAND_BUFFER_BIT, VK_API_VERSION_1_1));",
+      "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 don't think we need to change this method. It's only used in `NodeJSGapicContext#fieldPropertyComment`, which is only used in `NodeJSGapicContext#convertField`. `NodeJSGapicContext#convertField` is only used in `NodeJSGapicContext#convertMessage`. `NodeJSGapicContext#convertMessage` is only used in `NodeJSGapicContext#convertClass`. `NodeJSGapicContext#convertClass` is only used in `NodeJSGapicContext#convertFile`. `NodeJSGapicContext#convertFile` is only used in `NodeJSGapicContext#convertPackage`. `NodeJSGapicContext#convertPackage` is only used in `NodeJSGapicContext#convert`. `NodeJSGapicContext#convert` is only used in `NodeJSGapicContext#generate`. `NodeJSGapicContext#generate` is only used in `NodeJSGapicContext#generateFile`. `NodeJSGapicContext#generateFile` is only used in `NodeJSGapicContext#generatePackage`. `NodeJSGapicContext#",
+      "reference": "Don't we still want to check it it's a keyword?"
+    },
+    {
+      "prediction": "I'm not sure why this change was made. The `<br>` and `<table>` tags are used to create a line break between the text and the image, and the `<strong>` and `<u>` tags are used to make the link text bold and underlined, respectively. Without these tags, the text and image will be displayed on the same line, and the link text will not be bold or underlined.",
+      "reference": "this can just me `link_text` :tomato:"
+    },
+    {
+      "prediction": "I think it's better to use `self.assert_eq` instead of `self.assert_eq(pser.argsort().sort_index(), kser.argsort().sort_index())` because `pser.argsort().sort_index()` and `kser.argsort().sort_index()` are not the same.",
+      "reference": "What if the Series contains null values?"
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it's better to keep the `AnalysisTarget` as a `List<PhysicalLocationComponent>`. The `AnalysisTarget` property is defined as a `List<PhysicalLocationComponent>` in the Sarif schema, so it makes sense to keep it that way.",
+      "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": "I think we should use `1e-12` instead of `EPS` in this file.",
+      "reference": "Like in atss_head"
+    },
+    {
+      "prediction": "I think this should be `pool.numPendingWholeBlocks - 1`",
+      "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": "Extra empty line detected at class body beginning.",
+      "reference": "Is there a reason we _wouldn't_ want to do this? A bit of a security hole :grimacing:"
+    },
+    {
+      "prediction": "I think you can use `Files.newInputStream` instead of `FileInputStream`.",
+      "reference": "Please add a copyright notice to all new files. Is there a tool to check and auto update copyright notices?"
+    },
+    {
+      "prediction": "I think this should be `t('Email address')`",
+      "reference": "A new translation dump should be made after this change (removing the \"Email\" message)"
+    },
+    {
+      "prediction": "`implicitSelector => implicitSelector + ':not([role])'` should be `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 we can just use `this` instead of `thing` here.",
+      "reference": "@JoviDeCroock I replaced the refs here with simple assignments."
+    },
+    {
+      "prediction": "I think this should be `private`.",
+      "reference": "nit: Could we use VisibleForTesting annotation here?"
+    },
+    {
+      "prediction": "I think this should be `must_equal topic.posts.order(created_at: :desc)`",
+      "reference": "Can we create some post records to assert sort_by"
+    },
+    {
+      "prediction": "Why do we need to reset the stream here?",
+      "reference": "is this necessary?"
+    },
+    {
+      "prediction": "I think this should be `private set;`",
+      "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 don't think we need these constants. We can just use the strings directly.",
+      "reference": "using a certain type instead of using string directly."
+    },
+    {
+      "prediction": "I think this should be \"alice and bob both rename the same file, causing a copy.\"",
+      "reference": "fix trailing ,"
+    },
+    {
+      "prediction": "I think this should be `mlat` and `mlon` instead of `latitude` and `longitude`.",
+      "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": "This validator is not used anymore.",
+      "reference": "The form no logger needs validation?"
+    },
+    {
+      "prediction": "Why is this check needed?",
+      "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": "I'm not sure this is the best way to test this, but I don't have a better idea at the moment.",
+      "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'm not sure if this is a typo or not, but it looks like it should be \"constructor type arguments\" instead of \"constructor type arugments\".",
+      "reference": "The test sources are ok though, you don't need to remove those changes"
+    },
+    {
+      "prediction": "I think this should be `maxWidth: Math.min(width * 2, 1920)`",
+      "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": "I think it would be better to use `String#contains(CharSequence)` instead of `String#contains(String)`.",
+      "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 `--force-qr` instead of `forceQR` to be consistent with the rest of the flags.",
+      "reference": "isn't the usual capitalization for subcommands force-qr? (don't care too much about it)"
+    },
+    {
+      "prediction": "I think we should keep this as a SUGGESTION, but I'm not sure if it's possible to make it a WARNING without also making it a WARNING in the baseline.",
+      "reference": "Do you know how many repos have hit this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this message, but I think it makes sense to have it here.",
+      "reference": "I'll be interested to see if our gettext rake tasks will pick up on this text to translate"
+    },
+    {
+      "prediction": "`setField(L, \"params\", spell.getHasParam());` -> `setField(L, \"params\", spell.hasParam());`",
+      "reference": "I think you need to change line 928 value from 6 to 7 too, since now table will have 7 values"
+    },
+    {
+      "prediction": "`f1[:, [\"cat\", \"dog\", \"mouse\"]].to_list() == [[1, 0, 0, 1], [1, 0, 0, 1], [0, 0, 1, 1]]` should be `f1[:, [\"cat\", \"dog\", \"mouse\"]].to_list() == [[1, 0, 0, 1], [1, 0, 0, 1], [0, 0, 1, 1], [0, 0, 0, 0]]`",
+      "reference": "Does `f0[::-1, :]` mean all the rows taken in a reverse order?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I don't think it's necessary.",
+      "reference": "Looks like there are strange characters here."
+    },
+    {
+      "prediction": "`GetPause` returns the name of the pause condition. If it is empty, the experiment is not paused.",
+      "reference": "It is better to have a check to ensure the format is valid."
+    },
+    {
+      "prediction": "nit: `next` -> `isNextAlreadyDefined`",
+      "reference": "How about `nextAlreadyDefined` variable name in method signature?"
+    },
+    {
+      "prediction": "Why is this number changed? I think it should be 3.",
+      "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": "nit: I think we can remove this line.",
+      "reference": "formatting... can this import go next to the other sources ones?"
+    },
+    {
+      "prediction": "I don't think we should change the order of these options.",
+      "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 verify that the `exception_type` tag is not added when there are no exceptions.",
+      "reference": "Should we add this to the other \"never\" verification in lines 71-73 above, too?"
+    },
+    {
+      "prediction": "I think this should be `resources = list(self.dao.iter_cai_assets(...))`",
+      "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 this should be `info.getMaximumSimultaneousSessions()` instead of `builder.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": "Metrics/LineLength: Line is too long. [103/80]",
+      "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 think this should be `var cg CgroupParams = CgroupParams{initpid: os.Getpid()}`",
+      "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": "I think it would be better to throw a `ParameterException` with a message like \"Problem with privacy-public-key-file: \" + e.getMessage() + \". Contents of privacy-public-key-file invalid: \" + e.getMessage(), e);",
+      "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 think we can remove this fixture as well.",
+      "reference": "Can we get a test with multiple directories as per the initial example?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to go about this, but it seems to work.",
+      "reference": "I assume this is because you were seeing cases where left was a negative number? When?"
+    },
+    {
+      "prediction": "I don't think you need to change the order of these includes.",
+      "reference": "this is undoing recent changes for increased consistency of include file statements. please restore to the original."
+    },
+    {
+      "prediction": "I think this should be `LambdaExpression` instead of `LambdaMethodReference`",
+      "reference": "If these are applied in order, should `LambdaMethoReference` come after `OptionalOrElseMethodInvocation`"
+    },
+    {
+      "prediction": "I think this should be `IReadOnlyDictionary<string, FileData> filesDictionary`.",
+      "reference": "Why this rename?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I don't think we should be patching `sys.modules` for Python 3.3 and above.",
+      "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 removed.",
+      "reference": "This can be removed too."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we need to make a new `MdID` here. I think we can just use `head.MD.MdID`.",
+      "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 don't think we need to import `map` from `lodash` anymore. We can just use `lodash/map` instead.",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "I don't think we need to change the encoding of this file.",
+      "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'm not sure this is the right place to do this. I think it should be done in `UvLoopHandle.Dispose` after `UvLoopHandle.Dispose` calls `UvLoopHandle.DisposeAsync`.",
+      "reference": "You could call OnStopRude() directly here to do the same thing."
+    },
+    {
+      "prediction": "I don't think we need to do this in this PR. Let's do this in a separate PR.",
+      "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 think this should be `mayactl snapshot revert --volname <vol> --snapname <snap>`",
+      "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\", taskId, containerId)`",
+      "reference": "should this be a separate/new functional test?"
+    },
+    {
+      "prediction": "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.<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.<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.<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.<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",
+      "reference": "Can these also assert at least the length of arrays returned and ideally contents?"
+    },
+    {
+      "prediction": "I think we can remove the `getattr` here and just use `task.param_visibilities = _get_default(param_visibilities, {})`",
+      "reference": "`hasattr(task, 'param_visibilities')` is more concise with the same effect, but either one works."
+    },
+    {
+      "prediction": "\"Attributes given to a function call have are being passed in a different order\" -> \"Attributes given to a function call 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": "Why is this needed? The `SnackBarHandler` already calls `show()` on the `Snackbar` instance.",
+      "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 you can use `folly::join` here.",
+      "reference": "concatHosts may be a better func name to explain this?"
+    },
+    {
+      "prediction": "I think it would be better to use `const stubString = () => '';` instead of `const stubString = always('');`",
+      "reference": "A function that returns empty string."
+    },
+    {
+      "prediction": "I think this should be `if filename == \"\" && packageName != \"\" || state.Config.FeatureFlags.RemovePleasings {`",
+      "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: I think we can remove the empty line here.",
+      "reference": "can delete this line"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be an error if there are multiple models, but I'm not sure what the right error message should be.",
+      "reference": "Will it still be an issue if all the models use the same data set?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "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 don't think this is needed.",
+      "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 this should be `Types.BooleanType.get().withDefault(false)`.",
+      "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. I think it should be done in the `create_dependencies` function.",
+      "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": "nitpick: `escape` -> `escape_slashes` or `escape_slashes_in_path` or `escape_slashes_in_uri_path` or `escape_slashes_in_uri_path_segments` or `escape_slashes_in_uri_path_segments_except_first` or `escape_slashes_in_uri_path_segments_except_first_and_last` or `escape_slashes_in_uri_path_segments_except_last` or `escape_slashes_in_uri_path_segments_except_first_and_last_except_last` or `escape_slashes_in_uri_path_segments_except_first_and_last_except_last_except_last` or `escape_slashes_in_uri_path_segments_except_first_and_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last_except_last",
+      "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 call it `NULL_ACCOUNTABLE` instead of `NOOP_ACCOUNTABLE`.",
+      "reference": "Interface members are always `static final` right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "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": "I'm not sure why this was removed, but it seems like it should be left in.",
+      "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": "Line is too long. [117/80]",
+      "reference": "nice extraction of the form for being used at new/edit. We should have more of these"
+    },
+    {
+      "prediction": "I think this should be `V6*(1 + s6*cos(6x))`",
+      "reference": "Are the coefficients the actual bug fix?"
+    },
+    {
+      "prediction": "`psycopg2.IntegrityError` is a subclass of `psycopg2.ProgrammingError`",
+      "reference": "Why are the error code options in a list and not a tuple?"
+    },
+    {
+      "prediction": "nit: `TEST_BAD_ANCESTORS` -> `TEST_INVALID_ANCESTORS`",
+      "reference": "Need alignment with the preceding line."
+    },
+    {
+      "prediction": "I don't think we need to make this a constant. It's only used in one place.",
+      "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 necessary. `empty()` returns an empty `TreeSet<T>`, not an empty `TreeSet<Object>`.",
+      "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": "`is_diff` and `add_thousands_sep` should default to `False` and `True` respectively.",
+      "reference": "Why change the default value of `is_diff`?"
+    },
+    {
+      "prediction": "I think this should be `tenacity>=6.3.0,<7.0.0`",
+      "reference": "I think tacking on `,<7` would be a wise precaution in case of future incompatible changes in tenacity."
+    },
+    {
+      "prediction": "I think we should keep the return type.",
+      "reference": "I don't think is necessary to remove the return type"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the client secret is null, we should probably just return an error to the user and not try to refresh the token.",
+      "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 should be done in the `getProfiles` resolver.",
+      "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 why this was removed, but it seems like it should still be there.",
+      "reference": "I wonder if we should trigger a PHP error and ensure it ends up in the log?"
+    },
+    {
+      "prediction": "I think we can remove these TODOs for now. We can always add them back later if we need to.",
+      "reference": "The `move_topic` action be a `post` request since it changes the state of the object."
+    },
+    {
+      "prediction": "I think this should be `if metadata == nil { metadata = &core.BuildMetadata{}`",
+      "reference": "Should we just return a non-nil metadata from `Test`?"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundMethodBody: Extra empty line detected at method body end.",
+      "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 don't think we need to import all of these.",
+      "reference": "please don't use wildcard imports"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "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": "This is not used anymore, so I removed it.",
+      "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 a `throw std::runtime_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 think this should be `node.store.tx_begin_read ()` instead of `node.store.tx_begin_write ()`.",
+      "reference": "Was changing this to a write transaction required?"
+    },
+    {
+      "prediction": "I think this should be `[]interface{}`",
+      "reference": "nit: I think \"They\"?"
+    },
+    {
+      "prediction": "I think it's better to use `ulong.MaxValue` instead of hardcoding the value.",
+      "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'm not sure if this is the best way to do it, but it works for now.",
+      "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": "Please remove this line.",
+      "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'm not sure this is the right place for this function. I think it would be better to have it in `bokeh.util` or `bokeh.util.string` or something like that.",
+      "reference": "Might be worth sorting the replacement items so this is deterministic."
+    },
+    {
+      "prediction": "It only works with ClusterScoped is false; -> It only works with ClusterScoped is true;",
+      "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 we can remove the `diagCollector` parameter from `createFieldNameConfigs` and `getRequiredFields` since they don't use it.",
+      "reference": "You'll also want to do this below, for the `optionalFieldConfigs`. See `GapicMethodConfig.java:180`."
+    },
+    {
+      "prediction": "`space` is not used in this function.",
+      "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 `LibSqreenDirectory => RootDirectory / \"packages\" / \"libsqreen.1.1.2.2\" / \"libsqreen.1.1.2.2\";`",
+      "reference": "I assume we have a package reference somewhere that pulls down the nuget package?"
+    },
+    {
+      "prediction": "I'm not sure why we need to subscribe to both the old and new subjects. I think we can just subscribe to the new one.",
+      "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": "(style) please, no space before tabs",
+      "reference": "can we use `nullptr` for consistency?"
+    },
+    {
+      "prediction": "`DomainRouter` is not a Symfony router, it's our custom router.",
+      "reference": "strange, for templating\\EngineBundle it was `\\Symfony\\Bundle\\FrameworkBundle\\`"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should always add the new deployment, even if it's not 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'm not sure if this is the best way to do this, but it seems to be the only way to get the `role` attribute as a token list.",
+      "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 rename this class to `ExpectBftProposals` or something like that.",
+      "reference": "rename ibftTwo arg to bft"
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "reference": "Should we go ahead and push `agent_is_crawler?` into the `skip_session_tracking?` method?"
+    },
+    {
+      "prediction": "`self.proposal_file is not None` -> `if self.proposal_file is not None:`",
+      "reference": "How about we raise a warning here and simply change to use self.data_infos = self.load_annotations(self.ann_file)"
+    },
+    {
+      "prediction": "I don't think we need to update the copyright year in this file.",
+      "reference": "Please remove this line to comply with the standard for copyright headers."
+    },
+    {
+      "prediction": "I don't think we need to make this configurable. It's not like we're going to change it in the middle of the epoch.",
+      "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": "nit: can we use `ScheduledExecutor` instead of `ScheduledExecutorService`?",
+      "reference": "Nit: please remove unnecessary whitespace changes."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this, but it seems to work for now.",
+      "reference": "Is this captured in a test?"
+    },
+    {
+      "prediction": "I don't think we need to make this public.",
+      "reference": "This is used in o.a.l.sandbox.search.LargeNumHitsTopDocsCollector."
+    },
+    {
+      "prediction": "I think we should keep the warmup_ratio as 1.0 / 3.",
+      "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": "nit: `initiatedChildExecutionsInSession` -> `initiatedChildExecutionsInWorkflowTask`",
+      "reference": "session is a new term. Should we call more like currentCommandBatch or something else?"
+    },
+    {
+      "prediction": "I think this should be `uuid.NewV4()` instead of `uuid.New()`.",
+      "reference": "Will go fmt before squash."
+    },
+    {
+      "prediction": "I think this should be `false` by default, and only set to `true` if we detect that we're running in a privileged container.",
+      "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 is this removed?",
+      "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 this should be `jsonschema>=3.0.0,<4.0.0`",
+      "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 think this should be `temporalImp=\"tm_py\"` until the C++ TM is finished.",
+      "reference": "No, we want to compare `tm_py` and `tm_cpp` in this test."
+    },
+    {
+      "prediction": "I think we should make this configurable as well.",
+      "reference": "Let me know if we'd rather have this as an env variable."
+    },
+    {
+      "prediction": "I think this should be `auth-delegator:%s:%s` instead of `auth-delegator:%s:%s:vault`",
+      "reference": "question: what is happening here?"
+    },
+    {
+      "prediction": "I don't think we need this method. We can just use `empty()` and add a `null` element to it.",
+      "reference": "\\[Checkstyle\\] ERROR: 'protected' modifier out of order with the JLS suggestions\\."
+    },
+    {
+      "prediction": "I think this should be `const getKmsProviders = localKey => EJSON.parse(process.env.CSFLE_KMS_PROVIDERS || 'NOT_PROVIDED');`",
+      "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'm not sure this is a good idea to remove the tick_format parameter. I think it's useful to be able to control the formatting of the tick labels.",
+      "reference": "So this parameter is now deprecated?"
+    },
+    {
+      "prediction": "I think this should be `#if defined(HAVE_RSEQ) && defined(X86)`",
+      "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": "nit: I think we can remove the empty line here.",
+      "reference": "I don't see a corresponding call to freeSelectorSet... am I missing something?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to redirect to the practice page instead of the dashboard page.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `db.statement_type` instead of `db.statement_type`.",
+      "reference": "Is this part of spec?"
+    },
+    {
+      "prediction": "I'm not sure I understand why we need to wait for `dr_app_started` here. If the app has already started, then `dr_app_started` will be set. If the app has not yet started, then `dr_app_started` will not be set, and we will wait forever. I don't think we need to wait for `dr_app_started` here. I think we just need to wait for `dynamo_initialized` to be set before we call `dynamo_thread_init`.",
+      "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": "I think it would be good to add a docstring to this function.",
+      "reference": "We can remove this function and use `lambda_function_name(..)` in this file instead."
+    },
+    {
+      "prediction": "I think this should be called `NewMetricsSender` or `NewMetricsTransport` or something like that.",
+      "reference": "nitpick: `ApplicationVersion` could be `AppVersion`. It's smaller but gives the same understanding for purposes of the field."
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "nit: maybe prefer `resp = &workflowservice.PollWorkflowTaskQueueResponse{}` over adding another function exit points especially for non-error cases?"
+    },
+    {
+      "prediction": "Nitpick: I think it would be better to use `client_options.merge(read: {mode: client_options[:read]}) if client_options[:read].is_a?(Symbol)` instead of `if client_options[:read].is_a?(Symbol); client_options = client_options.merge(read: {mode: client_options[:read]})`",
+      "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": "I think this should be `video` instead of `picture`",
+      "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 don't think you need these typedefs. You can just use `image_layout_map::ImageSubresourceLayoutMap::LayoutEntry` directly.",
+      "reference": "That doesn't make sense. LayoutEntry doesn't match the generator concept."
+    },
+    {
+      "prediction": "I think this should be `type: 'Load Balanced Web App'`",
+      "reference": "I think the customer visible strings should be \"Service\" instead of \"Svc\" to make it obvious to them"
+    },
+    {
+      "prediction": "I don't think you need to change this.",
+      "reference": "Explain why this property is necessary?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `_, _,` and just do `_, _, executions := action.ClassifyActions(blk.Actions)`",
+      "reference": "It should accept executions as the input"
+    },
+    {
+      "prediction": "I think it is better to use `assert batch_size == 1, 'show or out during test only support batch size of 1'` instead of `assert len(result) == 1, 'show or out during test only support batch size of 1'`",
+      "reference": "Is this limitation necessary?"
+    },
+    {
+      "prediction": "I think this should be `*SelfSignedIssuer` instead of `SelfSignedIssuer`.",
+      "reference": ":question: Should there be validation to ensure that only one of these attributes is used?"
+    },
+    {
+      "prediction": "I think this should be `use Laminas\\EventManager\\SharedEventManagerInterface;` instead of `use Laminas\\EventManager\\SharedEventManagerInterface as SharedEventManagerInterface;`",
+      "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'm not sure why this changed, but I think it should be reverted.",
+      "reference": "This would also change back to /var/www/html/docroot if we go that way."
+    },
+    {
+      "prediction": "I don't think this should be public.",
+      "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": "I don't think we need this anymore.",
+      "reference": "Thanks. Now that you've improved this we should just init and expose the StorageAPI on the node, like the other.s"
+    },
+    {
+      "prediction": "`nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next available rid.` -> `nextPrefix key will be key with next",
+      "reference": "it will be better if you can provide some examples"
+    },
+    {
+      "prediction": "I think this should be called `responseType` instead of `operationResponseType` to be consistent with `requestType` and `returnType`.",
+      "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": "This is not a cross-subpackage import, this is a subpackage import.",
+      "reference": "Import order has been violated here."
+    },
+    {
+      "prediction": "I'm not sure what this is for, but it looks like it should be removed.",
+      "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": "`use this key to set the customized labels for node. you can input customized labels like {\"key1\":\"value1\", \"key2\":\"value2\"}` -> `use this key to set the customized labels for node. 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": "Shouldn't this be `$this->allTabs->icons = $icons;`?",
+      "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) if isinstance(translatedMessage, str) else translatedMessage`",
+      "reference": "A docstring for this function please."
+    },
+    {
+      "prediction": "Should this be `IReadOnlyCollection<string>`?",
+      "reference": "heads up @JunTaoLuo"
+    },
+    {
+      "prediction": "`GetMethod()` can return `null` if the method is not public.",
+      "reference": ".GetMethod() cannot return null now, correct?"
+    },
+    {
+      "prediction": "I'm not sure why this was changed. I think it should be `volume_up`.",
+      "reference": "Same thing here, I think the proper fix is elsewhere."
+    },
+    {
+      "prediction": "I think this should be `if ((retcode < 0) && (errno == ENOENT) || (errno == EAGAIN))`",
+      "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 don't think this is the right place to do this. I think it should be done in the `__str__` method of the `PackageNode` class.",
+      "reference": "`hasattr` is kinda terrible; just append extra info in the subclass."
+    },
+    {
+      "prediction": "I think this should be `except ValueError as e:`",
+      "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": "Style/RedundantSelf: Redundant self detected.",
+      "reference": "This only updates the metadata and not the actual thing you are trying to solve for the docker hypervisor."
+    },
+    {
+      "prediction": "I don't think we should have a package for this. I think we should just have a `debug.go` file in the `internal` package.",
+      "reference": "should this package be internal? don't we want to use it e.g. in go-contrib?"
+    },
+    {
+      "prediction": "I think this should be `type Document interface{}`",
+      "reference": "Just curious, why did you make this change?"
+    },
+    {
+      "prediction": "I think we need to check if `prefix[column_label[0]]` is a string. If not, we need to raise an error.",
+      "reference": "Can you handle error cases such as `pd.get_dummies(pdf, prefix={\"A\": \"foo\"})`?"
+    },
+    {
+      "prediction": "I think this should be `os.Getenv(\"ECS_FTEST_FORCE_NET_HOST\") == \"true\"`",
+      "reference": "Where is this environment variable being set?"
+    },
+    {
+      "prediction": "`module_class` is a class, not an instance of a class, so I think we should use `type(module_class).__name__` instead of `module_class.__name__`.",
+      "reference": "`module_class ` is already a class and `type(module_class)` always returns `<class 'type'>`"
+    },
+    {
+      "prediction": "I don't think we need to convert the path to `UTF8String` here.",
+      "reference": "Nit: The lambda can be replaced with a method reference: `SparkOrcWriter::new` I think."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it would be better to have it in `bokeh/core/properties.py` where `is_number` is defined.",
+      "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": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "How do we manage updates to this view down the road?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "We can revert this to reduce the diff of the PR."
+    },
+    {
+      "prediction": "I think it would be better to use `assertThrows` instead of `expectThrows` here.",
+      "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 think we can remove this line as well.",
+      "reference": "not sure I follow why all this goes away?"
+    },
+    {
+      "prediction": "I think this should be `:insert-text {clipboard}`",
+      "reference": "Is this `{clipboard}` or `{primary}`, as the deprecation message for `:paste-primary` says?"
+    },
+    {
+      "prediction": "Typo: \"Sets the price of the OpenVPN service per minute.\" -> \"Sets the price of the OpenVPN service per GiB.\"",
+      "reference": "Human unreadable. IMHO user should input MYST value: 0.0006 @chompomonim, opinions?"
+    },
+    {
+      "prediction": "I don't understand why this was removed. It seems like it should still be here.",
+      "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": "I think this should be `playerTradeItem` and `tradePartnerTradeItem`",
+      "reference": "This is not consistent. Either rename the first to `tradePlayerItem` (which is ugly) or the second to `partnerTradeItem` (:+1:)"
+    },
+    {
+      "prediction": "I think this should be `go get -u github.com/xeipuuv/gojsonschema/cmd/gojsonschema`",
+      "reference": "is there a reason we don't want to gx this dependency?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the database query, so that we don't have to do it for every host.",
+      "reference": "I'd suggest to add \"ATI \" and \"Intel(R) \" to this list"
+    },
+    {
+      "prediction": "I think this should be `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 don't think we need the `*` at the end of the regular expression.",
+      "reference": "shouldn't this work? nodesRegex = regexp.MustCompile(api.SpecNodes + \"=(('[0-9A-Za-z,_-]+')|([0-9A-Za-z_-]+)),?\")"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it makes sense to do this in the `before_validation` callback instead of the `before_create` callback. The reason is that the `before_validation` callback is called before the `before_create` callback, so if we do this in the `before_validation` callback, we can be sure that the `before_validation` callback will be called before the `before_create` callback, so we can be sure that the `before_validation` callback will be called before the `before_create` callback, so we can be sure that the `before_validation` callback will be called before the `before_create` callback, so we can be sure that the `before_validation` callback will be called before the `before_create` callback, so we can be sure that the `before_validation` callback will be called before the `before_create` callback, so we can be sure that the `before_validation` callback will be called before the `before_create` callback, so we can be sure that the `before_validation` callback will be called before the `before_create` callback, so we can be sure that the `before_validation` callback will be called before the `before_create` callback, so",
+      "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 don't think we need to log here. We can just ignore the value.",
+      "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 `$( [ -d \"/cygdrive/c/Program Files (x86)\" ] && echo \"/cygdrive/c/Program Files (x86)/Puppet Labs/Puppet Enterprise/bin\" || 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 if this is the right place to put this, but I think it should be in the same menu as \"Move to Folder\" and \"Move to Trash\".",
+      "reference": "It's not async"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the temporality. I think it should be set on the `OtlpExporterOptions` instead of the `MetricReaderOptions`.",
+      "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": "I think this should be `everyKey` to be consistent with the rest of the codebase.",
+      "reference": "Maybe would be better to rename it to \"onEveryKey\" or \"fireOnEveryKey\"."
+    },
+    {
+      "prediction": "I don't think this is the right place to call this method. This method is called from `FilePage` and `FilePage` is not aware of `FileDownloadHelper`. I think this method should be moved to `FileDownloadHelper` and called from `FileDownloadHelper` instead of `FilePage`.",
+      "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 don't think we should throw an exception if the method config is not found. I think we should just return `null`.",
+      "reference": "is `fullName` still required as a parameter?"
+    },
+    {
+      "prediction": "I think this should be `LOG.fine(\"Bound legacy RC support\");`",
+      "reference": "This informational message is important to users. Please leave."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should default this to `true` for now. We can always add a flag in the future to allow the user to opt-out.",
+      "reference": "Why do we want to remove this one? Are we planning to substitute it?"
+    },
+    {
+      "prediction": "nit: I think we can remove the nolint now that WithDetails takes v2.",
+      "reference": "We may create an issue to track this so we don't forget?"
+    },
+    {
+      "prediction": "Trailing whitespace detected.",
+      "reference": "This file has the executable bit set."
+    },
+    {
+      "prediction": "I don't think we need to log this.",
+      "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": "I don't think you need to unlock the mutex here.",
+      "reference": "This lock cannot be released here, as it is protecting m_RTPSParticipants and the participant reference."
+    },
+    {
+      "prediction": "I think we should move this to `Query.prototype.cast` so that we don't have to call it in `Query.prototype.find` and `Query.prototype.findOne`",
+      "reference": "we'll need this in `_findAndModify` too"
+    },
+    {
+      "prediction": "I don't think this is necessary. If `bundle.getString(DB_NAME)` returns `null`, then `bundle.getString(DB_NAME, DBOpenHelper.DEFAULT_DB_NAME)` will return `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'm not sure this is the right place to do this. I think it should be done in `event_textChange` of `WordDocumentNode` instead.",
+      "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.getProductConfig().getPackageName() + \".\" + context.getNamer().getLocalPackageName() + \";;\";`",
+      "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'm not sure if this is the right thing to do, but it seems to be the only way to get this to work on Windows 10 1903.",
+      "reference": "Can't you just use runnvda.bat here?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "these definitions of UNUSED aren't the same as the one you imported, so callers should be fixed (if any)"
+    },
+    {
+      "prediction": "Should this be `ECSIPAMPluginName` or `ECSIPAMPluginType`?",
+      "reference": "is type a free string too?"
+    },
+    {
+      "prediction": "`WrapFunction(model_func)` should be `model = WrapFunction(model_func)`",
+      "reference": "here input could be a normal function or an instance of torch.nn.Module."
+    },
+    {
+      "prediction": "I don't think this is necessary. We already have `nix-env --version` and `nix-instantiate --version`.",
+      "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": "Why is this removed?",
+      "reference": "Class import is therefore redundant I guess :)"
+    },
+    {
+      "prediction": "I don't think this is a warning, it's an error.",
+      "reference": "If parsing fails, don't act as no IP specified. Error out."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "... here the semicolon wanted to run away from the enum ;)"
+    },
+    {
+      "prediction": "I think this should be `replicationConfig.add(\"replication\", replicationConfig);`",
+      "reference": "Note that this is in the context of SolrCloud, so \"secondary\" doesn't apply and should be instead follower"
+    },
+    {
+      "prediction": "Line is too long. [81/80]<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.",
+      "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": "nit: extra space before {",
+      "reference": "Need to ensure this is set to TRU for LOB datatype access too not just for hive ."
+    },
+    {
+      "prediction": "`numeric_only` is not used by this implementation, but passed down by stats functions.",
+      "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": "I think it would be good to add a test for this case.",
+      "reference": "Don't really need this since the rest of the method will no-op but it's also safe to return early."
+    },
+    {
+      "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 don't think this is needed.",
+      "reference": "this file can be removed"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Do we have test cases that name is null?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "in a separate PR, I'll check for a table that is empty/all-null and skip this function call completely."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "We really, really want to catch _everything_ here. It is safe to leave `azkabanEventReporter` set to null."
+    },
+    {
+      "prediction": "I think we should use `parentDom.lastChild === newDom` instead of `parentDom.lastChild !== newDom`.",
+      "reference": "Can't we add the check here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the request fails, the connection is closed, so the next request will fail as well. I think it would be better to close the connection only if the request is successful.",
+      "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 we should call this `Secrets` instead of `Encryption` since it's not really about encryption.",
+      "reference": "Tell me if you have a better field name for this."
+    },
+    {
+      "prediction": "Shouldn't this be `Framework == TargetFramework.NET6_0 || Framework == TargetFramework.NET5_0 || Framework == TargetFramework.NETCOREAPP3_1 || Framework == TargetFramework.NETCOREAPP3_0 || Framework == TargetFramework.NETCOREAPP2_1 || Framework == TargetFramework.NETCOREAPP2_0 || Framework == TargetFramework.NETCOREAPP1_1 || Framework == TargetFramework.NETCOREAPP1_0 || Framework == TargetFramework.NET48 || Framework == TargetFramework.NET472 || Framework == TargetFramework.NET471 || Framework == TargetFramework.NET47 || Framework == TargetFramework.NET462 || Framework == TargetFramework.NET461 || Framework == TargetFramework.NET46 || Framework == TargetFramework.NET452 || Framework == TargetFramework.NET451 || Framework == TargetFramework.NET45 || Framework == TargetFramework.NET40 || Framework == TargetFramework.NET35`?",
+      "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 this should be `const auto ai = ...`",
+      "reference": "This is just `mol->atomNeightors()`, is it?"
+    },
+    {
+      "prediction": "Why do we need `task::clustering` here?",
+      "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 this should be `RenderPassCreateAttachmentMismatchingLayoutsColor`",
+      "reference": "@cnorthrop -- here is a test with a name change. Is this going to affect your internal CI?"
+    },
+    {
+      "prediction": "I think we can remove this constant. It's only used in one place.",
+      "reference": "Nit: `Headless` is one word, not two, and so doesn't need camelcasing in this way."
+    },
+    {
+      "prediction": "I think this should be `None` by default, and we should set it to the number of available CPUs in the `__init__` method of the `core` class.",
+      "reference": "I think this will cause a warning, can you set the default to zero (`0`)?"
+    },
+    {
+      "prediction": "I think it's better to use `github.com/mysterium/node/identity/identity` instead of `github.com/mysterium/node/identity`",
+      "reference": "confusing naming. Identity from \"id\" package :/ why we need alias here? (and in other imports)"
+    },
+    {
+      "prediction": "I think this should be `SettingValue(typ.List(none_ok=True), [])`",
+      "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": "I'm not sure if this is the right thing to do, but it seems to be the only way to make this test pass.",
+      "reference": "Perhaps use the new `.keys()` method here instead?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we need to remove the `diff.WithIgnoreAddingMapKeys()` option from the `provider.DiffList` call. The reason is that the `provider.DiffList` call is used for both `pulumi up` and `pulumi preview`. In the case of `pulumi up`, we want to ignore adding map keys, but in the case of `pulumi preview`, we don't want to ignore adding map keys.",
+      "reference": "We don't need this option because plan-preview is comparing between 2 commits, not using the live state."
+    },
+    {
+      "prediction": "Metrics/BlockLength: Block has too many lines. [31/25]",
+      "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 this should be called `Metrics` to be consistent with the rest of the codebase.",
+      "reference": "So `Status` returns metrics?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "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": "Why do we need this?",
+      "reference": "Why would this be called to copy a non-Date object?"
+    },
+    {
+      "prediction": "I think this should be `unix.NLA_ALIGN(unix.NLA_HDRLEN + 1)`",
+      "reference": "This alignment issue had not appear because no multiple `Boolmsg`s had been serialized at once."
+    },
+    {
+      "prediction": "I think 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": "Metrics/LineLength: Line is too long. [123/120]",
+      "reference": "Replaced this method with a ruby idiomatic setter."
+    },
+    {
+      "prediction": "I think you meant to write \"This program is free software; you can redistribute it and/or modify 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": "I'm not sure if this is the best way to do it, but it seems to work for me.",
+      "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": "Metrics/LineLength: Line is too long. [118/100]",
+      "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 `tunnel->destroy(tunnel, stream->tunnel);`",
+      "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": "Please remove the `\\n` at the end of the string.",
+      "reference": "Did you mean to remove the periods here while newly adding to other locations?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "the check can now be `if newLen < 0`"
+    },
+    {
+      "prediction": "nit: I think we can remove this import.",
+      "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'm not sure if this is the best way to do it, but it seems to work.",
+      "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'm not sure if this is the best way to do it, but it seems to work for me.",
+      "reference": "form => from"
+    },
+    {
+      "prediction": "I'm not sure if we want to do this in 7.0.0. I think it would be better to wait for 8.0.0.",
+      "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": "nit: I think it's better to use `GapicProductConfig.create(model, configProto, \"google.example.myproto.v1\", TargetLanguage.GO)` instead of `GapicProductConfig.create(model, configProto, \"google.example.myproto.v1\", TargetLanguage.GO)`.",
+      "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 don't think we need to create a new `String` here. We can just use `word.subSequence(0, word.offset + breakPos)`.",
+      "reference": "Include all compounds parts from the very beginning for the case check to work later"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we need to add the `unix://` prefix here.",
+      "reference": "if the user specifies a url scheme on the cli this is going to cause problems, no?"
+    },
+    {
+      "prediction": "I think this should be 0.75",
+      "reference": "How about that `scaleFactor`?"
+    },
+    {
+      "prediction": "`$result` should be typed as `int|float`.",
+      "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 `getQuantifiedProductsIndexedByItemIds` or `getQuantifiedProductsIndexedByItemIds()`",
+      "reference": "I'm unfortunately unable to review whether you've changed everything that used to use cartIds"
+    },
+    {
+      "prediction": "This import is not needed anymore.",
+      "reference": "Please don't touch unrelated files to keep the git history clean"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Per our style guide, I don't think we indent `private` keyword. Would you mind fixing that?"
+    },
+    {
+      "prediction": "why do we need this line?",
+      "reference": "Will this change the input `heatmap`? Is this behavior expected or not?"
+    },
+    {
+      "prediction": "I think this should be `stringArgV` instead of `chars`.",
+      "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": "Please remove this file.",
+      "reference": "Need copyright/license info."
+    },
+    {
+      "prediction": "why remove this import?",
+      "reference": "Is this `info log` necessary? If it is not useful, can you consider deleting it?"
+    },
+    {
+      "prediction": "I think it would be good to add a test for the case where the header is correct but the rest of the file is not base64 encoded.",
+      "reference": "Unrelated change, but seems to be the correct purpose of the test."
+    },
+    {
+      "prediction": "I think this should be `id={this.id}` instead of `id={'input-' + this.id}`",
+      "reference": "Should it be `'uppy-input-`?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "what's the reasoning to keep these always true or dead-code blocks?"
+    },
+    {
+      "prediction": "I think this should be `$('.widget-content').removeClass('hide-zoom');` instead of `$('.widget-content').addClass('hide-zoom');`",
+      "reference": "It would be a good practice to do `.off('click', '.tab-buttons > div').on('click', '.tab-buttons > div', function() {`"
+    },
+    {
+      "prediction": "I think this should be `false` by default, and we should enable it if we discover the Gateway API.",
+      "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 moved to the `configureLogging` method, which is called before `instantiateSignatureAlgorithmFactory` and `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 don't think this is needed.",
+      "reference": "Why remove this?"
+    },
+    {
+      "prediction": "Nit: I think it would be better to call this `setIeDriverService` or `setIeDriverServiceBuilder` to make it clear that this is the service for the driver, not the service for managing the geckodriver process.",
+      "reference": "`this.ieService_` should be initialized to null in the constructor."
+    },
+    {
+      "prediction": "nit: I think it would be better to use `std::unique_ptr<ebpf::BPF> bpf = std::unique_ptr<ebpf::BPF>(new ebpf::BPF());`",
+      "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: I think this should be `cert-manager/pkg/api/util` rather than `jetstack/cert-manager/pkg/api/util`",
+      "reference": "nit: move to last block"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to use the logger from `pkg/common/logger` instead of the one from `pkg/agent/logger`. The reason is that the logger from `pkg/agent/logger` is a wrapper around the logger from `pkg/common/logger` and it adds some extra functionality that we don't need in the agent.",
+      "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 think this should be `System.Data.IsolationLevel? IsolationLevel { get; set; }` instead of `System.Data.IsolationLevel? ? IsolationLevel { get; set; }`",
+      "reference": "Why is IsolationLevel fully qualified? System.Data is already in scope? What do I miss?"
+    },
+    {
+      "prediction": "Why do we need to remove `Serialize`?",
+      "reference": "do we still need SanityCheck()? seems they all return nil now"
+    },
+    {
+      "prediction": "I think it's better to keep the description complete sentence.",
+      "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 think this should be `https://#{ENV.fetch(\"APP_DOMAIN\")}/admin/video/#{video.id}/edit`",
+      "reference": "Line is too long. [100/80]"
+    },
+    {
+      "prediction": "I think this should be `p.V(\"no parent snapshot found, will read all files\\n\")`",
+      "reference": "I suggest \"no parent snapshot found, will read all data\\n\"."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "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 correct. If the string contains a single quote, you need to escape it with a backslash.",
+      "reference": "It should be return \"''\"; I will fix it"
+    },
+    {
+      "prediction": "I don't understand why this change is needed. The `m` variable is declared as `Map<NameDeclaration, List<NameOccurrence>> m = s.getDeclarations();` so it is already a `Map` of `NameDeclaration` to `List<NameOccurrence>`, so the `entry.getKey()` and `m.get(entry.getKey())` should be the same as `m.keySet().iterator().next()` and `m.get(m.keySet().iterator().next())`.",
+      "reference": "This is wrong, should be `entry.getValue()`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix, but it seems to be the only way to get the patch to work.",
+      "reference": "nit: why can't we combine the two add operations into one so that we add `{\"nameCertificates\": []}` to `/spec/servingCerts` ?"
+    },
+    {
+      "prediction": "nit: extra blank line",
+      "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 don't think this should be passed in as a parameter. It should be a member of the TheatrePlugin class.",
+      "reference": "why does this suddenly need a config ref"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a separate test file that loads the adapter and then runs the tests that need it. That way we don't have to load it for all tests.",
+      "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 don't think we need to exclude all of these. We can just exclude the ones that don't have a `__init__.py` file.",
+      "reference": "Those functions are not currently wrapped, so I don't think it is necessary to exclude them (at least for now)."
+    },
+    {
+      "prediction": "I think this should be `var afterFunc = func(c clock.Clock, d time.Duration, f func()) stoppable {`",
+      "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": "I think this should be `int(255.0 * col.get<0>()), int(255.0 * col.get<1>()), int(255.0 * col.get<2>())`",
+      "reference": "I'm assuming that MolDraw2Qt drops the alpha channel?"
+    },
+    {
+      "prediction": "I don't think we need to test `int` here.",
+      "reference": "the code looks good to me, but I just wonder if it is a real use case: (de)serializing a single integer."
+    },
+    {
+      "prediction": "Please remove this print statement.",
+      "reference": "Pls remove the print statements. Also, you do not need the if statement."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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 disable sniffing?",
+      "reference": "Should this be instead a configurable option?"
+    },
+    {
+      "prediction": "I think this should be `if snapshot, ok := w.snapshots.get(cd.SourceSnapshot); ok { cd.SourceSnapshot = snapshot.link } else { return fmt.Errorf(\"source snapshot %q not found\", cd.SourceSnapshot) }`",
+      "reference": "unit test needs an update?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "I realised that this is not used anywhere"
+    },
+    {
+      "prediction": "nit: I think we can remove the `default` case here.",
+      "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 place to call `sendZlp()`. I think it should be called in the `else` branch of the `if` statement on line 1475.",
+      "reference": "This line appears to have caused the regression. What is it supposed to be doing?"
+    },
+    {
+      "prediction": "I don't think this is correct. If `attr_value['string']` is `None` then `attr_value['string'] == value` will be `False` even if `value` is `None`.",
+      "reference": "# `attr_value['string'] is not None` probably not required as `attr_value['string'] == value` check is already being done"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the `exporter.CollectMetric = metricReader.Collect;` so that the `PrometheusExporterMetricsHttpServer` can call `exporter.CollectMetric()` to collect metrics.",
+      "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 think it would be better to have a timeout on the `SubmitRequest` method, so that we don't have to wait for 5000ms in case the server doesn't respond.",
+      "reference": "As for Owin, I think we should exclude this URL in `IsNotServerLifeCheck` too?"
+    },
+    {
+      "prediction": "I think this should be `IHttpClientFactory` instead of `IHttpClientFactoryExporterOptions`.",
+      "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": "`\u06a9\u0686\u06be \u0633\u06cc\u06a9\u0646\u0688 \u0645\u06cc\u06ba` should be `:seconds \u0633\u06cc\u06a9\u0646\u0688 \u0645\u06cc\u06ba`",
+      "reference": "here is `:seconds` missing"
+    },
+    {
+      "prediction": "I think this should be `func2` instead of `func3`.",
+      "reference": "change of name is not needed"
+    },
+    {
+      "prediction": "Why did you remove these lines?",
+      "reference": "Hm .. so this case doesn't work anymore?"
+    },
+    {
+      "prediction": "I think we should use `from PyQt5.QtCore import pyqtSignal, pyqtSlot, QObject, QProcess, QFileSystemWatcher` here.",
+      "reference": "Please indent this so it lines up with the `(`"
+    },
+    {
+      "prediction": "Can we remove this line?",
+      "reference": "We should still register the script. Just not enqueue it."
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "reference": "I think this can stay as a method. And we can write our own attribute_change(name) as required."
+    },
+    {
+      "prediction": "I think this should be `ItemTemplateSelector.GetItemLayoutId(viewType)`",
+      "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 `__repr__` instead of `__str__`. `__str__` should return a human-readable string, and `__repr__` should return a string that can be used to recreate the object.",
+      "reference": "Can you change this to use `.format()`?"
+    },
+    {
+      "prediction": "I think this should be `if dropEncap && (r.Config.DropVXLANPacketsFromWorkloads || r.Config.DropIPIPPacketsFromWorkloads) {`",
+      "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 think this should be `GeneratePoStRequest` and `GeneratePoStResponse` instead of `GeneratePoSTRequest` and `GeneratePoSTResponse`.",
+      "reference": "new casing is less ELiTE"
+    },
+    {
+      "prediction": "I think this should be `GLuint wlr_renderer_renderbuffer_from_image(struct wlr_renderer *r, EGLImageKHR image) { if (!r->impl->renderbuffer_from_image) { return 0; } return r->impl->renderbuffer_from_image(image); }`",
+      "reference": "This leaks EGL/GL implementation details into the generic renderer interface."
+    },
+    {
+      "prediction": "I think this should be `\\\\`sources\\\\` and an array as a value.` instead of `\\\\`sources\\\\` and an array as a value.`",
+      "reference": "Maybe add *key* i.e. `\"The \\\\`deploy.on\\\\` key should be a hash (dictionary).`"
+    },
+    {
+      "prediction": "I don't think this is correct. The code should be the same as the id.",
+      "reference": "its be good also change $code na $id, actual name is misleading and that was probably actual error generator"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "I changed the name of the key, to make deployment easier."
+    },
+    {
+      "prediction": "I think this should be `ECS_ENABLE_SPOT_INSTANCE_DRAINING` instead of `ECS_SPOT_INSTANCE_DRAINING_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": "This is not needed. The `__future__` imports are only needed in Python 2, and this file is only used in Python 3.",
+      "reference": "Do you think we might need this anywhere else in HoloViews?"
+    },
+    {
+      "prediction": "I think this should be `session.TranscodingInfo.IsVideoDirect && session.TranscodingInfo.IsAudioDirect && session.PlayState.PlayMethod === 'Transcode'`",
+      "reference": "@MrTimscampi don't we want to remove this term entirely? Might as well do it now if that's the case."
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "reference": "I have followed the convention of `cp`"
+    },
+    {
+      "prediction": "`?string` should be `string|null`",
+      "reference": "The docs say it's `false|null`"
+    },
+    {
+      "prediction": "The default is 5s. -> The default value is 5s.",
+      "reference": "Thinking on it more, I actually really like your idea to rename `type` to `factor`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do here, but it seems to be the only way to get rid of the crash.",
+      "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'm not sure if this should be `r\"\"\"` or just `\"\"\"`.",
+      "reference": "Feel free to add your own name while at it."
+    },
+    {
+      "prediction": "I think this should be `--signer` instead of `--signer-addr`",
+      "reference": "`signer` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I don't think we need to pass `opts` here, we can just pass `localize` directly.",
+      "reference": "shouldn't this be `&& opts[:localize]` (i.e. both nil and false skip localization?)"
+    },
+    {
+      "prediction": "I think we should avoid importing `pickle` in this file.",
+      "reference": "This can also be removed."
+    },
+    {
+      "prediction": "I think this should be `const char *key, const char *ns, const char *target);`",
+      "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 `always` instead of `when_searching`.",
+      "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": "This should be `os.path.realpath(fs_path).decode(sys.getfilesystemencoding())`",
+      "reference": ".decode(sys.getfilesystemencoding()) will be better. Although I think rrdtool should accept bytes.."
+    },
+    {
+      "prediction": "Should this be `IHttpHeadersHandler<T>` where `T` is the type of the headers?",
+      "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": "`targetsForLevel` should be renamed to `targets`",
+      "reference": "Resharper annotations, always +1"
+    },
+    {
+      "prediction": "Why hide this flag?",
+      "reference": "same here, CryptoSm2 won't be changed once command is compiled"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to call `loadLoginPage()`. I think it should be called in `onKeyChainAliasComplete()` instead.",
+      "reference": "Loading the login page after the certificate has been loaded. Views can only be loaded from UI thread."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's better to keep the line number in the log.",
+      "reference": "Technically it doesn't belong here; let me remove it."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "reference": "remember to update the hash once the upstream PR is merged"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this logic. I think it should be in the `PushNotificationDecryptor` class.",
+      "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 I understand why this is needed. If `params` is `nil`, then `params.(adt_spec.EmptyValue)` will panic. If `params` is not `nil`, then `params.(adt_spec.EmptyValue)` will return `false`.",
+      "reference": "FYI This will go the other way, with nil being the correct value for \"no params\""
+    },
+    {
+      "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's a more efficient way to do this.",
+      "reference": "Is this necessary? Are there duplicates in your LibKeyIndex?"
+    },
+    {
+      "prediction": "I think this should be `m.ctrl.RecordCallWithMethodType(m, \"Next\", reflect.TypeOf((*MockActionIterator)(nil).Next)).Return(0, 0)`",
+      "reference": "Why will the gomock files be regenerated? It seems to be irrelevant"
+    },
+    {
+      "prediction": "Why is this a buffered channel?",
+      "reference": "I'm very curious about why using buffered-channel. Is there something wrong to use an unbuffered channel with zero capacity?"
+    },
+    {
+      "prediction": "I think this should be `from .helpers import LuigiTestCase, RunOnceTask`",
+      "reference": "It seems like you accidentally pulled some unrelated changes to util_test into this."
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "do not toucm NDM please - there will be lots of conflicts"
+    },
+    {
+      "prediction": "`v` should be `val` here.",
+      "reference": "This looks legitimate - Would cause a classCastException. But have we ever seen it in the wild?"
+    },
+    {
+      "prediction": "I think this should be `True` by default.",
+      "reference": "This looks out of scope for this particular PR."
+    },
+    {
+      "prediction": "I think this should be `leafDefaultFirstPingInterval` instead of `leafFirstPingInterval`.",
+      "reference": "Don't need () if only one."
+    },
+    {
+      "prediction": "I think it would be better to use `mailPort` instead of `mailHost` to avoid confusion.",
+      "reference": "Why not have mailPort as int since your are already parsing while calling t.connect ?"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "In the future, do not remove these trailing commas as they have the purpose of making append-diffs easier to read. :)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `SolrDefault::getLccn()` method.",
+      "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 a `map[string]struct{}` instead of a `map[string]bool`.",
+      "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 think we should call this `API` instead of `Server` to avoid confusion with the `Server` service.",
+      "reference": "This is very unclear what Server means. We need a better name, maybe ServerExtension?"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but it seems like it should still be there.",
+      "reference": "I wonder if we should trigger a PHP error and ensure it ends up in the log?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "Hmmm not sure about this. Is it fine @ar2rsawseen?"
+    },
+    {
+      "prediction": "I think this should be `void run();`",
+      "reference": "Why do you prefer `go` over `run`?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Unintentional, please remove"
+    },
+    {
+      "prediction": "I think this should be `where(deactivated_on: nil)` instead of `where.not(deactivated_on: nil)`",
+      "reference": "What was the reason behind moving this?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I couldn't figure out how to do it otherwise.",
+      "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": "This should be `@since 3.0.0`",
+      "reference": "`Indicates if this Realm has been closed.`?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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'm not sure if this is the right thing to do, but I think it's the right thing to do. We don't need to load the entire state tree, we just need the state root.",
+      "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 don't think we need to add `where T : IComparable<T>` here. We already have `[GenericTypeArguments(typeof(IntClass))]` which is a custom reference type that implements `IComparable<T>`, so I don't think we need to add `where T : IComparable<T>` here.",
+      "reference": "@billwert I have kept the `1000` InvocationsPerIteration, but do note this is very small for the small types."
+    },
+    {
+      "prediction": "Please don't remove empty lines.",
+      "reference": "Unrelated whitespace change."
+    },
+    {
+      "prediction": "I don't think this is necessary. `transaction_p` is a pointer to a `transaction` object, not a pointer to a pointer to a `transaction` object. The `transaction_p->open_for_write` check should be sufficient.",
+      "reference": "Slight formatting issue here. There should be a space after the asterisk."
+    },
+    {
+      "prediction": "`node` is not used in this function, so it can be removed.",
+      "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's better to use `private static final List<Throwable> errorList = Collections.synchronizedList(new ArrayList<Throwable>());` instead of `private static final List<Throwable> errorList = new CopyOnWriteArrayList<Throwable>();`",
+      "reference": "this is a duplicate file of the one in demo-schema. please remove this file."
+    },
+    {
+      "prediction": "I think this should be `utils.Eth1L1FeeWalletAddressFlag`",
+      "reference": "Nit: unified names between geth and contracts"
+    },
+    {
+      "prediction": "I think this should be `config.NumThreads = 5`",
+      "reference": "Necessary? Could you just use DefaultBuildState here?"
+    },
+    {
+      "prediction": "This change is not related to this PR. Can you please revert it?",
+      "reference": "linked hashmap retains order"
+    },
+    {
+      "prediction": "I think this should be `extern const struct wlr_drm_interface *const iface_atomic;` and `extern const struct wlr_drm_interface *const iface_legacy;`",
+      "reference": "And the old names for these variables made more sense imo."
+    },
+    {
+      "prediction": "I think this should be `<=` instead of `>=`",
+      "reference": "Are you sure this change is correct?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix. I don't know what the problem is with the ICO file, and I don't know if this is the right way to fix it. I'm just trying to get the tests to pass.",
+      "reference": "The previously linked bug does not apply anymore? If so, this is LGTM!"
+    },
+    {
+      "prediction": "I think this should be `TestFailOver` instead of `Test_FailOver`",
+      "reference": "Why are we renaming all of the tests?"
+    },
+    {
+      "prediction": "I think this should be `strings.TrimSpace(nameserver)` instead of `nameserver = strings.TrimSpace(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'm not sure this is the right thing to do here. I think it would be better to just return `res` and let the caller handle the exception.",
+      "reference": "Why not also replace this one with calls to `GetProp<ROMol,T>`?"
+    },
+    {
+      "prediction": "Remove this empty line.",
+      "reference": "Still need to remove trailing spaces on this line"
+    },
+    {
+      "prediction": "I think this interface should be moved to `net.sourceforge.pmd.lang.apex.ast` package.",
+      "reference": "Please use the `@Experimental` annotation on this"
+    },
+    {
+      "prediction": "nit: I think this should be `if err != nil && err != io.EOF {` instead of `if err != nil && err != io.EOF {`",
+      "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 `metricNamespaceOverride`.",
+      "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 this should be `methodInfo.responsePageTokenName()` instead of `methodInfo.requestPageTokenName()`.",
+      "reference": "Just checking: no need for case manipulation?"
+    },
+    {
+      "prediction": "I think we should keep the name `apiConfigs` for consistency with the rest of the codebase.",
+      "reference": "Same as above, not that descriptive"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what else to do either. We don't want to call `special_ibl_xfer_tgt` twice, but we also don't want to call it once and store the result in a local variable and then use that local variable twice.",
+      "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 don't think this is the right place to cancel the thread. I think it should be done in `flb_stop()`.",
+      "reference": "Should we need to invoke pthread_cancel ?"
+    },
+    {
+      "prediction": "I think this should be `System.getProperty(ZMSConsts.ZMS_PROP_USER_DOMAIN, ZMSConsts.USER_DOMAIN) + \".\" + name + \":domain\"`",
+      "reference": "this is auto generated file so no changes are allowed here"
+    },
+    {
+      "prediction": "I think this should be `:not([type=\"text\"])` instead of `:not(textarea)`.",
+      "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 don't think this is the right way to do it. I think it should be `config[:perl] == 5.1 ? \"5.10\" : config[:perl] == 5.2 ? \"5.20\" : config[:perl] == 5.22 ? \"5.22\" : config[:perl] == 5.24 ? \"5.24\" : config[:perl] == 5.26 ? \"5.26\" : config[:perl] == 5.28 ? \"5.28\" : config[:perl] == 5.30 ? \"5.30\" : config[:perl] == 5.32 ? \"5.32\" : config[:perl] == 5.34 ? \"5.34\" : config[:perl] == 5.36 ? \"5.36\" : config[:perl] == 5.38 ? \"5.38\" : config[:perl] == 5.40 ? \"5.40\" : config[:perl] == 5.42 ? \"5.42\" : config[:perl] ==",
+      "reference": "can you please make this multi line, this version is hard to read."
+    },
+    {
+      "prediction": "`goog.isNumber(xyRatio.x)` is equivalent to `typeof xyRatio.x === 'number'`.",
+      "reference": "For consistency, it's best to either always use strict equality (`===`) or loose equality (`==`) in `typeof` checks, and to not mix them."
+    },
+    {
+      "prediction": "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": "Work Orders would never have only one approver, right?"
+    },
+    {
+      "prediction": "I think it would be nice to have a `fromCode(int code)` method that throws an `IllegalArgumentException` if the code doesn't match any of the enum values.",
+      "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": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "This is the only \"extra\" change here."
+    },
+    {
+      "prediction": "I think this should be `false` by default.",
+      "reference": "Please TODO and link to an issue for changing this."
+    },
+    {
+      "prediction": "I think we can remove the `eslint-disable` here.",
+      "reference": "Can we make the script output trailing commas?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work for now.",
+      "reference": "This file should get a file header"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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": "Why do we need to return 0 here?",
+      "reference": "we need to pay attention not to use this flag everywhere."
+    },
+    {
+      "prediction": "Why 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 think we should keep the `WithStamp(dataItem.Stamp)` here.",
+      "reference": "wait, are we not mising `WithStamp` here?"
+    },
+    {
+      "prediction": "I think this should be `stages.Finish` instead of `stages.TxPool`.",
+      "reference": "oh. didn't know we store this stage progress."
+    },
+    {
+      "prediction": "Shouldn't this be `'\\uE000-\\uF8FF'` instead of `'\\uE000-\\uF8FF'`?",
+      "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": "Shouldn't this be `new Type\\Atomic\\TLiteralInt($atomic_types['array']->count)`?",
+      "reference": "This change is un-tested and requires testing"
+    },
+    {
+      "prediction": "This should be `ipywidgets` and `vscode`",
+      "reference": "Should the docstring mention the vscode option?"
+    },
+    {
+      "prediction": "I don't think this is needed. The gapic config name is always `gapic.yaml` and the short name is always the package name.",
+      "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 should keep the import as `github.com/docopt/docopt-go` so that it's clear where it comes from.",
+      "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 `where T : unmanaged`",
+      "reference": "Should this be `ISocketOutput<T> where T : struct`?"
+    },
+    {
+      "prediction": "I don't think we need this method.",
+      "reference": "In which case will this actually return 0? If there is no snapshot, then there is no `Snapshot` object, right?"
+    },
+    {
+      "prediction": "I think it's better to use `isinstance(dataset.CLASSES, str)` instead of `type(dataset.CLASSES) is not str`",
+      "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'm not sure why this was pinned to 2.8, but I don't think we need to pin it to a specific version.",
+      "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 don't think we need this anymore.",
+      "reference": "could you please explain a bit where these numbers come from? for example: why is there no `swarm.EncryptedSectionSize`?"
+    },
+    {
+      "prediction": "I think this should be `return r.r.Attrs().ContentType`",
+      "reference": "s/object content/blob object/ (for consistency with the `Size` docs)"
+    },
+    {
+      "prediction": "I don't think this is correct. If you have `useJUnitPlatform()` in your `build.gradle` then you can use JUnit 4 and JUnit 5 in the same test suite.",
+      "reference": "Rather than deleting this entirely, could we just emit it as a `warn` or `info` log line?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I don't think we should change the format of the lock key. I think it would be better to add a new field to the `acmeClient` struct and use that in the `lockKey` function.",
+      "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 in the `finally` block, so that it is always called, even if an exception is raised in the `try` block.",
+      "reference": "Thanks for this... is this the only place where the rollback is needed? Are there others?"
+    },
+    {
+      "prediction": "I think this should be `&crypto.Signature{Type: crypto.SigTypeSecp256k1, Data: []byte{}}` instead of `crypto.Signature{Type: crypto.SigTypeSecp256k1, Data: []byte{}}`",
+      "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 we should keep the `err != nil` check here, and return `err` instead of `perr`.",
+      "reference": "This was definitely a subtle bug right here."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it works for now. I think we should be able to do this with a single API call, but I couldn't figure out how to do it.",
+      "reference": "If this method gets all the emails, it should be named appropriately. **all_emails** or just **emails**."
+    },
+    {
+      "prediction": "Why is this not random?",
+      "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` by default.",
+      "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": "Align the parameters of a method call if they span more than one line.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "what if a purchaser is also an approver? or would that not happen?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. It should be done in `estimate_fee` instead.",
+      "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 don't think we need this method.",
+      "reference": "sorry, I wasn't familiar with this TopLevelCondition... What it would be in this case? The AuditsLogReady condition?"
+    },
+    {
+      "prediction": "I think we can get rid of the `cache` field and just use `execute.NewTableBuilderCache()` in the `NewDiffTransformation` function.",
+      "reference": "Where does the `alloc` field get set?"
+    },
+    {
+      "prediction": "I think we can remove the `countQuery` variable and just use `opts.selectAs` in the `runQueryPostgres` function.",
+      "reference": "looks expensive to do it on each query"
+    },
+    {
+      "prediction": "I think this should be `nyc_taxi.csv` instead of `nycTaxi.csv`.",
+      "reference": "@rhyolight -- but weren't we supposed to be standardizing on using underscores in file names (versus camelcase)?"
+    },
+    {
+      "prediction": "I'm not sure about this change. `in_memory` is an array, so `count` and `length` should be the same.",
+      "reference": "Can this simply call `in_memory.empty?` ?"
+    },
+    {
+      "prediction": "I think this should be `if app.OmitSettingsPhp && app.OmitSettingsYaml`",
+      "reference": "The styling/wording here probably needs some thought. It's more of a placeholder."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [127/120]",
+      "reference": "Using a plural route helps in detecting the path automatically for `= form_for @password_reset`."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "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 `from.toPath().toString()`.",
+      "reference": "seems reasonable to also want to include the 'to' location?"
+    },
+    {
+      "prediction": "I think this should be `if (!wasDeleted) { Runtime.getRuntime().removeShutdownHook(shutdownHook); }`",
+      "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 need to check if the volumePolicy is nil or not. If it is nil then we should return an error.",
+      "reference": "Better to push getting policy details into a func we can reuse later."
+    },
+    {
+      "prediction": "I don't think we need to do anything here, but I'm not 100% sure.",
+      "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": "nit: I think we can remove this import.",
+      "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 should be `assert.Panics(t, func() { ... })` instead of `assert.Panics(t, func() { ... })`",
+      "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'm not sure this is the right way to do this. I think we should be able to do something like: `f.Set(reflect.ValueOf(replacer.Replace(*raw.(*WaitForInstancesStopped))))`",
+      "reference": "Didn't we have specific logic for handling slices vs structs?"
+    },
+    {
+      "prediction": "This should be `new_message` instead of `new_message.c_str()`.",
+      "reference": "`c_str()` is not needed."
+    },
+    {
+      "prediction": "I think we should move this to `if features.DefaultFeatureGate.Enabled(features.AntreaIPAM)` block.",
+      "reference": "The handler will be installed twice if you don't remove the below one."
+    },
+    {
+      "prediction": "I think this should be `getEpochLengthActivationBlock().orElse(0L)`",
+      "reference": "How about putting `epochLength` (or something like that) in place of `a` just to make it easier to read"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "reference": "Wouldn't this still fire afterwards if you created one called `pleasings`? or am I missing something?"
+    },
+    {
+      "prediction": "I think this should be `kbp_record_prefix` instead of `kbp_record`.",
+      "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": "I don't think you need to suppress the unchecked warning here. The `mapper` parameter is a `BiFunction<? super T, ? super U, ? extends R>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Tuple2<T, U>>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Tuple2<?, ?>>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Object>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Object>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Object>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Object>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Object>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Object>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Object>`, which is a subtype of `BiFunction<? super T, ? super U, ? extends Object>`, which is a subtype",
+      "reference": "Does a type-hint work instead of casting? `return ofAll(toList().<U> zipWith(that, mapper));` (Probably not, just a question.)"
+    },
+    {
+      "prediction": "nit: remove the extra space after `new`",
+      "reference": "Are these required?"
+    },
+    {
+      "prediction": "I think this should be `c.mcl = int32(opts.MaxControlLine)`. If `opts.MaxControlLine` is 0, then `c.mcl` will be `MAX_CONTROL_LINE_SIZE`, but if `opts.MaxControlLine` is greater than 0, then `c.mcl` will be 0.",
+      "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": "Shouldn't this be `|| javaToken.kind == JavaTokenKinds.DECIMAL_LITERAL`?",
+      "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": "This should be `github.com/nats-io/nuid/v2`",
+      "reference": "Don't need extra line here."
+    },
+    {
+      "prediction": "I don't think this is correct. We have 3 accounts in the test, but only 2 of them are used in the test. The third one is not used at all, so it should not be included in the result.",
+      "reference": "For the first look, this change is strange. Were we passing these tests before?"
+    },
+    {
+      "prediction": "I think this should be `strings.HasSuffix(appname, processname + \".exe\")`",
+      "reference": "On Windows, doesn't `os.Args[0]` include the `.exe`? What if you run the program like `caddy.exe`?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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 it would be good to have a test for the case where there is no line break at the end of the string.",
+      "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_list() || ttype->is_set() || ttype->is_map())`",
+      "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 think this should be `os.Exit(1)`.",
+      "reference": "I would do an s.Notice(\"Initiating Shutdown...\") then after Shutdown() do the original Server Exiting."
+    },
+    {
+      "prediction": "I don't think we need to log this.",
+      "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 this should be `field.Required(computePath.Child(\"platform\").Child(\"aws\"), \"must specify platform for compute machine sets\")`",
+      "reference": "Why is this required? The user should be able to omit it and use the defaults."
+    },
+    {
+      "prediction": "This should be `&Handle mouse control from other &applications`",
+      "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 think it would be better to log the `collectorUri` as a `string` rather than a `Uri`. The `Uri` class is not serializable and will cause an exception to be thrown if you try to log it.",
+      "reference": "Didn't went with backwards compatibility since it's still in beta"
+    },
+    {
+      "prediction": "I don't think this is the right thing to do here. If `ob` is not a list of ints, then `pythonObjectToVect<unsigned int>(ob)` will return `nullptr` and `v` will be `nullptr`. Then `if (!v)` will be `true` and `throw_value_error(\"invalid value for bondsToCut\");` will be executed. But `v` is `nullptr` so dereferencing it will cause a segfault.",
+      "reference": "Perhaps bondsToCut must be None or non empty."
+    },
+    {
+      "prediction": "I think this should be `line` instead of `linechart`.",
+      "reference": "See above, we could avoid adding that (same applies below)."
+    },
+    {
+      "prediction": "I don't think you need to cast to `Long` here. `parameters.get(\"type\")` is already a `Long`.",
+      "reference": "should use Number instead of Long"
+    },
+    {
+      "prediction": "I think this should be `const expect = require('chai').expect;`",
+      "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'm not sure this is the right place to do this. I think it should be done in the `ModifyStatus` method.",
+      "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": "`random.uniform(0, 1) > self.prob` -> `random.uniform(0, 1) < 1 - self.prob`",
+      "reference": "it seems `random.uniform(0, 1)` similar to `random.randint(2)`, all have 1/2 probabilities."
+    },
+    {
+      "prediction": "I think this should be `if isinstance(v, (dict, list)):`",
+      "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\"` instead of `var 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": "This lock is not needed here.",
+      "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": "`key.reserve(kTagsTable.size() + sizeof(GraphSpaceID) + sizeof(TagID));` -> `key.reserve(kTagsTable.size() + sizeof(GraphSpaceID) + sizeof(TagID) + sizeof(int32_t));`",
+      "reference": "Typically, to be simple, we could reserve a const length space."
+    },
+    {
+      "prediction": "I think we can remove this file.",
+      "reference": "Need to add a license and copyright header to each file."
+    },
+    {
+      "prediction": "I think this should be `hidden_field_tag \"coupon_id\", @coupon.id`",
+      "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's better to use `DBPDataSource` instead of `PostgreDataSource` here.",
+      "reference": "Please remove the unused import of DBPDataSource."
+    },
+    {
+      "prediction": "Why is this a warning instead of an error?",
+      "reference": "I think you should add some extra message here saying it was originally sent as an error, for diagnosability."
+    },
+    {
+      "prediction": "I'm not sure why this is different, but it seems like it should be the same.",
+      "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 `updateRole` method of the `DBService` class. The `updateRole` method is called from the `updateRole` method of the `RoleManagerImpl` class, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `RoleManager` interface, which is called from the `updateRole` method of the `",
+      "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 don't think we need to have this as a constant. It's only used in one place.",
+      "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": "(style) code indent should use tabs where possible (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please, no space before tabs (style) please",
+      "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": "I think this should be `IsWrappedFatalError` or `IsWrappedFatalErrorCause` or something like that. `IsFatalError` sounds like it's checking if the error is a fatal error, not if it's a wrapped fatal error.",
+      "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": "I think it's better to use `klog` instead of `log` in this file.",
+      "reference": "redundant empty line"
+    },
+    {
+      "prediction": "This should be `houseOwnedByAccount` instead of `HOUSE_OWNED_BY_ACCOUNT`",
+      "reference": "The config.lua.dist still shows `houseAccountOwner`"
+    },
+    {
+      "prediction": "I don't understand why you remove this line ?",
+      "reference": "I think we can leave this import and use non FQNs in the code."
+    },
+    {
+      "prediction": "I think it would be better to use `this.driver_.getExecutor().w3c ? 'handle' : 'name'` instead of `this.driver_.getExecutor().w3c ? 'handle' : 'name'`.",
+      "reference": "I'd rather just send the parameter twice than break encapsulation here. There's already precedence with webelement IDs"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "We should probably exit early if there are no items in the list."
+    },
+    {
+      "prediction": "Shouldn't this be `missing-raises-doc`?",
+      "reference": "As we can't use old names we should warn in whats new for 2.12 that this can happen."
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I don't think we want to run `main` as part of the test suite. I think we want to run `bincover` as part of the test suite, but I'm not sure how to do that.",
+      "reference": "good job finding this package, I hope it's actively maintained"
+    },
+    {
+      "prediction": "Why is this unsetenv removed?",
+      "reference": "same changes as in hipEnvVar.cpp"
+    },
+    {
+      "prediction": "I think this should be `#define _POSIX_C_SOURCE 200809L` instead of `_POSIX_C_SOURCE 200809L`",
+      "reference": "NULL check should be before this line"
+    },
+    {
+      "prediction": "This should be `client.getLocalPlayer() != null && client.getLocalPlayer().getWorldLocation() != 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 don't think this change is correct. If the first operand is NULL and the second operand is not NULL, then the result should be UNKNOWN, not NULL.",
+      "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 if this is the right place to get the detected locale from, but it seems to be the only place where we have access to the request object.",
+      "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 think this should be `http.StatusAccepted` instead of `http.StatusContinue`.",
+      "reference": "Why setting the status to Continue as at the end it is written OK? Status should be written only once."
+    },
+    {
+      "prediction": "I think we can remove the `hydrateParamCache` function now.",
+      "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": "nit: I think this should be `mayactl snapshot create --volname <vol> --snapname <snap>` instead of `mayactl snapshot create --volname <vol> --snapname <snap>`",
+      "reference": "nit : Remove the extra lines"
+    },
+    {
+      "prediction": "I think this should be `Listen` instead of `DontListen`.",
+      "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 don't think this is the right place to do this. I think it should be done in the `analyze` method of the `ClangTidyAnalyzer` class.",
+      "reference": "Should we add some default case if non of the above return some analyzer name? Something like `unknown analyzer`?"
+    },
+    {
+      "prediction": "I don't think this is correct. If `mix_case` is true and `min_length` is 1, then `min_alpha` should be 1, not 2.",
+      "reference": "Nvm. The original version is best... *hides from the angry rubocop*"
+    },
+    {
+      "prediction": "I'm not sure why this was here, but I think it's safe to remove.",
+      "reference": "we can drop this and use the returned body above, right?"
+    },
+    {
+      "prediction": "`echo: true` is the default, so this can be removed.",
+      "reference": "Is echoing enabled on purpose here or it's a remainder of your tests?"
+    },
+    {
+      "prediction": "I think this should be `@logger.info(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 it would be better to add the `urlParams` parameter to the docblock of the `getServiceAccountSiteURL` function, instead of adding it to the docblock of the `getServiceAccountSiteURLWithParams` function. The `getServiceAccountSiteURLWithParams` function is just a wrapper around the `getServiceAccountSiteURL` function, so the docblock of the `getServiceAccountSiteURLWithParams` function doesn't need to mention the `urlParams` parameter.",
+      "reference": "This also needs to be reverted."
+    },
+    {
+      "prediction": "I don't think we need this type. You can just use `journalPausedFromConflict` and `journalPausedFromSignal` directly.",
+      "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 changed?",
+      "reference": "This is fix from another PR #120."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `__init__` method of the `MultiInterface` class.",
+      "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": "Shouldn't this be `sonata.classification.manager.category`?",
+      "reference": "lol that variable name"
+    },
+    {
+      "prediction": "I don't think we should change the names of these chains.",
+      "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 set the default representative. I think it should be set when the wallet is opened, not when a new account is created.",
+      "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 || {};`",
+      "reference": "We should add the definition to `.storybook/config.js` as well."
+    },
+    {
+      "prediction": "I think this should be `if (*(Lng32 *)op_data[1] == 0 || *(Lng32 *)op_data[1] == -1 || *(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 best way to do it, but I think it's better than what we had before.",
+      "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 don't think we need to add `fullResult` here.",
+      "reference": "Can we remove `raw`, `fullResult`, and `serializeFunctions`?"
+    },
+    {
+      "prediction": "I don't think we want to call this `Security` as it's not really about security, it's about authentication and authorization. I would call it `Auth` or `Authz` or `Authorization` or something like that.",
+      "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 = 'Debug'`",
+      "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 `if (this->GetTarget() == this)`",
+      "reference": "Shouldn't need this-> here."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Why delete this?"
+    },
+    {
+      "prediction": "I think this should be `if (typeof ref === 'function' || typeof ref === 'object' && ref.hasOwnProperty('current'))`",
+      "reference": "This line is here to be compatible with the current way `refs` work, right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Hmm, I wonder if we can safely assume all decimals should be displayed as $$...ok for now I suppose."
+    },
+    {
+      "prediction": "Line is too long. [85/80]<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.",
+      "reference": "Good spec :). Very clear and easy to see what it's doing."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Shouldn't we need to sanitize the `params[:filter]` from a defined expected values?"
+    },
+    {
+      "prediction": "I think this should be `ECS_EVENTUAL_CONSISTENCY_TIMEOUT_SECONDS` or something like that.",
+      "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": "I think we should revert this change. I don't think we should be using Guava's `MoreObjects` in our codebase.",
+      "reference": "Guava changed `Objects` -> `MoreObjects`"
+    },
+    {
+      "prediction": "I think we can remove the `import mock \"github.com/stretchr/testify/mock\"` line.",
+      "reference": "All these hand generated mocks can be deleted. I think we already moved everything to gomocks."
+    },
+    {
+      "prediction": "I think this should be `this.get('_id').toString() === doc.get('_id').toString()`",
+      "reference": "what about string/number etc _ids? this will fail."
+    },
+    {
+      "prediction": "I think this should be `eMS.Spec.Template.Spec.Annotations` and `rMS.Spec.Template.Spec.Annotations`",
+      "reference": "Looks like the test had a bug prior to this."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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`",
+      "reference": "PR works great. Just out of curiosity: why all the id changes?"
+    },
+    {
+      "prediction": "I think this should be `offsetof(struct flb_config, backoff_cap)`.",
+      "reference": "config map first value is wrong, it aims to define the \"value type\", that value will lead to an undefined behavior."
+    },
+    {
+      "prediction": "nit: `*blob.Bucket` -> `blob.Bucket`",
+      "reference": "This doesn't make sense. The thing being constructed is a blob.Bucket. \"Constructing\" the pointer is trivial."
+    },
+    {
+      "prediction": "I don't think we need to disable this here.",
+      "reference": "Outch. A configuration/parsing refactor is on the radar but this seems more pressing."
+    },
+    {
+      "prediction": "I think this should be `moment().add(-this.membersStats.days + 1, 'days').format(DATE_FORMAT);`",
+      "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 if this is the right thing to do, but I think it makes sense to have the user sign in and then click on the video trail from the landing page.",
+      "reference": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure why this was here, but it seems like it should be removed.",
+      "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 `ServiceImpactingChanges`",
+      "reference": "is this rule serviceimpacting? I'm thinking it's mostly around the modeling of pageable in the swagger, thoughts?"
+    },
+    {
+      "prediction": "I think this should be `duckv1.KResource` instead of `duckv1.Source`.",
+      "reference": "same here, replace `ks` and `kc` to something more related to Source."
+    },
+    {
+      "prediction": "Shouldn't this be `parameter is IDictionary<TKey, TValue> dict`?",
+      "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": "Align the elements of a hash literal if they span more than one line.",
+      "reference": "this should only be indented 2 spaces"
+    },
+    {
+      "prediction": "I think this should be `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 this should be `Boolean downloadOriginalTracker = false;` instead of `private Boolean downloadOriginalTracker = false;`",
+      "reference": "shouldn't every `Boolean` in this pr be `boolean`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think we should also add a `minimize_window` method to the `Remote` class in `webdriver/remote/remote_connection.py`.",
+      "reference": "This `if` is not necessary"
+    },
+    {
+      "prediction": "I think this should be `System.Action<System.Web.Http.HttpConfiguration>`",
+      "reference": "Are you using this anywhere? (I couldn't find it.)"
+    },
+    {
+      "prediction": "I think this should be `defer s.recordStepTime(startTime)()` so that we don't call `s.recordStepTime` if `s.stepImpl` returns an error.",
+      "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": "This import is not used in this file.",
+      "reference": "Please use `kinto.core.utils.json`"
+    },
+    {
+      "prediction": "I'm not sure if this should be an error or just a warning. I think it should be a warning, but I'm not sure if it should be logged here or in the `upstreamHost`'s `ServeHTTP` method.",
+      "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": "This should be `typing_extensions.final` instead of `typing.final`.",
+      "reference": "What happens if someone import `typing` and thus uses `@typing.final` instead?"
+    },
+    {
+      "prediction": "I'm not sure why this is needed, but it seems to be required for the tests to pass.",
+      "reference": "Better remove this dependency altogether"
+    },
+    {
+      "prediction": "nitpick: `h2o__http2client_wbuf_buffer_prototype` -> `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 `required`?",
+      "reference": "We will also need to add this to the spec."
+    },
+    {
+      "prediction": "I'm not sure why we need to make a copy of the header and timestamp here. I think we should be able to pass in a pointer to the header and timestamp.",
+      "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 think this should be in the constructor of the `By` class.",
+      "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": "Can you add a benchmark for `JsonSerializer.Serialize<TValue>(TValue value)` as well?",
+      "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": "I don't think we need to add this bundle to the `config/bundles.php` file.",
+      "reference": "only dev and local"
+    },
+    {
+      "prediction": "I don't think we need to make this column nullable. It should always have a value.",
+      "reference": "This should not be nullable, instead it should be `->default(0);`"
+    },
+    {
+      "prediction": "I think this should be renamed to `selectColumns` to be consistent with the rest of the codebase.",
+      "reference": "What about `selectedColumns`? I believe it is frequently used throughout the project."
+    },
+    {
+      "prediction": "I think this should be `synonymQueryBoostByPayload`",
+      "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": "Align the parameters of a method call if they span more than one line.<br>Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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 `SecretsCount` to be consistent with the rest of the codebase.",
+      "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": "Shouldn't this be `secondaryOk: true`?",
+      "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": "nit: I think we can remove the `to` and `from` fields from the `OutboundCall` struct. The `to` field is only used in the `NewOutboundCall` function, and the `from` field is only used in the `NewOutboundCallWithFrom` function. I think we can just pass the `to` and `from` fields as arguments to the `NewOutboundCall` and `NewOutboundCallWithFrom` functions, and remove the `to` and `from` fields from the `OutboundCall` struct.",
+      "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 *= abs(count)`. Otherwise, `times` will be negative if `count` is negative.",
+      "reference": "No blank line after the docstring."
+    },
+    {
+      "prediction": "nit: remove extra blank line",
+      "reference": "nit: move this to the group below."
+    },
+    {
+      "prediction": "I don't think we need to do this. If the sort record length is greater than the max buffer size, then we will fail the assertion below.",
+      "reference": "This is fine, but maybe we should change the variable name to \"bufferSize_as_uint\"? Or \"...uint32\"?"
+    },
+    {
+      "prediction": "`term_info['suggestion'].first.is_a?(Hash)` should be `term_info['suggestion'].first.is_a?(Hash) || suggestions.index(\"correctlySpelled\")`",
+      "reference": "Use snake_case for variable names."
+    },
+    {
+      "prediction": "Why is this test disabled?",
+      "reference": "Obviously, don't merge this to master."
+    },
+    {
+      "prediction": "I think this should be `expect(reporter).to have_received(:report).with(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": "Nearly there! This should assert on the output, take the result of \"start\" and \"finish\" as an example :)"
+    },
+    {
+      "prediction": "I think this should be `ms := s.grabQueue(maxMessages, ctx.Done())` so that we can break out of the loop if the context is canceled.",
+      "reference": "This loop is going to burn CPU if len(s.q) == 0. Maybe a small sleep?"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "can you give a reason why session should be contained in RegionManager?"
+    },
+    {
+      "prediction": "I think this test should be renamed to `TestAuthenticate_WhenBase64SignatureIsCorrect`",
+      "reference": "You are comparing two signerIDs here, error message should be \"Signers should match\""
+    },
+    {
+      "prediction": "I think we can remove the `GetIAMRoleCredentials` method and just use `role.IAMRoleCredentials` directly.",
+      "reference": "Who's acquiring this lock for writes?"
+    },
+    {
+      "prediction": "I don't think we should remove the `compressed` parameter. It's used in the `fromX` 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 best way to test this, but I think it's good enough for now.",
+      "reference": "Should there be a `trail.topic_name` method so we don't annoy Demeter?"
+    },
+    {
+      "prediction": "`team` is not used in this function.",
+      "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": "I think this should be `require_once __DIR__ . '/Psalm/Internal/exception_handler.php';` instead of `require_once __DIR__ . '/Psalm/Internal/exception_handler.php';`",
+      "reference": "Would it make sense to include this file via composer files-autoloading instead?"
+    },
+    {
+      "prediction": "I think this should be `elementText != null && elementText.contains(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 we can remove this line.",
+      "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": "Shouldn't this be `if (\"8bit\".equals(body.getEncoding()) || \"7bit\".equals(body.getEncoding())) {` ?",
+      "reference": "Should we ignore case when doing the comparison?"
+    },
+    {
+      "prediction": "This should be `#if WLR_HAS_X11_BACKEND`",
+      "reference": "The same `ifdef` should apply to `#include <wlr/backend/x11.h>` line 14"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think it would be better to just log the error and return `false` so that the profiler doesn't try to call the method again.",
+      "reference": "We're instrumenting a static method, so there's no instance in this case. `diagnosticSource` is the first argument."
+    },
+    {
+      "prediction": "This should be a `boolean isHighPriority = false;` instead of `private boolean isHighPriority = false;`",
+      "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": "This should be 1.4.0.1 or 1.4.0-1.",
+      "reference": "Let's go back to 1.4.0 until the release is being prepared."
+    },
+    {
+      "prediction": "I don't think you need to initialize this to 0.",
+      "reference": "I would avoid the term \"errno\" as it implies the glibc variable. Use error_code or status or something."
+    },
+    {
+      "prediction": "I think we can remove the `d` parameter from this function. It's not used.",
+      "reference": "@abhinav @kriskowal what do you think about putting the handler on `d`?"
+    },
+    {
+      "prediction": "This should be `restic.TagList` instead of `[]string`.",
+      "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 we can just use `time.Second` here.",
+      "reference": "An hour seems a significant amount of time to wait for within a test, does it need to be that long?"
+    },
+    {
+      "prediction": "I think this is the only place that needs to be changed.",
+      "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 `MsmqTransport` instead of `Msmq`",
+      "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 `apiutil.CertificateRequestHasInvalidRequest(req)`",
+      "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 we can use `Ergonode\\Core\\Application\\Messenger\\Stamp\\UserStamp` instead of `Ergonode\\Core\\Application\\Security\\User\\CachedUser`",
+      "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": "remove empty line",
+      "reference": "What's the meaning of null URIEndpointObject?"
+    },
+    {
+      "prediction": "I don't think it's a good idea to set a default log level. It should be left to the user.",
+      "reference": "Why did you remove the time? It can be useful if some actions are ran in jobs?"
+    },
+    {
+      "prediction": "This should be `github.com/t3rm1n4l/go-mega/mega`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think it's better to use `github.com/ethereum/go-ethereum/accounts/abi/bind/types` instead of `github.com/ethereum/go-ethereum/accounts/abi/bind/bind_types`",
+      "reference": "What the reason to change imports?"
+    },
+    {
+      "prediction": "I think we should also check for `TTransportException` instead of `TTransportException`",
+      "reference": "Nit: continuation indents are 2 indents / 4 spaces."
+    },
+    {
+      "prediction": "I think this should be `netloc.rfind(':', 0, netloc.rfind('@'))` so that we don't get the port from the username:password part of the URL.",
+      "reference": "You can use `netloc.rsplit(\":\")` for that."
+    },
+    {
+      "prediction": "I think this should be an error.",
+      "reference": "nits, I don't get the point of this change"
+    },
+    {
+      "prediction": "I think we can remove this file, as it's not used anywhere.",
+      "reference": "`warningPrefix` looks the same on both platforms, we could move it up to `log.go`."
+    },
+    {
+      "prediction": "Filters and Interceptors that will be applied to all outgoing and incoming requests respectively. -> Filters and Interceptors that will be applied to all outgoing and incoming connections respectively.",
+      "reference": "What's the motivation behind switching to plural here? I think we want singular cc @abhinav"
+    },
+    {
+      "prediction": "I think this should be `Wrapf` to match `Newf`.",
+      "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 it's a good idea to set default value to the same value as `Port`. I think it should be `null` or `-1` or something like that.",
+      "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, but it seems to be the only way to get the tests to pass.",
+      "reference": "I'm still not convinced dropping Chrome logs is the right thing to do, but if you insist"
+    },
+    {
+      "prediction": "I don't think we should throw an exception here. I think we should just return `null`.",
+      "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 set the log destination. I think it should be done in `main.go` instead.",
+      "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 use `@wordpress/i18n` in this file.",
+      "reference": "This belongs under \"WordPress dependencies\"."
+    },
+    {
+      "prediction": "I think we can remove the `#if !NET452` and `#endif` from this file.",
+      "reference": "Was there an issue on `net452`?"
+    },
+    {
+      "prediction": "Shouldn't this be `if (_platformNotSupportedExceptionMessage != null && _platformNotSupportedExceptionMessage.StartsWith(\"SR.\"))`?",
+      "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 to use `deepmerge` here. We can just use `Object.assign` or `lodash/merge`.",
+      "reference": "This should use the `merge` function added in #473 instead of introducing a new dependency"
+    },
+    {
+      "prediction": "This should be `VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_DEDICATED_BUFFER_MVIEW_OPAQUE_VALVE` and `VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_DEDICATED_IMAGE_MVIEW_OPAQUE_VALVE`",
+      "reference": "Shouldn't this say \"a binding with descriptor type VK_DESCRIPTOR_TYPE_MUTABLE_VALVE\""
+    },
+    {
+      "prediction": "I think this should be `static const struct wlr_tablet_tool_v2_grab_interface default_tool_interface = {};`",
+      "reference": "This variable name should probably contain `grab`"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "reference": "Which initializer should we be using for open_aire? The one in `external_apis` or this one in `initializers`?"
+    },
+    {
+      "prediction": "I think this should be `gx/ipfs/QmTiRqrF5zkdZyrdsL5qndG1UbeWi8k8N2pYxCtXWrahR2/go-libp2p-routing/routing`",
+      "reference": "remove todo on line 28?"
+    },
+    {
+      "prediction": "Please remove the `_base_` and `model` in this file.",
+      "reference": "clean unnecessary arguments."
+    },
+    {
+      "prediction": "I think this should be `EXECUTION_FAILED` instead of `EXECUTION_STOPPED`.",
+      "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": "This should be `from molecule.provisioners import DockerProvisioner` or `from molecule.provisioners.docker import DockerProvisioner`",
+      "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 `sentryAddr` and use `sentryAddrs[0]` instead.",
+      "reference": "We don't need that one anymore?"
+    },
+    {
+      "prediction": "This import is not used in this 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 don't think you need to escape the `#` here.",
+      "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 don't think we need to add `aria-label` and `title` here, since it's a button with an icon.",
+      "reference": "can we localise these? not sure if we have easy access to a translator instance from `generic-provider-views`"
+    },
+    {
+      "prediction": "I think this should be `0x8028`",
+      "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 thing to do here. The `scope` string is used to keep track of the current scope. If we truncate it here, we'll lose information about the current scope.",
+      "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 don't think we need to log the error twice here.",
+      "reference": "please change to `could not connect to peer: %s`. it is not relevant to say \"from kademlia\" in a warning"
+    },
+    {
+      "prediction": "`time::TimeConversion::unixSecondsToDateTime(timestamp).toString()`",
+      "reference": "Why need to remove this? Datetime has microseconds in fact."
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. I think you need to pass in a `t *testing.T` to the function and then call `t.Fatal(err)` if there is an error.",
+      "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 don't think this is mutation free, as it calls `parent::getAssertionString()` which is not mutation free.",
+      "reference": "this feels weird"
+    },
+    {
+      "prediction": "I think this should be `PRE_MERGE` or `PRE_MERGE_TESTNET` to be consistent with the other network names.",
+      "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 `if (get_role() == \"train\" && debug.is_open())`",
+      "reference": "debug? I suspect this will be removed before merge?"
+    },
+    {
+      "prediction": "I don't think we need to repeat this twice.",
+      "reference": "should this be here twice?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I think it would be nice if we could have a way to specify that a particular test case should be skipped. This would allow us to avoid having to use `SKIP_IF` and `REQUIRE` in the same test case, and it would also allow us to use `REQUIRE` in cases where we don't want to use `SKIP_IF` (e.g. if we want to use `REQUIRE` in a `for` loop).",
+      "reference": "`generate_special` - a meaningless name for me. Can we provide more detailed naming?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it seems to be the only way I could get the test to pass.",
+      "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.finer` instead of `finest`. `finest` is the lowest level of logging, and I don't think we want to log this at that level. `finer` is the next lowest level, and I think it's appropriate for this level of logging.",
+      "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 thing to do. The id is supposed to be a unique identifier for the remote host, not the remote host's URL.",
+      "reference": "any reason you're assigning to the local variable 'id' too?"
+    },
+    {
+      "prediction": "I think this should be `from molecule import validators`",
+      "reference": "Need to fix this import :)"
+    },
+    {
+      "prediction": "I think this should be `&types.Actor{Code: types.AccountActorCodeCid, Balance: big.NewInt(100000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "sorry these tests suck so much we are working to kill them"
+    },
+    {
+      "prediction": "I'm not sure why this is necessary. I don't think we need to multiply by 10000 here.",
+      "reference": "Why not adjust `currentTime` to be in ms directly?"
+    },
+    {
+      "prediction": "nit: extra space after `if`",
+      "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 we can remove the `ChainDBPath` and `TrieDBPath` as well.",
+      "reference": "this flag not used"
+    },
+    {
+      "prediction": "I don't think we need to convert the date here. The date is already in the correct format. We only need to convert the date when we get it from the API.",
+      "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 this file, it's not used 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 `Optional[Union[float, int]]`",
+      "reference": "Nit: this should actually always be float since we convert it before we construct the namedtuple."
+    },
+    {
+      "prediction": "I think we should check the return value of `FileUtils::makeDir(path)` and `FileUtils::fileType(path.c_str()) == FileType::DIRECTORY`",
+      "reference": "The result of `FileUtils::fileType` could be saved, instead of calling twice."
+    },
+    {
+      "prediction": "I don't think this is a good idea. If an exception is thrown, it should be logged.",
+      "reference": "Can you log the exception ?"
+    },
+    {
+      "prediction": "I think we can remove this debug message.",
+      "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 think we can remove this import.",
+      "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 is this change needed?",
+      "reference": "It would be nice all these tweaks to be configurable from cmd line, with sensible defaults"
+    },
+    {
+      "prediction": "I think we can remove the `connect_args={'timeout': 600}` here.",
+      "reference": "Why do we need this `check_same_thead` to be false? I feel a bit uncomfortable about this."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to test this. I think it would be better to have a separate test case for this. What do you think?",
+      "reference": "We should probably test with constructors which aren't in the schema, and functions which aren't constructors."
+    },
+    {
+      "prediction": "I don't think we need to pass `arg_logical_size` to `cudaMalloc`.",
+      "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 `/opt/puppet/bin` instead of `/opt/puppetlabs/puppet/bin`",
+      "reference": "This winds up flipping the desired values. Looks like I gave you a bum steer @kevpl"
+    },
+    {
+      "prediction": "`iotexAddress` -> `producerAddress`",
+      "reference": "nit: let's call our address ioAddr from now on"
+    },
+    {
+      "prediction": "I think this should be `pPhysicalDeviceProperties->limits.maxBoundDescriptorSets != 0`",
+      "reference": "You're killing this \"else\" case here which currently flags an error when maxBoundDescriptorSets == 0."
+    },
+    {
+      "prediction": "I think this should be `const char *checksum = g_variant_get_string (value, NULL);`",
+      "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": "why not use `ImmutableList` here?",
+      "reference": "or you can use `filter(x => !x.isView)`"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Instead of NewBuffer on the line above, do NewReader to avoid having to wrap later."
+    },
+    {
+      "prediction": "I don't think we need to disable pylint here. We can just add `# pylint: disable=too-many-locals` and `# pylint: disable=too-many-statements` to the lines above.",
+      "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 if this is the right thing to do, but I think we should be using `GetEncodedLength` instead of `len(protocol.Encode(&txib))` here. `GetEncodedLength` returns the number of bytes that would be written by `Encode`, while `len(protocol.Encode(&txib))` returns the number of bytes that were actually written by `Encode`.",
+      "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. [117/80]",
+      "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: please remove this extra blank line",
+      "reference": "similar here for imports, these should be consolidated (and also consolidated with the grouping below)"
+    },
+    {
+      "prediction": "I think it is better to move this line to `__init__`",
+      "reference": "Maybe there is no need to modify it. Because if you don't use the mask, it won't be initialized."
+    },
+    {
+      "prediction": "I'm not sure I understand why this was changed. The `%` is used to match any number of characters, so I don't see how it could cause a substring match. Also, the `\u2591` character is not a special character in SQL, so I don't see how it could cause a problem.",
+      "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 it would be better to keep the original name of the setting and just change the value of the setting.",
+      "reference": "There should be a migration for that as well to reflect the change on in-production instances"
+    },
+    {
+      "prediction": "I think this should be `\"\"` instead of `\"\"`",
+      "reference": "Why does it have to be `\"\"` instead of an empty string"
+    },
+    {
+      "prediction": "I think this should be `unless pending? or 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'm not sure this is the right thing to do. I think we should keep the `Kernel.srand RSpec.configuration.seed` and `Kernel.srand # reset random generation` lines, and just replace `list.sort_by { Kernel.rand(list.size) }` with `list.shuffle`.",
+      "reference": "Will this obey the seed set on Kernel?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be able to call `add_facts` without having a `bolt_inventory` or `bolt_executor` in the environment.",
+      "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 don't think this should be a runtime exception.",
+      "reference": "Why the switch from period to colon?"
+    },
+    {
+      "prediction": "I think this should be `if numImagesToDeletePerCycleEnvVal != \"\" && numImagesToDeletePerCycleEnvVal != \"0\" && err != nil {`",
+      "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 think it would be better to have a separate config section for Baseline DBs.",
+      "reference": "let us not add this"
+    },
+    {
+      "prediction": "Should this be `omitempty` too?",
+      "reference": "Is there a need for group as well? crun sets both."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but this is the only way I could get this test to work.",
+      "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 this is equivalent to calling `pre()` on `save` and `findOneAndUpdate`. `pre('save')` will run on `save`, `findOneAndUpdate`, `findOneAndReplace`, and `findOneAndDelete`. `pre(['save', 'findOneAndUpdate'])` will run on `save` and `findOneAndUpdate`, but not `findOneAndReplace` or `findOneAndDelete`.",
+      "reference": "`this.getFilter()` won't work on `pre('save')`. Perhaps make this `toySchema.pre(['updateOne', 'findOneAndUpdate'])`?"
+    },
+    {
+      "prediction": "I think this should be `1593493828_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000",
+      "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 why this was removed, but I think it should be added back in.",
+      "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 `tlsConfig.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 think this should be `tm.tm_year + 1900;`",
+      "reference": "Just occurred to me to ask: is the condor log date in localtime or UTC?"
+    },
+    {
+      "prediction": "This should be `SULLIUSCEP_TREE_LOCATION`",
+      "reference": "Can you remove the sculliscep varbits from here too?"
+    },
+    {
+      "prediction": "I think this should be `host_id` instead of `host_cpid`.",
+      "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": "This should be `duraton` -> `duration`",
+      "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_error_printf` 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 should be `it('should record roundTripTime', function (done) {`",
+      "reference": "is this one all platforms?"
+    },
+    {
+      "prediction": "I think this should be `NotFound` instead of `Not Found`.",
+      "reference": "why capitals grr?"
+    },
+    {
+      "prediction": "I don't think we need this function. We can just call `Error.Fatal(e)` directly.",
+      "reference": "Surely all the above should go into `common`."
+    },
+    {
+      "prediction": "I think this is an empty line that shouldn't be there.",
+      "reference": "This is an unrelated change, but was failing CI... probably introduced in master."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the best we can do for now.",
+      "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`",
+      "reference": "MB is presumably 1024 * 1024."
+    },
+    {
+      "prediction": "I don't think we need to import `Suspense` and `lazy` from both `react` and `@wordpress/element`",
+      "reference": "Didn't we extract this logic to a `react-features` helper?"
+    },
+    {
+      "prediction": "Why is this no longer needed?",
+      "reference": "But But, overlapping async writes will get corrupted"
+    },
+    {
+      "prediction": "I think this should be `github.com/filecoin-project/go-filecoin/internal/pkg/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 why this was removed, but it seems like it should still be here.",
+      "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": "Why do we need to check the depth of the expression here?",
+      "reference": "It seems that you only need to do this `checkExprDepth()` inside `deduceExprType()`. So you don't have to add this check everywhere."
+    },
+    {
+      "prediction": "nit: please remove the extra space",
+      "reference": "why split into two lines?"
+    },
+    {
+      "prediction": "I don't think this needs to be public.",
+      "reference": "Does this need to be public or can it be package-private?"
+    },
+    {
+      "prediction": "I think this should be `showEditButton = 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": "I don't think we need to check for `null` here, because `Stream.ofAll` will do it for us.",
+      "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 this should be `self.molecule._print_valid_platforms()` and `self.molecule._print_valid_providers()`",
+      "reference": "Just to be consistent, can we use the print function `print()` instead of the keyword. Same goes for line 437."
+    },
+    {
+      "prediction": "I think this should be `wallet[0]` instead of `wallet[0].String()`",
+      "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 don't think we need to pass the src here. We can just pass an empty object.",
+      "reference": "Could you move it after `stop`?"
+    },
+    {
+      "prediction": "I don't think we need to change the class name here. The class name should be the same as the method name.",
+      "reference": "The variable `method` seems not used. Maybe we can remove the parameter `bodyParamName` and generate it by invoking `ParamUtils.generateBodyParameterName(method)` ?"
+    },
+    {
+      "prediction": "I don't think this test needs to be marked as `@Slow` anymore.",
+      "reference": "I don't think this should be here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the least bad of the options we have.",
+      "reference": "Either I'm missing something or specs actors should remove this call cc @anorth"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `foreach (var counter in topCounters.OrderBy(c => c.MetricName))`",
+      "reference": "Please undo all these changes as they are white-space only."
+    },
+    {
+      "prediction": "I don't think this is the best way to get the alpha2 code from the alpha3 code. I would use `Countries::getAlpha3ToAlpha2Mapping()` instead.",
+      "reference": "You forgot an optimization? `$countries = CountryField::FORMAT_ISO_3166_ALPHA3 === $countryCodeFormat ? Countries::getAlpha3Names() : $countriesAlpha2;`"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be `value = \".\".join(str(item) for item in sys.version_info[:2])`",
+      "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 this should be `time_range` instead of `time_range`.",
+      "reference": "> the time range the listen search the time range of the listen search?"
+    },
+    {
+      "prediction": "I don't think this is the right place to set the flag. The flag should be set when the instruction is decoded, not when the prefix is read.",
+      "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": "Should this be `--conn_retries <number>` or `--connect_retries <number>`?",
+      "reference": "would call it connect_retries"
+    },
+    {
+      "prediction": "I think it's better to use `assertRaises` instead of `assertRaisesRegex`.",
+      "reference": "Why are we using `repr`?"
+    },
+    {
+      "prediction": "I think this should be `bind(ZMSImplFactory.getZmsInstance()).to(ZMSHandler.class).in(Singleton.class);`",
+      "reference": "I use the same zms instance to check authentication in swagger endpoints. Same thing in ZTS."
+    },
+    {
+      "prediction": "nit: I think it would be better to use `ProcessJob.EXECUTE_AS_USER` instead of `CommonJobProperties.EXECUTE_AS_USER` here.",
+      "reference": "Consider consolidating the common code in a common setup method in tests?"
+    },
+    {
+      "prediction": "This import is not needed anymore.",
+      "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 think this should be `Schema` instead of `DiscoveryField`.",
+      "reference": "What is the motivation for switching from `Schema` to `DiscoveryField` everywhere?"
+    },
+    {
+      "prediction": "I think this should be `containerMetadataEnabled := utils.ParseBool(os.Getenv(\"ECS_ENABLE_CONTAINER_METADATA\"), true)`",
+      "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": "This should be `WITH_XCB_ERRORS` instead of `HAS_XCB_ERRORS`.",
+      "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's better to use `recoverutil` instead of `util/recoverutil`",
+      "reference": "alias is not needed"
+    },
+    {
+      "prediction": "I think it's better to put the pid file in the same directory as the log file.",
+      "reference": "Do we need to include the path?"
+    },
+    {
+      "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": "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": "Why is this needed?",
+      "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 `GetKeyForUser` instead of `HasKeyForUser`.",
+      "reference": "We maintain that each (logical) key generation has the same set of device keys, so no need to plumb through `keyGen`."
+    },
+    {
+      "prediction": "I think this should be `mock>=2.0.0, <3.0.0`",
+      "reference": "Please use `from unittest import mock` instead of this package."
+    },
+    {
+      "prediction": "I don't think we should be doing this here. This should be done in the `exec` method of the `Query` class.",
+      "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": "Metrics/LineLength: Line is too long. [121/120]",
+      "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 think we can remove the `#if DEBUG` and the `Debug.Assert` here.",
+      "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 think this should be `props.value == null && props.value !== 0 && props.value !== false`",
+      "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": "This should be `import { __, sprintf } from '@wordpress/i18n';`",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "I think we can remove the `OpenEBSLocalPVProvisionerLabelSelector` from here.",
+      "reference": "The latest yaml in openebs/master has changed to name=admission-webhook. cc: @prateekpandey14"
+    },
+    {
+      "prediction": "Nitpick: I think it would be more accurate to say \"the AWS SDK for Ruby v3\".",
+      "reference": "using **an** Amazon Simple... Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "Why is this here?",
+      "reference": "Probably don't intend for this `1 ||` to still be here"
+    },
+    {
+      "prediction": "I don't think we need to remove this check.",
+      "reference": "Looks like this is failing to compile on Windows (VS 2015 I think)."
+    },
+    {
+      "prediction": "nit: I think it would be better to call this `WithBeforeQuery` or `WithBeforeQueryFunc` to make it clear that it's a modifier.",
+      "reference": "nit: call the arg something simple like `f` to avoid repeating \"beforeQuery\"."
+    },
+    {
+      "prediction": "I think this should be moved to `pkg/correlation/correlation.go` and `pkg/correlation/correlation_test.go` should be moved to `pkg/correlation/correlation_test.go`",
+      "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": "I think this should be `Message = string.Empty;` instead of `Message = \"\";`",
+      "reference": "Why are you passing out the Message?"
+    },
+    {
+      "prediction": "I don't think we need to use a mutex here. The `spanProcessors` map is only used in `RegisterSpanProcessor` and `UnregisterSpanProcessor` which are both protected by the `Provider`'s mutex.",
+      "reference": "Yikes, this was a bug :grimacing:"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to call `Finish` on the subscriber. I think it should be called in the `for` loop below, after the `select` statement.",
+      "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'm not sure if this is the right thing to do, but I don't think we need to open the file if we're not scaffolding.",
+      "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'm not sure this is the right thing to do. If the condition doesn't exist, it means we haven't checked yet, so I think we should return `unreachable = false` and `lastCheck = time.Time{}`.",
+      "reference": "This seems unrelated, should be a separate PR."
+    },
+    {
+      "prediction": "Why do you need to include engine_create_dispatcher.h here?",
+      "reference": "Never use relative includes, prefer full path"
+    },
+    {
+      "prediction": "`Zero_count Argument cannot exist without count!`",
+      "reference": "This should fit on the line above without going over the 79-char limit, no?"
+    },
+    {
+      "prediction": "Block has too many lines. [31/25]",
+      "reference": "Is this related to the other contents of this pull request?"
+    },
+    {
+      "prediction": "I think this should be a `map[string]string` instead of a `string`.",
+      "reference": "Should this be `*string`?"
+    },
+    {
+      "prediction": "nit: I think we can remove the empty string here.",
+      "reference": "In the old code here there was no Source Selector, but CreateDefaultProfile will specify a Source Selector of `\"\"`. Is that equivalent?"
+    },
+    {
+      "prediction": "I think this should be `caller.find { |line| line !~ %r{/lib/rspec/(core|mocks|expectations|matchers|rails)/} }` instead of `caller.find { |line| line !~ %r{/lib/rspec/(core|mocks|expectations|matchers|rails)/} }`",
+      "reference": "We should pick this across for all our deprecation specs"
+    },
+    {
+      "prediction": "I think this should be `kubectl` instead of `kubectlBinary` to be consistent with the rest of the codebase.",
+      "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\")`",
+      "reference": "Is this a related change?"
+    },
+    {
+      "prediction": "I think this should be moved to the end of the function, right after the call to `a.reconcileLBAttachment(scope, machine, instanceDescription)`.",
+      "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": "I don't think we need to do this here. The `UndoRedo.ChangeAction` constructor will take care of it.",
+      "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` to be consistent with `--bfbs-comments`.",
+      "reference": "I hate to ask, but can we make this flag shorter so it lines up? :("
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "Note to reviewers: latent, dedup import"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "Seems this func is no more needed after your refactoring."
+    },
+    {
+      "prediction": "I think this should be `Tuple.of(\"^io\\\\.usethesource\\\\.\", \"Steindorfer persistent @ \")`",
+      "reference": "already added (see several rows below)"
+    },
+    {
+      "prediction": "I think this should be `if ifaceName == \"wireguard.cali\" || ifaceName == \"wg-quick.cali\"`",
+      "reference": "nit: a switch perhaps?"
+    },
+    {
+      "prediction": "I think this should be `false` since this is not a retryable failure.",
+      "reference": "Great you caught this. Super critical to not retry these errors."
+    },
+    {
+      "prediction": "I think this should be `Path instancePath = it.handler.coreContainer.getCoreRootDirectory().resolve(coreName);`",
+      "reference": "Just a little change to make the var effectively final, which is clearer."
+    },
+    {
+      "prediction": "I think this should be `sc.ip_address[0]` and `sc.ip_address[1]`",
+      "reference": "Using `\"{}\".format(...)` is a bit beside the point..."
+    },
+    {
+      "prediction": "`obj` should be renamed to `record` for consistency with the rest of the codebase.",
+      "reference": "I think these should remain as they are."
+    },
+    {
+      "prediction": "I don't think we need to make this a class variable. We can just make it a local variable in the method that uses it.",
+      "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": "I think this should be `return Device[DeviceIndex].Number;` instead of `return (validDeviceIndex(DeviceIndex) ? Device[DeviceIndex].Number : INVALID_PLUGIN_ID);`",
+      "reference": "We don't have a `getPluginIDfromTaskIndex` function? I think it should be implemented in Globals/Plugins.h / .cpp"
+    },
+    {
+      "prediction": "Shouldn't this be `rb.req.getSchema().getUniqueKeyField().getName()` instead of `rb.req.getSchema().getUniqueKeyField().getName()`?",
+      "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 }}-{{.name }}-db`",
+      "reference": "Should we be using version 3 now?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. If the config file is not found, it should throw an exception.",
+      "reference": "Would be better to throw a more specific exception, but for now it will do."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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'm not sure if this is the right thing to do, but I think it is. If the file doesn't exist, we don't want to try to set the permissions on it.",
+      "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 called `NewParserWithBuiltins` to match the naming of `newAspParser` above.",
+      "reference": "This function isn't useful? it's just a clone of GetParserWithBuiltins?"
+    },
+    {
+      "prediction": "I don't think we need to validate the cpu circuit breaker threshold. It's not used in the same way that the memory circuit breaker threshold is used.",
+      "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 think this should be `RewriteFiles rewriteDeletes(Set<DeleteFile> deletesToDelete, Set<DeleteFile> deletesToAdd);` instead of `RewriteFiles rewriteFiles(Set<DeleteFile> deletesToDelete, Set<DeleteFile> deletesToAdd);`",
+      "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 this is the right place for this caching. I think it would be better to cache the result of `t('blacklight.application_name', locale: I18n.default_locale)` in `config/initializers/i18n.rb`.",
+      "reference": "Layout/TrailingWhitespace: Trailing whitespace detected."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it works for now.",
+      "reference": "Is it possible that the shortcuts list is empty?"
+    },
+    {
+      "prediction": "I think this should be `if errors.Cause(err) == context.Canceled || errors.Cause(err) == context.DeadlineExceeded { return errors.Fatal(\"canceled\") }`",
+      "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 `Content-Length` header should always be set by the server. If it's not set, the server should set it. If it is set, the server should not set it again. If it is set and the client tries to set it again, the server should ignore the client's setting.",
+      "reference": "maybe it's better to remove HttpHeaders.CONTENT_LENGTH after the loop"
+    },
+    {
+      "prediction": "I think this should be `if (n < 0) {` instead of `if (n <= 0) {`",
+      "reference": "Very cool. We are now collecting the fruits after your length implementation!"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do this, but it seems to work.",
+      "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": "nit: I think it would be good to add a `defer eniAttachHandler.stop()` here.",
+      "reference": "Please break this into multiple lines."
+    },
+    {
+      "prediction": "I think we can remove the `scope.Span.SetHeaderTags<IHeadersCollection>(httpContext.Response.Headers.Wrap(), Tracer.Instance.Settings.HeaderTags, defaultTagPrefix: SpanContextPropagator.HttpResponseHeadersTagPrefix);`",
+      "reference": "Shouldn't this new line replace the few lines above?"
+    },
+    {
+      "prediction": "I think this should be `flux.RegisterBuiltInOption(\"now\", universe.SystemTime())`",
+      "reference": "Perhaps this belongs in universe?"
+    },
+    {
+      "prediction": "I think this should be 1e-10 or 1e-11. 1e-12 is way too small.",
+      "reference": "there is a `kEpsilon` you can use directly."
+    },
+    {
+      "prediction": "nit: I think we can remove the empty line here.",
+      "reference": "This should be grouped with the below imports, and the two separate groups of imports below should likely also be grouped together"
+    },
+    {
+      "prediction": "This should be `error_code = \"VUID-VkCopyDescriptorSet-dstSet-parameter\";`",
+      "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": "`static const std::string prefix = kPartsTable;` -> `static const std::string prefix(kPartsTable);`",
+      "reference": "If return by value, then why bother to make it `static`?"
+    },
+    {
+      "prediction": "I think this should be `if paste: utils.pastebin_version()` instead of `if paste: utils.pastebin_version()`",
+      "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 import.",
+      "reference": "`sdktrace` or `tracesdk`, pick one."
+    },
+    {
+      "prediction": "I don't think we should return an empty array here. I think we should return `null` instead.",
+      "reference": "can we return null instead?"
+    },
+    {
+      "prediction": "I don't think we need to wait for the child to die. We can just destroy the event.",
+      "reference": "You mean XXX and \"we should\" or \"NYI\" or sthg. Maybe put it on the ifdef line."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be in `pkg/importer/importer.go`.",
+      "reference": "minor: can be in-lined in return"
+    },
+    {
+      "prediction": "I think we should remove the `\\n` from the expectation here.",
+      "reference": "why the additional \"\\n\"?"
+    },
+    {
+      "prediction": "I think this should be `return err`",
+      "reference": "This seems to be a big omission. If the first buffer reports an error, stream seems to just discard it."
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "Maybe move this one down too?"
+    },
+    {
+      "prediction": "I think we can remove the `readonly` here, since it's not used outside of the constructor.",
+      "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 we need to keep the original error message in the new error message. I think it's enough to keep the original error message in the `errs` field of `dErrImpl`.",
+      "reference": "What if formatPrefix doesn't include a space? Will the resulting error have e.Error() stuck at the end without any space?"
+    },
+    {
+      "prediction": "nit: \"pushsync protocol implementation\" -> \"pushsync protocol\"",
+      "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 `VK_IMAGE_LAYOUT_DEPTH_STENCIL_ATTACHMENT_OPTIMAL` instead of `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 should be `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": "I'm not sure why we need to convert this to a dict here, but it seems to be necessary to get the `items()` method to work.",
+      "reference": "what does this change do exactly?"
+    },
+    {
+      "prediction": "I think this should be `ctx context.Context` and `backgroundCtx := ctx.Background()`",
+      "reference": "What's the purpose of this `backgroundCtx` field?"
+    },
+    {
+      "prediction": "I don't think we need to have this constant here. It's only used in one place.",
+      "reference": "non contract incarnation"
+    },
+    {
+      "prediction": "I'm not sure about this change. If `value` is not a dictionary, then `value.get('message', None)` will raise a `TypeError`.",
+      "reference": "I would recommend to leave out `None` because None is already the default."
+    },
+    {
+      "prediction": "I think this should be an error, not a debug message.",
+      "reference": "Should the \"bad hello message\" case above also disconnect?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these. I think they should be in the `InsulinOrefBasePlugin` class.",
+      "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": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Should probably say REGEX or PATTERN rather than string."
+    },
+    {
+      "prediction": "I think this should be `10000000` instead of `10000001`",
+      "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": "`isset($metadata['associationType'])` is always true.",
+      "reference": "In theory `isset($metadata['associationType'])` is `false` if the linked field is not an association field, isn't?"
+    },
+    {
+      "prediction": "I'm not sure if we want to remove the `authorization_code` grant flow.",
+      "reference": "@tute the only thing I needed to enable this like as `password` is not in default flow anymore I think."
+    },
+    {
+      "prediction": "I think this should be `allAuthenticatedUsers` instead of `allAuthenticatedUsers`.",
+      "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 `if (options.verbose) args.push('--verbose=1')`",
+      "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 `Log system events`",
+      "reference": "Change the key to `default_tab` instead and I'll merge this"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 correct. If the value is an enum, it should be converted to a string, not wrapped in a StringValue.",
+      "reference": "I think yes, this makes sense to expose this. +1"
+    },
+    {
+      "prediction": "I think this should be `OPENEBS_IO_LOCALPV_HOSTPATH_DIR` instead of `OPENEBS_IO_LOCALPV_HOSTPATH_DIR`.",
+      "reference": "let us put the sample yaml snippet i.e. maya api server deployment that makes use of this env & value"
+    },
+    {
+      "prediction": "I think we can remove this test.",
+      "reference": "What changed to cause this?"
+    },
+    {
+      "prediction": "I think this should be `StoreSession(bucketName string, key string, value interface{}) error` and `GetAll(issuer string, data interface{}) error`",
+      "reference": "Repeating big interface in many places shows smtg is wrong with design"
+    },
+    {
+      "prediction": "I'm not sure about the name of this controller. I think it should be `ChartsController` instead of `Accounts::ChartsController` because it's not related to the `Account` model.",
+      "reference": "Inheriting from `AccountsController` to get access to the `redirect_if_disabled` filter."
+    },
+    {
+      "prediction": "I don't think we need to add the `@param s3_client [Aws::S3::Client] An initialized Amazon S3 client.` line here.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "nit: please remove the extra space after `Class<?>`",
+      "reference": "Enter at the end of the file! :-D (and at the end of lines ;-) )"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we want to rename this column to `scheduled_for_deactivation_on` so that it's clear that we're deactivating the subscription, not cancelling it.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I think we can just do `newRoot.RemoveSuccessor(root)` and then `newRoot.AddSuccessors(newSucc...)`",
+      "reference": "Don't you want to place this check before you add `succ` to `newSucc`?"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I think it's better to remove unused imports than to leave them in the code.",
+      "reference": "Commit 676eeeab10884456a5f70ce708a8aa5116ecbaf7 contains a lot of unrelated code style changes. Please get rid of these."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get this to work with Bokeh 0.12.6 and 0.12.7.",
+      "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": "Why remove this? It's useful to know how many peers are in the list.",
+      "reference": "Consider instead moving this into an _test file so it's public but only usable in tests."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the only way to make the tests pass.",
+      "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": "I think this should be `tstzrange($3, $4, '[]'), $4, $4, NOW())`",
+      "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 to call `setup_profiler`. I think it should be called after `@setup_default.call`, but before `@setup = true`.",
+      "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'm not sure if this is the right thing to do, but I don't think we want to expose this command to users yet.",
+      "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 think we can remove this file now.",
+      "reference": "perhaps file can be removed altogether (trying to avoid an empty package for these build tags)"
+    },
+    {
+      "prediction": "I think this should be `NULL` instead of `0`.",
+      "reference": "NULL change, could you move that out too?"
+    },
+    {
+      "prediction": "I don't think we need this function. We can just call `formatList` directly.",
+      "reference": "A brief explanation about this func is required."
+    },
+    {
+      "prediction": "I think it would be better to use `MergedCellsCollection.IS_OVERLAPPING_WARNING(newMergedCell)` instead of `MergedCellsCollection.IS_OVERLAPPING_WARNING(newMergedCell.row, newMergedCell.col)`.",
+      "reference": "Could you add description for below static function?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "Since this is only counting the failed attempts, can you call it failedAttemptsCount or retryCount instead of attemptsCount?"
+    },
+    {
+      "prediction": "I'm not sure why this is here.",
+      "reference": "I don't think this will work."
+    },
+    {
+      "prediction": "Should this be `IHttpParser<T>`?",
+      "reference": "Since http2 is coming do we want to call this something more specific?"
+    },
+    {
+      "prediction": "I'm not sure why this was wrapped in a `content_tag` in the first place, but it's not needed.",
+      "reference": "I don't think we want to revert the accessibility issues the tool caught, do we?"
+    },
+    {
+      "prediction": "I think this should be `!rt.config.DisableLocalhostConnectionRateLimit || !isLocalhost(trackerRequest.remoteHost)`",
+      "reference": "I prefer naming this rateLimitedRemoteHost This can be local host but reads remote Host Is Non Local."
+    },
+    {
+      "prediction": "I think we can remove the `else if (type != null && typeof type === 'object')` check here.",
+      "reference": "Hmm I think I might've messed up this condition here... Need to look more at it"
+    },
+    {
+      "prediction": "I don't think we need to check for `package_stats is None` here. If `package_stats` is `None` then `{'total_files', 'total_bytes'}.issubset(package_stats)` will be `False` anyway.",
+      "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 `False` to be consistent with the 1D case.",
+      "reference": "Is this correct? The auto indexing behavior caused problems fixed in a recently merged PR..."
+    },
+    {
+      "prediction": "I think this should be `cryptography>=1.6, <2.0`",
+      "reference": "Hmm, didn't we just find out that 1.6 *does NOT* work? We should bump the minimum as well."
+    },
+    {
+      "prediction": "This should be `std::cerr << \"Assertion (\" << check_expr << \") failed \" << file << \":\" << line << std::endl; abort();`",
+      "reference": "Will calling `abort()` cause problems for our test suite?"
+    },
+    {
+      "prediction": "I think this should be `if (!expr_object(options, astp)) return AST_FATAL;`",
+      "reference": "for my own edificaton, what's this?"
+    },
+    {
+      "prediction": "I think this should be `if ws == nil || sf.currentChainHeight != ws.Version() {`",
+      "reference": "make it harmless to commit ws == nil"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put these, but I don't know where else to put them.",
+      "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 if this is the right thing to do, but I think it's the right thing to do for the purposes of this test.",
+      "reference": "That's actually not as bad as I thought it was going to be."
+    },
+    {
+      "prediction": "I think this should be `ORDER BY o.id DESC` so that the most recently created datasets are at the top of the list.",
+      "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": "This should be `frozenset(...)`, not ``frozenset(```.",
+      "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 think this should be `IHttpMinRequestDataRateFeature` and `IHttpMinResponseDataRateFeature` instead of `IHttpMinDataRateFeature` and `IHttpMinDataRateFeature`.",
+      "reference": "I would change \"should be sent\" to \"must be received\"."
+    },
+    {
+      "prediction": "nit: `DeleteStoragePool receives StoragePoolClaim object and deletes it.` -> `DeleteStoragePool receives a StoragePoolClaim object and deletes it.`",
+      "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 it is better to use `cse.rest.upload` as prefix.",
+      "reference": "It's better to use servicecomb now."
+    },
+    {
+      "prediction": "I think this should be `defaultRenewBeforeExpiryDuration time.Duration` instead of `defaultRenewBeforeExpiry time.Duration`.",
+      "reference": "Make the flag derived default available in the readiness controller"
+    },
+    {
+      "prediction": "I think we should have a separate test case for this.",
+      "reference": "Confirmed this test fails when the fix is reverted"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to return an error if the file logging is disabled.",
+      "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": "`conv_offset` can be replaced by `self.conv_offset`",
+      "reference": "DCN should not be able to use `init_cfg`."
+    },
+    {
+      "prediction": "I think this should be an `ArgumentOutOfRangeException`.",
+      "reference": "It is a bit weird to see a message like \"something must be valid\"."
+    },
+    {
+      "prediction": "I'm not sure this is the right level of logging for this message. I think it should be at INFO level.",
+      "reference": "Again. This is part of the start up of the server, and the diagnostics are helpful."
+    },
+    {
+      "prediction": "I think this should be `include_sexp [:cmd, /.*R CMD check.* --no-manual.*/, echo: true, timing: true]` instead of `include_sexp [:cmd, /.*R CMD check.* --no-manual.*/, echo: true, timing: true]`",
+      "reference": "I think you need a `,` at the end of the line?"
+    },
+    {
+      "prediction": "I think we should revert this change. It is not necessary and makes the code harder to read.",
+      "reference": "Please avoid wildcard imports."
+    },
+    {
+      "prediction": "Is there a way to make this configurable?",
+      "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": "nit: I think it's better to use `nodeConfig.NodeIPv4Addr.IP` and `nodeConfig.NodeIPv6Addr.IP` instead of `nodeIP` in the error message.",
+      "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 think this should be `if server.sni is not None:` instead of `if server.sni:`.",
+      "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'm not sure why this number is so high. Is it because we're using a different dictionary?",
+      "reference": "Hmm what are these magical constant numbers? And why does this change mean they should increase?"
+    },
+    {
+      "prediction": "This should be 2.9.0-dev.",
+      "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 do this. I think it should be done in the `responseWriter` itself.",
+      "reference": "nit: let's move this right under where we create the `responseWriter`, before `h.callHandler`"
+    },
+    {
+      "prediction": "I don't think we should be using `release_assert_success` here. It's not a release assert, it's a debug assert. We should be using `debug_assert_success` instead.",
+      "reference": "The way it was implemented, needed to pass the template values. Please check whether we want this."
+    },
+    {
+      "prediction": "I think this should be \"time a node waits after the end of an epoch for blocks to arrive\"",
+      "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": "This is not needed. You can remove it.",
+      "reference": "Is this import needed?"
+    },
+    {
+      "prediction": "I think this should be `padding: 20px;` to match the other stories.",
+      "reference": "Dashboard stories also need to have the default padding."
+    },
+    {
+      "prediction": "I think this should default to `true` and we should remove the `DEFAULT` constant.",
+      "reference": "Is there a reason why we wouldn't use this as the default?"
+    },
+    {
+      "prediction": "`edgeName` -> `req.get_edge_name()`",
+      "reference": "if there is an tag of the same name?"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I think it's needed for `exec.LookPath` below.",
+      "reference": "I'd add `// +build go1.10` to prevent from accidentally compiling with older Go."
+    },
+    {
+      "prediction": "I think this should be `from __future__ import unicode_literals` at the top of the file.",
+      "reference": "Not sure how this import relates to the PR but seems like a good idea anyway..."
+    },
+    {
+      "prediction": "Why is this included here? It's not used in this file.",
+      "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.AddDebug(LogLevel.Debug);`",
+      "reference": "nit: Space between these lines."
+    },
+    {
+      "prediction": "I think it would be better to have a default translator here.",
+      "reference": "Why aggregator should have this translator? It's Selenium specifics for now..."
+    },
+    {
+      "prediction": "I think this should be `void setReverse(boolean reverse)` and the default implementation should be `void setReverse(boolean reverse) { this.reverse = reverse; }`",
+      "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": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "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 don't think this is the right place to set the `content-type` header. The `content-type` header should be set in the `init` object passed to the `fetchWithTimeout` function.",
+      "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 think this should be `WLR_ERROR` instead of `WLR_INFO`.",
+      "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'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Should we have an avatar, truncate, etc examples, too?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "could you add this to `docs/source/reference/indexing.rst` ?"
+    },
+    {
+      "prediction": "This import is not needed.",
+      "reference": "Why the change from rdkit.six to six?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `extractPathExpressions` instead.",
+      "reference": "Looking good, let's just move this inside the `if` to avoid doing the extra work if we don't need it."
+    },
+    {
+      "prediction": "Shouldn't this be `expect(err).to.match(/Batch cannot be re-executed/);`?",
+      "reference": "I think we should change this to an instanceOf check for MongoBatchReExecutionError"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get a `conversion.Converter` with the new API.",
+      "reference": "Required by the libcalico-go changes"
+    },
+    {
+      "prediction": "I think this should be `blockNonExistentError` instead of `BlockNonExistentError`.",
+      "reference": "Explain why we don't reuse the corresponding server error locally?"
+    },
+    {
+      "prediction": "I think this should be `this.releaseDate` instead of `this.releaseOrCreateDate`.",
+      "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": "This should be `if (is_single() || is_page() && !is_front_page() && !in_array(basename(get_permalink()), $classes)) {`",
+      "reference": "can you replace the tabs with 2 spaces please to match the rest of the project?"
+    },
+    {
+      "prediction": "I think this should be `timeSrc` and `timeDst` instead of `TimeSrc` and `TimeDst`.",
+      "reference": "Why are we adding it to the integral function?"
+    },
+    {
+      "prediction": "I think this should be `nl.Swap32(uint32(state.Spi))`",
+      "reference": "Same for this file"
+    },
+    {
+      "prediction": "`fileCharacteristics` should be renamed to `fileInfo`.",
+      "reference": "Looks good! But why it's still a out parameter? We can refactor the whole tree because it's internal?"
+    },
+    {
+      "prediction": "I think this should be `if (curr_grab != NULL) { curr_grab->interface->cancel(curr_grab); }`",
+      "reference": "probably better to use `wlr_seat_keyboard_end_grab()`."
+    },
+    {
+      "prediction": "I think this should be `!configuration.isUnprotectedTransactionsAllowed() && transaction.getChainId().isEmpty()`",
+      "reference": "Such a large test impact for such a little 3 line change ;)"
+    },
+    {
+      "prediction": "I think this should be `@ORM\\ManyToMany(targetEntity=\"Shopsys\\FrameworkBundle\\Model\\Transport\\Transport\", inversedBy=\"payments\", cascade={\"persist\"})`",
+      "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 don't think this should be an error",
+      "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 should be `cert-manager.k8s.io/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 this should be `public Props getCommonPluginLoadProps() {`",
+      "reference": "No need to add this file change for this PR. Please remove it."
+    },
+    {
+      "prediction": "I think it would be better to use `java.lang.*` instead of `java.lang.`",
+      "reference": "this extra dot at the end is causing a couple tests to fail on Travis"
+    },
+    {
+      "prediction": "I don't think you need to use a mutable object for Python 2 compatibility. You can use `nonlocal` in Python 3.",
+      "reference": "Did you try using `global` with a regular integer ?"
+    },
+    {
+      "prediction": "I think this should be 24, not 20.",
+      "reference": "windows size might be off"
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "This should still be guarded with an `#ifdef` since the DataMan code will only be supported for shared lib builds."
+    },
+    {
+      "prediction": "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.<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.<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.<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.<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",
+      "reference": "thanks for cleaning up these old length checks"
+    },
+    {
+      "prediction": "I think this should be `e.config.Output.Namespace` instead of `e.config.Input.Namespace`.",
+      "reference": "What will be logged when the `Input.Namespace` was not specified?"
+    },
+    {
+      "prediction": "I don't think this is a good name. I would call it `fixViolations` or something like that.",
+      "reference": "I wouldn't expose this until the feature is ready. We are most likely to cut releases while still in development"
+    },
+    {
+      "prediction": "Do we need this in axe-core? I don't see it being used anywhere.",
+      "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 this should be `internal static int IncrementIfTriggered()` instead of `public static int IncrementIfTriggered()`.",
+      "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 we can remove the `Trigger.TriggerBuilder` and just use the `Trigger` constructor.",
+      "reference": "would be good to turn these into constants."
+    },
+    {
+      "prediction": "I think this should be `//go:generate gowire` and `// Code generated by gowire. DO NOT EDIT.`",
+      "reference": "Seems like it should be `generated by Wire` here and in the SUT?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we need to pass `false` here, otherwise the deal will be rejected by the miner.",
+      "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": "nit: I think it would be better to use `nameof(Timeout.Infinite)` instead of `Timeout.Infinite` in the error message.",
+      "reference": "`Timeout.Infinite` is actually `-1`. Here the correct message can be taken from Line:67."
+    },
+    {
+      "prediction": "`using Xunit.Extensions;` -> `using Xunit;`",
+      "reference": "I think we need `#if !SILVERLIGHT` here and at the test."
+    },
+    {
+      "prediction": "This import should be removed.",
+      "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 the right thing to do, but I think we need to remove the `GenerateServiceAccountName` function and just use the `kServiceAccount` here.",
+      "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 think this should be `-A` instead of `--append`.",
+      "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 `uploads with indeterminate file size`",
+      "reference": "I presume this isn't some special term..."
+    },
+    {
+      "prediction": "I think this should be `return undefined;` instead of `undefined;`",
+      "reference": "Did you mean to return here? ;) You'll also need to add an `incomplete` message to the rule metadata."
+    },
+    {
+      "prediction": "Should this be `JivaCloneReplicaTypeHolder`?",
+      "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 call `onStartSendRequest()`. It should be called in `RestClientSender#send()`.",
+      "reference": "`invocation.getInvocationStageTrace().startSend()` is inside `RestClientSender.send`"
+    },
+    {
+      "prediction": "I think this should be `actHash` instead of `actionHash`.",
+      "reference": "use \"actionHash\" to be consistent with other places"
+    },
+    {
+      "prediction": "I don't think you need to change this class at all.",
+      "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 this should be `m_errorMonitor->SetUnexpectedError(\"VUID-VkSamplerYcbcrConversionCreateInfo-xChromaOffset-01651\");`",
+      "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.SiteRunning`",
+      "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": "Shouldn't this be `sns_backend.sms_messages.clear()`?",
+      "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 it's better to use `github.com/mysteriumnetwork/node/client/stats/dto/stats_dto` instead of `github.com/mysteriumnetwork/node/client/stats/dto`",
+      "reference": "Some weird blocks formation again."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the only way to get the test to pass.",
+      "reference": "This is worrisome."
+    },
+    {
+      "prediction": "I think this should be `AutoFilesystemTrimUsage` instead of `AutoFSTrimUsage`",
+      "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 if this is the right thing to do, but I think we need to add `/opt/puppetlabs/bin` to the `puppetbindir` so that we can run `puppet` and `puppetserver` commands.",
+      "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 think this should be `Ember.inject.service('pouchdb')` instead of `Ember.inject.service()`.",
+      "reference": "any point in keeping this here ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do for now.",
+      "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": "I don't think you need to include `routing/routing_base_offline.hpp` if you don't have `OSRM_EXTERNAL_MEMORY` defined.",
+      "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": "Why is this change necessary?",
+      "reference": "Then, where is it removed?"
+    },
+    {
+      "prediction": "I think this should be set to `false` by default, and only set to `true` in the `CaseSensitivePath` test.",
+      "reference": "The tests are not run concurrently? (I don't know. Better check.)"
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "reference": "Regardless of scale?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it works for now.",
+      "reference": "Nit: should not we hide them from printing out?"
+    },
+    {
+      "prediction": "I think it's better to keep the `ImageToTensor` and `Collect` in the pipeline.",
+      "reference": "What is the reason to modify those things? I noticed it breaks the examples using `asyncio`"
+    },
+    {
+      "prediction": "I think this should be `Encoding transport.Encoding = \"rpc-protobuf-raw-response\"`",
+      "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 don't think we need to add `wtimeoutMS` and `journal` to this set. `wtimeoutMS` is just an alias for `wtimeout` and `journal` is just an alias for `j`.",
+      "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 don't think this is correct. The `underlayer` of a `GTPPDUSessionContainer` is a `GTPHeader`, not a `GTPPDUSessionContainer`. The `underlayer` of a `GTPPDUSessionContainer` is a `GTPHeader`, and the `underlayer` of a `GTPHeader` is a `GTPPDUSessionContainer`. The `underlayer` of a `GTPPDUSessionContainer` is a `GTPHeader`, and the `underlayer` of a `GTPHeader` is a `GTPPDUSessionContainer`. The `underlayer` of a `GTPPDUSessionContainer` is a `GTPHeader`, and the `underlayer` of a `GTPHeader` is a `GTPPDUSessionContainer`. The `underlayer` of a `GTPPDUSessionContainer` is a `GTPHeader`, and the `underlayer` of a `GTPHeader` is a `GTPPDUSessionContainer`. The `underlayer` of a `GTPPDUSessionContainer` is a `GTPHeader`, and the `underlayer` of a `GTPHeader` is",
+      "reference": "You probably don't want to assume that `self.underlayer` has a `seq` attribute."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the only way to make this work.",
+      "reference": "This apparently breaks the language selection."
+    },
+    {
+      "prediction": "Line is too long. [81/80]",
+      "reference": "since we're not using mysql, do we need this migration?"
+    },
+    {
+      "prediction": "I think this should be `PARTITION_ADJUSTMENT_ENABLED` or something like that. `PARTITION_ADJUSTMENT` sounds like it's a boolean, but it's not.",
+      "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 this should be `this.activeExecutions` instead of `this.activeFlows`.",
+      "reference": "unfinishedFlows might not be the same as activeFlows. Will it cause confusion here?"
+    },
+    {
+      "prediction": "I think we can remove the `_, _ = snapshot, err` line here.",
+      "reference": "Same, remove the error handling."
+    },
+    {
+      "prediction": "I think this should be `Task<bool> Start(object hint = null);` so that it can return whether the start was successful or not.",
+      "reference": "We should not change this at this time."
+    },
+    {
+      "prediction": "I don't think it's necessary to specify the order of w and h here.",
+      "reference": "We may also update the docstring in the AnchorGenerator's arguments at around line 15."
+    },
+    {
+      "prediction": "I think this should be `inserted.first->second->commandBuffer` instead of `inserted.first->second`.",
+      "reference": "FormatHandle isn't current typesafe w.r.t. requiring *VULKAN* handles... so this slipped through."
+    },
+    {
+      "prediction": "I think it's better to use `github.com/asaskevich/event` instead of `github.com/asaskevich/EventBus`",
+      "reference": "What kind of dependency is here, I think we have own implementation of event wo external dependencies"
+    },
+    {
+      "prediction": "I think this should be `SYCL::SYCLDevice::SYCLDevice(size_t id) : m_device(sycl::device::get_devices(sycl::info::device_type::gpu)[id]) {}`",
+      "reference": "Throw an exception instead"
+    },
+    {
+      "prediction": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "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 don't think we should use `Storage::url()` here. We should use `Url::asset()` instead. `Storage::url()` is used when we want to get the URL of a file that is stored in the `storage` directory. `Url::asset()` is used when we want to get the URL of a file that is stored in the `public` directory. In this case, we want to get the URL of a file that is stored in the `public` directory, so we should use `Url::asset()` instead of `Storage::url()`.",
+      "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": "`itemmethod.parent.frame().name` is the same as `itemmethod.parent.name`.",
+      "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": "Why is this needed?",
+      "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": "Why is this increased?",
+      "reference": "I am not sure what is 120 and what is 10. Hence unable to review."
+    },
+    {
+      "prediction": "I think this should be `EnableDNSMasq` or `EnableDNSMasqSupport` or something like that. `EnableDNS` is too vague.",
+      "reference": "Not needed anymore"
+    },
+    {
+      "prediction": "I don't think we need to expose this package.",
+      "reference": "Please not abbreviations"
+    },
+    {
+      "prediction": "I think this should be `!aws.IsReaderSeekable(r.GetBody())` instead of `!aws.IsReaderSeekable(r.Body)`.",
+      "reference": "This shouldn't get stomped over on line 207"
+    },
+    {
+      "prediction": "I think this should be \"All properties in Position are one indexed\"",
+      "reference": "are they not 1-indexed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be `currentTime > lastTime.CheckTimestamp` and `currentTime > lastTime.Timestamp`",
+      "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's better to keep the `short_version` the same as `__version__`.",
+      "reference": "`short_version = __version__`"
+    },
+    {
+      "prediction": "`AWS IAM` -> `AWS Identity and Access Management`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think we can remove the `Serializable` interface from this class.",
+      "reference": "Required for the SparkDataFile SerDe"
+    },
+    {
+      "prediction": "I think this should be `return !in_value.IsBaseClass() && !SwiftASTContext::IsFullyRealized(in_value.GetCompilerType());`",
+      "reference": "Do you want to make this an assertion instead?"
+    },
+    {
+      "prediction": "This should be 2015-2016, not just 2016.",
+      "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 should be `if head.mdID == md.mdID` instead of `if head != (ImmutableRootMetadata{}) && head.mdID == md.mdID`.",
+      "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 this change is necessary.",
+      "reference": "Why are you changing this?"
+    },
+    {
+      "prediction": "Why do we need to pass a pointer to `json.RawMessage` here?",
+      "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 think we can remove this line as well.",
+      "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": "nit: please remove the extra blank line.",
+      "reference": "pylint might complain that this isn't alphasorted?"
+    },
+    {
+      "prediction": "I'm not sure why this is needed, but it seems to be required for the tests to pass.",
+      "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 don't think we need this function. We can just use `TaskStatus.String()` in the backend.",
+      "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::idForInstance<Kokkos::Experimental::HIP>(reinterpret_cast<uintptr_t>(this));`",
+      "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 this should be `map[string]common.HclPluginConfig` instead of `map[string]map[string]common.HclPluginConfig`.",
+      "reference": "nit: `PluginConfigs` feels a little better"
+    },
+    {
+      "prediction": "Why do we need to remove the error check 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 I understand why we need to do this synchronously. If we don't do it in a goroutine, then the caller of `Get` will be blocked until the block is cached. If we do it in a goroutine, then the caller of `Get` will be unblocked immediately, and the block will be cached in the background. Why do we need to block the caller of `Get`?",
+      "reference": "Please expand \"later behavior\". Presumably something to do with the prefetch logic?"
+    },
+    {
+      "prediction": "I think this should be `return swarm.ZeroAddress, err`",
+      "reference": "The `response` variable is not used, it can be removed, also the fmt.Errorf."
+    },
+    {
+      "prediction": "`value?.Length > 0 ? string.IsNullOrEmpty(value.Trim()) : true;` can be simplified to `string.IsNullOrEmpty(value?.Trim())`",
+      "reference": "Do you find this really easier to read?"
+    },
+    {
+      "prediction": "I think you can remove the forceGroupByAgg parameter and the GroupByAggNodeAdded variable. The only caller of addOneRowAggregates() passes TRUE for forceGroupByAgg, so you can just add the GroupByAggNodeAdded variable to that caller.",
+      "reference": "A small nit: This variable should start with a lower case letter."
+    },
+    {
+      "prediction": "I think we should use `context.getNamer().getResponseTokenFieldName(pageStreaming)` instead of `context.getNamer().getResponseTokenFieldName(pageStreaming)`.",
+      "reference": "Where is this used for Node?"
+    },
+    {
+      "prediction": "I don't think we need to change this message.",
+      "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 should be `github.com/aws/amazon-ecs-agent/agent/config/config` instead of `github.com/aws/amazon-ecs-agent/agent/config`",
+      "reference": "There's a lot of changed tests -- but no new tests. You need unit and integration tests for this."
+    },
+    {
+      "prediction": "I think it's better to return `AcceptTxResult.GasLimitExceeded` instead of `AcceptTxResult.Accepted` here.",
+      "reference": "WithMessage what gas limit is?"
+    },
+    {
+      "prediction": "I think we can remove the `@ignore` and `@api private` tags.",
+      "reference": "Why are these skipped?"
+    },
+    {
+      "prediction": "I think it's better to put `Enable` and `Debug` in `MetaManager` instead of `MetaServer`.",
+      "reference": "We'd better add the `json:\"metaServer,omitempty\"` here"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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'm not sure if this is the right thing to do, but I think we should return an error if `syscall.Exec` returns an error.",
+      "reference": "not sure how this got in master, but the linter is unhappy about it so I fixed it here"
+    },
+    {
+      "prediction": "I don't think we need to increase this. It's only used for reconnecting to the MD server, and we don't expect that to happen very often.",
+      "reference": "No need for this TODO anymore, right?"
+    },
+    {
+      "prediction": "I think this should be `if 'kinto.plugins.default_bucket' not in settings['includes'] and 'kinto.plugins.default_bucket' not in settings['excludes']:`",
+      "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 `unset($this->plugins[ $this->normalizeIdentifier($id) ]);` instead of `unset($this->plugins[$id]);`",
+      "reference": "Would be better to normalize it at the start of the method to pass through to all the other calls"
+    },
+    {
+      "prediction": "I think this should be `$this->parentForm->model->isSafeMode() || $this->parentForm->model->isAdmin()`",
+      "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 think this should be `except COMError as e: log.exception(\"Can't scroll: %s\", e)`",
+      "reference": "This line is unnecessary."
+    },
+    {
+      "prediction": "I think this should be removed from the header file as well.",
+      "reference": "How did this compile in the first place? Is there a matching declaration in a header that should be removed too?"
+    },
+    {
+      "prediction": "nitpick: I think it would be better to use `h2o_mruby__abort_exc(mrb, \"h2o_mruby_new_str:failed to create string:\", file, line);` instead of `h2o_mruby__abort_exc(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": "Why are these 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 `report(EventType eventType, Map<String, Object> metadata)` instead of `report(EventType eventType, Map<String, String> metadata)`.",
+      "reference": "The convention is to use <p> see the google style guide and use checkstyle to check. You can change it later."
+    },
+    {
+      "prediction": "I think this should be `self.list_marker, self.item_marker` instead of `self.list_marker, self.item_marker, self.item_marker`",
+      "reference": "Using non-keyword args after the keyword arg (parent-self) will cause a SyntaxError."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I think you'll want to add some YARD docs to this method to get the PR approved"
+    },
+    {
+      "prediction": "I think this should be \"If it doesn't exist, @out_variant will be set to %NULL and the function will still return TRUE.\"",
+      "reference": "Hmm, AFAICT `*out_variant` isn't explicitly set to `NULL` in the `ENOENT` case."
+    },
+    {
+      "prediction": "I think this should be `b.logger.Fatalw(\"Failed to create SDK client\", \"error\", err)`",
+      "reference": "Nit: \"Failed to configure TLS for SDK client\"?"
+    },
+    {
+      "prediction": "I think you can use `f-string` here as well.",
+      "reference": "please update your branch to the latest `master`. This change was already made in #4359"
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "This line should be formatted"
+    },
+    {
+      "prediction": "I don't think this is a good idea to have a command line option for this. This is something that should be configured in the config file.",
+      "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 don't think this is needed.",
+      "reference": "remove unnecessary whitespace changes"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Are these conditions covered elsewhere ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it would be better to have a function that checks if the names are the same, and then call that function here.",
+      "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": "(style) code indent should use tabs where possible (style) please, no space before tabs",
+      "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": "I think this should be `#include \"plugins/map_matching.hpp\"`",
+      "reference": "reorder includes to be alphabetically ordered"
+    },
+    {
+      "prediction": "`need_unmap` -> `unmap`",
+      "reference": "Rename `need_unmap` to `unmap`."
+    },
+    {
+      "prediction": "why remove this import?",
+      "reference": "In line 28, it has been declared twice, so removing `cm` should not impact, can you explain me little bit? :)"
+    },
+    {
+      "prediction": "I don't think we need to do this. We can just call `startTopologyCache` in the `New` function.",
+      "reference": "i don't think these unexported functions belong on the public interface, they can exist solely on the struct."
+    },
+    {
+      "prediction": "I think this should be `WithDispenser(d *caddyfile.Dispenser) (Helper, error)`.",
+      "reference": "I'm not entirely clear on why this is necessary?"
+    },
+    {
+      "prediction": "I don't think this is necessary. `get_temp_file` already returns a `str`.",
+      "reference": "Why do we need to cast to `str` here?"
+    },
+    {
+      "prediction": "Why is this test ignored?",
+      "reference": "The ignore import above should be unused now and can be removed."
+    },
+    {
+      "prediction": "Shouldn't this be `Occur.MUST`?",
+      "reference": "redundant null check?"
+    },
+    {
+      "prediction": "Why do we need to set `Available` to the same value as `Total`?",
+      "reference": "initialization of available ram is here"
+    },
+    {
+      "prediction": "I think this should be `this.sourceBlock_.editable_ && this.sourceBlock_.type === 'math_number' && this.sourceBlock_.getField('NUM') === this`",
+      "reference": "Can have negative numbers with the '-' on the right (`10-`) by just using line 610."
+    },
+    {
+      "prediction": "`AWS IAM` -> `AWS Identity and Access Management`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think we should keep the `command = ' '.join(sys.argv)` line.",
+      "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 think this should be `!@exclusion_patterns.empty?` so that it returns `true` when there are exclusion patterns and `false` when there are none.",
+      "reference": "Given that this is a boolean predicate, maybe it should be `full_backtrace?`."
+    },
+    {
+      "prediction": "I think this should be `from django.contrib.auth import get_user_model as User`",
+      "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 I like this. I think we should have a `Dictionary<string, string>` and then serialize it to a string when we need it.",
+      "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": "This import is not used.",
+      "reference": "`PY310` should probably be imported from `pylint.const`"
+    },
+    {
+      "prediction": "I think this should be `GetValueAsBool(key string) (bool, error)` so that the caller can handle the error.",
+      "reference": "Can we propagate this error up the stack?"
+    },
+    {
+      "prediction": "Why do we need to create a homogen table here? Can't we just use `reference`?",
+      "reference": "Why reference is converted to table?"
+    },
+    {
+      "prediction": "I think this should be `desktop->views = list_create(); assert(desktop->views);`",
+      "reference": "You shouldn't use assert for runtime errors."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `vendor` folder and loaded in the `vendor.js` file.",
+      "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 think this test should be updated to match the new output format.",
+      "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 `Used when something is compared against itself, which is usually what you do not want to do, since you can compare to a different literal than what was expected altogether.`",
+      "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": "I don't think we need to make these virtual.",
+      "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 think this should be `mmn-api-url` or `mmn-api-address` to be consistent with the rest of the flags.",
+      "reference": "This looks a bit cryptic to me, and also it contains a dash. How about `mymysterium.url`?"
+    },
+    {
+      "prediction": "I think this should be `if (var == 0) {`",
+      "reference": "You did notice the no newline at the end of this file, right?"
+    },
+    {
+      "prediction": "I'm not sure why this is an `ISet` instead of a `List`.",
+      "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 this should be `from .config import BAN_PUBLIC_USERS` instead of `from .config import BAN_PUBLIC_USERS as BAN_PUBLIC_USERS`",
+      "reference": "Sorry, one more thing... You should use `app.config` instead of importing it directly. See the code below."
+    },
+    {
+      "prediction": "I think this should be `void releaseResources()` or `void release()` or something like that. `close()` has a specific meaning in Java.",
+      "reference": "This was correct before; \"setup\" is a noun and \"set up\" is the verb form."
+    },
+    {
+      "prediction": "I think this should be `maybeSetAccount: true` 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 think this should be `q === '~'` instead of `q === '~'`",
+      "reference": "I wonder why ~ and not a wildcard instead?"
+    },
+    {
+      "prediction": "`if (beanCls == null) { return; }` -> `if (beanCls == null) return;`",
+      "reference": "when will this happened? if happened, just ignore it Silently?"
+    },
+    {
+      "prediction": "I think this is a typo, it should be 12000.",
+      "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 `isinstance(input, datetime.datetime) or isinstance(input, datetime.date)`",
+      "reference": "@scottpurdy, strictly speaking, should this scenario raise a ValueError exception instead of AssertionError?"
+    },
+    {
+      "prediction": "I think this should be `KUBERNETES_EXECUTOR` and `KUBERNETES_SCHEDULER`.",
+      "reference": "Can you please add open source disclaimer?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `wireError = &wirepb.Error{Message: appErr.Error(), Code: int32(appErr.Code())}`",
+      "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": "I think this should be `svr.cfg.KamchatkaBlockHeight = 10`",
+      "reference": "don't need this line, since it is not tested in api_test?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Does this mean the list operation is done for all the namespaces? How to list the volume replicas per namespace?"
+    },
+    {
+      "prediction": "`nms_op = eval(nms_type)` -> `nms_op = eval(nms_type)`",
+      "reference": "I suggest adding `class_agnostic` as an argument of `batched_nms()`, with the default value False."
+    },
+    {
+      "prediction": "I think this should be `DELETE_FILE_ROW_FIELD_NAME` instead of `DELETE_FILE_ROW_FIELD_ID`",
+      "reference": "@rdblue, did we not add the name on purpose?"
+    },
+    {
+      "prediction": "I don't think we need to make these enums public.",
+      "reference": "Why is this now public?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I think it's better than what we had before. I'm open to suggestions on how to do this better.",
+      "reference": "I guess we should backport some deprecation for this?"
+    },
+    {
+      "prediction": "I think we can remove the `when /^Path=(.+)$/` block and the `path` variable.",
+      "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 don't think this is needed. You can just use `opnd_create_reg(DR_REG_##reg)`.",
+      "reference": "I would say just change REGARG: no need for a separate thing."
+    },
+    {
+      "prediction": "I think we can remove the `inset` class from the `footer` div.",
+      "reference": "we can remove this template, it is no longer in use"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "This doesn't look like it has as much coverage of the recursive code as it once did."
+    },
+    {
+      "prediction": "I think this should be `+kubebuilder:validation:MinItems=1`",
+      "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": "This import is not used in this file.",
+      "reference": "This seems like a new dependency on rometools. Do we need it?"
+    },
+    {
+      "prediction": "Should this be `String.format(\"Pull of %s:%s complete\", name, tag)`?",
+      "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": "nit: I think we can remove this import.",
+      "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": "(style) line over 80 characters",
+      "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 should be `ReceiveAndDelete` instead of `ReceiveAndDeleteMode`.",
+      "reference": "Instead of just \"See SubscriberOptions....\", say something like \"Use ... to choose between the two.\""
+    },
+    {
+      "prediction": "This file should not be included in this PR.",
+      "reference": "Use with statement for `f`"
+    },
+    {
+      "prediction": "I think this should be `Any iso` instead of `Any tag`.",
+      "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 if this is the right thing to do, but I don't think we need to implement this for 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 don't think this is correct. `findElements` should return `List<WebElement>`, not `List<T>` where `T` extends `WebElement`.",
+      "reference": "This change should also probably go into the corresponding method of the abstract By class?"
+    },
+    {
+      "prediction": "please remove this empty line",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "`WriteTo` returns the number of bytes written and an error. We should check the error here.",
+      "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 it would be better to call this `is_upfront_shutdown_enabled` or something like that. `upfront_shutdown_script_enabled` is a bit confusing.",
+      "reference": "this method is not used"
+    },
+    {
+      "prediction": "I don't think we need to emit `title_changed` if the title is empty.",
+      "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": "I'm not sure I understand why this is false. I would have expected it to be true.",
+      "reference": "Why this change (test name still says \"must pass with metadata\")?"
+    },
+    {
+      "prediction": "I think this should be `vers = force_187_p371 config[:rvm].to_s.gsub(/-(1[89]|2[01])mode$/, '-d\\1')`",
+      "reference": "Another nitpick: parens around the arg pretty please"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I think it should be kept.",
+      "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 `RewriteOriginalURICtxKey` 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 think this should be `cmapi.CertificateOutputFormatPEMKey` instead of `cmapi.CertificateOutputFormatCombinedPEMKey`.",
+      "reference": "Not added by you, but we don't really need the '5' here.."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "The vendor prefix is still being used on Chromium based browsers like Edge Chromium and Chrome. Did you mean to remove this?"
+    },
+    {
+      "prediction": "This should be `func.emuGesture = emuGesture`",
+      "reference": "I don't think this is needed anymore?"
+    },
+    {
+      "prediction": "I think we can just pass `filter.ByBanned.Value` here, and remove the `byBanned` variable.",
+      "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": "why remove this?",
+      "reference": "Need you to delete the `version_` from `HostInfo` struct ?"
+    },
+    {
+      "prediction": "I think this should be `node.getAttribute('lang') || node.getAttribute('xml:lang') || ''`",
+      "reference": "This could be the cleanest looking check we've got. Great job Jey."
+    },
+    {
+      "prediction": "`strlen` is faster than `empty` or `!== ''`",
+      "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 don't think we need to remove the part from the engine if it is already loaded.",
+      "reference": "When could this happen?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the content type. I think it should be set in the `PostSegmentAsync` method. What do you think?",
+      "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 think this should be `SOCKS4` and `SOCKS5` instead of just `SOCKS`",
+      "reference": "Here is a newline missing"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get R 4.0.0 on High Sierra.",
+      "reference": "Did you mean to make this http rather than https?"
+    },
+    {
+      "prediction": "I don't think this change is necessary. The `obj.name` will be set to `winUser.getWindowText(obj.windowHandle).replace('&','')` in `NVDAObject`'s `__init__` method, which is called before `AppModule`'s `event_NVDAObject_init` method.",
+      "reference": "The updated copyright header should be: # Copyright (C) 2006-2018 NV Access Limited, yourname"
+    },
+    {
+      "prediction": "`$performanceTestSample = null;`",
+      "reference": "wow :+1: , i do not even know how this test works."
+    },
+    {
+      "prediction": "This should be `[MvxPlugin]` not `[MvxPlugin]`",
+      "reference": "File should be renamed `PlugIn` -> `Plugin`"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "I scanned the new test cases quickly. Do we have one for an egressIPPool with multiple different IP ranges?"
+    },
+    {
+      "prediction": "I think this should be `should_component_be_loaded` or `should_component_be_enabled` or something like that. `active` is a term used for plugins and themes, not for components.",
+      "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 don't think this is necessary. `bulkExecute` will always return an object with `options` and `callback` properties.",
+      "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 way to do this, but I couldn't find a way to check for a flash message in a feature spec.",
+      "reference": "Useless assignment to variable - `user`."
+    },
+    {
+      "prediction": "I don't think this is correct. `archive_name(release)` is the name of the tarball, not the path to the tarball.",
+      "reference": "I think we would want to keep `-O` instead. I wonder how this worked before, though."
+    },
+    {
+      "prediction": "I don't think we need to store the tagKeyCombinations in the AggregatorStore. It can be computed on the fly from the tagKeysInteresting.",
+      "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": "Why do we need to update the symbol table here?",
+      "reference": "if (out != nullptr)"
+    },
+    {
+      "prediction": "I think this should be `need_exact = request.args.get(\"exact\", \"false\") == \"true\"`",
+      "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'm not sure this is the best way to do this. I think it would be better to have a `TabbedBrowser.close_tab()` and a `TabbedBrowser.remove_tab()` method. The `TabbedBrowser.close_tab()` method would call the `TabbedBrowser.remove_tab()` method and then call the `TabbedBrowser.close()` method. The `TabbedBrowser.remove_tab()` method would call the `TabbedBrowser.close_tab()` method and then call the `TabbedBrowser.close()` method. The `TabbedBrowser.close_tab()` method would call the `TabbedBrowser.close_tab()` method and then call the `TabbedBrowser.close()` method. The `TabbedBrowser.close_tab()` method would call the `TabbedBrowser.close_tab()` method and then call the `TabbedBrowser.close()` method. The `TabbedBrowser.close_tab()` method would call the `TabbedBrowser.close_tab()` method and then call the `TabbedBrowser.close()` method. The `TabbedBrowser.close_tab()` method would call the `TabbedBrowser.close_tab()` method and then call the `TabbedBrowser.close()` method. The `TabbedBrowser.close_tab()` method would call the `TabbedBrowser.close_tab()` method and then call the",
+      "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 we can remove this import.",
+      "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 this should be `bool` instead of `?bool`",
+      "reference": "I reverted that one in a previous PR because of a CI failure but it was actually unrelated"
+    },
+    {
+      "prediction": "I don't think we need to do this. `latestVersion` is not used anywhere else.",
+      "reference": "The for loop is useless here, since any error will make the function return."
+    },
+    {
+      "prediction": "Why is this file empty?",
+      "reference": "Fix file endings here and elsewhere."
+    },
+    {
+      "prediction": "I think this should be `%.127s` to leave room for the null terminator.",
+      "reference": "I'd better increase `buf` length to MAXPATHLEN and not cut p->get_project_name() output twice"
+    },
+    {
+      "prediction": "This should be removed.",
+      "reference": "Please remove the prints, even though this code will be trashed."
+    },
+    {
+      "prediction": "nit: I think it's better to keep the `linux` and `external` packages in the `felix/dataplane/` directory.",
+      "reference": "Not sure if you added deliberately but I've seen these popping up; is goimports adding them?"
+    },
+    {
+      "prediction": "I think this should be `dst.Spec.NetworkSpec.SecurityGroupOverrides = restored.Spec.NetworkSpec.SecurityGroupOverrides.DeepCopy()`",
+      "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": "Please remove the `_base_` and `model` in this file.",
+      "reference": "Is this from a third-party library?"
+    },
+    {
+      "prediction": "I think this should be `if (el.tabIndex > -1 && dom.isFocusable(el) && !dom.isNativelyFocusable(el)) { return true; } else { return false; }`",
+      "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 don't think we need to import `github.com/mysteriumnetwork/go-openvpn/openvpn3` here.",
+      "reference": "`identity` is unused now, is it for future usage?"
+    },
+    {
+      "prediction": "I don't think we need to mention `out::message` here, since it's not part of this PR.",
+      "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 `b'c' + module_name.encode('utf-8')`",
+      "reference": "I think it's better we keep `(` and encode with default encoding (utf-8)."
+    },
+    {
+      "prediction": "I don't think this is correct. The `trimTrailingWhitespace` call should be done before the `importOrder` call.",
+      "reference": "this is gonna be different in an IDE vs from gradlew??"
+    },
+    {
+      "prediction": "Line is too long. [85/80]",
+      "reference": "Do we not already have something for isolating syntax?"
+    },
+    {
+      "prediction": "I think this should be 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024, 2025, 2026, 2027, 2028, 2029, 2030, 2031, 2032, 2033, 2034, 2035, 2036, 2037, 2038, 2039, 2040, 2041, 2042, 2043, 2044, 2045, 2046, 2047, 2048, 2049, 2050, 2051, 2052, 2053, 2054",
+      "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": "I don't think we need to change the description of this example.",
+      "reference": "how **to** deny"
+    },
+    {
+      "prediction": "Should these be `internal`?",
+      "reference": "The only beef I have with this is that it's app domain global."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to work for me on OpenBSD 6.5.",
+      "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": "This should be `0.0f` and `1.0f`, not `-0.0f` and `+1.0f`.",
+      "reference": "Typical to report the value found as well as the valid bounds."
+    },
+    {
+      "prediction": "I think this should be `transactionLastValidRound int64` instead of `transactionLastValidRound uint64`.",
+      "reference": "maybe we want 0 as a default value?"
+    },
+    {
+      "prediction": "I don't think we need to return the block identifier here.",
+      "reference": "can we get height from sf at very beginning of the function and use that as height of the block?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Does this have to be public and in the primary namespace?"
+    },
+    {
+      "prediction": "I think this should be `fbo.log.CDebugf(ctx, \"Head was already set, aborting: %v\", err)`",
+      "reference": "I'm a bit confused. Doesn't `mdWriterLock` already synchronizes calls to this function?"
+    },
+    {
+      "prediction": "Shouldn't we throw an exception here?",
+      "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": "nit: I think we can remove the empty line here.",
+      "reference": "Why the separation of imports here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should keep the `deps` argument around, but make it optional. If it's not provided, we can use the default dependencies.",
+      "reference": "Random extra newline."
+    },
+    {
+      "prediction": "`descendents` -> `descendants`",
+      "reference": "No need for this variable."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it seems to work.",
+      "reference": "Why not use `msg` directly below?"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "I think other places changed to use `Deque` instead of suppressing the warning."
+    },
+    {
+      "prediction": "`#include <boost/range/algorithm_ext/erase.hpp>` -> `#include <boost/range/algorithm/erase.hpp>`",
+      "reference": "missing: utility, cstddef also: vector is already included in the header (interface requires vectors as in/out types)"
+    },
+    {
+      "prediction": "I think this should be `'fetchMock': '<rootDir>/node_modules/fetch-mock/dist/cjs/index.js'`",
+      "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 != 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. [91/80]",
+      "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)s:%(user)s/%(pkg)s is private. To share it with the team, run `quilt access add %(team)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 done in the `on_request` method.",
+      "reference": "Nit: Don't move it between `require_dummy_response` definition and usage, this can live above or below :)"
+    },
+    {
+      "prediction": "I think this should be `EnableBlockServiceFallbackToArchiverEndpoints`",
+      "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'm not sure I understand why this is necessary. `mergedNodeCache` is only used in this function, so it can't be garbage collected until the function returns, right?",
+      "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 don't think we need to start the DNS daemon here. It should be started in the `start()` method.",
+      "reference": ":+1: amazing that this has been missing since it was originally implemented"
+    },
+    {
+      "prediction": "I think we can remove this method. It is not used anywhere.",
+      "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 this should be `if (!(other instanceof PartitionField)) {`",
+      "reference": "nit: I feel multiple `if (...) { return ... }` is more readable than `if ... else ...`."
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "This is needed for us to work in IDE UTs"
+    },
+    {
+      "prediction": "I think this should be \"Only show files that are starred in their original directory structure.\"",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be `label=disable` not `label:disable`",
+      "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 `r` instead of `R` to be consistent with the rest of the flags.",
+      "reference": "You need to declare variable `recurse` above, same place as `localTime`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think it should be `xdensity=kwargs.get('xdensity', element.xdensity)` and the same for `ydensity`.",
+      "reference": "Why does `clone` not already copy `xdensity` and `ydensity` from what it is cloning?"
+    },
+    {
+      "prediction": "I think this should be `if (meterSources == null || meterSources.Count() == 0) { throw new ArgumentException(\"No meter 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": "This function is only called by `SdMmcHcStartSdClock()` and `SdMmcHcStopSdClock()`, so I think we can remove the `@param[in] PciIo` and `@param[in] ControllerVer` parameters.",
+      "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 think this should be `webelem.Group.all, webelem.Group.prevnext, webelem.Group.url`",
+      "reference": "You'll also need to add `webelem.Group.all` everywhere as that matches as well"
+    },
+    {
+      "prediction": "Use single-quoted strings instead of double-quoted strings.<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long. [81/80]<br>Line is too long",
+      "reference": "Does this need to be an ERB template?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `feedbackv1` import and just use `Feedback` and `FeedbackMetadata` directly.",
+      "reference": "do we want to register this in the mock server for testing?"
+    },
+    {
+      "prediction": "I'm not sure why this change is necessary, but it seems like it should be reverted.",
+      "reference": "shouldnt this be `0:996` (+4) so the 1000s below fit?"
+    },
+    {
+      "prediction": "I think this should be `MultiDimensionalMapping._sort` instead of `MultiDimensionalMapping.sort` to be consistent with `MultiDimensionalMapping._sorted`.",
+      "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": "Why is this necessary? I don't think it should be.",
+      "reference": "This tests failed unless I swapped the order here. Perhaps some new class transform broke the old form?"
+    },
+    {
+      "prediction": "`ops.split(\"address=\")[1]` should be `opts.split(\"address=\")[1]`",
+      "reference": "extract the port and set to `debug_java_port`"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(os.path.abspath(entry['directory']), os.path.relpath(sourcefile, entry['directory']))`",
+      "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 think this should be `status.Errorf(codes.Unimplemented, \"EnablePiped is not implemented\")`",
+      "reference": "`ctx` is unused in EnablePiped"
+    },
+    {
+      "prediction": "I don't think we need to store the `authenticated` flag in the session. We can just check if `self.user` is `None` or not.",
+      "reference": "Why do we have both a list of valid sessions and an instance variable if a session is destroyed?"
+    },
+    {
+      "prediction": "Why is this needed? I don't see it being used anywhere.",
+      "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 should be `assertEquals(0, new NullRepository().fetchInstallableApplications().size());` instead of `assertEquals(null, new NullRepository().fetchInstallableApplications());`",
+      "reference": "You may want to use assertNull"
+    },
+    {
+      "prediction": "I think this should be `1593493828_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000",
+      "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 set to 1.",
+      "reference": "Why did you remove this?"
+    },
+    {
+      "prediction": "Shouldn't this be `log.Errorf`?",
+      "reference": "Why we need another error printouts? We already have it inside addMapping.."
+    },
+    {
+      "prediction": "This should be `pipeline->rp_state->dynamic_rendering_pipeline_create_info.colorAttachmentCount` instead of `location_map.size()`.",
+      "reference": "This LGTM, but I'm curious if this fixed a specific error you were hitting?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "Actually we rely on this functionality in cgroup v2, where the subsystem is empty."
+    },
+    {
+      "prediction": "I think this should be `go.opentelemetry.io/api/core`",
+      "reference": "suggestion: use `core` package name"
+    },
+    {
+      "prediction": "I don't think we should change the default value of `limit` to `get_option(\"compute.max_rows\")` in this PR. I think we should keep the default value of `limit` as `1000` and add a new parameter `max_rows` to `get_option` and `set_option`.",
+      "reference": "I think we can entirely remove this `limit` parameter for now to be consistent with other APIs."
+    },
+    {
+      "prediction": "I think this should be `Resource Groups` instead of `Resource Groups Tagging API`.",
+      "reference": "nit: `Resource Groups Tagging API` -> `Resource Groups API`"
+    },
+    {
+      "prediction": "nit: I think we can remove the empty line here.",
+      "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 we can remove this line.",
+      "reference": "You can remove this since ES6 modules are strict by default. And thanks for contributing to the Jellyfin Project."
+    },
+    {
+      "prediction": "ad565b60006000f35b600054600160a060020a031632600160a060020a031614156100bd576100be565b60006000f35b600054600160a060020a031632600160a060020a031614156100ce576100cf565b60006000f35b600054600160a060020a031632600160a060020a031614156100df576100e0565b60006000f35b6000546001",
+      "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(), expectedErr)` instead of `strings.HasSuffix(err.Error(), expectedErr)`.",
+      "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": "Why do we need this?",
+      "reference": "nit: Remove this assignment because it is not necessary."
+    },
+    {
+      "prediction": "I think this should be `glnx_openat_rdonly (repo->repo_dir_fd, \"summary\", TRUE, &fd, error)`",
+      "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 instanceof FeedMedia && ((FeedMedia) media).fileExists()`",
+      "reference": "Wouldn't this crash the app if the user is currently listening to a stream?"
+    },
+    {
+      "prediction": "I think this should be `prefix != \"\" && !strings.HasSuffix(prefix, \"/\")` and `directory != \"\" && !strings.HasSuffix(directory, \"/\")`",
+      "reference": "This looks like an unrelated change? What is it for?"
+    },
+    {
+      "prediction": "I think this should be `return d.empty || atomic.LoadInt32(&d.used) != 0`",
+      "reference": "Wait, what is this doing? Is this a way to create methods that are only accessible from tests?"
+    },
+    {
+      "prediction": "I think we should return 403 here, instead of 200 with an empty list.",
+      "reference": "Is it more reasonable to return some errors here, such as returning error codes (403 and so on?) ?"
+    },
+    {
+      "prediction": "I'm not sure I understand why this needs to be initialized to `undefined` instead of `null`. Can you explain?",
+      "reference": "Thinking out loud... would `_nextDom` be a better name for this?"
+    },
+    {
+      "prediction": "I think we should keep this test case. It's testing that we don't get a false positive when the parens are part of a generator expression.",
+      "reference": "Turns out the unittests also had a false positive."
+    },
+    {
+      "prediction": "I think this should be `CreateInt64Counter` instead of `CreateCounter<Int64>`.",
+      "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": "`getScriptEnv()->getThingByUID(getNumber<uint32_t>(L, -1)) != nullptr` is not the same as `getScriptEnv()->getThingByUID(getNumber<uint32_t>(L, -1))`",
+      "reference": "it pushes a boolean, so it has to be explicitly casted to boolean with != nullptr, same with all other similar stuff"
+    },
+    {
+      "prediction": "I don't think this is the right place to rebuild the instances. It should be done when the publisher is created, not when a writer is added to it.",
+      "reference": "It would be nice to have an equivalent of this call on `DataWriterImpl::enable`"
+    },
+    {
+      "prediction": "I think this should be `DEPLOYMENT_BLOCK_PENDING`",
+      "reference": "Isn't this PENDING?"
+    },
+    {
+      "prediction": "I don't think we need to sanitize the `title` here. `aria.labelVirtual` will sanitize the `aria-label` and `aria-labelledby`.",
+      "reference": "minor detail, why aim to sanitize title if tile is empty (in some cases)? worth adding an && to check for that."
+    },
+    {
+      "prediction": "This import is no longer needed.",
+      "reference": "Code shared by ParentChildrenSyncDownTarget and ParentChildrenSyncUpTarget moved to ParentChildrenSyncTargetHelper"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be the other way around: if `capacity` is set, use it, otherwise use `sequential`.",
+      "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 should be `tf.IntegrationTest(t)`",
+      "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 `return version(mver.GetVersion())` instead of `return version(version(mver.GetVersion()))`",
+      "reference": "Can you provide corresponding UT for this."
+    },
+    {
+      "prediction": "I don't think we need to import `golang.org/x/net/context` here.",
+      "reference": "nit: this should be below the yarpc group"
+    },
+    {
+      "prediction": "I don't think this is the right thing to do. If the error is `AlreadyRegisteredError` it means that the metric is already registered, so we should not register it again. Instead, we should get the already registered metric and use it.",
+      "reference": "singleCaseSwitch: should rewrite switch statement to if statement (from `gocritic`)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this. I think we should be able to do something like `healthAggregator.Report(healthName, &health.HealthReport{Live: false, Ready: false})` in the `typhaConnection.Start` function.",
+      "reference": "Please log once here at Error level \"Failed to connect to Typha, will retry...\""
+    },
+    {
+      "prediction": "I think we can remove the `false` here.",
+      "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'm not sure about this change. I don't think we should allow underscores in names.",
+      "reference": "So we fix it only for kinto and not for all cliquet resources?"
+    },
+    {
+      "prediction": "why change the import order?",
+      "reference": "No empty line between"
+    },
+    {
+      "prediction": "I think this should be done before the `ConfigureLogging` call, so that we don't log to the wrong place.",
+      "reference": "We have `ResolveRootfs` in `libcontainer/utils` so maybe use that one here?"
+    },
+    {
+      "prediction": "I think this should be `LongRunningOperationsWithLongRunningExtensionTrue` or `LongRunningOperationsWithLongRunningExtensionTrueValidation`",
+      "reference": "Just curious, do we have positive test covered somewhere?"
+    },
+    {
+      "prediction": "I don't think this is correct. If `serial` is 0, then we should not set `pending_move_resize_configure_serial` to 0.",
+      "reference": "Style error, put a space between `if` and `(`"
+    },
+    {
+      "prediction": "I think this should be `EnableEndpointHostPrefix` instead of `DisableEndpointHostPrefix`.",
+      "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 think this should be 2 * pullImageTimeout",
+      "reference": "Why did you pick two hours for this?"
+    },
+    {
+      "prediction": "I think this should be \"enter password for new key (new):\" and \"enter password for new key (again):\"",
+      "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 you need to import the whole module, just the `ContextHelpMixin` class.",
+      "reference": "Another redundant import"
+    },
+    {
+      "prediction": "I don't think we need to release the client here. The client will be released when the connection is closed.",
+      "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 do we need to change the type here?",
+      "reference": "What is the difference between GraphType and graph_type. Naming should be more accurate."
+    },
+    {
+      "prediction": "`EntityManagerInterface` is not a repository, it's an entity manager.",
+      "reference": "hmmm in docblock of contructor it is EntityManagerInterface....maybe it should be that way.... :confused:"
+    },
+    {
+      "prediction": "I don't think this is necessary. You can just call `GetPos` with `confId = -1` to get the default position.",
+      "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 should use `I18n.t` instead of `statusLabel`",
+      "reference": "Does this need to change `'active'` to `'Active'` and that? Or is that done in CSS or something?"
+    },
+    {
+      "prediction": "This should be `public const int MaxAllowedMaxFrameSize = 16 * 1024 * 1024 - 1;`",
+      "reference": "Why are these constants public? They should also be listed above members."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. This is the place where we are checking the signature. We have to fix the `v` value if this is a private transaction.",
+      "reference": "This TODO isn't related to this change. We should remove it."
+    },
+    {
+      "prediction": "I think this should be `default Stream<FieldMetrics> metrics() { return Stream.empty(); }`",
+      "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 should keep the `SetMetricProcessor` and `SetMetricExporter` methods on `MeterProviderBuilder` for consistency with `Sdk.CreateTracerProviderBuilder`.",
+      "reference": "@cijothomas do we want this to be `Set` or `Add`?"
+    },
+    {
+      "prediction": "I think it should be 100_000_000",
+      "reference": "Why are we dropping GasCap?"
+    },
+    {
+      "prediction": "I think this should be `nameof(this.Span.Context)`, not `nameof(this.Span.Context.Context)`.",
+      "reference": "should we end the text with a .?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this check. I think it should be in the `if (dcontext->whereami == DR_WHERE_DISPATCH)` block below.",
+      "reference": "Could this instead keep the `go_native` and add to it \"or the last exit was the special reset exit\"?"
+    },
+    {
+      "prediction": "I don't think you need to capture `this` here.",
+      "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 we need to call `kill` twice here.",
+      "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 think this should be `if r.ContentLength == 0 {`",
+      "reference": "when would this happen? backward compatibility?"
+    },
+    {
+      "prediction": "I think this should be `node instanceof HTMLElement` instead of `node.namespaceURI === 'http://www.w3.org/1999/xhtml'`",
+      "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 `mayactl snapshot list` instead of `mayactl snapshot list --volname <vol>`",
+      "reference": "This command displays available snapshots on a volume."
+    },
+    {
+      "prediction": "nit: `stpb` -> `storagepb`",
+      "reference": "`spb` is the standard package name we use for this proto."
+    },
+    {
+      "prediction": "I think this should be `handle` instead of `name`.",
+      "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": "Is this TODO still valid? If not, can we remove it?",
+      "reference": "is this pr support parallelly process application?"
+    },
+    {
+      "prediction": "I think this should be `if ctx.Repo.Repository.CanEnablePulls() && !ctx.Repo.Repository.EnablePulls {`",
+      "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": "Shouldn't this be `projectName` instead of `project`?",
+      "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": "Metrics/LineLength: Line is too long. [121/120]",
+      "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": "I think we should add `@throws \\InvalidArgumentException` to this method as well.",
+      "reference": "There is no docblock on the parent method. Also, not related, but I forgot to typehint the `$templating` constructor argument above."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to handle this, but it works for now.",
+      "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": "nit: I think we can remove the `queryRelation == QueryRelation.INTERSECTS` check and just pass `true` to `relateRangeBBox`",
+      "reference": "Shouldn't this work as well for Disjoint?"
+    },
+    {
+      "prediction": "I think this should be `err` instead of `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 this is necessary. `SocketError` is a subclass of `Errno::ECONNREFUSED` and `Errno::ENOTCONN`.",
+      "reference": "Doesn't this need to include `Errno::ECONNRESET` to fix the issue?"
+    },
+    {
+      "prediction": "I think this should be `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 `ListPowerSelectorEntries`.",
+      "reference": "This is a hard one, naming-wise... `ListPowerSelectorEntries` is logical given `ListSelectorEntries`. Another option could be `ListMatchingEntries` or even `FindMatchingEntries`. :shrug:"
+    },
+    {
+      "prediction": "I don't think we need this type. We can just use `uint32`.",
+      "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": "`pygment` -> `pygmentize`",
+      "reference": "No need for this when you don't need to customize anything about the argument."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could think of to get the first section to be draggable. If there's a better way to do this, please let me know!",
+      "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": "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": "did you mean to leave this in here?"
+    },
+    {
+      "prediction": "I think this should be `--sysprep-windows` instead of `--sysprep_windows` to be consistent with the rest of the flags.",
+      "reference": "Mention that it only applies to Windows. This is kind of implied, but better to be explicit."
+    },
+    {
+      "prediction": "I think you can use `big.NewInt(0).String()` and `big.NewInt(-10).String()` instead.",
+      "reference": "again: it is a bad practice to have some global parameters with such common names for unit test purpose."
+    },
+    {
+      "prediction": "I think this should be `private \\DateTimeInterface $expiresAt = null;`",
+      "reference": "Should we use `\\DateTimeInterface` or `DateTimeInterface` and declaration of `DateTimeInterface` in `use`?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `task-` prefix here.",
+      "reference": "What do you think of moving this stack related constant to the `stack` pkg?"
+    },
+    {
+      "prediction": "I think this should be `if proto.SupportRekeying && (s.AuthAddr != basics.Address{}) { return errors.New(\"nonempty AuthAddr but rekeying not supported\") }`",
+      "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'm not sure why this is necessary, but it seems to be required for the tests to pass.",
+      "reference": "This story also needs the default padding."
+    },
+    {
+      "prediction": "I don't think this change is correct. The `randomState.nextBoolean()` was there for a reason.",
+      "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": "I think this should be `!= 0` as well.",
+      "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'm not sure if this is the right place to put this, but I think it would be nice to have a `K_KEYWORD` enum value. This would allow us to check if a `Kind` is a keyword by doing `kind >= K_KEYWORD && kind < K_KEYWORD + KEYWORD_COUNT`. It would also allow us to remove the `KEYWORD_NONE` enum value.",
+      "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 don't think we need this anymore.",
+      "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": "This should be `return 200` instead of `return 0`",
+      "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 don't think we need to change this. We can just use `iota - 4` here.",
+      "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": "Why do we need to set the encoding to \"proto\" here?",
+      "reference": "should we use the constant above?"
+    },
+    {
+      "prediction": "I think this should be `ASSERT(result && \"bad file handle?\");`",
+      "reference": "I think something like `size_res` would be more descriptive."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this, but I think it would be nice to have a `#to_h` method on this class that returns a hash that can be passed to `Bolt::Event.new`.",
+      "reference": "Oh lol so this was...really more of a bug?"
+    },
+    {
+      "prediction": "I don't think we need to suppress this one.",
+      "reference": "Well, it is providing a platform neutral way of calling Task or TaskEx. We could just call it PlatformTask."
+    },
+    {
+      "prediction": "I think this should be `wlr_xwayland_server_destroy(xwayland->server);`",
+      "reference": "Typo: this should free `xwayland` instead"
+    },
+    {
+      "prediction": "I think you can remove the `rand` import here.",
+      "reference": "Is it ok to use \"math/rand\" instead of \"crypto/rand\" here? Do we need to protect this ID somehow?"
+    },
+    {
+      "prediction": "I think it's better to use `github.com/cihub/seelog/log` instead of `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": "Shouldn't this be `getPersistentId().getAuthority()`?",
+      "reference": "Is there a reason to use `getUNF()` and `UNF` in the same block?"
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it should be `/c$/` instead of `/c/`.",
+      "reference": "Again this seems odd that megacheck cares"
+    },
+    {
+      "prediction": "I think this should be `@service_url` and `@token_file` instead of `:service_url` and `:token_file`.",
+      "reference": "I assume this will resolve to nil if this isn't in the config?"
+    },
+    {
+      "prediction": "I think it would be better to have a `ParseCustomKeyValues(string data, IDictionary<string, string> mappings, bool allowOptionalMappings = false)` method that takes the mappings as a parameter.",
+      "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": "`in_channels (List[int]): The number of input channels per scale.` -> `in_channels (list[int]): The number of input channels per scale.`",
+      "reference": "` per scale.` -> ` per scale.` need to clean the unnecessary blank space"
+    },
+    {
+      "prediction": "Please remove the extra backtick.",
+      "reference": "there's no import for Export"
+    },
+    {
+      "prediction": "Why do we need to register the node API on the UDS server?",
+      "reference": "i don't think we want/need to make the node API available over UDS"
+    },
+    {
+      "prediction": "I think this should be `find` instead of `findById` because it's not guaranteed that the article exists.",
+      "reference": "How about to skip if and returning null and return a result of `ArticleFacade::findById` directly?"
+    },
+    {
+      "prediction": "I don't think we need to override `map` and `filter` methods. They are not used in our codebase.",
+      "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 don't think this is the right way to get the version. I think we should have a `VERSION` constant in `lib/travis.rb` that is set to the value of `git rev-parse HEAD` or something like that. Then we can just use `Travis::VERSION` to get the version.",
+      "reference": "how about \"unknown\" instead of \"fafafaf\"?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it should be in the `api2` package, but I don't think it should be in the `api2` sub-package.",
+      "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, but it seems to work.",
+      "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 do this. I think it should be done in `fast.NewEnvironmentMemoryGenesis` and `fast.NewEnvironmentGenesis`.",
+      "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 this should be `elif metricName == 'negativeLogLikelihood':`",
+      "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 call `parseFunction()`. I think it should be called in `readTokenFull()` instead.",
+      "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": "Why is this needed?",
+      "reference": "Remove the finally?"
+    },
+    {
+      "prediction": "This should be `const wchar_t *fn`",
+      "reference": "Build on AArch64 fails because WCHAR is not defined."
+    },
+    {
+      "prediction": "I think this should be `envName := \"test\"` instead of `envName = \"test\"`.",
+      "reference": "Do we use this const?"
+    },
+    {
+      "prediction": "I think this should be `SetAnnotations(annotations map[string]string) CertificateRequestModifier { return func(cr *v1alpha2.CertificateRequest) { cr.SetAnnotations(annotations) } }`",
+      "reference": "I think this may need to be `SetCertificateRequestAnnotations` as it returns a `CertificateRequestModifier`"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "reference": "it would likely be better to do this in the Identifier.by_scheme_name method itself."
+    },
+    {
+      "prediction": "This should be `case 'Date created (newest first)':` and `case 'Date created (oldest first)':`",
+      "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 think this should be `/ghost/api/(.+)` instead of `/ghost/api/(.+)`",
+      "reference": "Would this break things if Ghost is run in a subdirectory or no?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "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": "This is not used in this file.",
+      "reference": "Is this an oversight for some code in the body?"
+    },
+    {
+      "prediction": "I don't think you need to set the product again here. It's already set in the `onCreate` method.",
+      "reference": "Reason for adding this? I couldn't find where you are retrieving it back."
+    },
+    {
+      "prediction": "I think this should be `h2o_mruby_eval_file(mrb, \"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": "I'm not sure if this is the best way to do it, but it works for now.",
+      "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 think this should be `if (!path || path === '_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 think we can remove `ApplicationID` and `ApplicationName` from this struct.",
+      "reference": "Passing only needed data to make it can be reused by `planpreview` package where there is no deployment data."
+    },
+    {
+      "prediction": "I think this should be `CustomClientAuth` and `CustomServerAuth`.",
+      "reference": "Could we rename those to `CustomClientAuthentication` (same for Router) to remove ambiguity between Authentication and Authorization (permissions)?"
+    },
+    {
+      "prediction": "I don't think we need to add these methods to the interface. They are only used in `MvxFormsViewPresenter` and `MvxFormsPagePresenter`.",
+      "reference": "Is the name \"IMvxFormsViewPresenter\" too close to \"IMvxFormsPagePresenter\"? Perhaps something like \"IMvxFormsNativeViewPresenter\" or \"IMvxFormsPlatformViewPresenter\" or inline with Forms naming \"IMvxFormsOnPlatformViewPresenter\""
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `if (child != null && child !== true && child !== false && typeof child !== 'number')`",
+      "reference": "`typeof child != 'boolean'` maybe?"
+    },
+    {
+      "prediction": "I think this should be `rawErrorData.get(CLASS) == null || rawErrorData.get(STACK_TRACE) == null`",
+      "reference": "Why && changed to || ?"
+    },
+    {
+      "prediction": "I think this should be `InitAddress, err = NewIDAddress(1)`",
+      "reference": "nit: not a big deal right now but spec assigns ID 0 to InitAddress"
+    },
+    {
+      "prediction": "I think this should be `handle` instead of `name`.",
+      "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 this is necessary. `setZoomFactor` is already called by `setZoom`.",
+      "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 if this is the best way to do this, but it seems to work for now. I'm not sure if there's a better way to do this.",
+      "reference": "Is there a benefit to keeping this in RefinementSettings? Or could we move it into its own class in TestUtil?"
+    },
+    {
+      "prediction": "I think this should be `Results()` instead of `Results() <-chan Result`.",
+      "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 think we should keep the throws clause here.",
+      "reference": "why? what if we want to access those methods outside of azkaban-common?"
+    },
+    {
+      "prediction": "I think it's better to rename this method to `CreateEngineRpcModule`",
+      "reference": "Rename file too"
+    },
+    {
+      "prediction": "I think this should be `r.HTTPResponse.Status` instead of `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": "Why are these not final?",
+      "reference": "Why is the `final` being removed here?"
+    },
+    {
+      "prediction": "@abumusamq can you please update the copyright year to 2018?",
+      "reference": "Maybe let's take this opportunity to change easylogging to a git submodule?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems to work.",
+      "reference": "Should we just be generous and elide all extra whitespace? `strings.TrimSpace`"
+    },
+    {
+      "prediction": "This method is not used anymore, so it can be removed.",
+      "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": "I don't think this should be `finest`.",
+      "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 what's going on here, but it seems like the patch needs to be started before `getToken` is called.",
+      "reference": "You should just use `with patch(...):`, so it unpatches it automatically. Also, much simpler."
+    },
+    {
+      "prediction": "I think this should be `_viewsContainer = Mvx.Resolve<IMvxViewsContainer>();`",
+      "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 add this constructor. I think it would be better to add a constructor that takes a `String` and a `NodeList<Expression>` as parameters.",
+      "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 `ptags=[\"Need action\", \"Pending\"]`",
+      "reference": "@jalavik should be have somewhere enum/list/registry of possible tags?"
+    },
+    {
+      "prediction": "I think we should remove the `# pylint: disable=logging-too-many-args` line.",
+      "reference": "why is this pylint disable needed?"
+    },
+    {
+      "prediction": "I think this should be `POLICY_LIBRARY_SYNC_ENABLED = 'true'`",
+      "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's better to put `warmup_iters` in `__init__` instead of `extra_repr`.",
+      "reference": "This type of return (f'xxx') is not recommended, it is recommended to return 'xxx'"
+    },
+    {
+      "prediction": "I think this should be `LOG(THREAD, LOG_INTERP, 2, \"Detected AVX-512 code in use\\n\");`",
+      "reference": "I would make this level 1 or 2."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Is that path in a variable anyway? Not a big fan of hard coding it here."
+    },
+    {
+      "prediction": "nit: remove extra blank line",
+      "reference": "nit: Move `IdentityTemplateLabel` to be under `IdentityTemplate` to match the struct in config_crd.go"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 `if err != nil && err != context.DeadlineExceeded { 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'm not sure this is the right thing to do, but I don't know what else to do.",
+      "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 should avoid wildcard imports.",
+      "reference": "please inline imports"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should keep the default timeout of 30s.",
+      "reference": "Should IptablesPostWriteCheckIntervalSecs be set back to its previous smaller value, if use of the iptables lock is disabled?"
+    },
+    {
+      "prediction": "I don't think it's a good idea to include the package version in the error message. It's not something that the user can do anything about.",
+      "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 should probably return 0 or -1 or something to indicate that the file doesn't exist. I don't think returning the current timestamp is the right thing to do.",
+      "reference": "this seems unrelated to the rest of the PR, although related to the original report"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Would a check for added columns with non-null default be sufficient for aligned format?"
+    },
+    {
+      "prediction": "I think this should be `--compute-service-account` instead of `--compute_service_account` to be consistent with the rest of the 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": "I think it should be `elements/{element}`",
+      "reference": "Why change to element ?? in whole application use plural convention ??"
+    },
+    {
+      "prediction": "I think this should be `tabindex=\"0\"` instead of `tabindex=\"-1\"`.",
+      "reference": "Tiniest nit but this can be `tabindex={-1}`, `width={11}`, `height={11}`"
+    },
+    {
+      "prediction": "I think this should be `itk.Vector[itk.UC,components]`",
+      "reference": "Unrelated to this PR, we should remove this exception. We now require `numpy`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this, but I think it's the only place we can do it.",
+      "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": "Metrics/MethodLength: Method has too many lines. [21/20]",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `dagster_pandas` instead of `dagster.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'm not sure if this is the right place to do this, but I think it would be nice to have a `newBlockRetrievalQueueWithConfig` function that takes a `Config` as an argument. That way, we don't have to call `config.Codec()` and `config.BlockCache()` in two different places.",
+      "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 don't think we need to close the stream here. The error is returned to the caller, and the caller is responsible for closing the stream.",
+      "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": "Why is this here?",
+      "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 `result.success? ? 'success' : 'failure'`",
+      "reference": "This seems reversed... if `success?` is true, wouldn't we use `success`?"
+    },
+    {
+      "prediction": "Do we need this enum? Can't we just use `LogLevel` from NLog?",
+      "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": "Please use the `link_to_unsubscribe` helper here.",
+      "reference": "Should we disallow entry authors to unsubscribe to their entries?"
+    },
+    {
+      "prediction": "I don't think this should be changed.",
+      "reference": "2016 should be 2017."
+    },
+    {
+      "prediction": "`curSchemaSumary == null || microservice.getInstance().getEnvironment().equalsIgnoreCase(\"development\")` -> `microservice.getInstance().getEnvironment().equalsIgnoreCase(\"development\") || curSchemaSumary == null`",
+      "reference": "boolean exists = (existSchema != null) && (existSchema .getSummary() != null);"
+    },
+    {
+      "prediction": "I don't think we need to pass `cmd` to `runPoolDescribe` anymore.",
+      "reference": "It does not look idiomatic."
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "Generally, we do not use newlines between project imports, we only separate them from the stdlib imports"
+    },
+    {
+      "prediction": "I think this should be `if (nowPlayingItem.RunTimeTicks && nowPlayingItem.RunTimeTicks > 0) {`",
+      "reference": "can this long line be wrapped?"
+    },
+    {
+      "prediction": "I think this should be `W0144` to be consistent with `W0142` and `W0143`.",
+      "reference": "I'd use the same message that I suggested in the release changelog."
+    },
+    {
+      "prediction": "nit: add a space after `}`",
+      "reference": "Keep semi-colons at end of function defs please"
+    },
+    {
+      "prediction": "I think this should be `this.explain = options.explain || false;`",
+      "reference": "A little skeptical of `explain` being here. Why is this now a property of all commands?"
+    },
+    {
+      "prediction": "Is this change intentional?",
+      "reference": "License copied two times!!"
+    },
+    {
+      "prediction": "I think this should be `match` instead of `matcher` to be consistent with the rest of the codebase.",
+      "reference": "... their **superclass**"
+    },
+    {
+      "prediction": "I think this should be `|| 1`",
+      "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 don't think this is correct. If the dependencies are loading, then the context will be `undefined`. If the dependencies are loaded, then the context will be `null` if there is no registered context string for the given module. If the dependencies are loaded and there is a registered context string for the given module, then the context will be the registered context string, screenWidgetContext for the given module.",
+      "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 if this is the best way to do this, but it seems to work.",
+      "reference": "Layout/EmptyLinesAroundBlockBody: Extra empty line detected at block body beginning."
+    },
+    {
+      "prediction": "I don't think we need to create a new exception for this. Let's just use `InvalidElementStateException`.",
+      "reference": "We keep who wrote the code anonymous."
+    },
+    {
+      "prediction": "I think this should be `log.Debug` instead of `log.Info`",
+      "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 to set the profile ID. I think it should be set in the `setPropertyID` action.",
+      "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 `REQUIRE(m->getNumAtoms() == 14);`",
+      "reference": "It might be reasonable to add a check that sanitization still fails here."
+    },
+    {
+      "prediction": "I don't think this is correct. If `attemptNo` is 0, then we should not print \"retry: 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": "Align the elements of a hash literal if they span more than one line.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Yuck. Rails bug."
+    },
+    {
+      "prediction": "I think this should be `l = l.payload.getlayer(0, subclass=True)`",
+      "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 think this should be `http.StatusAccepted` instead of `http.StatusContinue`.",
+      "reference": "DRY these 5 lines into a function call, no?"
+    },
+    {
+      "prediction": "I think this should be `logger.Error` instead of `logger.Log`.",
+      "reference": "should we be trying to close here? or just return directly @fabxc ? same goes for `stopc`..."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [123/120]",
+      "reference": "just re-ordered for clarity"
+    },
+    {
+      "prediction": "I think this should be `fmt.Printf(\"Volume: %s not found in namespace: %q error: %s\\n\", volname, namespace, http.StatusText(resp.StatusCode))`",
+      "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.IndexByteString`.",
+      "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 don't think this change is needed.",
+      "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 `$this->content` instead of `$contend`",
+      "reference": "typo in `$contend`"
+    },
+    {
+      "prediction": "I think we can remove the `import org.apache.hadoop.fs.Path;` as it is not used in this file.",
+      "reference": "Nit: empty line"
+    },
+    {
+      "prediction": "I don't think we should panic here. We should return an error instead.",
+      "reference": "I figured this was more fool-proof than trying to plumb the error up from everywhere."
+    },
+    {
+      "prediction": "I think this should be `&c.srv.logging.debug` and `&c.srv.logging.trace`",
+      "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(AppRoot, \".ddev\", \"config.yaml\")` instead of `app.ConfigPath = app.GetConfigPath(\"config.yaml\")` because `app.GetConfigPath(\"config.yaml\")` calls `app.ConfigPath = filepath.Join(AppRoot, \".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 removed?",
+      "reference": "Should BuildBase be removed?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `internal_test` suffix from the file name.",
+      "reference": "this was a bad Goland auto-import that I missed in the previous PR. Thanks to @rkowalick and @acruikshank for catching it"
+    },
+    {
+      "prediction": "I don't think you need to do this. `MolOps::removeHs(m2);` should delete the `QueryAtom`.",
+      "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'm not sure this is the right place for this function. I think it should be in `wlr_output.c` or `wlr_output_private.h`.",
+      "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": "Style/AccessorSyntacticSugar: Use attr_reader instead of defining the annualized_savings method.",
+      "reference": "I don't see tests for this (or the similar method on `Team`)."
+    },
+    {
+      "prediction": "I don't think you need to cast to `Long` here. `parameters.get(\"type\")` is already a `Long`.",
+      "reference": "should use Number instead of Long"
+    },
+    {
+      "prediction": "I think this should be `if err != nil && err != ErrMaxPayload && err != ErrAuthorization {`",
+      "reference": "We had the checks for those two errors for a reason, why do we not need them anymore?"
+    },
+    {
+      "prediction": "This import is not used in this file. Please remove it.",
+      "reference": "Why didn't you just switch to `range` here too?"
+    },
+    {
+      "prediction": "I think we should be able to pass in a `*http.Client` instead of a `*gcp.HTTPClient` here. The `*gcp.HTTPClient` is just a wrapper around an `*http.Client`, so we should be able to pass in the `*http.Client` directly.",
+      "reference": "I don't even think `nil` should be allowed, since it uses global state to get the client."
+    },
+    {
+      "prediction": "I don't think the backticks are needed here.",
+      "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": "I think this should be `if (delta_from_revision && !g_str_equal (delta_from_revision, to_revision))`",
+      "reference": "Minor, can we drop the leading `!` and just swap the if and the else blocks?"
+    },
+    {
+      "prediction": "why do we need to check the number of proposals is not 0?",
+      "reference": "`bbox_feats.shape[0] > 0` requires the number of proposal is not 0."
+    },
+    {
+      "prediction": "I'm not sure why this was changed. I don't think it should be removed.",
+      "reference": "Just FYI, I had to remove this unused import because Travis/`npm run lint` were complaining."
+    },
+    {
+      "prediction": "I don't think you need to include `fmt/format.h` here.",
+      "reference": "format.h is not a system header but bundled with LAMMPS. This should be `#include \"fmt/format.h\"`"
+    },
+    {
+      "prediction": "nit: I think it would be better to have a single `visitDocValues` method that takes a `boolean cardinalityLeaves` argument.",
+      "reference": "could we always call visitDocValuesWithCardinality? It seems to include the version check already?"
+    },
+    {
+      "prediction": "I think this should be `if (port == -1) { throw new GridConfigurationException(\"No port was specified in -hub parameter. Example - http://\" + hub + \":4444/grid/register\"); } else if (port == 0) { throw new GridConfigurationException(\"Invalid port was specified in -hub parameter. Example - http://\" + hub + \":4444/grid/register\"); } else if (port < 0 || port > 65535) { throw new GridConfigurationException(\"Invalid port was specified in -hub parameter. Example - http://\" + hub + \":4444/grid/register\"); }`",
+      "reference": "I think if they _don't_ specify a port, we should assume grid default (which is 4444)"
+    },
+    {
+      "prediction": "`res.tautomers` is a `std::vector<ROMOL_SPTR>`, not a `std::vector<TautomerEnumerator::Tautomer>`.",
+      "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 this should be `if len(m.Bucketlist) == 0 { return \"Empty bucketlist with given address\" }`",
+      "reference": "Perhaps remove this line"
+    },
+    {
+      "prediction": "I think this should be `if (!ps.removeHydrides && atom->getAtomicNum() == 1 && atom->getFormalCharge() == -1 && atom->getDegree() == 0)`",
+      "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 don't think it's a good idea to make the node sleep for a long time.",
+      "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 why we need to sort the CommRs here. I think we can just pass them in the same order as we get them from the sector store.",
+      "reference": "This 32 could go away too."
+    },
+    {
+      "prediction": "I'm not sure if we want to pass this as a prop. I think it would be better to have it as a constant in the component.",
+      "reference": "These props should be less specifically named since the component is generic, for example `anchorLink`, `anchorLinkLabel`."
+    },
+    {
+      "prediction": "I think this should be `Unable to find EC2 instance matching deleted AWSMachine`",
+      "reference": "In the future, it'd be great if we can separate commits or PRs that aren't in the same scope"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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 don't think we need to add the `ANSIBLE_CONFIG` environment variable here. The `ansible-galaxy` command will automatically pick up the `ansible.cfg` file in the current working directory.",
+      "reference": "Can we write a unit test for this."
+    },
+    {
+      "prediction": "I don't think this is the right place to save the session. The session should be saved when the user closes the tab, not when the page is loaded.",
+      "reference": "Same as above, use `.delete('_autosave')`, not the command handler."
+    },
+    {
+      "prediction": "I think we can remove the `projectsIdentityPartitionColumns` variable now.",
+      "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": "This should be `Command.W3C_MINIMIZE_WINDOW`",
+      "reference": "Update after command rename"
+    },
+    {
+      "prediction": "This import can be removed.",
+      "reference": "@KroArtem This is the Checkstyle violation. This import should be grouped with other `net.sourceforge.pmd` imports below, in alphabetical order"
+    },
+    {
+      "prediction": "This should be `rclone/fs` not `github.com/ncw/rclone/fs`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I don't think we need this method. We can just call `mHeaderContainer.setOnFlagListener(listener);` directly.",
+      "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 `WL_KEYBOARD_KEYMAP_FORMAT_NO_KEYMAP`",
+      "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 think this should be `yarpcconfig.MaxMsgSize` instead of `yarpcconfig.ServerMaxRecvMsgSize` and `yarpcconfig.ClientMaxRecvMsgSize`.",
+      "reference": "might be good for the units (bytes?) to be in the variable name."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get this to work.",
+      "reference": "Metrics/BlockLength: Block has too many lines. [166/25]"
+    },
+    {
+      "prediction": "I'm not sure this is correct. I think it should be \"for each field, the name, type, and a special indicator for those fields that serve as the reset field, sequence id field, and timestamp field.\"",
+      "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": "why remove this empty line?",
+      "reference": "These empty lines can be kept."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should be asserting that the output is exactly what we expect, not just that it's not empty.",
+      "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 don't think we need to make this a global variable. We can just use `TraceContext{}` as the default propagator.",
+      "reference": "It is better to have a method than a var."
+    },
+    {
+      "prediction": "I don't think we need to change these constants. They are only used in tests.",
+      "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": "Please remove the `./` in the `work_dir`.",
+      "reference": "List is redundant."
+    },
+    {
+      "prediction": "I don't think you need to check for `null` here. `getSupportActionBar()` will never return `null` if `setSupportActionBar()` has been called.",
+      "reference": "Please use a string resource here rather than the hardcoded string \"user\". This allows the app to be multilingual."
+    },
+    {
+      "prediction": "remove unused imports",
+      "reference": "@aswathkk Why did we remove this import? I guess we are using it."
+    },
+    {
+      "prediction": "Why is this no longer needed?",
+      "reference": "Why? I thought this is for graceful shutdown."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we should print empty stdout/stderr.",
+      "reference": "Can `result.stdout` or `result.stderr` be nil?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "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 `MaxIPSetSizeBytes` to make it clear that it's in bytes.",
+      "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 `messages.delete_attachment_singular`",
+      "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 don't think we need to throw an exception here. I think we can just return null.",
+      "reference": "why not be consistent with `isNull` and `notNull` and return null?"
+    },
+    {
+      "prediction": "This should be `yaml:\"dockerLabels,flow\"`",
+      "reference": "Did we forget to remove this field?"
+    },
+    {
+      "prediction": "I think this should be `this.values().stream().filter(Objects::nonNull).mapToInt(elem -> elem.elementCount()).sum();`",
+      "reference": "What if there is no element?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it would be good to add a mutex to the `Client` struct and lock it here. This will prevent multiple concurrent builds of the same target from uploading the same inputs multiple times.",
+      "reference": "I think you need to be careful to honour isTest and stamp here."
+    },
+    {
+      "prediction": "I don't think you need to import `com.google.api.codegen.viewmodel.SimpleParamDocView` here.",
+      "reference": "why zero value for `arrayType` for the default value of the optional args?"
+    },
+    {
+      "prediction": "I think this should be `return $this->render();` instead of `return ['.control-filter' => $this->render()];`",
+      "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 `cniNetConfig` and `cniIpamConfig`",
+      "reference": "I've never seen us do this anywhere else - why here? Why not `address, meshAddress string`?"
+    },
+    {
+      "prediction": "nit: I think it would be better to use `JsonQueryConverter.class` instead of `JsonQueryConverter.class.getName()`.",
+      "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 `getBackendConfig(): array`",
+      "reference": "Same here than [there](#r26168925)"
+    },
+    {
+      "prediction": "I don't think we should allow dots in the application id.",
+      "reference": "Again we should think about moving this pattern to a static final field"
+    },
+    {
+      "prediction": "I don't think this is the right place to inject the `controller_resolver` service. The `NormalizerConfigPass` should not depend on the `ControllerResolver` service.",
+      "reference": "haven't been implemented the `__constructor` for `NormalizerConfigPass` ?"
+    },
+    {
+      "prediction": "Wrap dict fields into lists -> Wrap dict fields to lists",
+      "reference": "Docstring is missing."
+    },
+    {
+      "prediction": "This should be `if (!success || !registerEvent(event, node)) { delete event; return false; }`",
+      "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 this should be `AddPeer` instead of `AddPeers`",
+      "reference": "change from AddPeer to AddPeers added a space in some occurrences"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "nit: `NewIPTableRules` -> `NewIPTableRulesClient`",
+      "reference": "Do you remove \"s\" from \"iptables\" to IPTable\" on purpose? I feel it should be \"IPTablesRules\"."
+    },
+    {
+      "prediction": "I think we can remove the `PrivacyApiMethod` superclass here as it's no longer used.",
+      "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 think this should be `GetMeterProvider().Meter(instrumentationName, opts...)` instead of `otel.GetMeterProvider().Meter(instrumentationName, opts...)`",
+      "reference": "These changes have been resolved upstream. We can revert changes to this file when synced."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this docstring. I think it should be on the `Partitioned` class, and the `Partition` class should have a docstring that describes what a `Partition` is in the context of a `Partitioned` object.",
+      "reference": "The description predates this diff but I think we can come up with something more grounded."
+    },
+    {
+      "prediction": "I think this should be `VolumeName string `yaml:\"volumeName\"` instead of `VolumeName string `yaml:\"volumeName\"`",
+      "reference": "Let the Kind, APIVersion & Metadata follow the rules followed by Volume. What is the reason for a change for snapshot struct ?"
+    },
+    {
+      "prediction": "I think it's better to use `com.netflix.spectator.api.Registry` instead of `com.netflix.spectator.servo.ServoRegistry`.",
+      "reference": "seems that we did not strong depend servo no need to keep it? change to another one?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I think it's the best we can do for now.",
+      "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 think we can remove the `testSuites.forEach(testSuite => {` and just use `for (const testSuite of testSuites) {`",
+      "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.buildArgsToString(buildConfig.buildArgs())`",
+      "reference": "this is wrong and should be reverted"
+    },
+    {
+      "prediction": "I think this should be `runtime.GOMAXPROCS(0)` 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 rename this class to `SyncManagerTest`",
+      "reference": "Unrelated typo fix."
+    },
+    {
+      "prediction": "I think this should be `user, _ := r.request.Context().Value(RemoteUserCtxKey).(string)`",
+      "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 `http://${ hostname }:6006/`",
+      "reference": "Why was this changed?"
+    },
+    {
+      "prediction": "`std::string& value` -> `std::string* value`",
+      "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'm not sure if this is the right thing to do, but I think it makes sense to use the `email` from the `@current_user` if it exists, otherwise use the `email` from the `@invitation`.",
+      "reference": "Use the return of the conditional for variable assignment and comparison."
+    },
+    {
+      "prediction": "`AnyKey` should be imported from `selenium.webdriver.common.keys` instead of `selenium.types`.",
+      "reference": "just realised... This breaks things since there is not selenium types module, is there a PR for this?"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "move to line 23 below"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "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 is not None:`",
+      "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 if this is the right place to put this, but I think it would be nice to have a `ResubscriptionService` class that takes care of all the logic for creating a resubscription. This way, we can have a `ResubscriptionController` and a `ResubscriptionService` and we don't have to worry about the controller having too much logic in it.",
+      "reference": "1 trailing blank lines detected."
+    },
+    {
+      "prediction": "I think this should be `require(\"./pv-tests\")` instead of `node_require(\"./pv-tests\")`",
+      "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 think this should be `if (parameters.Count() > 0 && bundle == null)`",
+      "reference": "This check looks odd to me. I can't quite grok it. Why is it needed now? What's changed?"
+    },
+    {
+      "prediction": "I think this should be `s.enableJetStream(&cfg)`",
+      "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 think we can remove the `session` variable and just call `session.endSession()` in the `finally` block.",
+      "reference": "is this necessary? doesn't `client.close()` imply all sessions will be ended?"
+    },
+    {
+      "prediction": "nit: please remove this extra blank line",
+      "reference": "Extra line, it's not needed here."
+    },
+    {
+      "prediction": "I think this should be `Bolt::PAL.load_puppet` instead of `Bolt::PAL.load_puppet`",
+      "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 don't think we need to create a new `StringBuilder` here. We can just use the existing one.",
+      "reference": "or is it `i < length - back.length()`? `\"12345\".padTo(10, 'a')` should be `12345aaaaa`"
+    },
+    {
+      "prediction": "`cfg.get('custom_hooks', None)`?",
+      "reference": "custom_imports -> custom_hooks"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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 don't think we need to pass `preload` to `openInput`. It's only used in `MMapIndexInput`, which is only used in `MMapDirectory`. So we can just make `MMapDirectory.openInput` call `MMapIndexInput.openInput` with `preload` set to `true`.",
+      "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 don't think it's a good idea to move MurmurHash2 from common to base.",
+      "reference": "This has to be in front of all other includes"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be `pStart->calcImplicitValence(true);` and `pEnd->calcImplicitValence(true);`",
+      "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": "Why is this change necessary?",
+      "reference": "`window.self === window`"
+    },
+    {
+      "prediction": "I think this should be `&workspace.ErrFileNotExists{FileName: wantedSvcName}`",
+      "reference": "maybe we should keep one test case for ErrFileNotExists and update only one file for ErrWorkspaceNotFound?"
+    },
+    {
+      "prediction": "I don't think we need to force flush here, since we are shutting down the exporter anyway.",
+      "reference": "Please add `ConsigureAwait(false)` here and on `ShutdownAsync` below."
+    },
+    {
+      "prediction": "I don't think this is correct. `augvar` and `vardel` are used in the body of `bad_default`, so they should not be flagged as `unused-variable`.",
+      "reference": "Isn't this a false positive? Why are we reporting `unused-variable` on a `del` operation with a `undefined-variable`."
+    },
+    {
+      "prediction": "I think it should be `SessionStats` instead of `SessionDto`",
+      "reference": "Is it correct that `SessionDto` should be here?"
+    },
+    {
+      "prediction": "nit: add a space after `=`",
+      "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 this should be `false` instead of `'false'`",
+      "reference": "Make this consistent with the other definition please `'false'` vs `false`"
+    },
+    {
+      "prediction": "I think this should be `os.Exit(code)`, not `Exit(code)`.",
+      "reference": "I think this is a good way to do it. :+1:"
+    },
+    {
+      "prediction": "I don't think we need to change the import here. We can keep it as `github.com/mitchellh/go-homedir` and just use `homedir` as an alias.",
+      "reference": "Why this is needed?"
+    },
+    {
+      "prediction": "nit: I think it would be better to pass `ovsctlClient` as an argument to `NewConnTrackOvsAppCtl` instead of creating it inside the function.",
+      "reference": "Can we define this directly in the function `NewConnTrackOvsAppCtl` instead of passing this as an argument?"
+    },
+    {
+      "prediction": "I think this is a bug in unparam. I don't think the first argument is ever used.",
+      "reference": "False positives are annoying..."
+    },
+    {
+      "prediction": "I think it should be `string $name, int $iterationIndex, string $extension): 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 ulong _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 `exc_value.__class__.__name__` instead of `exc_type.__name__`.",
+      "reference": "feel free to make this a \"typed\" named tuple (overriding __new__) if you are feeling frisky"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way to make this work in IE11.",
+      "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 `liveManifests, headManifests`",
+      "reference": "The bug was caused by this order change in the last refactoring."
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "reference": "This is already imported as docker. You're safe to simply use `docker.utils.kwargs_from_env()` below."
+    },
+    {
+      "prediction": "I don't think we need this method. We can just call `AddAction` directly on the `RunnableActions` object.",
+      "reference": "No need to add this fun. Using RunnableActionsBuilder#AddActions instead"
+    },
+    {
+      "prediction": "I don't think it's a good idea to move hooks into core. The hooks are not part of the core, they are part of the compat layer. If we move them into core, we will have to maintain them in core, which is not a good idea. I think it's better to keep the hooks in the compat layer, and if we want to move them into core, we should do it in a separate PR, and only if we are sure that it's a good idea to move them into core.",
+      "reference": "I don't think it's possible to do this for hooks"
+    },
+    {
+      "prediction": "I'm not sure why we need to add 1 to the wait time.",
+      "reference": "wait(1ms) when sec=0. wait(0ms) waits forever so that's why this is better."
+    },
+    {
+      "prediction": "nitpick: I think it would be better to use `socklen_t` instead of `int` for `domain` and `so_reuseport`.",
+      "reference": "Do we need `domain` and `so_reuseport` now that we open new sockets immediately after calling `add_listener`?"
+    },
+    {
+      "prediction": "nit: remove the extra space after `* Closes the connections associated with this client. `",
+      "reference": "The formatting seems different from the rest of the code"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this function. I think it should be in `pkg/cluster/cluster.go` or `pkg/cluster/cluster_test.go` since it's used in `pkg/cluster/cluster.go`.",
+      "reference": "What about calling this package \"parameters\"? utils\\parameters are utils related to parameters."
+    },
+    {
+      "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 don't think we should change the order of the options in the help message.",
+      "reference": "what is the actual difference here?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think you should use `UriComponentsBuilder.fromUriString(registrationId).build().getQueryParams()` to get the query parameters.",
+      "reference": "Parsing URLs is hard. Any fixes should avoid manually parsing the URL."
+    },
+    {
+      "prediction": "I think this should be `@group_level -= 1 if @group_level > 0`",
+      "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 we can remove the `if (value_type)` check here.",
+      "reference": ".GetValueOr(0) ... but really, shouldn't m_value_strife also be optional?"
+    },
+    {
+      "prediction": "I think this should be `Record<string, string | undefined>`",
+      "reference": "Working around a possible limitation in the JSDoc plugin's typescript syntax checking. This means the same thing"
+    },
+    {
+      "prediction": "I think we should make these `readonly` and initialize them in the constructor.",
+      "reference": "Should be readonly"
+    },
+    {
+      "prediction": "I think this should be `from rdkit.six.moves import cPickle as pickle`",
+      "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 think this should be `Copyright 2016-2020 The Kythe Authors. All rights reserved.`",
+      "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 don't think this is the right change. The `allowedOrigins` parameter should be set to `*` and the `allowedMethods` parameter should be set to `GET,POST,PUT,DELETE,HEAD`.",
+      "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 `Preconditions.checkNotNull(warehouseLocation, \"Warehouse location is not set: hive.metastore.warehouse.dir=null\");`",
+      "reference": "Style: indentation should be 4 spaces (2 indents) from the start of `Preconditions`."
+    },
+    {
+      "prediction": "I think this should be `if (in_array($metadata['associationType'], $toManyAssociations)) {` instead of `if ('association' === $metadata['type'] && in_array($metadata['associationType'], array(ClassMetadataInfo::ONE_TO_MANY, ClassMetadataInfo::MANY_TO_MANY))) {`",
+      "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": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Probably want something more generic here like 'DMPRoadmap' so that other installations aren't using UoE by default."
+    },
+    {
+      "prediction": "I think this should be `InternalLogger.Error(exception, \"Failed to add type '{0}'.\", t.FullName);`",
+      "reference": "Must this not change to `MustBeRethrownImmediately`?"
+    },
+    {
+      "prediction": "I think we should keep the `client` and `lastTf` variables, but remove the `graph` variable since it's not used anymore.",
+      "reference": "Can it be pointer?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Not related to this issue but this condition seems weak."
+    },
+    {
+      "prediction": "I think this should be `if (isset($this->mainConfig->Record->preferredMarcField) && array_key_exists($this->mainConfig->Record->preferredMarcField, $this->fields) && $this->mainConfig->Record->preferredMarcField != 'fullrecord')`",
+      "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": "nit: I think we can remove the `np = import_numpy()` line and just use `import_numpy()` directly.",
+      "reference": "does this still allow this file to be used with Python installations that don't have numpy?"
+    },
+    {
+      "prediction": "I think this should be `\\\\/v([a-z0-9]+);` instead of `\\\\/v([a-z1-9]+);`",
+      "reference": "what about 0? Version probably can' start with 0, but v10 is theoretically possible."
+    },
+    {
+      "prediction": "Can we use `Path.GetFullPath` instead of `Directory.GetCurrentDirectory` and `Path.Combine`?",
+      "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: \"Init\"`",
+      "reference": "Do we need to use quote? Why not `status: Init`"
+    },
+    {
+      "prediction": "This function is not used anymore, so it can be removed.",
+      "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 this should be `assert.Equal(t, getClusterRegion(cd), \"us-east-1\", \"incorrect cluster region label\")`",
+      "reference": "Remove this line."
+    },
+    {
+      "prediction": "I'm not sure why this was needed.",
+      "reference": "This is no longer called in any of our unit tests, right?"
+    },
+    {
+      "prediction": "I think this should be `uint16` 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 the right way to do it, but I think we need to add an index on `grant_id` so that we don't have to do a full table scan when we're trying to find all the plans for a given grant.",
+      "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": "This should be `WinImageSVG` instead of `WinImageSVG`.",
+      "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 think this should be `ECS_ENABLE_CONTAINER_METADATA_FILE` instead of `ECS_ENABLE_CONTAINER_METADATA`",
+      "reference": "This is not checked below?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be in the `Stack` struct, not in the `CloudFormation` struct. The `CloudFormation` struct should only be responsible for talking to the CloudFormation API.",
+      "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 don't think we need to import `get_package_registry` from `backends` anymore.",
+      "reference": "minor: at this scale, one import per line reads better"
+    },
+    {
+      "prediction": "Layout/AlignHash: Align the elements of a hash literal if they span more than one line.",
+      "reference": "Should remove the default templates as well."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Why not `let`?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. We should be able to rely on Spring Boot for this.",
+      "reference": "ns -> DNS My first reading was this is referring to System.nanoTime and thought hrm?"
+    },
+    {
+      "prediction": "I think this should be `check.opt_int_param(start_timestamp, \"start_timestamp\")`",
+      "reference": "is there a reason not to remove the param now? I think the serdes will still work?"
+    },
+    {
+      "prediction": "Shouldn't this be 3.321928094887362?",
+      "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 it would be more useful to log the name of the project that was not found in the cache, rather than just saying \"Project doesn't exist in cache\".",
+      "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.math import GetNTAReal, SparseMatrix, NTAReal`",
+      "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 don't think we need to import `ResourceChanged` and `ACTIONS` here.",
+      "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 don't think we need this method. We can just use `Truncate.withLength(16)` directly.",
+      "reference": "I'd probably move this into `MetricsConfig` as a private method. Seems like we only use it there."
+    },
+    {
+      "prediction": "I don't think we need to change the docstring here.",
+      "reference": "Not clear here what you mean with this list"
+    },
+    {
+      "prediction": "I don't think we need to suppress the checkstyle rule here.",
+      "reference": "Curious, is there a way we can get around this without suppressing?"
+    },
+    {
+      "prediction": "I think this should be `sessionInstance.ProposalID = manager.currentProposal.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 think this should be `RequestParsingData.InnerLoopCount` instead of `RequestParsingData.InnerLoopCount * RequestParsingData.Pipelining`.",
+      "reference": "Can't have two benchmarks with `Baseline = true`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be in the `tests/e2e/api.js` file, as it's specific to the e2e tests.",
+      "reference": "Same here: Empty line before and no `Internal dependencies` docblock (mandatory anyway once #217 lands)."
+    },
+    {
+      "prediction": "I think we should keep the default value here.",
+      "reference": "sorry, the default value for this field is no longer allowed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we should be using `Path.GetTempPath()` instead of `Environment.GetFolderPath(Environment.SpecialFolder.ApplicationData)` here. `Path.GetTempPath()` returns the path to the current user's temporary folder, which should be writable by the current user. `Environment.GetFolderPath(Environment.SpecialFolder.ApplicationData)` returns the path to the current user's application data folder, which may not be writable by the current user.",
+      "reference": "Why does this work? I'd've expected that an admin could create a file anywhere."
+    },
+    {
+      "prediction": "I think this should be `config[:platform] || 'AnyCPU'`",
+      "reference": "falling back to `x64` seems like a really bad idea given that the VS templates default to x86 these days."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but it seems to be the only way to get the tests to pass.",
+      "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. `node.as_string().startswith('from .')` will be `True` for `from .foo import bar`, but `importedname` will be `'foo'`, not `'.foo'`.",
+      "reference": "Check modname instead, but only if it's a ImportFrom."
+    },
+    {
+      "prediction": "I think this should be `write.metadata.previous-log-max-size` or `write.metadata.previous-log-max-size-bytes` to make it clear that this is about the size of the previous log, not the number of entries.",
+      "reference": "How about `write.metadata.previous-versions-max`? No need to refer to these as a log."
+    },
+    {
+      "prediction": "I think this should be `config.filter_run_excluding windows: true if ENV['BOLT_WINDOWS']`",
+      "reference": "We generally do this via tags in the Rakefile, not separate environment variables."
+    },
+    {
+      "prediction": "This should be `Command.W3C_MINIMIZE_WINDOW`",
+      "reference": "Update after command rename"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "reference": "`if X not in Y:`"
+    },
+    {
+      "prediction": "I think this should be `metastore.hiveConf().getIntVar(HiveConf.ConfVars.METASTORETHRIFTCONNECTIONRETRIES)`",
+      "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 `Bolt::Logger.lower_level?(configured_level, :notice)`",
+      "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": "I think this should be `address_string.erase (pos, 7);` instead of `address_string.replace (pos, 7, \"\");`",
+      "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 don't think we need to log this at all.",
+      "reference": "This is wildly unhelpful to users --- they need to know when a proxy has been registered."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think we should validate that the image is not too large.",
+      "reference": "Why do we need `do_not_validate_attachment_file_type` is we're doing it in the previous line?"
+    },
+    {
+      "prediction": "I think this should be `if (!shouldUseCertBasedAuth())` so that we don't call `choosePrivateKeyAlias` when we don't need to.",
+      "reference": "Loading login page right away only for the regular use case."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these constants. I think they should be defined in `src/constants.js` and exported from there.",
+      "reference": "IIRC this flag disables re-rendering entirely (`s/recursively //`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should be able to go back to the dashboard after finishing the wizard.",
+      "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 was this removed?",
+      "reference": "FYI, this returned `StringValues` which is a value type (aka can never be null)."
+    },
+    {
+      "prediction": "I think this should be `sizeHint, err = c.parseConvert(expr.Reserve.Type(), types.Typ[types.Int], sizeHint, expr.Pos())`",
+      "reference": "Oh no, that's a bug."
+    },
+    {
+      "prediction": "I think this should be `ApplicationError error` 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 think it would be better to call this `mappedName` or something like that. `mapTo` sounds like it's a function that takes a string and returns a string.",
+      "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 think we can remove this now.",
+      "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": "Why do we need to add the table name here?",
+      "reference": "We can use `getQualifiedKeyName` here instead. I will update."
+    },
+    {
+      "prediction": "`x-tar` and `x-gzip` are not archive types.",
+      "reference": "Could you clarifty why some of those are removed?"
+    },
+    {
+      "prediction": "I don't think we need to implement this in the fake.",
+      "reference": "`ctx` is unused in ListDeploymentConfigTemplates"
+    },
+    {
+      "prediction": "I don't think we need the `&nbsp;` here.",
+      "reference": "nice to see the usage of block for a more readable link name"
+    },
+    {
+      "prediction": "I don't think you need to cast to `unsigned long long` here. `bytes_to_read` and `bytes_read` are both `int64_t`.",
+      "reference": "Can these use `static_cast<unsigned long long>` instead? Other than that, it's fine."
+    },
+    {
+      "prediction": "I don't think we should remove this. It's useful for debugging.",
+      "reference": "This output is only present when `DEBUG=True`. Does it really bother so much?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we need to make sure that we don't call `receive_confirmed` on the same block twice. I think the best way to do this is to move the call to `receive_confirmed` into the `block_confirmed` callback. That way we can guarantee that we don't call `receive_confirmed` on the same block twice.",
+      "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": "I think this should be `MMFF::Utils::calcUnscaledVdWWellDepth(mmffVdWParams.R_ij_starUnscaled, mmffVdWParamsIAtom, mmffVdWParamsJAtom);`",
+      "reference": "RDKit::Utils is now in the namespace for localeswitcer... We could change it to something else."
+    },
+    {
+      "prediction": "I don't think we need to flatten the tuple here. The `order` is used to reorder the columns of the result, and we don't need to flatten the tuple here.",
+      "reference": "@itholic, can we fix it in `_normalize_keyword_aggregation`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best place for this constant, but I couldn't think of a better place to put it.",
+      "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 think this should be in the constructor of the `By` class.",
+      "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 `optionNameClefSignerEnable` and `optionNameClefSignerEndpoint`",
+      "reference": "do we need both flags? maybe infer that `clef-signer-enabled` whenever `clef-signer-endpoint` is defined?"
+    },
+    {
+      "prediction": "I think this should be `\\\\.(css|scss)$`",
+      "reference": "Do we need this? I didn't see css files (only scss)."
+    },
+    {
+      "prediction": "I think this should be `if settings.REMOTE_EXCLUDE_LOCAL and 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": "I'm not sure if this is the right place to put this, but I think it would be nice to have a link in the email that allows the user to opt out of annual billing.",
+      "reference": "Align the parameters of a method call if they span more than one line."
+    },
+    {
+      "prediction": "I think this should be renamed to `OptionsFacetPackage` to make it more clear that this is a facet package.",
+      "reference": "`@package *` can be removed as well."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 remove this?",
+      "reference": "Where is the other place?"
+    },
+    {
+      "prediction": "I think we can just use `list` instead of `UserList` here. `UserList` is a subclass of `list` anyway.",
+      "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": "I'm not sure this is the right place for this helper. I think it should be in `kudos_helper.rb` since it's related to kudos.",
+      "reference": "We can DRY the above three functions"
+    },
+    {
+      "prediction": "I think it would be better to use `Collections.toListAndThen(HashMap::new, HashMap::ofEntries)` here.",
+      "reference": "Only no good is good code. Nice that you were able to remove all this duplicate stuff!"
+    },
+    {
+      "prediction": "I don't think we need to remove the backslash here.",
+      "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 this is the right place for these interfaces. I think they should be in the `storage` package.",
+      "reference": "these interfaces are defined twice. we should use only one and have it in `storage` package maybe"
+    },
+    {
+      "prediction": "I don't think we need to make this global. I think it would be better to pass it in as an argument to the constructor.",
+      "reference": "This is the wrong context for usage of `global` keyword. It need only be used inside functions."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think it should be `if (targetType.isAssignableFrom(child.getClass()) && !child.getClass().equals(targetType)) {`",
+      "reference": "this change should be described in the changelog. I'll update it when merging if nothing else arises."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this, but I think it would be nice to have a `CreateDispatcherForOutbound` method that takes a `yarpc.Outbound` as an argument. This would allow us to create a dispatcher for an outbound without having to know the name of the service or the name of the host.",
+      "reference": "Just to stay consistent with naming let's call it CreateGRPCDispatcherForOutbound. Also rename 'CreateDispatcherForOutbound' to 'CreateTChannelDispatcherForOutbound'"
+    },
+    {
+      "prediction": "I think this should be `INFO` level.",
+      "reference": "This will always need to be displayed to users."
+    },
+    {
+      "prediction": "I think this should be `bin_safe(s, encoding='utf-8')` to be consistent with the rest of the code.",
+      "reference": "This looks very much like `repr` now (we miss 127 though) - maybe just do `repr(x)` + strip outer parantheses?"
+    },
+    {
+      "prediction": "I don't think we need to set visibility time here. The visibility time is set when the task is added to the queue.",
+      "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 don't think we need to check for `self.isNotified()` here. If the cursor has been notified, it means that it has already been closed, so there is no need to call `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 think it would be better to use `Assert.Throws<NLogRuntimeException>(() => LogManager.Flush(0));`",
+      "reference": "should users also do a triple-flush?"
+    },
+    {
+      "prediction": "I'm not sure why these were removed. I think they should be added back.",
+      "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) and not isinstance(vals[0], list):`",
+      "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 don't think we should ignore errors here.",
+      "reference": "The body can just be `os.RemoveAll(dir)`"
+    },
+    {
+      "prediction": "I think we should throw an exception here.",
+      "reference": "Is this covered by regression tests? Also you might make a patch onto refactor branch."
+    },
+    {
+      "prediction": "I don't think this is correct. This is a ping request, not a query request.",
+      "reference": "This is maybe more of an admin request? WDYT?"
+    },
+    {
+      "prediction": "I think this should be `HashMap.ofAll(toList().map(f::apply))`",
+      "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'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "maybe localProjectAppNames? Just a thought."
+    },
+    {
+      "prediction": "I think this should be `nextHeightWidth.height - Blockly.BlockSvg.NOTCH_HEIGHT` instead of `nextHeightWidth.height - 2 * Blockly.BlockSvg.NOTCH_HEIGHT`.",
+      "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'm not sure why we need to wait 10 rounds here. I think 5 rounds should be enough.",
+      "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": "Line is too long. [101/80]",
+      "reference": "This section needs to be modified; crontab format differs between OS releases (Debian and CentOS, at least)."
+    },
+    {
+      "prediction": "I think this should be `resume_from_failure: Optional[bool] = False`",
+      "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 we can remove `affine_grid_generator` and `affine_grid` in this 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 `if config.AttackTimeout < 0 { attackTimeout = 0 }`",
+      "reference": "Is there a reason why you need to support negative `AttackTimeout`? Why not just throw an error?"
+    },
+    {
+      "prediction": "I think we should avoid importing `pyspark.sql.functions` directly. Instead, we should use `from pyspark.sql import functions as F` and then use `F.col` instead of `F.col`.",
+      "reference": "nit: an unnecessary change?"
+    },
+    {
+      "prediction": "I think this should be a `RequestStream` instead of a `quic.Stream`.",
+      "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'm not sure this is the right place to do this. I think it should be done in the `detach_volume` method.",
+      "reference": "Any chance of a wait-forever situation here? Is there a reasonable timeout?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the node is following the wrong chain, we need to rollback the blockchain first, and then start syncing from the correct chain.",
+      "reference": "line is 165 characters"
+    },
+    {
+      "prediction": "I don't think we need to poke the cache if we're not using the in-process cache.",
+      "reference": "is this for cross DC case?"
+    },
+    {
+      "prediction": "`mmcv.ops` -> `mmcv._ext`",
+      "reference": "Remove this line."
+    },
+    {
+      "prediction": "I think this should be `webmAudioCodecs.includes('opus') ? 'opus' : 'vorbis'`",
+      "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": "nit: I think this should be \"secondaries\" instead of \"primaries\".",
+      "reference": "Interestingly, this has nothing to do with replication, no clue what this means here"
+    },
+    {
+      "prediction": "I think this should be `if r.h.Type() == tlf.Public || r.h.Type() == tlf.Private && len(r.h.ResolvedWriters()) > 1 {`",
+      "reference": "Is that because we don't provide a UI for these?"
+    },
+    {
+      "prediction": "I don't think we need to import `github.com/influxdata/flux/stdlib/csv` here.",
+      "reference": "Standard is either to use `HappyPath` (CamelCase) or `happy path` (lowercase sentence). I usually prefer the latter."
+    },
+    {
+      "prediction": "why is this nil?",
+      "reference": "instead of using blkMemDao, we could generate a blockMemDao, and then use it as the second parameter."
+    },
+    {
+      "prediction": "I don't think we need to check for `topology.description != null` here. If `topology.description` is `null` or `undefined` then `topology.description.servers` will also be `null` or `undefined` and the `for` loop will not run.",
+      "reference": "nit(2/10): `topology && topology.description != null`;"
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "reference": "why the space? I may not know all the style conventions."
+    },
+    {
+      "prediction": "Shouldn't this be `'<dl role=\"menubar\"><dt id=\"target\">My list item</dt></dl>'`?",
+      "reference": "Should be \"should fail if the **dt element** has a parent <dl> with a changed role\""
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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 don't think we should remove this docblock.",
+      "reference": "not sure about removing this ping @Soullivaneuh"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I think we should be able to remove the `allow_nil: true` from the `scheduled_for_deactivation_on` delegate. I think the `scheduled_for_deactivation_on` method will return `nil` if `subscription` is `nil`, so I think we should be able to remove the `allow_nil: true` from the `scheduled_for_deactivation_on` delegate and the `scheduled_for_deactivation?` delegate.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I think this should be `if err != nil {`",
+      "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": "Why is this initialized to `Tuple.Create((string)null, new HashSet<string>())` instead of just `Tuple.Create(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": "I think this should be `[CLY]_star_rating` instead of `[CLY]_action`",
+      "reference": "I think there is no point providing action key here, as it will be called specifically, rather than randomly"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to be the only way to get the current runtime description.",
+      "reference": "`XUnitIntegration` doesn't need to cache this anymore."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think it should be `files[\"themes/*.lua\"].ignore = {\"631\"}` or `files[\"themes/**.lua\"].ignore = {\"631\"}`.",
+      "reference": "Can we have something line `# noqa` there instead? I think it's fine like this though."
+    },
+    {
+      "prediction": "Line is too long. [115/80]",
+      "reference": "this variable is used in both branches so I've decided to declare it above"
+    },
+    {
+      "prediction": "I don't think you need to add the flags, just set them.",
+      "reference": "`setFlags()` overwrites the previous flag, we should be using `addFlags()` to append flags."
+    },
+    {
+      "prediction": "Shouldn't this be `ctx.Repo.Mirror, err = models.GetMirrorByRepoID(repo.ID)`?",
+      "reference": "Need `return` after this."
+    },
+    {
+      "prediction": "I don't think we should be adding cache headers to all JSON responses. I think we should only add cache headers to responses that are meant to be cached.",
+      "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 `FINE` instead of `FINER` or `FINEST`.",
+      "reference": "This change is unhelpful: it precludes users from knowing which ports are being used for what purpose within the system."
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>",
+      "reference": "Whoa, we weren't doing this before?? Derp."
+    },
+    {
+      "prediction": "I think you can just use `f.Flow.CopyActionsToNewFlow(&flow)` here.",
+      "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 if this is the right thing to do, but I think we should be able to enable XDP even if BPF is not enabled.",
+      "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 don't think we need to implement `ASTAnyTypeDeclaration` in this class. I think it's enough to implement `ASTAnyTypeDeclaration` in `ASTClassOrInterfaceDeclaration` and `ASTEnumDeclaration`.",
+      "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 this is the right place for this constant. I think it would be better to have it in the `http2` package, since it's specific to HTTP/2.",
+      "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 thing to do here. I think we should be able to assert that the `endSessions` command was sent.",
+      "reference": "Is this change implying something or just seems fit b/c regardless of outcome the sessions should still be length 1?"
+    },
+    {
+      "prediction": "I think this should be `configProto.getTemplatesList().get(0).getLanguageProvider()`.",
+      "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 think this should be `get \"/new-language-confirmation\", to: \"new_language_confirmations#new\"`",
+      "reference": "Prefer single-quoted strings when you don't need string interpolation or special symbols."
+    },
+    {
+      "prediction": "I think this should be `return s == \"\" || s == \"linux\"`",
+      "reference": "I feel the name of this function could be better, conveying it's a predicate. Something like `HasDockerTestEnvironment` or `CanRunLocalServerTests`, etc."
+    },
+    {
+      "prediction": "This should be `\"/MyGame.Example.MonsterStorage/Retrieve\"`",
+      "reference": "This seems unrelated to your PR, how did this end up in here?"
+    },
+    {
+      "prediction": "I don't think we should change the behavior here.",
+      "reference": "Is `username` expected to be case insensitive?"
+    },
+    {
+      "prediction": "why change the import order?",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I think this should be `@Deprecated(since = \"7.0.0\", forRemoval = true)`",
+      "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 `QItemSelection` instead of `str`.",
+      "reference": "You can remove the `QItemSelection` import in this file now."
+    },
+    {
+      "prediction": "`spaceLock.unlock_shared();` should be in `finally` block.",
+      "reference": "Use `std::shared_lock` and `std::try_to_lock_t`. BTW. Why we fail on contention?"
+    },
+    {
+      "prediction": "I don't think this is necessary. The default is `[\"\", \".js\", \".json\"]` and `.js` is already included.",
+      "reference": "Not sure what this actually does, but do you need to add `\"\"` here? Or does `[\".jsx\", \".js\"]` work too?"
+    },
+    {
+      "prediction": "please remove this empty line",
+      "reference": "let's group imports in the order: std third party first party"
+    },
+    {
+      "prediction": "nit: can we remove this method?",
+      "reference": "So the tests run only when the db exists? Else everything passes?"
+    },
+    {
+      "prediction": "I think this should be `io.WriterCloser` instead of `io.Writer` and `io.Closer`.",
+      "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 think this should be `MinerID` instead of `Miner` to avoid confusion with the `Miner` type.",
+      "reference": "Mentioned elsewhere but repeating again: would be nice to have units (ideally `Size *Bytes`)."
+    },
+    {
+      "prediction": "I think this should be `--no-resolve` instead of `--no-resolving`",
+      "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 was removed, but I think it should be kept.",
+      "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 `cb != NULL` instead of `cb == NULL`.",
+      "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": "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.<br>",
+      "reference": "why are we using an instance variable instead of accessing the visibility method?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think it's a big deal.",
+      "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')`",
+      "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": "Why is this removed?",
+      "reference": "Fixing `lint` warnings that have existed for a while."
+    },
+    {
+      "prediction": "I think this should be `OPENEBS_IO_CAS_DEFAULT_CSTOR_SPARSE_POOL`",
+      "reference": "Better to rename this to CASDefaultCstorSparsePool."
+    },
+    {
+      "prediction": "I think this should be `awsClientFactory.close()`.",
+      "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 if this is the best way to check if the name is empty or not, but it works for now.",
+      "reference": "And here - what happens if name id is invalid"
+    },
+    {
+      "prediction": "I think this should be `echo $query->getRawQuery();`",
+      "reference": "__toString() is not a part of QueryInterface"
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "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 don't see this error code in the CUDA 10.2 docs. Is this a new error code in CUDA 11.0?",
+      "reference": "Please remove `HIP_UNSUPPORTED`"
+    },
+    {
+      "prediction": "I think this should be `returnNumber ? parseInt( daysMatch[ 1 ], 10 ) : daysMatch[ 1 ];`",
+      "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 not needed anymore?",
+      "reference": "Is this not required?"
+    },
+    {
+      "prediction": "I think we should keep the `AccountState` and `Balance` methods.",
+      "reference": "savedStates is for confirmed states, not needed in working set"
+    },
+    {
+      "prediction": "I think this should be `wireguard` instead of `Wireguard`",
+      "reference": "Why should we move it into `bootstrap.go` file?"
+    },
+    {
+      "prediction": "I don't think we need to export this. It's only used in one place.",
+      "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 think this should be `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 thing to do, but I don't have a better idea.",
+      "reference": "hrm, unsigned headers now won't be included."
+    },
+    {
+      "prediction": "`strlen($value) <= 255;` is not the same as `strlen($value) < 256;`",
+      "reference": "what if `$value = ' ' `?"
+    },
+    {
+      "prediction": "I think we should also log the number of files that are being committed.",
+      "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": "I think this should be `const std::string& Container::getName() const`",
+      "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": "`type` is always `0` in this case, so we don't need to pass it.",
+      "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": "`h2o_buffer_reserve` returns `h2o_iovec_t`, not `h2o_iovec_t *`.",
+      "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": "I'm not sure if this is the best way to get the timezone, but it seems to work.",
+      "reference": "Does this mean the times will show up in emails as UTC?"
+    },
+    {
+      "prediction": "I think this should be `color='black'` instead of `color=Cycle()`.",
+      "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 `Split(Environment.NewLine, StringSplitOptions.RemoveEmptyEntries)`",
+      "reference": "What will be an empty entry now?"
+    },
+    {
+      "prediction": "Shouldn't this be `this.hot.getSettings().rtl`?",
+      "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": "Please add a license header to this file.",
+      "reference": "`this.` seems unnecessary"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "nit: can be removed before merging..."
+    },
+    {
+      "prediction": "I think you can just use `$('#testContainer')` instead of `$(<div id=\"testContainer\"></div>)`",
+      "reference": "Maybe a single quote would be compatible with airbnb style."
+    },
+    {
+      "prediction": "I think this should be `return sessions.size();`",
+      "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'm not sure about this change. I think it's better to keep the formatting as it was before.",
+      "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'm not sure this is the right thing to do. I think we should return an error here, and let the caller decide what to do with it.",
+      "reference": "I think we probably need a util.Warning() here."
+    },
+    {
+      "prediction": "I don't think we need to check if `path` is a `PackageRegistry` here. If it is, we can just return it. If it isn't, we can assume that it is a `PhysicalKey`.",
+      "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 `var imports = (axe.imports = {});` instead of `var imports = axe.imports = {};`",
+      "reference": "What is this directive for?"
+    },
+    {
+      "prediction": "I think this should be `chained-comparison` instead of `simplify-chained-comparison`.",
+      "reference": "I'd rephrase it as `Simplify chained comparison between the operands`."
+    },
+    {
+      "prediction": "I think this should be `1` instead of `2`.",
+      "reference": "so in the performance data fixtures, there will be references for the second domain only?"
+    },
+    {
+      "prediction": "I don't think we need to log this at all.",
+      "reference": "This is wildly unhelpful to users --- they need to know when a proxy has been registered."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do here, but it seems to fix the issue for me.",
+      "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 should still return a 500 here.",
+      "reference": "is there any unit test that can be updated to verify this?"
+    },
+    {
+      "prediction": "I think you can remove the `else` and just do `lua_call(ctx->lua->state, 3, 3);`",
+      "reference": "exiting at this point is leaking memory, take a look at the valgrind output. the sbuffer must be destroyed."
+    },
+    {
+      "prediction": "I don't think we need to do this. We can just use `@pytest.mark.poppler` in the tests that need it.",
+      "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: can we keep the empty line here?",
+      "reference": "this needed to change? or it was just some formatting thing?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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 `project_id` instead of `project_number`.",
+      "reference": "typo: if -> of"
+    },
+    {
+      "prediction": "Shouldn't this be `return \"\", err`?",
+      "reference": "It'll be interesting to see who or what this breaks..."
+    },
+    {
+      "prediction": "I think this should be `log.fine`",
+      "reference": "This change is incorrect: the current log level is correct."
+    },
+    {
+      "prediction": "I think we should use `log.NewDefaultLogger()` instead of `log.NewTestLogger()` here.",
+      "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 don't think this is a good idea to throw an `IllegalArgumentException` here. I think it would be better to throw a `NullPointerException` in case of a `null` value and a `StringIndexOutOfBoundsException` in case of an empty string.",
+      "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 the best way to do it, but it seems to be the only way to get the `id` of a `dict` object.",
+      "reference": "index by memory address? I realize I don't understand why we don't just build a list :)"
+    },
+    {
+      "prediction": "Please 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 think this should be `aspnet-mvc.request.child-action`",
+      "reference": "Does not appear to be used."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here, but I think it's the best we can do for now. We don't have a good way to know if we're running on Windows or not, and we don't have a good way to know if we're running on a Windows node or not, so we have to assume we're running on a Windows node if we're running on Windows.",
+      "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 right thing to do, but I think we want to use `@video_page.video` instead of `@video_page.purchaseable.video` because `@video_page.purchaseable` could be either a `Video` or a `Collection` and we don't want to assume it's a `Video`.",
+      "reference": "Can we add a `collection?` method to the `VideoPage` so we don't violate Law of Demeter here?"
+    },
+    {
+      "prediction": "I don't think we need to remove this. It is used in `ProjectManagerServlet.java` line 1255.",
+      "reference": "Deleting unused code."
+    },
+    {
+      "prediction": "I don't think we need to import `PseudoSampler` from `..bbox.samplers.pseudo_sampler` here.",
+      "reference": "`PseudoSampler` can also be imported from `..bbox`"
+    },
+    {
+      "prediction": "I think this should be `protocol.ConsensusVersion` instead of just `ConsensusVersion`.",
+      "reference": "Could you move this import to where the other `github.com/algorand/go-algorand` imports are?"
+    },
+    {
+      "prediction": "I think this should be `l2tpv3` instead of `l2f`.",
+      "reference": "I'd rather not activate this protocol by default for now, unless there is a good reason for that."
+    },
+    {
+      "prediction": "nit: please add a blank line between imports.",
+      "reference": "Remove these 2 blank lines."
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "Want to remove this?"
+    },
+    {
+      "prediction": "I think this should be moved to `sql/postgres.go`",
+      "reference": "is this a golint requirement? huh. mixed feelings :)"
+    },
+    {
+      "prediction": "I think it would be better to use `BCryptPasswordEncoder.BCRYPT_VERSION_2A` instead of `BCryptPasswordEncoder.BCRYPT_VERSION_2B` and `BCryptPasswordEncoder.BCRYPT_VERSION_2X` instead of `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": "Why is this lock passed in as an argument?",
+      "reference": "Should something use the lock passed in?"
+    },
+    {
+      "prediction": "I think this should be `volumes` instead of `volumes`",
+      "reference": "Is this dependant on some version of Docker? Do we need to include both or `volumes` is deprecated (or just wrong!?)?"
+    },
+    {
+      "prediction": "I think this should be `if (leader != null && leader.getState() == State.ACTIVE && zkShardTerms.getTerm(leader.getName()) == zkShardTerms.getHighestTerm()) {`",
+      "reference": "I know this is not new code, but should we change `leader.getState() == State.ACTIVE` to `leader.isActive(liveNodes)`?"
+    },
+    {
+      "prediction": "I don't understand why this change is necessary. `node.getScope()` returns a `ClassScope` when `node` is an `AbstractAnyTypeDeclaration`.",
+      "reference": "Why not use ASTAnyTypeDeclaration? AbstractAnyTypeDeclaration is deprecated"
+    },
+    {
+      "prediction": "I think this should be `from .tools.util import save as _save` so that it doesn't conflict with the built-in `save` function.",
+      "reference": "A blank line at the very end of each file should eliminate the \"No EOF\" warning we see above"
+    },
+    {
+      "prediction": "nit: I think we can remove the `isCurrentWorkflowConditionFailedErr` variable and just use `isWorkflowConditionFailedErr` instead.",
+      "reference": "These would both be more robust with `errors.As` so that we could safely chain errors."
+    },
+    {
+      "prediction": "I think this should be `_TtCs23_ContiguousArrayStorage` instead of `_TtCs23_ContiguousArrayStorage`.",
+      "reference": "This is checking the same condition twice?"
+    },
+    {
+      "prediction": "I think this should be `>= 16` instead of `>= 16`.",
+      "reference": "Aren't we able to handle the case where we have MPI-CUDA without NCCL, even if it's suboptimal?"
+    },
+    {
+      "prediction": "This should be a `constexpr` instead of a `static const`.",
+      "reference": "Wouldn't we want to namespace this `AA` or something similar"
+    },
+    {
+      "prediction": "I think this should be `if len(envCfg.Profile) == 0 && !envCfg.EnableSharedConfig && envCfg.Creds.HasKeys() {`",
+      "reference": "This probably should look beyond envConfig, and include `aws.Config.Credentials` as well."
+    },
+    {
+      "prediction": "I don't think we need to add it to the filter pipeline. It should be enough to add it to the `AddTransaction` method.",
+      "reference": "How does this filter differ from TooExpensiveTxFilter?"
+    },
+    {
+      "prediction": "I think it would be better to return a `Map<String, Permission>` instead of a `List<String>`.",
+      "reference": "Use \" List<String>\" instead of ArrayList<String> in declaration."
+    },
+    {
+      "prediction": "I think we should be able to remove the `runtime_type.name is not None` check here, since `runtime_type.name` will always be `None` if `runtime_type` is not in `self.TYPE_REGISTRY`.",
+      "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'm not sure if this is a good idea to use `BundleUtil` in this class. I think it would be better to use `MessageUtil` instead.",
+      "reference": "Just noticed this - why \"like\" and not straight \"=\""
+    },
+    {
+      "prediction": "I think this should be `travis_vars.count.should == 11`",
+      "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 don't think this is the right way to use `String.format`. It should be `String.format(\"Failed to setInstances, AppId=%s, MicroserviceName=%s.\", getAppId(), getMicroserviceName())` or `String.format(\"Failed to setInstances, appId=%s, microserviceName=%s.\", getAppId(), getMicroserviceName())`",
+      "reference": "when will lost exception? by my test, never happened."
+    },
+    {
+      "prediction": "I think this should be `urlstr = url.toDisplayString()`",
+      "reference": "You should re-stringify it here with `QUrl.FullyEncoded`."
+    },
+    {
+      "prediction": "I don't think we need to pass a quit channel here. We can just return an error if the context is cancelled.",
+      "reference": "i'm not sure why the quit channel is needed in this context"
+    },
+    {
+      "prediction": "I think this should be `if args.CPULimit != \"\"` instead of `if len(args.CPULimit) > 0`.",
+      "reference": "When `args.MemoryLimit` is empty, will `MustParse` panic?"
+    },
+    {
+      "prediction": "I think this should be `r.inSync = false`",
+      "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": "Why do we need this?",
+      "reference": "I add a `idx` property, and it is only use for benchmark `heap.Remove`, is it acceptable?"
+    },
+    {
+      "prediction": "Please remove the extra `t` from 2019.",
+      "reference": "daemonic t got inserted."
+    },
+    {
+      "prediction": "I think this should be split into 2 lines.",
+      "reference": "This edit is non-essential."
+    },
+    {
+      "prediction": "I think we can just use `obj.__proto__ = proto; return obj;` here.",
+      "reference": "Might be safer to ponyfill this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should check if the config file exists, and if it does, return an error. If it doesn't exist, then we should create it.",
+      "reference": "why was this join not necessary before?"
+    },
+    {
+      "prediction": "I'm not sure this is a good test. I don't think we should be testing the internals of the `CountingQueryTimeout` class. I think we should be testing that the `ExitableDirectoryReader` calls the `shouldExit` method on the `QueryTimeout` instance passed to it.",
+      "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 don't think we need to include `--configfile` here.",
+      "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": "I think this should be `bool RecordException { get; set; } = true;`",
+      "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'm not sure if this is the right thing to do, but I think it's better to catch `Throwable` instead of `IOException` here. `IOException` is a subclass of `RuntimeException`, so it's already caught by the `catch (RuntimeException e)` block above.",
+      "reference": "usually I do either ` IOException|RuntimeException` or if doing Throwable, use propagateIfFatal"
+    },
+    {
+      "prediction": "I think this should be `other.clone(shared_data=False, callback=callback, streams=[])`",
+      "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'm not sure this is the right place to put this. I think it should be in the `SolrSecurity` class.",
+      "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 don't think we need this constant. We can just use `address.AddressLength` instead.",
+      "reference": "IoAddrLen Golang prefers short variable names"
+    },
+    {
+      "prediction": "I don't think this is correct. The alias should be `a` only for the `list` context. For the `show` context, the alias should be `o`.",
+      "reference": "PHPCR-ODM does not allow blank aliases.."
+    },
+    {
+      "prediction": "I think this should be `request.getContext().toString()` instead of `getContext().toString()`.",
+      "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": "I think this should be `#include <Kokkos_SYCL_Instance.hpp>`",
+      "reference": "Why are you including `<Kokkos_Core.hpp>`?"
+    },
+    {
+      "prediction": "I think this should be `/api/v1/health` instead of `/health`",
+      "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": "I think this should be `len(rule.DstIpPortSetIds) > 0` instead of `len(rule.DstIpPortSetIds) == 0`.",
+      "reference": "Yep, looks right; your new selector _should_ disqualify the rule from being XDP accelerated."
+    },
+    {
+      "prediction": "I'm not sure about the name of this feature. `IConnectionHeartbeatTickFeature` sounds like it's going to be called every time the heartbeat is ticked, but that's not the case. It's only going to be called once when the connection is opened.",
+      "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 `f\"{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 what you mean by \"the alignment of xstate ultimately depends on the alignment of the special allocator's blocks\". Can you elaborate?",
+      "reference": "Should we assert that xstate is aligned here?"
+    },
+    {
+      "prediction": "I don't think we need to suppress Lucene3x, Lucene40, Lucene41, Lucene42, Lucene45, and Appending anymore.",
+      "reference": "any particular reason for suppressing this change?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in `DifficultyCalculatorFactory`.",
+      "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 this should be `blob://myvar.txt?bucket=fileblob:///foo/bar&decoder=string&wait=30s`.",
+      "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 `min_ts` instead of `max_ts`.",
+      "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 think this is a copy/paste error. I don't think it should say \"There might be something wrong with default web browser on this computer..\". I think it should say \"There might be something wrong with the default web browser on this computer..\".",
+      "reference": "Seems like a lot of unintended changes. Are your line break settings causing these?"
+    },
+    {
+      "prediction": "I don't think we need to pin `ca-certificates` here.",
+      "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 this should be changed.",
+      "reference": "I'm not arguing with it, but how did you decide to make this change?"
+    },
+    {
+      "prediction": "I don't think we need to update the copyright year for this file.",
+      "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 this should be `+kubebuilder:validation:Enum=DST Root CA X3,ISRG Root X1`",
+      "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 `ColdStartEnabled` or something like that. `ColdStart` sounds like it's a config struct.",
+      "reference": "do we really need this behavour to be configurable?"
+    },
+    {
+      "prediction": "I don't think you need to require `objectid` here. You should be able to use `mongoose.Types.ObjectId` instead.",
+      "reference": "Hmm interesting. Do the tests check out on this?"
+    },
+    {
+      "prediction": "I think we should make this configurable.",
+      "reference": "For another PR, another `--X` candidate."
+    },
+    {
+      "prediction": "I think this should be `Releases` to be consistent with the other menu items.",
+      "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": "Please use `org.junit.Assert` instead of `junit.framework.TestCase`.",
+      "reference": "this is a junit 3 import, not a junit 4"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we need to create a `ClusterSync` resource in the `cdBuilder` so that the `Reconcile` function doesn't return an error when it can't find the `ClusterSync` resource.",
+      "reference": "shouldn't there be a reque expected here?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "I prefer to make it `protected set`. What do you think?"
+    },
+    {
+      "prediction": "I don't think these should be json tags.",
+      "reference": "Maybe rename this to JetStreamStoreDir?"
+    },
+    {
+      "prediction": "I'm not sure what's going on here, but I think we should investigate.",
+      "reference": "You can start by logging out the error"
+    },
+    {
+      "prediction": "Why do we need to pass `baseFeePerGas` as `in`?",
+      "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 we should keep the `Function1<K, V>` interface.",
+      "reference": "follow-up of #2002"
+    },
+    {
+      "prediction": "Why do we need to divide by `conf.getPartitionPerSplit()`?",
+      "reference": "could you give some explanations of what problem do you want to solve by this change, and how/why it works?"
+    },
+    {
+      "prediction": "I'm not sure why this needs to be `LongLivedMarshalByRefObject`.",
+      "reference": "LongLivedMarshalByRefObject exists in both the netstandard and netfx xunit?"
+    },
+    {
+      "prediction": "I don't think we need the `&&=` here. We can just do `options[:page_ranges] = Array(options[:page_ranges])`",
+      "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'm not sure if this is the right thing to do, but I think we want to kill the containerd process when we kill the containerd service.",
+      "reference": "Testing on linux with this change, containerd no longer stops when k3s stops."
+    },
+    {
+      "prediction": "I think this should be called `AllowConnectionWindowDecrease` to be consistent with `AllowStreamWindowDecrease` and `AllowStreamWindowIncrease`.",
+      "reference": "Maybe [u]int64 to be consistent with other byte-valued vars in this struct?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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 do this. `prefetch_` is a `std::vector<std::unique_ptr<Batch<Dtype>>>`, so `prefetch_[i]` is a `std::unique_ptr<Batch<Dtype>>`. You can just do `prefetch_[i].reset(new Batch<Dtype>());`",
+      "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 don't think we need to return an error here. If we don't know what type of transaction it is, we can just return an empty `v1.Transaction` object.",
+      "reference": "nit : move this into a default clause in the previous switch statement."
+    },
+    {
+      "prediction": "I don't think this is correct. If the string contains a single quote, you need to escape it with a backslash.",
+      "reference": "It should be return \"''\"; I will fix it"
+    },
+    {
+      "prediction": "I think we should keep `Id` and `ID` the same.",
+      "reference": "Go standard linter hates `Id`s and likes `ID`s, please do not resist."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it would be nice to have a retry for `git checkout` as well.",
+      "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 think this should be `SetCertificateRenewalTime` instead of `SetCertificateRenewalTIme`",
+      "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 think this should be `lambda pkt: pkt.extension is not None`",
+      "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[:order] = \"rand:#{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 don't think this is the right place to set the flag. I think it should be set in the `privload_insert()` call above.",
+      "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'm not sure this is the right place to do this. I think it should be done in `MoleculeAccessionNumbers.FormatAccessionNumbers()`.",
+      "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 would prefer to keep the imports in alphabetical order.",
+      "reference": "As Miklas stated, please avoid wildcard imports."
+    },
+    {
+      "prediction": "I think we should keep the `app` and `companionOptions` exports.",
+      "reference": "Maybe change it to `module.exports = server` ? Becase there's nothing else to export here"
+    },
+    {
+      "prediction": "nit: I think this can be removed.",
+      "reference": "This may be an automated change, but the import seems to fit better where it was previously."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it should be in the `application.html.erb` file, so that it's loaded on every page.",
+      "reference": "Might we want to separate including this snippet and sending the data for signed in users into different partials?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. It's not clear to me why we would want to skip some peers, and it's not clear to me why we would want to return an error in case the base is the closest to the chunk. I think it would be better to just return the base as the closest peer, and let the caller decide what to do with it. What do you think?",
+      "reference": "closest to the address"
+    },
+    {
+      "prediction": "The `AutoRowSize` plugin allows you to set row heights based on their highest cells. By default, the plugin is declared as `undefined`, which makes it disabled (same as if it was declared as `false`). Enabling this plugin may decrease the overall table performance, as it needs to calculate the heights of all cells to resize the rows accordingly. If you experience problems with the performance, try turning this feature off and declaring the row heights manually. But, to display Handsontable's [scrollbar](https://handsontable.com/docs/8.0.0/demo-scrolling.html) in a proper size, you need to enable the `AutoRowSize` plugin, by setting the [`autoRowSize`](@/api/options.md#autoRowSize) option to `true`. Row height calculations are divided into sync and async part. Each of this parts has their own advantages and disadvantages. Synchronous calculations are faster but they block the browser UI, while the slower asynchronous operations don't block the browser UI.",
+      "reference": "It should be much more clear for end-user"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get rid of the pylint warning.",
+      "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": "nit: I think it would be nice to add a test for `NewRangeReader` with `gcerrors.OK` as well.",
+      "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": "Align the elements of a hash literal if they span more than one line.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "I would put that constant under config/initializers/constant.rb and I would leave application.rb for rails specific."
+    },
+    {
+      "prediction": "nit: I think it would be clearer to use `if _, ok := doNotUnref[ptr.ID]; ok { return ops }`",
+      "reference": "Is there a reason not to do something like `if _, ok := doNotUnref[ptr.ID]; ok { return ops }` here?"
+    },
+    {
+      "prediction": "I don't think this is correct. `self._execute(Command.GET_ELEMENT_RECT)['value']` returns a dictionary with `x`, `y`, `width` and `height` keys. `self.size` and `self.location` return dictionaries with `x` and `y` keys.",
+      "reference": "This would return a tuple of two dictionaries. You need to combine them and return a dictionary"
+    },
+    {
+      "prediction": "I don't think we need to pass `asa` here.",
+      "reference": "Why not always forward it to `_data`?"
+    },
+    {
+      "prediction": "I think this should be a `GET` request, not a `POST` request.",
+      "reference": "there are tests that use `client.get` which will fail now"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do for now.",
+      "reference": "Whoa, TIL about this syntax."
+    },
+    {
+      "prediction": "I think this should be `&yarpc.Response{ApplicationErrorInfo: &yarpcerror.ExtractInfo(h.applicationErr)}`",
+      "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": "Please remove the `load_from` line.",
+      "reference": "This URL is outdated, add a `TODO` to fix it."
+    },
+    {
+      "prediction": "Why do we need to add chainID to envelope?",
+      "reference": "`ChainID()` is a getter method, move up to `Nonce()`"
+    },
+    {
+      "prediction": "I think it would be better to have a separate method for this, so that we don't have to repeat the same code in multiple places.",
+      "reference": "If this condition is not met, shall we just throw the exception rather than logging it? @li-ygerchikov Your thoughts?"
+    },
+    {
+      "prediction": "I think this should be `ruamel.yaml>=0.15.0`",
+      "reference": "nit: alpha-sort `ruamel.yaml` in this list please"
+    },
+    {
+      "prediction": "I don't think we need to check `openFileCost >= 0` here. It's already checked in `BinPacking.PackingIterable`.",
+      "reference": "Remove this because we already added it in another pr"
+    },
+    {
+      "prediction": "I think this should be `getClass().getClassLoader().getResource(\"phoenicis.png\").toExternalForm()`",
+      "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": "`allow_blank: true` is not necessary here, since `presence: true` already implies `allow_blank: true`.",
+      "reference": "I think we want to require the association with the user...?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to be the only way to get the size of the message 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 can remove this line.",
+      "reference": "nit: no need this change anymore."
+    },
+    {
+      "prediction": "I don't think we need this change.",
+      "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": "I don't think we need to import `javaslang.collection.Seq` here.",
+      "reference": "[Checkstyle] ERROR: Unused import - javaslang.collection.Stream."
+    },
+    {
+      "prediction": "I'm not sure why this is changing, but I'm guessing it's because of the addition of the `ssto` field.",
+      "reference": "The numbers were changed since now _root_ is also added to the doc, increasing the field count by 1."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "TTemplateParamClass is handled above with the same content in the conditional"
+    },
+    {
+      "prediction": "Shouldn't this be `if (isset($newConfig['Captcha'][$key]) && !isset($newConfig['Captcha']['recaptcha_' . $key]))`?",
+      "reference": "Do we need to add `theme` here?"
+    },
+    {
+      "prediction": "I think this should be `*string` instead of `string` to be consistent with the rest of the codebase.",
+      "reference": "i think you have to regen the api stuff, `make generate`"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `Signer` class.",
+      "reference": "DataHash shouldn't be needed here"
+    },
+    {
+      "prediction": "I'm not sure this is the right level of logging. I think it should be INFO.",
+      "reference": "`info` was chosen deliberately to mirror the old behaviour that users expected."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way to get an array from the ember-selectize component.",
+      "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 think we should rename this to `originalIcebergType` to make it clear that this is the original type of the column in the iceberg schema.",
+      "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 think this should be `*storage.ObjectHandle` instead of `**storage.ObjectHandle`.",
+      "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 the `done` here.",
+      "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": "I think this should be `bundle != null && bundle.getBoolean(IS_GLOBAL_STORE, false);`",
+      "reference": "SmartStoreInspectorTest was throwing a NPE here"
+    },
+    {
+      "prediction": "I think we can remove the `CloseableIterable` here and just use `Iterables.transform` instead.",
+      "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 `session.TranscodingInfo.HardwareAcceleration` instead of `session.TranscodingInfo.HardwareAccelerationType`",
+      "reference": "I think it would be a good idea to wrap this in an if like the stats above, just in case."
+    },
+    {
+      "prediction": "I don't think this is the right place to call `selectAllPhotosUpToFav()`. I think it should be called in `selectAllPhotosUpToFav()` itself.",
+      "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 think this should be `join(path, r.BlkioDeviceReadBps, r.BlkioDeviceReadIOps, r.BlkioDeviceWriteBps, r.BlkioDeviceWriteIOps)`",
+      "reference": "Looks like the actual rename is missing in this commit edit: Arf, nevermind, I was at the wrong commit"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this. I think it would be better to have a `defaultOptions` object that we can merge with the user-provided options. This way, we don't need to check for `getAssemblyOptions` at all. What do you think?",
+      "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": "nit: I think we can remove the `import_` prefix from the regexes.",
+      "reference": "networkVarPattern (etc) then?"
+    },
+    {
+      "prediction": "nitpick: I think it would be clearer to write this as `metadata[:block] || new_metadata[:block]`",
+      "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 necessary.",
+      "reference": "Just realized this should be fixed instead of changing here. Concatenating int should still result in lowercase string."
+    },
+    {
+      "prediction": "I think this should be `new({browser_name: 'firefox', marionette: true}.merge(opts))`",
+      "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": "nit: I think we can remove the `dstP` arg here and just use the one from the `DefaultSyncerTestParams` struct.",
+      "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 think this should be `deviceName = ad.DeviceName` instead of `deviceName := ad.DeviceName`",
+      "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": "I think this should be `FeedLockTimeoutSeconds` to be consistent with the rest of the codebase.",
+      "reference": "typo: \"be hold\" -> \"be held\""
+    },
+    {
+      "prediction": "Why is this changed to 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 this should be `>=1.14.5`",
+      "reference": "Why did this change?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [123/120]",
+      "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": "nit: I think it would be better to make the `PropertyBagDictionary` constructor that takes a `PropertyBagDictionary` private.",
+      "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 `Values` instead of `Vector`.",
+      "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 we can remove this import now.",
+      "reference": "Thinking this should be `removeNode()` exported from `dom/index.js`. Thoughts?"
+    },
+    {
+      "prediction": "nit: `memoryCPULimitsEnabledLock` -> `memoryCPULimitsEnabledMutex`",
+      "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 `atomic.LoadUint64(&s.cacheHits) / float64(st.NumMatches)`",
+      "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 think this should be `wifiStatus |= ESPEASY_WIFI_SERVICES_INITIALIZED;` instead of `wifiStatus &= ~ESPEASY_WIFI_SERVICES_INITIALIZED;`",
+      "reference": "@TD-er use bitRead/bitWrite macros, do you use here inversed logic?"
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "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 think this should be `if (context.EndStreamReceived && !context.ExpectData) { return MessageBody.ZeroContentLengthClose; } else { return new ForHttp2(context); }`",
+      "reference": "Is ExpectData still used anywhere?"
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it should be `[~!.]` instead of `[~!.]?`. The `?` makes the whole thing optional, which is not what we want.",
+      "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'm not sure if this is the best way to test this, but it seems to work.",
+      "reference": "`eachAsync()` should pass a doc, not an array of docs, to the callback. This would be a massive breaking change."
+    },
+    {
+      "prediction": "I don't think we should log this as a warning. It's not an error, it's just an informational message.",
+      "reference": "please start log with caps letters"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it seems like a good place to me.",
+      "reference": "Is this where we want it in relation to the other scripts?"
+    },
+    {
+      "prediction": "I think we should use `REPORTS_FEATURE_NAME` instead of `FEATURE_NAME`",
+      "reference": "Please wrap this variable within a self invoking anonymous function. (function(){ FEATURE_NAME = \"reports\"; //and rest of the js goes here. })();"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in `Pause` and `Unpause` in `cgroup_manager_linux.go`.",
+      "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 don't think we should be throwing an error here. I think we should be using `expect.throws` or `expect(() => { ... }).to.throw(MongoURIError);`",
+      "reference": "no need for custom errors in tests unless the tests are intended to mock a specific sort of error"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "This const has already existed in the code, no need to define a new one"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to just log the error and not exit the process.",
+      "reference": "Would Errorf be better than Warnf?"
+    },
+    {
+      "prediction": "I don't think you need to initialize `currentContent` to an empty string here.",
+      "reference": "A good candidate for StringBuilder? Also make sure to add white space around the operators (`path + \"/\" + file`)"
+    },
+    {
+      "prediction": "I think this should be `String flowName` instead of `int flowVersion`.",
+      "reference": "Can we have the same argument order as `getUploadedFlowFile`?"
+    },
+    {
+      "prediction": "I think we should keep this as `dynamicconfig.DurationPropertyFnWithTaskQueueInfoFilters`",
+      "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": "This should be `obj.scrollIntoView(align=\"top\")`.",
+      "reference": "Should these cases be logged? `if not obj` used to be"
+    },
+    {
+      "prediction": "I don't think we need to use `LANG=C` here.",
+      "reference": "Is it OK to assume that keys.gnupg.net has the new key?"
+    },
+    {
+      "prediction": "I don't think we need to remove these notes.",
+      "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 don't think this test is needed anymore.",
+      "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 thing to do. I think we want to use a `CachedThreadPool` so that we don't have to create a new thread for every task, but we also want to limit the number of threads to the number of cores on the machine. So I think we want to use `Concurrent::FixedThreadPool.new(@config[:concurrency])` instead.",
+      "reference": "It's a little surprising that CachedThreadPool overrides the max_threads argument. This makes sense as a solution though."
+    },
+    {
+      "prediction": "`elif reduction == 'none':` -> `elif reduction != 'mean':`",
+      "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": "I don't think this is correct. The stride should be `stride` and the padding should be `dilation`.",
+      "reference": "`padding` should be 0 instead. Actually you can just remove padding and dilation to use default values."
+    },
+    {
+      "prediction": "I think this should be `$balance += $finesListSort[$key][\"amount\"];` instead of `$balance = $finesListSort[$key][\"amount\"];`",
+      "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` is not a keyword argument of `plot` method.",
+      "reference": "Why do we need to explicitly use keyword arguments?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should still log the build result here, but I'm not sure what the best way to do that is.",
+      "reference": "Might still want to log this as `TargetBuilding` but with a slightly different message?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "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 `nodes` here.",
+      "reference": "Well done, we forget that one apparentely."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "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 if this is the best way to do this, but I think it's better than having a bunch of `if`/`elif`/`else` statements.",
+      "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 this should be `athenz.zms.principal_state_updater_frequency` instead of `athenz.zms.principal_state_updater_frequency_default`",
+      "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": "Shouldn't this be `if (this.opts.lazy && this.opts.waitForThumbnailsBeforeUpload)`?",
+      "reference": "Should be `this.opts` instead of `defaultOptions`"
+    },
+    {
+      "prediction": "I think this should be `export const DIST_TAGS = 'dist-tags';` instead of `export const USERS = 'users';`",
+      "reference": "You can reuse `USERS` above as well."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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": "I'm not sure this is a good idea. `List.ofAll` is more efficient than `Stream.ofAll`.",
+      "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 `return $check->getListTagSetting() === 'enabled' || $check->getTagSetting() === 'enabled';`",
+      "reference": "Since this is only used in the MyResearchController, does it need to be placed at the AbstractBase level?"
+    },
+    {
+      "prediction": "I don't think we need to change the name of this field. We can keep it as `PodPhaseSelectors` and add a new field called `PhaseSelectors`.",
+      "reference": "why `PhaseSelector` ?"
+    },
+    {
+      "prediction": "I think this should be `AllowlistedModules` instead of `allowlisted_modules` to be consistent with the rest of the codebase.",
+      "reference": "Does` AllowlistedModules` have the same meaning of `WhitelistedModules`?"
+    },
+    {
+      "prediction": "I think this should be `list<array<mixed, mixed>>`",
+      "reference": "I missed the $preserve_keys=true case when adding this to Phan. For psalm, two separate signatures may make sense"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "TcpServer will be created for each HighwayServerVerticle instance so the counter number in server is not correct."
+    },
+    {
+      "prediction": "I think this should be `cmdarray[i]` instead of `cmdarray[1]`.",
+      "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 don't think this is correct. If `subsystem` is empty, it means that we are looking for the cgroups2 path, so we should return `hybridMountpoint`. If `subsystem` is not empty, it means that we are looking for the cgroups1 path, so we should return `legacyMountpoint`.",
+      "reference": "I'd say \"cgroup2 hybrid path\" instead."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we want to keep the `plugin_hooks` in the `config` hash, but I'm not sure if we want to merge them with the `plugin_hooks` in the `options` hash. I think we want to keep the `plugin_hooks` in the `options` hash, but I'm not sure if we want to merge them with the `plugin_hooks` in the `config` hash. I think we want to keep the `plugin_hooks` in the `options` hash, but I'm not sure if we want to merge them with the `plugin_hooks` in the `config` hash. I think we want to keep the `plugin_hooks` in the `options` hash, but I'm not sure if we want to merge them with the `plugin_hooks` in the `config` hash. I think we want to keep the `plugin_hooks` in the `options` hash, but I'm not sure if we want to merge them with the `plugin_hooks` in the `config` hash. I think we want to keep the `plugin_hooks` in the `options` hash, but I'm not sure if we want to merge them with the `plugin",
+      "reference": "I dont see a reason to print URI for `Target`. we do not expose that in inventory v1"
+    },
+    {
+      "prediction": "I think we should just use `*int` here. I don't think we need to distinguish between `0` and `nil`.",
+      "reference": "Do we still need this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Application` struct.",
+      "reference": "`campaignSource` is unused (from `structcheck`)"
+    },
+    {
+      "prediction": "I think it would be better to make this `0.3.0` since that is the latest version on maven central.",
+      "reference": "we should discuss how to not hard-code this version"
+    },
+    {
+      "prediction": "I don't think we need to concatenate `CHANGE_STREAM_OPTIONS` with `CURSOR_OPTIONS` here.",
+      "reference": "do we have a doclet for this event?"
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "Why don't we want to catch segfaults by default?"
+    },
+    {
+      "prediction": "I think this is a bug. The `ZipkinExporter` should be disposing of the `HttpClient` it creates.",
+      "reference": "You could probably dispose request & content if you `await` the SendAsync."
+    },
+    {
+      "prediction": "I think you can remove the `**kwargs` and `sep` arguments from this function.",
+      "reference": "why is this test case here? this file is for testing functionalities in utils.py"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "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'm not sure this is the right thing to do. I think we should just return `nil` and let the caller handle the error.",
+      "reference": "drop the todo?"
+    },
+    {
+      "prediction": "Please use `std::string::compare` instead of `strncmp`",
+      "reference": "Maybe use { } here like we are most other places now."
+    },
+    {
+      "prediction": "Can you add a `@Override` annotation here as well?",
+      "reference": "I want to get PartitionData field type, I don't know how to get it in other way."
+    },
+    {
+      "prediction": "I think this should be `Get(addr basics.Address, cidx basics.CreatableIndex, ctype basics.CreatableType, withPendingRewards bool) (basics.AccountData, error)`",
+      "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 `len > 0` instead of `len > 1`.",
+      "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": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "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": "Line is too long. [101/80]",
+      "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 think this should be `true` to do a Store-Release.",
+      "reference": "I think this should read \"is always observed before\" or \"is never observed after\"."
+    },
+    {
+      "prediction": "Please revert this change.",
+      "reference": "You might consider adding yourself as an author of the class."
+    },
+    {
+      "prediction": "I think this should be `if (freopen(STDERR_FILE, \"a\", stderr) == NULL) { perror(\"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 think this should be `1593493828_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000_000",
+      "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 about this change. I think it should be `DHCPRevOptions.has_key(o) and DHCPRevOptions[o][1] is None`",
+      "reference": "I think your indentation is wrong here."
+    },
+    {
+      "prediction": "I think this should be a `map[string]set.Set[cnet.IPNet]` instead of a `map[string]set.Set[string]`.",
+      "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 if this is the right place to do this, but it seems like a good place to me.",
+      "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 don't think we need this method. We can just set the attribute directly.",
+      "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 don't think this is correct. `findElements` should return `List<WebElement>`, not `List<T>` where `T` extends `WebElement`.",
+      "reference": "This change should also probably go into the corresponding method of the abstract By class?"
+    },
+    {
+      "prediction": "I don't think this should be an error. It should be a warning.",
+      "reference": "Should this be set to something else?"
+    },
+    {
+      "prediction": "I'm not sure what this file is for.",
+      "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 don't think we need the `backend` argument here.",
+      "reference": "Glad to see this generalized to support the backend argument."
+    },
+    {
+      "prediction": "I think we should return an empty list here instead of an unimplemented error.",
+      "reference": "`ctx` is unused in ListDeploymentChains"
+    },
+    {
+      "prediction": "I think this should be `fmt.Errorf(\"failed to prepare running deploy source data: %w\", err)`",
+      "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 it's better to set it to `manageTickerTime` or `manageTickerTime * 2`.",
+      "reference": "This seems relatively high, is it the recommended value from the etcd folks? Or is this debugging cruft?"
+    },
+    {
+      "prediction": "I think this should be `PeerTrackerAttestor` instead of `PeerTrackerAttestor`.",
+      "reference": "This requires updating the name in the `endpoints` tests, please run `make test`."
+    },
+    {
+      "prediction": "`FLAGS_data_path.empty()` -> `FLAGS_data_path.empty() || FLAGS_data_path[0] == '/'`",
+      "reference": "LOG(FATAL) means coredump, LOG(ERROR) is better here."
+    },
+    {
+      "prediction": "I think this should be `handle` instead of `name`.",
+      "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": "nit: I think we can remove the `newDefaultScheduledJob()` here and just use `&ScheduledJob{}`.",
+      "reference": "It's a bit weird to have `svc.go` to include a `ScheduledJobType`...should we rename this file?"
+    },
+    {
+      "prediction": "I think this should be `vendor/` instead of `tpb/`.",
+      "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 `!= 0` instead of `== 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 don't think we need to change the name of this struct.",
+      "reference": "is there anything still referencing `checksumAddress` or can we just delete it?"
+    },
+    {
+      "prediction": "I think this should be `hivev1.ClusterProvisionedCondition` instead of `hivev1.RequirementsMetCondition`.",
+      "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": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `if a.Key == nil || a.Kind == Get {`",
+      "reference": "I believe Key is not necessarily nil, it could be empty string. Probably better check a.Kind == Create"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the cancellation token. I think it would be better to add it to the `WriteAsync` overload that takes a `byte[]` and then call that overload from the `WriteAsync` overload that takes a `Func<PipeWriter, T, long>`. That way the cancellation token would be passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]` and then passed to the `WriteAsync` overload that takes a `byte[]`",
+      "reference": "This is only used for headers and therefore isn't used."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "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 can remove this file.",
+      "reference": "Why do we have this here? I'm not concerned about it really, just curious."
+    },
+    {
+      "prediction": "I'm not sure why these were swapped, but I think it makes more sense to have `SetUintAction` be 1 and `SetBytesAction` be 2.",
+      "reference": "@justicz Why are these switched here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we want to make sure that the help page loads successfully.",
+      "reference": "Maybe verify that an anchor was created -- that the markdown was processed?"
+    },
+    {
+      "prediction": "I think it would be better to make `TwoPhaseIterator.approximation()` return a `DocIdSetIterator` instead of a `DocIdSet`. This way, we don't need to create a new `DocIdSetIterator` in `TwoPhaseIteratorAsDocIdSetIterator`'s constructor.",
+      "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": "why remove this method?",
+      "reference": "why delete this method?"
+    },
+    {
+      "prediction": "nit: I think it would be better to rename this class to `JavaSampleImportTransformer` to make it clear that it's specific to 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": "This import is not used in this file. Please remove it.",
+      "reference": "Could you please clean-up this line while at it?"
+    },
+    {
+      "prediction": "I think this should be `+kubebuilder:validation:MinLength=1`",
+      "reference": "no need to use a pointer. Just string and check for != \"\""
+    },
+    {
+      "prediction": "I think this should be `if (endpointObject.isHttp2Enabled() && endpointObject.isSslEnabled())`",
+      "reference": "this means h2 mode? but how h2c can work?"
+    },
+    {
+      "prediction": "I don't think you need to add your name here.",
+      "reference": "This line should not be there"
+    },
+    {
+      "prediction": "Why do we need to lock here?",
+      "reference": "why do we need lock here? if needed, then we also need to lock in Start()?"
+    },
+    {
+      "prediction": "I think this should be done in `ProtocolGame::playerExecuteParsePacketEvent` instead of here.",
+      "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 if this is a breaking change or not. I think it is, but I'm not sure if it is a breaking change in a good way or a breaking change in a bad way. I think it is a breaking change in a good way, but I'm not sure if it is a breaking change in a good way or a breaking change in a bad way. I think it is a breaking change in a good way, but I'm not sure if it is a breaking change in a good way or a breaking change in a bad way. I think it is a breaking change in a good way, but I'm not sure if it is a breaking change in a good way or a breaking change in a bad way. I think it is a breaking change in a good way, but I'm not sure if it is a breaking change in a good way or a breaking change in a bad way. I think it is a breaking change in a good way, but I'm not sure if it is a breaking change in a good way or a breaking change in a bad way. I think it is a breaking change in a good way, but I'm not sure if it is a breaking change in a good way or a breaking change in a bad",
+      "reference": "unrelated minor fix"
+    },
+    {
+      "prediction": "I don't think we need to disable the spinner here. It's only disabled for the `npm install` command below.",
+      "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 it would be good to have a test for the case where `count` is greater than the number of occurrences.",
+      "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 don't think we need to set the block as a tag here.",
+      "reference": "should just be able to defer the call directly here too"
+    },
+    {
+      "prediction": "I think this should be `y=0.95` instead of `y=1.05`",
+      "reference": "Is this a magic number? I assume this makes it look better."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the only way to get the error message to show up in the journal. I'm not sure why `AsBigInt()` doesn't work here.",
+      "reference": "Would welcome a stringification method on big.Int in specs-actors"
+    },
+    {
+      "prediction": "Why do we need this method?",
+      "reference": "PersistBlockLevelInfo -> UpdateBlockLevelInfo"
+    },
+    {
+      "prediction": "Please add `pretrained=True` in `init_cfg`.",
+      "reference": "Type is redundant since it is inherited."
+    },
+    {
+      "prediction": "`isNotView` -> `isNotViewTable`",
+      "reference": "better use `!isView()` so that we don't need to modify multiple lines in future."
+    },
+    {
+      "prediction": "I think it would be better to make these private methods. I don't think we need to expose them to the user.",
+      "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 don't think we need to check for `item.Type === 'Season' || itemType === 'Studio'` here. The `if (item.Type === 'Program' || item.Type === 'Timer' || item.Type === 'SeriesTimer' || item.Type === 'CollectionFolder' || item.Type === 'UserView' || item.Type === 'Channel' || itemType === 'Season' || itemType === 'Studio') {` line above already checks for `item.Type === 'Program' || item.Type === 'Timer' || item.Type === 'SeriesTimer' || item.Type === 'CollectionFolder' || item.Type === 'UserView' || item.Type === 'Channel' || itemType === 'Season' || itemType === 'Studio'`",
+      "reference": "There's no need to create a new variable here, the other checks just use the item type directly."
+    },
+    {
+      "prediction": "I think it would be better to use `scope` instead of `def self.workshops`",
+      "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'm not sure if this is the right place to put the error handling, but I think it makes sense to do it here.",
+      "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 `from googlecloudprofiler import start as googlecloudprofiler_start` so that we don't have to do `googlecloudprofiler.start()` below.",
+      "reference": "nit: should go before `grpc`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the only way to get the tests to pass.",
+      "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'm not sure if this is the right thing to do, but it seems to work for me.",
+      "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 `config:\"string;local;non-zero,die-on-fail\"`",
+      "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'm not sure about this change. I don't think it's necessary, and I don't think it's a good idea to change the behaviour of the regular expression.",
+      "reference": "This might be a silly question, but what happens to files with durations > 24 hours?"
+    },
+    {
+      "prediction": "I think we should keep the `Func<ActivitySourceAdapter, TInstrumentation> instrumentationFactory` overload.",
+      "reference": "`AddDiagnosticSourceInstrumentation` method can now be eliminated and simply use `AddInstrumentation`"
+    },
+    {
+      "prediction": "I think this should be `expect { Bolt::Project.new(config, pwd) }.to raise_error(/Invalid project name 'puppetlabs-foo' in bolt-project.yaml/)`",
+      "reference": "I think we also need to add `name:` to the bolt-project.yaml files in `spec/fixtures/projects`."
+    },
+    {
+      "prediction": "I don't think we need to return the revision here.",
+      "reference": "worth mentioning the same warning here as in `md_util.go`, I think."
+    },
+    {
+      "prediction": "I don't think we need to check for `nodeList.Items == nil` here. `len(nodeList.Items) == 0` will be true if `nodeList.Items` is nil.",
+      "reference": "Do we need both the checks, for `Items` not nil and `len(Items)`"
+    },
+    {
+      "prediction": "I'm not sure what this table is for, but it seems like it should be removed.",
+      "reference": "Do you know what this is coming from? It keeps getting deleted/created @nickbristow"
+    },
+    {
+      "prediction": "I don't think we need to remove `WAREHOUSE_LOCATION` since it is still used in `HiveCatalogFactory`.",
+      "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 don't think we should deprecate this method. It's not used anywhere in the codebase and it's not part of the public API. I think we should just remove it.",
+      "reference": "`The %s is deprecated...` -> `The %s method is deprecated...`"
+    },
+    {
+      "prediction": "I don't think this should be in specs-actors. I think it should be in go-filecoin.",
+      "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.MinimumSegmentSize / 2`",
+      "reference": "Nit: Might as well make this same change to AdaptedPipeline."
+    },
+    {
+      "prediction": "I don't think this is correct. `OffsetDateTime.of` creates a new `OffsetDateTime` with the given `LocalDateTime` and `ZoneOffset.UTC` as the offset. This means that if the original `OffsetDateTime` had a different offset, it will be lost. For example, `OffsetDateTime.of(LocalDateTime.of(1970, 1, 1, 0, 0, 0), ZoneOffset.ofHours(1))` will be converted to `OffsetDateTime.of(LocalDateTime.of(1970, 1, 1, 0, 0, 0), ZoneOffset.UTC)`.",
+      "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": "I think this should be `a-zA-Z0-9_` instead of `a-zA-Z0-9`",
+      "reference": "Based on username requirements for forge.puppet.com - only letters and digits permitted."
+    },
+    {
+      "prediction": "I don't think this change is correct. If the sizes are different, we should return false.",
+      "reference": "Was this removed to ignore the extra columns coming from the file projection?"
+    },
+    {
+      "prediction": "I think this should be `// +build !windows,linux`",
+      "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": "I think it would be better to inject the `AWSEmailProvider` into the `EmailNotificationService` instead of instantiating it here.",
+      "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": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "+1 for creating this script to help people get boot strapped"
+    },
+    {
+      "prediction": "I think this should be `data-preact-` instead of `preact-`.",
+      "reference": "Do you need the prefix?"
+    },
+    {
+      "prediction": "I think we need to check the return code of deleteFromTextTable() and return an error if it fails.",
+      "reference": "if retcode is < 0, it should return -1 indicating an error."
+    },
+    {
+      "prediction": "I think this should be `# pylint: disable=locally-disabled,useless-suppression,suppressed-message,line-too-long,missing-docstring,unsubscriptable-object,too-few-public-methods,redefined-outer-name,use-dict-literal,iterating-modified-dict`",
+      "reference": "Are those necessary? I feel like `useless-suppression` could be avoided here?"
+    },
+    {
+      "prediction": "I don't think you need the `return` here.",
+      "reference": "Is this equivalent in JavaScript?"
+    },
+    {
+      "prediction": "`LOG(INFO)` -> `LOG(WARN)`",
+      "reference": "get/list operation is not schema change, it is supposed to use VLOG"
+    },
+    {
+      "prediction": "I think this should be `ReadOnlySpan<KeyValuePair<string, object?>>` instead of `ReadOnlySpan<KeyValuePair<string, object>>`.",
+      "reference": "we'll need to delete this whole file, right?"
+    },
+    {
+      "prediction": "I think it would be good to add a note here that this refresh is a no-op if the table is not associated with a TransactionalCatalog.",
+      "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 these TODOs now?",
+      "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": "Metrics/LineLength: Line is too long. [101/80]",
+      "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": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "reference": "These files should in `/javascript/brython`"
+    },
+    {
+      "prediction": "I think this should be `break runLoop`",
+      "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 don't think we need this anymore.",
+      "reference": "Can we fix this, or is it going to stay broken."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't think we want to run these tests on postgres.",
+      "reference": "Is there a reason to remove this attr?"
+    },
+    {
+      "prediction": "`MessageLayer` should be `MessageLayerType` or `MessageLayerType` should be renamed to `MessageLayer`",
+      "reference": "log message should be started with upper-case word."
+    },
+    {
+      "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": "I guess this is not needed anymore now?"
+    },
+    {
+      "prediction": "I think it would be better to keep `flags` as an argument, and set it to `H2O_HEADER_FLAG_NONE` in the caller.",
+      "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 think this should be `this.startWorkspace_.setResizesEnabled(true);`",
+      "reference": "Is this a different fix?"
+    },
+    {
+      "prediction": "I think it's better to use `if req.WithoutNS { ... }` instead of `if !req.WithoutNS { ... }`",
+      "reference": "Why not add a `EnterNS` filed instead of `WithoutNS`?"
+    },
+    {
+      "prediction": "I don't think these are used anywhere.",
+      "reference": "Same points as in other test file."
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I think it would be better to have a separate setting for each task, so that you can have different settings for different tasks.",
+      "reference": "I have a PR pending, can you change this bit index to `16`?"
+    },
+    {
+      "prediction": "I don't understand why this change is needed. `classOrInterfaceType.toString()` returns the fully qualified name of the type, so `typeName.indexOf('.') > -1` should always be true.",
+      "reference": "we should probably have some utility class for this sort of things"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "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 `static exists(config: Realm~Configuration): boolean`",
+      "reference": "does it return something?"
+    },
+    {
+      "prediction": "Shouldn't this be `data` instead of `error`?",
+      "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 `DescribeIdentifier(sha string) (string, error)`.",
+      "reference": "super nit: `revision string` (the passed value might not be a SHA hash)."
+    },
+    {
+      "prediction": "Why is this changed to internal?",
+      "reference": "@snakefoot why is internal needed?"
+    },
+    {
+      "prediction": "Why do we need to disable health check here?",
+      "reference": "Why were we disabling health checks before?"
+    },
+    {
+      "prediction": "I think this should be `'/conftest.py'` instead of `'/conftest.py'`.",
+      "reference": "did `conftest.py` actually show up for you? `test_files` should never contain it..."
+    },
+    {
+      "prediction": "I don't think we need to check for the external schema name here. If it is an external schema name, then it will not be a reserved schema name. So, we can just check for the reserved schema name here.",
+      "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'm not sure this is the right place to add this column. I think it should be added to the `onUpgrade` method of the `PodDBAdapter` class.",
+      "reference": "Please change to `1070400`. I promise to look into this PR in more detail before the 1.7.4 release ;)"
+    },
+    {
+      "prediction": "Why do we need to cast `fStruct` to `RowType` here?",
+      "reference": "Can we change to use `LogicalType.getChildren`?"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "I'm confused as to how these are both passing."
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "Prefer single-quoted strings when you don't need string interpolation or special symbols."
+    },
+    {
+      "prediction": "I think this should be `string[]|null` instead of `string[]|null[]`",
+      "reference": "there is `\"` sign in commit message and colon. can you rename also `descrition` to plural `descriptions` and add there `$` ?"
+    },
+    {
+      "prediction": "I don't think we need to record the userLatencyDuration here. It's already recorded in the `metricsScope.RecordTimer(metrics.ServiceLatencyNoUserLatency, timerNoUserLatency.Seconds())` line above.",
+      "reference": "timerNoUserLatency is not being used?"
+    },
+    {
+      "prediction": "I think this should be `isinstance(attr, six.string_types)`.",
+      "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             AssertEx.DeserializeNoError<StaticMod>(\"<static_modification name=\\\"Loss4\\\" aminoacid=\\\"T, S\\\" formula=\\\"HPO3\\\" explicit_decl=\\\"true\\\"><potential_loss formula=\\\"HP3O4\\\"/><potential_loss formula=\\\"H2O\\\"/><potential_loss formula=\\\"NH3\\\"/></static_modification>\", true, true, structuralModificationType);\n             AssertEx.DeserializeNoError<StaticMod>(\"<static_modification name=\\\"Loss5\\\" aminoacid=\\\"T, S\\\" formula=\\\"HPO3\\\" explicit_decl=\\\"true\\\"><potential_loss formula=\\\"HP3O4\\\"/><potential_loss formula=\\\"H2O\\\"/><potential_loss formula=\\\"NH3\\\"/></static_modification>\", true, true, structuralModificationType);\n             AssertEx.DeserializeNoError<StaticMod>(\"<static_modification name=\\\"Loss6\\\" aminoacid=\\\"T, S\\\" formula=\\\"HPO3\\\" explicit_decl=\\\"true\\\"><potential_loss formula=\\\"HP3O4\\\"/><potential_loss formula=\\\"H2O\\\"/><potential_loss formula=\\\"NH3\\\"/></static_modification>\", true, true, structuralModificationType);\n```\nReview comment?",
+      "reference": "Why change the name in the XML?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [118/100]",
+      "reference": "This is much cleaner. makes it easier to tell what happens on a save."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "reference": "Could you add return error unsupported operator here"
+    },
+    {
+      "prediction": "I don't think we need to import `Empty` and `Cons` here.",
+      "reference": "(Mhh, we seem to use different formatters - we should unify them. I like the wildcards)"
+    },
+    {
+      "prediction": "I don't think we need to set `kind` here, it will default to `normal` if not set.",
+      "reference": "Should `kind` be a command line parameter also?"
+    },
+    {
+      "prediction": "I think it would be better to use `transaction.AccessList is null ? null : AccessListItemForRpc.FromAccessList(transaction.AccessList);` instead of `TryGetAccessListItems()`",
+      "reference": "can we have it inside FromAccessList?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to just return the inner error and let the caller decide what to do with it.",
+      "reference": "Why not embed filterASCII in this function?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the flow is still in the queue, it means that it has not been processed yet, so it should not be counted as processed.",
+      "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` instead of `# pylint: enable=protected-access`",
+      "reference": "You can probably move this up after the `._Numeric` line."
+    },
+    {
+      "prediction": "This is not needed. `super` is called implicitly in Python 3.",
+      "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": "Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>",
+      "reference": "Do we need indices for this table?"
+    },
+    {
+      "prediction": "I don't think this is needed. The fuzzers should be able to call `process_data()` and `check_connection()` directly.",
+      "reference": "what about making this function static inline to avoid the extra declaration ?"
+    },
+    {
+      "prediction": "Why not use `uint32` instead of `GroupIDType`?",
+      "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 think this should be `Will always be 1.0 unless specified otherwise.`",
+      "reference": "would be `DEPRECATED Percentage of the graph (in vertices) to contract [0..1]` better?"
+    },
+    {
+      "prediction": "I think we should rename this to `CAST_TAB_MIRRORING` to be consistent with the rest of the commands.",
+      "reference": "These command names are specific to Chromium-based browsers. Please move to `ChromiumDriverCommand`"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "if we won't use, should we remove this?"
+    },
+    {
+      "prediction": "I think this should be `MvvmCross.Platforms.Android.Views` and `MvvmCross.Platforms.Android.Binding.Views`",
+      "reference": "Having 2 items with the same key (Mvx) won't work in a dictionary"
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "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": "I'm not sure if this is the right thing to do, but I think we should limit the length of `message` to 512 characters.",
+      "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: extra blank line",
+      "reference": "Nit: unnecessary whitespace change."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do.",
+      "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 an `if (label >= 0 && label < m_num_labels) { ... } else { ... }` instead of `if (label >= 0 && label < m_num_labels) { ... } else { ... }`.",
+      "reference": "I would use static_cast<label_t>(0) and static_cast<label_T>(m_num_labels) just in case."
+    },
+    {
+      "prediction": "I think this should be `toLowerCase(Locale.ENGLISH)`.",
+      "reference": "Even if we want to keep in original case - domain will be lower-cased."
+    },
+    {
+      "prediction": "I think this migration is not needed.",
+      "reference": "Based on discussion with Chad it feels like trail_map_json could be a good column name."
+    },
+    {
+      "prediction": "This should be `backend::lang.permissions.allow` etc.",
+      "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": "I'm not sure this is the right thing to do, but it seems to be the only way to get it to work.",
+      "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 best way to do this, but I think it should be fine for now.",
+      "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": "Why do we need to change this to `WriteBatch` instead of `Commit`?",
+      "reference": "sloppyReassign: re-assignment to `err` can be replaced with `err := kv.WriteBatch(batchForBlock)` (from `gocritic`)"
+    },
+    {
+      "prediction": "I don't think this is correct. If `part` is `null` we should not return a `CompletableFuture` that is already completed. Instead we should return a `CompletableFuture` that will be completed with an exception.",
+      "reference": "using factory method. return CompletableFuture.completedFuture"
+    },
+    {
+      "prediction": "I think this should be `Google Analytics` instead of `Analytics`.",
+      "reference": "The `_x` function needs to be imported at the top of the file (in addition to `__`)"
+    },
+    {
+      "prediction": "I think we should keep the `ofctrl.Flow` as a field of `typeFlow` instead of a pointer to `ofctrl.Flow`. This way, we don't have to dereference `ofctrl.Flow` all the time, and we don't have to worry about whether the `ofctrl.Flow` is nil or not.",
+      "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 don't think this is needed.",
+      "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 for now.",
+      "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 `time.Now().UnixNano() / 1000000`",
+      "reference": "Isn't the point of the clock module to encapsulate access to `time.Now()`? Coming later?"
+    },
+    {
+      "prediction": "I think this should be `fastParams.AgreementFilterTimeoutPeriod0 = 500*time.Millisecond`",
+      "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": "I'm not sure if this is the right thing to do, but it seems to be the only way to get `curl` to follow redirects.",
+      "reference": "Pretty sure the single quotes here should be double quotes, single quotes are not expanded by the shell."
+    },
+    {
+      "prediction": "Shouldn't this be `Validate.notBlank(stringArg, \"message %s %s\", \"msg\", \"msg\");`?",
+      "reference": "these are not really constants, right?"
+    },
+    {
+      "prediction": "This is not needed, please remove it.",
+      "reference": "import below copyright statement"
+    },
+    {
+      "prediction": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `array|string|Markup`",
+      "reference": "`if (is_string($value) && $this->getDefinition()->get('allow_html')) {`"
+    },
+    {
+      "prediction": "I don't think we should use `Debug.Assert` here. We should use `ArgumentNullException.ThrowIfNull(key, nameof(key));` instead.",
+      "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": "Please remove this line.",
+      "reference": "I thought that the main was in catch_qt.cpp?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do it, but it works for now.",
+      "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": "nit: I think we can remove the `containerEvent.ManagedAgents != nil` check and just use `assert.Equal(t, apicontainerstatus.ManagedAgentRunning, containerEvent.ManagedAgents[0].Status, \"expected managedAgent container state change event did not match actual event\")`",
+      "reference": "I'd prefer to timeout on this, but I know it's not a pattern that is being followed"
+    },
+    {
+      "prediction": "nit: remove extra blank line",
+      "reference": "hmm let's not use * imports please"
+    },
+    {
+      "prediction": "I think we should add a check for `am.BearerTokenFile != \"\" && am.BearerToken != \"\"` and return an error if both are set.",
+      "reference": "is this configuration key already exist in prometheus ?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [84/80]",
+      "reference": "As @gravitystorm said new tests need to be using factories, not fixtures."
+    },
+    {
+      "prediction": "I think we should keep the `addr_` and `port_` in the prompt.",
+      "reference": "You can replace IP by hostname."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it should be done here. The reason is that if the device is a braille display, we want to use the braille driver, not the keyboard driver.",
+      "reference": "This should say \"why\" NVDA should do things in this order."
+    },
+    {
+      "prediction": "I think this should be `Bucket: aws.String(\"bucket\")`",
+      "reference": "nit these will get changed back during next release."
+    },
+    {
+      "prediction": "I think we can remove this file. It's not used anywhere.",
+      "reference": "it is actually not needed to make the member public. the build tags can also set a package scoped variable."
+    },
+    {
+      "prediction": "I think this should be `iconClass` instead of `icon` to be consistent with the other properties.",
+      "reference": "We already have a method for it 'icon' => $content->getIcon()"
+    },
+    {
+      "prediction": "nit: `WrapEntry` -> `wrapEntry`",
+      "reference": "Instead of importing `golang.org/x/net/context` The SDK should use `aws.BackgroundContext()` instead of `context.Background()`"
+    },
+    {
+      "prediction": "`findOneBy` should be used instead of `find`",
+      "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": "I'm not sure if this is the best way to go about it, but it seems to be the only way to get all the topics.",
+      "reference": "I'm not 100% sure, but I think this should be `explorable`. If not, I think `explorable` can be removed entirely."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to be the only way to get this to work on Python 3.",
+      "reference": "Note to self: Why was this not originally not using `stderrbytes`?"
+    },
+    {
+      "prediction": "I think it's better to use `LOG(WARNING) << \"Rename \" << src << \" to \" << dst << \" failed, the errno: \" << ::strerror(errno);` instead of `LOG_IF(WARNING, status < 0) << \"Rename \" << src << \" to \" << dst << \" failed, the errno: \" << ::strerror(errno);` because `LOG_IF(WARNING, status < 0) << \"Rename \" << src << \" to \" << dst << \" failed, the errno: \" << ::strerror(errno);` will only log when `status < 0`, but `LOG(WARNING) << \"Rename \" << src << \" to \" << dst << \" failed, the errno: \" << ::strerror(errno);` will always log.",
+      "reference": "Is betterLOG_IF(ERROR, status != 0) ?"
+    },
+    {
+      "prediction": "I think this should be `self.collection = self.bucket = 'default'`",
+      "reference": "I wouldn't call it default too."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get this to work.",
+      "reference": "Why disable the rule for this file?"
+    },
+    {
+      "prediction": "I think this should be `external` instead of `ctaLinkExternal`.",
+      "reference": "Thanks for adding this :+1:. Could you please move it to go after the `ctaLink` prop?"
+    },
+    {
+      "prediction": "I don't think it's safe to assume that `_array[_index]` will never be null.",
+      "reference": "The other way around Current can be null."
+    },
+    {
+      "prediction": "I don't think this is the right place to add this.",
+      "reference": "why does it have to be here? Due to breadcrumb navigation?"
+    },
+    {
+      "prediction": "This function is not used in this PR. Can you please remove it?",
+      "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 think it would be better to use `str` instead of `repr` here.",
+      "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 it's better to use `model.getConfig()` instead of `model.getConfig().get(key)` in `getProperty(String key)` method.",
+      "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": "`action.GasLimit` should be `blockchain.GasLimit`",
+      "reference": "line is 165 characters"
+    },
+    {
+      "prediction": "Why do you need this class?",
+      "reference": "Please add a copyright notice. Also, could you tell me please why do we need this empty provider here?"
+    },
+    {
+      "prediction": "I think this should be `if (maybePrivacyGroup.isEmpty() || !maybePrivacyGroup.get().getMembers().contains(enclavePublicKey))`",
+      "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'm not sure this is the right thing to do.",
+      "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 this is the right thing to do. I think we should allow dragging an input to an output, but not dragging an output to an input.",
+      "reference": "Glad we fixed this as well. Blockly is going to be left as-is for this case, right?"
+    },
+    {
+      "prediction": "This class is not deprecated.",
+      "reference": "I had accidentally deprecated this class. This is meant to stick around. Only `SalesforceObjectType` goes away."
+    },
+    {
+      "prediction": "I don't think this is necessary. `hex(i)[2:].rstrip('L')` will always return a string that starts with '0x'.",
+      "reference": "`hex(i)[2:]` is trying to do the same thing above. what is `i` in your malformed case?"
+    },
+    {
+      "prediction": "I don't think this should be part of the method name. It should be part of the method signature.",
+      "reference": "I think this change is redundant; see the previously defined method \"writeUnescapedXML\"."
+    },
+    {
+      "prediction": "I think this should be `Commitment` instead of `Commitments`.",
+      "reference": "Ah now I have at least some idea what these are for."
+    },
+    {
+      "prediction": "I think this should be `this.readSchema = Schema.applyAliases(schema, readSchema);`",
+      "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 don't think this is the right place to do this. I think it should be done in the caller of this function.",
+      "reference": "Do we really need separate error in manager if utils.ErrRequestCancelled is the only error which indicates cancelation ?"
+    },
+    {
+      "prediction": "I think this should be `tm_cpp`",
+      "reference": "Leave as `cpp` since that still gives better results."
+    },
+    {
+      "prediction": "I think this should be `func PausedPredicates(logger logr.Logger) predicate.Funcs {`",
+      "reference": "temp change so the predicate can be used by the `exp` package"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `__init__` method of the `KeyboardInputGesture` class. That way, you don't have to worry about the `isCharacter` property being set to `True` after the `__init__` method has been called.",
+      "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'm not sure this is the right place to put this TODO. I think it should be in the `deployClusterImageSets` function.",
+      "reference": "@dgoodwin Will it cause an issue when the issue get fixed in OLM?"
+    },
+    {
+      "prediction": "I don't think we need `All` anymore.",
+      "reference": "This option is now unnecessary"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "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 `deleteTerms.keySet().forEach(term -> {`",
+      "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 this should be 4 instead of 8.",
+      "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 don't think we need to join the keys here. The server will do that for us.",
+      "reference": "Also update line 2205 below"
+    },
+    {
+      "prediction": "This should be `STATE_KEY_FIRST_TIME_EMPTY_SUBJECT`",
+      "reference": "Typo, should read `firstTimeEmptySubject`"
+    },
+    {
+      "prediction": "I'm not sure this is correct. The transport layer can handle both unary and streaming requests.",
+      "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 thing to do. I think we should just raise the exception and let the caller handle it.",
+      "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 `False` by default.",
+      "reference": "I wonder if this is a fitting name. Why \"expands user\"? Wouldn't be `user_expanded` or so be more fitting?"
+    },
+    {
+      "prediction": "`_isHidden` should be `isHidden`",
+      "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 this should be `csharp/gapic_samples_csproj.snip`",
+      "reference": "nit: for consistency, have this line and the previous formatted similarly"
+    },
+    {
+      "prediction": "I think it would be better to use `Path` instead of `String` here.",
+      "reference": "As before, shouldn't mention these as being orion options I don't think. Perhaps just enclaveKeyStoreFile etc."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be good to add a `console.log` here to indicate that the test is running. Something like `console.log( 'Running tests...' );`",
+      "reference": "What's the thinking here? Only logging when the test has not ended yet?"
+    },
+    {
+      "prediction": "I think it's better to use `is_log_op = f.__name__ in log_ops` instead of `any(f == getattr(spark.Column, f'__{log_op}__') for log_op in log_ops)`",
+      "reference": "@HyukjinKwon @ueshin (cc @itholic @charlesdong1991 ) Not sure if this is the right implementation ..."
+    },
+    {
+      "prediction": "I don't think this is the right thing to do here. If the URL is malformed, we should just return the error to the user, and let them know that the URL is not valid.",
+      "reference": "Do we want to add any specific error on how the url is malformed?"
+    },
+    {
+      "prediction": "I don't think this is correct. If `lastLogIdSent_` is less than `logIdToSend_`, then we should not increment `lastLogIdSent_`. Instead, we should set `lastLogIdSent_` to `logIdToSend_` and then increment `logIdToSend_`.",
+      "reference": "why NOT self->lastLogIdSent_ = self->logIdToSend_"
+    },
+    {
+      "prediction": "I think this should be `hivev1.ProvisionedProvisionedReason` instead of `hivev1.ProvisionedProvisionedReason`.",
+      "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'm not sure if this is the right thing to do, but it seems to be the only way to get the tab to open in the foreground.",
+      "reference": "This seems like another unrelated change I've done in `master`."
+    },
+    {
+      "prediction": "I think this should be `expect { config.add_formatter :documentation }.not_to change { config.formatters.length }`",
+      "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 don't think we need 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": "This should be `mIsFirstContentfulPaint`",
+      "reference": "nit: `Contentful` is the word used in the GV API, without the uppercase in F"
+    },
+    {
+      "prediction": "I think it's better to throw an exception here.",
+      "reference": "It just does the check, maybe we can name it as checkOperations."
+    },
+    {
+      "prediction": "I think this should be `<div id=\"t1\"><span></span></div><div id=\"t2\"><span></span></div>`",
+      "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 best place to put this, but I think it makes the most sense here.",
+      "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": "nit: `recieveUpdates` -> `receiveUpdates`",
+      "reference": "add some form of docs in docstring or on :74 for what the effect is"
+    },
+    {
+      "prediction": "I don't think we need to import `json` here. We can just use `json.loads` and `json.dumps`.",
+      "reference": "Couldn't you import `json` from `kinto.core.utils` here too?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [105/80]",
+      "reference": "Minor: Hmmm...is it worth setting this explicitly for the tests that it affects?"
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "reference": "This was added to fix a race condition and you are reverting it here, right?"
+    },
+    {
+      "prediction": "I don't think this is correct. `eth_coinbase` should always be available, even in `--chaindata` mode. It's just that it will always return `0x0000000000000000000000000000000000000000`.",
+      "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": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "`new_ncr_work_order_path` ? (I've been slowly moving specs over to that way of calling paths)"
+    },
+    {
+      "prediction": "I'm not sure about the `?` here. I think it should just be `application/x-shockwave-flash`",
+      "reference": "Should flash be included in the patterns too? It currently isn't."
+    },
+    {
+      "prediction": "Shouldn't this be `Left` and `Right`?",
+      "reference": "I suspect I will have to keep looking whether Left/Right refers to contract/Implementation while working on the rules code."
+    },
+    {
+      "prediction": "Shouldn't this be `Microsoft.Rest.Azure.Core.IResource`?",
+      "reference": "(ignore this. testing a codeflow bug)"
+    },
+    {
+      "prediction": "I think this should be `CASSANDRA_TLS_VERIFY_SERVER_HOSTNAME`",
+      "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. [84/80]",
+      "reference": "this will still fail on systemd"
+    },
+    {
+      "prediction": "This should be `if isinstance(body, bytes):`",
+      "reference": "The convention in most of the codebase is to use `if isinstance(body, bytes):` instead."
+    },
+    {
+      "prediction": "I think this should be `comp.rsplit('.', 1)[-1]` so that we don't split on `.` in the middle of a word.",
+      "reference": "I think this would be better as: `label = comp.rsplit('.', 1)[-1]`"
+    },
+    {
+      "prediction": "I think we can remove the `Element` type parameter here, since it's not used in the method.",
+      "reference": "We have been creating a language-level context interface for each language instead of parameterizing the generate function."
+    },
+    {
+      "prediction": "`$media->getContext() ?? ''` should be `$media->getContext() ?: ''`",
+      "reference": "Same thing about context here."
+    },
+    {
+      "prediction": "why do we need this?",
+      "reference": "there is a function to get namespace entry from mutable state directly"
+    },
+    {
+      "prediction": "I think this should be `@inventory.config.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 don't think this is correct. You can omit the dba and ddev-ssh-agent containers globally, but not the db container.",
+      "reference": "Sorry, typo s/unusuable/unusable/"
+    },
+    {
+      "prediction": "I think this should be `if (options.containsKey(\"type\") && options.get(\"type\").equalsIgnoreCase(\"hive\") && options.containsKey(\"uri\") && !options.get(\"uri\").equals(sparkSession.conf().get(\"spark.sql.catalog.\" + name + \".uri\"))) {`",
+      "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 think this should be `return 1 if content_type and content_type.startswith(\"multipart/form-data\") else 0`",
+      "reference": "see above - this is only used to select the correct view, we don't need to handle the boundary information here."
+    },
+    {
+      "prediction": "I think this should be `gen.SetCertificateRequestCertificateAuthority(certPEM)`",
+      "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": "Shouldn't this be `if (excessDomChildren != null && excessDomChildren.indexOf(dom) > -1)`?",
+      "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 don't think this is necessary. The `$destroy` event is only fired when the element is removed from the DOM. If the element is not removed from the DOM, then the `mousewheel` event will still be bound to the element.",
+      "reference": "this looks like it may be something that needs additional refactoring"
+    },
+    {
+      "prediction": "nit: remove this empty line",
+      "reference": "Nit: Remove whitespace (and below)"
+    },
+    {
+      "prediction": "This should be `Only files are allowed. You've selected a directory.`",
+      "reference": "\"Failed opening\" or \"Failed to open\"."
+    },
+    {
+      "prediction": "I think this should be `if d.SourceImage == \"\" && d.SourceSnapshot == \"\" {`",
+      "reference": "and -> or"
+    },
+    {
+      "prediction": "I think this should be `yield User.collection.createIndex({ password: 1 });` and `yield User.collection.createIndex({ email: 1 });` instead of `yield User.collection.createIndex({ password: 1 });` and `yield User.collection.createIndex({ email: 1 });`",
+      "reference": "Add a `yield User.init()` here to wait for all indexes to finish building"
+    },
+    {
+      "prediction": "I think it's better to use `DEFINE_string(local_ip, \"0.0.0.0\", \"local ip\")`",
+      "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 think this should be `if (!$this->hasCustomData()) { return false; } return $this->getCustomData()->delete();`",
+      "reference": "@DanHarrin please add a `return true` at the bottom of this method, as the docblock indicates a boolean return value."
+    },
+    {
+      "prediction": "I don't think we need to make this public.",
+      "reference": "This is used in o.a.l.sandbox.search.BM25FQuery."
+    },
+    {
+      "prediction": "I think this should be `queryRelation == QueryRelation.CONTAINS`",
+      "reference": "Shouldn't this work as well for Disjoint?"
+    },
+    {
+      "prediction": "I think this should be `except RuntimeError as e:` so that we can print the error message.",
+      "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 `#include \"lbann/data_store/data_store_jag.hpp\"` instead of `#include \"lbann/utils/lbann_library.hpp\"`.",
+      "reference": "Why do you need to load a specific data reader in lbann_library?"
+    },
+    {
+      "prediction": "I don't think we need this step at all. We can just remove it.",
+      "reference": "why not ValueTask?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we want to requeue if the installer image is not set, but we don't want to requeue if the installer image is set but the CLI image is not set.",
+      "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 think we can remove the `location` parameter from `newTableMetadata` and just use `ops.location()`.",
+      "reference": "Is it necessary to change this file? Doesn't `tableMetadata` call `newTableMetadata`?"
+    },
+    {
+      "prediction": "I'm not sure why these tests were removed.",
+      "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 don't think we need to log the entire list of tables here. I think we can just log the namespace and the number of tables.",
+      "reference": "What is the purpose of this debug message? I'm not sure how it would help."
+    },
+    {
+      "prediction": "nit: I'd prefer to keep the `extutil` package around for now, even if it's not used in this extractor.",
+      "reference": "Is there a corresponding BUILD dependency to prune?"
+    },
+    {
+      "prediction": "nit: `goodContentType` and `badContentType` could be `goodMime` and `badMime`",
+      "reference": "The \"good\" and \"bad\" descriptions confused me a bit. Maybe `allowedContentType` and `disallowedContentType`? Not sure if that's better."
+    },
+    {
+      "prediction": "I don't think this is the right message. It should be something like \"Session ID is null. Did you call quit()?\" or \"Session ID is null. Did you call close()?\" or \"Session ID is null. Did you call shutdown()?\" or \"Session ID is null. Did you call stop()?\" or \"Session ID is null. Did you call destroy()?\" or \"Session ID is null. Did you call terminate()?\" or \"Session ID is null. Did you call exit()?\" or \"Session ID is null. Did you call kill()?\" or \"Session ID is null. Did you call stopServer()?\" or \"Session ID is null. Did you call stopWebDriver()?\" or \"Session ID is null. Did you call stopDriver()?\" or \"Session ID is null. Did you call stopBrowser()?\" or \"Session ID is null. Did you call stopApplication()?\" or \"Session ID is null. Did you call stopProcess()?\" or \"Session ID is null. Did you call stopService()?\" or \"Session ID is null. Did you call stopApplicationServer()?\" or \"Session ID is null. Did you call stopApplicationDriver()?\" or \"Session ID is null. Did you call",
+      "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": "Why is this increased?",
+      "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": "I'm not sure this is the right thing to do here. If `flux_respond_error` fails, it will return -1 and `flux_respond` will be called with `errno` set to `EPROTO`, which is probably not what we want.",
+      "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 think we can remove the `sendAnalyticsTrackingEvent` function now that we're using `trackEvent` instead.",
+      "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": "Why do we need to create a new variable here?",
+      "reference": "Good catch, and if we don't declare the newSchedule here, will the last schedule be executed in multi goroutine?"
+    },
+    {
+      "prediction": "I think this should be `project is not null` instead of `project != null`",
+      "reference": "`project` should never be null here, unless we have projects that aren't in the solution? Which we shouldn't have"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Simple `for` loop instead of `for_each` would help here by making it more readable."
+    },
+    {
+      "prediction": "I think this should be \"Conf height processing\"",
+      "reference": "Looks like this falls through, should break."
+    },
+    {
+      "prediction": "I think this should be `true` to overwrite the existing manifest list file.",
+      "reference": "Let's default manifest lists and manifests to overwrite. These use UUID-based file names and should never conflict."
+    },
+    {
+      "prediction": "This should be `showSpeechViewerAtStartup = boolean(default=false)`",
+      "reference": "Please rename [speechView] to [speechViewer]."
+    },
+    {
+      "prediction": "I think this should be `PercpuUsageInKernelmode` and `PercpuUsageInUsermode`",
+      "reference": "I would use shorter yet still descriptive names, e.g. `KernelPerCpu` and `UserPerCpu`."
+    },
+    {
+      "prediction": "I think this should be `if format.lower().startswith('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": "This should be `self._commands[\"ExecuteCDP\"] = ('POST', '/session/$sessionId/goog/cdp/execute')`",
+      "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 don't think we need to set the `language` tag here. The `language` tag is already set in the `StartActive` method.",
+      "reference": "This is also missing in CreateScopeFromExecuteAsync. Can you add that there too?"
+    },
+    {
+      "prediction": "I don't understand why this is needed. Can you provide an example?",
+      "reference": "I hope that this does not break anything... Probably needs detailed beta tests."
+    },
+    {
+      "prediction": "I think this should be `expect(config.start_time).to be_an_instance_of ::Time`",
+      "reference": "@JonRowe in what scenario do you see this being manually set?"
+    },
+    {
+      "prediction": "I think this should be `if (origin_a == nano::block_origin::local)`",
+      "reference": "I think the work \"trigger\" makes more sense than \"check\" because this is taking an action based on status."
+    },
+    {
+      "prediction": "`RegistrationManager.INSTANCE.getMicroservice().getProperties().put(DefinitionConst.INSTANCE_PUBKEY_PRO, rsaKeyPairEntry.getPublicKeyEncoded());` should be `RegistrationManager.INSTANCE.getMicroserviceInstance().getProperties().put(DefinitionConst.INSTANCE_PUBKEY_PRO, rsaKeyPairEntry.getPublicKeyEncoded());`",
+      "reference": "add a configuration , default put public key in Microservice, if the configuration is set, then put in Instance."
+    },
+    {
+      "prediction": "I think this should be `ASSIGNMENT` instead of `DEFINITION`.",
+      "reference": "WDYT about the idea above of making these all verbs that correspond to the input spec?"
+    },
+    {
+      "prediction": "This should be `public static final ServerSettings.Type STORE_TYPE = ServerSettings.Type.IMAP;`",
+      "reference": "With the introduction of the enum(s) the constants in the store classes should be removed."
+    },
+    {
+      "prediction": "I think this should be `ip_header->ihl > 5 * sizeof(u32)`",
+      "reference": "I would say `!= 5`; A packet with <5 would be malformed."
+    },
+    {
+      "prediction": "Why is this change needed? I don't see it in the PR description.",
+      "reference": "I think you missed removing `Stateful` from `push/config.go`"
+    },
+    {
+      "prediction": "I don't think we need `internal` here.",
+      "reference": "why would you make it internal? if so then at least protected internal"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this test. I think it should be in `test_tcp_stream_modify.py` in the `scripts` directory.",
+      "reference": "We should check if the response (`d`) contains bar as response, screw the log. :smile:"
+    },
+    {
+      "prediction": "I think this should be `transferred` instead of `transfered`",
+      "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().parent`",
+      "reference": "I don't understand how this work, could you explain ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we need to make sure that we don't write to a SIMD slot that we didn't read from. So we need to make sure that `ci->num_simd_used` is at most `MCTX_NUM_SIMD_SLOTS`. I'm not sure if this is the right thing to do, but I think we need to make sure that we don't write to a SIMD slot that we didn't read from. So we need to make sure that `ci->num_simd_used` is at most `MCTX_NUM_SIMD_SLOTS`. I'm not sure if this is the right thing to do, but I think we need to make sure that we don't write to a SIMD slot that we didn't read from. So we need to make sure that `ci->num_simd_used` is at most `MCTX_NUM_SIMD_SLOTS`. I'm not sure if this is the right thing to do, but I think we need to make sure that we don't write to a SIMD slot that we didn't read from. So we need to make sure that `ci->num_simd_used` is at most `MCTX_NUM_SIMD_SLOTS`.",
+      "reference": "The abbreviation for \"context\" used extensively inside DR is \"cxt\", not \"ctx\", so s/MCTX/MCXT/."
+    },
+    {
+      "prediction": "`\u06a9\u0686\u06be \u0633\u06cc\u06a9\u0646\u0688 \u0645\u06cc\u06ba` instead of `:seconds \u0633\u06cc\u06a9\u0646\u0688 \u0645\u06cc\u06ba`",
+      "reference": "here is `:seconds` missing again"
+    },
+    {
+      "prediction": "I think it would be better to have a default value instead of `null` here.",
+      "reference": "initializing to null makes me uncomfortable. let's have a default value"
+    },
+    {
+      "prediction": "Shouldn't this be `if (pathStart == -1 || pathEnd == -1)`?",
+      "reference": "This is the `GET % HTTP/1.1` scenario right?"
+    },
+    {
+      "prediction": "This should be `use Backend\\Facades\\BackendAuth as Auth;`",
+      "reference": "No need to use the fully qualified path the BackendAuth facade, just `use BackendAuth` is fine."
+    },
+    {
+      "prediction": "This should be `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 think we can remove this line.",
+      "reference": "Please give a more descriptive name and add a command to explain what it does."
+    },
+    {
+      "prediction": "I don't think we need to keep the old key around. We can just remove it.",
+      "reference": "we can never get rid of this..."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add the buckets. I think it should be in the `NewHistogramOpts` call above.",
+      "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 --sanitize-tchannel myservice.thrift`",
+      "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 don't think this is correct. `MaxFeatureIdx()` returns the largest feature index, not the number of features. The number of features is `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 don't think we need to store the `DbRound` in the error. We can just return `fmt.Sprintf(\"round %d before earliest stored db entry %d\", e.Round, e.DbRound)`.",
+      "reference": "rename Round -> requestedRound and DbRound -> dbRound. we don't need to export the fields here, only the error struct."
+    },
+    {
+      "prediction": "I think this should be `assert.Equal(t, getClusterRegion(cd), cd.Labels[hivev1.HiveClusterRegionLabel], \"incorrect cluster region label\")` instead of `assert.Equal(t, getClusterRegion(cd), cd.Labels[hivev1.HiveClusterRegionLabel], \"incorrect cluster region label\")`",
+      "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": "nit: remove empty line",
+      "reference": "TODO: update unit tests to exercise base64-inline path"
+    },
+    {
+      "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 a `boolean` and a method that returns a `boolean` that takes a `boolean` as a parameter.",
+      "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 this should be `w.every == 0` instead of `w.every.IsZero()`.",
+      "reference": "We could utilize the new isZero method in this if-statement, right?"
+    },
+    {
+      "prediction": "I think this should be `FilesWithHolesDataVer` by default.",
+      "reference": "should probably have config have a `t` or a `logger.NewTestLogger(t)`, and return the latter"
+    },
+    {
+      "prediction": "nit: I think it would be clearer to do `for root := range inputPlan.Roots { nodeStack = append(nodeStack, root) }` instead of `nodeStack := make([]PlanNode, 0, len(inputPlan.Roots)) for root := range inputPlan.Roots { nodeStack = append(nodeStack, root) }`",
+      "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 if this is the right thing to do, but I think it makes sense to have the GAPIC config take precedence over the proto annotations.",
+      "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 be `default:\"8080\" description:\"The port to run the server on\"`",
+      "reference": "not related to this change?"
+    },
+    {
+      "prediction": "I think this should be `ANDROID_NDK_HOME || ANDROID_NDK` so that if `ANDROID_NDK_HOME` is set but `ANDROID_NDK` is not, we still use `ANDROID_NDK_HOME`.",
+      "reference": "Why is this needed?"
+    },
+    {
+      "prediction": "I think this should be `<= 0` instead of `== 0`",
+      "reference": "This one could be replaced with a `std::max`, no?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to return `nil` if we can't find the extension id, rather than raising an exception.",
+      "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 right way to do it. If the scope is a checkbox or a switch, then the value should always be true or false. If the scope is not a checkbox or a switch, then the value should always be a string.",
+      "reference": "Please use strict type comparisons (`===`), not loose type comparisons."
+    },
+    {
+      "prediction": "I don't think we need to check for `getToBlock().isLatest()` here. If `getFromBlock().isLatest()` is true, then `getToBlock().isLatest()` must be true as well.",
+      "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 unused includes.",
+      "reference": "Is that necessary?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [121/120]",
+      "reference": "yess so much easier to read in order like this!"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this rule. I think it would be better to add it to `org.apache.lucene.util.TestRuleIgnoreTestSuites` so that it is applied to all tests.",
+      "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 we should keep this at 6. 4 is too short.",
+      "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": "This should be `int(cell // self.cellsPerColumn)`",
+      "reference": "Why was this change necessary?"
+    },
+    {
+      "prediction": "I don't think we should change the name of the logger.",
+      "reference": "I kind of did this deliberately, so everything from the distributor appears in the same set of logs."
+    },
+    {
+      "prediction": "I think this should be `pkt[TCP].seq + pkt[TCP].dataofs * 4` to account for the size of the TCP header.",
+      "reference": "We already have a `seq = pkt[TCP].seq` so you can just re-use it."
+    },
+    {
+      "prediction": "I don't think we need to import all of these classes.",
+      "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 don't think you need to rename these.",
+      "reference": "Why are these variables renamed? Is there a conflict with some Windows specific variable?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. If the file exists, it should be included. If it doesn't exist, it shouldn't be included.",
+      "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 it's better to use `string` instead of `int64` for `duration`.",
+      "reference": "`TimeChaos` does not contains a field .spec.action"
+    },
+    {
+      "prediction": "This should be `AbstractArchive` not `RSAbstractArchive`",
+      "reference": "You need to override where possible in rs-api"
+    },
+    {
+      "prediction": "Shouldn't this be an `IDictionary<string, string>` instead of a `Dictionary<string, string>`?",
+      "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 we need this method.",
+      "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 think this should be `case e.Timeout() && !e.Temporary()`.",
+      "reference": "Returning here bypasses all the health check and load balancing features."
+    },
+    {
+      "prediction": "This should be 2020-2022.",
+      "reference": "Should revert this copyright change, when the file isn't changing in any other way."
+    },
+    {
+      "prediction": "Why is this change necessary? I don't think it's related to this PR.",
+      "reference": "Interesting that filepath.Dir(\"thing\") returns \".\" whereas this returns \"\""
+    },
+    {
+      "prediction": "I think this should be `+ELLIPSIS +NORMALIZE_WHITESPACE`",
+      "reference": "Nice, I did not know about that"
+    },
+    {
+      "prediction": "I think this should be `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 don't think we should return the read state height here. The read state height should be the same as the block height.",
+      "reference": "also fetch block hash of this height and return hash"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Why do we need this in this sample app and not the others?"
+    },
+    {
+      "prediction": "I think it would be better to have `addEvent(String name, Map<String, AttributeValue> attributes)` and `addEvent(String name, long timestamp, Map<String, AttributeValue> attributes)`",
+      "reference": "We don't need this additional method."
+    },
+    {
+      "prediction": "I think this should be `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 `SupportedDiskTypes` to be consistent with the rest of the codebase.",
+      "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 think this should be `if cd.Spec.ClusterInstallRef != nil && cd.Spec.ClusterInstallRef.Kind != \"ClusterProvision\"`",
+      "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": "`bbox_feats.shape[0] > 0` -> `bbox_feats.size(0) > 0`",
+      "reference": "\\`bbox_feats.shape[0] > 0\\` requires the number of proposal is not 0."
+    },
+    {
+      "prediction": "nit: I think we can remove the `table().location()` here since it's not used.",
+      "reference": "Same here. Any way to fit on one line?"
+    },
+    {
+      "prediction": "I think this should be `startsWith(\"set \") || startsWith(\"register \")`",
+      "reference": "using folly::to is better ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but it seems to be the only way to get this to work. I'm not sure why `init_profile` needs to be called here, but it seems to be the only way to get this to work.",
+      "reference": "I'm guessing this is unintended?"
+    },
+    {
+      "prediction": "I don't think this is correct. If `startAtOperationTime` is set, we should always use `startAtOperationTime` instead of `resumeAfter` or `startAfter`.",
+      "reference": "This looks a little suspicious to me, can you explain what's going on here?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it works for now.",
+      "reference": "Can you talk about this? I'm not sure I follow why this is necessary."
+    },
+    {
+      "prediction": "Please use `elseif` instead of `else if`",
+      "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 think it would be good to add a `@since` annotation to this method.",
+      "reference": "Can you say what exception will be thrown in that case (or add an `@throws`, below)?"
+    },
+    {
+      "prediction": "I think this should be `2.6` instead of `2.7`.",
+      "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 we should be importing from `nupic.data.generators` anymore. We should be importing from `nupic.data.pattern_machine`.",
+      "reference": "Should we put these tests in a `generators` directory?"
+    },
+    {
+      "prediction": "I don't think this change is correct. `x1 >= 0.5` is not equivalent to `x1 != zero && !std::isnan(x1)`. For example, if `x1 = -1.0`, then `x1 >= 0.5` is false, but `x1 != zero && !std::isnan(x1)` is true.",
+      "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 `['bob', 'system.Authenticated', 'system.Everyone']`",
+      "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": "Why do we need a new mock for dirty block cache?",
+      "reference": "this reads like \"mock database cache\", maybe a clearer name"
+    },
+    {
+      "prediction": "nit: I think we can remove this empty line.",
+      "reference": "Nit: move this with the other CM imports."
+    },
+    {
+      "prediction": "I think it would be better to use `Symfony\\Component\\HttpFoundation\\BinaryFileResponse` instead of `Symfony\\Component\\HttpFoundation\\StreamedResponse`.",
+      "reference": "This change should be mentioned in the CM"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this. I think it should be in the `Tracer` class.",
+      "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 is this change needed?",
+      "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 this should be 1000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "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').get('url'):`",
+      "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 don't think you need the spaces in the regular expression.",
+      "reference": "on my android \".a.m\" and \".p.m\" is used add `||` here to support both variants"
+    },
+    {
+      "prediction": "I think this should be `if (!(image_flags & VK_IMAGE_CREATE_BLOCK_TEXEL_VIEW_COMPATIBLE_BIT_KHR) && !FormatIsMultiplane(image_format)) {`",
+      "reference": "`!FormatIsMultiplane(image_format)` will always be true at this point due to the test on line 4639."
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "reference": "Maybe inline the CachedAccessor code? Is it similar to our lazy property? If yes, can we reconcile the two?"
+    },
+    {
+      "prediction": "I think this should be `AttachSmokeTests attaches the smoke test cases to the API model.`",
+      "reference": "Nit this file has unintended changes."
+    },
+    {
+      "prediction": "I think this should be `!selectedItem.getMedia().fileExists()`",
+      "reference": "Potential NPE? `hasMedia && selectedItem...`"
+    },
+    {
+      "prediction": "I think this should be `zone=None` instead of `zone = None`",
+      "reference": "PEP-8/consistency with the rest of the code."
+    },
+    {
+      "prediction": "I don't think we need this class. We can just use `MongoDbInstrumentMethodAttribute` instead of `MongoDbCommandInstrumentMethodAttribute` and `MongoDbBulkOperationInstrumentMethodAttribute`.",
+      "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": "Why do we need `TxGasPriceList` and `FallbackGasPrice`?",
+      "reference": "Is exposing SpecProvider needed here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it would be better to have this in the `MvxNotifyPropertyChanged` base class.",
+      "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) pname->service = NULL;`",
+      "reference": "Tim, please add { } to this 'if'."
+    },
+    {
+      "prediction": "I think this should be `0.0.0.0` instead of `127.0.0.1`.",
+      "reference": "Suggest moving to `enable.go` since thats only place used."
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it should be left as it was.",
+      "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": "This should be `// LocalCache is used to handle the local cache.`",
+      "reference": "Local cache of what? What's stored in it?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the sort order id is null, then the sort order is unsorted, so the sort order id should be `SortOrder.unsorted().orderId()`. If the sort order id is not null, then the sort order is not unsorted, so the sort order id should not be `SortOrder.unsorted().orderId()`.",
+      "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(os.path.dirname(os.path.abspath(__file__)), 'html_files')` instead of `os.path.join(os.path.dirname(os.path.abspath(__file__)), 'html_files')`",
+      "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'm not sure if this is the right thing to do, but I think it makes sense to include the name of the environment in the error message.",
+      "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 think this should be `!mimxrt1062 && !mimxrt1052`",
+      "reference": "Maybe it would be better to explicitly include devices instead of explicitly excluding devices?"
+    },
+    {
+      "prediction": "I think this should be `auto &atom` to avoid unnecessary copies.",
+      "reference": "This could be `for (auto bond: bonds()){`, right?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the view.",
+      "reference": "Is $config no longer being used? Can we remove this line?"
+    },
+    {
+      "prediction": "I think this should be `char* remove_ext(const char* path, char dot, char sep, size_t* allocated_size) __attribute__((warn_unused_result));`",
+      "reference": "can you revert changes to this file."
+    },
+    {
+      "prediction": "Why do we need to remove the domain prefix here?",
+      "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 `Task<bool> ExecuteOnMainThreadAsync(Func<Task<bool>> action, bool maskExceptions = true);`",
+      "reference": "Please add this to IMvxMainThreadDispatcher as well"
+    },
+    {
+      "prediction": "I think this should be `selected[opt].is_a?(String) || selected[opt].is_a?(Pathname)`",
+      "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-settings__notification`",
+      "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 `3.2.0`",
+      "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 think this should be `time.Time` instead of `string` so that we don't have to convert it back and forth.",
+      "reference": "We have RequestStart which seems redundant a bit to this one, WDYT?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the only way to get this to work.",
+      "reference": "I think we can remove this corresponding export from `preact` now! Double check no other s using though lol"
+    },
+    {
+      "prediction": "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.<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.<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.<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.<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",
+      "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 think this should be `DEFAULT_GITHUB_ENDPOINT = \"https://api.github.com\"`",
+      "reference": "Maybe `DEFAULT_GITHUB_API_ENDPOINT` would be a better name, since we refer to this as an API endpoint elsewhere?"
+    },
+    {
+      "prediction": "I think this should be `com.amazonaws.ecs.capability.task-cpu-mem-limit`",
+      "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 should be `assert.Equal(t, \"bytecount 1\", mockConnection.LastLine)`",
+      "reference": "What is the test case here?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we want to wait for the caches to sync before starting the flow aggregator, but I don't think we want to wait for the caches to sync before starting the informers.",
+      "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 think this should be `optional(512, \"added_rows_count\", Types.LongType.get())`",
+      "reference": "Can we add these up by the data files counts?"
+    },
+    {
+      "prediction": "I'm not sure about this change. `list.props.revealOrder[0] === 't' && list._map.size` is the same as `!list.props.revealOrder || list.props.revealOrder[0] === 't' && list._map.size`.",
+      "reference": "Most of the time assigning won't save bytes unless used 3+ times (var adds 3bytes)"
+    },
+    {
+      "prediction": "I'm not sure if this is correct, but I think the artist should be the title and the title should be the artist.",
+      "reference": "I would rather find the code that inverts the logic and remove that."
+    },
+    {
+      "prediction": "I think we can remove the `else cls._timeout` and just return `cls._timeout`",
+      "reference": "The else doesn't return anything?"
+    },
+    {
+      "prediction": "Why are we ignoring the error here?",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "I'm not sure why this change is necessary. I don't think it's related to this PR.",
+      "reference": "why not simply `paths = search_path_opts.LibrarySearchPaths` ?"
+    },
+    {
+      "prediction": "I don't think this should be translated. It's the name of the item.",
+      "reference": "Are we sure this should be translated by the web client? It was unclear in chat exactly what's getting translated here."
+    },
+    {
+      "prediction": "This is the only change in this file.",
+      "reference": "same oneof issue."
+    },
+    {
+      "prediction": "I think this should be `@plugins.resolve_references(config.project.module_install)`",
+      "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 `if (globalWorkSizeX > UINT32_MAX || globalWorkSizeY > UINT32_MAX || globalWorkSizeZ > UINT32_MAX || localWorkSizeX > UINT32_MAX || localWorkSizeY > UINT32_MAX || localWorkSizeZ > UINT32_MAX)`",
+      "reference": "it would be better to change `kds` into a `std::vector` then we don't need to explicitly free it"
+    },
+    {
+      "prediction": "I don't think we need to get the config every time we send an email. We can get it once at the start of the app and store it in a global variable.",
+      "reference": "Looks like you're also changing logic and not only tests."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think we should be able to remove this line and the test should still pass. I think the problem is that the test is too sensitive to small changes in the actual interval limit. I think we should be able to remove this line and the test should still pass. I think the problem is that the test is too sensitive to small changes in the actual interval limit. I think we should be able to remove this line and the test should still pass. I think the problem is that the test is too sensitive to small changes in the actual interval limit. I think we should be able to remove this line and the test should still pass. I think the problem is that the test is too sensitive to small changes in the actual interval limit. I think we should be able to remove this line and the test should still pass. I think the problem is that the test is too sensitive to small changes in the actual interval limit. I think we should be able to remove this line and the test should still pass. I think the problem is that the test is too sensitive to small changes in the actual interval limit. I think we should be able to remove this line and the test should still pass. I think the problem is that the",
+      "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 way to do it. The `lazy` option should be a boolean, not a string. If you want to use the `data-` prefix, you should add it to the `attr` option.",
+      "reference": "Shouldn't there be some kind of validation somewhere? IMO there should be an exception if `$options['lazy']` is not a boolean."
+    },
+    {
+      "prediction": "Can you remove the empty line here?",
+      "reference": "nit: please move it inside namespace."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it's the only way I could think of to make sure we don't miss any options in the `RUN_AS_OPTIONS` constant.",
+      "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": "This should be a single string, not two.",
+      "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": "Line is too long. [84/80]",
+      "reference": "This was raising intermittent errors, since Lorem ipsum contains the word \"error\""
+    },
+    {
+      "prediction": "I think this should be `assert((p->mp_impl != nullptr) && \"This participant has been previously invalidated\");`",
+      "reference": "Shouldn't we then add `mp_impl = nullptr` to the destructor of `RTPSParticipant`"
+    },
+    {
+      "prediction": "Why do we need to add `()` after `MagicMock`?",
+      "reference": "Are the additional brackets necessary?"
+    },
+    {
+      "prediction": "I think it's better to throw `NoSuchMethodException` instead of `IllegalStateException`.",
+      "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 think it would be good to add a test for the case where there are 0 failures and 1 pending.",
+      "reference": "this one is unrelated right? (Don't mind including it, just making sure I understand)"
+    },
+    {
+      "prediction": "why not use the config in `configs/_base_/schedules/schedule_1x.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 a good example. I think it's better to have an example that shows how to do something useful, rather than something that does nothing.",
+      "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 think this should be removed.",
+      "reference": "Please remove this blank line."
+    },
+    {
+      "prediction": "Why do you need this trait here ?",
+      "reference": "the cartTrait is not used anymore"
+    },
+    {
+      "prediction": "I don't think we need to return anything from this method.",
+      "reference": "Cn you please make the message more human friendly."
+    },
+    {
+      "prediction": "This import is not used.",
+      "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 `#define ENCFAIL ((uint)0)`",
+      "reference": "Isn't ENCFAIL used in codec.c below? Wouldn't this make it no longer compile?"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "this looks very weird"
+    },
+    {
+      "prediction": "I think this should be `@since 7.0.0` and `@deprecated since 7.0.0`",
+      "reference": "So, the culprit was actually this class \"ConsoleLogHandler\", correct? Because it simply wrote to stdout..."
+    },
+    {
+      "prediction": "I think this should be `return LogFactory.CreateConfig(_originalFileName, LogFactory.Configuration);`",
+      "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'm not sure if this is the right place to add the cleanup step.",
+      "reference": "Should this say \"lint, cleanup, destroy, dependency, syntax, create, prepare, converge, idempotence, side_effect, verify, cleanup, destroy\"?"
+    },
+    {
+      "prediction": "I don't think this is correct. The error here is that we failed to send the message, not that we failed to read the response.",
+      "reference": "I am not sure why these log messages are changed."
+    },
+    {
+      "prediction": "I don't think we need to import `reflect` here.",
+      "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 think this should be `DR_MAX_OPTIONS_LENGTH - 1` instead of `DR_MAX_OPTIONS_LENGTH`.",
+      "reference": "See below: let's use the existing convention _tcsnlen."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Why don't we need this any more?"
+    },
+    {
+      "prediction": "I think we should use `resnet101_caffe` as the pretrained model.",
+      "reference": "clean unnecessary comma"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way to ensure that the `_original` property is not enumerable.",
+      "reference": "This is the default value for enumerable."
+    },
+    {
+      "prediction": "`auto` should be `RDKit::MOL_SPTR_VECT::const_iterator`",
+      "reference": "Just above, you have `const auto` for a `RDKit::MOL_SPTR_VECT::const_iterator`, here it is only `auto`. Why is this?"
+    },
+    {
+      "prediction": "`static_cast<Action_ptr>(event)` would be more idiomatic.",
+      "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 don't think this is necessary. `cfg.Get(key).(*cli.StringSlice)` is already a `*cli.StringSlice`, so `cast.ToStringSlice([]string(*cfg.Get(key).(*cli.StringSlice)))` is equivalent to `cast.ToStringSlice([]string(*cfg.Get(key).(*cli.StringSlice)))`.",
+      "reference": "Why to `Get()` value twice, while u have it 2 lines before"
+    },
+    {
+      "prediction": "This should be `System\\Console\\OctoberFresh` not `System\\Console\\OctoberMirror`",
+      "reference": "There is a typo here, this needs to state `october.fresh`"
+    },
+    {
+      "prediction": "I think this should be `Assert.assertEquals(\"Table name must match\", tableDir.toURI().toString(), table.name());`",
+      "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 think this should be `this.workspace_.scrollY - delta`",
+      "reference": "Can you explain why you went from a positive value (metrics.viewTop) to a negative value?"
+    },
+    {
+      "prediction": "I think it would be better to have a single query parameter `withTasks` which can take the values `locked`, `unlocked` and `retriesLeft`. This would make it easier to extend the API in the future.",
+      "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 it would be better to rename this method to `getScriptById(String id)`.",
+      "reference": "Do you plan to remove the `ScriptDTO getScript(List<String> path);` method long-term?"
+    },
+    {
+      "prediction": "I think this should be `if (!this.enabledTypes || !this.enabledTypes.length) {`",
+      "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": "Please remove this line.",
+      "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 `throw new UnsupportedOperationException(\"getTypeNameForTypedResourceName not supported by PHP\");`",
+      "reference": "Why implement this? It duplicates the value in FeatureConfig."
+    },
+    {
+      "prediction": "I think we can just use `SparkMicroBatchStreamScan` for both batch and micro-batch.",
+      "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 don't think this is correct. If `typeConverter.convert(claim)` returns `null` then we should not add the claim to the result.",
+      "reference": "Will you please update these files to use a copyright end date of 2021?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way to get the `checks` and `commons` objects.",
+      "reference": "Adding fixture to the body as we don't have the `runner.tmpl` anymore"
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "nit: just for consistency we can move this import block down. Right above openshift/hive imports"
+    },
+    {
+      "prediction": "I think this should be `about a managed message of the linter`.",
+      "reference": "Perhaps clean up the commit history and don't squash?"
+    },
+    {
+      "prediction": "I'm not sure why this was changed. I don't think it should have been.",
+      "reference": "Just `writeBytes(content.array())` Back to my original, somewhat broken version ;)"
+    },
+    {
+      "prediction": "I think this should be `ctxpropagation` instead of `ctxpropagation`.",
+      "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": "nit: remove extra blank line",
+      "reference": "File is not `goimports`-ed"
+    },
+    {
+      "prediction": "I don't think we need this check anymore.",
+      "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 we need to throw an exception here. We can just return `name.toOriginal()`.",
+      "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 think this should be `assert self.put_json(\"/options\", {\"anticache\": True}).code == 200`",
+      "reference": "We should also test type confusion here (e.g. `{\"anticache\": \"foo\"}`)"
+    },
+    {
+      "prediction": "I think this should be `sh.if '$? -ne 0' do`",
+      "reference": "I'd love to trim the content of `unformatted` too. Any ideas?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "can you remove this unrelated change?"
+    },
+    {
+      "prediction": "I don't think we need to use `conf` here. We can just use `colors.hints.bg` and `colors.hints.fg` directly.",
+      "reference": "Would using `HintLabel` here instead (without `setObjectName`) work?"
+    },
+    {
+      "prediction": "I think this should be `serviceAccountFile string, serviceAccount []byte, dns01Nameservers []string, ambient bool`",
+      "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 if this is the right place to put the sign in link, but I think it makes sense to have it in the same place as the sign up link.",
+      "reference": "Why do we remove the Sign in link?"
+    },
+    {
+      "prediction": "I'm not sure why these were removed, but I think they should be added back in.",
+      "reference": "holy crap, were we really missing a 'break' statement here?"
+    },
+    {
+      "prediction": "`debugForcedLogMaker` -> `debugForceEnableLogMaker`",
+      "reference": "This doesn't seem to be used anywhere, probably doesn't need to be a new interface."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better than failing the whole Statfs() call.",
+      "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": "I'm not sure about the `:class => \"geolink #{link_class}\"` here. I think it should be `:class => \"geolink #{link_class} dropdown-item\"`",
+      "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 think this should be `@source = step_body['upload'] || step_body['source'] || step_body['source']`",
+      "reference": "Does this fail validation if you now use the `upload` key instead of `source`?"
+    },
+    {
+      "prediction": "I don't think we need to pass NULL as the first argument.",
+      "reference": "Can we use GetCliGlobals(false)->exCollHeap() here instead"
+    },
+    {
+      "prediction": "Why do we need to wait until all elected?",
+      "reference": "I suggest you move the waitUntilAllElected into init::KV() method."
+    },
+    {
+      "prediction": "I think these should be `internal` instead of `public`.",
+      "reference": "no need of pub;lic."
+    },
+    {
+      "prediction": "I don't think we need to keep this around. We can just remove it after the upgrade is done.",
+      "reference": "Do you think we can name this variable in a less ...negative.. language ? maybe `ProposalTieBreakerEqualizer`"
+    },
+    {
+      "prediction": "I think we can remove the `inputSchema` parameter from this constructor. It's only used to create the `accessors` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the `accessor` array, which is only used to create the",
+      "reference": "Can you add back these newlines? We try to avoid non-functional changes like this because they can cause commit conflicts."
+    },
+    {
+      "prediction": "I think this should be `searches = searches.filter(s => s.title !== action.search.title);`",
+      "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));`",
+      "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": "Why is this removed?",
+      "reference": "Why is this deleted too?"
+    },
+    {
+      "prediction": "I don't think we need to implement `IDisposable` here.",
+      "reference": "Given that IApiResponse is already made disposable, this is not necessary."
+    },
+    {
+      "prediction": "Please add braces to this if-statement.",
+      "reference": "client.getSelectedSpellName() can be used for this"
+    },
+    {
+      "prediction": "I think this should be `forseti_findings_{}.json`.",
+      "reference": "Please alpha sort these filenames."
+    },
+    {
+      "prediction": "I think we should keep the check for negative amount and gas price.",
+      "reference": "is this already called by actPool.validate()?"
+    },
+    {
+      "prediction": "I think this should be `if (product->getRingInfo()->numBondRings(pBond->getIdx()) > 0) { pBond->setStereo(Bond::BondStereo::STEREONONE); } else { pBond->setStereo(Bond::BondStereo::STEREOANY); }`",
+      "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 `svc-type` instead of `app-type`",
+      "reference": "Maybe we should add flag aliases for service as well."
+    },
+    {
+      "prediction": "I think this should be `ElectionProof *crypto.ElectionProof` so that it can be `nil` in the genesis block.",
+      "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": "`Blocklistener` -> `BlockListener`",
+      "reference": "can you move the [] into pubSub? so Blockchain just contains a pubSub, not []"
+    },
+    {
+      "prediction": "I think this should default to `true` and we should throw an error if it is set to `false`.",
+      "reference": "I don't think 100 is a safe default here, since later we check that the value is between 50-95."
+    },
+    {
+      "prediction": "I think this should be `ThreadAgnostic = 4 | ThreadSafe`",
+      "reference": "I would be nice if we could describe this without the word \"agnostic\""
+    },
+    {
+      "prediction": "I think this should be `if isSparse >= 0:`",
+      "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 `'<rootDir>/.storybook/?(*.)test.(js|jsx)'` so that we can test both `.js` and `.jsx` files.",
+      "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 we should keep the original message.",
+      "reference": "Most of the following test changes can be avoided if need be."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we can remove the `CellStorage` from the `MultiLevelPartition` class. The `MultiLevelPartition` should only contain the `MultiLevelPartitionData` and the `EdgeBasedGraph`. The `CellStorage` is only used in the `Partitioner::Run` method, so I think it makes sense to move it there. What do you think?",
+      "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, but I think we should return an error from `Kill` if we can't stop one of the services.",
+      "reference": "You're not really using named result variable - you can just leave `error`."
+    },
+    {
+      "prediction": "I think we can remove the `schema` parameter from this method.",
+      "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 this line ?",
+      "reference": "this namespace is never used"
+    },
+    {
+      "prediction": "I think this should be \"at-least-once semantics\", not \"at-most-once semantics\".",
+      "reference": "Well, if you call `Nack`, it _will_ be redelivered. So maybe reword."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it's the only way I could get the video to work.",
+      "reference": "Why did we make this a static string?"
+    },
+    {
+      "prediction": "I think this should be `google.cloud.bigquery.BigQueryDataset`",
+      "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 don't think we need to pass `hash.ZeroHash256` here.",
+      "reference": "`stateDB := NewStateDBAdapter(sm, 0, hash.ZeroHash256, NotFixTopicCopyBugOption())`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to fix the issue for me.",
+      "reference": "There are unneeded parens here, but I'll fix it up when merging."
+    },
+    {
+      "prediction": "I think it would be better to use `h2o_mruby_eval_file` instead of `h2o_mruby_eval_expr` here.",
+      "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 should be `flux_log (h, LOG_INFO, \"cron-%ju: task timeout at %.2fs. Killing\", e->id, e->timeout);`",
+      "reference": "Commit message: \"remove\" is misspelled."
+    },
+    {
+      "prediction": "I think this should be `value.slice(0, value.size() - 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 think this should be `config.apply_derived_metadata_to(@metadata)` instead of `config.apply_derived_metadata(@metadata)`.",
+      "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 don't think we need this constant anymore.",
+      "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 think this should be `playbackManager.currentTime(options.player) * 1000`",
+      "reference": "Converting playbackManager to operate in ms (preferred solution, imo) would allow us to completely drop this function."
+    },
+    {
+      "prediction": "The max edge number returnred searching vertex -> The max edge number returned searching vertex",
+      "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 an error, not a warning.",
+      "reference": "Probably missing return here as well"
+    },
+    {
+      "prediction": "I think this should be `PullInterval` instead of `PollInterval`.",
+      "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 don't think we need localization here.",
+      "reference": "Now you don't need the localization module here then :)"
+    },
+    {
+      "prediction": "I think this should be `return ROWS_CANNOT_MATCH;` instead of `return ROWS_MIGHT_MATCH;`",
+      "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 think we should raise an error here.",
+      "reference": "I think we still need this exception if doesn't have LIGHTGBM_BUILD_DOC"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it is better than what we had before.",
+      "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 think this should be a `Map<String, CamundaIntegration>` instead of a `Map<String, Object>`.",
+      "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'm not sure I understand the logic here. Why do we need to check if the task was pruned before checking if it's necessary?",
+      "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": "nit: extra blank line",
+      "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 this should be `finalOptions.readPreference = options.readPreference || ReadPreference.primary;`",
+      "reference": "Do we normally force ReadPreference primary on other write operations? Or is it just that we normally ignore it?"
+    },
+    {
+      "prediction": "I think this should be `return Expressions.apply(transform, Expressions.column(sourceName));` instead of `return null;`",
+      "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 we can remove the `if (prevBlockHeader.isPresent())` check and just call `resolveValidatorProvider(prevBlockNumber).getValidatorsForBlock(prevBlockHeader.get())`",
+      "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": "I don't think you need to do this. You can just use `sizeof(longopts) / sizeof(longopts[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`",
+      "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 if this is the best way to do this, but it's the only way I could get the SVGs to render in the storybook.",
+      "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 don't think we need to set these in the config file, they can be set on the command line.",
+      "reference": "Would it be possible to reuse RcJobExpireDuration and RcJobExpireInterval and auto delete them when job is deleted?"
+    },
+    {
+      "prediction": "I think this should be `from jinja2 import Environment, PackageLoader`",
+      "reference": "Python wants **future** imports to come first."
+    },
+    {
+      "prediction": "I don't think we need to call options_detach() from detach_on_permanent_stack(). It is already called from detach() which is called from detach_on_permanent_stack(). I think we can remove the options_detach() call from detach_on_permanent_stack().",
+      "reference": "Better to call this before EXITING_DR which is supposed to be last."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "Does this work at runtime?"
+    },
+    {
+      "prediction": "I think this should be 10, not 8. The default is 3 retries, which is 2 seconds. 10 retries would be 18 seconds.",
+      "reference": "Hmm why do we need to make this change? Like which one requires this change?"
+    },
+    {
+      "prediction": "Why is this no longer needed?",
+      "reference": "why remove this and other piece?"
+    },
+    {
+      "prediction": "I think this should be `assert.isTrue`",
+      "reference": "Yikes... That was probably me."
+    },
+    {
+      "prediction": "I think this should be \"Note attachments\" (without \"...\")",
+      "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": "I don't think you need to include all of these here.",
+      "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": "nit: can you add a blank line here?",
+      "reference": "`NewUpgradeTaskJob` -> `NewUpgradeResource`"
+    },
+    {
+      "prediction": "This should be `glibtop_get_mem(&mem, 0);`",
+      "reference": "Is that the memory usage of the system or the process / awesome?"
+    },
+    {
+      "prediction": "I think this should be `callback(Realm, string)|callback(Realm, Schema)`",
+      "reference": "\"change\" *and* \"schema\" events (A bunch of these; not marking them all)."
+    },
+    {
+      "prediction": "I don't think this is needed. The sequence number of the new snapshot will be the same as the sequence number of the snapshot that will be created.",
+      "reference": "I think there is probably a shorter, more descriptive name for this. Something like `commitAtSequenceNumber`?"
+    },
+    {
+      "prediction": "`allow_nil: false` is the default, so you don't need to specify it.",
+      "reference": "The `url_name` should definitely be more than just 1 character long at the shortest. How about 3 or 4 characters?"
+    },
+    {
+      "prediction": "This file is not used anymore, so it can be removed.",
+      "reference": "Please revert this change"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "reference": "Nice. I hadn't even heard of presenters in Rails before (learn something new every day) :)"
+    },
+    {
+      "prediction": "`@var array<string, DownloadStrategyInterface>`",
+      "reference": "`string[]` ? not sure, can you verify this @core23 ? Thank you"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to clear the timeout. I think it would be better to clear the timeout in the `onPlayPauseClick` function.",
+      "reference": "are we aware that this breaks older browsers like WebOS? :) they don't handle `let`"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think we should be able to use `touch` here, but I'm not 100% sure why it doesn't work.",
+      "reference": "This isn't really a Kinto record but a Memcached record. I could kind of go either way on this."
+    },
+    {
+      "prediction": "I think this should be moved to the `Account` class, so that it can be called from the `AccountManager` as well.",
+      "reference": "Kick off the download after login."
+    },
+    {
+      "prediction": "I think this should be `array<int, 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 == seat->pointer_state.grab_serial || serial == seat->touch_state.grab_serial || serial == seat->keyboard_state.grab_serial;`",
+      "reference": "TODO: fix this, probably in another PR"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Is there any case that tab is None?"
+    },
+    {
+      "prediction": "Why do we need to check the error here?",
+      "reference": "pls use `err` instead of `e`"
+    },
+    {
+      "prediction": "I don't think we need to check for `navigator.getGamepads` here. The `gamepadconnected` event will not fire if `getGamepads` is not available.",
+      "reference": "`attachGamepadScript` above can also be placed inside."
+    },
+    {
+      "prediction": "I think this should be `Signer` instead of `signer`.",
+      "reference": "when we use RegisterWriteCommand,this func should be exported,so we can use this to get signer at the outside action package"
+    },
+    {
+      "prediction": "Why is this line removed?",
+      "reference": "Dead code - we should remove it or use `log.Trace` for lower logging level. Same applies for change at line 138."
+    },
+    {
+      "prediction": "I think this should be `catch (const boost::bad_lexical_cast &e)`",
+      "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": "`Amazon Simple Storage Solution (Amazon S3)` -> `Amazon Simple Storage Service (Amazon S3)`",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I think this should be `defined('CACHE_ENABLED') || define('CACHE_ENABLED', APPLICATION_ENV != 'development' && !defined('VUFIND_PHPUNIT_RUNNING'));`",
+      "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 this should be `TargetNext` instead of `TargetDesired`.",
+      "reference": "has this changed? are we testing \"all possible situations\" now?"
+    },
+    {
+      "prediction": "I don't think we need the period at the end of the description.",
+      "reference": "ZK is properly spelled 'ZooKeeper' (capital 'K')."
+    },
+    {
+      "prediction": "I think this should be `~/.config/gcloud/application_default_credentials.json`, not `~/.config/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": "I'm not sure if this is the best way to do this. I think it would be better to have a `isColumnSelection` method in the `Selection` class and use it here.",
+      "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": "nit: remove the 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 should be `defer os.Setenv(noninteractive, os.Getenv(noninteractive))`",
+      "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 don't think this is correct. `shlex.split` should always be called with `posix=False` on Windows.",
+      "reference": "`sys.platform` could be equal to `darwin` which is posix. Use `not startswith('win')`?"
+    },
+    {
+      "prediction": "nit: I think it would be nice to have a `NewOverrideLookup` function that takes a `map[string]string` and returns an `OverrideLookup`. That way we don't have to expose the `NewOverrideLookup` function in the `overrides` package.",
+      "reference": "let's move this into `slack_helper.go`"
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "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 `> 360.`",
+      "reference": "this is the reason we might be seeing 360 as discrete bearing"
+    },
+    {
+      "prediction": "I don't think we should use `%%(prog)s` here.",
+      "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": "rowPosition is not used in this class. Can we remove it?",
+      "reference": "Is this needed? I don't see any uses."
+    },
+    {
+      "prediction": "nit: I think we can remove the `c.T.Helper()` calls here and in the `DeletePubSubOrFail` function.",
+      "reference": "Shouldn't this be V1beta1?"
+    },
+    {
+      "prediction": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.",
+      "reference": "I think there's an `approval_for`"
+    },
+    {
+      "prediction": "I don't think this should be part of the public API.",
+      "reference": "Here we should use <code>List</code> in generic"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to use the target compatibility instead of the toolchain version. The toolchain version is the version of the JDK used to compile the code, while the target compatibility is the version of the JVM that the code is intended to run on. If the target compatibility is lower than the toolchain version, then the code will be compiled with the toolchain version, but it will be compatible with the target compatibility. If the target compatibility is higher than the toolchain version, then the code will be compiled with the toolchain version, but it will not be compatible with the target compatibility. If the target compatibility is equal to the toolchain version, then the code will be compiled with the toolchain version, and it will be compatible with the target compatibility. So, I think it's better to use the target compatibility here, because it's the version of the JVM that the code is intended to run on, and we want to make sure that the code is compatible with that version of the JVM.",
+      "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 it should be `string $hash` instead of `string $name`",
+      "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 this should be `__u8 nat_level1_drop = 0;`",
+      "reference": "We have `stdbool` imported, might as well use that for clarity."
+    },
+    {
+      "prediction": "I think this file should be deleted.",
+      "reference": "File is empty, consider removing?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get this test to pass. The `writeConcernMax()` method returns an object that has a `writeConcern` property, which is what the `execute()` method expects.",
+      "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": "Metrics/LineLength: Line is too long. [101/80]",
+      "reference": "How about passing this through the `config`?"
+    },
+    {
+      "prediction": "I think this should be a `pytest.mark.skipif` instead of a 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": "Why do we need this?",
+      "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 an error, not a debug message.",
+      "reference": "would you please rename the message prefix to: [in_syslog] ...\""
+    },
+    {
+      "prediction": "nit: `molExt` is a pointer to a pointer to a `RWMol`. I think it would be clearer to call it `queryMol` or `queryMolPtr` or `queryMolRwPtr` or `queryMolRwPtrPtr` or `queryMolRwPtrPtrPtr` or `queryMolRwPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtrPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtrPtrPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtr` or `queryMolRwPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtrPtr",
+      "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 this should be renamed to `pendingCount` to be consistent with the rest of the codebase.",
+      "reference": "the name confusing, please rename to `pendingCountNoLock` or similar"
+    },
+    {
+      "prediction": "Why do we need to set the signature here?",
+      "reference": "line is 175 characters"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to call `loadReactAppOnceIfReady`, but it seems to be the only place where we have access to the `client` and the `reactActivityDelegate` at the same time.",
+      "reference": "@ivanbogdanov Does this fix the first time load gray screen issue that @wmathurin noticed?"
+    },
+    {
+      "prediction": "I don't think we need to remove this warning. It's just a reminder for us to remove it after 2.3.0 release.",
+      "reference": "Is `2.3.0` version OK?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it should be `driver.find_element(:id, 'iframe_page_heading').should be_kind_of(WebDriver::Element)` instead of `driver.find_element(:id, 'iframe_page_heading').should be_kind_of(Selenium::WebDriver::Element)`",
+      "reference": "I tested it only in Firefox (`./go //rb:firefox-test`)"
+    },
+    {
+      "prediction": "I don't think we need to override the methods on the new TestClient.",
+      "reference": "separate third party and builtin"
+    },
+    {
+      "prediction": "I think this should be `!isHybrid() || (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 if this is the right thing to do, but I think it's the right thing to do.",
+      "reference": "isn't this going to return a timeout error if the context is canceled?"
+    },
+    {
+      "prediction": "Shouldn't this be `if (argNameVal.length == 0)`?",
+      "reference": "`argNameVal.length` will still be !=2, so line 22 will give true and then continue on line 24"
+    },
+    {
+      "prediction": "I think this should be `$chars[$i + 1] === '\\\\' || preg_match('/\\w/', $char)`",
+      "reference": "It was `$chars[$i]` I believe."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `key.to_s.upcase` to be consistent with the rest of the code.",
+      "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 don't think this is necessary. `variableOptions` should always be defined.",
+      "reference": "Thanks for fixing this! I probably didn't test the playground when making changes here for cloud variables."
+    },
+    {
+      "prediction": "Can you add a check for `nameMapping` as well?",
+      "reference": "I will also check if `nameMapping` needs a precondition null check."
+    },
+    {
+      "prediction": "nit: remove the extra space after `String.format(`",
+      "reference": "Since RuntimeIOException is deprecated and you are touching this code, why not replace it?"
+    },
+    {
+      "prediction": "I think it would be good to have a test for each of the `getScalarMetricWithTimeOfDayAnomalyParams`, `getVectorMetricWithTimeOfDayAnomalyParams` and `getVectorMetricWithTimeOfDayAnomalyParams` methods.",
+      "reference": "You should also have a test that calls it with tm_cpp and make sure it returns the correct JSON file."
+    },
+    {
+      "prediction": "Why remove this?",
+      "reference": "One more minor change, the string `ForAdditionalLiveTvOptions` is probably also unused."
+    },
+    {
+      "prediction": "I think this should be an error.",
+      "reference": "How to distinguish between warning and error?"
+    },
+    {
+      "prediction": "This should be `if UIAHandler.handler.isUIAAvailable:`",
+      "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.enabled\", \"true\")`",
+      "reference": "can we make `replace-partitions` property a static variable in `SnaphotSummary.java`?"
+    },
+    {
+      "prediction": "I don't think we need to list all the bridges here.",
+      "reference": "Maybe call it `WireBridge`?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to send the welcome email. I think it would be better to send the email in the `_activate_user` function.",
+      "reference": "Should really be `one()`, not `one_or_none()`"
+    },
+    {
+      "prediction": "I think this should be `out.NodeRegistration.KubeletExtraArgs[\"node-labels\"] = fmt.Sprintf(\"%s,%s\", nodeRole, labels)`",
+      "reference": "Should we use strings.Split and strings.Join instead of manual concatenation?"
+    },
+    {
+      "prediction": "I think this should be `self.dagster_type if self.dagster_type is not None else NoValueSentinel`",
+      "reference": "rough that this needs to exist, but it is what it is"
+    },
+    {
+      "prediction": "I'm not sure why this test is failing. It seems like it should be passing.",
+      "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 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": "It is the maximum value that the configured emulator can support."
+    },
+    {
+      "prediction": "I don't think this is the right thing to do. If there is an autosave session, it should be loaded. If there is no autosave session, it should not be loaded. If there is no autosave session and no other session, it should not be loaded. If there is no autosave session and there is another session, it should be loaded.",
+      "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 think this should be `AccountData` instead of `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'm not sure this is the right thing to do, but I'm not sure what else to do either. I don't think we can wait on the mutex here, because that would cause a deadlock. I don't think we can wait on the condition variable either, because that would cause a race condition. I don't think we can wait on the condition variable, because that would cause a race condition. I don't think we can wait on the condition variable, because that would cause a race condition. I don't think we can wait on the condition variable, because that would cause a race condition. I don't think we can wait on the condition variable, because that would cause a race condition. I don't think we can wait on the condition variable, because that would cause a race condition. I don't think we can wait on the condition variable, because that would cause a race condition. I don't think we can wait on the condition variable, because that would cause a race condition. I don't think we can wait on the condition variable, because that would cause a race condition. I don't think we can wait on the condition variable, because that would cause a race condition. I don't think we can wait on the condition variable, because",
+      "reference": "I think this mutex is only used here. Remove it."
+    },
+    {
+      "prediction": "I don't think we need to do this here. We can do this in `NewMutableStateFromDB` and `NewMutableStateFromData` instead.",
+      "reference": "so this is now init to 1?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 should be `foo` instead of `foo-install`",
+      "reference": "Had to bring this up, will abandon this style of testing based on Joel's work soon."
+    },
+    {
+      "prediction": "I don't think we need to log the auth header here.",
+      "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 think it would be better to use `strings.SplitN` instead of `regexp.MustCompile` here.",
+      "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'm not sure if this is the right thing to do, but I think we should be able to get rid of all the `DeepCopy` calls in this file.",
+      "reference": "Can you please remove `codec` from the parameters too? I think it's unused now."
+    },
+    {
+      "prediction": "I don't understand why this is needed. Can you explain?",
+      "reference": "mocks needed this change so that it returns the \"right\" thing"
+    },
+    {
+      "prediction": "This should be `@Getter(AccessLevel.PACKAGE)` and `@Setter(AccessLevel.PACKAGE)`",
+      "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 seems to work for now. I'm not sure if there's a better way to get the wistia id from the `video.clip.wistia_id`.",
+      "reference": ":+1: to inlining this."
+    },
+    {
+      "prediction": "I don't think this is correct. `stream_get_meta_data` returns an associative array, not an object.",
+      "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 don't think this is needed anymore.",
+      "reference": "I can't recall if there was a fix specific to this in the past?"
+    },
+    {
+      "prediction": "I think this should be `ProjectID` to be consistent with the rest of the codebase.",
+      "reference": "Not sure what the todo is here? Is it to support some kind of defaulting based off of that?"
+    },
+    {
+      "prediction": "I think it would be better to name this `FlushOnConditionOnly` or `FlushOnlyOnCondition`",
+      "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 don't think this is necessary. The `pytest.mark.qt_log_ignore` should be enough.",
+      "reference": "I don't think that works - you can't mark a fixture."
+    },
+    {
+      "prediction": "I think this should be `~/.puppetlabs/etc/bolt/bolt.yaml` and `~/.puppetlabs/bolt.yaml`",
+      "reference": "We should not include this path."
+    },
+    {
+      "prediction": "nit: I think it would be good to add a test for the case where `violation_data` is not a string.",
+      "reference": "Add newline at end of file"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add this column. I think it should be added to the `PodDBAdapter.TABLE_NAME_EPISODES` table instead. What do you think?",
+      "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 don't think this is the right place for this constant. I think it should be in `pkg/operator/manifests/manifests.go`.",
+      "reference": "suggest hive.openshift.io/cluster-machine-management or something to make it more obvious what it is."
+    },
+    {
+      "prediction": "Why do we need this dependency?",
+      "reference": "We explicitly don't want to do that. This was a complaint from users before."
+    },
+    {
+      "prediction": "I don't think this is a good idea to add a new method to the type collection. It's not a type, it's a filter.",
+      "reference": "Hello, The PHPDoc is `string[]` not `int[]`"
+    },
+    {
+      "prediction": "`enterNS` is not used in this function.",
+      "reference": "Same issues with parameters order in `ipset_server.go`"
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "Wow, did it literally mean the retention days is a huge number?"
+    },
+    {
+      "prediction": "This should be `DirExists(opts.project_root.c_str())` instead of `DirExists(opts.project_root)`.",
+      "reference": "you probably didn't intend to touch those files in `scripts/`"
+    },
+    {
+      "prediction": "`tpfp_func` -> `tpfp_fn`",
+      "reference": "Similar to `collate_fn`, we may rename it to `tpfp_fn`."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Awesome! We may also use negative rule `never` to fix new lines after `return`: `{ blankLine: \"never\", prev: \"return\", next: \"*\" }`"
+    },
+    {
+      "prediction": "I don't think we need this function. We can just call `config.BlockServer().(journalBlockServer).jServer` directly.",
+      "reference": "Why can't this be a function on the `Config` interface like all the others?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's better than the current behavior.",
+      "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": "nit: I think we can remove the `allowIncompatibleChanges` call here, as it is already called in the `addColumn` method.",
+      "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::get('backend::dashboard.status.widget_title_default')`",
+      "reference": "Early translation , should be logic-less"
+    },
+    {
+      "prediction": "nit: I think it would be better to call this `NullInterceptor` instead of `nullInterceptor` to be consistent with the rest of the codebase.",
+      "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": "Metrics/LineLength: Line is too long. [123/120]",
+      "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 it would be better to use `assertThat(gen).isIn(1, 2, 3);` instead of `assertThat(gen.apply(RANDOM)).isIn(1, 2, 3);`",
+      "reference": "Any suggestions on how I would even _approach_ writing tests for arbitrary values...."
+    },
+    {
+      "prediction": "I don't think we should make these constants private. They are used in the `Theme` class and in the `ThemeManager` class.",
+      "reference": "Scope declarations for class constants was not added until 7.1, this will not be accepted."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the client is already in the `deadClientMap_` then it will be removed from the `activeClientMap_` and then added to the `deadClientMap_` again. I don't think this is a problem, but I'm not 100% sure.",
+      "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 don't think this is correct. If the key is missing, you should return `nil, nil`. If the collection cannot generate a fresh key, you should return `nil, err`.",
+      "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": "I think this should be \"The system was in the wrong state, and the operation could not be completed.\"",
+      "reference": "Doesn't `gcerr_string.go` need to be updated for this?"
+    },
+    {
+      "prediction": "I think this should be `stringToRuneSlice`",
+      "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 don't think this should be called `load_prepared_history`. It's not loading anything, it's just clearing the prepared history.",
+      "reference": "You can simply do `if tab.history_prepared:` here as empty lists are falsey."
+    },
+    {
+      "prediction": "I think this should be `github.com/jetstack/cert-manager/cmd/ctl/pkg/create/certificatesigningrequest/install`",
+      "reference": "Remove this whitespace"
+    },
+    {
+      "prediction": "I think we should keep the original image for visualization.",
+      "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 don't think we should be clamping the value here. If the user wants to set the offset to 10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "Why is this bounded between -30 and 30?"
+    },
+    {
+      "prediction": "I think this should be `EXCLUDE_HIP_PLATFORM nvcc` instead of `EXCLUDE_HIP_PLATFORM all`",
+      "reference": "I think even you can exclude to run it on nvcc"
+    },
+    {
+      "prediction": "Is this change related to this PR?",
+      "reference": "G104: Errors unhandled. (from `gosec`)"
+    },
+    {
+      "prediction": "I think this should be `response['failures'][0]['reason']`",
+      "reference": "will `failure` always include `arn` and `reason` in its dictionary? If so, :+1:"
+    },
+    {
+      "prediction": "I think this should be `action: action || :run_task` or something like that. I don't think we want to default to `:run_task` if the user passes in `nil` for `action`.",
+      "reference": "Should this be optional, or default to 'action' as well?"
+    },
+    {
+      "prediction": "I think it's better to use `chaos-mesh` instead of `github.com/chaos-mesh/chaos-mesh`",
+      "reference": "how about formating this import?"
+    },
+    {
+      "prediction": "I think this should be `isCaret is False` instead of `not isCaret` since `isCaret` is a boolean.",
+      "reference": "Comma police. :) Also, the docstring needs updating."
+    },
+    {
+      "prediction": "I think this should be `volume.CloudMigrateDriver` instead of `volume.CloudMigrateDriver`.",
+      "reference": "Implement an in-memory implementation of this in the fake driver."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Same question as the other view spec regarding not actually saving records."
+    },
+    {
+      "prediction": "I think this should be `InternalLogger.Warn(ex, \"Auto loading assembly file: {0} failed! Skipping this file.\", extensionDll);`",
+      "reference": "Rethrow for `MustBeRethrown()`-exceptions?"
+    },
+    {
+      "prediction": "`device = 'cpu' if not torch.cuda.is_available() else torch.cuda.current_device()`",
+      "reference": "May clean the unnecessary modification."
+    },
+    {
+      "prediction": "I don't think we need to check for flux-security here. If it's not found, then flux_sec_incdir will be empty, and we'll get an error later when we try to use it.",
+      "reference": "Should this line set the value to `x` since that is checked below?"
+    },
+    {
+      "prediction": "I don't think we need this method. We can just call `copy()` and then set the stats to null.",
+      "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 this should be `if (idx == -1 || idx == 0 || idx == ann.length() - 1) {`",
+      "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 don't think we need to `encodeURIComponent` the `recording_start` here.",
+      "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'm not sure why this was removed, but I think it should be restored.",
+      "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": "Please remove unused imports.",
+      "reference": "nit: strip away those imports? seems not used in the new code."
+    },
+    {
+      "prediction": "I think it's better to use `github.com/mysteriumnetwork/payments/contracts/generated` instead of `github.com/ethereum/go-ethereum/accounts/abi/bind`",
+      "reference": "redundant whitespace :octocat:"
+    },
+    {
+      "prediction": "I think this should be `self.assertIsInstance(api_response, dict)`",
+      "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": "nitpick: I think it would be better to move `h2o_mem_release_shared(ret);` to the end of the function.",
+      "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 don't think we need to make this abstract.",
+      "reference": "Wouldn't it make sense to already have title, owner etc in this class?"
+    },
+    {
+      "prediction": "Line is too long. [84/80]",
+      "reference": "New generators should have version `next`"
+    },
+    {
+      "prediction": "I think this should be `metricsConfiguration.getPushProtocol().name()`",
+      "reference": "Wrap these two lines inside an `if (node.isMetricsEnabled() || metricsConfiguration.isPushEnabled()) { ... }`"
+    },
+    {
+      "prediction": "I don't think this is correct. If `end_level == self.num_ins`, then `self.backbone_end_level` should be `self.num_ins - 1`. Otherwise, `self.backbone_end_level` should be `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": "Why do we need to divide by 2 here?",
+      "reference": "No CLR overhead?"
+    },
+    {
+      "prediction": "nit: you can use `t.Fatal` instead of `t.Fatalf`",
+      "reference": "Consider faking the TestClient's GetProject/GetZone/GetMachine methods."
+    },
+    {
+      "prediction": "I'm not sure I understand what you mean by \"the concrete peer type is private and intrinsic to the HTTP transport\". Can you elaborate?",
+      "reference": "Returning a private type as part of the public API? Isn't this impossible for users to implement?"
+    },
+    {
+      "prediction": "I think this should be `unsigned int rankNum = 0;`",
+      "reference": "prefer common_properties::_canonicalRankingNumber if possible, should be quicker in general."
+    },
+    {
+      "prediction": "I think it's better to set the default value of `loss_normalizer` to `None`, and set the default value of `loss_normalizer_momentum` to `0.9`.",
+      "reference": "The meaning of `loss_normalizer` and `loss_normalizer_momentum` should be reflected in docstring."
+    },
+    {
+      "prediction": "`ioctlsocket` returns `0` on success, so this should be `if (ioctlsocket(fds[1], FIONBIO, &nonblock) != 0) {`",
+      "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": "I don't think we should use `import java.util.*;`",
+      "reference": "There are no changes in this file. Please reset."
+    },
+    {
+      "prediction": "I think it would be clearer to say \"Given the normal distribution specified by the mean and standard deviation in distributionParams, return the probability of getting samples at least as far from the mean as x. For values above the mean, this is the probability of getting samples > x and for values below the mean, the probability of getting samples < x. This is the Q-function: the tail probability of the normal distribution.\"",
+      "reference": "Should we rename this to tailProbability?"
+    },
+    {
+      "prediction": "I think you mean `if (index == REG_NULL && opnd_get_disp(memref) != 0) {`",
+      "reference": "Add the issue number too i#4400"
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "reference": "Why make this modifiable?"
+    },
+    {
+      "prediction": "0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "Why this file is changed?"
+    },
+    {
+      "prediction": "I don't think we need to add a new capability for this. The `history` capability should be enough.",
+      "reference": "I'm not sure that we need this"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "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'm not sure if this is the right place to put this function, but I think it should be in the `aws` package, not the `secretsmanager` package.",
+      "reference": "Does this have retries? Is there a possibility that customer would hit throttle errors here?"
+    },
+    {
+      "prediction": "This should be `PLATFORM_IS_BSD` instead of `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 if this is the right place to do this, but I think we should also remove the `network-audit-results.json` file if it exists.",
+      "reference": "Deleting the files before starting the audit helps avoid stale results if the new file is not created."
+    },
+    {
+      "prediction": "nit: I think we can remove the `if err != nil` check here as we are already returning the error from `CreateStoragePool` function.",
+      "reference": "Remove this blank line"
+    },
+    {
+      "prediction": "This import is not used.",
+      "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 don't think this is the right place to set the background color.",
+      "reference": "why are these colors needed to be specified here? when are they used?"
+    },
+    {
+      "prediction": "I think this should be `Cannot manage snapshots from a metadata table`",
+      "reference": "Nit: no need to change this method."
+    },
+    {
+      "prediction": "nit: I think we can remove the empty line here.",
+      "reference": "Why a blank line?"
+    },
+    {
+      "prediction": "I'm not sure about this change. The `html_safe` is there for a reason. If we remove it, we'll need to make sure that `current_user.org.feedback_email_msg` doesn't contain any HTML.",
+      "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": "I don't think we need to change the API here. The attach case should be handled internally in `dynamorio_app_init()` and `dynamorio_app_start()`.",
+      "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 don't think we need to add this to the `ProtocolMsg` struct. We can just use `ProtocolOverlayControlMsg` directly.",
+      "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 think this should be `if a.NoValidataShapeMethods`",
+      "reference": "This is never set anywhere in the `cli/gen-api` folder. Should it be?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `corev1` import and just use `clientv1` in the rest of the file.",
+      "reference": "All the changes in this commit are mechanical. Produced by the script in the previous commit."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get this to work.",
+      "reference": "What's the difference between these 2 versions?"
+    },
+    {
+      "prediction": "This test should be updated to check for the correct number of violations.",
+      "reference": "@red2k18 Are we sure its correct to only have 1 now?"
+    },
+    {
+      "prediction": "I think this should be `newB->setProp(RDKit::common_properties::_unspecifiedOrder, bondOrder);`",
+      "reference": "@greglandrum interesting, is this related to #3307 by any chance?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be good to have a `force_user_id` method in the `ApplicationController` that returns `ENV['FORCE_USER_ID']` if `Rails.env.development?` or `Rails.env.test?`, and `nil` otherwise. Then we can call `force_user_id` instead of `ENV['FORCE_USER_ID']` in the `find_current_user` method.",
+      "reference": "Minor: how about moving the trailing `if` to an `elsif` above?"
+    },
+    {
+      "prediction": "Shouldn't this be `DEFAULT 0` instead of `DEFAULT -1`?",
+      "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 don't think we need to lock the global account here.",
+      "reference": "Safe to reference c.srv without capturing it first under a client lock?"
+    },
+    {
+      "prediction": "I don't think you need to add the license header to this file.",
+      "reference": "@vincepri blame tells me this was you, any objections to the change?"
+    },
+    {
+      "prediction": "I think we should be able to wait for the server to be ready.",
+      "reference": "@abhinav here is where i need to be able to block/wait until the server is started"
+    },
+    {
+      "prediction": "I think it would be better to say \"the name of the parameter used to specify the jsonp callback, or an empty string, if no such parameter exists.\"",
+      "reference": "It seems to me as an unrelated change. Could you please revert it?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "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 don't think this is correct. If you have `class X extends Object { }` then `getExtendedTypes()` will return an empty list, not a list with `Object` in it.",
+      "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 why this change was made. I don't think it's necessary.",
+      "reference": "For a followup - it is better to use acceptlist/denylist. I realize this is a historical artifact of our codebase."
+    },
+    {
+      "prediction": "Looks like this is a copy-paste error.",
+      "reference": "Looks like glassfish went a bit nuts doing a refactor"
+    },
+    {
+      "prediction": "Can you add a test for `display: inline-block` as well?",
+      "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 this should be `DateTimeUtils.nanosToMillis(args.getLong(1))`",
+      "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": "(style) line over 80 characters",
+      "reference": "This one had me stumped!"
+    },
+    {
+      "prediction": "I think we should keep the `id` and `name` attributes on the `Optin` component.",
+      "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 this should be `protected` instead of `public` as it's not meant to be used outside of this class.",
+      "reference": "Why public and not protected?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "We definitely need to address this duplication after MVP"
+    },
+    {
+      "prediction": "I think this should be `aria-label` instead of `screen-reader-text`.",
+      "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": "I think this should be `NULL` instead of `\"default\"`.",
+      "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 don't think we need `ref` here. `Bloom` and `LogEntryStructRef` are value types.",
+      "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 == self.keys[0]) and not self.drawn:`",
+      "reference": "Is the idea that ``self.keys[0]`` here normally matches ``init_key``?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that `UiaHasServerSideProvider` is being called on the desktop window, which is not a UIA window. The problem is that `UiaHasServerSideProvider` is being called on the desktop window, which is not a UIA window. The problem is that `UiaHasServerSideProvider` is being called on the desktop window, which is not a UIA window. The problem is that `UiaHasServerSideProvider` is being called on the desktop window, which is not a UIA window. The problem is that `UiaHasServerSideProvider` is being called on the desktop window, which is not a UIA window. The problem is that `UiaHasServerSideProvider` is being called on the desktop window, which is not a UIA window. The problem is that `UiaHasServerSideProvider` is being called on the desktop window, which is not a UIA window. The problem is that `UiaHasServerSideProvider` is being called on the desktop window, which is not a UIA window. The problem is that `UiaHasServerSideProvider` is being called on the desktop window, which is",
+      "reference": "Can this be clarified to \"Exit early when hwnd is the windows desktop handle, UiaHasServerSideProvider would return false anyway.\" ?"
+    },
+    {
+      "prediction": "I think this should be renamed to `CpuLimit` to be consistent with `MemoryLimit`.",
+      "reference": "I'm okay with leaving VCPULimit and MemoryLimit on our internal model"
+    },
+    {
+      "prediction": "I don't think we need to limit this to 10 tiers.",
+      "reference": "5 or 10?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this. I think it should be in the `ExportAsync` method of the `ActivityExporter` class, not in the `MyExporter` class. The `MyExporter` class is just an example of how to use the `ActivityExporter` class.",
+      "reference": "Should we explain more here?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `*v1alpha1.AWSClusterProviderStatus` from the `CreateOrGetMachine` method.",
+      "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 think this should be `@InternalApi` instead of `@Deprecated`.",
+      "reference": "Why not `@InternalApi`?"
+    },
+    {
+      "prediction": "I don't think we need this try/except. `_GenericAlias` is only available in Python 3.7+ and we don't support Python 3.6 anymore.",
+      "reference": "FYI: `GenericMeta` is renamed to `_GenericAlias` in Python3.7."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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 thing to do. I think we should be able to call `clear_examples` multiple times in the same process, and we should be able to call `configuration.reporter.reset` multiple times in the same process. I think we should be able to call `configuration.reset_reporter` multiple times in the same process. I think we should be able to call `configuration.start_time = ::RSpec::Core::Time.now` multiple times in the same process. I think we should be able to call `configuration.reset_filters` multiple times in the same process. I think we should be able to call `configuration.reporter.reset` multiple times in the same process. I think we should be able to call `configuration.reset_reporter` multiple times in the same process. I think we should be able to call `configuration.start_time = ::RSpec::Core::Time.now` multiple times in the same process. I think we should be able to call `configuration.reset_filters` multiple times in the same process. I think we should be able to call `configuration.reporter.reset` multiple times in the same process. I think we should be able to call `configuration.reset_reporter` multiple times in the same process. I think we should",
+      "reference": "Is `Reporter#reset` no longer used? If so, can we remove it?"
+    },
+    {
+      "prediction": "I'm not sure this is correct. The base fee is a `UInt256`, not a `LongScalar`.",
+      "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 think this should be `MvxTvosView` not `MvxTvosView`",
+      "reference": "Can we remove this file?"
+    },
+    {
+      "prediction": "I think this should be `sErr.ErrStatus.Details.Causes[0].Reason` instead of `sErr.ErrStatus.Details.Causes[0].Message`.",
+      "reference": "Why listing only first error reason (`ErrStatus.Details.Causes[0].Message`)?"
+    },
+    {
+      "prediction": "I think this should default to `false` to be consistent with the rest of the codebase.",
+      "reference": "Now that this is true, do we have to ignore it with V1 Tables?"
+    },
+    {
+      "prediction": "I think it would be better to use `Blockly.Colours[Blockly.Colours.colourProperty]` instead of `Blockly.Colours[colourProperty]`.",
+      "reference": "prefer if (condition) { stuff; } to if !(condition) { continue; } stuff"
+    },
+    {
+      "prediction": "I'm not sure about this, but I think it would be better to put the RSS link in the `content_for :heading` block, so that it's only shown when the page has a heading.",
+      "reference": "Should this section not be converted to a secondary action list?"
+    },
+    {
+      "prediction": "`result is False or result is None` can be simplified to `not result`",
+      "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.EnableMediaConversion), $(\"#chkEnablePlaybackRemuxing\", page).checked(user.Policy.EnablePlaybackRemuxing), $(\"#chkEnableSubtitlePlayback\", page).checked(user.Policy.EnableSubtitlePlayback), $(\"#chkEnableSubtitleTranscoding\", page).checked(user.Policy.EnableSubtitleTranscoding), $(\"#chkEnableSubtitleRemuxing\", page).checked(user.Policy.EnableSubtitleRemuxing), $(\"#chkEnableSubtitleDownloading\", page).checked(user.Policy.EnableSubtitleDownloading), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#chkEnableSubtitleEncoding\", page).checked(user.Policy.EnableSubtitleEncoding), $(\"#",
+      "reference": "Can we deuglify this?"
+    },
+    {
+      "prediction": "I don't think this is correct. `flatMap` expects a `PartialFunction<T, Option<R>>`, not a `PartialFunction<T, R>`.",
+      "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": "nit: `schema` -> `schema`",
+      "reference": "Looks like this is an unnecessary change."
+    },
+    {
+      "prediction": "This change is not related to this PR. Please revert it.",
+      "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 this should be `__u8 nat_level1_drop = 0;`",
+      "reference": "We have `stdbool` imported, might as well use that for clarity."
+    },
+    {
+      "prediction": "I think we should keep the `Samples.SqlServer-{SqlServerIntegration.ServiceName}` as the service name.",
+      "reference": "Why do we not also have integration tests for postgres?"
+    },
+    {
+      "prediction": "Why is this 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 `0x16` instead of `22`.",
+      "reference": "Can you make this hexadecimal to match the other constants?"
+    },
+    {
+      "prediction": "I'm not sure if we want to remove this or not. I think it's useful to warn users that they're using a deprecated file format, but I'm not sure if we want to remove the ability to use `project.yaml` entirely.",
+      "reference": "Can this get moved to `Bolt::Project#validate`?"
+    },
+    {
+      "prediction": "I think it would be more efficient to do this in a single pass over the list.",
+      "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` instead of `TAKES_HEAP_SNAPSHOTS`",
+      "reference": "This is not a standard capability and should be hidden behind a vendor prefix: -chromium-takesHeapSnapshot"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we need to pass the algorithm to the execution context so that the execution context can call the algorithm's `get_max_mini_batch_size` method.",
+      "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 it would be good to add a link to the GCC bug report here as well.",
+      "reference": "Sidenote: some day we'll drop the avr-gcc and avr-libc dependencies so that this case becomes supported."
+    },
+    {
+      "prediction": "I don't think this change is needed.",
+      "reference": "do we still need this method?"
+    },
+    {
+      "prediction": "nit: extra space after `)`",
+      "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": "This import is not used.",
+      "reference": "Again, you don't want to depend on glassfish."
+    },
+    {
+      "prediction": "I'm not sure why this is necessary, but it seems to be required for the tests to pass.",
+      "reference": "This shouldn't be required. What was going on here?"
+    },
+    {
+      "prediction": "I think this should be `tls.VersionTLS13`",
+      "reference": "Why add this here?"
+    },
+    {
+      "prediction": "I think this should be `{ReturnType.Headers.AsNullableType(HttpMethod != HttpMethod.Head && IsXNullableReturnType)}`",
+      "reference": "nullability does not apply to headers type (that is always an object)"
+    },
+    {
+      "prediction": "I think this should be `(uint64_t) addr` instead of `(uint64_t) NULL`.",
+      "reference": "Do we also need to add the iova, or is that done somewhere else?"
+    },
+    {
+      "prediction": "I think it would be better to rename this interface to `IdentitySelector` or `IdentitySelectorHandler` to make it more obvious what it is.",
+      "reference": "why change this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get rid of the compiler warning.",
+      "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 `strings.HasSuffix(opts.DockerfilePath, \"Dockerfile\")`",
+      "reference": "Should we validate the suffix before going to the filesystem?"
+    },
+    {
+      "prediction": "I think this should be `default` instead of `ddev_default`",
+      "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": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "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):`",
+      "reference": "This may change \"\" into None. Better return filename."
+    },
+    {
+      "prediction": "I don't think this is a good idea to have this in the `Resources` struct. I think it would be better to have it in the `ContainerConfig` struct.",
+      "reference": ":+1: That does resolve my security concerns about this feature."
+    },
+    {
+      "prediction": "I don't think we need the `from scapy.arch import SIOCGIFHWADDR` here.",
+      "reference": "Could you add a docstring?"
+    },
+    {
+      "prediction": "I think this should be `Region()` instead of `GetUserData()`.",
+      "reference": "Shouldn't this still be `Region()`?"
+    },
+    {
+      "prediction": "I think this should be a `pyqtSignal` as well.",
+      "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 `create_no_window=True` by default.",
+      "reference": "This would be better served as a method/property that is set when people don't want to a window."
+    },
+    {
+      "prediction": "I think this should be `context.getStreamingMethods()` instead of `context.getNonStreamingMethods()`.",
+      "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 `parse_version` from `packaging.version`",
+      "reference": "LooseVersion class can help here"
+    },
+    {
+      "prediction": "I don't think you need to do this. You can just pass `shape` as the default value.",
+      "reference": "`const Dims zeros(shape.size(), 0);`"
+    },
+    {
+      "prediction": "I think this should be `options[:profile_examples] = count.nil? ? 10 : count.to_i`",
+      "reference": "Would be good for this not to say `10` anymore..."
+    },
+    {
+      "prediction": "I don't think we need to pass the service name here, it's already passed in the `getDataErrorComponent` call above.",
+      "reference": "See above, we don't need to pass the module name here."
+    },
+    {
+      "prediction": "This is not needed anymore.",
+      "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 think this should be `if (typeof featureToken !== 'string') { throw new Error(\"featureToken should be a string\"); }`",
+      "reference": "My personal taste: \"depreciated\" -> \"deprecated\""
+    },
+    {
+      "prediction": "I don't think we want to add this to the list of labels. This is something that can be derived from the `destinationServiceName` and `destinationClusterIP`.",
+      "reference": "this includes the port as well right? should the name be `destinationServicePortName`?"
+    },
+    {
+      "prediction": "I think we should use `tenacity.retry` instead of `tenacity.retry`",
+      "reference": "We can import all of them in one line."
+    },
+    {
+      "prediction": "Shouldn't this be `Objects.requireNonNull(elements, \"elements is null\")`?",
+      "reference": "this was wrong, the rest were just inconsistent :)"
+    },
+    {
+      "prediction": "I think this should be `proposals-completed` instead of `past-pr`",
+      "reference": "I gave them the EXACT SAME NAMES in my branch :)"
+    },
+    {
+      "prediction": "I think this should be `0750` instead of `0755`.",
+      "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'm not sure this is the right thing to do, but I don't see a way to get the `TraceContext` from the `SpanContext`.",
+      "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": "I don't think you need all these imports.",
+      "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 `json:\"Msg\"` and `json:\"Level\"` to be consistent with the rest of the struct.",
+      "reference": "You should probably include a `json:...` annotation here."
+    },
+    {
+      "prediction": "I think this should be `ecsMaxRuntimeIDLength` instead of `ecsMaxRuntimeIDLength`.",
+      "reference": "can container ID > 255 chars? why are we doing this check?"
+    },
+    {
+      "prediction": "`This field field prohibits :other from being present.` -> `This field prohibits :other from being present.`",
+      "reference": "*\"This **field field** prohibits :other from being present.\"*, **field** word repeated is right?"
+    },
+    {
+      "prediction": "I think we should move this to pkg/agent/npl.",
+      "reference": "nit: no need to add an alias for this case."
+    },
+    {
+      "prediction": "This should be `QUIC_GO_LOG_LEVEL` instead of `QUIC_LOG_LEVEL`.",
+      "reference": "Or `QUIC_LOG_LEVEL`. Which one do you prefer?"
+    },
+    {
+      "prediction": "nit: I think it would be better to have a constant for \"byte\" instead of hardcoding it here and in the `Parse` function.",
+      "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 it would be good to have a test that checks that the `fingerprint` attribute is not returned in the response when it is not present in the request.",
+      "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'm not sure this is a good test. I think it would be better to have a test that checks that `asMap()` returns a `Map<Integer, Set<Integer>>` and that `asMap().get(1)` returns a `Set<Integer>` that contains `2` and `3`.",
+      "reference": "Thank you! That's better, especially the conversion test should not test the mkString method :)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the correct coordinates.",
+      "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": "Shouldn't this be `fs.Config.MaxTransferMode == fs.MaxTransferModeCautious && accounting.Stats(ctx).GetBytesWithPending()+src.Size() >= int64(fs.Config.MaxTransfer)`?",
+      "reference": "This needs to be done for server-side copies too, earlier in the function."
+    },
+    {
+      "prediction": "nit: I think we can remove the `// TODO(maybe): make scopes part of config?` since it's now done.",
+      "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": "Metrics/BlockLength: Block has too many lines. [21/20]",
+      "reference": "I am not sure what the difference between approving and full approving is..."
+    },
+    {
+      "prediction": "I think we should keep the import as `github.com/docopt/docopt-go` so that it's clear where it comes from.",
+      "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 `dept_ss_dv,name_s_dv,depts:[subquery]` instead of `dept_ss_dv,name_s_dv,depts:[subquery]`",
+      "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 think this should be `GENEVE` instead of `Generic Network Virtualization Encapsulation (GENEVE)`",
+      "reference": "Is this really needed?"
+    },
+    {
+      "prediction": "I'm not sure I understand why this is necessary, but it seems like it should be possible to remove the `if method_defined?(name)` check and the `remove_method(name)` call. Can you help me understand why that's not the case?",
+      "reference": "Is this necessary with line 38?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do. `pytest-benchmark>=3.1` requires `pytest>=3.1`, so we don't need to specify both.",
+      "reference": "DO we want to have >= here? Or =="
+    },
+    {
+      "prediction": "I don't think we should use the CoreTags constant here, because this is not a core tag, it's a custom one.",
+      "reference": "Can you also add this `\"version\"` tag to the CoreTags? Now that we're targeting the service/env/version trio"
+    },
+    {
+      "prediction": "I think this should be `testInvalidValue` instead of `testInvalidVAlue`",
+      "reference": "This entry will be incompatible with the currently implemented formatting rules. Maximum 120 characters per line."
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.",
+      "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": "`getNodeAttributes` returns a `NamedNodeMap` which has a `getNamedItem` method that can be used to get an attribute by name.",
+      "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": "I don't think we need to keep track of the last playback speed. We can just use the current playback speed.",
+      "reference": "I think a float value fits better."
+    },
+    {
+      "prediction": "I don't think we need to handle POST here, right?",
+      "reference": "This should be (POST,DELETE) `/pss/subscribe/{topic}` for subscriptions and there should be `/pss/send/{topic}` for sending."
+    },
+    {
+      "prediction": "I don't think this is necessary. `shouldComponentUpdate` is only called when `props` or `state` change, and neither of those can change here.",
+      "reference": "Closing over the closure arguments is a neat trick :+1: Love it :100:"
+    },
+    {
+      "prediction": "nit: can you add a space after the `->`?",
+      "reference": "Does this need to be a block or can it be an expression?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. This check should be done in `LoginAbstractAzkabanServlet#doGet` and `LoginAbstractAzkabanServlet#doPost`.",
+      "reference": "req.getQueryString() != null is not necessary since same check is already done in allowedPostRequest"
+    },
+    {
+      "prediction": "I think this should be `public abstract boolean isGrpcOnlyTransport();`",
+      "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": "`\u06a9\u0686\u06be \u0633\u06cc\u06a9\u0646\u0688\u0632` should be `\u06a9\u0686\u06be \u0633\u06cc\u06a9\u0646\u0688`",
+      "reference": "here is `:seconds` missing again"
+    },
+    {
+      "prediction": "Line is too long. [83/80]",
+      "reference": "Changed this helper to also get class name."
+    },
+    {
+      "prediction": "I think this should be `APPLICATION_NAME` instead of `applicationName`.",
+      "reference": "I think there's another spot for this in DefaultCapabilityMatcher"
+    },
+    {
+      "prediction": "This should be `PONY_ACTOR_DEFAULT_BATCH` instead of `PONY_ACTOR_DEFAULT_BATCH`.",
+      "reference": "Why the rename from PONY_SCHED_BATCH ? ACTOR_DEFAULT_BATCH is less meaningful to me than SCHED_BATCH."
+    },
+    {
+      "prediction": "I think this should be `0.00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "Let's be sure we have an issue that tracks this, filed against testnet."
+    },
+    {
+      "prediction": "This should be `payload.length !== 20`",
+      "reference": "There should be no space before !Buffer"
+    },
+    {
+      "prediction": "I'm not sure why this is necessary. I don't see any padding applied to the `p` element.",
+      "reference": "Same here, we need the default padding for this story."
+    },
+    {
+      "prediction": "I think this should be `t0 = time.perf_counter()` instead of `time.time()`. `time.perf_counter()` is more precise than `time.time()`.",
+      "reference": "I think this should go away, its noise in the grand scheme of things."
+    },
+    {
+      "prediction": "I think you should check if the file exists before trying to remove it.",
+      "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 call this `setIeDriverService` or `setIeDriverServiceBuilder` to make it clear that this is the service for the driver, not the service for managing the geckodriver process.",
+      "reference": "`this.ieService_` should be initialized to null in the constructor."
+    },
+    {
+      "prediction": "I think this should be `Match().OutInterface(ifaceMatch).IPVSConnection()` instead of `Match().OutInterface(ifaceMatch).IPVSConnection()`.",
+      "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 `getPojoProducers` or `getPojoProducersMeta`",
+      "reference": "This is public method , we need to deprecated this method first and add new updated method for it."
+    },
+    {
+      "prediction": "I think this should be `newNode.NetAddress = \"0.0.0.0:4160\"` instead of `newNode.NetAddress = \"0.0.0.0:4160\"`",
+      "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 this is the right place to set the thanos block number. I think it should be set in the `ProtocolScheduleBuilder` constructor.",
+      "reference": "This field is not needed."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way to pass the reason to the `Cancellation` model.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "remove this line",
+      "reference": "move to line 18 below"
+    },
+    {
+      "prediction": "I'm not sure this is a good example of how to use S3. It's not a common use case to encrypt an object before uploading it to S3. I think it would be better to have an example that shows how to encrypt an object using KMS and then upload it to S3. That would be a more common use case.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to have the `message` output be indented, so that it's clear that it's part of the `root` context.",
+      "reference": "Thats odd, I'd actually not expect this output at all..."
+    },
+    {
+      "prediction": "I think this should be `1.8.3`",
+      "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 test should be renamed to `createFlexibleConfigWithAccountPermissioningDisabledShouldReturnEmpty`",
+      "reference": "this class is permissioning so let's keep \"Onchain\" here"
+    },
+    {
+      "prediction": "Is this change intentional?",
+      "reference": "we can be more specific to call out the it is \"AutoRest C# code generator\"?"
+    },
+    {
+      "prediction": "I'm not sure if this is necessary, but it doesn't hurt to have it.",
+      "reference": "Ditto for not adding this to the axe namespace."
+    },
+    {
+      "prediction": "I think this should be `package registration` (without the backticks).",
+      "reference": "Should we just remove this file? Having it present but empty feels misleading"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to log this. I think it should be logged in the `Trust` function.",
+      "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": "This should be `protected static $createRecordCallback = 'getSolrRecord';`",
+      "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": "Please remove unused use statements.",
+      "reference": "remove this line"
+    },
+    {
+      "prediction": "I think this should be `array<string, 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 don't think this is correct. `API-MS-Win-Core-Kernel32-private-L1-1` should be `kernel32.dll` and `API-MS-Win-Core-Appinit-L1-1` and `API-MS-Win-Core-Com-L1-1` should be `ole32.dll`.",
+      "reference": "nit: to match the others: s/private/Private/"
+    },
+    {
+      "prediction": "I don't think it's a good idea to make this property public. I think it's better to have it as a private property and use it in the `ConnectedPeers` property.",
+      "reference": "Maybe just IEnumerable? Not sure myself."
+    },
+    {
+      "prediction": "nit: can we remove this line?",
+      "reference": "do we need this change?"
+    },
+    {
+      "prediction": "This is a bit of a hack, but I don't see a better way to do this.",
+      "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 this should be `GetVolumeStatusReason` or `GetVolumeStatusMessage` or something like that. `GetVolumeStatus` sounds like it's going to return `volumeStatusOK` or `volumeStatusError` or something like that.",
+      "reference": "The descriptions are incorrect"
+    },
+    {
+      "prediction": "I think this should be `if (debuggerAddress.startsWith(\"ws://\") || debuggerAddress.startsWith(\"http://\")) {`",
+      "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 don't think this is the right place to do this. I think it should be done in the `message` function itself.",
+      "reference": "Please only indent this by four spaces."
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. I don't think we want to use the same set of characters for all of our random strings. I think we should use a different set of characters for each type of random string. For example, we should use a set of characters that doesn't contain any special characters for things like passwords. We should use a set of characters that doesn't contain any numbers for things like usernames. We should use a set of characters that doesn't contain any letters for things like API keys.",
+      "reference": "What's the reason behind these characters?"
+    },
+    {
+      "prediction": "Metrics/BlockLength: Block has too many lines. [31/25]",
+      "reference": "Minor: The name of this specs seems a little off now"
+    },
+    {
+      "prediction": "nit: trailing whitespace",
+      "reference": "I'd suggest using `ast_error_continue` here to get the `apply` message in the `Info` part of the error report."
+    },
+    {
+      "prediction": "This should be `protected $detailConfig;`",
+      "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 don't think we need to handle user data here. The user data will be handled by the user service.",
+      "reference": "Use const for \"user\", same as below"
+    },
+    {
+      "prediction": "I think this should be set to `false` instead of `0`",
+      "reference": "->setUnsubscribed(false) would be better :)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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 why this is necessary, but it seems to fix the issue.",
+      "reference": "Don't you need = something ?"
+    },
+    {
+      "prediction": "I think this should be `VK_QUEUE_SPARSE_BINDING_BIT_KHR` and `VK_QUEUE_PROTECTED_BIT_KHR` instead of `VK_QUEUE_SPARSE_BINDING_BIT` and `VK_QUEUE_PROTECTED_BIT`.",
+      "reference": "Looks like this could be `const` (I realize it wasn't like that before)?"
+    },
+    {
+      "prediction": "I think this should be `FAILED_TO_LOAD(\"failed_to_load\", PreviewType.FAILED_TO_LOAD);`",
+      "reference": "Maybe just `FAILED`. We can't be sure loading was the thing that failed."
+    },
+    {
+      "prediction": "Please revert this change. It's not needed.",
+      "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": "I think this should be `PREF_QUEUE_ADD_TO_FRONT_WHEN_NO_MEDIA`",
+      "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": "nitpick: `counter` should be `index` or `position` or something like that. `counter` is too vague.",
+      "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 think we should move this to constructor as well.",
+      "reference": "At some point, we would like to get rid of these custom injections as well. Looks Ok for now."
+    },
+    {
+      "prediction": "I don't think we need a flag for this. We can just enable it by default.",
+      "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).url().split('://', 1)[0]`",
+      "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 we can remove this import.",
+      "reference": "alphasort this import"
+    },
+    {
+      "prediction": "Please add a space after the `#include` directives.",
+      "reference": "remove this file!"
+    },
+    {
+      "prediction": "I think this should be `suffixLength := fs.Config.StatsFileNameLength / 3` and `prefixLength := 2 * fs.Config.StatsFileNameLength / 3`",
+      "reference": "You don't need to say `rune(' ')` - `' '` is already a `rune`."
+    },
+    {
+      "prediction": "I think we can remove the `.freeze` here as well.",
+      "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": "Why is this needed?",
+      "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.capacity * swarm.ChunkSize / 1024 / 1024`",
+      "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 this should be `AZKABAN_HADOOP_JOB_LINK` and `AZKABAN_SPARK_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": "Why is this change needed? I don't see it in the PR description.",
+      "reference": "This seems to be duplicated with line 19?"
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "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": "`DeleteAccessKey.java demonstrates how to delete an access key from an AWS Identity and Access Management (AWS IAM) user.`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it works for now.",
+      "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": "`NumericType` is already imported from `pyspark.sql.types`, so we don't need to import it again.",
+      "reference": "nit: shall we keep alphabetical order?"
+    },
+    {
+      "prediction": "I think it would be better to have a `typeForConstructor` method on the `Realm` object itself, rather than on the `objects` object. That way, you can do `realm.typeForConstructor(MyClass)` instead of `objects.typeForConstructor(realm, MyClass)`.",
+      "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": "Nit: \"to transform the word\" sounds more natural to me than \"to convert the word\".",
+      "reference": "A correct by unrelated change. We'll slide this one in without another word ;)"
+    },
+    {
+      "prediction": "This should be `if (param_.solver_mode() == Caffe::GPU)`",
+      "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 we need this test.",
+      "reference": "let's replace this example with something that won't fail in Python. Like \"for+if+print\""
+    },
+    {
+      "prediction": "I don't think you need to check for `isinstance(value, str)` here. If `precision` is not `None` and `value` is a string, then `float2str` will raise a `ValueError`.",
+      "reference": "Please remove the brackets and use `string_type` from `compat.py` module instead of `str` in `isinstance()`."
+    },
+    {
+      "prediction": "I don't think we should remove the `srcAddr` field from the `AbstractAction` struct. The `srcAddr` is the address of the sender of the transaction, and the `dstAddr` is the address of the receiver of the transaction. In this case, the sender is the executor, and the receiver is the contract. So, the `srcAddr` should be the address of the executor, and the `dstAddr` should be the address of the contract.",
+      "reference": "line is 143 characters (from `lll`)"
+    },
+    {
+      "prediction": "I think we should return an error here and not just log it.",
+      "reference": "It should follow a `continue`."
+    },
+    {
+      "prediction": "I think this should be `{{ .WorkingDir }}` instead of `{{ .WebWorkingDir }}`",
+      "reference": "Should this be DBWorkingDir?"
+    },
+    {
+      "prediction": "I think this should be `TriggerEventType eventType = TriggerEventType.valueOf(eventTypeStr.trim().toUpperCase(Locale.ROOT));` instead of `TriggerEventType.valueOf(eventTypeStr.trim().toUpperCase(Locale.ROOT));`",
+      "reference": "I think you can remove the whole line."
+    },
+    {
+      "prediction": "remove the empty line",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be \"DELETE FAILED\" instead of \"DELETE IN PROGRESS\"",
+      "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": "I think we should use `MetricsConfig.forTable(table)` instead of `MetricsConfig.fromProperties(properties)`.",
+      "reference": "Can we also update the equality delete branch below?"
+    },
+    {
+      "prediction": "I think this should be `if step_output_handle.step_key == record.dagster_event.event_specific_data.step_key:`",
+      "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": "This is not used in this file, can you remove it?",
+      "reference": "I also do not see any usages of this import. Is this a fix for a crash?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like it should be done before the task is started.",
+      "reference": "Feels like this needs to be invoked in `PostUnmarshalTask` since that's where most of the task setup happens."
+    },
+    {
+      "prediction": "Shouldn't this be `DB__SYSDBA`?",
+      "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 `strings.Split(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 `nameOrHandle` instead of `handle`.",
+      "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'm not sure if this is the best way to do it, but it works for now.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this test should be called `test_generate_ancestors_for_folder` and `test_generate_ancestors_for_org` should be renamed to `test_generate_ancestors_for_project`.",
+      "reference": "nit: folders are always a numeric ID, projects as given by CAI use project number for ancestry"
+    },
+    {
+      "prediction": "I don't think we should have this as a global variable. I think it's better to have it as a field of the service.",
+      "reference": "is this used anywhere?"
+    },
+    {
+      "prediction": "This import is not used in this file. Please remove it.",
+      "reference": "change pkg_errors to errors. Keep it consistent across all the files."
+    },
+    {
+      "prediction": "I think this should be `[\"--url\", self._url, \"--review-status\", \"unreviewed\", \"confirmed\", \"false_positive\"]` instead of `[\"--url\", self._url, \"--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": "Shouldn't this be `if (method.DefaultResponse != null && method.DefaultResponse is CompositeType && !client.Exceptions.Contains((CompositeType)method.DefaultResponse))`?",
+      "reference": "This belongs to SwaggerModeler.cs `public override ServiceClient Build()`"
+    },
+    {
+      "prediction": "I think it would be better to say that this option will disable `AutoColumnSize` if it is enabled.",
+      "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 think this should be `m_renderPass_dependencies.insert(m_renderPass_dependencies.end(), m_additionalSubpassDependencies.begin(), m_additionalSubpassDependencies.end());` instead of `m_renderPass_dependencies.reserve(m_additionalSubpassDependencies.size() + m_renderPass_dependencies.size()); m_renderPass_dependencies.insert(m_renderPass_dependencies.end(), m_additionalSubpassDependencies.begin(), m_additionalSubpassDependencies.end());`",
+      "reference": "I think a `std::copy` with `std::back_inserter` will get this done in one call."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this error. I think it should be in the `quic` package, since it's a QUIC error.",
+      "reference": "In the future: s/QUIC 36/Version36/ to make grepping easier"
+    },
+    {
+      "prediction": "I think this should be `ev.target.checked` instead of `ev.target.checked ? 'on' : 'off'`",
+      "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": "Why is this removed?",
+      "reference": "we should also skip the e2e flow aggregator tests if the Flow Exporter feature is disabled."
+    },
+    {
+      "prediction": "I think this should be `script.ObserveScripts(self, i)` instead of `script.ObserveScripts(self, i)`.",
+      "reference": "We should initialize the observation in the constructor of the Script class - otherwise, we don't have this feature for mitmdump."
+    },
+    {
+      "prediction": "`type(None)` -> `None`",
+      "reference": "I would remove this line. Otherwise looks fine cc @ueshin"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "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 thing to do, but I think we need to make sure that the compiler doesn't optimize this away. I'm not sure if this is the right way to do it, but I think we need to make sure that the compiler doesn't optimize this away. I'm not sure if this is the right way to do it, but I think we need to make sure that the compiler doesn't optimize this away. I'm not sure if this is the right way to do it, but I think we need to make sure that the compiler doesn't optimize this away. I'm not sure if this is the right way to do it, but I think we need to make sure that the compiler doesn't optimize this away. I'm not sure if this is the right way to do it, but I think we need to make sure that the compiler doesn't optimize this away. I'm not sure if this is the right way to do it, but I think we need to make sure that the compiler doesn't optimize this away. I'm not sure if this is the right way to do it, but I think we need to make sure that the compiler doesn't optimize this away. I'm not sure if this is the right",
+      "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)`.",
+      "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 `host.check_output('hostname -s').strip() == 'instance'`",
+      "reference": "Why is this flipped? Looks unrelated and our pattern is `expected == returned`."
+    },
+    {
+      "prediction": "I think it would be better to use `workshop.online? ? t('.heading.online') : t('.heading.in_person')`",
+      "reference": "Can the `h5` tags be pulled outside of the `if` block?"
+    },
+    {
+      "prediction": "I think it would be better to use `Shopsys\\FrameworkBundle\\Component\\Image\\DirectoryStructureCreator` and `Shopsys\\FrameworkBundle\\Component\\UploadedFile\\DirectoryStructureCreator`",
+      "reference": "From @PetrHeinz review: this should be also aliased (`ImageDirectoryStructureCreator`)"
+    },
+    {
+      "prediction": "Why do we need to deep copy here?",
+      "reference": "Isn't AdditionalTags already of `Tags` type? If not, we should make it so, if it's not a breaking change"
+    },
+    {
+      "prediction": "This file is auto-generated. Please don't make changes to this file.",
+      "reference": "Add blank line below license header."
+    },
+    {
+      "prediction": "I think this should be `initPWMClock` (singular).",
+      "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 don't know what this is for, but it's not needed.",
+      "reference": "No idea why it ended up in the code. You can remove the entire line."
+    },
+    {
+      "prediction": "nit: I think we can remove the `namedPortExists` parameter from `toAntreaPeerForCRD` as it is only used here.",
+      "reference": "\"i\" and \"From\" is a bit duplicate, and should it be singular given the method name? I mean \"fromPeer\" or just \"from\"?"
+    },
+    {
+      "prediction": "I don't think we need to add `openid` as a default 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 len(claims.Namespaces) == 0 { return Result{Decision: DecisionDeny}, nil }` instead of `if claims.Namespaces == nil { return Result{Decision: DecisionDeny}, nil }`",
+      "reference": "technically not necessary since reading from a nil dictionary will always return found = false, so this should already be handled"
+    },
+    {
+      "prediction": "I don't think we need to return an api.RelaxedReclaimPurge object here. We can just return an error.",
+      "reference": "This can return bool. RelaxedReclaimPurge(nodeID string) (bool, error)"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it would be nice to have a way to specify which EIPs we want to run.",
+      "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 this should be `AddSourceBlock(calcGenBlk)` instead of `AddSourceBlocks(calcGenBlk)`.",
+      "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",
+      "reference": "should we do the same for os.Mkdir(criuOpts.WorkDirectory, 0755)?"
+    },
+    {
+      "prediction": "I think this should be `Automatically sends usage statistics and crash reports to Google`",
+      "reference": "@bbondy Two messages because the label and description differ very slightly in their wording..."
+    },
+    {
+      "prediction": "I think this should be `proc_num_simd_sse_avx_avx512_registers()`.",
+      "reference": "\"currently\" implies it should and will be changed: so TODO i#?"
+    },
+    {
+      "prediction": "I think you can remove the `computed` here and just use `t('imaging.pageTitle')`",
+      "reference": "This should be `computed('i18n.locale'....`"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "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'm not sure why this was here.",
+      "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. I think it should be hidden in `onCreateOptionsMenu`.",
+      "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 renamed to `executionpb`",
+      "reference": "And for \"execution \"."
+    },
+    {
+      "prediction": "I think this should be `File.directory?(type_files)`",
+      "reference": "Do agents usually download these from puppetserver when applying a catalog?"
+    },
+    {
+      "prediction": "This should be `ev->flags & ASIO_DISPOSABLE == ASIO_DISPOSABLE`",
+      "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": "Shouldn't this be `o.PeriodicExportingMetricReaderOptions.PeriodicExportingMetricReaderOptions.ExportIntervalMilliseconds = options.DefaultCollectionPeriodMilliseconds;`?",
+      "reference": "this is not required, right? its the default already..."
+    },
+    {
+      "prediction": "I think this should be StatusCodeServerErrorRequiredLockNotHeld.",
+      "reference": "\"contingent to\" -> \"contingent on\""
+    },
+    {
+      "prediction": "`GroupByError` should be a base class of `DataError`, and `DataError` should be a base class of `SparkPandasIndexingError`.",
+      "reference": "Does pandas throw an exception like this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what the right thing is either. I think we need to be able to get the device data from the state tracker, but I'm not sure how to do that.",
+      "reference": "The point of the exercise has been to eliminate CoreChecks as an object dependency for DescriptorSet et. al."
+    },
+    {
+      "prediction": "Why do we need to make this class public?",
+      "reference": "Why do we need to make so much more of this ScanContext public?"
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "reference": "is choice_name not needed anymore?"
+    },
+    {
+      "prediction": "I think this should be done before the `ParquetSchemaUtil.convert` call. Otherwise, if the schema is invalid, we will not be able to convert it to a `MessageType` and we will not be able to create the `ParquetWriter`.",
+      "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 `i` instead of `(i - 1)`.",
+      "reference": "Why is this necessary? Just in order to get an indexed loop?"
+    },
+    {
+      "prediction": "I think this should be `zpoolListRequestRejectCounter` instead of `zpoolRequestRejectCounter`.",
+      "reference": "instead of using `prometheus.Gauge`, using `promethus.Counter` will be better."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "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": "Trailing whitespace detected.",
+      "reference": "Why is this necessary?"
+    },
+    {
+      "prediction": "I think we can remove this method and use `Spark3Util.getDistributionMode` instead.",
+      "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 `self.focus` instead of `self.f`.",
+      "reference": "This needs a more self-explaining name. Maybe `.is_focused`?"
+    },
+    {
+      "prediction": "I think this should be 'regenerate:expected:csazurefluent'",
+      "reference": "minor: I'd add a space between `,` & `'paginggroup'`."
+    },
+    {
+      "prediction": "I think this should be `--force` instead of `--reupload`.",
+      "reference": "\"Re-upload all fragments (even if fragment is already in registry)\""
+    },
+    {
+      "prediction": "I think it would be better to have a `PartialResponseStrategy` field in the `PrometheusRuleSpec` struct, and set the `PartialResponseStrategy` field in the `RuleGroup` struct to the value of the `PartialResponseStrategy` field in the `PrometheusRuleSpec` struct. This way, the `PartialResponseStrategy` field in the `PrometheusRuleSpec` struct can be used to set the `PartialResponseStrategy` field in the `RuleGroup` struct, and the `PartialResponseStrategy` field in the `PrometheusRuleSpec` struct can be used to set the `PartialResponseStrategy` field in the `RuleGroup` struct, and the `PartialResponseStrategy` field in the `PrometheusRuleSpec` struct can be used to set the `PartialResponseStrategy` field in the `RuleGroup` struct, and the `PartialResponseStrategy` field in the `PrometheusRuleSpec` struct can be used to set the `PartialResponseStrategy` field in the `RuleGroup` struct, and the `PartialResponseStrategy` field in the `PrometheusRuleSpec` struct can be used to set the `PartialResponseStrategy` field in the `RuleGroup` struct, and the `PartialResponseStrategy`",
+      "reference": "I don't see us ignoring the field. Are we sure Prometheus wouldn't refuse to load the rules?"
+    },
+    {
+      "prediction": "I think this should be `!mSession.hasDisplay() || !SettingsStore.getInstance(getContext()).getLayersEnabled()`",
+      "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 it would be better to have a default of 1000MB (or 1GB) for this setting. This is the default for `EXE_MEMORY_FOR_PROBE_CACHE_IN_MB` and `EXE_MEMORY_FOR_UNPACK_ROWS_IN_MB` is only used when `EXE_MEMORY_FOR_PROBE_CACHE_IN_MB` is set to 0.",
+      "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 the `bool` here anymore.",
+      "reference": "think we should pass in ctx, and use ctx.GreenlandHeight inside bucketPool to determine if create or not"
+    },
+    {
+      "prediction": "Nit: \"to transform the word\" sounds more natural to me than \"to convert the word\".",
+      "reference": "A correct by unrelated change. We'll slide this one in without another word ;)"
+    },
+    {
+      "prediction": "I'm not sure why this was here in the first place.",
+      "reference": "Why remove this?"
+    },
+    {
+      "prediction": "`Twig_Environment` -> `Twig\\Environment`",
+      "reference": "I still see some usages of not namespaced variants (look for Twig_ in project, about 42 matches). Is it intentional?"
+    },
+    {
+      "prediction": "I think we should keep the `github.com/aws/amazon-ecs-agent/agent/api` import here. It's not used in this file, but it's used in `cgroup_linux.go` and `cgroup_windows.go`.",
+      "reference": "Is this a new dependency? Does this require a dep update?"
+    },
+    {
+      "prediction": "I think this should be `type.base_type == BASE_TYPE_FLOAT || type.base_type == BASE_TYPE_DOUBLE || type.base_type == BASE_TYPE_LONG_DOUBLE`",
+      "reference": "> opts.generate_hexfloat_in_json && **IsFloat**(type.base_type)"
+    },
+    {
+      "prediction": "I think this should be `AppName() (string, error)` so that we can return an error if the app name cannot be determined.",
+      "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 `harparser>=1.0.0`",
+      "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 this is the best way to do this. I think it would be better to have a `gulpfile.js` in the root of the project and have it require the `gulpfile.iced` in the `.gulp` folder. That way you don't have to change the `gulpfile.js` every time you want to add a new task.",
+      "reference": "makes it work nice with vscode. all logic is now in `.gulp/*.iced` files"
+    },
+    {
+      "prediction": "This should be `List<? extends WebElement> findElements(By by);`",
+      "reference": "should a similar change be done for findElement?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "`+` -> ` + ` I guess. Same below"
+    },
+    {
+      "prediction": "I think this should be `opts.feeSat || opts.fee * util.COIN`",
+      "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'm not sure if this is the right place to put this. I think it should be in `test/test_helper.rb` in the `windows?` method, but I'm not sure if that's the right place to put it.",
+      "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'm not sure this is the right place for this field. I think it should be in the `ClusterSyncStatus` struct.",
+      "reference": "how about \"all matching SyncSets and SelectorSyncSets\""
+    },
+    {
+      "prediction": "I think this should be `io::FileReader::HasNoFingerprint`",
+      "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": "This import is not used, please remove it.",
+      "reference": "This looks like an unused import to me?"
+    },
+    {
+      "prediction": "I think we should keep the `sqlblobs` prefix in the function name.",
+      "reference": "`RW` means read/write. Why do we have it here, as part of a func name?"
+    },
+    {
+      "prediction": "I don't think we need to log this at `finest` level.",
+      "reference": "I'd keep this at `info` level..."
+    },
+    {
+      "prediction": "I think this should be `H2O_HTTP2_ERROR_INTERNAL_ERROR`",
+      "reference": "I think you missed this."
+    },
+    {
+      "prediction": "I think this should be `#if NETCOREAPP2_1 || NETSTANDARD2_1`",
+      "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 `req.GetXattr() != nil` instead of `req.GetXattr() >= 0`",
+      "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'm not sure if this is the right place to set the position to 0. I think it should be set to 0 when the media is added to the queue.",
+      "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 don't think this is correct. If `demangled` is `nullptr` then `name.assign(demangled == nullptr ? \"\" : demangled);` will assign `name` to an empty string, but `free(demangled);` will cause a segfault because `demangled` is `nullptr`. I think this should be `if (demangled != nullptr) { name.assign(demangled); free(demangled); }`",
+      "reference": "Please remove it or remove all `std::` namespace prefixes."
+    },
+    {
+      "prediction": "This file is empty. Can we remove it?",
+      "reference": "wait! I'm confused - is that an empty file ?!"
+    },
+    {
+      "prediction": "`virtualNode.actualNode` should be `virtualNode.actualNode.actualNode`",
+      "reference": "These should all call `findUpVirtual`."
+    },
+    {
+      "prediction": "I think this should be `file_dict[\"mime-type\"] = mime`",
+      "reference": "This changes the behavior. Original behavior was \"set if not set\", while new is \"just set\"."
+    },
+    {
+      "prediction": "I don't think this is a good idea. If the `Authorization` header is not in the expected format, this will throw a `NullPointerException`.",
+      "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": "`delegate` is not a good name for this field. It's not a delegate, it's a backing array. `back` would be a better name.",
+      "reference": "All the operations are delegated to this entity, hence the rename. `back` can have too many meanings."
+    },
+    {
+      "prediction": "Style/RedundantElse: Redundant else detected.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br",
+      "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 don't think we need to import `protocol` here.",
+      "reference": "Could you delete this constant?"
+    },
+    {
+      "prediction": "I don't think we need to add `Disassemblers` namespace here.",
+      "reference": "looks like this snuck in from your other change?"
+    },
+    {
+      "prediction": "Extra empty line detected at class body end.",
+      "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'm not sure why we need to pass in the `template` here. I think we can remove it.",
+      "reference": "this partial is used also for previewing a template, did you test if still works?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this, but I think it would be good to add a note here about the fact that the app name will be used as the ECR repository name, and that it must be unique across all ECR repositories in your AWS account.",
+      "reference": "Would it be better to put like `Which Dockerfile would you like to use for %s?`"
+    },
+    {
+      "prediction": "I don't think we need to change this.",
+      "reference": "You mistakenly removed `/docs` here, please add it"
+    },
+    {
+      "prediction": "It is important that the returned list is sorted in ascending order. -> It is important 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 this should be `alerts.Filterer` instead of `alerts.FilterDeleter`.",
+      "reference": "Please change this to non-exported."
+    },
+    {
+      "prediction": "(style) 'drstats' may be misspelled - perhaps 'drstats' ?",
+      "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` to account for the round change message itself.",
+      "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 don't think we need to include `stdlib/runtime` here.",
+      "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": "I don't think it is necessary to change the namespace here.",
+      "reference": "Annotations in this hunk should follow PhpDoc style (above the occurrence, type first)"
+    },
+    {
+      "prediction": "I don't think we need to throw an exception here. We can just return `\"? extends \" + boundedType.describe()`.",
+      "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'm not sure why this was changed, but I think it should be `*os.File` instead of `int`. `os.File` implements `io.Closer`, which is what we want here.",
+      "reference": "nit: `procfd` seems more appropriate now?"
+    },
+    {
+      "prediction": "I don't think we need to pass the auth header here.",
+      "reference": "Let's not call it django. Maybe `auth_provider_headers`? `auth_headers`?"
+    },
+    {
+      "prediction": "I'm not sure I agree with this change. I think it's better to keep the order as it was before.",
+      "reference": "Should we mention TLS map here?"
+    },
+    {
+      "prediction": "This should be `@ConfigItem(keyName = \"clanColor\", name = \"Clan color\", description = \"Color for clan members\")`",
+      "reference": "remove this empty line pl0x"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "Why not have this in the `Dispatcher`? Doesn't seem like NCR-specific functionality."
+    },
+    {
+      "prediction": "I don't think you need to return the chainpacks here.",
+      "reference": "If you need only test db, use `memdb.NewTestDB(t)`"
+    },
+    {
+      "prediction": "Why is this change necessary? I don't think it's related to this PR.",
+      "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": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 think this should be `0x0408` instead of `0x0804`.",
+      "reference": "Is this change needed?"
+    },
+    {
+      "prediction": "I think this should be `tmpInt > 0 && tmpInt < 128`",
+      "reference": "Isn't the % 128 redundant here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add these, but I think it makes sense to add them here.",
+      "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 `#if defined(MPI) && defined(BATCH_QUEUE_MPI)`",
+      "reference": "Please change MPI to CCTOOLS_WITH_MPI"
+    },
+    {
+      "prediction": "I don't think we should use `domain_error` here. We should use `invalid_argument` or `runtime_error` instead. `domain_error` is used for errors related to mathematical domain, e.g. `sqrt(-1)`, `log(-1)`, `atan2(0, 0)`, etc. `invalid_argument` is used for errors related to invalid arguments, e.g. `sqrt(NAN)`, `log(NAN)`, `atan2(NAN, 0)`, etc. `runtime_error` is used for errors related to runtime conditions, e.g. `malloc(0)`, `malloc(INT_MAX)`, etc.",
+      "reference": "Why you can't use function here?"
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "We should scope this down to only the methods we use."
+    },
+    {
+      "prediction": "I think this should be `mkString(\"[\", \", \", \"]\")`",
+      "reference": "Oh, thanks for catching - I thought I've catched all after changing `mkString(infix, prefix, suffix)` to `mkString(prefix, infix, suffix)`."
+    },
+    {
+      "prediction": "@deads2k @liggitt @smarterclayton I think we need to add a `CertificateRequestReasonPending` to indicate that the request is waiting for an approver to approve it.",
+      "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": "Shouldn't this be `if ( ! accountID ) { return null; }`?",
+      "reference": "We should use `! isValidAccountID( accountID )` for the `accountID` checks."
+    },
+    {
+      "prediction": "I don't think this is wrong. I think this is the right way to do it. The alternative would be to pass the inventory through the executor to the transport, which seems like a lot of extra work for not much gain.",
+      "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": "I don't think we need to install `puppet-common` if we're going to install `puppet`.",
+      "reference": "Missing the `-y` argument which all other `apt-get install` commands have."
+    },
+    {
+      "prediction": "This should be PRIu32, not PRIu64.",
+      "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'm not sure if this is the right thing to do, but it seems to fix the issue.",
+      "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 don't think we need the `_, _, _` here.",
+      "reference": "I think we want to provide getAction API instead"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do for now.",
+      "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": "Style/RedundantTrue: Redundant true detected.<br>Metrics/LineLength: Line is too long. [114/100]",
+      "reference": "This doesn't seem like it should be part of this pull request."
+    },
+    {
+      "prediction": "Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "Is the idea that `@product` (and `@workshop` for `workshops_controller`) would eventually go away here?"
+    },
+    {
+      "prediction": "I don't think we need to hardcode the pre-verified hashes for BSC mainnet. If we want to do that, we should also do it for the other networks.",
+      "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": "nit: I think we can remove the `config_instance` argument here, as it's not used in the test.",
+      "reference": "And how is that related?"
+    },
+    {
+      "prediction": "`c.BoolT(FlagConnectionEnableAlias)` -> `c.Bool(FlagConnectionEnableAlias)`",
+      "reference": "this should be required"
+    },
+    {
+      "prediction": "I think this should be `@Nullable` instead of `@Null`.",
+      "reference": "This appears to be unused"
+    },
+    {
+      "prediction": "I think this should be `fromSnapshotId, toSnapshotId]` to be consistent with `ancestorIds`",
+      "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 don't think we need to implement these.",
+      "reference": "This looks like a bug: pretty sure there are scatter-gather instructions on AArch64. Ditto below."
+    },
+    {
+      "prediction": "nit: I think it would be nice to add a `//nolint:govet` to this line to avoid the linter complaining.",
+      "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 `m.Name` instead of `m.AppName()`.",
+      "reference": "Should this be in the parent struct? `AppManifest` since it's embedded to `LBFargateManifest` it'll get the method as well."
+    },
+    {
+      "prediction": "nit: please remove the extra blank line here.",
+      "reference": "\"Openvpn check\" we could move this to separate prefix."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the request fails, the connection is closed, so the next request will fail as well. I think it would be better to close the connection only if the request is successful.",
+      "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 remove this?",
+      "reference": "We don't return now?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should only retry when the build fails, not when it times out.",
+      "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 think we should keep the copyright year as 2020.",
+      "reference": "Is there a BOM change?"
+    },
+    {
+      "prediction": "why remove it?",
+      "reference": "It's not a good practise to delete the public static constant."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should still check the well-formedness of the transaction before verifying the signature.",
+      "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 think this should be `>=3.4,<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 this should be `if instance.Spec.MaintenanceMode == nil || !*instance.Spec.MaintenanceMode {`",
+      "reference": "This can be simplified somewhat to `pointer.Int32Ptr(0)`. But it is not necessary."
+    },
+    {
+      "prediction": "I think this should be `bs1 != Bond::STEREOANY && !(bs2 != Bond::STEREOANY)`",
+      "reference": "Clever but perhaps confusing."
+    },
+    {
+      "prediction": "I think this should be `if err != nil { return reconcile.Result{}, err }` so that we don't try to remove the finalizer if there was an error deleting the secret references.",
+      "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 this is the right thing to do. I think we should return `false` here and let the caller decide what to do.",
+      "reference": "Should you capture o.asflr before releasing consumer's lock?"
+    },
+    {
+      "prediction": "I don't think this is the right place for this check. It should be done at the beginning of the `initialize` method.",
+      "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": "Layout/EmptyLinesAroundClassBody: Extra empty line detected at class body end.",
+      "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 don't think this is correct. If `enabled_features_found` is `nullptr`, then `device_data->enabled_features.core` will be uninitialized.",
+      "reference": "I'd move the empty assignment and non-null case into the if check directly above (adding an else clause as needed)"
+    },
+    {
+      "prediction": "I don't think we should add this here. We should add it to the `config/system.php` file.",
+      "reference": "Please remove this extra line of whitespace"
+    },
+    {
+      "prediction": "Shouldn't this be `p.id >= :lastProcessedId`?",
+      "reference": "maybe it's time to rename `ProductSearchExportWithFilter` to something better, what do you think?"
+    },
+    {
+      "prediction": "I think this should be `appendNameOnlyAttribute(capabilities, attributePrefix+capabilityFirelensFluentd, attributePrefix+capabilityFirelensFluentbit)`",
+      "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_` to avoid shadowing the built-in `hash` function.",
+      "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 right thing to do, but I think we can remove the `github_team` attribute from the `GithubRemovalJob` struct. We don't use it in the `perform` method, and we don't use it in the `enqueue` method either, so I think it's safe to remove it.",
+      "reference": "Don't extend an instance initialized by `Struct.new`."
+    },
+    {
+      "prediction": "I think this should be `pmix_list_remove_item(&t->nslist, &nm->super); PMIX_RELEASE(nm);`",
+      "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 this is necessary. The `ApexRuleSetFactoryTest` already tests that Apex rules are found in the classpath, and the `VFRuleSetFactoryTest` already tests that VF rules are found in the classpath.",
+      "reference": "I think, we should fix/improve AbstractRuleSetFactoryTest. I guess, both apex and visualforce rules are now tested, which is unnecessary."
+    },
+    {
+      "prediction": "I don't think this is correct. The parameters should be passed as an associative array, not as an indexed array.",
+      "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": "Please remove unused imports.",
+      "reference": "easily possible to get that information without exposing internal information"
+    },
+    {
+      "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 need tristate logic (pe, foss, nil)?"
+    },
+    {
+      "prediction": "nit: `mux` instead of `mutex`?",
+      "reference": "this change needs to be reverted to what is on `master`"
+    },
+    {
+      "prediction": "I think we can remove the `variableInMap` variable and just do `if (this.getVariable(name) == null) { this.refreshToolboxSelection_(); }`",
+      "reference": "Do you still need to call the superclass `createVariable` if you've already determined that the variable exists?"
+    },
+    {
+      "prediction": "I think it would be better to have a separate function for this. The name `EachByPack` is a bit misleading.",
+      "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": "`std::find_if` would be better here.",
+      "reference": "This is a revert, is it really an issue?"
+    },
+    {
+      "prediction": "Why is this commented out?",
+      "reference": "can we delete this file altogether?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I don't know what else to do.",
+      "reference": "Suggestion: log some additional info here so we know where we are i.e 'Failed to prepare coverage framework..'"
+    },
+    {
+      "prediction": "I don't think we need this class. We can just use `TracerConstants.Language` directly.",
+      "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": "`UpdateDeviceStatusBuffer` is the size of channel which save update device status message from edge",
+      "reference": "Log message should be started with upper-case word."
+    },
+    {
+      "prediction": "I think this should be `Layout<int?>` instead of `Layout<int>`.",
+      "reference": "changes for example usage in this class"
+    },
+    {
+      "prediction": "This should be `elements/emby-collapse/emby-collapse.css`",
+      "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": "Should this return an error if the application doesn't exist?",
+      "reference": "`ctx` is unused in EnableApplication"
+    },
+    {
+      "prediction": "I think this should be `return stringtab(def->name);` 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 don't think we need this import anymore.",
+      "reference": "Should not be a static import."
+    },
+    {
+      "prediction": "`AttachmentARN` should be `AttachmentARN` instead of `AttachmentARN`",
+      "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 don't think we need to pass `startDate` and `endDate` to `getServiceReportURL` as it's already being passed to `getReport` below.",
+      "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 this should be `False` by default.",
+      "reference": "`default=True` shouldn't be used for BoolParameters iirc."
+    },
+    {
+      "prediction": "I think this should be `strings.HasPrefix(refStr, \"refs/heads/\") || strings.HasPrefix(refStr, \"refs/tags/\")`",
+      "reference": "It would be better to use `return refStr[12:]`. or `11`... I have problem with counting.."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should keep the test outputs as part of the rule hash, but I'm not sure how to do that.",
+      "reference": "Pretty sure these should only contribute to the runtime hash."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get rid of the `#ncr-layout` id.",
+      "reference": "Just verifying: this change doesn't break the 18f layout, right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to remove this.",
+      "reference": "we should keep this counter using the \"isYarpcError\" api"
+    },
+    {
+      "prediction": "I think we should use `CaseInsensitiveStringMap` instead of `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 think this should be `return slot as RuntimeContextSlot<T>;`",
+      "reference": "I think this should be a throw. Or the method should be TryGetSlot?"
+    },
+    {
+      "prediction": "I think this should be `std::cerr` instead of `std::cout`.",
+      "reference": "We don't usually put spaces between ( and [, and I think `path` should be passed by reference."
+    },
+    {
+      "prediction": "I think this should be `if (IsStackable && charges > 0) || (!IsStackable && !sold) {`",
+      "reference": "Make sure to `snake_case` locally scoped variable names in the future"
+    },
+    {
+      "prediction": "I think this should be `HealthCheckPath` instead of `HealthcheckPath`.",
+      "reference": "Maybe we should use underscores like `health_check` - what do you think?"
+    },
+    {
+      "prediction": "I think this should be `if _, ok := vol.Annotations[string(v1alpha1.StorageClassKey)]; !ok {`",
+      "reference": "Do a TrimSpace before setting."
+    },
+    {
+      "prediction": "nit: I think this should be `SubnetGatewayIPv4Address` to be consistent with the other fields.",
+      "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": "`Link.find(link.id)` would be enough here, right?",
+      "reference": ".find(id) is being deprecated to Rails 5."
+    },
+    {
+      "prediction": "These imports are not used in this file. Please remove them.",
+      "reference": "Will be good to have the iris interface moved to geoviews. Could this be done for 1.10.6?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `if err != nil { return err }` here and just `return utils.SignHTTPRequest(request, cs.AgentConfig.AWSRegion, ServiceName, cs.CredentialProvider, nil)`",
+      "reference": "does this get wrapped in the calling method?"
+    },
+    {
+      "prediction": "I'm not sure what you mean by \"successful group authorization\". Do you mean \"successful authorization of a group\" or \"successful authorization of a user\"? If the former, then I don't think this test is doing what you want it to do. If the latter, then I don't think this test is doing what you want it to do either.",
+      "reference": "Do we still need this test case? If your new test cases test this, then we can remove it."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [118/100]",
+      "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 I follow this. If the CA is an intermediate, then the root should be in the chain. If the CA is not an intermediate, then the root should not be in the chain. Am I missing something?",
+      "reference": "Do we need any update on the test side?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this, but I don't have a better idea.",
+      "reference": "There's a few `return nil` cases above this -- we should probably move this above the big switch."
+    },
+    {
+      "prediction": "I think this should be `fmt.Sprintf(\"transaction %v using an overlapping lease %s\", lile.txid, lile.lease.lease.String())`",
+      "reference": "`lile *LeaseInLedgerError` to reduce copying?"
+    },
+    {
+      "prediction": "I think this should be `/se/grid/newsessionqueuer/queue`",
+      "reference": "should it be `newsessionqueue` or `newsessionqueuer`? In case we'd like to be consistent"
+    },
+    {
+      "prediction": "Why is this defaulting to ReadWrite?",
+      "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": "This import is unused.",
+      "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": "Shouldn't this be `neve-transparent-header` instead of `neve-tansparent-header`?",
+      "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": "I think we can remove this line.",
+      "reference": "Out of curiosity.. is there a rule to determine the package name as prevDirectory+currentDirectory."
+    },
+    {
+      "prediction": "I think this should be `else if (YMM_ENABLED() || ZMM_ENABLED())`",
+      "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 don't think we need to link to the support page here.",
+      "reference": "Why not use the JSdoc annotation as it did before?"
+    },
+    {
+      "prediction": "I think it would be better to use `ProductPropertiesUtil.getProductVersion()` instead of `packageImplementationVersion`.",
+      "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::set<const FieldDef*> union_fields;` to avoid duplicates.",
+      "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 `copilot-taskgroup` instead of `copilot-task`.",
+      "reference": "nit: What do you think of `copilot-task-group`?"
+    },
+    {
+      "prediction": "This should be `if (e.type === 'input' || e.type === 'change') { this.value = this._prevValue || ''; }`",
+      "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": "This should be (2021, 2, 0)",
+      "reference": "I think there should be some reasoning below this line about the version updating."
+    },
+    {
+      "prediction": "I think it would be better to use `strncasecmp(val, \"ON\", len)` and `strncasecmp(val, \"OFF\", len)` instead of `!strncasecmp(\"ON\", val, len)` and `!strncasecmp(\"OFF\", val, len)`.",
+      "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 if this is the right thing to do, but I think it makes sense to have the block cache be nil in minimal mode. This way, the user can't try to do any data reads or writes, and we don't have to worry about the block cache being in an inconsistent state.",
+      "reference": "I prefer if possible for `if`/`else` statements to have the positive case first."
+    },
+    {
+      "prediction": "I don't think we need this function. We can just use `isinstance(x, type) and issubclass(x, t)` directly.",
+      "reference": "Docstring would be nice"
+    },
+    {
+      "prediction": "I think this should be added to the baseline file instead of the extension.",
+      "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": "I think this should be `KOKKOS_ARCH_VOLTA70 || KOKKOS_ARCH_PASCAL60 || KOKKOS_ARCH_TURING75 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80 || KOKKOS_ARCH_AMPERE80",
+      "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 use `modelzoo://resnet50` instead of `torchvision://resnet50`",
+      "reference": "Since we have specified `load_from`, `pretrained` can be left None."
+    },
+    {
+      "prediction": "I think this should be `len(args) > 4`.",
+      "reference": "I prefer to translate these `else if` into `switch` for more readability."
+    },
+    {
+      "prediction": "I'm not sure if this is the best place to do this, but I think it makes sense to update the trails state when the status is created.",
+      "reference": "This can also be a local variable in `create`."
+    },
+    {
+      "prediction": "I think you can use `os.OpenFile(daemonExportFile, os.O_CREATE|os.O_WRONLY, 0644)` instead of `os.Create`.",
+      "reference": "Why not use `path.Join`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be working for now.",
+      "reference": "Why not `export default API` instead? This ensures we export the same that we export on the global."
+    },
+    {
+      "prediction": "I don't think this is correct. The `itk.ImageIOFactory.FileModeType_ReadMode` should be `itk.ImageIOFactory.ReadMode`.",
+      "reference": "This is a good change I think. It greatly helps reduce name conflicts."
+    },
+    {
+      "prediction": "I think this should be a `map[string][]string` so that we can have multiple tags per cluster.",
+      "reference": "cluster not having omitempty, does this imply that cluster is required? seems empty is valid so just checking"
+    },
+    {
+      "prediction": "I think this should be `c.lock.RLock()` and `c.lock.RUnlock()` instead of `c.lock.Lock()` and `c.lock.Unlock()`.",
+      "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 don't think we need to make `aspects` a `Set`. We can just use an array.",
+      "reference": "nit: aspects are defined in `OperationBase`, should `defineAspects` live there as well?"
+    },
+    {
+      "prediction": "I don't think we need to do this. The `Migrator` class is already imported in `storage/postgresql/__init__.py` so it should be available in the `Permission` class.",
+      "reference": "Why the `os.path.abspath`? `os.path.dirname` should always give a valid directory path."
+    },
+    {
+      "prediction": "I think this should be a `Path` instead of a `String`.",
+      "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 think we need to add a test case for this.",
+      "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 `nameof(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": "This should be `v30.0` now.",
+      "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 anymore.",
+      "reference": "there is no \"DEFAULT_REVISION\" logic, no need to define this."
+    },
+    {
+      "prediction": "Why is this needed? It looks like `loginOptions.setUrl(url);` is already called on line 294.",
+      "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 to do this. I think it should be done in the `focusElement_` method.",
+      "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: please remove this empty line",
+      "reference": "We don't add blank lines in imports."
+    },
+    {
+      "prediction": "This should be 2018-2019, not 2012-2019.",
+      "reference": "Please add your name to the copyright list."
+    },
+    {
+      "prediction": "I think this should be `if reqid == \"\" { return \"\" }` instead of `if reqid != \"\" { return reqid }`",
+      "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": "I don't think we need to change the labels here, as they are not used anywhere else in the code.",
+      "reference": "Different i18n keys. By mistake or on purpose?"
+    },
+    {
+      "prediction": "I think this should be `ACMEChallengeSelfCheckFailurePolicyRetryForever` and `ACMEChallengeSelfCheckFailurePolicyIgnore`",
+      "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'm not sure this is the right thing to do here. I think we should be using `ot_gpgme_error_to_gio_error` here.",
+      "reference": "Nit: let's add a more descriptive prefix in line with the other ones? E.g. `Unable to import key` ?"
+    },
+    {
+      "prediction": "This should be `Dir.glob(\".vagrant/**/private_key\").first`",
+      "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": "Why do we need this file?",
+      "reference": "Do we need this?"
+    },
+    {
+      "prediction": "I think this should be `MatchPriority(priority uint32) FlowBuilder` instead of `MatchPriority(priority uint16) FlowBuilder`.",
+      "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 \"packaged\" instead of \"packaged\".",
+      "reference": "In this case we probably meant \"package\"."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Just export `createFile` directly."
+    },
+    {
+      "prediction": "This should be `//tools/go_buildid_replacer:go_buildid_replacer`",
+      "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": "`opts.network = networks['btc'].testnet;`",
+      "reference": "interface for bitcoin should not change if possible. i.e: networks.testnet should return networks['btc'].testnet"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I think this should be `int64_t timespec_to_msec(const struct timespec *a) { return (int64_t)a->tv_sec * 1000 + a->tv_nsec / 1000000; }`",
+      "reference": "We have this functions in a couple of places. It should probably live in util or something."
+    },
+    {
+      "prediction": "I think this should be `replacement.new_record = true` instead of `_target.new_record = true`.",
+      "reference": "maybe this should be inside the destroy?"
+    },
+    {
+      "prediction": "I think it's better to use `img_shape_for_onnx` instead of `img_shape`.",
+      "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 `return nil, errors.New(\"Instance does not have an instance id\")`",
+      "reference": "I think we want to return an error here, not nil."
+    },
+    {
+      "prediction": "I think this should be `user_input_state: 'complete'`.",
+      "reference": "Nitpicking, but I think it would be more accurate to use an actually supported value, i.e. `completed`."
+    },
+    {
+      "prediction": "Line is too long. [85/80]",
+      "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 don't think we need to create a new factory here. We can just use the one we created above.",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "I think this should be `return nil, nil` instead of `return nil, cache.ErrUnimplemented`",
+      "reference": "`k` is unused in PutHash"
+    },
+    {
+      "prediction": "I think this should be `?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 = 1;`",
+      "reference": "Should we remove this entirely now then?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this interface. I think it would be better to have it in the `ring` package, and have the `ring` package depend on the `monitor` package, rather than the other way around. That way, the `ring` package doesn't need to know anything about the `monitor` package, and the `monitor` package doesn't need to know anything about the `ring` package.",
+      "reference": "That is way too generic a name."
+    },
+    {
+      "prediction": "Can you add a test for the case where the `id` is on a different element than the `data-testelm` attribute?",
+      "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": "This should be `VK_COMMAND_BUFFER_USAGE_ONE_TIME_SUBMIT_BIT | VK_COMMAND_BUFFER_USAGE_RENDER_PASS_CONTINUE_BIT`",
+      "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 think it would be better to move this to `ElasticsearchHttpSpanConsumer` and call it from `ElasticsearchHttpSpanConsumer.resetIndexToServiceSpansCache()` and `ElasticsearchHttpSpanConsumer.resetIndexToServiceSpansCache()`",
+      "reference": "es.clear should call this"
+    },
+    {
+      "prediction": "nit: extra blank line",
+      "reference": "please remove. this is true by the definition of the Python language"
+    },
+    {
+      "prediction": "`<Ctrl-Return>` is already bound to `follow`, so I think this should be `<Ctrl-Shift-Return>` instead.",
+      "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": "nit: I think it would be better to have a `icmpPkt_with_type_code(src, dst string, icmpType, icmpCode int) packet` helper function that calls `icmpPkt_with_type_code(src, dst string, icmpType, icmpCode int) packet` instead of duplicating the code.",
+      "reference": "Golang naming convention is to use camel case `icmpPktWithTypeCode` Often the linter will complain"
+    },
+    {
+      "prediction": "I think it would be better to make these properties on `KestrelServerOptions` instead of `HttpConnectionContext`.",
+      "reference": "nit: Probably should renamed to `InitializeBody`"
+    },
+    {
+      "prediction": "I think this should be `blogIcon || 'favicon.ico'` so that if `blogIcon` is an empty string, it defaults to `favicon.ico`",
+      "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 think this should be `RegisterBootstrapContainer` instead of `RegisterBootstrapContainerProvider`",
+      "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 / 1024 / 1024`",
+      "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 think this should be `!!label && axe.commons.text.sanitize(label).trim() !== '' && !!text && axe.commons.text.sanitize(text).trim() !== ''`",
+      "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 `unset($options[\"tax_engine\"])` instead of `unset($options[\"tax_engine\"])`",
+      "reference": "Why did you change the signature of this method ? You can break BC doing that"
+    },
+    {
+      "prediction": "I think this should be `if _, err := os.Stat(cfg.ConfigTemplate); err == nil {`",
+      "reference": "can you catch the error here please and return err if it can't read the template"
+    },
+    {
+      "prediction": "I don't think this is the right place to call this method. It should be called when the library is loaded, not when the app is loaded.",
+      "reference": "Is this method called at another location too? I'm just asking because you didn't remove another call to `updateLibrary`"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [123/120]",
+      "reference": "Not necessary but you should be able to add a `?` on `is_default?` and `published?`"
+    },
+    {
+      "prediction": "I'm not sure this is a good benchmark. I think we should be benchmarking `decodeList` with a `ByteBuf` and `decodeList` with a `ByteBuffer`. I don't think we should be benchmarking `decodeList` with a `ByteBuffer` and `decodeList` with a `ByteBuffer`. I think we should be benchmarking `decodeList` with a `ByteBuf` and `decodeList` with a `ByteBuffer`. I don't think we should be benchmarking `decodeList` with a `ByteBuffer` and `decodeList` with a `ByteBuffer`. I think we should be benchmarking `decodeList` with a `ByteBuf` and `decodeList` with a `ByteBuffer`. I don't think we should be benchmarking `decodeList` with a `ByteBuffer` and `decodeList` with a `ByteBuffer`. I think we should be benchmarking `decodeList` with a `ByteBuf` and `decodeList` with a `ByteBuffer`. I don't think we should be benchmarking `decodeList` with a `ByteBuffer` and `decodeList` with a `ByteBuffer`. I think we should be benchmarking `decodeList` with a `ByteBuf` and `decodeList` with a `ByteBuffer",
+      "reference": "here is us decoding directly the bytebuffer in benchmarks"
+    },
+    {
+      "prediction": "I think we should still set the default values, but only if the underlying logging framework supports it. If it doesn't support it, we can just ignore it and not set the default values. What do you think?",
+      "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 `JsonRpcError.OFFCHAIN_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 `Vectorized bool` instead of `Vectorized Expression`.",
+      "reference": "This could be made into a function expression, yes?"
+    },
+    {
+      "prediction": "I think this should be `abort ();`",
+      "reference": "Oops, I missed this initially."
+    },
+    {
+      "prediction": "I think it would be better to use `short_uid()` instead of `uuid.uuid4()` here.",
+      "reference": "Can we add an assertion here the policy actually contains both statements created above?"
+    },
+    {
+      "prediction": "`disable` is not a valid value for `rocksdb_bottommost_compression`.",
+      "reference": "a space after comma?"
+    },
+    {
+      "prediction": "I think this should be `if (skewFromParent != null && !isLocalSpan(node.value()))`",
+      "reference": "nit formatting here and below"
+    },
+    {
+      "prediction": "nit: I think we can remove the `v1` from the import path.",
+      "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 `return false, \"\", fmt.Errorf(\"not implemented\")`",
+      "reference": "`ctx` is unused in RunQuery"
+    },
+    {
+      "prediction": "This should be `report.files.get(event['location']['file'], '')` or `report.files.get(event['location']['file']) or ''`.",
+      "reference": "Couldn't we use type hints to express this?"
+    },
+    {
+      "prediction": "This should be `icon.classList.add('material-icons favorite');`",
+      "reference": "why you do `.classList.add()` here but `.replace()` in other places?"
+    },
+    {
+      "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": "What is this sleep here for?"
+    },
+    {
+      "prediction": "I think this test should be moved to `autoRowSize.spec.js` file.",
+      "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": "I'm not sure this is the best way to do this, but I think it's the best we can do for now. We need to make sure that we don't try to access `unflushedPaths` while it's being populated, and we need to make sure that we don't try to access `unflushedPaths` after it's been cleared.",
+      "reference": "Maybe this should just be passed in on construction time"
+    },
+    {
+      "prediction": "I don't think this is correct. The monitor should be destroyed when the backend is destroyed, not when the multi-backend is destroyed.",
+      "reference": "We should create the monitor after the `if (!primary_drm)` check."
+    },
+    {
+      "prediction": "I don't think this should be whitelisted. It's a private method, and it's not clear to me why it needs to be whitelisted.",
+      "reference": "Can be removed now that slots are used"
+    },
+    {
+      "prediction": "I think this should be `must_respond_with :not_found`",
+      "reference": "The description should match the test"
+    },
+    {
+      "prediction": "I think this should be `memset(wlr_xwayland->display_name, 0, sizeof(wlr_xwayland->display_name));`",
+      "reference": "rootston needs to unset this now"
+    },
+    {
+      "prediction": "I think this should be `Delete` instead of `Pop`.",
+      "reference": "why do you rename this function?"
+    },
+    {
+      "prediction": "Style/RedundantElse: Redundant else detected.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br",
+      "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 set the `RestInvocationContext` in the `RestProducerInvocation`. The `RestProducerInvocation` should not know anything about the `RestInvocationContext`. The `RestInvocationContext` should be set in the `RestConsumerInvocation` and passed to the `RestProducerInvocation` as a parameter. The `RestProducerInvocation` should not know anything about the `RestInvocationContext`. The `RestInvocationContext` should be set in the `RestConsumerInvocation` and passed to the `RestProducerInvocation` as a parameter. The `RestProducerInvocation` should not know anything about the `RestInvocationContext`. The `RestInvocationContext` should be set in the `RestConsumerInvocation` and passed to the `RestProducerInvocation` as a parameter. The `RestProducerInvocation` should not know anything about the `RestInvocationContext`. The `RestInvocationContext` should be set in the `RestConsumerInvocation` and passed to the `RestProducerInvocation` as a parameter. The `RestProducerInvocation` should not know anything about the `RestInvocationContext`. The `RestInvocationContext` should be set in the `RestConsumerInvocation` and passed to the `RestProducerInvocation` as a parameter. The `",
+      "reference": "both edge and normal need to set this, so set it to be default action?"
+    },
+    {
+      "prediction": "Is there a reason to add a newline at the end of the file?",
+      "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 don't think you need the `{}` here.",
+      "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 `WLR_INPUT_DEVICE_TABLET_PAD` instead of `WLR_INPUT_DEVICE_TABLET_TOOL`.",
+      "reference": "Hmm, why is this needed? `wlr_input_device_destroy` should destroy the tablet tool."
+    },
+    {
+      "prediction": "I don't think we need to support this for now.",
+      "reference": "I think it's better to throw `UnsupportedOperationException`"
+    },
+    {
+      "prediction": "Shouldn't this be `if (Array.isArray(props.value) && props.multiple && type === 'select')`?",
+      "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 don't think we need this line anymore.",
+      "reference": "Please just remove this line. We have the git history if we want the old version."
+    },
+    {
+      "prediction": "I think this should be `SpinScriptEnvResolver.class.getSimpleName()`",
+      "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 we can remove this example, since it's just a copy of the one in the `runtimevar` package.",
+      "reference": "Remove the handling here as well."
+    },
+    {
+      "prediction": "I think we should use `int.MaxValue` instead of `50000` here.",
+      "reference": "Do we want to make this configurable?"
+    },
+    {
+      "prediction": "I think this should be `> 0` instead of `>= 0`.",
+      "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": "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": "Is this going to break things for 18F proposals, or will these fields just be ignored?"
+    },
+    {
+      "prediction": "I don't think we need to add COMMAND here. It's not a modifier key, it's a command key.",
+      "reference": "Keys.COMMAND seems to be an alias to Keys.META. That isn't mentioned?"
+    },
+    {
+      "prediction": "I think this should be `omitempty` instead of `omitemptuy`",
+      "reference": "What's the reason for moving this to config from status?"
+    },
+    {
+      "prediction": "I think this should be `return False` instead of `return True`.",
+      "reference": "Shouldn't we re validate/update the in memory suppress data here?"
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it should be `fixtures.TestAddresses[0]` instead of `fixtures.TestAddresses[2]`.",
+      "reference": "curious why the renumbering is required here and elsewhere?"
+    },
+    {
+      "prediction": "I think this should be `allErrs = append(allErrs, field.Invalid(awsPath.Child(\"credentialsAssumeRole\"), aws.CredentialsAssumeRole, \"cannot specify assume role when credentials secret is provided\"))`",
+      "reference": "s/must specify secrets/must specify secrets or Role info/"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but it seems like it should be here.",
+      "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 \"AlwaysOffActivitySampler\".",
+      "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 thing to do. If the replacement is empty, I think we should remove the header. What do you think?",
+      "reference": "This applies only in the \"set\" headers case, but what about \"adding\" a header (field prefixed with `+`)?"
+    },
+    {
+      "prediction": "I think this should be `@Deprecated(since = \"7.0.0\")`",
+      "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 `MarkNotSetError` instead of `MarkNotSetError`.",
+      "reference": "You'll need to adjust the docstring :wink:"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Add a TODO explaining why this is being done"
+    },
+    {
+      "prediction": "I think we can remove these two lines.",
+      "reference": "This file still uses Blockly.VariableModel..."
+    },
+    {
+      "prediction": "I think this should be `Public bool `json:\"public,omitempty\"` and not `Public bool `json:\"public,omitempty\"`",
+      "reference": "Small nit, in the PR description its `publicAccess` but here its `public`. Guessing the preferred naming is public?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be good to add a note here that the `Type` should be a pointer type. This is because the `reflect.ValueOf` call in `Decode` will return a `reflect.Value` that is a pointer to the object, and if the `Type` is not a pointer type, then the `reflect.ValueOf` call will return a `reflect.Value` that is a copy of the object, and the `reflect.ValueOf` call in `Decode` will not be able to set the value of the `reflect.Value` returned by the `reflect.ValueOf` call in `Decode` to the value of the `reflect.Value` returned by the `reflect.ValueOf` call in `Decode`, and the `reflect.ValueOf` call in `Decode` will not be able to set the value of the `reflect.Value` returned by the `reflect.ValueOf` call in `Decode` to the value of the `reflect.Value` returned by the `reflect.ValueOf` call in `Decode`, and the `reflect.ValueOf` call in `Decode` will not be able to set the value of the `reflect.Value` returned by the `reflect.ValueOf",
+      "reference": "Maybe `driver.Watcher` -> provider (2x)? This is the concrete type, this user doesn't really know anything about the driver."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "can we remove this since `getDocValues` already returns an iterator? (we might need to do `T extends DocIdSetIterator` above)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Don't check this in. This is already implemented in core_extensions/ruby/string.rb"
+    },
+    {
+      "prediction": "I think this should be `writable` instead of `enumerable`.",
+      "reference": "This is used for both `id`, and `hash` we may not want these both to be enumerable."
+    },
+    {
+      "prediction": "I think this should be `@return [Array<Formatter>] the formatters currently loaded`",
+      "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 we should add `ai.Version` to `GetActivityScheduledEvent` as well.",
+      "reference": "@wxing1292 to double check this is the right version to use?"
+    },
+    {
+      "prediction": "Shouldn't this be `($this->min_bound === null && $this->max_bound === null) || ($this->min_bound === null && $this->max_bound >= $i) || ($this->max_bound === null && $this->min_bound <= $i) || ($this->min_bound <= $i && $this->max_bound >= $i)`?",
+      "reference": "Shouldn't there be another case for when both min and max are specified?"
+    },
+    {
+      "prediction": "I think this should be 2014, not 2013.",
+      "reference": "Don't miss this one `2: I102 copyright year is outdated, expected 2014 but got 2013`. Thanks"
+    },
+    {
+      "prediction": "I think you can remove the `if (tail_)` check here and just do `str.append(tail_, blockPt_);` since `blockPt_` will be 0 if `tail_` is null.",
+      "reference": "the tail_ maybe nullptr?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do for now.",
+      "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. [85/80]",
+      "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 10s or 20s. 1s is way too short.",
+      "reference": "LGTM, curious though why exactly this is in place, and are there any possible issues from not waiting long enough?"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "?: is unnecessary because we already force preserve_hosts to be a string with to_s."
+    },
+    {
+      "prediction": "I think this should be `/se/grid/newsessionqueuer/queue`",
+      "reference": "should it be `newsessionqueue` or `newsessionqueuer`? In case we'd like to be consistent"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it seems to work.",
+      "reference": "Since lazy is re-rendered when mounting the fallback, these checks get triggered an additional time."
+    },
+    {
+      "prediction": "nit: `is_content_editable_prop` -> `is_content_editable_property`",
+      "reference": "There is a `is_content_editable()` method just above this, what's the difference?"
+    },
+    {
+      "prediction": "I don't think this is a warning, it's a deprecation.",
+      "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": "I'm not sure this is the right thing to do, but I don't have a better idea. The problem is that `flux_subprocess_get_cmd (p)` returns a pointer to a static buffer that is overwritten by the next call to `flux_subprocess_get_cmd (p)`. So we need to make a copy of the command string before the next call to `flux_subprocess_get_cmd (p)`.",
+      "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 this should be `plusMinutes(1)`.",
+      "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 it's better to use `LOG_IF(WARNING, code != kvstore::ResultCode::SUCCEEDED) << \"put failed, ret \" << static_cast<int32_t>(code);`",
+      "reference": "what to do if some failed? just logging?"
+    },
+    {
+      "prediction": "I think it would be nice to have an example for this one as well.",
+      "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\"`",
+      "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": "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.<br>",
+      "reference": "Extra empty line detected at class body beginning."
+    },
+    {
+      "prediction": "I don't think we need to intern the key here.",
+      "reference": "This is smart but hacky! I would probably prefer an alternate solution that would be more obvious to read/understand."
+    },
+    {
+      "prediction": "I think this should be `*dynamodbv1.TableStatus` instead of `*dynamodbv1.Status`.",
+      "reference": "i would not return a pointer for status here which is an int"
+    },
+    {
+      "prediction": "I don't think we need to exclude all of these. We can just exclude the ones that don't have a `__init__.py` file.",
+      "reference": "Those functions are not currently wrapped, so I don't think it is necessary to exclude them (at least for now)."
+    },
+    {
+      "prediction": "I think this should be `The folder %{folder} was already added`",
+      "reference": "Should we specify the name of the folder?"
+    },
+    {
+      "prediction": "I don't think we need to import `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 we can remove this now.",
+      "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 `default:\"debug\"`",
+      "reference": "Out of the scope."
+    },
+    {
+      "prediction": "I don't think you need to use `simulate` here. You can just call `element.trigger('mousedown')` and `element.trigger('mouseup')`.",
+      "reference": "Maybe `sortByColumnHeader` would be more precise? The plugin has a method with the same name and it could be confusing."
+    },
+    {
+      "prediction": "`refs` is not used in this function, so it can be removed.",
+      "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 think this should be `newOp = &resolutionOp{}` instead of `newOp = op`.",
+      "reference": "Remind me why this is needed?"
+    },
+    {
+      "prediction": "I think this should be `BooleanClause.Occur.MUST`",
+      "reference": "same as above, can this be null now?"
+    },
+    {
+      "prediction": "I don't think we need to cast `e` to `int` here. `e` is already an `int`.",
+      "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'm not sure if this is the best way to do this, but it seems to work for now.",
+      "reference": "str.Substring(3).Split(' ')[0] would also work and is more concise."
+    },
+    {
+      "prediction": "I think this should be `node_l->config.allow_local_peers && node_l->peers.not_a_peer (endpoint, true)`",
+      "reference": "Did you mean to negate the not_a_peer check?"
+    },
+    {
+      "prediction": "Please revert this change. It's not related to this PR.",
+      "reference": "Copyright (C) 2010-2021 DBeaver Corp and others"
+    },
+    {
+      "prediction": "This import should be moved to the top of the file.",
+      "reference": "This seems like an unrelated change"
+    },
+    {
+      "prediction": "I think this should be `defer func() { label.SetKeyLabel(l.config.ProcessLabel) }()`",
+      "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": "I think this should be in the `SharedMemTransportDescriptor` class.",
+      "reference": "I thought we were removing the FAIL policy altogether...."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. The `txt2tags` command line tool doesn't have an option to specify the output file name, so this will always overwrite the original file.",
+      "reference": "Could you elaborate on why you made this change?"
+    },
+    {
+      "prediction": "nit: please add a space after `\\\"` and before `\\\"`",
+      "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 should be `if !c.Data[\"IsApiToken\"].(bool) {`",
+      "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 think this should be `applyMode == hivev1.UpsertResourceApplyMode || applyMode == \"\"`",
+      "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 don't think we should enable sm2 by default.",
+      "reference": "add an entry in makefile to generate binary with diff name?"
+    },
+    {
+      "prediction": "I think this should be `if err := conf.HTTPConfig.Validate(); err != nil { return errors.Wrap(err, \"HTTPConfig validation failed\") }`",
+      "reference": "hmm so we didn't validate the HTTP config for all receivers?"
+    },
+    {
+      "prediction": "I think this should be `bool TryExtract<T>(T carrier, Func<T, string, IEnumerable<string>> getter, [NotNullWhen(true)] out SpanContext? spanContext);`",
+      "reference": "ITextFormatActivity had this, can I add to have the same effect?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `containerresource` prefix here and in the `containerstatus` import above.",
+      "reference": "not blocking: can we remove extra lines here?"
+    },
+    {
+      "prediction": "`data === 'random'` should be `data === 'random' || typeof(data) === 'undefined'`",
+      "reference": "This just feels weird... can we use some kind of constant here? something like: `if (_.isUndefined(data) || data === PrivateKey.Random)`"
+    },
+    {
+      "prediction": "I think we should throw an exception if the type is not a class.",
+      "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 if this is the right thing to do, but I think it makes sense to use a `Set` instead of a `List` here, because the order of the fields in the field mask doesn't matter.",
+      "reference": "LOL: use your GitHub username?"
+    },
+    {
+      "prediction": "I think it's better to use `next` instead of `nextPrefix` to avoid confusion.",
+      "reference": "any tests for nextPrefix() ?"
+    },
+    {
+      "prediction": "I think this should be `n.getMaximumCommonType().ifPresent(t -> t.accept(this, arg));` instead of `n.getMaximumCommonType().ifPresent(t -> t.accept(this, arg));`",
+      "reference": "In general I'm saying \"if there is no common type I'm just going to skip some stuff and fail silently.\""
+    },
+    {
+      "prediction": "nit: I think we can remove this import.",
+      "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 think we should keep the `operation` field. It is used to set the operation name on the span.",
+      "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 don't think this is correct. The source IP should be the IP address of the interface on which the packet is being sent.",
+      "reference": "Can you merge #466? That would make the current `None` default value working and more relevant that `\"127.0.0.1\"`."
+    },
+    {
+      "prediction": "I don't think this is correct. The error message should be something like \"Invalid value for search attribute `<saName>`. Expected `<expectedType>`, but got `<actualType>`.\"",
+      "reference": "is reserved filed name the only possible reason this could fail?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the env var isn't set, we should probably default to `false`.",
+      "reference": "Would a KBFS_DISABLE_JOURNAL make more sense?"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this message.",
+      "reference": "BTW. What's the message meaning?..."
+    },
+    {
+      "prediction": "Use 2 (not 4) spaces for indentation.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.<br>Space inside { missing.<br>Space inside } missing.<br>Prefer single-quoted strings when you don't need",
+      "reference": "this should say false"
+    },
+    {
+      "prediction": "I think we can remove this file now.",
+      "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'm not sure this is the right place to do this. I think it should be done in the `open_url` function.",
+      "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": "This class is not used anymore, so it can be 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 `GrapheneMaterializationOrObservation`",
+      "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": "Why is this change necessary?",
+      "reference": "Can also cast them as `set`s and compare them."
+    },
+    {
+      "prediction": "I think this should be `package trigger`",
+      "reference": "Still TODO I guess?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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 don't understand why this change is needed. Can you explain?",
+      "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": "I don't think this is needed.",
+      "reference": "Is this necessary?"
+    },
+    {
+      "prediction": "Is there a reason to have a protected constructor?",
+      "reference": "How about this NS inside query NS?"
+    },
+    {
+      "prediction": "I think this should be `protocol.StateReader` instead of `protocol.StateManager`",
+      "reference": "in line 79, sm also can be stateReader"
+    },
+    {
+      "prediction": "It's better to use `std::unordered_map<HostAddr, std::vector<PartitionID>>` instead of `std::map<HostAddr, std::vector<PartitionID>>`.",
+      "reference": "maybe can not return a map directly, it will be modified by inner thread."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it is.",
+      "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. I think it would be better to do this in `RGroupDecomposition::RGroupDecomposition(const ROMol &inmol, const RGroupDecompositionOptions &options)`.",
+      "reference": "Might be better to use a tag here. I for one have used 1000 isotopes as a tag in the past..."
+    },
+    {
+      "prediction": "I think this should be `if span.IsRecording() == true {`",
+      "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 call this `isShowVersion` or `isShowVersionRequested` or something like that. `isVersion` sounds like it's about the version of PMD, not about the `--version` command line option.",
+      "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 an error, not a debug message.",
+      "reference": "I'm worried about we hide some important log here"
+    },
+    {
+      "prediction": "I think this should be 365 * 24 * 60 * 60 * 1000",
+      "reference": "Python 3 will throw an exception for inf, but not providing a timeout will result in no timeout exception."
+    },
+    {
+      "prediction": "`/bzz-tag/uuid/{uuid}` should be `/bzz-tag/{uuid}`",
+      "reference": "As this method changes the state, it should be `POST`. Also, maybe to rename it to `createTag`?"
+    },
+    {
+      "prediction": "Shouldn't this be `model.StageStatus_STAGE_SUCCESS`?",
+      "reference": "`ctx` is unused in rollbackTraffic"
+    },
+    {
+      "prediction": "Is this used anywhere? If not, can we remove it?",
+      "reference": "Roslyn conventions have const in PascalCase."
+    },
+    {
+      "prediction": "I don't think this is correct. If the client port is null, then it should be 2181. If the client port is not null, then it should be the value of the client port.",
+      "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 think we can remove the `Datadog.Trace.Vendors.MessagePack.Formatters` namespace and just use `Datadog.Trace.Vendors.MessagePack`",
+      "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": "Is this a typo?",
+      "reference": "Aren't you duplicating this option?"
+    },
+    {
+      "prediction": "I think this should be `strings.HasSuffix(base, \".build_defs\") || strings.HasSuffix(base, \".build_defs.bzl\")`",
+      "reference": "I'm not sure we should be doing this based on the extension? Calling them `.build_defs` is just a convention"
+    },
+    {
+      "prediction": "I think we should keep `T` as it is used in `__getitem__` and `__setitem__`.",
+      "reference": "Should remove in `_MissingPandasLikeMultiIndex:` too"
+    },
+    {
+      "prediction": "I think this should be `+kubebuilder:validation:MinItems=1` and `+kubebuilder:validation:MaxItems=20`",
+      "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 `+kubebuilder:validation:Enum=us-east-1,cn-northwest-1`",
+      "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 should be fixed rather than skipped.",
+      "reference": "Stephen already fixed this one. Only the OverSize test is flaky now right?"
+    },
+    {
+      "prediction": "I don't think we need the `@activity` variable here.",
+      "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'm not sure this is the right place to do this, but I'm not sure where else to do it.",
+      "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 return an empty list instead of null.",
+      "reference": "Is Java smart enough to know the generic type here? Never knew this."
+    },
+    {
+      "prediction": "I don't think we need to pass the `PluginInfo` to the `newInstance` method. We can just pass the `className` and `initArgs` to the `newInstance` method.",
+      "reference": "no thought given to reload"
+    },
+    {
+      "prediction": "I think this should be `!disabledMetricAttributes.contains(attribute)`",
+      "reference": "Please, don't remove the `getDisabledMetricAttributes` method. All changes must be backward-compatible."
+    },
+    {
+      "prediction": "I think we can remove the `err.message.match(/no such (cmd|command)/)` check as it is already covered by the `err.code === 11601` check.",
+      "reference": "Is this left over from debugging?"
+    },
+    {
+      "prediction": "I think this should be `done` instead of `client.close(done)`. `done` is a function that needs to be called when the test is done. `client.close(done)` will call `done` when the client is closed, which is not necessarily when the test is done.",
+      "reference": "I've fallen for this before myself :) We _do_ call `close` inside of `done` on L65"
+    },
+    {
+      "prediction": "I think this should be `rclone/%s` where `%s` is the version.",
+      "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 don't think we need to do this. `torch.clamp(neg_prob, min=1e-12)` should be enough.",
+      "reference": "Adding one line `prob = prob.clamp(min=EPS, max=1-EPS)` already works."
+    },
+    {
+      "prediction": "I think this should be `AsyncRequestQueue` instead of `ConcurrentRequestQueue`",
+      "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 `&sibling->parent_pending_link.prev`.",
+      "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 2. there with blank line.`",
+      "reference": "why did this change?"
+    },
+    {
+      "prediction": "I think this should be `tlf.FakeID(1, tlf.Private)`",
+      "reference": "Do you think we should have some tests here for `tlf.SingleTeam` too?"
+    },
+    {
+      "prediction": "I don't think we need to import the `table` package in this file.",
+      "reference": "No problem with this but might as well revert this file since nothing else changed."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I don't have a better idea at the moment.",
+      "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(message);` instead of `throw new Error(message); 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 think it would be better to have `addEvent(String name, Map<String, AttributeValue> attributes)` and `addEvent(String name, long timestamp, Map<String, AttributeValue> attributes)`",
+      "reference": "We don't need this additional method."
+    },
+    {
+      "prediction": "I don't think we need to mock `DefaultOAuth2UserService` here.",
+      "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": "I don't think we need to change the name of the variable here.",
+      "reference": "nit: rename to nativeStk? proto might lead to think protobuf"
+    },
+    {
+      "prediction": "I don't think this should be part of the ShardParams interface.",
+      "reference": "Why not `queryUUID` (and reference the same constant as in other places)?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I don't think we should change the casing of the first letter of the first word in the string.",
+      "reference": "This is also used to log, of which our convention is leading uppercase..."
+    },
+    {
+      "prediction": "I think we can remove this `less_precise` argument.",
+      "reference": "What about we just name it `check_exact`?"
+    },
+    {
+      "prediction": "I think it would be good to have a test for the case where there are no endpoints in either annotations or binary annotations.",
+      "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": "Style/TrailingWhitespace: Trailing whitespace detected.<br>Layout/TrailingWhitespace: Trailing whitespace detected.",
+      "reference": "Where does the `7` come from?"
+    },
+    {
+      "prediction": "I'm not sure why this is necessary. The `setValues` function doesn't depend on `value`, so I don't see why it needs to be included in the dependency array.",
+      "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 this should be `ALTER TABLE ... SET TBLPROPERTIES ('sort-order'='...')`",
+      "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": "I think we can remove the `initialState` import here, as it's not used in this 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": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "This was used in `app/views/video_tutorials/_video_tutorial_details.html.erb` how are we handling that now?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be nice to have a `github.com/pulumi/pulumi/pkg/util/regex` package that contains all of these common regexes.",
+      "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 don't think we need to test these anymore.",
+      "reference": "This block of changes looks like you removed tests without replacing them anywhere. Did I miss something?"
+    },
+    {
+      "prediction": "I think this should be `python::arg(\"includePrivate\") = false` and `python::arg(\"includeComputed\") = false`",
+      "reference": "n.b. private and computed values are now exposed to the API. They were hidden/not exposed before."
+    },
+    {
+      "prediction": "I think this should be `if self._elem.tagName().lower() == 'a':`",
+      "reference": "Nitpick: Please add a period after `link` :wink:"
+    },
+    {
+      "prediction": "Shouldn't this be `if err != nil`?",
+      "reference": "we got no chance that `err` is not nil here, need to revisit how `NewUpstreamController` is defined."
+    },
+    {
+      "prediction": "I don't think we need to add this to the `nameToUrl` map. It's only used in the `execute` method, which doesn't seem to use the `nameToUrl` map at all.",
+      "reference": "Instead of building in routing for a browser-specific command, could you refactor the HttpCommandExecutor to allow arbitrary commands to be registered?"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "It'd be nice to assert the right deprecation is being raised, just add a third argument of `/\"its\" method/`"
+    },
+    {
+      "prediction": "I think this should be `write.location-provider.impl` instead of `write.object-storage.location-provider.impl`.",
+      "reference": "Do we want to add this to the docs?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to compare the string representations of the codes, rather than the codes themselves. This is because the `codes.Code` type is an `int32`, and the `assert.Equal` function doesn't know how to compare two `int32`s.",
+      "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": "Shouldn't this be `if (size && *size == 0)`?",
+      "reference": "Because of this new early exit."
+    },
+    {
+      "prediction": "I don't think we need to list all the bridges here.",
+      "reference": "Maybe call it `WireBridge`?"
+    },
+    {
+      "prediction": "I don't think we need to check for `privacyParameters.getGoQuorumPrivacyParameters().isPresent()` here. If it's not present, then `privacyParameters.isEnabled()` will be false and we will return a `DisabledPrivacyRpcMethod`.",
+      "reference": "can this be removed now?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If you have two processes with the same UID and PID, you're going to have a problem.",
+      "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 don't think we need this constant. We can just use `fmt.Sprintf(AppCfnTemplateConfigurationNameFormat, app.Name, app.Environment.Name)`.",
+      "reference": "What do you think about moving this to the `cloudformation` pkg?"
+    },
+    {
+      "prediction": "I think this should be `protected $clones = [];` instead of `protected $clones = null;`",
+      "reference": "Please make this private by default"
+    },
+    {
+      "prediction": "This import is not used.",
+      "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 we should rename this to `LocationCache` to make it more clear that it's a cache for locations.",
+      "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 think this should be `except optparse.OptionConflictError as e:`",
+      "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 think this should be `if status == Stopped && !c.cgroupManager.Exists() && !c.cgroupManager.IsSystemdCgroup() { return pids, nil }`",
+      "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.name()` here and just pass `identifier` to the `createTableTransaction` method.",
+      "reference": "Did you mean to use `identifier.toString` here as well?"
+    },
+    {
+      "prediction": "Extra empty line detected at block body end.",
+      "reference": "interesting! didn't realize this was needed in RSpec 3"
+    },
+    {
+      "prediction": "nit: I think it's better to use `ks.Series` instead of `'ks.Series'` here.",
+      "reference": "This is to avoid circular reference"
+    },
+    {
+      "prediction": "I think this should be `@property` and `@notebook.setter`.",
+      "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": "I don't think we need to remove the MongoError import here.",
+      "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": "I don't think we need to pass in the full path and filename separately. We can just pass in the full path and then extract the filename from it.",
+      "reference": "Mirror the `OPTIONAL OUT` of fullpath"
+    },
+    {
+      "prediction": "I think this should be `ConsensusVTEALv1` or `ConsensusVTEALv1LogicSig` or something like that.",
+      "reference": "Please merge with master and replace with with \"future\" version."
+    },
+    {
+      "prediction": "I'm not sure why this line was removed, but it seems like it should be there.",
+      "reference": "Nit: Change is not needed for this PR :slightly_smiling_face:"
+    },
+    {
+      "prediction": "I think this should be `self.vcap_data['application_uris'].min_by(&:length)` instead of `self.vcap_data['application_uris'].min_by(&:length)`",
+      "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": "I don't think we need to change the name of the class.",
+      "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'm not sure this is the right thing to do. If the file doesn't exist or is empty, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for some other reason, we should return false. If the file can't be accessed for",
+      "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": "Shouldn't this be `wsc.connection.SetWriteDeadline(time.Now().Add(wsc.config.WriteDeadline))`?",
+      "reference": "Looks good, but seems `ReadMessage` hadn't used this Deadline in Underlying `WSConnection`?"
+    },
+    {
+      "prediction": "I think this should be in `ProjectProperties` instead of `Constants`.",
+      "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 `private` instead of `final private`.",
+      "reference": "We don't need this anymore after #2151"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "this is this removed? I don't understand in the context of this PR"
+    },
+    {
+      "prediction": "This should be `ip6 = next(iter(ip6))`",
+      "reference": "`isinstance(ip6, Net6)` instead"
+    },
+    {
+      "prediction": "`std::vector<Status>` -> `std::vector<StatusOr<std::string>>`",
+      "reference": "Do we really need the param here? If only partial results returned, we could return ResultCode::PARTIAL_RESULTS;"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do. If the output name is the same as the package name, then we want to make sure that we don't accidentally overwrite the package directory with the output file.",
+      "reference": "this logic seems duplicated from the other new function. I think it should probably be a member function on `BuildTarget`"
+    },
+    {
+      "prediction": "Why did this change?",
+      "reference": "size reduce to 1/4 of using Gob"
+    },
+    {
+      "prediction": "I think this should be `h.h.Write(p[:n])` instead of `h.h.Write(p)`.",
+      "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 don't think we need the `2>/dev/null` here.",
+      "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": "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": "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 don't think we need to do this. `ConstantScoreWeight` is only used by `ConstantScoreQuery` and `ConstantScoreQuery` doesn't support `ScoreMode.TOP_DOCS` or `ScoreMode.TOP_DOCS_WITH_SCORES`.",
+      "reference": "maybe add a `isExhaustive()` method on the enum to avoid these large conditions?"
+    },
+    {
+      "prediction": "I don't think we need to make these constants public.",
+      "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 this should be `if (empty($params) && null === $source && null !== $method) {`",
+      "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'm not sure this is the right place to set `binary` and `compressed`. I think it should be done in the `open_file` method of each derived class.",
+      "reference": "Should we add error info for not supporting the compressed binary?"
+    },
+    {
+      "prediction": "nit: I think this should be `var clusterRegionGetter = utils.NewClusterRegionGetter()`",
+      "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 don't think we need to create a new instance of `JsonSerializer` every time we need to serialize or deserialize something. We can create a static instance of `JsonSerializer` and reuse it.",
+      "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.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 about this change. I think it should be `separator = \"#{'-' * 80}\"`",
+      "reference": "Funny that we were wrapping this with string interpolation before..."
+    },
+    {
+      "prediction": "`guid` should be `model.guid`",
+      "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 if this is a breaking change or not. I think it is, but I'm not 100% sure. @stephentoub can you confirm?",
+      "reference": "unsure if the methodname can still be `GetMetricPoints()` as before..."
+    },
+    {
+      "prediction": "I think this should be `rule_id` instead of `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 should be done in the `SmartStoreTestCase.tearDown()` method, which is called after each 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 we need to bump the model version for this change.",
+      "reference": "Is new model format backward compatible with current v2?"
+    },
+    {
+      "prediction": "I think this should be `expect { get '/auth/myusa/callback' }.to change { ActionMailer::Base.deliveries.length }.from(0).to(1)`",
+      "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 call this `ECSStopTimeoutSeconds` or `ECSContainerStopTimeoutSeconds` to make it clear that it only applies to ECS containers.",
+      "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 this import is needed.",
+      "reference": "JMX offers normal APIs for this. I don't think you want the glassfish one."
+    },
+    {
+      "prediction": "I think this should be `name: yamllint` instead of `name: ...`",
+      "reference": "We should really leave a TODO or something or else we will forget them ..."
+    },
+    {
+      "prediction": "I think this should be `runtime_type=DataFrame` instead of `dagster_type=DataFrame`",
+      "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 don't think this is necessary. If the selection is not anchored at the start, then it must be anchored at the end.",
+      "reference": "It might help here if you give a brief explanation of what you mean with a selection being anchored at the start."
+    },
+    {
+      "prediction": "remove unused import",
+      "reference": "move cachedStates to handle function"
+    },
+    {
+      "prediction": "I think this should be `privacyMode` instead of `keyguardPrivacy`",
+      "reference": "I don't think this can just be renamed; it's saying that privacyMode is a BooleanSetting, which it's not."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should be using the thread id. The thread name is not guaranteed to be unique and can be changed at any time.",
+      "reference": "what was the reason for this change?"
+    },
+    {
+      "prediction": "Why are these removed?",
+      "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/client-go/pkg/api/v1` instead of `k8s.io/apimachinery/pkg/apis/meta/v1`",
+      "reference": "I think I would just pass in `len(peers)`, on the principle of minimum information."
+    },
+    {
+      "prediction": "I think this should be `read.parquet.vectorization.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 we need to add the `@author` here.",
+      "reference": "Please update copyright year."
+    },
+    {
+      "prediction": "I think it would be better to use `std::chrono::duration_cast<std::chrono::seconds>(expansion).count()` instead of `TIMER_SEC(expansion)`.",
+      "reference": "Same here, could just be `std::setprecision`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to just remove this code for now. We can always add it back later if we need it.",
+      "reference": "Do we still want the \"admins\" gate?"
+    },
+    {
+      "prediction": "`Amazon Simple Storage Solution (Amazon S3)` -> `Amazon Simple Storage Service (Amazon S3)`",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I think this should be `EnableAutoObservabilityMiddleware`",
+      "reference": "nit: `it is assumed`"
+    },
+    {
+      "prediction": "I think this should be `#include \"h2o/http2_internal.h\"`",
+      "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 `searchScope` instead of `manage[searchScope]`",
+      "reference": "@danielbidala What happens if `searchScope` is not defined in `config_relation.yaml` ?"
+    },
+    {
+      "prediction": "I don't think we need to launch these in the background, since we're already in a background goroutine.",
+      "reference": "Do we want maybe a 1min timeout?"
+    },
+    {
+      "prediction": "I think this should be `if len(a) != 2 {`",
+      "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 don't think we need to make this constructor public.",
+      "reference": "Doesn't this no-arg constructor need to be public for serialization to work?"
+    },
+    {
+      "prediction": "`Hierdie gebied is reeds aangeheg.` -> `Hierdie veld is reeds aangeheg.`",
+      "reference": "Just use the word \"veld\" instead of \"gebied\" here"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "This change (import org.shredzone.flatter4j.model.User;) does not seem relevant to this fix / commit."
+    },
+    {
+      "prediction": "I think this should be a `Bool()` instead of a `String()`.",
+      "reference": "This should be a Bool() rather than a String(). (See the other PR for an example)"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "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 don't think it's a good idea to have `SyncMode` as a property of `SyncState` class. I think it's better to have `SyncState` as a property of `SyncMode` class. What do you think?",
+      "reference": "We need to switch from BeamSync to FullSync when we download all the needed headers, blocks, receipts and state"
+    },
+    {
+      "prediction": "Shouldn't this be `g_value_get_object (value)`?",
+      "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": "I don't think this is the right way to get the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be the status of the chaos experiment, not the status of the chaos instance. The status of the chaos instance should be",
+      "reference": "`in.Status.Experiment.Phase`. we can omit `ChaosStatus`"
+    },
+    {
+      "prediction": "I think we can remove this.",
+      "reference": "Just curious... why the alias here?"
+    },
+    {
+      "prediction": "This line is not needed.",
+      "reference": "Seems like this should conflict with changes I made during merging of the 20.2 RESX file translation"
+    },
+    {
+      "prediction": "I don't think we should enable this by default. I think it should be disabled by default.",
+      "reference": "Why disable this rule?"
+    },
+    {
+      "prediction": "I think this should be `arg.replace('{}', self._filename)` instead of `arg.replace('{}', self._filename) if '{}' in arg else arg`",
+      "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 1000 or 10000. 20 is way too small.",
+      "reference": "nit: Should we make this an attribute of the LoadGroupMembersPIpeline class?"
+    },
+    {
+      "prediction": "I don't think we need this constant.",
+      "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 to call `determineNonce`. I think it should be called in `PrivateTransactionHandler` and the result should be passed to `getSenderNonce` and `getRecipientNonce`.",
+      "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 don't think this is needed anymore.",
+      "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 don't think this is correct.",
+      "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": "This should be `msrest>=0.2.0, 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 `public static boolean hasAnyResourceNameParameter(FlatteningConfig flatteningConfig) {`",
+      "reference": "This was copied from JavaMethodViewGenerator; only the `public static` method modifiers were added."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this method. I think it would be better to have it in the `Transaction` interface and have the `TransactionImpl` implement it.",
+      "reference": "this is going to throw for frontier transactions post-london"
+    },
+    {
+      "prediction": "I don't think this is a good idea to remove this line.",
+      "reference": "The catch clause is not needed. The finally clause alone is enough."
+    },
+    {
+      "prediction": "I don't think this is correct.",
+      "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": "Why is this change necessary?",
+      "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": "nit: I think we can remove the `te := api.TaskStateChange{...}` line and just pass `api.TaskStateChange{...}` to `AddStateChangeEvent`",
+      "reference": "Please use more meaningful names than `te`here and in other places."
+    },
+    {
+      "prediction": "I think this should be `ConfigFor(interface{}) (*PluginConfig, error)` so that we can return an error if the plugin is not managed by the catalog.",
+      "reference": "Perhaps this would be more idiomatic as `ConfigFor(interface{}) (*PluginConfig, bool)`?"
+    },
+    {
+      "prediction": "I think this should be `application/x-amz-json-1.0` instead of `application/x-amz-json-1.1`.",
+      "reference": "nit: We could import `APPLICATION_AMZ_JSON_1_1` from `constants.py` here."
+    },
+    {
+      "prediction": "I think this should be `Task<bool> ExecuteOnMainThreadAsync(Func<Task<bool>> action, bool maskExceptions = true);`",
+      "reference": "Should this not inherit from IMvxMainThreadDispatcher?"
+    },
+    {
+      "prediction": "Why do we need to pass `blockInterval` to `UpdateRound`?",
+      "reference": "line is 167 characters (from `lll`)"
+    },
+    {
+      "prediction": "Can you remove this line?",
+      "reference": "nit: consider adding a blank line between L15 and L16."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it would be nice to have a `ReadVectorSize<T>()` function that reads the size of a vector of type `T` from the file. This way, we don't need to know the size of `T` in order to read the size of a vector of type `T` from the file.",
+      "reference": "This is another bug fix: Without skipping the bytes of the vector this would read garbage data."
+    },
+    {
+      "prediction": "I think this should be `actions.CreatorPubKey` instead of `actions.Creation.PubKey`",
+      "reference": "why need this? i don't see Gen.CreatorPubKey being used?"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "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 am not sure if this is the right place to put this. I think this should be in the `pkg/version/version.go` file.",
+      "reference": "is this supposed to be `Desired`?"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I think it should be restored.",
+      "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": "`os.path.exists(folder) == False` is equivalent to `not os.path.exists(folder)`.",
+      "reference": "Is there too much spaces there? (should be 4 I think)"
+    },
+    {
+      "prediction": "I think this should be `private static final String REVISION = \"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": "Shouldn't this be `mixins` instead of `mixins`?",
+      "reference": "This needs to be removed since we removed the mixin."
+    },
+    {
+      "prediction": "why remove this?",
+      "reference": "Can `weight_` be deleted?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. The `design_config_pass` service should not depend on the `locale` parameter. The `locale` parameter should depend on the `design_config_pass` service.",
+      "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 would be better to have it in `Puppet::Util::Execution` or `Puppet::Util::Execution::Unix`.",
+      "reference": "The host object already has a nice way of querying configprint. Try `puppet('master')['user']`"
+    },
+    {
+      "prediction": "I think this should be `from mitmproxy.addons.upload import Upload`",
+      "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'm not sure why this is a dialog instead of a mini frame.",
+      "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 `IF_AARCH64_ELSE(md->emitted_size += fixup_indirect_trace_exit(dcontext, trace), md->emitted_size += fixup_indirect_trace_exit(dcontext, trace));`",
+      "reference": "So is this invoked every time we extend the trace?"
+    },
+    {
+      "prediction": "I don't think this is correct. The microsecond field should be between 0 and 999999, not 0 and 999.",
+      "reference": "Why not use switch here?"
+    },
+    {
+      "prediction": "`AuthFree` is not a good name. I think `Public` would be better.",
+      "reference": "I think `NoAuth` is a better name"
+    },
+    {
+      "prediction": "I think this should be `if err != nil { r.Error = awserr.New(eventstreamapi.InputWriterCloseErrorCode, err.Error(), r.Error) }` instead of `if err != nil { r.Error = awserr.New(eventstreamapi.InputWriterCloseErrorCode, err.Error(), r.Error) }`",
+      "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'm not sure why this was removed, but I think it should be restored.",
+      "reference": "Please keep this blank line. PEP8 says: > Method definitions inside a class are surrounded by a single blank line."
+    },
+    {
+      "prediction": "`params_for_search(params.merge(:per_page => count))` should be `params_for_search(params.merge(per_page: count))`",
+      "reference": "You are not passing in a `:params` key here."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Probably change \"ovs\" to \"OVS\"."
+    },
+    {
+      "prediction": "`$this->getRequest()->get($uniqueId)` should be `$this->getRequest()->query->get($uniqueId)`",
+      "reference": "Please linebreak this"
+    },
+    {
+      "prediction": "I don't think it's a good idea to make it nullable. If it's empty, it should be `default(T)`, not `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 this should be `wsAppManager` instead of `wsProjectManager`.",
+      "reference": "Maybe it's time to fix the param name for this interface?"
+    },
+    {
+      "prediction": "I think we should keep `ClientLibraryVersion`, `ClientFeatureVersion` and `ClientImpl` as they are useful for debugging.",
+      "reference": "Why are we removing it? Looks like useful info. Obviously, field names should change."
+    },
+    {
+      "prediction": "This should be `protected virtual IValueFormatter ValueFormatter => _valueFormatter ?? (_valueFormatter = Resolve<IValueFormatter>());`",
+      "reference": "Missing fallback to MessageTemplates.ValueFormatter.Instance"
+    },
+    {
+      "prediction": "I think this should be `axis=0` instead of `axis=1`.",
+      "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": "nit: I think we can remove this import.",
+      "reference": "This also runs a patch if its in the middle of one."
+    },
+    {
+      "prediction": "I think this should be `Debugf` instead of `Debug`.",
+      "reference": "Could we rename this method to `updateKnownStatus` ?"
+    },
+    {
+      "prediction": "I think we should only set the origin tag if it's not already set.",
+      "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 think we can remove the `networkingv1` import here, since it's not used.",
+      "reference": "nit: these imports were in a separate group to ones from this repo"
+    },
+    {
+      "prediction": "I think it would be better to have a single method `isPushMethodEnabled(String dispatchMethod)` that returns `true` if the dispatch method is either `PUSH` or `PUSH_CONTAINERIZED`.",
+      "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 this should be `numAttempts < maxAttempts` instead of `numAttempts <= maxAttempts`.",
+      "reference": "Don't penalize everyone with 250ms delay b/c some machines have problems."
+    },
+    {
+      "prediction": "I think this should be `log.error(\"Failed parsing 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": "I think this should be `if s == nil || !s.eventsEnabled() { return false }`",
+      "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 `Ember.Handlebars.Utils.escapeExpression(erroredEmails.join(', '))`",
+      "reference": "The `import` statement for `Ember` is missing in this file."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems to fix the issue for me.",
+      "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 should be `2012-1-1 12:00:00`",
+      "reference": "For testing some of the rounding functions"
+    },
+    {
+      "prediction": "I think this should be moved to `Version20180603135342` migration.",
+      "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 what you mean by \"any other kind of bytes we care about\". Can you elaborate?",
+      "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": "I don't think you need to lock the mutex here. It's only used in `hasReaderAttached()`, which is already protected by the mutex.",
+      "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": "`tab is not None` is redundant.",
+      "reference": "This fits on one line without the parentheses :wink:"
+    },
+    {
+      "prediction": "I'm not sure why we need 4 nodes for this test. 2 should be enough.",
+      "reference": "Looks like `TestCloudRecovery` changes are unrelated?"
+    },
+    {
+      "prediction": "Why do you remove this interface?",
+      "reference": "Try to separate it to different methods. Invoke is huge :)"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Looks like another spot that would benefit from `truncate`"
+    },
+    {
+      "prediction": "I think this should be `token = h2o__tokens + (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": "`no_docstring_rgx.pattern` should be `no_docstring_rgx`",
+      "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 user-read-currently-playing playlist-read-private playlist-read-collaborative playlist-modify-public playlist-modify-private user-follow-read user-follow-modify user-library-read user-library-modify user-read-email user-read-private user-read-birthdate user-read-playback-state user-modify-playback-state user-read-playback-position user-top-read user-read-recently-played playlist-read-private playlist-read-collaborative playlist-modify-public playlist-modify-private user-follow-read user-follow-modify user-library-read user-library-modify user-read-email user-read-private user-read-birthdate user-read-playback-state user-modify-playback-state user-read-playback-position user-top-read user-read-recently-played playlist-read-private playlist-read-collaborative playlist-modify-public playlist-modify-private user-follow-read user-follow-modify user-library-read user-library-modify user-read-email user-read-private user-read-birthdate user-read-playback-state user-modify-playback-state user-read-playback-position user-top-read user-read-recently-played playlist-read-private playlist-read-collaborative playlist-modify-public playlist-modify-private user-follow-read user-follow-modify user-library-read user-library-modify user-read-email",
+      "reference": "Should this be in a config file?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `driver` from the description.",
+      "reference": "purely preference, but have you considered using parentheses to indicate that a method is being tested? e.g., `describe('eachAsync()')`"
+    },
+    {
+      "prediction": "I think we should return `ERR_CHECKPOINT_ERROR` instead of `ERR_IO_ERROR` here.",
+      "reference": "Why not return success?"
+    },
+    {
+      "prediction": "`$linkFormat` should be `?string`",
+      "reference": "Please use snake_case for properties to keep it consistent with the rest of the codebase."
+    },
+    {
+      "prediction": "I don't think we need to log the address here.",
+      "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 it would be better to have a `@Before` and `@After` method to set and reset the clock.",
+      "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 think we should keep the `Pagination` component in the footer, even if there is no `footerText` to display.",
+      "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 don't think we need to make this enum public.",
+      "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 `dataset.interface.iloc(dataset.interface, (rows, cols))`",
+      "reference": "Had to fix this to get my tests passing (should have been a new PR sorry)."
+    },
+    {
+      "prediction": "I think this should be `CABundleBase64` to be consistent with the other `Base64` fields.",
+      "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 `False` by default.",
+      "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 correct. If both are set, the created date will be used.",
+      "reference": "Perhaps note here that `--drive-use-created-date` takes precedence if both set?"
+    },
+    {
+      "prediction": "I think it is better to use `eps=1e-15` as the default value of `eps`.",
+      "reference": "Use EPS=1e-15 as that in atss_head or FCOS head."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the only way to get this test to pass.",
+      "reference": "Is there a need to replace the Index with a MasterIndex?"
+    },
+    {
+      "prediction": "I think this should be `verifyExecAgentRunningStateChange(t, taskEngine, false)`",
+      "reference": "are we adding stop state change in a different PR?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [121/120]",
+      "reference": "How much difference is there between sort and sort_by ?"
+    },
+    {
+      "prediction": "I'm not sure why this is needed. I don't see it being used anywhere.",
+      "reference": "count doesn't seem to be used. Let's remove it."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `Mongoid::QueryCache#cache_table` instead.",
+      "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 think this should be `add_column :plans, :funder_id, :integer, index: true, foreign_key: true`",
+      "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": "nit: `process.env.FB_TS_TEST === 'true'`",
+      "reference": "Use !! to cast to bool, rather than the ternary bool antipattern."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to only show the first path in the table.",
+      "reference": "why is it in the list twice to begin with?"
+    },
+    {
+      "prediction": "This should be `if (m_pzLocked) { return; }`",
+      "reference": "This way you will **never** be kicked while monsters are around you."
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "Can you please revert changes to files in the `thoughtworks` package? This is legacy code and we will eventually phase out RC."
+    },
+    {
+      "prediction": "Shouldn't this be `moduleSlugs` instead of `moduleSlug`?",
+      "reference": "This just fixes an unrelated prop type error, correct?"
+    },
+    {
+      "prediction": "I think this should be `adds` instead of `dels`?",
+      "reference": "This was a bug in KBFS-2071 -- oops."
+    },
+    {
+      "prediction": "I think this should be `RedeemedVoucher` or `RedeemedVoucherHash` or something like that. `Redeemed` is too vague.",
+      "reference": "// Condition is a condition for ... Condition"
+    },
+    {
+      "prediction": "Layout/MultilineMethodCallIndentation: Use 2 (not 4) spaces for indenting an expression spanning multiple lines.<br>Layout/AlignArray: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the elements of an array literal if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the parameters of a method call if they span more than one line.<br>Layout/AlignHash: Align the elements of a hash literal if they span more than one line.<br>Layout/AlignParameters: Align the",
+      "reference": "Might be good to test that no emails were sent out, either. You should be able to use `deliveries` for this"
+    },
+    {
+      "prediction": "I think this should be `finer` instead of `finest`.",
+      "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, but it works for now.",
+      "reference": "Is this new list being consumed yet?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "moved to `errors.go` which is the only places `math/rand` is used in this package"
+    },
+    {
+      "prediction": "I think this should be `use Shopsys\\FrameworkBundle\\Component\\Environment\\EnvironmentType;` instead of `use Shopsys\\FrameworkBundle\\Component\\Environment\\EnvironmentType as EnvironmentType;`",
+      "reference": "This use is wrong, corrent is: `Shopsys\\FrameworkBundle\\Component\\Environment` It is the same in classes below"
+    },
+    {
+      "prediction": "I don't think we need to do this here. The `fixUpHost` method will take care of setting the `remoteHost` to `localhost` if it is not set.",
+      "reference": "This is not really needed, the line added in `GridNodeConfiguration.java` is what really fixes the `remoteHost` regression."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I don't think we should be updating the mount points here. The mount points should be set when the container is created, not when the container is updated.",
+      "reference": "For my understanding, why was this removed?"
+    },
+    {
+      "prediction": "I don't think it's a good idea to load all the dlls from the plugins directory. It's better to load only the ones that are referenced in the config file.",
+      "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 think this should be `if (TextUtils.isEmpty(password) || password.length() < 6) {`",
+      "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 if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think we need to specify the protocol here. I'm not sure if this is the right thing to do, but I think",
+      "reference": "Shouldn't we pass here a proposal too?"
+    },
+    {
+      "prediction": "I think it would be good to add a test for the case where the solid is defined in the same pipeline twice.",
+      "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 don't think we need to have a separate endpoint for this. We can just use the `/stamps/{id}/{amount}` endpoint and add a `?topup=true` query param.",
+      "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']) is None`",
+      "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": "Is this a typo?",
+      "reference": "It seems that the two words mean the same thing."
+    },
+    {
+      "prediction": "Why do we need to set `term_` and `proposedTerm_` to `lastLogTerm_` here?",
+      "reference": "Do we need to set lastLogTerm_ to committedLogTerm when `lastLogId_ < committedLogId_`, on line 286"
+    },
+    {
+      "prediction": "I think this should be `RamUsageTester.sizeOf(dic.words) + RamUsageTester.sizeOf(dic.wordHashes) + RamUsageTester.sizeOf(dic.affixes) + RamUsageTester.sizeOf(dic.prefixes) + RamUsageTester.sizeOf(dic.suffixes) + RamUsageTester.sizeOf(dic.stemmers) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) + RamUsageTester.sizeOf(dic.inflectors) +",
+      "reference": "About ~7% memory usage increase on average, at most 512KB"
+    },
+    {
+      "prediction": "I'm not sure what you mean by \"they should be part of the Handle interface\". Can you elaborate?",
+      "reference": "Was this too big of a pain to do now?"
+    },
+    {
+      "prediction": "I don't think it's a good idea to use `zeros` instead of `create` here. `zeros` is a special case of `create` where all elements are initialized to 0.",
+      "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": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Can this be `null:false, default: ''` to avoid the nil vs blank issue?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Had to add this to get macOS to work."
+    },
+    {
+      "prediction": "I think we can remove this class and use `TemporaryFolder` directly in the test class.",
+      "reference": "What is the usage of this class?"
+    },
+    {
+      "prediction": "I'm not sure why we need to pass `null` here. I think we can just remove it.",
+      "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": "Why do we need to clear tombstones here?",
+      "reference": "Ideally this error should not be swallowed (can do in the next PR)"
+    },
+    {
+      "prediction": "I don't think we need to pass `timeout` to `_update_auth`. It's only used in `_update_auth` and `_update_auth` is only called from `update_auth`, so we can just pass `timeout` to `update_auth` and remove `timeout` from `_update_auth`.",
+      "reference": "Wait a minute... You're passing it as a POST parameter. There's no way that can work."
+    },
+    {
+      "prediction": "Should this be `w.Name` or `n`?",
+      "reference": "minor: to be consistent with other logs, either use \"cleaning up\" (preferred) or \"cleanup\"."
+    },
+    {
+      "prediction": "I think this should be `listenbrainz` instead of `listenbrainz_`",
+      "reference": "Adding the config changes to the consul config template `consul_config.py.ctmpl` would be helpful too."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to add this, but I think it would be good to have a `DATASET` type as well.",
+      "reference": "While we're at it should we add Department? See Dataverse Category: Add Department #2829"
+    },
+    {
+      "prediction": "I don't think we should be adding this to the public API. It's not part of the W3C spec, and it's not part of the Selenium spec.",
+      "reference": "Files should have an extra line at the end of them."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but it seems to work.",
+      "reference": "TODO: check whether this is needed or not"
+    },
+    {
+      "prediction": "I think it's better to put this in the `Config` struct, so we don't have to introduce a new global variable.",
+      "reference": "`ReadConfig` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think this should be `//go:build linux && !cgo`",
+      "reference": "This is duplicated with `+build !cgo`?"
+    },
+    {
+      "prediction": "I think this should be `expect { driver.find_element(id: 'box0') }.to raise_error(WebDriver::Error::NoSuchElementError)`",
+      "reference": "This should just be `.not_to raise_error` otherwise it potentially hides errors"
+    },
+    {
+      "prediction": "`numeric_only : boolean, default False` -> `numeric_only : bool, default False`",
+      "reference": "Line 175 needs to be indented so that it's under `sfun`."
+    },
+    {
+      "prediction": "I don't think it's necessary to have a private constructor here.",
+      "reference": "It looks strange a bit. What is the reason?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "Is this one of those `LexicalPreservingPrinter` issues?"
+    },
+    {
+      "prediction": "This should be `if strtype == \"[]uint8\" || strtype == \"[]byte\" {`",
+      "reference": "Can we also add the length of the slice here? Some like `<binary> len %d`"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "move this line together to after line 16, group internal packages together"
+    },
+    {
+      "prediction": "Use the new Ruby 1.9 hash syntax.",
+      "reference": "How about using a reference(project_id) instead of project name"
+    },
+    {
+      "prediction": "`UseRDS` -> `EnableRDS`",
+      "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 done in the `ExternalToolHandler` constructor. The `ExternalToolHandler` is responsible for creating a `DataFile` object, so it should be responsible for making sure it has all the data it needs.",
+      "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": "This should be `static bool usingLocalPrefs;`",
+      "reference": "Make this a member variable instead (move to header inside `class` and rename to `m_bUsingLocalPrefs`)."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the best we can do for now. The problem is that the `pin-mongos` tests are failing because the `pin-mongos` server is not running, and we don't have a way to check if the `pin-mongos` server is running or not. So, we need to skip the `pin-mongos` tests if the `pin-mongos` server is not running. I think the best we can do for now is to skip the `pin-mongos` tests if the `pin-mongos` server is not running, and then we can fix the `pin-mongos` tests in a separate PR.",
+      "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 `EXCLUDE_HIP_PLATFORM=clang` instead of `EXCLUDE_HIP_PLATFORM clang`",
+      "reference": "Why are we skipping these tests? they should be passing in HIP-Clang."
+    },
+    {
+      "prediction": "I don't think you need the `object` base class here.",
+      "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 it would be better to move this check to the `GoQuorumCompatibilityMode` class. This way, we don't need to pass the `chainId` to the `BesuCommand` class, and the `GoQuorumCompatibilityMode` class is the one that knows what to do with the `chainId`.",
+      "reference": "what needs to happen if it is not present?"
+    },
+    {
+      "prediction": "I think we should move this to `provider.NewProvider` so that we don't have to repeat this logic in other places.",
+      "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 think this should say \"bucket\" instead of \"object\".",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I don't think this is correct. If the filter throws an exception, then the request will be filtered out.",
+      "reference": "This line doesn't seem to be right?"
+    },
+    {
+      "prediction": "I think this should be `cvc.Spec.Policy = volumePolicy.Spec.DeepCopy()`",
+      "reference": "this need to be moved up after getting policy.."
+    },
+    {
+      "prediction": "I think this should be `platform\\.gcp\\.project.* invalid project ID`",
+      "reference": "I'm curious - is there a way to test these?"
+    },
+    {
+      "prediction": "I think we can remove the `catalogName` parameter from this method. The `fullTableName` already contains the catalog name.",
+      "reference": "This was broken before as the name of the metadata table started with its type, not catalog."
+    },
+    {
+      "prediction": "I think it would be better to move this line to the end of `Bootstrap` function.",
+      "reference": "Maybe remake `VersionAsString()` function, so that that we would have build info in all places"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `TSocket::hasPendingDataToRead()`. The reason is that `TSocket::hasPendingDataToRead()` is called by `TSocket::read()`, and `TSocket::read()` is called by `TSSLSocket::read()`. So if we do this in `TSocket::hasPendingDataToRead()`, we can avoid calling `TSSLSocket::hasPendingDataToRead()` from `TSSLSocket::read()`.",
+      "reference": "This should probably say something other than SSL_peek?"
+    },
+    {
+      "prediction": "I think this should be \"Search Console module initialization.\"",
+      "reference": "Nitpicking, but this should be capitalized since it's a product name :)"
+    },
+    {
+      "prediction": "I think this should be `RawTransactions []byte` instead of `RawTransactions [][]byte`.",
+      "reference": "Why do we need RawTransactions?"
+    },
+    {
+      "prediction": "I think it's better to use `github.com/iotexproject/go-pkgs/util/byteutil` instead of `github.com/iotexproject/iotex-core/pkg/util/byteutil`",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `BroadcastSgvs.handleNewSgv()`. The reason is that `BroadcastSgvs.handleNewSgv()` can be called from other places, not only from `NSClientService`.",
+      "reference": "should not be this lessThan15MinAgo ?"
+    },
+    {
+      "prediction": "I think we can remove the `catch` here and just let the exception bubble up to the caller.",
+      "reference": "Would it work if we changed the return type of 'RejectRequest*' methods to Exception and instead did `throw RejectRequest(...` ?"
+    },
+    {
+      "prediction": "I think we can remove the `options.supportsDts` check as well.",
+      "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 don't think this is correct. dstCtx and srcCtx can be the same context.",
+      "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 = [];` instead of `protected $indexingConfigurationsToReIndex = array();`",
+      "reference": "Can't use short array syntax for the 3.1 release branch."
+    },
+    {
+      "prediction": "I don't think this should be removed.",
+      "reference": "@rhc54 Is this intentional or for debug purposes?"
+    },
+    {
+      "prediction": "I think it would be better to use `config.registry.settings.get('statsd.host')` instead of `config.registry.statsd`.",
+      "reference": "Maybe it should be `plugins.history` here instead of `listeners.X`, since we use those for listeners configured via `.ini` files?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "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": "Shouldn't this be `if (ivLength != 16 && ivLength != 32) { throw new IOException(\"Can't decrypt file: incorrect iv length found in file: \" + ivLength); }`?",
+      "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/RedundantSelf: Redundant self detected.",
+      "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 think this should be `setuptools>=12,!=34.4.3,<40`",
+      "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 we should rename this to `fakeConnectionManager` to make it clear that this is a fake implementation of the `ConnectionManager` interface.",
+      "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]`",
+      "reference": "Code style. Btw, it's weird Codacy didn't catch that."
+    },
+    {
+      "prediction": "I think this should be called `NewEthAddressFromBytes` or `NewEthAddressFromSlice` or `NewEthAddressFromBytesSlice` or something like that. `NewEthAddress` is ambiguous.",
+      "reference": "no need in BigInt just m.Unwrap().Big().Bitlen() == 0"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it would be nice to have a method on `ConjunctiveMatchFlow` that returns a `Flow` that can be used with the `FlowOperations` interface. That way, we don't have to worry about the implementation details of `ConjunctiveMatchFlow` in the `ConjunctiveMatchFlowContext`, and we don't have to worry about the implementation details of `ConjunctiveMatchFlowContext` in the `FlowOperations` interface.",
+      "reference": "I assume you want to change NetworkPolicy to use bundles in a separate PR later?"
+    },
+    {
+      "prediction": "I think this should be `a4adc8f4da9c7ec0d93b50ca5b73dd14df791c0e` instead of `a4adc8f4da9c7ec0d93b50ca5b73dd14df791c0e/`",
+      "reference": "nit: We could parameterize the commit hash, as it's used in multiple places."
+    },
+    {
+      "prediction": "Can we use lodash instead of underscore?",
+      "reference": "move this above the \"Constants\" as this is 3rd party ones"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to check for the capability. I think it would be better to have a test in `test_homepage.py` that checks for the presence of the `capabilities` key in the response.",
+      "reference": "`assert not in`"
+    },
+    {
+      "prediction": "I think we should emit the container event regardless of whether we restarted the agent or not. If we didn't restart the agent, then the agent is still running, so we should still emit a state change event.",
+      "reference": "just curious, why was this modified?"
+    },
+    {
+      "prediction": "nit: please add a space after the comma.",
+      "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 don't think we should store this as a string, but as an integer.",
+      "reference": "looks like you forgot to actually increase the version"
+    },
+    {
+      "prediction": "I think this should be called `EnableHttpServerSemanticConventions` or something like that.",
+      "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 `config.sync.fullSynchronized !== undefined && !config.sync.fullSynchronization && config.schema === undefined`",
+      "reference": "I think the check should be `config.sync.fullSynchronization === false` - otherwise this will get triggered even when full sync is `true`."
+    },
+    {
+      "prediction": "I think we can remove the `sealedSecretEncryption` field from the `piped` struct and just use the `secretEncryption` field.",
+      "reference": "this blown my mind"
+    },
+    {
+      "prediction": "I think this should be `MayaAPIServiceOutputLabel` instead of `MayaAPIServiceOutputLabel = \"vsm.openebs.io/replica-cont-status\"`",
+      "reference": "this should be controller-container-status"
+    },
+    {
+      "prediction": "This import is not used. Please remove it.",
+      "reference": "Sort import. You can use IDE's organize import feature."
+    },
+    {
+      "prediction": "This import is not used.",
+      "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 should be `t.Skipf(\"Unskip when we have implemented production drand component and local drand network for functional tests\")`",
+      "reference": "This PR is off to a rough start. :)"
+    },
+    {
+      "prediction": "I think this should be `config.pacUrl = config.proxyAutoconfigUrl;`",
+      "reference": "Actually, it looks like the entire `normalizeProxyConfiguration` function isn't needed anymore"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it works for now.",
+      "reference": "How about `unless current_path == carts_path`?"
+    },
+    {
+      "prediction": "This should be `Or Bahari <orbahari@mail.tau.ac.il>`",
+      "reference": "You need to modify the copyrite aliases so it's done automatically."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in `vnode.js` where the `vnode` is created.",
+      "reference": "Quick question: Is this true for all `onAnimation*` and all `onTransition*` events?"
+    },
+    {
+      "prediction": "I don't think this is necessary. If you pass an `Address` object to the `Address` constructor, it will just return the same object.",
+      "reference": "Should sending an address into address error? Since the call isn't needed, and could be fixed easily."
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "This is the wrong MDC; see MDCLoggingContext which imports `org.slf4j.MDC`"
+    },
+    {
+      "prediction": "I think this should be `HistoricDetailQuery initial(boolean initial)`.",
+      "reference": "Maybe we can clarify a bit what that means."
+    },
+    {
+      "prediction": "Why is this needed? I don't see where `tcs` is being canceled.",
+      "reference": "This entire method can be deleted; it isn't used and it isn't an override of Stream."
+    },
+    {
+      "prediction": "nit: I think we can remove this static import.",
+      "reference": "We avoid static method imports. Can you call `MetadataTableUtils.createMetadataTableInstance` instead?"
+    },
+    {
+      "prediction": "I don't think we need to disable pylint here.",
+      "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 about this one, but I think it's better to keep the `includes_mentor?` method in the `Purchaseable` module, because it's not related to the purchase itself, but to the type of purchaseable it is.",
+      "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(\"Nullability doesn't match\", expectedValue == null, arrowVector.isNull(rowId));`",
+      "reference": "Why change this?"
+    },
+    {
+      "prediction": "I think this should be `if (isSomeFileRemote) { throw new Error('Can\u2019t bundle remote files when bundle: true is set') } else { return this.uploadBundle(files) }`",
+      "reference": "Maybe change it to `'Can't UPLOAD remote files when bundle: true is set'`?"
+    },
+    {
+      "prediction": "I think it would be better to use `OpenTelemetrySdkEventSource.Log.SpanProcessorException(nameof(this.OnExport), ex);` instead of `OpenTelemetrySdkEventSource.Log.SpanProcessorException(nameof(this.OnExport), ex);`",
+      "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 would be better to do this in the miner actor's `SubmitPoSt` method, so that we can check that the PoSt is well-formed.",
+      "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": "`ReflectionFunctionAbstract::getReturnType` and `ReflectionMethod::getReturnType` should be `ReflectionFunctionAbstract::getReturnType` and `ReflectionMethod::getReturnType`",
+      "reference": "To be consistent, these should go at the top between `DateTimeZone::listIdentifiers` and `SQLite3::createFunction`."
+    },
+    {
+      "prediction": "I think this should be `long fileModified(String name) throws IOException;` instead of `long fileModified(String name) throws IOException { throw new UnsupportedOperationException(); }`",
+      "reference": "I think we should avoid changing any lucene classes for the moment - fileModified() can probably stay where it is?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to hide the view instead of just making it invisible. The reason is that when the view is invisible, it will still take up space in the layout, but when the view is gone, it will not take up any space in the layout.",
+      "reference": "GONE breaks formating of \":\". or is it fixed below?"
+    },
+    {
+      "prediction": "I don't think we need to deprecate them.",
+      "reference": "What do you mean with this? Won't the deprecation be inherited?"
+    },
+    {
+      "prediction": "I think this should be `--object-class-suffix` instead of `--object-suffix`.",
+      "reference": "line between \" and \" should be <80 chars."
+    },
+    {
+      "prediction": "I'm not sure if we want to remove this. I think we should keep it around for a while and see if anyone complains about it.",
+      "reference": "What will happen if neither `rvm` nor `ruby` is present in the .travis.yml?"
+    },
+    {
+      "prediction": "I don't think this is correct. `len(p)-4` is not the same as `len(pay)`.",
+      "reference": "@p-l- Hey it seems that this was len-4 ?!"
+    },
+    {
+      "prediction": "I don't think we need to use `async/await` here anymore.",
+      "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": "Why is this gone?",
+      "reference": "i have removed this line because whenever a video is played in AntennaPod this line was making play button invisible"
+    },
+    {
+      "prediction": "remove this test",
+      "reference": "remove temporary code"
+    },
+    {
+      "prediction": "I think this should be a `BigInteger` instead of a `Long`.",
+      "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 `const TEST_STORE_NAME = 'test/' + MODULE_SLUG;`",
+      "reference": "Minor detail, but let's use the interpolated template string syntax instead."
+    },
+    {
+      "prediction": "I think it's better to use `[]ClockID` instead of `[]string`.",
+      "reference": "this line is so long, split it to multi lines."
+    },
+    {
+      "prediction": "I think this should be `if (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": "Why do we need this? `std::size_t` is already defined as `std::uint64_t` on Apple.",
+      "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 think this should be called `GetUriTagValueFromRequestUriString` since it takes a `string` and not a `Uri`.",
+      "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 debug level.",
+      "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\").ShouldNot(BeAnExistingFile())`",
+      "reference": "maybe \"app init creates an copilot directory and a workspace file\"? since you validate for the file as well."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "This should be `@template TReturn`",
+      "reference": "Nit: Could rename this file from SabreEvent.php now that it's not for Sabre"
+    },
+    {
+      "prediction": "I'm not sure what you mean by \"started up successfully\". I don't think we have a way of knowing that.",
+      "reference": "I believe this `Stop` function _is_ called with deferred."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the best we can do for now. If we don't have an `accountID`, we don't want to make a request to the API, but we also don't want to throw an error, so we just return an empty promise.",
+      "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 don't think we need to link to yamllint anymore.",
+      "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 `CmpCommon::getDefaultString(HIVE_DEFAULT_CHARSET) == CmpCommon::getDefaultString(HIVE_FILE_CHARSET)`",
+      "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 think this should be a `ConcurrentDictionary` instead of a `Dictionary`.",
+      "reference": "isn't `HttpWebRequest` an expensive key value?"
+    },
+    {
+      "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.<br>",
+      "reference": "couldn't we run this query whether there are observations or not?"
+    },
+    {
+      "prediction": "I don't think we should change the order of arguments here.",
+      "reference": "This shouldn't be needed to change right?"
+    },
+    {
+      "prediction": "I think this should be `[\"examples/bundle/*\"]`",
+      "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 don't think we need to keep the `DDWAF_RET_CODE Action` field in the `DdwafResultStruct` anymore.",
+      "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. I think it would be better to do this in `RGroupDecomposition::add(const ROMol &inmol, const std::map<std::string, std::string> &options)`.",
+      "reference": "Definitely not required, but it would be better if you had a constexpr for `\"INPUT_DUMMY\"`"
+    },
+    {
+      "prediction": "I think this should be reverted. The archive file name should be calculated before the file is opened for writing. If the archive file name is calculated after the file is opened for writing, then the archive file name might be different from the file name that was opened for writing.",
+      "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 don't think we need to add this to the genesis config. The genesis config is only used to create the genesis block, and the genesis block is only used to create the genesis state. The genesis state is only used to create the genesis account. The genesis account is only used to create the genesis balance. The genesis balance is only used to create the genesis nonce. The genesis nonce is only used to create the genesis timestamp. The genesis timestamp is only used to create the genesis difficulty. The genesis difficulty is only used to create the genesis gas limit. The genesis gas limit is only used to create the genesis extra data. The genesis extra data is only used to create the genesis mix hash. The genesis mix hash is only used to create the genesis nonce. The genesis nonce is only used to create the genesis timestamp. The genesis timestamp is only used to create the genesis difficulty. The genesis difficulty is only used to create the genesis gas limit. The genesis gas limit is only used to create the genesis extra data. The genesis extra data is only used to create the genesis mix hash. The genesis mix hash is only used to create the genesis nonce. The genesis nonce is only used to create the genesis timestamp. The genesis timestamp is only used to create the genesis difficulty. The",
+      "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 an error, not a warning.",
+      "reference": "util.Warning()? Easier to say."
+    },
+    {
+      "prediction": "Why do we need to remove clpsK and clpsV?",
+      "reference": "these 2 no longer needed after refactor"
+    },
+    {
+      "prediction": "I think this should be `specCSIRawBlock` and `specCSIFsType`",
+      "reference": "most spec options appear to be snake case - `fa_direct_raw_block`, let's stick to that convention"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "Looks like we can then drop this dependency."
+    },
+    {
+      "prediction": "I think this should be `for s in self.scripts[:]` to avoid modifying the list while iterating over it.",
+      "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 it would be better to have a `getBody()` method that returns a `String` instead of a `StringBuilder`. This way, you don't have to create a new `StringBuilder` every time you want to get the body of the email.",
+      "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": "I think this should be `package project`",
+      "reference": "accidental? i guess it's the same as any other boilerplate"
+    },
+    {
+      "prediction": "I don't think we need to import `ThreadPoolExecutor` and `TimeoutError` here.",
+      "reference": "shadowing the builtin, let's use function or func or callable_"
+    },
+    {
+      "prediction": "Shouldn't this be `[unhashable-dict-key, unhashable-dict-key, unhashable-dict-key]`?",
+      "reference": "Can you add some good examples, for instance integers, strings and whatnot?"
+    },
+    {
+      "prediction": "Please remove the empty lines.",
+      "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 think this should be `std::vector<uint32_t> pgm` instead of `std::vector<unsigned int> pgm`.",
+      "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'm not sure why we need to pass `column.getPrimitiveType()` to `fromParquetPrimitive()`. It seems to me that `stats.genericGetMin()` and `stats.genericGetMax()` already return values of the same type as `column.getPrimitiveType()`.",
+      "reference": "Nit: continuation lines should be indented 4 spaces from the start of the statement."
+    },
+    {
+      "prediction": "I think this should be `sysfs_format_ptr = NULL;` instead of `_sysfs_format_ptr = NULL;`",
+      "reference": "Are these protected by any kind of lock?"
+    },
+    {
+      "prediction": "I think this should be `self.hierarchical_name = kwargs.get('firewall_rule_hierarchical_name')`",
+      "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 think this should be `A reference to a Kubernetes [IngressClass](https://kubernetes.io/docs/concepts/services-networking/ingress/#ingress-class) object`",
+      "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 if this is the right thing to do, but it seems to be the only way to get the tests to pass.",
+      "reference": "@durran Was this change intended to be included in this PR?"
+    },
+    {
+      "prediction": "Why is this `null`?",
+      "reference": "Feels like this could have a default null value to remove the need for this explicit \"null\" use."
+    },
+    {
+      "prediction": "I think this should be `firestore` instead of `filestore`.",
+      "reference": "\"failed to create ...\""
+    },
+    {
+      "prediction": "I think this should be `glog.Errorf(\"Fake pod object could not be created: %v\", err)`",
+      "reference": "No formatting directives, `glog.Error` will do just fine."
+    },
+    {
+      "prediction": "I think this should be `expect { driver.find_element(id: 'box0') }.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 this is the right thing to do. I think it's better to fail fast and let the pod be restarted by k8s. If we want to retry, I think we should do it in the `Start` method of `typhaConnection` and not here.",
+      "reference": "Need an `if err == nil {break}` above this line so that we don't log/sleep if the retry succeeds."
+    },
+    {
+      "prediction": "Why do you remove this interface?",
+      "reference": "Separate it to different methods :D Invoke method look's like old fashion portal class :D"
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it should be `const height = elem.offsetHeight + 'px';` instead of `var height = elem.offsetHeight + 'px';` because `var` is function-scoped and `const` is block-scoped.",
+      "reference": "Collapse/expand animation of filter is broken - not smooth."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to have `mvn install` instead of `mvn test` here. `mvn test` only runs the tests, while `mvn install` builds the project and runs the tests.",
+      "reference": "so during the `script` stage we run `mvn install`?"
+    },
+    {
+      "prediction": "remove this empty line",
+      "reference": "we may need to move CandidateNamespace to this protocol as it is only used here."
+    },
+    {
+      "prediction": "I think we should also set the node status to KILLED if the flow status is EXECUTION_STOPPED due to e.g. pod failure.",
+      "reference": "Just to confirm, this will take care of all the nodes in graph. right?"
+    },
+    {
+      "prediction": "I think this should be `StringAssert.Contains(startInfo.Arguments, string.Format(\"--{0} {1}\", Microsoft.VisualStudio.TestPlatform.CrossPlatEngine.Constants.ParentProcessIdOption, parentProcessId));`",
+      "reference": "nit: remove this line. All \"arrange\" may be in one code block."
+    },
+    {
+      "prediction": "I think it would be better to use `[]byte` instead of `string` for `XattrName`.",
+      "reference": "This name is not a good fit, what about `ExtendedAttribute`? Or even just `Attribute`?"
+    },
+    {
+      "prediction": "Shouldn't this be `assert(testing_fails == 0);`?",
+      "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": "why remove this import?",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "I was saving this white-space for my retirement :nit:"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this deprecation warning. It's not related to the `discard_flash_if_xhr` method, it's related to the `after_action :discard_flash_if_xhr` filter.",
+      "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 if this is the right thing to do, but I think we should be using the hash160 instead of the full address here.",
+      "reference": "change `accountNonceMap` to map[address.Address][]uint64"
+    },
+    {
+      "prediction": "I think this should be `MainGapicProviderFactory.CLIENT_CONFIG` instead of `\"clientconfig\"`.",
+      "reference": "The `CLIENT_CONFIG` is \"client_config\" with underscore, so we can't reuse it in the second arg."
+    },
+    {
+      "prediction": "I think this should be `adminFeatureList[session.UID]` instead of `!adminFeatureList[session.UID]`.",
+      "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 we should throw an exception here. We should just ignore the parameter.",
+      "reference": "Set default value as empty array instead of null"
+    },
+    {
+      "prediction": "I think this should be `if routeUpd.Dst == nil || routeUpd.Gw == nil {`",
+      "reference": "Would be good to UT this case"
+    },
+    {
+      "prediction": "Why do we need to exclude `Specs` here?",
+      "reference": "this was updated after linking the latest ethereum/tests after a submodules recursive update"
+    },
+    {
+      "prediction": "I think this should be `this.__data = Object.create(null);`",
+      "reference": "Todo: find a way to only do this in one spot, ideally with defaults for the stuff we use"
+    },
+    {
+      "prediction": "I don't think we need to import `ParquetFileReader` here.",
+      "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": "I think it's better to use `if not isinstance(feval, list): feval = [feval]` instead of `if callable(feval): feval = [feval]`",
+      "reference": "Please move this to L3115 to group the code logically - all code for `feval` in one place."
+    },
+    {
+      "prediction": "I think this should be `final` as well.",
+      "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'm not sure this is the right thing to do here. I think it should be `time.Now().Add(time.Duration(cr.Expires) * time.Millisecond)`.",
+      "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 don't think we need to call `cds.cdsObjectToUpnpavObject` here. We can just call `cds.cdsObjectToUpnpavObject(obj, obj, host)`.",
+      "reference": "ineffectual assignment to `err` (from `ineffassign`)"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do this. I think we should just have a `map[string]func() (string, error)` and return an error if the function returns an empty string.",
+      "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 needed? `listenSocket_` is initialized to `THRIFT_INVALID_SOCKET` in the constructor.",
+      "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 do this, but it seems to work. I'm not sure if there's a better way to do this, or if this is the best way to do this, but it seems to work.",
+      "reference": "Could this line use `url: current_user.subscription`?"
+    },
+    {
+      "prediction": "I don't think we need to escape the `/` in the `preg_quote` call.",
+      "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": "Please remove this empty line.",
+      "reference": "The changes in this file are not necessary."
+    },
+    {
+      "prediction": "util should be in pkg/apiserverlite/util/",
+      "reference": "run `make lint` to fix"
+    },
+    {
+      "prediction": "I think this should be `assert result.success`",
+      "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": "can we use int here instead of string?",
+      "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 think this should be `github.com/opencontainers/selinux/go-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": "Why is this not nullable?",
+      "reference": "nit: It might be worth defining these elsewhere as a python-type?"
+    },
+    {
+      "prediction": "I think this should be `ENABLE_QUARTZ_SCHEDULER` or `ENABLE_QUARTZ_SCHEDULER_ENABLED`.",
+      "reference": "prefix it with azkaban. Maybe azkaban.server.schedule.enable_quartz?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it would be nice to have a `Bolt::Errors::LookupError` class that we could use here and in the `Puppet::DataBinding::LookupError` rescue.",
+      "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 this should be `panic(\"NOT IMPLEMENTED\")`",
+      "reference": "Probably need to understand/implement this one now."
+    },
+    {
+      "prediction": "nit: I think it would be better to use `Optional.ofNullable` instead of `if (d != null) { depValidationStatuses.put(d, v); }`",
+      "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": "I'm not sure this is the right thing to do, but I think it's the least bad option we have right now.",
+      "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": "I think this should be `dr_stats_get_num_built_blocks()` to be consistent with the rest of the API.",
+      "reference": "style violation: { on own line"
+    },
+    {
+      "prediction": "I think this should be `H2O_COMPRESS_HINT_DISABLE` if `headers[i].value_len` is 0.",
+      "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": "nit: `GetTipSetAndStateRoot` would be more descriptive than `GetTipSetStateRoot`",
+      "reference": "The sender doesn't use `GetTipSet`, so please remove it from this interface."
+    },
+    {
+      "prediction": "I don't think we need this import.",
+      "reference": "We can get rid of this import then."
+    },
+    {
+      "prediction": "I think this should be `\"/var/azkaban/private\"` instead of `\"/var/azkaban/private/conf\"`.",
+      "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 *zone == \"\" && mgce.OnGCE() {` instead of `if *zone == \"\" && mgce.OnGCE() {`",
+      "reference": "Why is error ignored here?"
+    },
+    {
+      "prediction": "I'm not sure this is needed. I think we can just use `PartitionFlagBiosGrub` instead of `PartitionFlagBiosGrubLegacy` and `PartitionFlagBiosGrubLegacy` can be removed.",
+      "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 `table.currentSnapshot().allManifests()`",
+      "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": "Shouldn't this be `if (Tools.ContainsKey(toolName) && Tools[toolName].osSpecific[Os] != null && !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 `event.Content.String()`",
+      "reference": "I think event.Content is of type interface{}, so we may need %v here?"
+    },
+    {
+      "prediction": "This should be `RemovedInInvenio23Warning`.",
+      "reference": "Note: `editor` depends on legacy `BibEdit`"
+    },
+    {
+      "prediction": "This should be `duraton` -> `duration`",
+      "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 it's better to use `self.assertRaises` instead of `self.assertRaisesRegex`.",
+      "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": "I think this should be \"Checks whether this {@code Value} is not empty, i.e. if the underlying value is present.\"",
+      "reference": "I think it was reversed"
+    },
+    {
+      "prediction": "I don't think we should remove the constructor with `JspParser` parameter.",
+      "reference": "The setter \"setName\" can be package-private as well"
+    },
+    {
+      "prediction": "I think this should be `const int t = this->iter_ + 1;`",
+      "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 `#ifdef _WIN32` instead of `#ifndef _WIN32`.",
+      "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": "nit: I think it would be better to use `isRequestStreaming()` and `isResponseStreaming()` instead of `getRequestStreaming()` and `getResponseStreaming()`.",
+      "reference": "If you use an enum for the streaming type, you wouldn't need to add each boolean as a separate attribute."
+    },
+    {
+      "prediction": "I think this should be `from sqlalchemy.exc import IntegrityError`",
+      "reference": "can you check if you can use `sqlalchemy.exc.IntegrityError` instead?"
+    },
+    {
+      "prediction": "Why do we need a recursive mutex here?",
+      "reference": "Why we need to use recursive_mutex here?"
+    },
+    {
+      "prediction": "I think we should keep the `acmeIssuer` variable here. It's easier to read and understand.",
+      "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": "I think this should be `import org.openqa.selenium.grid.sessionmap.SessionMap;` instead of `import org.openqa.selenium.grid.sessionmap.SessionMap;`",
+      "reference": "I think that this is an unused import"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what else to do either.",
+      "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 think this should be called `GetCommitsViaComparison` or something like that. `ListCommitsViaComparison` sounds like it's going to return a list of commits, but it doesn't.",
+      "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 think this should be `if apis.IsInUpdate(ctx) && !current.DeletionTimestamp.IsZero() {`",
+      "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 don't think we need a new version for this. We can just write the nextWriteDocValuesGen to the file.",
+      "reference": "hmm should this be `VERSION_81` instead?"
+    },
+    {
+      "prediction": "I think this should be `trieRebuildAccountChunkSize * trieRebuildCommitFrequency`",
+      "reference": "What is the reason for increasing this?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems 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 don't think we need to expose these in the API. They are only used internally and can be set directly on the `HttpSessionCsrfTokenRepository`.",
+      "reference": "The formatting is a little off here."
+    },
+    {
+      "prediction": "nit: I think this should be `expected` instead of `sw`",
+      "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": "I think this should be `for (int i = CONST_SLOT_FIRST; i < CONST_SLOT_LAST; ++i)`",
+      "reference": "You should use C++-type casts instead."
+    },
+    {
+      "prediction": "I don't think we need to do this for `CreateWorkflowExecution` since it's a new workflow. We only need to do this for `UpdateWorkflowExecution` since it's an existing workflow.",
+      "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 think this should be `telemetry.SPIFFEID` instead of `telemetry.SPIFFEID`.",
+      "reference": "why are you replacing log here? it is defined above and contains EntryID and SPIFFEID"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this check. I think it should be done in the `Start` method of the `openvpn.Server` struct.",
+      "reference": "Why not `checkOpenvpn` as in client command?"
+    },
+    {
+      "prediction": "I don't think this is the right way to check if mocha is loaded. `mocha.reporter('base')` will throw an error if mocha is not loaded.",
+      "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:%d HTTP/1.1\\r\\n\\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 what you're trying to do here, but you can't call `this->Save(1);` from within the `Handle_OP_PetCommands` function. The `this` pointer refers to the `Client` object, not the `NPC` object. You need to call `mypet->Save(1);` instead.",
+      "reference": "No need to use `this->` here"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the cluster install is complete, but the cluster is not yet installed, then we should not set the `ClusterInstallCompleted` condition to `True`. The `ClusterInstallCompleted` condition should only be set to `True` when the cluster install is complete and the cluster is installed.",
+      "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": "I think this should be `namespaceCache_mock.go`",
+      "reference": "doh. My fault, sorry."
+    },
+    {
+      "prediction": "I think this should be `proposal.ServiceID` instead of `proposal.ID`",
+      "reference": "Why do we returning `interface{}` here? Don't we have a predefined type for `ID`?"
+    },
+    {
+      "prediction": "I don't think we need to import `map` from `lodash` anymore.",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "This should be `import _ \"github.com/rclone/rclone/fs/config/provider/goconfig\"` and `import _ \"github.com/rclone/rclone/fs/config/provider/viper\"`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I don't think we need to check for `null` here. `noteIds` is an array, so it will never be `null`.",
+      "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 don't think we need to expose this as a constant. It's only used in one place.",
+      "reference": "Almost makes me wonder if this should default to Rpc-Context so everything defaults under Rpc-"
+    },
+    {
+      "prediction": "I think it's better to keep the import path as `github.com/iotexproject/iotex-core/action/protocol/account/util`",
+      "reference": "File is not `goimports`-ed (from `goimports`)"
+    },
+    {
+      "prediction": "I think this should be `mset.jsa.limits.MaxConsumers <= 0 || mset.config.MaxConsumers < mset.jsa.limits.MaxConsumers`",
+      "reference": "would these require any account lock here?"
+    },
+    {
+      "prediction": "I think we can remove the `else cls._timeout` and just return `cls._timeout`",
+      "reference": "The else doesn't return anything?"
+    },
+    {
+      "prediction": "I think this should be `if (opts.fee === 0 || opts.feeSat === 0) this.givenFeeSat = 0;`",
+      "reference": "if (typeof (opts.fee) !== 'undefined' || typeof (opts.feeSat) !=='undefined') {"
+    },
+    {
+      "prediction": "I think this should be `if d.Permissions == \"\" { p = \"rwm\" } else { p = d.Permissions }`",
+      "reference": "I not sure we need to change this method's logic. Isn't no perms and invalid config?"
+    },
+    {
+      "prediction": "I think this should be 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 this should be `o.targetEnvironment.AccountID` instead of `o.targetApp.AccountID`.",
+      "reference": "Can we add this code to `job_package.go` and `svc_package.go`?"
+    },
+    {
+      "prediction": "`nodeIDKey` is not used in this PR.",
+      "reference": "While you're here, can you change the casing on this so it is unexported? `nextPlanNodeIDKey` instead."
+    },
+    {
+      "prediction": "I think this should be `// +build !windows` instead of `// +build !bootstrap`",
+      "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 `True` instead of `False`.",
+      "reference": "Again, `no_duplicates=False` would be clearer here..."
+    },
+    {
+      "prediction": "I think it would be better to have a single channel that is closed when the subscription is cancelled. This way the caller doesn't have to check two channels.",
+      "reference": "shouldnt we stop these routines BEFORE the db closed?"
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "It fills me with endless sadness that we can't use Selenium's own `SessionId` here."
+    },
+    {
+      "prediction": "I think this should be `INFO` level.",
+      "reference": "This will always need to be displayed to users."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Maybe we make an `ORIGINS` constant on the Cart model instead?"
+    },
+    {
+      "prediction": "I think this should be `ManagedAgents` instead of `ManagedAgentsUnsafe`.",
+      "reference": "looks like this is not needed anymore with boltdb migration. lets confirm with @fenxiong"
+    },
+    {
+      "prediction": "why do we need to pass `false` to `NewWorkingSet` here?",
+      "reference": "shadow: declaration of \"err\" shadows declaration at line 408 (from `govet`)"
+    },
+    {
+      "prediction": "I think this should be `finer` instead of `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging. `finer` is one level above `finest`. `finest` is the lowest level of logging.",
+      "reference": "Again, the existing level is correct --- we're letting users know about something that might cause there tests to fail."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get a `conversion.Converter` with the new API.",
+      "reference": "Required by the libcalico-go changes"
+    },
+    {
+      "prediction": "I think this should be `current_min_count` instead of `current_mode_min_count`.",
+      "reference": "You want \"current_min_count\""
+    },
+    {
+      "prediction": "I think we should keep the `GROUP BY` clause.",
+      "reference": "Why changes in this file?"
+    },
+    {
+      "prediction": "I don't think this is correct. I think it should be `Pattern.compile(\"\\\\[([^\\\\]]+)\\\\]\\\\[([A-Za-z_][A-Za-z_.0-9]*)*\\\\]\", Pattern.CASE_INSENSITIVE);`",
+      "reference": "This doesn't look right. I think what you want here is `\\\\[([^\\\\]]+)\\\\]\\\\[[A-Za-z_][A-Za-z_.0-9]*]*\\\\]`."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "@AndrewRayCode thanks for contributing   Could you write/modify the specs to make sure that this description is working properly?"
+    },
+    {
+      "prediction": "I think this should be `if error.content_type == 'application/json' and not is_redirect:`",
+      "reference": "I would have tested 304 only there."
+    },
+    {
+      "prediction": "I'm not sure why this was removed. It seems like it should still be there.",
+      "reference": "`TimeoutAfter` to be safe."
+    },
+    {
+      "prediction": "I don't think we should be asserting here. If the syscall method is sysenter, then we shouldn't have a displaced pc. If the syscall method is not sysenter, then we should have a displaced pc. I think we should assert that the syscall method is sysenter if and only if the displaced pc is NULL.",
+      "reference": "So the 32-bit tests on the new kernel hit this case? Won't they fail in debug build?"
+    },
+    {
+      "prediction": "I don't think this is necessary. The default charset is already UTF-8.",
+      "reference": "Is there a recommended quote style for attributes? I see single and double here, double further down."
+    },
+    {
+      "prediction": "I think this should be `the number of seconds taken to boot RSpec and load the spec files` instead of `the number of seconds taken to boot RSpec and load the spec files`.",
+      "reference": "The description of `load_time` here is different from the description below..is that intentional?"
+    },
+    {
+      "prediction": "Shouldn't this be `<span class=\"cardImageIcon material-icons folder\"></span>`?",
+      "reference": "Actually it could also be a video playlist. But music is used more often."
+    },
+    {
+      "prediction": "I think this should be `if (writeConcern.w === 0 || writeConcern.wtimeout === 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'm not sure this is the best way to calculate the average cycle time. I think it would be better to calculate the average cycle time as the average of the cycle times for each point in the peak.",
+      "reference": "You can simplify this by doing: return (EndTime - StartTime) * 60 / PointsAcrossPeak;"
+    },
+    {
+      "prediction": "I don't think this change is correct. The `QFontMetrics::horizontalAdvance` returns the advance of a single character, not the advance of a string.",
+      "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 think this should be `boltdir_type: :cd10`",
+      "reference": "I already set up cd10 for \"inventory_version in google analytics. I've added cd11 for Boltdir Type now"
+    },
+    {
+      "prediction": "I don't think we need to call `untaint` here.",
+      "reference": "@henrikhodne We are lazily skipping `jruby-*` here (assuming that JRuby all of a sudden start using these version numbers)."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. You should be able to use `wxString::Format(_T(\"%.2f\"), y_start_val + ny * d_oy_val)` instead of `wxString::Format(wxT(\"%s\"), format_number(y_start_val + ny * d_oy_val, 2))`.",
+      "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'm not sure this is the best solution, but it seems to be the only way to get the tests to pass reliably.",
+      "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 for this. I think it should be in the `scripts/` folder, but I'm not sure if it should be in `scripts/utils/` or `scripts/monorepo/` or `scripts/monorepo/utils/` or `scripts/utils/monorepo/` or `scripts/monorepo/utils/monorepo/` or `scripts/monorepo/utils/monorepo/utils/` or `scripts/monorepo/utils/monorepo/utils/monorepo/` or `scripts/monorepo/utils/monorepo/utils/monorepo/utils/` or `scripts/monorepo/utils/monorepo/utils/monorepo/utils/monorepo/` or `scripts/monorepo/utils/monorepo/utils/monorepo/utils/monorepo/utils/` or `scripts/monorepo/utils/monorepo/utils/monorepo/utils/monorepo/utils/monorepo/` or `scripts/monorepo/utils/monorepo/utils/monorepo/utils/monorepo/utils/monorepo/utils/` or `scripts/monorepo/utils/monore",
+      "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. I think it should be done in `sigpending_t` constructor, which is called from `sigpending_t` constructor.",
+      "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": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 to do this. I think it would be better to do this in the `value` property of the `UIProperty` class.",
+      "reference": "Might as well use translate here, as @jcsteh suggested"
+    },
+    {
+      "prediction": "I think it would be better to have this as `title_format = param.String(default=\"{label} {group}\\n{dimensions}\", doc=\"\"\"The formatting string for the title of this plot, allows defining a label group separator and dimension labels.\"\"\")`",
+      "reference": "Definitely an improvement as long as the old tests pass (i.e backwards compatible)."
+    },
+    {
+      "prediction": "nit: can you move this import to the top of the file?",
+      "reference": "nit: non functional change"
+    },
+    {
+      "prediction": "I think this should be `/root/.aws/credentials` instead of `/home/.aws/credentials`",
+      "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": "Why is this change necessary?",
+      "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": "I think we can remove this interface and just use `MiningCoordinator` directly.",
+      "reference": "New files should be copyright Hyperledger Besu Contributors."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "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": "`list` is not a valid type for `to_replace` or `value`.",
+      "reference": "@itholic can you also update the parameters in the docs?"
+    },
+    {
+      "prediction": "I think this should be `Comparator<? super T> comparator = original.getComparator(); return comparator == null ? null : new Comparator<T>() { public int compare(T o1, T o2) { return comparator.compare(o1, o2); } };`",
+      "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 don't think this is correct. The metrics config should be derived from the table's properties, not from the table itself.",
+      "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 think this should be `--java-primitive-has-method` to be consistent with the other flags.",
+      "reference": "can you make all the `_` into `-` to be consistent with the other options?"
+    },
+    {
+      "prediction": "I think we can remove this interface and just use `*cloudformation.TemplateBody` as the type for the `Template` field in the `Stack` struct.",
+      "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+[assembly: InternalsVisibleTo(\"OpenTelemetry.Exporter.OpenTelemetryProtocol.AspNetCore.Tests, PublicKey=0024000004800000940000000602000000240000525341310004000",
+      "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 think this should be `let(:transport) { Bolt::Transport::Local }`",
+      "reference": "Why does this include `bolt/inventory`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should use `Objects.equals` instead of `==` in the `containsKey` and `remove` methods.",
+      "reference": "We called `t._1.equals(...)` where `t._1` potentially could be `null`."
+    },
+    {
+      "prediction": "I think this should be `node.hasAttribute('aria-valuenow') && node.hasAttribute('aria-valuemin') && node.hasAttribute('aria-valuemax')`",
+      "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 it should be `--file` instead of `--file`",
+      "reference": "Shouldn't it be optional?"
+    },
+    {
+      "prediction": "I think this should be \"API Gateway REST API\" or \"API Gateway V2 REST API\" or \"API Gateway V2 HTTP API\" or \"API Gateway V2 WebSockets API\".",
+      "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 don't think this is a good idea. If the file does not exist, it should throw an exception.",
+      "reference": "Duplicate call to the method. Better to put in a variable."
+    },
+    {
+      "prediction": "I'm not sure I understand the purpose of `lifetime` here. It seems like it's only used in `PrefetchAfterBlockRetrieved` and `PrefetchAfterBlockRetrieved` doesn't do anything with it, so I'm not sure what it's for.",
+      "reference": "\"and returns that\" -- it looks like this method has no return value."
+    },
+    {
+      "prediction": "I don't think we need to support diagonal flip for bitmap masks.",
+      "reference": "Modifications are also needed for PolygonMask."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Does this really make a difference?"
+    },
+    {
+      "prediction": "I think this should be `new RenameFieldValueValidator(\"_root_\", \"id\")`",
+      "reference": "Added a validator for _root_, which is now added automatically since the schema used here declares _root_."
+    },
+    {
+      "prediction": "I think this should be \"The estimated cost to open a split, used as a minimum weight when combining splits. By default this will use the \"read.split.open-file-cost\" value in the table properties of the table being updated.\"",
+      "reference": "The other properties are `file-open-cost`, not `open-file-cost`."
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. You should be able to use `PKG_CHECK_MODULES(libunwind, libunwind, [AC_DEFINE([HAVE_LIBUNWIND], [1], [Define to 1 if you have the <libunwind.h> header file])], [AC_MSG_WARN([libunwind not found])])`",
+      "reference": "confused by this: what we need here is not libunwind but libunwind-dev right? Why skipping detection?"
+    },
+    {
+      "prediction": "I think this should be `property.SetMethod != null && !property.SetMethod.IsPrivate`",
+      "reference": "just in case helps, it this related with this PR?"
+    },
+    {
+      "prediction": "I think this should be `data[\"server_conn\"][\"ip_address\"] = data[\"server_conn\"][\"ip_address\"][\"address\"] if data[\"server_conn\"][\"ip_address\"] else None`",
+      "reference": "This is incomplete I think (at least source_address and ip_address are missing)"
+    },
+    {
+      "prediction": "Why do we need to call `System.exit(0)` here?",
+      "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 `RealPath = new string[] { parameter.RealPath }`",
+      "reference": "fixes ArgNullEx when using both `x-ms-parameter-grouping` and media type `application/xml` in the same operation (issue #2236)"
+    },
+    {
+      "prediction": "The specified bucket does not exist. -> The specified bucket does not exist.",
+      "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 this is the right place for this test. I think it should be in the `pkg.rb` file.",
+      "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 `fmt` instead of `println`?",
+      "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 think this should be `requestEntity` instead of `httpEntity`.",
+      "reference": "Just fixing a bunch of outdated java docs in this file"
+    },
+    {
+      "prediction": "Why do we need to change this to a non-concurrent dictionary?",
+      "reference": "I think it may be safer not to touch Concurrent to normal."
+    },
+    {
+      "prediction": "I don't think this is the right place to emit the error. I think it should be in the `drain` event handler in the `Pool` class.",
+      "reference": "is this for everything, or just legacy?"
+    },
+    {
+      "prediction": "I don't think we need to check if `eventEmitter` is `false` here. If `eventEmitter` is `false` then `callback` will be `null` and the `if (typeof callback === 'function') return callback(error, null);` line will be skipped.",
+      "reference": "I think this should be a `MongoError` right?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I think it's the only way to do it. The problem is that we need to download the files from the release page, but the release page doesn't have a tarball of the files. So we need to download the tarball of the files, and the tarball of the database, and the tarball of the site.",
+      "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 this should be `allow-all-egress` instead of `allow-all-egress-normal`.",
+      "reference": "Why these changes to existing tests?"
+    },
+    {
+      "prediction": "I think this should be `source?.GetDictionary(ConfigurationKeys.ServiceNameMappings)?.Where(kvp => !string.IsNullOrEmpty(kvp.Key) && !string.IsNullOrEmpty(kvp.Value)).ToDictionary(kvp => kvp.Key.Trim(), kvp => kvp.Value.Trim()) ?? new Dictionary<string, string>();`",
+      "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 we should rename `provided_features` to `additional_features` to be consistent with the rest of the codebase.",
+      "reference": "This is just for consistency + searchability with the transports"
+    },
+    {
+      "prediction": "I don't think we need to create a new directory for this. We can just use the same directory as `CertFile` and `KeyFile`.",
+      "reference": "please run gofmt."
+    },
+    {
+      "prediction": "I think you can remove this line.",
+      "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": "I'm not sure if this is the right thing to do, but I think it is. If the app doesn't exist in the store, then we don't want to add the resource to the app's `dependedResources` map.",
+      "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 don't think we need to validate the config name here. We already do it in the `createConfNode` method.",
+      "reference": "just observing that this innocent looking change seems important to this PR. Previously this data had disappeared from the state."
+    },
+    {
+      "prediction": "I don't think this is correct. If errorRowCount is 1000000000 and cond->getOptionalInteger(0) is 2000000000, then errorRowCount will be set to 2000000000.",
+      "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 don't think we need this. We can just use the name of the protocol.",
+      "reference": "it is confusing to return ID as Name"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the request fails, the connection is closed, so the next request will fail as well. I think it would be better to close the connection only if the request is successful.",
+      "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 don't think this is the right place to do this. This should be done in the `web/app.php` file, which is the entry point of the application.",
+      "reference": "commit mesasge: I would append \"...Kernel::loadClassCache() method call\""
+    },
+    {
+      "prediction": "I don't think we need to add COMMAND here. It's not a modifier key, it's a command key.",
+      "reference": "Keys.COMMAND seems to be an alias to Keys.META. That isn't mentioned?"
+    },
+    {
+      "prediction": "Please revert this change. It is not related to this PR.",
+      "reference": "You can move this import down to before line 8 (import org.junit.Test;)."
+    },
+    {
+      "prediction": "I think this should be `inventory_index_id` instead of `scanner_index_id`.",
+      "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": "This should be `afterInteractingWith(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 go about it, but it works for now.",
+      "reference": "Should this be I18n'd?"
+    },
+    {
+      "prediction": "I think this should be `ImmutableList.copyOf(fieldErrorLists)` instead of `ImmutableList.of(fieldErrorLists)`. `ImmutableList.of` will create a list with a single element that is `fieldErrorLists`.",
+      "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 think this should be `type(m2) == type(m)`",
+      "reference": "@greglandrum Is this test sufficient?"
+    },
+    {
+      "prediction": "I think we should keep the `GITHUB_USER` and `GITHUB_PASSWORD` environment variables around for now, just in case we need to use them again in the future.",
+      "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 way to do it, but I think we need to add a `plan_type` column to the `subscriptions` table so that we can make the `plans` table polymorphic.",
+      "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 `waitForContainerTransitions` instead of `waitForContainerTransition`.",
+      "reference": "I don't think we need this line. You're logging in `waitForContainerTransition()` anyway"
+    },
+    {
+      "prediction": "Please revert this change.",
+      "reference": "Shouldn't the `:attribute` placeholder appear?"
+    },
+    {
+      "prediction": "I think it's better to rename `arg` to `struct` here.",
+      "reference": "Could you update this to `structLike` instead of `arg`?"
+    },
+    {
+      "prediction": "I think this should be `gc.leaking,!gc.conservative`",
+      "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 `g_build_filename (g_get_tmp_dir (), \"tmpostree-deltaobj-XXXXXX\")`",
+      "reference": "Ugh, we were using cwd? =("
+    },
+    {
+      "prediction": "I think this should be removed from the struct, as it's not used anymore.",
+      "reference": "This can probably be removed?"
+    },
+    {
+      "prediction": "`std::make_unique<InterimResult>(std::move(resultColNames_))` -> `std::make_unique<InterimResult>(resultColNames_)`",
+      "reference": "put this sentence in `if (onResult_) { }`"
+    },
+    {
+      "prediction": "I think this should be `gsutil cp -r gs://{scanner_bucket}/config_validator_constraints /home/ubuntu/config_validator_constraints`",
+      "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 this should be \"preceding\" instead of \"preceeds\".",
+      "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 think this should be a `DeprecationWarning` instead of an `ImportWarning`.",
+      "reference": "This doesn't seem related to docs? :)"
+    },
+    {
+      "prediction": "I think this should be `notificationDuringQuietTimeEnabled = false;` and `private static boolean notificationDuringQuietTimeDisabled = false;`",
+      "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 think this should be `ForceTraceWriteLine` instead of `ForceTraceFail`",
+      "reference": "Think you need to keep `RawWrite` around as obsolete until NLog6 (Property that just assigns `ForceTraceWriteLine`)"
+    },
+    {
+      "prediction": "I don't think we need to pass in the icache line size here. The icache line size is always the same as the dcache line size on AArch64.",
+      "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 we can remove the `pipeline_` prefix here.",
+      "reference": "maybe should rename this to be `run_status_sensor_definition.py`"
+    },
+    {
+      "prediction": "I don't think we need this class. We can use `SparkWriteOptions.FANOUT_ENABLED` directly.",
+      "reference": "Do we want to add the other write options in this PR or keep the refactor separate?"
+    },
+    {
+      "prediction": "I think it would be better to raise an exception if the token is not refreshed after 5 retries.",
+      "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 you need the `> 0.5` here. You can just pass `y_pred` to `recall_score` and `precision_score`.",
+      "reference": "Same about new metrics. Can we avoid adding them?"
+    },
+    {
+      "prediction": "I think this should be `--vpc-name` and `--subnet-name` instead of `--vpc` and `--subnet`.",
+      "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": "Why not use `setDataAtCell` instead of `setSourceDataAtCell`?",
+      "reference": "The test description says it's \"by reference\". We should change the description"
+    },
+    {
+      "prediction": "I think this should be `public abstract ValidationCategory ValidationCategory { get; }`",
+      "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 this should return an empty map instead of `nil`",
+      "reference": "just looks like you should return an **empty map** with nil error"
+    },
+    {
+      "prediction": "Shouldn't this be `os.Exit(0)`?",
+      "reference": "could you also update the starting log at L146?"
+    },
+    {
+      "prediction": "I think this should be `strings.Contains(dbusError.Name, \"org.freedesktop.systemd1.Error.ConnectionClosed\")`",
+      "reference": "Also this probably should be `error.As()` or something like it."
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "This is Safari specific and should be defined somewhere in the `safaridriver` namespace"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. I think it should be done in `ostree-prepare-root` itself.",
+      "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/format.hpp>`.",
+      "reference": "Not sure if this doesn't require explicitly adding this library to cmake."
+    },
+    {
+      "prediction": "This should be `duraton` -> `duration`",
+      "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 `from.toPath().toString()`.",
+      "reference": "seems reasonable to also want to include the 'to' location?"
+    },
+    {
+      "prediction": "Line is too long. [82/80]",
+      "reference": "This `if` statement can be merged with the above `if opts[:collect_perf_data]`."
+    },
+    {
+      "prediction": "I think this should be `MatchFold` instead of `Match`. `MatchFold` is case-insensitive, and `Match` is case-sensitive.",
+      "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'm not sure if this is the best way to go about it, but I added a new `workshop-type` class to the `hgroup` so that we can style it differently from the product title.",
+      "reference": "I believe this should be on `workshops/show` now, not `products/show`"
+    },
+    {
+      "prediction": "I think this should be `NL` as well.",
+      "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'm not sure this is the right place to do this. I think it should be done in `GetTracerHomeDirectory` instead of `GetTracerFrameworkDirectory`. The reason is that `GetTracerFrameworkDirectory` is called from `GetTracerHomeDirectory` and `GetTracerFrameworkDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and `GetTracerHomeDirectory` is called from `GetTracerHomeDirectory` and",
+      "reference": ".NET Framework: Fallback to `DD_INTEGRATIONS` if `DD_DOTNET_TRACER_HOME` was not set."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to handle the case where there is no path in the URL.",
+      "reference": "Can you use `authority, _, path = rest.partition(b\"/\")` here? That should make stuff a bit cleaner."
+    },
+    {
+      "prediction": "I think it would be better to have this as `--stress-test SECONDS` instead of `--stress-test LIMIT`.",
+      "reference": "I don't think this needs to be a float if its a number of seconds... :)"
+    },
+    {
+      "prediction": "`$values` should be `array $values`",
+      "reference": "Please add \"Array\" back, right now it's not a proper sentence."
+    },
+    {
+      "prediction": "I think this should be `BooleanClause.Occur.MUST_NOT` instead of `BooleanClause.Occur.SHOULD`.",
+      "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 think this should be `if telemetryConfig.Enable && log.GetTelemetryURI() == \"\" {` instead of `if log.GetTelemetryURI() == \"\" {`",
+      "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'm not sure this is the right thing to do. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not overwrite it. If the user has already set a content-type, we should respect that and not",
+      "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 don't think you need to create a new registry here. You can just use the one from `protocol.DefaultRegistry`.",
+      "reference": "importShadow: shadow of imported from 'github.com/stretchr/testify/require' package 'require' (from `gocritic`)"
+    },
+    {
+      "prediction": "I'm not sure why this change is needed.",
+      "reference": "Wouldn't it be enough to change the condition in the for loop to `link_state != IBL_UNLINKED`?"
+    },
+    {
+      "prediction": "I think this should be `G_OPTION_FLAG_HIDDEN` so that it doesn't show up in `git commit --help`.",
+      "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 are we ignoring the error here?",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "I think this should be `get \"/auth/failure\" => \"auth#failure\", as: :auth_failure`",
+      "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 + '\"><i class=\"md-icon\">info_outline</i></button>'), 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": "I think this is not needed anymore.",
+      "reference": "unnecessary blank line"
+    },
+    {
+      "prediction": "I don't think we need to have a constant for the partition column id. We can just use `MetadataColumns.PARTITION_COLUMN_ID` instead of `MetadataColumns.PARTITION_COLUMN_ID`.",
+      "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 don't think you need to include the `test_SharedMemTransportDescriptor` anymore.",
+      "reference": "We can just include the public UDPv4TransportDescriptor header here ..."
+    },
+    {
+      "prediction": "shouldn't it be `self._session['data-address'] = data_address`?",
+      "reference": "Why do we need this change?"
+    },
+    {
+      "prediction": "I don't think this is correct. If you have a file and its manifest, you should have 2 tags, one for the file and one for the manifest. The manifest tag should have the root hash of the manifest, and the file tag should have the root hash of the file.",
+      "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 think this should be `var slotIndex = (Slot)(startSlot % timeParameters.SlotsPerHistoricalRoot);`",
+      "reference": "Is this because you don't have % on your version of Slot? I don't really care either way."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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\", err, cachedLibPath.data());`",
+      "reference": "If cachedLibPath > 200 bytes, it will overflow errString."
+    },
+    {
+      "prediction": "I don't think this is correct. `aria-label` and `aria-labelledby` are allowed on `generic` roles.",
+      "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": "Why is this check not needed for the OVM?",
+      "reference": "Let's keep it to avoid the diff"
+    },
+    {
+      "prediction": "I think we should add `location` to `RESERVED_PROPERTIES` instead of `CAPABILITIES`.",
+      "reference": "nit: not directly related to this PR, but shall we use `ImmutableSet` for consistency?"
+    },
+    {
+      "prediction": "I think we can remove the `reverse_merge` and `merge` calls and just do `@params = processed_parameters.merge(@reverse_merged_params).merge(@merged_params).tap { clear_changes }`",
+      "reference": "Layout/DotPosition: Place the . on the previous line, together with the method call receiver."
+    },
+    {
+      "prediction": "I think we should check if the passcode hash is null before calling decryptWithPasscode. If the passcode hash is null, it means the user has not set a passcode, so we should not try to decrypt the instance URL.",
+      "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 it would be better to have a `BftBlockHeaderFunctions.CmsSignature` class that extends `BftBlockHeaderFunctions` and overrides the `hash` method. This way we don't have to create a new instance of `BftBlockHeaderFunctions` every time we want to use the CMS signature hash function.",
+      "reference": "Can this be moved to QbftBlockHeaderFunctions class as it is only used for qbft"
+    },
+    {
+      "prediction": "Why is this not implemented?",
+      "reference": "`Tags` now comes from the `PropertyBagHolder` base class."
+    },
+    {
+      "prediction": "`range_type` is not used in this method, so it can be removed.",
+      "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 think we can remove this import.",
+      "reference": "please revert all of this stuff in core.. I'll take another pass after. cheers and thanks for the help!"
+    },
+    {
+      "prediction": "nit: I think it's better to use `sourceTable` instead of `sourceName` here.",
+      "reference": "Why was this needed?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to call Prefetcher.PrefetchAfterBlockRetrieved. I think it should be called in BlockRetrievalQueue.GetWithPrefetch.",
+      "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 don't think this is correct. The domainConfig is not an instance of DomainConfig, it's an instance of DomainConfigInterface.",
+      "reference": "I do not understand this change"
+    },
+    {
+      "prediction": "I don't think we need to mention `aws.Config.LowerCaseHeaderMaps` here. We can just say \"By default keys received from the service api response will be formatted using net/http.CanonicalHeaderKey.\"",
+      "reference": "It would be helpful to clarify this is only for unmarshaling a response. not marshaling a request."
+    },
+    {
+      "prediction": "I'm not sure why this test is failing, but it seems like it should be passing. I'll look into it.",
+      "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": "I think this should be `#endif // HAVE_CONFIG_H`",
+      "reference": "Remove the \"rm\" from the end of \"HAVE_CONFIG_H\""
+    },
+    {
+      "prediction": "I'm not sure this is the best name for this struct, but I'm not sure what a better name would be.",
+      "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 `(char *)\"BOOL\"` instead of `(char *)\"BOOLEAN\"`.",
+      "reference": "Not sure why the ternary expressions are used here, since the true and false results are the same."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "reference": "Is `vertigo` another name for Vertica databases?"
+    },
+    {
+      "prediction": "I think this should be `urlstr = urlstr.lstrip('mailto:').lstrip('http://').lstrip('https://')`",
+      "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 we can remove this empty line.",
+      "reference": "Why the white space changes?"
+    },
+    {
+      "prediction": "Style/ClassAndModuleChildren: Use a guard clause in LicenseMailerPreview#fullfillment_error instead of wrapping the code inside a conditional expression.",
+      "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 thing to do. I think we need to hold the lock while we're talking to the SchedQueue, otherwise we could end up in a race condition where two goroutines try to add the same name to the SchedQueue at the same time. I think we need to hold the lock while we're adding the name to the SchedQueue, and then release the lock after we've added the name to the SchedQueue. That way, we can guarantee that only one goroutine will add the name to the SchedQueue at a time.",
+      "reference": "This is now outside the mutex, so could race."
+    },
+    {
+      "prediction": "I don't think we need to import `org.apache.lucene.util.BytesRefBuilder` here.",
+      "reference": "hmmm did this pass spotless check? I don't think we typically use wildcard imports"
+    },
+    {
+      "prediction": "I think this should be `lockBeforeLock *keybase1.LockID, lockAfterLock *keybase1.LockID`",
+      "reference": "`lockBeforeLock` -> `lockBeforeGet` (here and everywhere below)."
+    },
+    {
+      "prediction": "I think this should be `import net.sourceforge.pmd.lang.java.ast.JavaTokenTypes;`",
+      "reference": "Do we need to internalize net.sourceforge.pmd.lang.java.ast.JavaParserConstants on master, so that we can rename it?"
+    },
+    {
+      "prediction": "Why is this commented out?",
+      "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 no longer needed?",
+      "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_JOB_LOGGING_KAFKA_GLOBAL_DISABLE",
+      "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 think it would be better to return `true` in this case. If the platform does not support sharable mutexes, then it should be safe to assume that concurrent writes are supported.",
+      "reference": "Maybe change into `return _concurrentWrites ?? PlatformDetector.SupportsSharableMutex` ?"
+    },
+    {
+      "prediction": "I don't understand why this is needed.",
+      "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 if this is the right thing to do, but I think it makes sense to return an error if we can't encode the error. If we can't encode the error, then we don't know how to send it to the client, so we should return an error.",
+      "reference": "Do we still need to return the `Response` here?"
+    },
+    {
+      "prediction": "I think this should be `global.VERDACCIO_API_URL = 'http://localhost:4873/-/verdaccio/'`",
+      "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 this should be `--auth-url` instead of `--auth_url` to be consistent with the rest of the options.",
+      "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 `DD_TRACE_CONVENTION` instead of `DD_CONVENTION`",
+      "reference": "This should be `DD_TRACE_CONVENTION` to follow our ... conventions."
+    },
+    {
+      "prediction": "Why do we need to remove `getters` here?",
+      "reference": "Be careful memory leaks. memory leaks occur when getters.getAliasProp == nullptr . right?"
+    },
+    {
+      "prediction": "nit: `lastOrphanSubsCheck` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTimestamp` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSubsCheckTime` or `lastOrphanSub",
+      "reference": "Why not just a time.Time?"
+    },
+    {
+      "prediction": "why do we need to convert gt_masks to tensor?",
+      "reference": "Please do not submit a commit that is not part of this PR."
+    },
+    {
+      "prediction": "nit: I think it would be better to use `updatedInfo.Description != \"\"` instead of `updatedInfo.GetDescription() != \"\"`",
+      "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 don't think we should be returning an error here. If we can't find the CNAME, we should just return the original domain.",
+      "reference": "Can you open a separate PR with this patch? It seems valuable outside the context of this PR!"
+    },
+    {
+      "prediction": "I think this should be `/chunks/{address}` instead of `/chunk/{address}`",
+      "reference": "I would suggest to have plurals in the api `\"/chunks/{address}\"`"
+    },
+    {
+      "prediction": "I think this should be `BioMassCalc.C13` instead of `BioMassCalc.C14`.",
+      "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 think this should be `approval.mini_approval('Accept', event, %s);`",
+      "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": "Shouldn't this be `if (file_exists($carousel->getUploadDir() . DS . $carousel->getFile())) {`?",
+      "reference": "Adding a warning or an error in the Thelia log would be a nice idea."
+    },
+    {
+      "prediction": "I think this should be `const CONNECTION_EVENTS = Object.freeze(['error', 'close', 'timeout', 'parseError', 'connect', 'message']);`",
+      "reference": "nit: `Set` (and swap `forEach`s for `for (const i of CONNECTION_EVENTS)`)"
+    },
+    {
+      "prediction": "I think we can remove the `using System.Runtime.InteropServices;` line.",
+      "reference": "nit: Is this `using` statement still needed?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "Dont need to cast `string` -> `string`"
+    },
+    {
+      "prediction": "I don't think we should be logging this at debug level.",
+      "reference": "why doesn't the API return an error in this case?"
+    },
+    {
+      "prediction": "This should be `NoRegionError(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": "Redundant `self` detected.",
+      "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 should use `Map<Integer, PartitionField>` instead of `Map<String, PartitionField>`.",
+      "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": "nit: I think we can remove the `or None` as it is the default value.",
+      "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'm not sure why this is necessary. I don't see any padding applied to the `Settings` component.",
+      "reference": "Please, use the default padding here."
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "reference": "let's actually use the direct include: `require('../core/error').MongoError;`"
+    },
+    {
+      "prediction": "I think this should be `last-opened` instead of `last-focused`. I don't think we want to open new tabs in the most recently focused window, but in the most recently opened one.",
+      "reference": "You're missing a space before the `\"` here and below, but I'll fix it up when merging."
+    },
+    {
+      "prediction": "I think this should be `void dump(const STR_VECT &p) {`",
+      "reference": "remove this debugging code?"
+    },
+    {
+      "prediction": "`addressOffset` should be renamed to `streamAddressOffset` or something like that.",
+      "reference": "s/`terms text`/`terms's text`"
+    },
+    {
+      "prediction": "`replaceResultInFormWithNew` is a bit of a mouthful. How about `replaceResultInFormOnSuccess` or `replaceResultInFormAfterSuccess`?",
+      "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'm not sure why we need `show_email_and_username_form` here, but it seems to be necessary for the test to pass.",
+      "reference": "Line is too long. [87/80]"
+    },
+    {
+      "prediction": "I don't think you need to import `argparse` here.",
+      "reference": "Are you using an automated tool for sorting the imports?"
+    },
+    {
+      "prediction": "I think this should be `project_id` instead of `project_number`.",
+      "reference": "If this is `project_number` now, can we also update the `id` the description?"
+    },
+    {
+      "prediction": "I don't think we need to stop the timer here. It will be stopped when the stream is closed.",
+      "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 `application/x-msgpack` instead of `application/msgpack` (see https://github.com/dotnet/corefx/blob/master/src/System.Net.Http/src/System/Net/Http/Headers/HttpMediaTypeHeaderValue.cs#L125)",
+      "reference": "Should we rename the traces parameter now that this isn't just traces?"
+    },
+    {
+      "prediction": "I think this should be `DRM_MODE_ATOMIC_ALLOW_MODESET` instead of `DRM_MODE_ATOMIC_TEST_ONLY`.",
+      "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; point->surface_destroy.notify = NULL; point->surface_destroy.link.prev = NULL; point->surface_destroy.link.next = NULL; point->surface_destroy.link.container = NULL; point->surface_destroy.link.list = NULL; point->surface_destroy.notify = NULL; point->surface_destroy.link.prev = NULL; point->surface_destroy.link.next = NULL; point->surface_destroy.link.container = NULL; point->surface_destroy.link.list = NULL; point->surface_destroy.notify = NULL; point->surface_destroy.link.prev = NULL; point->surface_destroy.link.next = NULL; point->surface_destroy.link.container = NULL; point->surface_destroy.link.list = NULL; point->surface_destroy.notify = NULL; point->surface_destroy.link.prev = NULL; point->surface_destroy.link.next = NULL; point->surface_destroy.link.container = NULL; point->surface_destroy.link.list = NULL; point->surface_destroy.notify = NULL; point->surface_destroy.link.prev = NULL; point->surface_destroy.link.next = NULL; point->surface_destroy.link.container = NULL; point->surface_destroy.link.list = NULL; point->surface_destroy.notify = NULL; point->surface_destroy.link.prev = NULL; point->surface_destroy.link.next =",
+      "reference": "Need to remove the surface destroy listener (and `wl_list_init` it so that `touch_point_destroy` still works)"
+    },
+    {
+      "prediction": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "create ticket to investigate this"
+    },
+    {
+      "prediction": "This function is not used outside of this file, so it doesn't need to be `PONY_API`.",
+      "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 don't think we need to add this to the constants. It's only used in one place and we can just use the string \"DEV_FUSE_MOUNT\" there.",
+      "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 an array of `IntPtr` instead of a single `IntPtr`.",
+      "reference": "`bool` vs `BOOL` in native side :)"
+    },
+    {
+      "prediction": "I think this should be `program.command('lint').description('lint the code').action(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 think this should be `DBWriter.markItemPlayed(getActivity(), FeedItem.UNPLAYED, checkedIds.toArray());`",
+      "reference": "Tsts, that's why... activity leak about to happen"
+    },
+    {
+      "prediction": "Why do we need to do this?",
+      "reference": "You should do a select here IMO just to be safe so we do not block forever."
+    },
+    {
+      "prediction": "I think this should be renamed to `Outbox` to be consistent with the rest of the codebase.",
+      "reference": "I'll replace this with the actual outbox object soon."
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "Providers are called *before* `setUp()`, so I don't think you can access properties here."
+    },
+    {
+      "prediction": "I think this should be `AttributeTargets.Method | AttributeTargets.Class | AttributeTargets.Assembly`",
+      "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 should be `that.is.a('string')` instead of `that.is.empty`.",
+      "reference": "I think these assertions want to be `.to.not.exist` which would pass for `null` or `undefined`"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the only way to get the example to compile.",
+      "reference": "Don't think you need this since this is not a URL example."
+    },
+    {
+      "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.<br>",
+      "reference": "Maybe a test to check if the generated value is within the boundaries would be nice, what do you think?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we want to use UTF-8 with BOM.",
+      "reference": "Suppresses UTF-8 BOM in outputs"
+    },
+    {
+      "prediction": "`Inbound` and `Outbound` would be more consistent with the rest of the codebase.",
+      "reference": "order wrong for \"outgoing and incoming\""
+    },
+    {
+      "prediction": "I think this should be `private final boolean isMemoryCircuitBreakerEnabled;`",
+      "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": "I think this should be `null: false, default: false`",
+      "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 this should be `logParamsMutex.RLock()` and `logParamsMutex.RUnlock()` instead of `Lock()` and `Unlock()`.",
+      "reference": "Why is this called update params when it's updating project info? Should p called project?"
+    },
+    {
+      "prediction": "Why do we need `ON DUPLICATE KEY UPDATE` here?",
+      "reference": "`upsertHistoryTreeQuery` is a better name for this query now. Is it ok to change history?"
+    },
+    {
+      "prediction": "I think it would be better to have `gitTag = gitTag.Insert(5, \"b\" + gitBranch);`",
+      "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": "I don't think this is needed anymore.",
+      "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 `RootfsPropagationMode`",
+      "reference": "Shouldn't this be something like `rootmountmode` to fit the pattern of the other fields' serialized representations?"
+    },
+    {
+      "prediction": "I don't think we need to remove this.",
+      "reference": "Why were we initializing here in the first place?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be `new Logger((logLevel, messageFunc, exception, formatParameters) => false);`",
+      "reference": "This should be removed, and instead set the logger to None."
+    },
+    {
+      "prediction": "I think this should be `getSourceDataArray()` instead of `getSourceData()`. `getSourceData()` returns an object with `columns` and `rows` properties, but `getSourceDataArray()` returns an array.",
+      "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 think this should be `https://cwiki.apache.org/confluence/display/TRAFODION/Tutorial%3A+The+object-oriented+UDF+interface#Tutorial:Theobject-orientedUDFinterface-DebuggingUDFcode`",
+      "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 don't think we need to convert RetryTaskV2Failure to RetryTaskFailure anymore.",
+      "reference": "maybe call it `fromRetryTaskV2Failure`"
+    },
+    {
+      "prediction": "This import is not used.",
+      "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": "Why do we need to do this?",
+      "reference": "We suggest keeping this logic in lines between 173-175 to restrict its influence."
+    },
+    {
+      "prediction": "I think this should be `@return string|null|GridView`",
+      "reference": "why not ? maybe another phpstan plugin"
+    },
+    {
+      "prediction": "I think it would be better to use `TimeSpan.FromMilliseconds(_jsonRpcConfig.TracerTimeout * 1000)` instead of `TimeSpan.FromMilliseconds(_jsonRpcConfig.TracerTimeout)`",
+      "reference": "That was a really bad CR from me before if I did not spot it."
+    },
+    {
+      "prediction": "Is there a reason to use `/usr/bin/flock` instead of just `flock`?",
+      "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, but I think it should be restored.",
+      "reference": "Does this happen somewhere else now?"
+    },
+    {
+      "prediction": "I'm not sure this is the right way to do it, but it seems to work.",
+      "reference": "Remove trailing space at this line, amend the commit, and then force-push."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "This should be grouped together with the import of \"go.opentelemetry.io/{api/core,sdk/export}\" below."
+    },
+    {
+      "prediction": "I think we should rename this to `ke_policy_scanner` and `ke_policy_rules.yaml` to be consistent with the other scanners.",
+      "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 think we should keep the `label` in the `for` loop.",
+      "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": "`other arguments should not be used` -> `other arguments except index should not be used`",
+      "reference": "nit: `.` at the end of line?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. If the message count is 0, it means there are no messages in the folder, so we should show the \"no email\" message. If the message count is greater than 0, then we should show the \"loading more\" message.",
+      "reference": "I'm not sure I agree that this should take priority over \"Loading\"."
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I think it should be added back in.",
+      "reference": "Was this an intentional deletion? Why not just allow overwrite of `deploy_mode`?"
+    },
+    {
+      "prediction": "I think we should return the error here and let the caller handle it.",
+      "reference": "better to call it as syncCVRStatus.. just syncCVR is very confusing"
+    },
+    {
+      "prediction": "nit: I think we can remove the `deleteAppVars` struct and just use `deleteAppOpts` instead.",
+      "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 it would be better to use `kinto.core.storage.redis` instead of `kinto_redis.storage`.",
+      "reference": "Does this mean that `kinto_redis` is required to run tests?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do. The original indexes are the indexes of the operations in the original batch, not the indexes of the operations in the batch that was actually sent to the server. So if the server returns an error with an index, we need to map that index back to the original batch.",
+      "reference": "will this ensure the indexes for ordered writes as well?"
+    },
+    {
+      "prediction": "I think this should be `BooleanArray` instead of `BoolArray`.",
+      "reference": "Put `BoolArray` first?"
+    },
+    {
+      "prediction": "I think this should be `appMode = 'standalone'` instead of `window.appMode = 'standalone'`.",
+      "reference": "I believe this whole file can be deleted now."
+    },
+    {
+      "prediction": "I think it would be better to have a default value of 100000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "We should specify format for a user. I.e: Integer, 1000 == 1 MYST"
+    },
+    {
+      "prediction": "I think this should be `if err := controller.Run(ctx, sc); err != nil { return errors.Wrap(err, \"controller\") }`",
+      "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 `if (file_identifier_.length() && opts.prefix_size) { builder_.FinishSizePrefixed(Offset<Table>(toff), file_identifier_.c_str()); } else { builder_.Finish(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'm not sure if this is the right place to set the tab indicator color, but it seems to work.",
+      "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 this should be `return this._localHooks[key];`",
+      "reference": "Could you add `s` to `@return`?"
+    },
+    {
+      "prediction": "I think we can remove the `when /^Path=(.+)$/` block and the `path` variable.",
+      "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": "`msg.GetContent()` returns `interface{}` type, so we need to check the type of `msg.GetContent()` before calling `json.Marshal(msg.GetContent())`.",
+      "reference": "Will cause any bugs without this change? I've seen this code block(L293-302) many times... ;-)"
+    },
+    {
+      "prediction": "I think this should be `public IReadOnlyDictionary<string, object> Properties { get { return _contextProperties ?? new ReadOnlyDictionary<string, object>(new Dictionary<string, object>()); } }`",
+      "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 skipped?",
+      "reference": "also seems we should not skip this test"
+    },
+    {
+      "prediction": "nit: `deletes` should be `deleteFiles` to be consistent with the rest of the codebase.",
+      "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 don't think we need to fetch all the flows that are queued for a long time. I think we should only fetch the flows that have been in the queue for more than 24 hours.",
+      "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 `if ci.StartupScript != \"\" && !sourceExists(ci.StartupScript) {`",
+      "reference": "Are you making it so startup script HAS to be in sources?"
+    },
+    {
+      "prediction": "I think this should be `else if (vals.LocalWidth() == 1 || vals.LDim() == vals.LocalHeight()) {`",
+      "reference": "Might be more clear to just have one if statement?"
+    },
+    {
+      "prediction": "I don't think we need the `exc_info=True` here.",
+      "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": "This function is not used in this PR.",
+      "reference": "Perhaps rename to `render_service_args_struct`?"
+    },
+    {
+      "prediction": "I think we can remove the `crypto2` import here.",
+      "reference": "clean the name"
+    },
+    {
+      "prediction": "I think we should also check for `http.StatusTooManyRequests` and `http.StatusRequestTimeout` as well.",
+      "reference": "I think we would expect StatusNotFound instead of StatusForbidden?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "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 we can remove the `DockerClient` field from the `ResourceFields` struct and pass the `DockerClient` as a parameter to the functions that need it.",
+      "reference": "Is it a concern that we initialize these fields irrespective of whether resources like cgroup/volumes are enabled or not?"
+    },
+    {
+      "prediction": "I don't think we should panic here. I think we should return an error and let the caller decide what to do with it.",
+      "reference": "Wouldn't this be the first panic in yarpc? What do we do for transport validation? Return errors?"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "have you envisioned other cancellation reasons?"
+    },
+    {
+      "prediction": "`class-string` is not a valid PHPStan type.",
+      "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": "Shouldn't this be a `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": "I think this should be `aws-tags` to be consistent with the rest of the flags.",
+      "reference": "This might be very confusing. Maybe `resource-tags`? I"
+    },
+    {
+      "prediction": "I think this should be `_pair(self.dilation)` instead of `self.dilation`.",
+      "reference": "We may use `_pair` to wrap `dilation`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in libcontainer.",
+      "reference": "what's with `setReadonly`? Why we deleted it totally?"
+    },
+    {
+      "prediction": "I think this should be `static const char * szTypename = \"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 don't think we need to import the ec2 sdk here.",
+      "reference": "Is this excessive?"
+    },
+    {
+      "prediction": "I think this should be `\"/etc\"` instead of `\"etc\"`.",
+      "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 think this should be `protected abstract Long sequenceNumber();`",
+      "reference": "I think this needs a more specific name, like `sequenceNumberOverride`"
+    },
+    {
+      "prediction": "I think this should be `connection_config[:adapter]` instead of `connection_config.fetch(:adapter)`. `fetch` will return `nil` if `:adapter` is not present in `connection_config`.",
+      "reference": "Ought we just raise an exception if the adapter isn't defined?"
+    },
+    {
+      "prediction": "I'm not sure why this was added in the first place. I don't see it referenced anywhere in the code.",
+      "reference": "Should this have been removed? We still set the region to \"aws-global\" in v3model.go#L115"
+    },
+    {
+      "prediction": "I think this should be `errno=errors.ERRORS.INVALID_PARAMETERS` if `qs` is empty.",
+      "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 don't think we need this anymore.",
+      "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 it would be better to pass in the `dataresidency.StoreSingleton` as a parameter to `NewConstructor` rather than creating it inside of `NewConstructor`.",
+      "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": "I think this should be \"0.######\" to be consistent with the other values.",
+      "reference": "This is probably more digits than needed - perhaps rename OneOverK0 to IonMobility instead"
+    },
+    {
+      "prediction": "I think these should be `FlattenedMethodAsyncCallSettings` and `FlattenedMethodAsyncCancellationToken`",
+      "reference": "End with 'Method', so: - FlattenedAsyncCallSettingsMethod - FlattenedAsyncCancellationTokenMethod"
+    },
+    {
+      "prediction": "nit: I think we can remove the `internal/secrets` import here.",
+      "reference": "`localsecrets` should be updated in the same way."
+    },
+    {
+      "prediction": "I think this should be `ConsensusV17`",
+      "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'm not sure this is the right thing to do, but I think we need to return an error from this function, so that the caller can decide what to do in the case of a failure.",
+      "reference": "Use `FullBlock`, it comes from the same package."
+    },
+    {
+      "prediction": "I think we can remove this import, as it's not used anymore.",
+      "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 `if as(&objp) { ... } else { return errors.New(\"Writer.As failed to get ObjectHandle\") }`",
+      "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": "Why do we need to build index for genesis block?",
+      "reference": "Error return value of `s.idx.BuildIndex` is not checked"
+    },
+    {
+      "prediction": "I think this should be `elif member in [configtypes.List, configtypes.LengthList, configtypes.Tuple]:`",
+      "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": "Why are we ignoring the error here?",
+      "reference": "Its this golint or govet stuff?"
+    },
+    {
+      "prediction": "nit: I think you can just use `DR_MC_ALL` instead of `DR_MC_ALL` and `DR_MC_INTEGER`",
+      "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 think we should add a benchmark for `Rgba32.Equals(Rgba32)` and `Rgba32.Equals(Rgba32, Rgba32)` as well.",
+      "reference": "This is just a random struct equality, no value here."
+    },
+    {
+      "prediction": "I think this should be `config:\"portrange;local\"`",
+      "reference": "Just spotted the `local` on here; that shouldn't be needed - no reason to limit this config to env vars only"
+    },
+    {
+      "prediction": "This import is not used in this file.",
+      "reference": "I had some weird issues when importing ``from .. import util`` getting the wrong utilities, hence I did this."
+    },
+    {
+      "prediction": "I think this should be `# Copyright (C) 2007-2021 NV Access Limited, Bill Dengler`",
+      "reference": "Please remove this line."
+    },
+    {
+      "prediction": "Line is too long. [116/80]",
+      "reference": "should this state that we're falling back to `:acceptable_exit_codes`?"
+    },
+    {
+      "prediction": "I think this should be `value.head().to_markdown()` and `value.tail().to_markdown()` instead of `dataframe.head().to_markdown()` and `dataframe.tail().to_markdown()`",
+      "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 this is the right thing to do. I think we should be able to have more than one standalone server in the topology description.",
+      "reference": "Why was it necessary to add this check against `this.servers.size`?"
+    },
+    {
+      "prediction": "Shouldn't this be `<= 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'm not sure if this is the right thing to do, but I think it's better than throwing an exception when `request.Method` is `null`.",
+      "reference": "Short circuits when it's something like a WebPackRequest, which we should instrument and test specifically."
+    },
+    {
+      "prediction": "I think this should be `use-disk-cert-cache-dev` so that we don't accidentally enable it in production.",
+      "reference": "Should this be true now by default?"
+    },
+    {
+      "prediction": "typo: stastics -> statistics",
+      "reference": "/DispatcherStatistics ? /statistics is a bit confusing with /stats"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `tail` command itself.",
+      "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 think this should be `fake-cluster-id` instead of `%s` for the `openshiftClusterID` field.",
+      "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 rename this function to `get_node_first_ancestor_of_type_and_its_predecessor`.",
+      "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 `t.Errorf(\"controller.Register failed (%s)\", err.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 think this should be `self.live = False` and `master.addons(\"kill\", self)` should be `master.addons.remove(\"kill\", self)`.",
+      "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 don't think this is correct. If `field.native_shared` is true, then `{{FIELD_NAME}}` will be a `std::shared_ptr<std::string>`, and `CreateString` expects a `const char*`.",
+      "reference": "this line can be lifted outside of the if-then? In fact only CreateString / CreateSharedString needs to be inside of it."
+    },
+    {
+      "prediction": "Why is this commented out?",
+      "reference": "I think the `EpisodesApplyActionFragment` is unused now. So please delete it :)"
+    },
+    {
+      "prediction": "I think this should be `debug, trace, info, warn, error, fatal, any.`",
+      "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": "Can you rename this to `oomWaitingJobMeter`?",
+      "reference": "This metrics is only exposed in executors. Should it be defined in azkaban.execapp.ExecMetrics instead?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work for now.",
+      "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": "Why do we need this?",
+      "reference": "I think this should never happen and should return an error also. Having `WorkflowBackoffTimerTask` with `UNSPECIFIED` type looks weird."
+    },
+    {
+      "prediction": "I think this should be `if (e.keyCode === 13 || (e.keyCode === 32 && !browser.tizen))`",
+      "reference": "Add `browser` import"
+    },
+    {
+      "prediction": "I think this should be `GenerateSHA256PasswordHash(plaintextPassword string) (string, error)` and return an error if the password is empty.",
+      "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 don't think this should be a public API.",
+      "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 think this should be `solver_ = boost::make_shared<SGDSolver<float> >(param_file);`",
+      "reference": "I've never seen this `boost::make_shared`...why use it instead of `reset`?"
+    },
+    {
+      "prediction": "I think this should be `return getDispatchMethod(exflow);`",
+      "reference": "Wasn't this added as a part of different PR already?"
+    },
+    {
+      "prediction": "I think it should be `FriendlyUrl[]` instead of `FriendlyUrl[][]`",
+      "reference": "hmmmmm, should not be in constructor in this case `$this->toDelete = [[]];` ??? maybe not"
+    },
+    {
+      "prediction": "I think this should be `if status != 200:`",
+      "reference": "We can just update it to be as follows: `if response: print(\"Success\") else: print(\"Error\")`"
+    },
+    {
+      "prediction": "I think this should be a `map[string]interface{}` instead of a `map[string]interface{}`.",
+      "reference": "Maybe this should be exported so (external/third-party) plugins can also use it."
+    },
+    {
+      "prediction": "I don't think we need to return an error here. If we can't enable IP forwarding, we should just log the error and continue.",
+      "reference": "Warning logging whould be exactly here in high logic. So that everybody understands why we swallow errors"
+    },
+    {
+      "prediction": "I don't think it's a good idea to pass `JsonRpcUrl` as a parameter to `JsonRpcWebSocketsServer` constructor. I think it would be better to have `JsonRpcUrl` as a property of `JsonRpcWebSocketsServer`.",
+      "reference": "if its optional, move it to last item an use JsonRpcUrl? url = null"
+    },
+    {
+      "prediction": "Metrics/MethodLength: Method client_model is too long. [21/20]",
+      "reference": "how is this different from the `client_slug` method already available on a `user` ?"
+    },
+    {
+      "prediction": "I don't think we need to log this at `finest` level.",
+      "reference": "I'd keep this at `info` level..."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "I believe this env var won't be set anywhere. Do we need this condition?"
+    },
+    {
+      "prediction": "I don't think we need to add this to the window object. We can just call `getParameterByName` directly.",
+      "reference": "why add function explicitly?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Is that intentional? If you don't create the cache here, it will never be created therefore used."
+    },
+    {
+      "prediction": "I'm not sure this test is doing what you want it to do. I think it should be something like `require.True(t, b.Before(now.Add(32*time.Second)), \"require that the given date for 10 retries is never more than the max interval\")`",
+      "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 don't think we need to change the prefix. I think it's fine to keep it as `/__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 don't think this is correct. I think it should be `if (blockNumber < ConstantinopleFixBlockNumber) { return ConstantinopleFix.Instance; } else if (blockNumber < IstanbulBlockNumber) { return Istanbul.Instance; } else if (blockNumber < ByzantiumBlockNumber) { return Byzantium.Instance; } else if (blockNumber < HomesteadBlockNumber) { return Homestead.Instance; } else if (blockNumber < Eip150BlockNumber) { return Eip150.Instance; } else if (blockNumber < Eip155BlockNumber) { return Eip155.Instance; } else if (blockNumber < Eip158BlockNumber) { return Eip158.Instance; } else if (blockNumber < SpuriousDragonBlockNumber) { return SpuriousDragon.Instance; } else if (blockNumber < ByzantiumBlockNumber) { return Byzantium.Instance; } else if (blockNumber < ConstantinopleBlockNumber) { return Constantinople.Instance; } else if (blockNumber < ConstantinopleFixBlockNumber) { return ConstantinopleFix.Instance; } else if (blockNumber < IstanbulBlockNumber) { return Istanbul.Instance; } else { return Frontier.Instance; }",
+      "reference": "need to add it for ropsten, rinkeby, goerli as well"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get `zfs` commands to work.",
+      "reference": "G204: Subprocess launching should be audited (from `gosec`)"
+    },
+    {
+      "prediction": "I think it's better to keep it as `private`.",
+      "reference": "curious why only making these 3 public. what about other package scope methods like `times`, `timestamps`, `array`?"
+    },
+    {
+      "prediction": "I think this should be `email` instead of `Email`.",
+      "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 don't think this is a good idea to have so many options. It's better to have a single option that enables all of them. Also, I don't think it's a good idea to have a single option that enables all of them. It's better to have a single option that enables all of them. Also, I don't think it's a good idea to have a single option that enables all of them. It's better to have a single option that enables all of them. Also, I don't think it's a good idea to have a single option that enables all of them. It's better to have a single option that enables all of them. Also, I don't think it's a good idea to have a single option that enables all of them. It's better to have a single option that enables all of them. Also, I don't think it's a good idea to have a single option that enables all of them. It's better to have a single option that enables all of them. Also, I don't think it's a good idea to have a single option that enables all of them. It's better to have a single option that enables all of them. Also, I don't think it's a good idea",
+      "reference": "Should it say `default 0` here? looks a bit confusing. Also following BODMAS we should put: `2048 * (signature checker threads + 1)`"
+    },
+    {
+      "prediction": "This import is not used in this file. Please remove it.",
+      "reference": "Add a newline. We group by 1: standard library, 2: 3rd party library, 3: application library"
+    },
+    {
+      "prediction": "I think this should be `RunQuery` instead of `Evaluate`.",
+      "reference": "So `queryRange.Step` will be ignored?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we need to make sure that we don't have any more `_X86_`, `_AMD64_`, `_ARM_`, or `_ARM64_` defines in the codebase.",
+      "reference": "It would be nice to replace these _HOST_* defines with HOST_*."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the best we can do for now.",
+      "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 `fastdds::dds::ParameterList`, not `eprosima::fastdds::dds::ParameterList`.",
+      "reference": "Don't use relative paths. src directory is already on the include search path"
+    },
+    {
+      "prediction": "I think this should be `namedTestTools map[string][]BuildInput `name:\"named_test_tools\"`",
+      "reference": "if we're introducing this now can we make this private?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this check. I think it would be better to do this check in the `IonMobilityFinder` class, where the `fileInfos` are created.",
+      "reference": "Something is wrong with the indentation here."
+    },
+    {
+      "prediction": "nit: can you sort the imports alphabetically?",
+      "reference": "This should go to the external imports group below"
+    },
+    {
+      "prediction": "I think this should be `log.V(4).Infof(\"processing yaml folder %s\", p)`",
+      "reference": "Intended to be checked in?"
+    },
+    {
+      "prediction": "The schemas in new version are less than the old version, which may cause compatibility problems. -> ATTENTION: The schemas in new version are less than the old version, which may cause compatibility problems.",
+      "reference": "I think this warning message is the same as above one. Anyway, it's fine to keep it."
+    },
+    {
+      "prediction": "I don't think it's a good idea to make this change. The `logsDir` variable is used in other places as well, and it's not guaranteed to be the same as the `logs` folder in the root of the project. I think it would be better to keep the original code and just make sure that the `logs` folder is created in the root of the project. What do you think?",
+      "reference": "This is done few times, replace with some well named method"
+    },
+    {
+      "prediction": "I think this should be `options.set(ToolOptions.GENERATOR_CONFIG_FILES, Lists.newArrayList(generatorConfigs));`",
+      "reference": "`configs` isn't very descriptive, especially when there are other config-ish things like `generatorConfigs`."
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to be the only way to make the test pass.",
+      "reference": "In my point we have to check conversion with specific (non-ASCII) characters in unicode string."
+    },
+    {
+      "prediction": "I think this should be `(\"python\", \"UnitTestExcludedVolume.py\", {}),` instead of `(\"python\", \"UnitTestExcludedVolume.py\", {})`",
+      "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 `ATTR_SELF_SERVE_ENABLED` or `ATTR_SELF_SERVE_ENABLED` to make it clear that it's a boolean.",
+      "reference": "we should keep the camel case format - selfServe"
+    },
+    {
+      "prediction": "nit: can we rename this to `metadataItems` to be consistent with the rest of the codebase?",
+      "reference": "consider the alias `compute` in case this ever references other apis."
+    },
+    {
+      "prediction": "I think this should be `return cgroups.PathExists(m.paths[\"devices\"]) || cgroups.PathExists(m.paths[\"memory\"]) || cgroups.PathExists(m.paths[\"cpuacct\"]) || cgroups.PathExists(m.paths[\"cpu\"]) || cgroups.PathExists(m.paths[\"blkio\"]) || cgroups.PathExists(m.paths[\"pids\"]) || cgroups.PathExists(m.paths[\"net_cls\"]) || cgroups.PathExists(m.paths[\"net_prio\"]) || cgroups.PathExists(m.paths[\"freezer\"]) || cgroups.PathExists(m.paths[\"hugetlb\"]) || cgroups.PathExists(m.paths[\"rdma\"]) || cgroups.PathExists(m.paths[\"perf_event\"]) || cgroups.PathExists(m.paths[\"cpuset\"])`",
+      "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 think this should be `gaAMPPropertyID` and `gaPropertyID` instead of `gaAMPPropertyID` and `gaPropertyID`.",
+      "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 think this should be `this.emit('executing')` instead of `this._fetchStatus({ diff: false })`",
+      "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 think this should be `if err != nil && m.rootless && m.config.Path == \"\" { return nil }`",
+      "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": "I think it would be better to call it `MaxDBSize` or `MaxDBCapacity` or something like that. `DbCapacity` is a bit confusing.",
+      "reference": "`DbCapacity` -> `DBCapacity`"
+    },
+    {
+      "prediction": "I think it's better to have a separate test case for London.",
+      "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 don't think we should be using `sudo` here. I think we should be using `sh.RunV(\"sudo\", \"build/myst/myst\", \"service\", \"--agreed-terms-and-conditions\", \"socks5\")`",
+      "reference": "Does `sudo` required for `darwin` only? Don't we need it for `linux` too?"
+    },
+    {
+      "prediction": "nit: I think we can remove this constant since we are not using it anywhere.",
+      "reference": "Can you add description to this exported struct?"
+    },
+    {
+      "prediction": "Why do we need to copy the previous state here?",
+      "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": "This should be `github.com/aws/aws-sdk-go/service/elbv2/elbv2iface`",
+      "reference": "Should this maybe directly start with ELBv2 aka NLBs or ALBs?"
+    },
+    {
+      "prediction": "I think this should be a `404` instead of a `200` with an empty body.",
+      "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'm not sure if this is the right place to put this. I think it should be in `src/utils/networkError.js` but I'm not sure if it should be a class or just a function that returns an error object.",
+      "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": "nit: I think we can remove the `GrpcGapicRetryMapping retryMapping` parameter from this method. It's only used in `GapicMethodConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicServiceConfigBuilder.build()`, which is only called from `GapicService",
+      "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": "I think we should add a check for `strlen(tmp) > 0` to avoid passing an empty string to `flb_standard_chain_provider_create()`.",
+      "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 it would be better to have `fill(int n, T element)` and `fill(int n, Function<? super Integer, ? extends T> f)`.",
+      "reference": "(...), where each element ~are~ **is the** given {\\@code element}."
+    },
+    {
+      "prediction": "nit: sortOrderId -> sortOrderIndex",
+      "reference": "I think this can be an int because we have a default that is always valid, `0`."
+    },
+    {
+      "prediction": "Style/RedundantSelf: Redundant self detected.<br>Metrics/BlockLength: Section#upcoming? has too many lines. [31/25]",
+      "reference": "I think you can use `build_stubbed` here for the same results but with more speed."
+    },
+    {
+      "prediction": "I think this should be `isinstance(rampup, numeric_types) and isinstance(hold, numeric_types) and rampup > 0 and hold > 0`",
+      "reference": "what if one of them is property and one is not?"
+    },
+    {
+      "prediction": "Why do we need to ignore this error?",
+      "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 think this should be `const owner = {}` and `session = topology.startSession({ owner });`",
+      "reference": "can't this just be left undefined/null?"
+    },
+    {
+      "prediction": "I'm not sure if we want to keep the `response` variable here, since it's not used.",
+      "reference": "changed bc we are using the `error` var below"
+    },
+    {
+      "prediction": "I think it's better to use `assertEqual` or `assertNotEqual` instead of `==` or `!=`.",
+      "reference": "Is there a reason why this is change is in the same commit?"
+    },
+    {
+      "prediction": "I think this should be `exec.Command(\"cp\", \"-rf\", r.dir, dest)` instead of `copyCommand(r.dir, dest)`.",
+      "reference": "@nghialv btw, perhaps was this method originally created for copying the repo root to the given `dest` as a subdirectory?"
+    },
+    {
+      "prediction": "I'm not sure why this was changed to `IProgressStatus` instead of `ProgressStatus`.",
+      "reference": "Just curious - why not var?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to load the image, but it's the only place I could find where the image is used.",
+      "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 errors [Integer] the number of errors that have occurred processing the spec suite`",
+      "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": "This bundle is not used anymore, so we can remove it.",
+      "reference": "there was no usage of the bundle in the framework? everything works exactly the same after the removal?"
+    },
+    {
+      "prediction": "I think this should be `infra.RemoveNodeAddresses(felixes[0])` instead of `infra.RemoveNodeAddresses(felixes[0])`",
+      "reference": "Why this difference?"
+    },
+    {
+      "prediction": "I think we should use `snprintf` instead of `sprintf` to avoid buffer overflows.",
+      "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 think this should be `AccountHistoryPrefix` instead of `AccountMaxVersionPrefix`",
+      "reference": "`AccountMaxVersionPrefix` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think we can remove this import as well.",
+      "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 think this should be `exceptions.BackendError` instead of `exceptions.ReadonlyError`.",
+      "reference": "Nit: Maybe merge these into one string while you are here?"
+    },
+    {
+      "prediction": "I don't think we need this anymore.",
+      "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 `ipvsSupportEnabled` instead of `kubeIPVSSupportEnabled` to be consistent with the rest of the code.",
+      "reference": "tiniest of nits: why would you abbreviate \"interface\" but not \"configuration\" ?"
+    },
+    {
+      "prediction": "Why is this not needed anymore?",
+      "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 a good fallback. I think it would be better to just return `None` and let the caller decide what to do in that case.",
+      "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 don't think this is the right thing to do. If the YAML file has a syntax error, we should raise a `CommandException` with the error message from `yaml.scanner.ScannerError`, not a `CommandException` with the error message from `yaml.scanner.ScannerError` prefixed with \"Syntax error while reading {!r}\".",
+      "reference": "duplicate code - can you move into build_from_path() ?"
+    },
+    {
+      "prediction": "Shouldn't this be `disable=invalid-name`?",
+      "reference": "very important test as we want to make sure the old name still work."
+    },
+    {
+      "prediction": "I think this should be `return 0`",
+      "reference": "I think returning 1 should some how be moved in to specific support for LibreOffice.."
+    },
+    {
+      "prediction": "I think this should be `static bool d_r_avx512_code_in_use = false;`",
+      "reference": "Should be inside `ifdef X86` I would think (or moved to ... I guess arch/x86/emit_utils.c)"
+    },
+    {
+      "prediction": "I don't think we need to store all the services in the cache. We only need to store the services that are used by the pod.",
+      "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 `^api[^a-zA-Z0-9_]*version`",
+      "reference": "nit, does it make sense to get this variables outside the func where they are used?"
+    },
+    {
+      "prediction": "This should be `sns-ruby-example-show-subscriptions.rb demonstrates how to list subscriptions to the Amazon Simple Notification Services (SNS) topic using the AWS SDK for Ruby.`",
+      "reference": "to **an** Amazon... Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "Nit: is \"top\" a short name we use frequently? Seems a bit mysterious, and saves only 2 chars."
+    },
+    {
+      "prediction": "I think it's better to use `const ProtoVersion = 4` instead of `const ProtoVersion = cassandraProtoVersion`.",
+      "reference": "This name is really confusing. May be just remove this const completely and hardcode `4` where it is used?"
+    },
+    {
+      "prediction": "I think we should keep this test, but change the expectation to `expect(config.files_to_run).to contain_exactly(file)`.",
+      "reference": "Why did you remove this spec?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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 it would be better to use `kwargs.get('composite_root_resources', '').split(',')` instead of `kwargs.get('composite_root_resources') or None`",
+      "reference": ".get will return None if there is no project_id key, so the or None part is redundant"
+    },
+    {
+      "prediction": "I think we can remove `longVal` and `doubleVal` here.",
+      "reference": "Do they need to be `internal`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's the right thing to do.",
+      "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 don't think this is correct. `hipMemcpy2D` and `hipMemcpyParam2D` are not the same thing. `hipMemcpy2D` is a macro that expands to `hipMemcpy2DAsync` with a NULL stream. `hipMemcpyParam2D` is a function that takes a `hipMemcpy2DParms` struct.",
+      "reference": "Is there any official CUDA API called cuMemcpy2D_v2?"
+    },
+    {
+      "prediction": "I think we can remove this class and use `ImageMgmtUtils.IMAGE_TYPE` and `ImageMgmtUtils.IMAGE_VERSION` instead.",
+      "reference": "Can you please move this interface as class in common Constants file in az-core module?"
+    },
+    {
+      "prediction": "I think this should be `::unlink(fs::FileUtils::dirname(pidFile.c_str()).c_str());` instead of `::unlink(fs::FileUtils::dirname(pidFile.c_str()).c_str());`",
+      "reference": "Why not TempDir?"
+    },
+    {
+      "prediction": "I think this should be `IMvxElement<TViewModel> : IMvxElement, IMvxView<TViewModel> where TViewModel : class, IMvxViewModel`",
+      "reference": "IMvxFormsView -> IMvxElement is a breaking change, isn't it?"
+    },
+    {
+      "prediction": "I think this should be `from molecule.provisioners import ansible_galaxy_install`",
+      "reference": "This should be imported alphabetically."
+    },
+    {
+      "prediction": "`Price` should be imported from `Shopsys\\FrameworkBundle\\Model\\Pricing\\Price`, not from `Shopsys\\FrameworkBundle\\Model\\Pricing\\PriceInterface`.",
+      "reference": "Please apply same changes as for Payment"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this hook. I think it should be in the `ScrollEffect` component itself.",
+      "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_str = token.token_str`",
+      "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": "This should be `m.Group(\"/branches\", func() { m.Get(\"\", repo.ListBranches) m.Get(\"/:id\", repo.GetBranch) m.Post(\"\", repo.CreateBranch) m.Patch(\"/:id\", repo.UpdateBranch) m.Delete(\"/:id\", repo.DeleteBranch) })`",
+      "reference": "Shouldn't we use `:branchname` here? Or just `:branch`?"
+    },
+    {
+      "prediction": "I'm not sure this is a good idea. The `AutoTime` class is intended to be used as a field, not as a value. I think it would be better to keep it as a field, and have it return a `RandNum` value. That way, you can still use it as a field, but you can also use it as a value if you want to.",
+      "reference": "`AutoTime` is a number but calling `int()` would make it crash. Tested through the docs"
+    },
+    {
+      "prediction": "I think it would be better to use `TryGetValue` instead of `FirstOrDefault` here.",
+      "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 this is the right thing to do. I think it should be `\"/\"` instead of `\"/*\"`, but I'm not 100% sure.",
+      "reference": "Please do the same for the DebugFilter a few lines above."
+    },
+    {
+      "prediction": "Do we need to export `setupRerender` from `@testing-library/react`?",
+      "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": "nit: I think it would be better to call it `useDefaultSources` instead of `createDefaultSources`",
+      "reference": "There are several uses of the word \"default\" in here, perhaps it can be reworded slightly?"
+    },
+    {
+      "prediction": "This should be `FormEvents::POST_SUBMIT`",
+      "reference": "`function (` is correct. A space must be there. Please run `make cs` to have correct coding style."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get this test to pass.",
+      "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 a good idea. If the subject is a large byte slice, it will be copied twice.",
+      "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 `kmsAliasPrefix` to be consistent with the rest of the fields.",
+      "reference": "Should this be optional for existing configs to work?"
+    },
+    {
+      "prediction": "I don't think this is the right place to set the available balance. The balance should be set when the cheque is received, not when it is issued.",
+      "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": "This should be `flatbuffers::IDLOptions::kJsonSchema` instead of `flatbuffers::IDLOptions::kJson`.",
+      "reference": "lets leave out the `-S` ?"
+    },
+    {
+      "prediction": "I think this should be `toLowerCase(Locale.ROOT)`.",
+      "reference": "Is this method needed?"
+    },
+    {
+      "prediction": "I don't think you need to do this. You can just use `this->layer_param_.pooling_param()` directly.",
+      "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 this should be \"desired to be running\" instead of \"desired to be running by a service\".",
+      "reference": "Does this mean it's only returning tasks associated with the primary deployment? or something else"
+    },
+    {
+      "prediction": "I think this should be `t.Logf(\"Orig Error: %#v of type %T\", origError, origError)`",
+      "reference": "nit, would be good to include prefix for this log statement."
+    },
+    {
+      "prediction": "I think this should be `['aria-bats', 'aria-valuenow', 'aria-valuemax', 'aria-valuemin']`",
+      "reference": "Same here, you should use a made up role. May I suggest `role=\"McCheddarton\"`? :)"
+    },
+    {
+      "prediction": "nit: I think it would be better to have a `validateNoConflictingDeletes` method in `OverwriteFiles` instead of `validateNoConflictingAppends`.",
+      "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": "nit: remove the extra blank 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 this is the right place to do this. I think it should be done in the `StartSpan` method of the `AppSecTracer` class.",
+      "reference": "`Tags.AppSecEnabled` should probably move to `Metrics.AppSecEnabled` I guess?"
+    },
+    {
+      "prediction": "I think this should be `super(atomic_file, self).__init__(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 thing to do, but I don't think we should be linking to Imperial anymore.",
+      "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": "@danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship @danwinship",
+      "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": "This should be `Command.W3C_MINIMIZE_WINDOW`",
+      "reference": "Update after command rename"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "What about adding an optional parameter that allows users to choose whether they want to use the format or not?"
+    },
+    {
+      "prediction": "I'm not sure why this was skipped, but I think it should be unskipped.",
+      "reference": "Lint didn't like this context being before the skip"
+    },
+    {
+      "prediction": "I don't think we should allow all origins. We should only allow `localhost:3000` and `localhost:3001`",
+      "reference": "so now basically anybody could drink my tequila? currently electron dev-env runs on `http://localhost:9080`"
+    },
+    {
+      "prediction": "I think this should be `mDesktopFile.name()` instead of `mDesktopFile.id()`. `id()` returns the full path to the `.desktop` file, while `name()` returns just the name of the plugin.",
+      "reference": "`mPluginLoader` is nullptr in static plugins, which causes segmentation faults"
+    },
+    {
+      "prediction": "This import is not needed anymore.",
+      "reference": "Better use `Optional<Consumer<String>> onSearch` here."
+    },
+    {
+      "prediction": "Why did you remove this test?",
+      "reference": "this is asserted as invalid input"
+    },
+    {
+      "prediction": "I think this should be `state.Graph` rather than `graph`.",
+      "reference": "Why do you need to create a new parser? Can't you just use `state.Parser` to answer this question?"
+    },
+    {
+      "prediction": "I don't think we need to create a new block here. We can just return `this`.",
+      "reference": "remove processing references from Nethermind.Core where possible, keep clean"
+    },
+    {
+      "prediction": "I think this should extend `MessagingException` instead of `Exception`.",
+      "reference": "Why not make this a subclass of `MessagingException` so we don't need to change all `throws` clause?"
+    },
+    {
+      "prediction": "I don't think you need to import `binascii` and `hmac_drbg` in this file.",
+      "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 think this should be `deletes.contains(schema.findField(parent + \".\" + name).fieldId())` instead of `deletes.contains(schema.findField(parent + \".\" + name).fieldId()) || deletes.contains(schema.findField(parent + \".\" + name).fieldId())`",
+      "reference": "Nit: indentation is 2 indents (4 spaces) from the previous line."
+    },
+    {
+      "prediction": "I think this should be `c.Platform` instead of `\"ddev\"`.",
+      "reference": "I don't understand why this is changing to a hard-coded string."
+    },
+    {
+      "prediction": "I think this should be `DevnetConfig` instead of `DevnetConfig`",
+      "reference": "nit: perhaps calling this `config` would be more clear?"
+    },
+    {
+      "prediction": "I think this should be `Lade %{smart_count} Dateien hoch` or `Lade %{smart_count} Dateien 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 0022",
+      "reference": "Hmm I would have expected this to go away? What role does it play now that we have a minimum?"
+    },
+    {
+      "prediction": "I don't think we need to remove these from the `GET_MORE_NON_RESUMABLE_CODES` set. I think we can just remove the `GET_MORE_NON_RESUMABLE_CODES` set and just use the `GET_MORE_RESUMABLE_CODES` set.",
+      "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 don't think this is needed anymore.",
+      "reference": "I think this mutex is only used here. Remove it."
+    },
+    {
+      "prediction": "I think it would be good to add a note here about the fact that the caller thread will be blocked until the task is completed.",
+      "reference": "it's unclear to me which method you are recommending overriding?"
+    },
+    {
+      "prediction": "I think this should be `if 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": "nit: please remove the extra blank line",
+      "reference": "Remove this file from the PR"
+    },
+    {
+      "prediction": "Why did you remove this?",
+      "reference": "I cannot find the changes / results of the eip2565 benchmark"
+    },
+    {
+      "prediction": "I think this should be `this.setState( { data: false, zeroData: true } );`",
+      "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": "Why not use `Task.Delay(0)` instead of `Task.FromResult(0)`?",
+      "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": "nit: `let err = null;`",
+      "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'm not sure if this is the right fix, but it seems to be the only way to get the tests to pass.",
+      "reference": "`action.ExecutionDataGas = 100`"
+    },
+    {
+      "prediction": "I don't think this is a legacy test.",
+      "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": "nit: remove empty line",
+      "reference": "getSpcResource should not be bothered about deletion. Can we have `separation of concerns` for this logic?"
+    },
+    {
+      "prediction": "I think the order should be `0, 1, 2, 3, 4, 5` instead of `0, 5, 1, 2, 3, 4`.",
+      "reference": "This is actually matched with pandas's result since the output is sorted."
+    },
+    {
+      "prediction": "I think this should be `ConfigDir = os.Getenv(\"HOME\") + \"/.config/ioctl/default\"`",
+      "reference": "struct field tag `yaml:\"endpoint\"\"` not compatible with reflect.StructTag.Get: key:\"value\" pairs not separated by spaces (from `govet`)"
+    },
+    {
+      "prediction": "I don't think you need to initialize `num_refs_` to 0. It will be set to 0 by the default constructor of `std::atomic<uint32_t>`.",
+      "reference": "This is not necessary. Topic inherits from DomainEntity which already has instance_handle_"
+    },
+    {
+      "prediction": "I think we can remove the `false` here.",
+      "reference": "Do we need the forXDP parameter in runBpfUnitTest? If not, I think better to revert in order to save a few changes."
+    },
+    {
+      "prediction": "can we rename this to `getCasSnapshot` ?",
+      "reference": "All the code that reference to this function must also be changed."
+    },
+    {
+      "prediction": "I think this should be `validateUserGroupMembership(String username, List<String> groupNames)` so that we can pass in a list of group names and check if the user is a member of any of them.",
+      "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 think it's better to use `Nethermind.Blockchain.BlockTree.BlockTree` instead of `Nethermind.Blockchain.BlockTree`.",
+      "reference": "Do we have namespace conflicts?"
+    },
+    {
+      "prediction": "I think we should pass `cfg.Cache.TTLDuration()` to `api.NewWebAPI` instead of `cfg.Cache.TTLDuration()` to `api.NewWebAPICache`.",
+      "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 it would be better to use `primary` and `secondary` instead of `master` and `slave`.",
+      "reference": "This is a BC break for extending classes, it can't be done on 3.x"
+    },
+    {
+      "prediction": "I think this should be `if server_side_encryption_customer_algorithm != 'AES256':` instead of `if server_side_encryption_customer_algorithm is None:`",
+      "reference": "Line length (PEP8, should be 80 chars max)"
+    },
+    {
+      "prediction": "I think this should be `private static final String HIGH_PRIORITY_TAG = \"X-Priority\";`",
+      "reference": "this is duplicate now. I suggest moving it into `MimeHeader`, we already have a bunch of other headers there"
+    },
+    {
+      "prediction": "This should be `fmt.Fprintln(os.Stderr, \"Your password contains leading/trailing whitespace - in previous versions of rclone this was stripped\")`",
+      "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": "I think this should be `MessageDlg.Show(this, message, Resources.SkylineWindow_FindIrtDatabase_The_database_file_specified_could_not_be_opened);`",
+      "reference": "I think \"this\" should be \"parent\"."
+    },
+    {
+      "prediction": "I think this should be `changed = TRUE;`",
+      "reference": "Why not init this at variable decl time instead?"
+    },
+    {
+      "prediction": "nit: I think this should be `ProjectIDFromEnv` or `ProjectIDFromEnvConfig` or something like that.",
+      "reference": "Do we still need this if we're doing everything through `envconfig`?"
+    },
+    {
+      "prediction": "I think this should be called `EnableIdentityCheck` or something like that. `IdentityCheck` sounds like it's a property of the node, not an option for the network.",
+      "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 works for now. We could also add a `DISABLE_SANDBOX_WARNING` environment variable to `config/environments/development.rb` and `config/environments/test.rb` and set it to `true`, but I'm not sure if that's the best way to do it.",
+      "reference": "Since we have the same logic in both forms, can we make a shared partial/helper for it?"
+    },
+    {
+      "prediction": "I don't think we need to store the market address in the key. We can just use \"ask_plans\" as the key.",
+      "reference": "Why to introduce this intermediate variable?"
+    },
+    {
+      "prediction": "I think this should be `client_type = luigi.ChoiceParameter(choices=['insecure', 'kerberos', 'token'], default='insecure', description='Type of client to use. One of insecure, kerberos or token')`",
+      "reference": "Maybe use ChoiceParameter?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "reference": "Not positive is this is needed, given the `TokenAuth` include"
+    },
+    {
+      "prediction": "This should be `NamedTuple`, not `Tuple`.",
+      "reference": "Why not use a docstring for it?"
+    },
+    {
+      "prediction": "I think it would be good to have a test for the case where one of the ports is 0 and the other is not.",
+      "reference": "Two more tests - no ports specified and every possible port specified - p2pPort, graphQLHttpPort, rpcHttpPort, rpcWsPort, metricsPort, metricsPushPort, stratumPort"
+    },
+    {
+      "prediction": "Please add your name in alphabetical order.",
+      "reference": "Don't worry about that next time, it's automated :)"
+    },
+    {
+      "prediction": "I don't think we need to disable pylint here.",
+      "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 if this is a breaking change or not, but I think it is. I think we should keep the `FLUENT_HOST` and `FLUENT_PORT` environment variables and add `OPEN_TELEMETRY_HOST` and `OPEN_TELEMETRY_PORT` environment variables.",
+      "reference": "Are these values expected to change over time? Why not inject these values in ECS backend instead of agent doing it?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set this env var. I think it should be set in `pkg/server/server.go` in the `NewServer` function.",
+      "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 this is the right thing to do. I think we need to fix these tests.",
+      "reference": "These seem to support the new test framework below, but the `.in.csv` files are still committed."
+    },
+    {
+      "prediction": "Why do we need this check?",
+      "reference": "!txReceipts[i].Removed and We can check it for first item outside of for loop, we don't expect mixed batches"
+    },
+    {
+      "prediction": "I think we should keep `githubURLFlag` and add `repoURLFlag` as an alias.",
+      "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 `Bolt::Error.unknown_plan(plan_name)` instead of `Bolt::Error.unknown_plan(plan_name)`",
+      "reference": "This provides file and line number if applicable."
+    },
+    {
+      "prediction": "Please remove unused import.",
+      "reference": "This is already imported. Please check the file."
+    },
+    {
+      "prediction": "I think this should be `GeckoSession.SelectionActionDelegate.ACTION_COPY.equals(action) && aSelection.isActionAvailable(GeckoSession.SelectionActionDelegate.ACTION_UNSELECT)`",
+      "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 think this should be `if len(hosts) == 0 {`",
+      "reference": "I don't think adding this value to _every_ connection policy is needed or useful... why is this necessary?"
+    },
+    {
+      "prediction": "I think it is better to use `Ergonode\\SharedKernel\\Domain\\Aggregate\\ProductId` instead of `Ergonode\\SharedKernel\\Domain\\AggregateId`",
+      "reference": "adding this constrain for null AggreagateId is redundant"
+    },
+    {
+      "prediction": "I think this should be `if act.RepoUserName == ctx.User.LowerName { repo.OwnerId = ctx.User.Id } else { repo.OwnerId, _ = models.GetUserByName(act.RepoUserName) }`",
+      "reference": "Why when current user is owner but still check access? Should we use `else if` here?"
+    },
+    {
+      "prediction": "I think this should be `10 * time.Second` or something like that. 60 seconds is way too long.",
+      "reference": "60 seems high. 10? 20?"
+    },
+    {
+      "prediction": "I think this should be `os.path.join(cfg_dir, 'Default.sqlite')` instead of `os.path.join(args.config_directory, '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": "nit: I think this should be `ControlPlaneLoadBalancerScheme` instead of `controlPlaneLoadBalancerScheme`",
+      "reference": "Let's make this a pointer given that's optional"
+    },
+    {
+      "prediction": "I think this should be `var CircularJSON = require('circular-json').stringify;`",
+      "reference": "Why is CircularJSON imported here? It is never used in this file."
+    },
+    {
+      "prediction": "I think we can remove this now, right?",
+      "reference": "Should be ARM"
+    },
+    {
+      "prediction": "I think we should keep the original error message here.",
+      "reference": "I don't see how removing \"open file bucket\" helps your purpose."
+    },
+    {
+      "prediction": "I think this should be `6.6.0` instead of `6.6.2`",
+      "reference": "Solr dependency was downgraded?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "or should we just autoload everything in `lib` ?"
+    },
+    {
+      "prediction": "I think we should use `anyMatch` instead of `count > 0`",
+      "reference": "Prefer `Stream.anyMatch` instead of iterating over all slots."
+    },
+    {
+      "prediction": "This should be `BLACKLIST_VIOLATION`",
+      "reference": "create_tables isn't used in 2.0 (should probably be deleted?)"
+    },
+    {
+      "prediction": "I think this should be `new List<string> { \"http://*:80/\" }`",
+      "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 don't think we need to test all the options here.",
+      "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": "Style/RedundantParentheses: Redundant parentheses detected.<br>Metrics/BlockLength: Block has too many lines. [21/20]",
+      "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` instead of `Backend\\\\Web\\\\Info\\\\ApacheSolrDocument`",
+      "reference": "We should remove this"
+    },
+    {
+      "prediction": "I think this should be `string` instead of `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": "nit: please remove this empty line",
+      "reference": "File is not `goimports`-ed with -local github.com/iotexproject/iotex-core (from `goimports`)"
+    },
+    {
+      "prediction": "I think it's better to return an empty list instead of `this.writer.writers()`. Because `this.writer.writers()` may return a non-empty list.",
+      "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'm not sure if this is the best way to do it, but it seems to be the only way to make the test pass.",
+      "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 we can remove this import.",
+      "reference": "Please don't use star imports, in the codebase we tend to use direct imports."
+    },
+    {
+      "prediction": "I think this should be `expect(username).toEqual(CREDENTIALS.user);` or `expect(username).toBe(CREDENTIALS.user);`",
+      "reference": "Filename can be `config.functional`"
+    },
+    {
+      "prediction": "I think this should be `std::move (thread).join ()`",
+      "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 `WalletCmd.AddCommand(WalletCreateCmd)` and `WalletCmd.AddCommand(WalletListCmd)` instead of `WalletCmd.AddCommand(walletCreateCmd)` and `WalletCmd.AddCommand(walletListCmd)`. This is because `WalletCreateCmd` and `WalletListCmd` are defined as `var WalletCreateCmd = &cobra.Command{...}` and `var WalletListCmd = &cobra.Command{...}` and `walletCreateCmd` and `walletListCmd` are defined as `var walletCreateCmd = &cobra.Command{...}` and `var walletListCmd = &cobra.Command{...}`. So `WalletCmd.AddCommand(WalletCreateCmd)` and `WalletCmd.AddCommand(WalletListCmd)` will add `WalletCreateCmd` and `WalletListCmd` to `WalletCmd` and `WalletCmd.AddCommand(walletCreateCmd)` and `WalletCmd.AddCommand(walletListCmd)` will add `walletCreateCmd` and `walletListCmd` to `WalletCmd` and `WalletCmd.AddCommand(walletCreateCmd)` and `WalletCmd.AddCommand(walletListCmd)` will add `walletCreateCmd` and `walletListCmd` to `WalletCmd` and `WalletCmd.AddCommand(walletCreateCmd",
+      "reference": "don't use `init` function (from `gochecknoinits`)"
+    },
+    {
+      "prediction": "This should be `sns-ruby-example-enable-resource.rb demonstrates how to enable an Amazon Simple Notification Services (SNS) resource using the AWS SDK for Ruby.`",
+      "reference": "Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "I don't think this is a good idea to add a crash test flag.",
+      "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 we need this method. We can just call `getRequestBodyVarName()` with an empty string.",
+      "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 `0o644` instead of `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": "I think this should be `details.html?id=...` instead of `details?id=...`",
+      "reference": "Why was this changed?"
+    },
+    {
+      "prediction": "`self.floor_thr == 0` -> `self.floor_thr = 0`",
+      "reference": "I think you meant, self.floor_thr = 0"
+    },
+    {
+      "prediction": "why change the order of import?",
+      "reference": "Can you revert these changes?"
+    },
+    {
+      "prediction": "Please remove the trailing comma.",
+      "reference": "Do we need to change this?"
+    },
+    {
+      "prediction": "I think this should be `cgroup.Resources.MemorySwap == uint64(configs.MemoryUnlimited)`",
+      "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 think this should be `const notOverFileLimit = props.totalFileCount < props.maxNumberOfFiles`",
+      "reference": "`props.totalFileCount < props.maxNumberOfFiles` makes the intent a bit more clear I think. And a check to see if maxNumberOfFiles even exists?"
+    },
+    {
+      "prediction": "I don't think we need this flag. We can just use `epochNum == 0` to check if it's the first epoch.",
+      "reference": "`epochNum` is a global variable (from `gochecknoglobals`)"
+    },
+    {
+      "prediction": "I think this should be `newRow = self.getRow(newEntity, oldRows);` instead of `newRow.entity = newEntity;`",
+      "reference": "This needs to be fixed before we can accept the PR"
+    },
+    {
+      "prediction": "I don't think we need to make these `constexpr` as they are only used in `nano::frontier_req::frontier_req`, `nano::bulk_pull_blocks::bulk_pull_blocks`, `nano::bulk_pull_account::bulk_pull_account` and `nano::frontier_req_client::frontier_req_client`.",
+      "reference": "Are these forward declarations needed if we `#include <node/common.hpp>` which includes the definition ?"
+    },
+    {
+      "prediction": "I think this should be `ConnectID` instead of `ConnectID`.",
+      "reference": "can you use either `connectID` or `ConnectID` i would preffer `connectID`"
+    },
+    {
+      "prediction": "This should be `?array` instead of `array|null`",
+      "reference": "It seems like it cannot be null?"
+    },
+    {
+      "prediction": "@jamiesjc I am not sure if this is the right way to deprecate a field.",
+      "reference": "should we consider add deprecated annotation here?"
+    },
+    {
+      "prediction": "I think this should be `username libkb.NormalizedUsername` instead of `name libkb.NormalizedUsername`.",
+      "reference": "Why can't the implementor get the username from the config?"
+    },
+    {
+      "prediction": "I think this should be `execution.GasLimit()` instead of `raCtx.ActionGasLimit`",
+      "reference": "@CoderZhi Please confirm if this change makes sense"
+    },
+    {
+      "prediction": "I think this should be `-50101` to be consistent with the rest of the error codes.",
+      "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'm not sure this is a good idea. `reduceLeft` and `foldLeft` have different semantics. `reduceLeft` throws a `NoSuchElementException` if the collection is empty, while `foldLeft` returns the identity value.",
+      "reference": "`tail()` is an expensive operation for certain collections"
+    },
+    {
+      "prediction": "nit: extra space after `['../../../TestServer/swagger/azure-parameter-grouping.json', 'ParameterGroupingModule'],`",
+      "reference": "do we need to add this one in this PR?"
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "where is a test for it?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `^\\/[a-zA-Z0-9\\/\\:_\\-\\*\\[\\]\\+\\?\\|\\.\\^\\\\\\$]*$`",
+      "reference": "Why has this spacing been adjusted?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should raise the exception and let the caller deal with it.",
+      "reference": "Sorry to do this, since you're touching this file can you resolve the pylint doc messages above?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's the only way to make sure we don't have duplicate rows in the BigQuery table.",
+      "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": "I think this should be `Gesamtwegstreckenz\u00e4hler Offset`",
+      "reference": "Wondering where this breaks btw (although pretty understandable)"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "is this needed? 10 is already the default"
+    },
+    {
+      "prediction": "I think this should be `'%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 don't think you need to initialize `security_attributes_` and `plugin_security_attributes_` to 0.",
+      "reference": "duplicated in both block of preprocessor condition. Can you join in one?"
+    },
+    {
+      "prediction": "`node.getAttribute('id')` can return `null` or `undefined` which will throw an error when passed to `escapeSelector`",
+      "reference": "Can we put the first part of the expression inline with the `if(` for consistency's sake?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to add these methods. I think they should be added to the `CallReqMeta` interface in `yarpc-go/transport/transport.go`.",
+      "reference": "Do servers have no need of accessing this information?"
+    },
+    {
+      "prediction": "I don't think we should change the name of this method. It's used in many places in the codebase.",
+      "reference": "It feels like this will break some stuff, but let it break. We need \"text\" version anyway."
+    },
+    {
+      "prediction": "I think this should be `FALSE` instead of `TRUE`.",
+      "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, 261, \"nan, inf 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'm not sure about this change, but I think it should be fine.",
+      "reference": "Mmm .... this affect #168 I'll need to research the collateral damage"
+    },
+    {
+      "prediction": "I'm not sure if we want to allow deletion of the beta network. I think we should only allow deletion of testnet, devnet, and mainnet. What do you think?",
+      "reference": "nit: refactor into a const map lookup."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the correct assembly.",
+      "reference": "I don't think this is going to work. We need to get the actual assembly of the app project. @johnnywebb thoughts?"
+    },
+    {
+      "prediction": "`use_memoise` -> `use_memoize`",
+      "reference": "@MSusik be careful about changing existing API. I think in this case you shouldn't rename the argument."
+    },
+    {
+      "prediction": "I think we should use `strncmp` instead of `strcmp` here. `strcmp` will fail if `val.via.str.ptr` is longer than `topic->name`.",
+      "reference": "This is bad. It is not a NULL-terminated string here."
+    },
+    {
+      "prediction": "I think this should be `@var string[]|bool`",
+      "reference": "This probably should be @param string|false."
+    },
+    {
+      "prediction": "I think this is a bug in the test, not in the code. The query should be `select mgr.{employees:salary}, e.{employees:salary} from {employees} as mgr, {employees} as e where e.{employees:lastName} = 'Thompson' and e.{employees:managerId} = mgr.{employees:employeeId}`",
+      "reference": "Unrelated test fix. Already in cordova34 branch."
+    },
+    {
+      "prediction": "why remove this?",
+      "reference": "validate() will be called by either V1 or V2"
+    },
+    {
+      "prediction": "I think this should be `#define NOEXCEPT noexcept(true)`",
+      "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 the right thing to do, but I think it's better to use a read-only transaction here. We don't need to write anything to the database, and we don't want to lock the database for writes while we're reading from it.",
+      "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": "Please remove the extra space at the end of the line.",
+      "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) { fpga_perf->previlege = false; return FPGA_OK; } else { fpga_perf->previlege = true; }`",
+      "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 it would be better to do `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: %v\", err))`",
+      "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": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "the text here is derived from the object type and not getting properly handled by get text"
+    },
+    {
+      "prediction": "nit: I think we can use `strings.HasPrefix(podName, \"antrea-\")` instead of `strings.HasPrefix(podName, \"antrea-agent\") || strings.HasPrefix(podName, \"antrea-controller\")`",
+      "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": "Why is this removed?",
+      "reference": "The corresponding action for this key needs to be removed as well, along with any unused translations."
+    },
+    {
+      "prediction": "I don't think we should use `Tracer.InternalInstance` here. We should use `Tracer.Instance`.",
+      "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": "Why do we need to remove `Rewards` here?",
+      "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": "This change is unrelated to this PR. Please revert it.",
+      "reference": "can you revert changes to this file."
+    },
+    {
+      "prediction": "nit: \"validator\" -> \"validator for the single-owner chunk implementation\"",
+      "reference": "suggestion to add: An soc is a chunks whose address is derived of (...)"
+    },
+    {
+      "prediction": "I don't think we need to add these to `APP_ERROR` as they are API errors.",
+      "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": "Why do we need to pass `blockNumber` here?",
+      "reference": "not a great fan of this change - maybe worth splitting into CommitTree() and CloseBlock(long blockNumber)?"
+    },
+    {
+      "prediction": "I think this should be `public virtual Task AbortAsync(ConnectionAbortedException abortReason)`",
+      "reference": "Ugh, if we're going to make a breaking change, I'd like this to be moved to ConnectionContext."
+    },
+    {
+      "prediction": "Why is this TODO? Is there a reason we can't import rc here?",
+      "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 in `src/Datadog.Trace/Datadog.Trace.nuspec` instead of `deploy/Nuget/Datadog.Trace.nuspec`",
+      "reference": "I think this entry can be removed now, right?"
+    },
+    {
+      "prediction": "I think this should be `provider.get_provider_name() == 'aws' or provider.get_provider_name() == 'google'`",
+      "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 these here. They are already defined in `org.openqa.selenium.remote.CapabilityType` and `org.openqa.selenium.remote.Timeouts`.",
+      "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 \"If the regionHint parameter is an empty string, GetBucketRegion will fallback to the ConfigProvider's region config. If no region is specified, hint or ConfigProvider, an error will be returned.\"",
+      "reference": "Should the last sentence be `If no region was found` rather than `specified`?"
+    },
+    {
+      "prediction": "I don't think you need to check for null here. `player.getName()` will never return null.",
+      "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 don't think we need to trim the `APP` suffix here.",
+      "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 this should be `CredentialRefreshTime` or something like that. `CredentialExpiration` implies that the credential will expire at this time, which is not the case.",
+      "reference": "Nit: missing closing \".\"."
+    },
+    {
+      "prediction": "I think this should be `EC.element_to_be_clickable(By.ID, \"disabled-button\")` instead of `EC.element_to_be_unclickable(By.ID, \"disabled-button\")`",
+      "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 `Marshal` instead of `MarshalBinary`.",
+      "reference": "Same here. Should this be pipeline YAML file? Like `MarshalPipelineManifest`"
+    },
+    {
+      "prediction": "I think this should be `InstanceRestartEvent`.",
+      "reference": "I will add this to my PR"
+    },
+    {
+      "prediction": "I think this should be `@role.plan.owned_by?(@user.id)` instead of `@role.user_id = @user.id`",
+      "reference": "Again not 100% sold on the name"
+    },
+    {
+      "prediction": "I think this should be `m.Get(\"/avatar\", user.SettingsAvatarGet)`",
+      "reference": "Use `m.Combo` for `Get` and `Post` methods."
+    },
+    {
+      "prediction": "I don't think this is the right place for this check. I think it should be done in the `DownloadManager` class, before the `DownloadItem` is created. The `DownloadItem` should not know anything about the file system.",
+      "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\", \"--local\"}`",
+      "reference": "shouldn't this be a const?"
+    },
+    {
+      "prediction": "I don't think this is correct. The functional tests are in the `functional` folder and the integration tests are in the `integration` folder.",
+      "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 `@interface SolrThreadSafe`",
+      "reference": "Please change the name of the file too."
+    },
+    {
+      "prediction": "why change the path here?",
+      "reference": "line is 161 characters (from `lll`)"
+    },
+    {
+      "prediction": "I think this should be `@outputter.print_action_step(\"Generating type references\") if ok`",
+      "reference": "When is this false?"
+    },
+    {
+      "prediction": "I think we should keep this test. It is testing that we don't fall back to the raw value if we have a converter.",
+      "reference": "note: also here ;)"
+    },
+    {
+      "prediction": "I think this should be `if (!wasDeleted) { Runtime.getRuntime().removeShutdownHook(shutdownHook); }`",
+      "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 this should be `Either GitRepo or LocalRepo should be set, not both.`",
+      "reference": "nit: blank between, for consistency."
+    },
+    {
+      "prediction": "I think this should be `replaceAll(\"\\\\\\\\$\", \"\\\\\\\\\\\\\\\\$\")` to escape the backslash as well.",
+      "reference": "ditto for RDoc or YARD link"
+    },
+    {
+      "prediction": "I don't think we should use `finest` here. `finer` would be more appropriate.",
+      "reference": "This method is designed to allow the condition runner to tell users something interesting. The current `INFO` log level is correct."
+    },
+    {
+      "prediction": "Why is this changed to `int64` instead of `uint64`?",
+      "reference": "Again, would not change that."
+    },
+    {
+      "prediction": "I think it's better to put this method in the `Status` class.",
+      "reference": "Is canceled considered failed?"
+    },
+    {
+      "prediction": "Why are we ignoring the error here?",
+      "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 think this should be `stream, err := wrapper.MintX509CA(ctx, &MintX509CARequest{...})`",
+      "reference": "We should add an additional call to Recv() somewhere after here that we assert returns io.EOF."
+    },
+    {
+      "prediction": "nit: I think we can remove the `defer` and just call `CheckConfigAndShutdown(ctx, t, config)` at the end of the function.",
+      "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": "I think this should be `InstanceReportUndestroyedObjects` instead of `DeviceReportUndestroyedObjects`.",
+      "reference": "Do we need HeadlessSurface and DisplayPlaneSurface here too?"
+    },
+    {
+      "prediction": "I think this should be `MongoGridFSStreamError: Cannot call abort() on a stream that has already been aborted`",
+      "reference": "Would it be appropriate for this to be a `MongoStreamClosedError`?"
+    },
+    {
+      "prediction": "This import is not used in this file. Please remove it.",
+      "reference": "Can you raise new issue that talks about refactoring webhook code. It should follow `idiomatic maya` patterns."
+    },
+    {
+      "prediction": "Do we need this interface? Can't we just use `ISpan`?",
+      "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": "Line is too long. [101/80]",
+      "reference": "Should we generate `gem \"rsolr\"` into the application Gemfile?"
+    },
+    {
+      "prediction": "I think this should be `-v` instead of `-v` and `-vv` instead of `-v` and `-vvv` instead of `-v` and `-vv` and `-vvv` and so on.",
+      "reference": "commit message, \"currently ignored\"? (-ed)"
+    },
+    {
+      "prediction": "I think this should be `options[:page_ranges] = Array(options[:page_ranges])` instead of `options[:page_ranges] &&= Array(options[:page_ranges])`",
+      "reference": "Same here. and all others."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "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 what you mean by \"for future linking\". I don't think we have any plans to change the exit stub format.",
+      "reference": "Use a named constant"
+    },
+    {
+      "prediction": "Why is this null? I don't think it should be.",
+      "reference": "The `?` is unnecessary since `Assert.Single` will fail if it can't find the stop event."
+    },
+    {
+      "prediction": "I think this should be `String dataLocation(String filename)`",
+      "reference": "What about providers that don't have a reliable location? Do they return null?"
+    },
+    {
+      "prediction": "I think this should be `public abstract void SetGetMetricFunction(Func<MetricItem> getMetric);`",
+      "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 it would be better to use `aws_stack.get_local_dynamodb_endpoint_url()` here.",
+      "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) { return (int)(num_raw & 0xffffff); /* Drop BSD bit */ } else { num = (int)num_raw; /* Keep Mach and Machdep bits */ }`",
+      "reference": "Convention is to use TEST"
+    },
+    {
+      "prediction": "Why did you remove the skip?",
+      "reference": "Just in case: remember to remove this."
+    },
+    {
+      "prediction": "I think this should be `baseFee = max(parentBaseFee - feeDelta, MIN_BASE_FEE);`",
+      "reference": "I am not certain if this should be 0 or 7 here. I think a basefee under 7 is pathological"
+    },
+    {
+      "prediction": "This should be `static constexpr char DEG_SIGN[] = \"\\xc2\\xb0\";`",
+      "reference": "it would be better to move the declaration of the variable with its initialization at line 68"
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "I don't think this is required."
+    },
+    {
+      "prediction": "I think this should be `service.updateContext()`",
+      "reference": "This rename should also ideally be in a separate PR."
+    },
+    {
+      "prediction": "I don't think this is correct. If `parseResponse` throws an exception, then `value` will be `null` and `onSuccess` will not be called.",
+      "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 don't think you need to make these final. They are only set in the constructor.",
+      "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 is this change necessary? I don't see a change in the default value of `ENABLE_QUARTZ` in the `ConfigurationKeys` class.",
+      "reference": "Yeah, why not use this method instead of checking with `containsKey`?"
+    },
+    {
+      "prediction": "I think it's better to use `format` instead of `f-string` here.",
+      "reference": "Compute `(time_forward + 1e-3) * 1e3 / loop_num` ahead."
+    },
+    {
+      "prediction": "I think this should be `InternalLogger.Error(exception, \"Exception in asynchronous handler.\");`",
+      "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": "nit: I think it would be better to have a `SnippetSetRunner<Method>` instead of a `SnippetSetRunner.Generator<Method>`.",
+      "reference": "Since this never changes, why not pass it into the constructor instead?"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "this is critical to avoid state root saving from beam"
+    },
+    {
+      "prediction": "`self.num_classes` should be `self.num_classes - 1`",
+      "reference": "To make it more clear, `fill=self.num_classes`."
+    },
+    {
+      "prediction": "I think this should be `len(m.Body) + len(m.Headers)`.",
+      "reference": "Nit: \"ByteSize estimates the size in bytes of the message for the purpose of restricting batch sizes.\""
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "It's weird to have this on the same level as kestrel. This needs to move to options."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work for now.",
+      "reference": "is it safe for us to use the `blob.type?.length` optional chaining with IE11 support? @goto-bus-stop"
+    },
+    {
+      "prediction": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "remove this debug statement"
+    },
+    {
+      "prediction": "I think this should be `else if (device_extensions.vk_khr_sampler_ycbcr_conversion && api_version < VK_API_VERSION_1_1)`",
+      "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": "(style) line over 80 characters",
+      "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'm not sure if this is the best way to handle this. I think it would be better to have a special case for the `progress` element in the `diffProps` function. What do you think?",
+      "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 remove this line.",
+      "reference": "Why is this being added here? Is this relevant to these changes?"
+    },
+    {
+      "prediction": "I think this should be `LOG.fine(\"Bound legacy RC support\");`",
+      "reference": "This informational message is important to users. Please leave."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to link to the proposal, but it's the only way I could get it to work.",
+      "reference": "why isn't rubocop failing for single quotes?"
+    },
+    {
+      "prediction": "I think this should be `pws.status_id` instead of `status_id`",
+      "reference": "Is `WHERE` should `Language` param ?"
+    },
+    {
+      "prediction": "I think it would be better to use `oneapi::dal::preview::spmd` instead of `oneapi::dal::preview::spmd`",
+      "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": "`abs` is a built-in function in Python, so we should not use it as a variable name. How about `abs_`?",
+      "reference": "Hm, should we show pandas' dtype instead of Spark data type?"
+    },
+    {
+      "prediction": "I think it would be better to use `0x11` for `HistogramWithBuckets`, `0x12` for `HistogramWithMinMax` and `0x13` for `HistogramWithMinMaxAndBuckets`.",
+      "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 `@psalm-taint-sink text $class` instead of `@psalm-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 `if isinstance(v, dict) or isinstance(v, list):`",
+      "reference": "Fails if obj is list. Please fix!"
+    },
+    {
+      "prediction": "This test should be moved to `test_ec2.py`.",
+      "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 `universal_newlines=True` instead of `universal_newlines=False`",
+      "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": "Why is this change needed?",
+      "reference": "These `Get`s are just cosmetic changes for consistency."
+    },
+    {
+      "prediction": "I think this test should be removed.",
+      "reference": "We should never skip unit tests, but instead fix them. Look at `pytest.raises` for cases where we expect an exception."
+    },
+    {
+      "prediction": "Shouldn't this be `Log.LogMessage(MessageImportance.Low, \"No items to push found in the items list.\");`?",
+      "reference": "Do we need to call ToList here?"
+    },
+    {
+      "prediction": "Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier if usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>Style/IfUnlessModifier: Favor modifier unless usage when having a single-line body. Another good alternative is the usage of control flow &&/||.<br>",
+      "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 about this change. I don't think it's a good idea to change the behavior of `toLispString` just to make it consistent with `toString`.",
+      "reference": "could you please separate computations from string concatenation?"
+    },
+    {
+      "prediction": "Shouldn't this be `Assert.Equal(\"code\", type.Properties[0].Name);`?",
+      "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 think it would be better to do this before the `h2o_reprocess_request` call, so that the rewritten headers can be used in the `h2o_reprocess_request` call.",
+      "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": "I think this should be a `time.Time` instead of a `time.Duration`.",
+      "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": "Why is this removed?",
+      "reference": "can we remove it ?"
+    },
+    {
+      "prediction": "I don't think we need to check the last character is an underscore. If it is, then it is a reserved column name. If it is not, then it is not a reserved column name.",
+      "reference": "I take it that by the time we get here, colName has already been case-shifted to upper case?"
+    },
+    {
+      "prediction": "I think it would be nice to have an example showing how to add multiple headers with the same name.",
+      "reference": "Super clear desc - thx."
+    },
+    {
+      "prediction": "I think this should be `swift_module->forAllVisibleModules` instead of `swift_module->forAllVisibleModules`.",
+      "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 should be `assert.True(t, cfg.SharedVolumeMatchFullConfig, \"Default SharedVolumeMatchFullConfig set incorrectly\")`",
+      "reference": "Can you also modify the `TestEnvironmentConfig` in `config_test.go` to cover this?"
+    },
+    {
+      "prediction": "I don't think we need to do anything in this method. We can just remove it.",
+      "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.finer` instead of `finest`. `finest` is the lowest level of logging, and I don't think we want to log this at that level. `finer` is the next lowest level, and I think it's appropriate for this level of logging.",
+      "reference": "We chose `info` in the test code for obvious reasons. Changing to `finest` makes debugging harder and noisier."
+    },
+    {
+      "prediction": "I don't think we need to add support for multiple modules. We can have a single global map for all modules.",
+      "reference": "I know this is not new here but we need to remove this global or protect access from multiple threads."
+    },
+    {
+      "prediction": "I don't think we need to change the order of parameters here.",
+      "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 don't think this is the right place to set `GOMAXPROCS`. It should be set in the `.travis.yml` file.",
+      "reference": "I think this should be `GOMAXPROCS=$(nproc 2>/dev/null || echo 2)`"
+    },
+    {
+      "prediction": "Why do we need to check working set is not nil?",
+      "reference": "Can we log if this actually happens."
+    },
+    {
+      "prediction": "I don't think we need to do this here. The `fixUpHost` method will take care of setting the `remoteHost` to `localhost` if it is not set.",
+      "reference": "This is not really needed, the line added in `GridNodeConfiguration.java` is what really fixes the `remoteHost` regression."
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "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 don't think we need to list all the bridges here.",
+      "reference": "Maybe call it `WireBridge`?"
+    },
+    {
+      "prediction": "nit: please remove unused imports",
+      "reference": "Looks like `org.json.*` can be removed?"
+    },
+    {
+      "prediction": "Please add a comma after `WidgetID.GauntletMap.CONTAINER`",
+      "reference": "remove the extra comma and newline"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to set the page title. I think it should be set in the `saved_searches_controller.rb` in the `index` action.",
+      "reference": "Can we try to use Ruby 1.9-style hashes?"
+    },
+    {
+      "prediction": "I don't think we need to panic here. We can just return an error from `RegisterGlobalOption` and let the caller handle it.",
+      "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": "This should be `if focusObject and focusObject.role == controlTypes.ROLE_PASSWORDEDIT:`",
+      "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 think this should be `EC.element_to_be_clickable(By.ID, \"disabled-button\")` instead of `EC.element_to_be_unclickable(By.ID, \"disabled-button\")`",
+      "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'm not sure if this is the right thing to do, but I think it's the least bad of the options we have.",
+      "reference": "Do DiscoveryContext and GoGapicContext have a more specific common ancestor than Object? If so, let's use that."
+    },
+    {
+      "prediction": "I think this should be `:disable_iptables => true`",
+      "reference": "is this the intended behavior now?"
+    },
+    {
+      "prediction": "I think this is a bug in the original 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 `time.Unix(0, 0)`",
+      "reference": "I'm curious why this is needed?"
+    },
+    {
+      "prediction": "I think this should be `if (caster && target && params.distanceEffect != CONST_ANI_NONE && target->getHealthPercent() > 0) {`",
+      "reference": "I do not see how that changes anything, as the target is operated on many times before that line?"
+    },
+    {
+      "prediction": "I think this should be \"sidebar.analytics.user-loyalty\"",
+      "reference": "Same user/visitor thing."
+    },
+    {
+      "prediction": "I think this should be `C:\\\\Program Files (x86)\\\\Puppet Labs\\\\Puppet\\bin\\\\environment.bat`",
+      "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 think it would be better to use `JaegerExporterEventSource.Log.FailedToSend(ex);` instead of `JaegerExporterEventSource.Log.FailedToSend(ex.ToString());`",
+      "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 to put this. I think it would make more sense to put it in the `PerField` class itself, so that it can be shared by all instances of `PerField` in the same `IndexingChain` instance.",
+      "reference": "should we set a smaller initial size to make sure that we exercise the growing logic in our tests?"
+    },
+    {
+      "prediction": "`f.A.max()` and `dt.max(f.A)` are not the same. `f.A.max()` is an expression, while `dt.max(f.A)` is a function.",
+      "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'm not sure why this was changed, but I think it should be `parser.UnqualifiedName(field->name()->str())`.",
+      "reference": "Why was this removed?"
+    },
+    {
+      "prediction": "I think we should make this required.",
+      "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 we need to add this overload. The existing `fill(int n, T element)` should be enough. We already have `fill(int n, Supplier<? extends T> s)` and `fill(int n, Function<? super Integer, ? extends T> f)`.",
+      "reference": "(...), where each element ~are~ **is the** given {@code element}."
+    },
+    {
+      "prediction": "I think we can remove the `AddAspNetCoreInstrumentation` and `AddHttpClientInstrumentation` calls here.",
+      "reference": "What if `AddOpenTelemetryTracing` is called multiple times?"
+    },
+    {
+      "prediction": "I think this should be `if (lastId >= 0 && lastId <= wal_->lastLogId()) {` instead of `if (lastId >= 0 && lastId <= wal_->lastLogId()) {`",
+      "reference": "Why is the condition of \"lastId <= wal _-> lastLogId() \" added here?"
+    },
+    {
+      "prediction": "I don't think we should use `path` here. We should use `github.com/pulumi/pulumi/pkg/util/contract` instead.",
+      "reference": "`environmentID` is unused in MakeEnvironmentURL"
+    },
+    {
+      "prediction": "I think this should be `length == 0` instead of `length > 0`",
+      "reference": "Just hide the textview"
+    },
+    {
+      "prediction": "I think this should be `PaymentVersionV1` and `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": "nit: `require.NotNil(stream, \"stream should not be nil\")`",
+      "reference": "We should add an additional call to Recv() somewhere after here that we assert returns io.EOF."
+    },
+    {
+      "prediction": "I think we can remove this import.",
+      "reference": "What is this used for?"
+    },
+    {
+      "prediction": "This is a breaking change. I don't think we should do this.",
+      "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 don't think we need to drop and recreate the tables here. We can just add the new columns to the existing tables.",
+      "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 think this should be `expect(page).to have_content(just_finished_trail.name)`",
+      "reference": "I'm confused why the expectations here would have inverted? Seems unrelated to topic stuff. Can you clarify?"
+    },
+    {
+      "prediction": "I think it's better to use `std::numeric_limits<int64_t>::max() - version - 1` instead of `std::numeric_limits<int64_t>::max() - version` to avoid the case that `version` equals to `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 `CloseableGroup` dependency here.",
+      "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": "I think this should be `const int kernel_extent = dilation_data[i] * (kernel_shape_data[i] - 1) + 1;` instead of `const int kernel_extent = dilation_data[i] * kernel_shape_data[i] + 1;`",
+      "reference": "Should this be `const`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I don't see any other way to run the benchmarks. The `BenchmarkRunner.Run<T>()` method doesn't seem to be able to take an assembly as an argument.",
+      "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(\"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": "This is not a real Ruby type, it's a class in the `activesupport` gem.",
+      "reference": "@devigned, please clarify"
+    },
+    {
+      "prediction": "nit: please remove this empty line",
+      "reference": "File is not `gofmt`-ed with `-s` (from `gofmt`)"
+    },
+    {
+      "prediction": "I don't think we need to install git here.",
+      "reference": "Should we break this up into a separate line?"
+    },
+    {
+      "prediction": "I think this should be `setEnvVar(\"USE_CI_ARTIFACTS\", \"true\", true)`",
+      "reference": "Why the override on the env variable, wouldn't we want to use the value set from Prow?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's the best we can do for now.",
+      "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 a good idea. The `subSequenceRangeCheck` method throws a `StringIndexOutOfBoundsException` which is not a subclass of `IndexOutOfBoundsException`. I think it would be better to throw a `StringIndexOutOfBoundsException` in the `subSequence` method.",
+      "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": "I think this should be `IllegalArgumentException(\"MetricRegistry must not be null\")`",
+      "reference": "I'd make this a `throw new NullPointerException(\"registry == null\");` instead"
+    },
+    {
+      "prediction": "I think this should be `config:\"duration;300s\"` and `config:\"duration;86400s\"`",
+      "reference": "Does adding things here require us also to extend the FelixConfiguration resource in libcalico-go?"
+    },
+    {
+      "prediction": "I think this should be `2018.09.0` and the git hash should be updated.",
+      "reference": "does this need to be public?"
+    },
+    {
+      "prediction": "nit: `FILEIDENTIFIER_LENGTH` -> `FILE_IDENTIFIER_LENGTH`",
+      "reference": "Please import this from a pre-existing definition."
+    },
+    {
+      "prediction": "I think this should be `127.0.0.1` instead of `0.0.0.0`.",
+      "reference": "File name does not represent class name... if we ever cared for such things?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "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 `ensureDataDir()` instead of `ensureSingleDataDir()`. `ensureSingleDataDir()` is used when we know there is only one data directory, but here we don't know if there is only one or not.",
+      "reference": "This exits if `-d` not specified which isn't exactly what we want, right?"
+    },
+    {
+      "prediction": "I think it would be better to use `re.escape` here.",
+      "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": "`SupportedType::INT` should be removed.",
+      "reference": "I want to know in which situation you need to traverse from \"Timestamp\" data....."
+    },
+    {
+      "prediction": "I think this should be `if err != nil { return err }` instead of `if err != nil { return err }`.",
+      "reference": "nit: hasBytes should be hashBytes"
+    },
+    {
+      "prediction": "I don't think we need to disable the `SpecialReturnValueCollector.checker_collect` here. It should be disabled by default.",
+      "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 think this should be `fee = 21000 * defaultGasPrice`",
+      "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 think this should be `!opc.StrictSigningKeyUsage && opc.Issuer != opc.Subject`",
+      "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 don't think we should be using `fmt.Fprintf` here. I think we should be using `w.WriteHeader(http.StatusUnauthorized)` and `json.NewEncoder(w).Encode(&api.ClusterResponse{Error: \"Access denied, token is malformed\"})`",
+      "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 this should be `output->format = format; wlr_output_schedule_done(output);`",
+      "reference": "Nothing would change here even if this `if` statement were removed."
+    },
+    {
+      "prediction": "I don't think we need to import `sql` from `databricks.koalas.sql`.",
+      "reference": "`sql` module is imported later."
+    },
+    {
+      "prediction": "`cfg.nms.get('type', 'nms') == 'nms'` -> `cfg.nms.type == 'nms'`",
+      "reference": "I think it's strange, whether `assert cfg.nms.type == 'nms'` would be better."
+    },
+    {
+      "prediction": "I think this should be `return Promise.reject(new Error('cancel'));`",
+      "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 think this should be `hu.IsPre(config.Aleutian, 0) || hu.IsPost(config.Fairbank, 0)`",
+      "reference": "let's use greenland in the unit tests as well"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be nice to have an option in the settings to turn this off.",
+      "reference": "Please split this into two lines"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a property on the `MvxTvosMoreNavigationController` that indicates whether or not it is currently showing the more menu. Then you can just check that property instead of hardcoding the selected index.",
+      "reference": "Is there any variable available instead of hardcoding 5?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "The onNewPasscode method was taking care of creating the user account. Now the user account will already have been created."
+    },
+    {
+      "prediction": "I don't think we need to make this configurable.",
+      "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'm not sure this is the best way to do this, but it works for now.",
+      "reference": "I wonder if you can use the controller's context instead of using it explicitly?"
+    },
+    {
+      "prediction": "I think this should be `e.LogPersister.AppendErrorf(\"Unable to remove canary resources: %v\", err)`",
+      "reference": "Can you add this log for `baseline` too?"
+    },
+    {
+      "prediction": "This should be `Blockly.Msg.NEW_BROADCAST_MESSAGE`.",
+      "reference": "Let's call this \"NEW_BROADCAST_MESSAGE\". Msg.NEW_MESSAGE will be confusing down the line."
+    },
+    {
+      "prediction": "I don't think we need to deprecate this. We can just remove it.",
+      "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 we can remove the `KeyError` here. I don't think it's possible to get a `KeyError` from `df['logical_key'].str.contains('/')`.",
+      "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 `DisableHTTPGit` instead of `DisableHttpGit` to be consistent with the rest of the codebase.",
+      "reference": "Change to `DisableHTTPGit`"
+    },
+    {
+      "prediction": "I'm not sure why this was here, but I think it's safe to remove it.",
+      "reference": "org was not used in the mail content"
+    },
+    {
+      "prediction": "I think this should be `$('link[title=light]').prop('disabled', true);`",
+      "reference": "Is this gonna cause a lint issue, because of unused var? Maybe we should log it out?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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'm not sure why this is needed. `mediatedSetInProgress` doesn't depend on `debouncedSetInProgress`, so I don't see why it needs to be included in the dependency array.",
+      "reference": "this dependency to a callback also has to be a callback (eslint now warns us)"
+    },
+    {
+      "prediction": "nit: `csp *apis.CStorPoolInstance` -> `cspInstance *apis.CStorPoolInstance`",
+      "reference": "directory name as well needs change.."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. The identifier should be normalized before it is passed to the plugin manager.",
+      "reference": "Why is the getIdentifier() call required here?"
+    },
+    {
+      "prediction": "I think this should be `file_being_parsed_.length() ? AbsolutePath(file_being_parsed_) : file_being_parsed_`.",
+      "reference": "Looks more appropriate to stick this in `AbsolutePath`.. actually that already has `FLATBUFFERS_NO_ABSOLUTE_PATH_RESOLUTION`"
+    },
+    {
+      "prediction": "I don't think you need to do this. You can just use `preact-hyperx` directly.",
+      "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 think this should be `if res.Status == keybase1.BlockStatus_ARCHIVED || res.Status == keybase1.BlockStatus_NOT_FOUND {`",
+      "reference": "Should this be inside the `else` below and right after `CDebugf` so that we still get the logs?"
+    },
+    {
+      "prediction": "I think this should be `dynamodb.ErrCodeValidationException` instead of `\"ValidationException\"`.",
+      "reference": "Weird they don't expose the type for dynamo and this seems a pretty common error."
+    },
+    {
+      "prediction": "I think this should be `IDisposable` instead of `IDisposable?`.",
+      "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": "`addtransaction` is not a very good name. I think it would be better to call it `commit` or `commit_transaction` or something like that.",
+      "reference": "The default should be \"False\", as \"True\" would be a significant change in behaviour."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it seems to be the only way to ensure compatibility between Python 2 and 3.",
+      "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 `str_replace(\"\\n\", \"\\n    \", $matches[2])`",
+      "reference": "This should consume less resources because str_replace don't need to assign arrays with all the values"
+    },
+    {
+      "prediction": "I don't think you need to call `ast_print_type` here. `a_type` and `p_type` are already `ast_t *`s, so you can just pass them to `ast_error_frame`.",
+      "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": "Shouldn't this be `System\\Classes\\MediaLibraryItem` instead of `SystemMediaLibraryItem::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": "I think this should be `$config->Mailer` instead of `$config`.",
+      "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": "These should be `SPI0_SCK_PIN = 25` etc.",
+      "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": "Shouldn't this be a `ClassNotFoundException` instead of a `LinkageError`?",
+      "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 think this should be `traceparent` instead of `Traceparent`.",
+      "reference": "nit:s/Traceparent/traceparent/ - As per the spec header name is all lowercase."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be in `pkg/paths/paths.go` or `pkg/paths/paths_test.go`.",
+      "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 think this import is not needed anymore.",
+      "reference": "We generally avoid wildcard imports. If you feel like changing it back, great. Otherwise, no big deal."
+    },
+    {
+      "prediction": "I think this should be `{\"outboundservice\": {Unary: myserviceOutbound}}` instead of `{\"outboundservice\": {Unary: myserviceOutbound}}`",
+      "reference": "We also have a typo on the line above! OUt or Out?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we can remove the `anchor: \"comments\"` from the `proposal_url` helper.",
+      "reference": "OH! These are text.... (Ignore above)"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "How was this set before? Was it just ignored?"
+    },
+    {
+      "prediction": "I don't think we want to use `redirect_to` here. We want to use `flash.now` so that we can display the error message on the same page as the form.",
+      "reference": "Should we `save!` here?"
+    },
+    {
+      "prediction": "I think this should be `snprintf(port, sizeof(port), \"%d\", port_);`",
+      "reference": "sizeof(port) seems ever-so-slightly clearer."
+    },
+    {
+      "prediction": "I'm not sure I like the name \"groups\" here. I think it would be better to call them \"selectors\" or \"labels\" or something like that.",
+      "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 - 1].startswith(\"--mode=\") and len(parts[cursor_part - 1]) > 7`",
+      "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 don't think we need to keep `ConditionFailedError` here.",
+      "reference": "Is this map used for durable state anywhere? If so we probably want to keep ConditionFailedError as 2"
+    },
+    {
+      "prediction": "I think this should be `const methodHEAD = \"HEAD\"` to make it clear that this is a constant.",
+      "reference": "You can use `http.MethodHead` here."
+    },
+    {
+      "prediction": "I don't think we need this file.",
+      "reference": "And another empty test. Please remove until there's content."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to use `console.debug` instead of `console.log`.",
+      "reference": "You really didn't need to do any of these, we need to remove the API client from source eventually anyway."
+    },
+    {
+      "prediction": "I think it's better to use `TiSession` instead of `TiConfiguration` here.",
+      "reference": "seems pretty strange here because TiSession contains PDClient."
+    },
+    {
+      "prediction": "This should be `assert_select \"h4\", :text => \"No notes\"` or `assert_select \"h4\", :html => \"No notes\"`",
+      "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 think this should be `/se/grid/newsessionqueuer/queue`",
+      "reference": "should it be `newsessionqueue` or `newsessionqueuer`? In case we'd like to be consistent"
+    },
+    {
+      "prediction": "nit: I think it would be better to call this `getTotalNumDocs` or `getTotalPendingNumDocs` to make it clear that this is the total number of documents in the index, not just the number of documents that have been added but not yet committed.",
+      "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": "I'm not sure this is the right thing to do here. I think we should be using `curl_with_retries` instead of `retry_on` here. `curl_with_retries` is designed to work with `curl` and `retry_on` is designed to work with `beaker` commands.",
+      "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 think this should be `libcalico-go/lib/selector/parser.Labels` instead of `libcalico-go/lib/selector/parser.Labels`.",
+      "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'm not sure this is the right place to call this. I think it should be called in `script_caret_changeSelection` and `script_caret_move`.",
+      "reference": "nit: This should perhaps be called `waitForAndReportSelectionChange`, as it affects braille as well, not just speech."
+    },
+    {
+      "prediction": "I don't think this is the right way to handle this. If the topic already exists with different tags, we should return a `ResourceAlreadyExistsException` instead of a `InvalidParameter` exception.",
+      "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": "I think it would be better to use `@Shopsys/Styleguide/styleguide.html.twig` instead of `@ShopsysShop/Styleguide/styleguide.html.twig`",
+      "reference": "I am missing return typehint"
+    },
+    {
+      "prediction": "Please remove the `_base_` and `model` in this file.",
+      "reference": "_base_ = 'retinanet_pvt-t_fpn_1x_coco.py'"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to call `onAuthFlowError` and `callback.finish()`. I think it should be in the `catch` block of the `performRequest` method.",
+      "reference": "We should remove `throws Exception` here, since we're catching it anyway and triggering `onAuthFlowError`."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `libp2p` package, not in the `ps` package.",
+      "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": "Should this be `String.format(\"Pull of %s:%s complete\", name, tag)`?",
+      "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": "This should be `cssutils>=1.0.1, <1.1`",
+      "reference": "We should set a 1.7 upper bound for this as well. :smiley:"
+    },
+    {
+      "prediction": "I don't think we should be using lodash here. We should be using `@wordpress/lodash` instead.",
+      "reference": "Lodash mixed in with WordPress dependencies"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to notify about this error than to ignore it.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to test this, but it works for now.",
+      "reference": "perhaps we should test that users cannot edit the profiles of others?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it's better to check for `SERVER_HANDSHAKE_TRAFFIC_SECRET` rather than `CLIENT_RANDOM`.",
+      "reference": "Does this mean the content changed with the new pyopenssl version? Is Wireshark already compatible with this new format?"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "line is 147 characters (from `lll`)"
+    },
+    {
+      "prediction": "I think this should be `true` instead of `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 it would be better to use `DateTimeInterface` instead of `DateTime`.",
+      "reference": "This is not a segment responsibility this query should be in exporter-File module"
+    },
+    {
+      "prediction": "I don't think this is the right place to add the `--export-fixes` option. It should be added to the `analyzer_cmd` list in the `analyze` method.",
+      "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": "`BlockDeviceList` -> `DiskList`",
+      "reference": "What if we do not change this file ? Are we using this `pkg/disk/v1alpha1` elsewhere for block device requirement?"
+    },
+    {
+      "prediction": "I think this should be `@port` instead of `port`",
+      "reference": "regarding dropping the port/host/uri methods on Nod: this would `options[:port] = target.port if target.port`"
+    },
+    {
+      "prediction": "I don't think we need this class at all. We can just call `FileUtil.determineFileType(file, file.getName())` 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 `super || 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": "Why do we need this check here?",
+      "reference": "[You can ignore it] Maybe a better method name would be IsSupported or HasSupportForAllocation?"
+    },
+    {
+      "prediction": "I think this should be `\\\\*` instead of `\\*`.",
+      "reference": "Why is this changed to `\\*`? The behavior of the code on line 42, shows that `*` is what's handled?"
+    },
+    {
+      "prediction": "Shouldn't this be `TestCase.assertNotEqual(Realm.schemaVersion(Realm.defaultPath), 0xFFFFFFFFFFFFFFFF);`?",
+      "reference": "I think this should either return `-1` or throw. I think my vote is on the former."
+    },
+    {
+      "prediction": "I think it would be better to have a default of `None` and set it to `5` in the `default_settings` in `kinto/__init__.py`.",
+      "reference": "I am confused, I've read 2, 3 and 5 seconds in various places"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work for me.",
+      "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 name.'`",
+      "reference": "this string should really be a local constant. it's repeated below."
+    },
+    {
+      "prediction": "I don't think we need to add short here. We already have `int` and `long` which should cover all the cases.",
+      "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": "I think this should be `Assert.Throws<ArgumentException>(() => HttpUtilities.ValidateHostHeader(host));`",
+      "reference": "`Assert.True(true)`...   xunit used to have Assert.DoesNotThrow"
+    },
+    {
+      "prediction": "I'm not sure if this is needed or not, but it seems like a good idea to me.",
+      "reference": "should this be an env var like the other `S3_*` values?"
+    },
+    {
+      "prediction": "I think this should be `/** @param {!Node} node The node to remove */`",
+      "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'm not sure about this change. I think it would be better to keep the original message, and add a new message for the `if key in dictionary` case.",
+      "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": "This should be `@var \\DateTime`",
+      "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 we should use `anyMatch` instead of `count > 0`",
+      "reference": "Prefer `Stream.anyMatch` instead of iterating over all slots."
+    },
+    {
+      "prediction": "This should be `drm_format_mod_explict != nullptr && drm_format_mod_list != nullptr`",
+      "reference": ":grimacing: yikes, thanks for catching this!"
+    },
+    {
+      "prediction": "I think this should be `IMvxRecyclerViewAdapterBindableHolder`",
+      "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 we need to pass diskList here?",
+      "reference": "can we have String() and GoString() implementation of `apis.CStorPool` struct to pretty print it."
+    },
+    {
+      "prediction": "I think this should be `if (name == \"$schema\" || name == \"type\") {` so that we ignore both `$schema` and `type` fields.",
+      "reference": "Like I said, please replace by `EXPECT(kTokenString)`"
+    },
+    {
+      "prediction": "I don't think you need this import.",
+      "reference": "Will you please update the copyright message to `2021` for classes that you modify?"
+    },
+    {
+      "prediction": "I don't think this is necessary. `gridRow.entity` is always an object, so it will always have a `$$hashKey` property.",
+      "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 don't think we need to check dynamo_exited_synched here. It is only set to true when dynamo_exited is true, so we can just check dynamo_exited.",
+      "reference": "No need to check dynamo_exited"
+    },
+    {
+      "prediction": "I think we should rename this to `SetTlsClientCertificate` and `SetTlsClientKey`",
+      "reference": "from linter: >method SetTlsCACertificate should be SetTLSCACertificate"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `(today + 1) * 1000`",
+      "reference": "this data had incorrectly aligned timestamps (which wasn't the point of the test)"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I don't know what else to do.",
+      "reference": "Since this is workflow agnostic, should we put this in common?"
+    },
+    {
+      "prediction": "I don't think we need to pass `environ` to this method.",
+      "reference": "using the `@abstractmethod` decorator wouldn't be better?"
+    },
+    {
+      "prediction": "I don't think we need to check if the property is a boolean here. If it's not a boolean, then it's not a boolean field, so we don't need to check if it's rendered as a switch.",
+      "reference": "This was crashing the app during EDIT on boolean Doctrine Embedded properties"
+    },
+    {
+      "prediction": "I don't think we need the `**/security/` here. I think `*dfl*/security/` should be enough.",
+      "reference": "Will this be backwards compatible with the previous path?"
+    },
+    {
+      "prediction": "I think this should be `IScope CurrentScope => CurrentSpan?.Scope;`",
+      "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 if this is the best way to do this, but it seems to work.",
+      "reference": "I think this is likely the only additional change needed on the original PR."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Please revert changes to unrelated file"
+    },
+    {
+      "prediction": "I think we can remove this class and just use `Server` directly.",
+      "reference": "Can test use a Mockito mock instance instead?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it would be better to keep the summary as it was before, and add a `@SuppressWarnings(\"InvocationHandlerDelegation\")` annotation to the `InvocationHandlerDelegation` class.",
+      "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 think this should be `IEnumerable<T>` instead of `IEnumerable<T>`.",
+      "reference": "we can remove this one as this is not used any more"
+    },
+    {
+      "prediction": "I think we can remove the `if (operationTypes == null) {` and `return null;` below.",
+      "reference": "we don't really need an extra error flag"
+    },
+    {
+      "prediction": "I think we can remove the `ErrorProneJavacPluginPlugin` import.",
+      "reference": "any chance you could just fix the processors plugin?"
+    },
+    {
+      "prediction": "`use_depthwise` should be `use_depthwise_conv`",
+      "reference": "We may keep this config unchanged since `use_depthwise=True` by default."
+    },
+    {
+      "prediction": "nit: missing space before `but`",
+      "reference": "Maybe add i#956 in this log as well."
+    },
+    {
+      "prediction": "I think this should be `if (storeInbox && storeInbox->getWeight() > 0)`",
+      "reference": "if (StoreInbox* storeInbox = getStoreInbox()) {"
+    },
+    {
+      "prediction": "I don't think we need to do this. We can just call `importFromMnemonic` twice, once with `opts.compliantDerivation = true` and once with `opts.compliantDerivation = false`.",
+      "reference": "what is this, is it going to be used later?"
+    },
+    {
+      "prediction": "Line is too long. [81/80]<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "@anodelman test failure is probably due to this, needs to be fixed to `(\"Host #{host.name}\")`"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [114/100]",
+      "reference": "I'm not sure that we need this anymore, but that can be a separate discussion."
+    },
+    {
+      "prediction": "nit: remove the extra space after `* Closes the connections associated with this client. `",
+      "reference": "The formatting seems different from the rest of the code"
+    },
+    {
+      "prediction": "I think this should be `proxyURL` instead of `proxy_url`.",
+      "reference": "Could we have a unit test for this?"
+    },
+    {
+      "prediction": "nit: I think it would be better to pass `stoppedCh` as an argument to `startHistory` instead of creating a new channel here.",
+      "reference": "nit: put params in new line"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "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 think this should be `LOG(THREAD, LOG_INTERP, 2, \"Detected AVX-512 code in use\\n\");` instead of `LOG(THREAD, LOG_INTERP, 2, \"Detected AVX-512 code in use\\n\");`",
+      "reference": "This should be inside the set routine IMHO: matches the others; all callers need it; simplifies code here."
+    },
+    {
+      "prediction": "I think we should make this configurable as well.",
+      "reference": "MAX_RETRY_TIMES makes much more sense."
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "`saveSession` allows easier mocking - just save function which You need, instead full interface"
+    },
+    {
+      "prediction": "`encoding_type` should be `None` by default.",
+      "reference": "No docs for the new param here?"
+    },
+    {
+      "prediction": "I think this should be `Copyright ConsenSys AG.`",
+      "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 think this should be `public virtual Task ForceFlushAsync(CancellationToken cancellationToken) => Task.CompletedTask;`",
+      "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 think this should be `log.Debug(\"completed ExecuteAndWaitProgress() for %s\", target.Label)`",
+      "reference": "is this deliberate? or testing?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "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 think you can remove the `false` here as well.",
+      "reference": "This isn't critical for test coverage, but it seemed off that we had disabled wrapping the reader."
+    },
+    {
+      "prediction": "I think this should be `from rdkit.Chem import Descriptors3D`",
+      "reference": "Please remove Descriptors3D from this file. We just did this in master."
+    },
+    {
+      "prediction": "`client_ = client;`",
+      "reference": "You could pass the MetaClient instance in ctor, and create a new instance if nullptr."
+    },
+    {
+      "prediction": "I'm not sure this is the right fix. I think the problem is that the `WaitForBuiltTarget` call is blocking, so it's holding up the parallelism limiter. I think the right fix is to make `WaitForBuiltTarget` non-blocking, and have it return a channel that will be closed when the target is built. Then the caller can wait on that channel to know when the target is built.",
+      "reference": "What does this do? Seems like a poor mans mutex/semaphore?"
+    },
+    {
+      "prediction": "I think this should be `error` instead of `err error`.",
+      "reference": "This is a breaking change, and cannot be made. We have committed to this API for v1."
+    },
+    {
+      "prediction": "`?Context $global_context = null` should be `Context $global_context = null`",
+      "reference": "Can also drop corresponding types from docblock here"
+    },
+    {
+      "prediction": "I think this should be `finer` or `finest`. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging. `finer` is the next level up. `finest` is the lowest level of logging.",
+      "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;`",
+      "reference": "why is this variable introduced? Please stick `_o.Test3.A` directly in the call below."
+    },
+    {
+      "prediction": "I don't think this change is necessary. The `gentleVerifyToken` middleware should be sufficient.",
+      "reference": "do you mind sharing what is the reason for this change?"
+    },
+    {
+      "prediction": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "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": "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": "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'm not sure if this is the right place to add these, but I think it would be useful to have the number of schedules and sensors in the telemetry data as well.",
+      "reference": "nit: type this"
+    },
+    {
+      "prediction": "I think we should rename this to `CAST_TAB_MIRRORING` to be consistent with the rest of the commands.",
+      "reference": "These command names are specific to Chromium-based browsers. Please move to `ChromiumDriverCommand`"
+    },
+    {
+      "prediction": "I think we should use `bd.Name` instead of `bd.Object.Name` here. `bd.Name` is the name of the block device, while `bd.Object.Name` is the name of the block device object.",
+      "reference": "lets print nodeName that got passed also"
+    },
+    {
+      "prediction": "I think this should be `if sys.version_info > (3, 0):`",
+      "reference": "Should be `if sys.version_info[0] > 2:`"
+    },
+    {
+      "prediction": "I think this should be `debug_config_file_size(string_metric_parse(\"0\"));` instead of `debug_config_file_size(string_metric_parse(\"0\"));//to set debug file size to \"don't delete anything\"`",
+      "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": "`print_option_arg` -> `print_options_arg`",
+      "reference": "Let's not use a form of hungarian notation in naming our variables"
+    },
+    {
+      "prediction": "I think this probe should be moved to `handle_incoming_request` as well.",
+      "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 think this should be `svc.Spec.IPFamilies = append(svc.Spec.IPFamilies, service.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's better to use `frozenset` instead of `set` here. `frozenset` is hashable and can be used as a key in a dictionary.",
+      "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 `core/` prefix in the name of the data store to make it clear that this is a core data store.",
+      "reference": "See above, same for all similar cases below."
+    },
+    {
+      "prediction": "I think this should be `disable_snapshot_annotations = {{ .NodeConfig.AgentConfig.DisableSnapshotAnnotations }}`",
+      "reference": "where is this coming from?"
+    },
+    {
+      "prediction": "Style/RedundantSelf: Redundant self detected.<br>Metrics/LineLength: Line is too long. [101/80]",
+      "reference": "We can also drop `plans.includes_screencasts`, right?"
+    },
+    {
+      "prediction": "I don't think we need to do this here. This is only needed for deterministic wallets, and we don't have deterministic wallets anymore.",
+      "reference": "wouldn't `int.to_bytes(counter, 32, 'little')` be equivalent, clearer and faster?"
+    },
+    {
+      "prediction": "I don't think we need to expose this in the API.",
+      "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'm not sure if this is the right place to do this, but I think it's better to do it here rather than in the `UserEntityStatMessage` class.",
+      "reference": "Could we only do this for all time? Because that's what is causing problems rn?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to fix the issue for me.",
+      "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": "nit: I think we can remove the `parseOptional` and just return `confParser.longConf().option(SparkReadOptions.SPLIT_SIZE).parse()`",
+      "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 don't think this is necessary. `os.makedirs` will raise `FileExistsError` if `parents=False` and the directory already exists.",
+      "reference": "To reduce complexity, please use OSError on Python3 as well."
+    },
+    {
+      "prediction": "I think this should be `cmdutils.argument('tab', value=cmdutils.Value.cur_tab, positional=False)`",
+      "reference": "That seems wrong."
+    },
+    {
+      "prediction": "I don't think this is needed here. The session timeout is set when the user logs in.",
+      "reference": "Yes, this is needed here. This is a builtin user who just reset their password."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but it seems to work for now.",
+      "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 think it would be better to call this `DisableConnectionIDTruncation` or something like that.",
+      "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, but it seems to be the only way to get the `hasResolvedAccounts` and `hasResolvedContainers` booleans.",
+      "reference": "I think the `undefined` checks still need to be removed from here."
+    },
+    {
+      "prediction": "I think this should be `self.kubernetes_config.backoff_limit`",
+      "reference": "Is there any particular reason to have 6?"
+    },
+    {
+      "prediction": "I think this should be `chunk.BiggestPow2AlignedRange()` instead of `chunk.BiggestPow2AlignedRange().BiggestPow2AlignedRange()`.",
+      "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 this is the right place to do this. I think it should be done in `substituteSourceVars` instead of `validateVarsSubbed`. `validateVarsSubbed` should only check that the variables are valid, and `substituteSourceVars` should do the substitution.",
+      "reference": "instead of doing the if err := __; err != nil { return err } thing, you can do errs = addErrs(errs, ___). If you want."
+    },
+    {
+      "prediction": "nit: I think it's better to have `errors` at the top of the file.",
+      "reference": "nit: Is this error message accurate?"
+    },
+    {
+      "prediction": "I think this should be `if (src_repo->mode == OSTREE_REPO_MODE_BARE || src_repo->mode == OSTREE_REPO_MODE_BARE_USER)`",
+      "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 think this should be `resp.Status = statusParts[1]`",
+      "reference": "Parentheses aren't needed here. `if len(statusParts) > 0 {` will suffice."
+    },
+    {
+      "prediction": "I think this should be `SkipOnServers []string // if the test needs to be skipped on particular server`",
+      "reference": "// the test is skipped on given servers because it will fail."
+    },
+    {
+      "prediction": "I think this should be `:unauthorized`",
+      "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.cancel == 1:`",
+      "reference": "Use `==` in this case instead of `is`."
+    },
+    {
+      "prediction": "This is not a CStorPool phase. This is a CStorPool status.",
+      "reference": "Please put it like this: `CStorPoolPhase is a typed string for phase field of CStorPool`"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. If you have 10 items in stock, and you want to buy 12, you should not be able to buy 12 items, but you should be able to buy 10 items.",
+      "reference": "variable must be in camelCase => $newStock"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "unnecessary conversion (from `unconvert`)"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `BitMappedTrieIterator` class that implements `Iterator<T>`. Then you can do `for (T t : trie)` or `while (iterator.hasNext()) { T t = iterator.next(); }`",
+      "reference": "... is probably automatically vectorized, since it has the same speed as `System.arraycopy`"
+    },
+    {
+      "prediction": "nit: I think this should be `var control cgroup.Control = cgroup.New()` so we don't have to call `setControl` in `SetupPlatformResources`",
+      "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 we should keep the `forall [t0, t1]` here.",
+      "reference": "Surprised to see this old syntax here, but not _that_ surprised."
+    },
+    {
+      "prediction": "I don't think we should deprecate this method.",
+      "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": "@aravindhp can you please check if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it. I am not sure if this is the right way to do it",
+      "reference": "exported func New returns unexported type *pool.pool, which can be annoying to use"
+    },
+    {
+      "prediction": "I think this should be `attr_reader :transports`",
+      "reference": "Sorry, I feel like I'm missing something - where does this get called?"
+    },
+    {
+      "prediction": "I don't think this is correct. If the node is not registered, we should not try to update the node status.",
+      "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": "I think this should be `#include <sys/stat.h>`, not `#include <sys/stat>`.",
+      "reference": "How portable is this? (e.g., Windows). I'm relaunching tests of windows because they failed for other reasons."
+    },
+    {
+      "prediction": "I don't think we need to iterate over all active slots here. We only need to iterate over the ones that are bound.",
+      "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 think it's better to return an empty array instead of an error.",
+      "reference": "should use `codes.InvalidArgument`, same as above line 1033 and 1036"
+    },
+    {
+      "prediction": "I think this should be `GetDefaultDockerAPIVersion` to be consistent with the other functions.",
+      "reference": "Please revert this file."
+    },
+    {
+      "prediction": "I think this should be \"For pubsub.OpenTopic and pubsub.OpenSubscription URLs, azurepubsub registers for the scheme \"azuresb\". The default URL opener will use a Service Bus Connection String based on the environment variable \"SERVICEBUS_CONNECTION_STRING\". To customize the URL opener, or for more details on the URL format, see URLOpener.\"",
+      "reference": "Probably in another PR, I think the package should be renamed to `servicebus` to be consistent."
+    },
+    {
+      "prediction": "I don't think this is the right place to put the link to the validation rules. I think it should be in the `SecurityDefinitionsStructureValidator` class.",
+      "reference": "curious: are we planning to add link to each of the rules in the message?"
+    },
+    {
+      "prediction": "I think we should return an error here.",
+      "reference": "line is 123 characters"
+    },
+    {
+      "prediction": "I think this should be `@param array $user` instead of `@param array $user of user data`",
+      "reference": "same here; `@param array $user Array of user data`"
+    },
+    {
+      "prediction": "I'm not sure what this test is trying to do, but it seems to be asserting that `b.String()` returns an empty string when `b.Size() == 0`. Is that correct? If so, I think it would be better to assert that `b.String()` returns an empty string when `b.Size() == 0` and `b.String()` returns a non-empty string when `b.Size() > 0`.",
+      "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": "This should be `MvvmCross.DroidX.RecyclerView.MvxGuardedLinearLayoutManager`",
+      "reference": "Register needs lowercasing"
+    },
+    {
+      "prediction": "I think we should make this true by default. I don't see any reason why we would want to disable it.",
+      "reference": "This should belong to FlowContainer class."
+    },
+    {
+      "prediction": "I don't think we need to change this. The while loop will break when `winUser.getMessage` returns -1.",
+      "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 should be `defer os.Setenv(noninteractive, os.Getenv(noninteractive))`",
+      "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 this should be `WRITE_MODE` instead of `PARQUET_WRITE_MODE` and `AVRO_WRITE_MODE`.",
+      "reference": "nice, having the \"overwrite\" default makes this change backwards compatible, right?"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The `onStatus` should be `onStatus((s) -> s != HttpStatus.OK && s != HttpStatus.NO_CONTENT, (response) -> { ... });`",
+      "reference": "It doesn't make sense to accept all 2xx status code. For example, `203 Non-Authoritative Information` or `205 Reset Content`, etc."
+    },
+    {
+      "prediction": "nit: I think it would be better to use `BytesRef.equals` instead of `BytesRef.bytesEquals` here.",
+      "reference": "Found a better choice here that doesn't involve object creation: StringHelper.startsWith()"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should be able to have multiple node attestors and pick the right one based on the attested data type.",
+      "reference": "Shouldn't we consider the situation where it doesn't find a plugin?"
+    },
+    {
+      "prediction": "I don't think we should throw an exception here. Instead, we should set `mode` to `preview::read_mode::table` by default.",
+      "reference": "Do not understand why this needed for table. All new classes defined in the `preview` namespace."
+    },
+    {
+      "prediction": "I think this should be `athenz.zts.notification_service_factory_class`",
+      "reference": "Kept the same property value even though it is now in Common to avoid breaking existing clients."
+    },
+    {
+      "prediction": "Why is this public?",
+      "reference": "TODO: this is temporary until we figure out case sensitivity for metrics evaluators."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I'm not sure what else to do either. We don't want to log the entire set of cids, but we also don't want to log just the first one. I think this is the best we can do for now.",
+      "reference": "can we just kill this TODO?"
+    },
+    {
+      "prediction": "I think this should be `assert.EqualError(t, err, \"invalid path \\\"awful\\\": selected encoding not supported\")`",
+      "reference": "This appears to be a change in error handling due to a dag upgrade."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "reference": "These changes look like a hosts file _must_ be provided?"
+    },
+    {
+      "prediction": "I don't think you need the `char fmt[] = \"O\";` anymore.",
+      "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 think this should be `--channel <target_channel>`, not `--channel <target_chanel>`.",
+      "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": "Why was this removed?",
+      "reference": "Also not for merge to master."
+    },
+    {
+      "prediction": "I think this should be `if len(self._cron_schedule.split(\" \")) != 5 and self._cron_schedule != \"* * * * *\":`",
+      "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": "Why was this removed?",
+      "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 don't think this is correct. If `checkSimpleWord` returns `false`, it means that the word is not in the dictionary, but it doesn't mean that the word is misspelled. For example, if you have a dictionary with only one word \"hello\" and you try to spellcheck \"helloo\", `checkSimpleWord` will return `false`, but \"helloo\" is misspelled.",
+      "reference": "extra word lookup removed"
+    },
+    {
+      "prediction": "nit: \"The resolution time for a promise is factored into whether a wait has timed out.\" -> \"The resolution time for a promise is not factored into whether a wait has timed out.\"",
+      "reference": "Will you link to {!webdriver.promise.Promise} here?"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `Intents.ACTION_NEW_PROFILE.equals(action) || Intents.ACTION_NEW_TREATMENT.equals(action) || Intents.ACTION_CHANGED_TREATMENT.equals(action) || Intents.ACTION_REMOVED_TREATMENT.equals(action) || Intents.ACTION_NEW_STATUS.equals(action) || Intents.ACTION_CHANGED_STATUS.equals(action) || Intents.ACTION_REMOVED_STATUS.equals(action) || Intents.ACTION_NEW_NOTE.equals(action) || Intents.ACTION_CHANGED_NOTE.equals(action) || Intents.ACTION_REMOVED_NOTE.equals(action) || Intents.ACTION_NEW_OBJECTIVE.equals(action) || Intents.ACTION_CHANGED_OBJECTIVE.equals(action) || Intents.ACTION_REMOVED_OBJECTIVE.equals(action) || Intents.ACTION_NEW_BOLUS.equals(action) || Intents.ACTION_CHANGED_BOLUS.equals(action) || Intents.ACTION_REMOVED_BOLUS.equals(action) || Intents.ACTION_NEW_BASAL.equals(action) || Intents.ACTION_CHANGED_BASAL.equals(action) || Intents.ACTION_REMOVED_BASAL.equals(action) || Intents.ACTION_NEW_TEMP_BASAL.equals(action) || Intents.ACTION_CHANGED_TEMP_BASAL.equals(action) || Intents.ACTION_REMOVED_TEMP_BASAL.equals(action) || Intents.ACTION_NEW_TEMP_BAS",
+      "reference": "@AdrianLxM finally found some your bug too :)"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "This crashed in an app using descriptor_indexing with some unbound descriptors."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the test to pass.",
+      "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'm not sure this is the right place to do this. I think it would be better to do this in the `OnAttach` method of the `MvxFragment<TViewModel>` class.",
+      "reference": "Throw `fragInfo.CachedFragment.ViewModel.GetType()` into a variable so it's not evaluated each time."
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [123/120]",
+      "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 think you can just do `el.remove()` now.",
+      "reference": "JS abstraction leaks so this should be `if (el.parentNode !== null) { el.parentNode.removeChild(el) }` @developit"
+    },
+    {
+      "prediction": "I think this should be `fmt.Sprintf(\"unexpected status 0x%x\", err)`",
+      "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 should be `const cstorRunTask = ` instead of `const cstorRunTask = `",
+      "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 `command_options << '--tty' if options[:tty] && !options[:stdin]`",
+      "reference": "@glennsarti you want to validate this on Windows? IIRC this won't work"
+    },
+    {
+      "prediction": "I don't think we need to make this a `@Value` anymore. We can just use `elasticsearch.getNamesLookback()`.",
+      "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'm not sure about this change. I think it should be `if (err) throw err;` and not `if (error) throw error;` because `err` and `error` are different variables.",
+      "reference": "Curly brackets here `{ (breakLine) xx (breakLine)}`"
+    },
+    {
+      "prediction": "I think this should be `if($(this).val() === countlyGlobal[\"member\"].username){` instead of `if($(this).val() == countlyGlobal[\"member\"].username){`",
+      "reference": "Can you check this case Assuming you have a dashboard with 2 users, test1 and test2 Currently logged in user is test1. You should not be able to change it to test2. You open dev console and paste in `countlyGlobal[\"member\"].username = \"test2\";` then you enter value test2 in the input and it is allowed and form would be submitted. Would server side validation deal with such case?"
+    },
+    {
+      "prediction": "I think this should be `DriverManager.getDriver(url).getClass().getName()`",
+      "reference": "since we're changing this class, let's fix the default class name as well. it should be com.mysql.cj.jdbc.Driver"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like a good place to me.",
+      "reference": "At load time, we ensure the config'd objects exist."
+    },
+    {
+      "prediction": "I think this should be `if (parsedAddress.Scheme.Equals(\"http\", StringComparison.OrdinalIgnoreCase) && parsedAddress.Scheme.Equals(\"https\", StringComparison.OrdinalIgnoreCase)) { throw new InvalidOperationException($\"Unrecognized scheme in server address '{address}'. Only 'http://' and 'https://' are supported.\"); }`",
+      "reference": "Is there a test for this case?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to go about it, but I think it's better than having a bunch of `if`/`else` statements in the view.",
+      "reference": "This copy doesn't match what I had in my mockup. I think it's better to call them episodes."
+    },
+    {
+      "prediction": "I think it's better to use `common.FrontendServiceName` and `common.MatchingServiceName` instead of `serviceName == common.FrontendServiceName || serviceName == common.MatchingServiceName`",
+      "reference": "For some reason it worked for frontend even before this change."
+    },
+    {
+      "prediction": "Why is this needed?",
+      "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 this is the right place to increment `total_blocks`. I think it should be after the `process_block` call, but before the `stop_pull` check.",
+      "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": "`DEFINE_int32(default_max_replica_factor, 1, \"The default max replica factor that a space can take\");` -> `DEFINE_int32(max_replica_factor, 1, \"The default max replica factor that a space can take\");`",
+      "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 `import com.google.common.collect.ImmutableList;` and `return ImmutableList.copyOf(generatedImports);` instead of `return generatedImports;`",
+      "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 this should be `tagMatchers` instead of `tagMatchers`.",
+      "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": "nit: I think this should be `knative.dev/pkg/apis/duck/v1alpha1`",
+      "reference": "same here, move this one down"
+    },
+    {
+      "prediction": "I think this should be `$this->getReference(AdministratorDataFixture::DEFAULT_ADMINISTRATOR)`",
+      "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 think we can remove this line.",
+      "reference": "Is this required?"
+    },
+    {
+      "prediction": "I think this should be `!c.Bool(\"public-volume-creation-disabled\")`",
+      "reference": "`--public-volume-create-allowed=true` and set the default to `true`, then use `!c.Bool('...')`"
+    },
+    {
+      "prediction": "I think this should be `conn->push_stream_ids.max_open = payload.last_stream_id + 1;`",
+      "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": "I think this should be `exit(exit_code)` instead of `exit(exit_code)`.",
+      "reference": "Might as well add a docstring for the new `exit_code` parameter too."
+    },
+    {
+      "prediction": "I think this should be `Optional.ofNullable(sessionRequests.poll(id));`",
+      "reference": "`poll` and `remove` have different meanings."
+    },
+    {
+      "prediction": "I think this should be `write.parquet.bloom-filter.expected.ndv` to be consistent with the other properties.",
+      "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": "`AWS IAM` -> `AWS Identity and Access Management`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think this should be `HIP_UNSUPPORTED`.",
+      "reference": "Remove `HIP_UNSUPPORTED` - it is an alias"
+    },
+    {
+      "prediction": "This method is not used in this PR, so I think we can remove it.",
+      "reference": "i think we should return an empty list instead of null here."
+    },
+    {
+      "prediction": "I'm not sure if we should have separate settings for HTTP and SOCKS proxies, or if we should just have one set of settings for all proxies. I think we should just have one set of settings for all proxies, but I'm not sure what the best way to do that would be. We could have one set of settings for HTTP proxies and one set of settings for SOCKS proxies, or we could have one set of settings for all proxies. I think we should have one set of settings for all proxies, but I'm not sure what the best way to do that would be. We could have one set of settings for HTTP proxies and one set of settings for SOCKS proxies, or we could have one set of settings for all proxies. I think we should have one set of settings for all proxies, but I'm not sure what the best way to do that would be. We could have one set of settings for HTTP proxies and one set of settings for SOCKS proxies, or we could have one set of settings for all proxies. I think we should have one set of settings for all proxies, but I'm not sure what the best way to do that would be. We could have one set of settings for HTTP proxies and one set of settings for",
+      "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'm not sure this is the best way to do this. I think it would be better to have a method that returns the collision energy for a given `TransitionDocNode` and `PeptideDocNode`. This method could then be called from the `GetCollisionEnergy` method, but also from other places where the collision energy needs to be calculated.",
+      "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 place to do this. I think it would be better to do this in the `get_plot_frame` function.",
+      "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 this is the best way to do this. I think it would be better to have a method on the `UserResource` class that returns the list of ids to filter by, and then override that method in the `ShareableResource` class to return an empty list if `forced_empty_list` is `True`.",
+      "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": "I'm not sure if this is the right place to put this, but I think we should add a `SimulateCustomPolicy` method to the `iam.Client` interface.",
+      "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": "Please revert this change.",
+      "reference": "Why are we doing that change?"
+    },
+    {
+      "prediction": "I think this should be `rawErrorData.get(CLASS) == null || rawErrorData.get(STACK_TRACE) == null`",
+      "reference": "Why && changed to || ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this function. I think it should be in `dask_ml.utils` or `dask_ml._utils` or something like that.",
+      "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": "Why do you need to use `StandardServiceContainer` here ?",
+      "reference": "This class is never used, don't import it."
+    },
+    {
+      "prediction": "I don't think this is necessary. The `OnNext` method is already wrapped in a try-catch block.",
+      "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": "I don't think we should be adding this to the public API. It's not part of the W3C spec, and it's not part of the Selenium spec.",
+      "reference": "Files should have an extra line at the end of them."
+    },
+    {
+      "prediction": "Why not just `LoggingRules.Add(rule);` ?",
+      "reference": "I think we could replace this (well the backing field) with `BlockingCollection` when we drop NET35?"
+    },
+    {
+      "prediction": "I think we need to check whether `pser_or_pdf` is a `DataFrame` or `Series`. If it is a `DataFrame`, we need to check whether `input_groupnames` is a subset of `pser_or_pdf.columns`. If it is a `Series`, we need to check whether `input_groupnames` is a subset of `pser_or_pdf.index.names`.",
+      "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 think this should be `if config && config['transports']` instead of `if config && config['transports']`",
+      "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 don't think this is used anywhere.",
+      "reference": "This probably isn't needed anymore now?"
+    },
+    {
+      "prediction": "I don't think this is correct. `rest` is not an array of `VNode`s, it's a single `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 we can remove the `@Override` annotation here.",
+      "reference": "are you sure that the privacyGroup cannot be null?"
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "reference": "Wrapped with list"
+    },
+    {
+      "prediction": "I don't think we need to return a bool here. If the base is 0, we can just return 0.",
+      "reference": "Please move this to `pkg/stats/v1alpha1`"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work. The `if: -> { url.match(URL_FORMAT) }` makes sure that the validation only runs if the URL is in the correct format.",
+      "reference": "a small nit pick ...this can go into `lib/patterns.rb` ?"
+    },
+    {
+      "prediction": "I think this should be `googlesitekit-cta-link--danger` instead of `mdc-tab--active`.",
+      "reference": "The above TODO is no longer necessary I believe"
+    },
+    {
+      "prediction": "Why do we need to change this to seconds?",
+      "reference": "Why is this change required ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place for this function. I think it should be part of the `summary` package, but I'm not sure if it should be part of the `summary` package or the `summary/markdown` package.",
+      "reference": "This really is an excerpt. ;-)"
+    },
+    {
+      "prediction": "I think this should be `return m.isWithinFunc(addr)`",
+      "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 don't think we need to load the sqlite plugin by default.",
+      "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 `int i = 0; i < NUM_SIMD_SLOTS; i++`",
+      "reference": "This should be all slots"
+    },
+    {
+      "prediction": "I don't think we need to change this method name.",
+      "reference": "Need to add deprecated annotation and call the new fixed method to avoid the API broken issue."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems like we don't need to pass in a `Configuration` object to `Glean.INSTANCE.initialize()` anymore.",
+      "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 think this should be `log.V(logf.DebugLevel).Info(\"not found\", \"err\", err)`",
+      "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 don't think we need to lock here. `mset.cfg.Storage` is immutable.",
+      "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 = currUserInfo`",
+      "reference": "I'd rather you do this under `if info.filledInMerkle {` since that's the only place it's relevant."
+    },
+    {
+      "prediction": "I think we should keep the `leaderBaseUrl` and `leaderCoreName` in the `ZkCoreNodeProps` object.",
+      "reference": "*NULL_DEREFERENCE:* object `leaderUrl` last assigned on line 348 could be null and is dereferenced at line 351."
+    },
+    {
+      "prediction": "Trailing whitespace detected.",
+      "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": "This import is not used in this file. Please remove it.",
+      "reference": "nit: We can revert this now."
+    },
+    {
+      "prediction": "I don't think you need to import `java.lang.ref.WeakReference`",
+      "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": "I think this should be `getContentDescription` instead of `getContentDescription`",
+      "reference": "Should be named getDescription, since the other description functions are named that."
+    },
+    {
+      "prediction": "I think this should be done in `onResume()` instead of `onCreate()`.",
+      "reference": "Don't we want to do this for g2 4k also?"
+    },
+    {
+      "prediction": "I don't think we need to use `string-to-arraybuffer` here. We can use `Buffer.from(str, 'utf-8')` instead.",
+      "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 this is a bit confusing. I think it would be better to have a `map[string]map[string]string` where the first key is the chain name, and the second key is the rule hash, and the value is the full rule.",
+      "reference": "I don't think this is used any more, please remove."
+    },
+    {
+      "prediction": "I think this should be `if inner_type.description is not None` rather than `inner_type.name is not None`.",
+      "reference": "actually use type_name=print_config_type_to_string(self, with_lines=False) to populate this"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `processResult` method. The reason is that the `processResult` method is called for each result, so if we delete the file here, it will be deleted for each result. If we delete the file in the `processResult` method, it will only be deleted once. Also, the `processResult` method is called for each result, so if we delete the file here, it will be deleted for each result. If we delete the file in the `processResult` method, it will only be deleted once. Also, the `processResult` method is called for each result, so if we delete the file here, it will be deleted for each result. If we delete the file in the `processResult` method, it will only be deleted once. Also, the `processResult` method is called for each result, so if we delete the file here, it will be deleted for each result. If we delete the file in the `processResult` method, it will only be deleted once. Also, the `processResult` method is called for each result, so if we delete the file here, it will be deleted for",
+      "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": "nit: extra backtick at the end of the line.",
+      "reference": "everywhere else in the project, `LOGGER` is used"
+    },
+    {
+      "prediction": "I think this should be `Settings.api_versions.max_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 think we can remove this file, it's not needed anymore.",
+      "reference": "typo --> gloabl"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `ServeHTTP` method of the `Server` struct.",
+      "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 don't think this is the right message. It should be something like \"Session ID is null. Did you call quit()?\" or \"Session ID is null. Did you call close()?\" or \"Session ID is null. Did you call shutdown()?\" or \"Session ID is null. Did you call stop()?\" or \"Session ID is null. Did you call destroy()?\" or \"Session ID is null. Did you call terminate()?\" or \"Session ID is null. Did you call exit()?\" or \"Session ID is null. Did you call kill()?\" or \"Session ID is null. Did you call stopServer()?\" or \"Session ID is null. Did you call stopWebDriver()?\" or \"Session ID is null. Did you call stopDriver()?\" or \"Session ID is null. Did you call stopBrowser()?\" or \"Session ID is null. Did you call stopApplication()?\" or \"Session ID is null. Did you call stopProcess()?\" or \"Session ID is null. Did you call stopService()?\" or \"Session ID is null. Did you call stopApplicationServer()?\" or \"Session ID is null. Did you call stopApplicationDriver()?\" or \"Session ID is null. Did you call",
+      "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 think this should be `@pinwheel = ['-', '\\\\', '|', '/']`",
+      "reference": "Is this going to be configurable? If not, it should just be removed for now."
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "Maybe just one import line - `from collections import OrderedDict, Mapping` ?"
+    },
+    {
+      "prediction": "I think this should be `PropertyFactory.intListProperty(\"multiInt\").desc(\"Multiple integer values\").requireEach(inRange(0, 10)).defaultValue(1, 2, 3, 4).build();`",
+      "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'm not sure why this is needed, but it seems to fix the test.",
+      "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 this should be `self.source_address = source_address` instead of `self.server_conn.source_address = 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 think this should be `self.{{FIELD_NAME}}().map(|u| {{U_ELEMENT_TABLE_TYPE}}::init_from_table(u)).unwrap_or_else(|| {{U_ELEMENT_TABLE_TYPE}}::init_from_table(self.{{FIELD_NAME}}()))`",
+      "reference": "Please flip the conditional to be `if (field.required) { ... }`."
+    },
+    {
+      "prediction": "nitpick: I think it would be better to use `expect { ... }.not_to change { User.count }` instead of `expect { ... }.to change { User.count }.by(0)`",
+      "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 this should be `Doc generateMethodSampleCode(PythonDocConfig config, Element element, Iterable<String> imports);`",
+      "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": "Why is this needed?",
+      "reference": "What about this: \"// Corresponding PMT does not exist.\""
+    },
+    {
+      "prediction": "I don't think we should raise an error here. If the config file doesn't exist, it's not a fatal error. It just means that the user didn't specify a config file, so we'll just use the defaults.",
+      "reference": "You don't need af ormat specified here."
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Nitpick: import ordering"
+    },
+    {
+      "prediction": "I don't think we need to pass `ITransactionProcessor` to the factory. We can just pass `IBlockchain` or `IBlockTree` instead.",
+      "reference": "code smell here, a function that create a reward calculator from transaction processor?"
+    },
+    {
+      "prediction": "I think this should be `A Client constructs runtime variables using the Runtime Configurator API.`",
+      "reference": "I don't think the `A` is idomatic is it?"
+    },
+    {
+      "prediction": "nit: `does not implement RemoveReachableFiles` -> `does not implement removeReachableFiles`",
+      "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 `runSILMandatoryOptPreDiffPasses(*sil_module);` instead of `runSILMandatoryOptPreDiffPasses(*swift_ast_ctx->GetSILModule());` because `swift_ast_ctx->GetSILModule()` returns a `std::shared_ptr<SILModule>`, and `runSILMandatoryOptPreDiffPasses` expects a `SILModule &`.",
+      "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": "Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "You left in a puts."
+    },
+    {
+      "prediction": "I don't think we need to override this method. The default implementation should be fine.",
+      "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 don't think we should remove the `gh-pages:ui-grid-site` task from the `release` task.",
+      "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 think this should be `// Process state values. Only values for Linux 3.14 and later are listed here.`",
+      "reference": "Nit: missing period."
+    },
+    {
+      "prediction": "I don't think we need to include the `# snippet-start:[s3.ruby.s3_add_bucket_ssekms_encryption_policy]` here.",
+      "reference": "how **to** deny Simple Storage **Service**"
+    },
+    {
+      "prediction": "I think this is too high. 15000 is the default and 25000 is too high. 10000 should be fine.",
+      "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 think this should be `if os.path.relpath(curSourceDir, fromPath) == \"addons\":`",
+      "reference": "I don't think this is necessary."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should return an error if we can't inspect a snapshot.",
+      "reference": "Instead of blindly ignoring all errors, this should just ignore the volume not found error."
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "The `rowPosition` will be ignored if the position column is not projected."
+    },
+    {
+      "prediction": "I think we need to use `zfs set io.openebs:livenesstimestamp='$(date)' cstor-$OPENEBS_IO_CSTOR_ID` instead of `zfs set io.openebs:livenesstimestap='$(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 if this is the right thing to do, but it seems to be the only way to get this to work.",
+      "reference": "This logic is already been performed in `CommandLine.defaultFactory().create(cls)` ... whats the point of repeating it here?"
+    },
+    {
+      "prediction": "Why do we need to take into account `countSourceCols` here?",
+      "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 thing to do, but I think we should check if `u_conn->fd` is not `-1` before calling `shutdown()`.",
+      "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": "Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Use the new Ruby 1.9 hash syntax.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br>Space inside } missing.<br>Space inside { missing.<br",
+      "reference": "Use the new Ruby 1.9 hash syntax."
+    },
+    {
+      "prediction": "This import should be moved to the bottom of the file.",
+      "reference": "Please host to a GitHub repo."
+    },
+    {
+      "prediction": "Shouldn't this be `s.Logger.Errorf(\"done split: %v\", err)`?",
+      "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 don't think we need to mention `loglevel-driver` and `loglevel-on-instance` here.",
+      "reference": "can we rename `fileLogLevelUsage` to be more generic like `instanceLogLevelUsage`? Same goes for other var below like fileLogLevelFlagName, FileLogLevel."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "Does this one not need the `tabContent` object used in the other files?"
+    },
+    {
+      "prediction": "Should this be `!b.IsManaged(clusterName)`?",
+      "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 should be `key == \"api_sig\" or key == \"signature\"`",
+      "reference": "Looks like all of these can be put into a list."
+    },
+    {
+      "prediction": "nit: I think it would be clearer to call this `originalKeyMapping` or something like that.",
+      "reference": "nit: \"*mapping\" for a map is unnecessary. Consider calling this `originalNames` or similar."
+    },
+    {
+      "prediction": "I think this should be `wx.NO | wx.CANCEL | wx.ICON_QUESTION, self` instead of `wx.OK | wx.CANCEL | wx.ICON_QUESTION, self`",
+      "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 thing to do. I think it should be `lastAutosensResult = new AutosensResult(\"autosens disabled\");` but I'm not 100% sure.",
+      "reference": "NPE here and in other APS plugins"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the router.",
+      "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 don't think you need to store these in the view. You can get them from the `ApplicationInfo` class.",
+      "reference": "These could be final"
+    },
+    {
+      "prediction": "I think this should be `invocation.addContext(cseContext);` instead of `invocation.setContext(cseContext);`",
+      "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())` instead of `Trace.TraceError(\"{0}\", sb.ToString())`.",
+      "reference": "Created a single error so it's not interleaved in log"
+    },
+    {
+      "prediction": "I don't think we should remove this check. It's not about IE9, it's about the fact that IE9 doesn't support `Blob` constructor.",
+      "reference": "`isIE9` was reverted. Shouldn't this condition be reverted as well?"
+    },
+    {
+      "prediction": "Why is this defaulting to `false` and not `true`?",
+      "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.kbfs.keybaseapi.com:443,bserver-1.kbfs.keybaseapi.com:443; bserver-0.kbfs.keybase.io:443,bserver-1.kbfs.keybase.io:443; bserver.kbfs.keybase.io:443`",
+      "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": "I'm not sure this is the right place to set the ping timer. The ping timer should be set when the client connects, not when the first ping is received. The ping timer should be reset when the first ping is received.",
+      "reference": "Why not go back to `c.setPingTimer()` here instead so you don't need the new boolean in setFirstPingTimer()."
+    },
+    {
+      "prediction": "This file is auto-generated. Please don't change it.",
+      "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": "I think we should keep the license header.",
+      "reference": "Was this our code @csrwng"
+    },
+    {
+      "prediction": "I don't think we want to do this in the shell, but rather in the ruby code.",
+      "reference": "I think you need `.untaint` here."
+    },
+    {
+      "prediction": "I don't think we need to check for `@path + 'tasks'` and `@path + 'files'` here. If the `plans_path` doesn't exist, then the `@path + 'tasks'` and `@path + 'files'` won't exist either.",
+      "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": "I don't think this import is 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 this should be `+kubebuilder:validation:Optional` instead of `+optional`",
+      "reference": "Can we rename this field to `Subject`? Looking around, it seems like 'subject' is the standard terminology for this stanza"
+    },
+    {
+      "prediction": "I think this should be `AddPeers`",
+      "reference": "Remove additional whitespace."
+    },
+    {
+      "prediction": "I think this should be `\"/\"` instead of `\"/v1/authn/login\"`.",
+      "reference": "We should add `Secure` as well"
+    },
+    {
+      "prediction": "I think this should be `TestWebdavNextcloud:`",
+      "reference": "I presume this isn't a big deal"
+    },
+    {
+      "prediction": "I don't think we need to pass in `true` for `update` here, since we're not actually updating the app's address space. I think we should pass in `false` for `update` and `true` for `image` instead.",
+      "reference": "Maybe {}, even though no multi-line body?"
+    },
+    {
+      "prediction": "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 `mbcs` instead of `utf-8`.",
+      "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 `NULL` instead of `0`.",
+      "reference": "This one is unrelated to the GetSymbolVendor removal ... the API to GetValueAsData was changed."
+    },
+    {
+      "prediction": "I think this should be `hostNetworkMode` instead of `bridgeNetworkMode`.",
+      "reference": "can you assert the value of this field below similar to other fields?"
+    },
+    {
+      "prediction": "I think this should be `| {@linkplain ASTClassLiteral ClassLiteral}`",
+      "reference": "I'll change that to \"linkplain\" for consistency :)"
+    },
+    {
+      "prediction": "I don't think this is the right fix. The `id_pattern` should be `\"[a-zA-Z0-9_-]{1,255}\"` and the `object_uri.replace('%2A', id_pattern)` should be `object_uri.replace('%2A', '[a-zA-Z0-9_-]{1,255}')`.",
+      "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'm not sure this is the right place to call this. I think it should be called in the `close()` 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": "Why is this change necessary? I don't see it in the PR description.",
+      "reference": "Use explicit swarm.Address when defining peerID, by using NewAddress, and remove conversion here."
+    },
+    {
+      "prediction": "I think this should be `/var/lib/rpm` and not `/var/lib/rpm/`",
+      "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": "rfp_inplanes is not used in BasicBlock, so it can be removed.",
+      "reference": "If we make a new backbone class, we don't need to support `BasicBlock`"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be an error if BPF is enabled and the platform doesn't support it.",
+      "reference": "This was an out-of-date dupe of the check on line 335"
+    },
+    {
+      "prediction": "why is this removed?",
+      "reference": "Let's put this back"
+    },
+    {
+      "prediction": "Why was this removed?",
+      "reference": "Things like these should have been addressed before submitting a pull request."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but it seems to be the only way to get the test to pass.",
+      "reference": "This one as well. Any better suggestion please?"
+    },
+    {
+      "prediction": "I think this should be `c.traceInOp(fmt.Sprintf(\"PUB %s %s %d\", c.pa.subject, c.pa.reply, c.pa.size), nil)`",
+      "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\") != 0` or `strcmp(P->short_name, \"s2\") == 0`",
+      "reference": "Why is this hack needed ? Ideally, we shouldn't need that."
+    },
+    {
+      "prediction": "I don't think we need the `MeshDriverName` field here.",
+      "reference": "Minor/Nitpick: replace `1` with a constant, esp. as used in `prog/weaver/http.go` in `{{if eq .Plugin.Version 1}}`"
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "nit: Looks like the whitespace got thrown off, can you fix this?"
+    },
+    {
+      "prediction": "I think this should be `CommandLine_ImportPeakBoundaries_Error__Failed_while_importing_boundaries_`",
+      "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,!plan9,!nacl,!js`",
+      "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": "I think it's better to use `TableColumn.physical(\"id\", DataTypes.BIGINT())` instead of `TableColumn.of(\"id\", DataTypes.BIGINT())`",
+      "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 don't think this is the right place to put this message. It should be in the `Messages` class.",
+      "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.receiveWindowSize/(1<<10)` instead of `inc/(1<<10)`",
+      "reference": "How often do we expect this to trigger? Should we maybe put it behind an if logger.Debug()?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "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 `kernel.hostname` instead of `hostname`.",
+      "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 this should be `req->res.content_length < self->args.min_size`",
+      "reference": "Should we better change the operator to `<` since the variable defines the minimum size that gets compressed?"
+    },
+    {
+      "prediction": "I think it would be better to have a separate test for `Stream#append` and `Stream#prepend`.",
+      "reference": "I'm really impressed by the AppendElements addition! Very cool! :-)"
+    },
+    {
+      "prediction": "I don't think this should be `finest`.",
+      "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 don't think this is a good idea. If we ever want to use more than 256 worlds, we'll have to change this again.",
+      "reference": "This should be in qtutils and imported to here and the other place instead of declaring it twice."
+    },
+    {
+      "prediction": "I think this should be `V(logf.DebugLevel).Info(\"Calling FetchCertAlternatives\")`",
+      "reference": "Seems a bit weird to add a timeout in \"logging\" middleware, but I see that that is done elsewhere, so fine."
+    },
+    {
+      "prediction": "I think this should be `messages.delete_singular` instead of `messages.delete_singular_form`",
+      "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 `if (Object.keys(updatedFiles).length === 0 && Object.keys(this.state.files).length === 0)`",
+      "reference": "Could we combine this with the `setState` call above, so we don't have to call it twice?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 don't think we need to pass `cfg.DB.DbPath` to `db.NewBoltDB(cfg.DB, cfg.DB.DbPath)`, because `db.NewBoltDB(cfg.DB, cfg.DB.DbPath)` will use `cfg.DB.DbPath` anyway.",
+      "reference": "Is it a doable way and does it make sense."
+    },
+    {
+      "prediction": "I think this should be `max_query_limit` instead of `max`",
+      "reference": "max was wrong here. As its buitlt-in, not a number, and a TypeError was thrown."
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "This change is not needed, please revert it."
+    },
+    {
+      "prediction": "I don't think this is the right place to add this. I think it should be in the docstring of the `train` function.",
+      "reference": "@ClimbsRocks Please add that it could be a string. `feval : callable, string or None, optional (default=None)`"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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 `platformType` rather than `ddev`",
+      "reference": "Not sure why this is changing to a hard-coded string."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `Command` class with a `run` method that takes a `command` string and returns a `CommandResult` object. The `CommandResult` object would have a `__str__` method that returns a string representation of the result. The `Command` class would also have a `__str__` method that returns a string representation of the command. The `Command` class would also have a `__repr__` method that returns a string representation of the command. The `Command` class would also have a `__eq__` method that returns `True` if the command is equal to another command. The `Command` class would also have a `__ne__` method that returns `True` if the command is not equal to another command. The `Command` class would also have a `__lt__` method that returns `True` if the command is less than another command. The `Command` class would also have a `__le__` method that returns `True` if the command is less than or equal to another command. The `Command` class would also have a `__gt__` method that",
+      "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": "Please revert this change. It is not related to this PR.",
+      "reference": "Why did we update the to this api?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but I think it's better than what we had before.",
+      "reference": "should probably just catch `IllegalArgumentException` here"
+    },
+    {
+      "prediction": "I don't think this is the right place to add the doc link. I think it should be added to the `NewWindow` function instead.",
+      "reference": "What does this error message mean?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do. I think we should be able to pass in a unicode string and have it be converted to a utf-8 byte string.",
+      "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": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "reference": "Can we move this into a method(s)?"
+    },
+    {
+      "prediction": "I think we can just use `bytes.IndexByte(n, 0)` here.",
+      "reference": "You could perhaps call this `strlen` (although I don't particularly care about the name as it is an implementation detail)."
+    },
+    {
+      "prediction": "I'm not sure why this was removed.",
+      "reference": "should we have a common `Escape()` method that can be used everywhere?"
+    },
+    {
+      "prediction": "I think this is a typo.",
+      "reference": "The whitespace removal above here is probably unintentional :slightly_smiling_face:"
+    },
+    {
+      "prediction": "I don't think we need to start the p2p agent before the chainservices.",
+      "reference": "should we start p2p even after dispatcher? b/c dispatcher handles msg from p2p"
+    },
+    {
+      "prediction": "I think this should be `assert.JSONEq(t, `[\"a\", \"b\", \"c\"]`, reqBuf.String())`",
+      "reference": "Did we decide once and for all to use req/resp throughout spring?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it's the only way I could think of to test that the link is rendered with the correct text.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I don't think we should be deleting the scratch bucket here. I think we should just return an error and let the user know that the scratch bucket is not in the project and they need to create a new one.",
+      "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 this should be `allow.restart.on.execution.stopped` instead of `allow.restart.on.execution.stopped`.",
+      "reference": "Lets move the whole string to next line for better readability"
+    },
+    {
+      "prediction": "I think this should be `Could not save WebCal-calendar: {reason}`",
+      "reference": "Please change this back to `Error saving WebCal-calendar` (and `Error saving WebCal-calendar` only)"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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 we can remove this constant now.",
+      "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 think this should be `EntrySize+crypto.Extension+(uint(headerLengthSize)*uint(len(p.blobs))) <= uint(maxHeaderSize)`",
+      "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": "This enum is not used anymore, so it can be removed.",
+      "reference": "Remove selector if only 1 version is supported"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it's better to keep all the items in the menu, even if they don't do anything for HTML notes. What do you think?",
+      "reference": "Why did you remove more than the asked command?"
+    },
+    {
+      "prediction": "I think this should be `if err != nil && utils.Debug() {`",
+      "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 this should be `sanitize=False` by default.",
+      "reference": "why is the `sanitize=True` here in the docs?"
+    },
+    {
+      "prediction": "Line is too long. [85/80]",
+      "reference": "Didn't end up using this, but I think it's useful anyway."
+    },
+    {
+      "prediction": "I think this should be `IndexOf(\"\\\\\", relativePath.IndexOf(\"\\\\\") + 1) + 1` so that we don't include the first backslash in the substring.",
+      "reference": "I'm pretty sure there's already a metadata item that contains what you're calculating here (PackageRelativePath)?"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [114/100]",
+      "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": "I don't think you need to include all of these here, just the new ones.",
+      "reference": "Why `2` in this and others?"
+    },
+    {
+      "prediction": "Is there a way to get the parent process id of the current process? I am not sure if we can get the parent process id of the current process.",
+      "reference": "This may cause a new allocation, please consider merging the concat in above line. Same applies to change in dotnethostmanager."
+    },
+    {
+      "prediction": "I don't understand why this is needed. `importedname` is already `.my_package1` or `.my_package2`.",
+      "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 these methods. The default implementation should be enough.",
+      "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 think this should be `notification` instead of `Notification` to be consistent with the rest of the codebase.",
+      "reference": "nit: `Additional configuration used while sending notification to external services.`"
+    },
+    {
+      "prediction": "I think this should be `nameOrHandle` instead of `handle`.",
+      "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": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "This should have failed before."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these types. I think they should be in the `paymentchannel` package, but I don't think they should be in the `paymentchannel/types` package. I think they should be in the `paymentchannel` package, but not in the `paymentchannel/types` package. I think they should be in the `paymentchannel` package, but not in the `paymentchannel/types` package. I think they should be in the `paymentchannel` package, but not in the `paymentchannel/types` package. I think they should be in the `paymentchannel` package, but not in the `paymentchannel/types` package. I think they should be in the `paymentchannel` package, but not in the `paymentchannel/types` package. I think they should be in the `paymentchannel` package, but not in the `paymentchannel/types` package. I think they should be in the `paymentchannel` package, but not in the `paymentchannel/types` package. I think they should be in the `paymentchannel` package, but not in the `paymentchannel/types` package. I think they should be in the `paymentchannel` package, but not in the `paymentchannel/types",
+      "reference": "the paymentchannel dir is where the paymentchannel manager will live. it will store the types below."
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "This is Safari specific and should be defined somewhere in the `safaridriver` namespace"
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "assignments should only be cuddled with other assignments (from `wsl`)"
+    },
+    {
+      "prediction": "I think this should be `fontWeight + ' ' + fontSize + ' ' + fontFamily;`",
+      "reference": "prefer changing the signature by adding params to the end, not the middle, I think."
+    },
+    {
+      "prediction": "I think this should be `p.Spec.RemoteWrite` instead of `p.Spec.Thanos`.",
+      "reference": "I guess we aim for hardcoded port for Prometheus operated ports right?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to work.",
+      "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 `--musicbrainz-ids` or `--musicbrainz-ids`",
+      "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 if this is correct, but I think it should be `ticks := uint32(d) / 100`",
+      "reference": "This code is necessary. It makes sure that you can sleep for longer durations than the timer supports."
+    },
+    {
+      "prediction": "nit: `executeDeleteWith` -> `executeDeleteManifestsAndDataFilesWith`",
+      "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 think this should be `if (backend->impl && backend->impl->destroy) { backend->impl->destroy(backend); } else { free(backend); }`",
+      "reference": "If you're going to start testing that impl itself is not NULL you should probably be doing it everywhere."
+    },
+    {
+      "prediction": "I think this should be `configFileNames` to be consistent with the rest of the code.",
+      "reference": "`ctx` is unused in LoadPlainYAMLManifests"
+    },
+    {
+      "prediction": "nit: I think we can remove the `final` here.",
+      "reference": "why is this called toBuilder when the method is called fromBuilder? (I realise you did not change this...)"
+    },
+    {
+      "prediction": "I think this should be `source.reader.fetch-batch-size`",
+      "reference": "nit: seems rather large."
+    },
+    {
+      "prediction": "I think this should be `int(time.time() * 1000)` to be consistent with the rest of the code.",
+      "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 `grpcClientTypeName` and `mockGrpcClientTypeName` to be consistent with the rest of the codebase.",
+      "reference": "Use existing `grpcStubCallString` instead"
+    },
+    {
+      "prediction": "I think this should be `ulong.MaxValue`",
+      "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 best way to do this. I think it would be better to have the catchpoint writing be a separate goroutine, and have the catchup process wait for it to complete. That way, we don't have to worry about the catchup process timing out, and we don't have to worry about the catchup process being interrupted by the catchpoint writing. What do you think?",
+      "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 (str): Model name to create.`",
+      "reference": "Do you still need the `name` in the arg description, if this is not `str` type anymore.?"
+    },
+    {
+      "prediction": "I think it's better to remove this class.",
+      "reference": "We may raise a warning."
+    },
+    {
+      "prediction": "I think this should be `icd10Code: DS.attr('string', {defaultValue: ''})`",
+      "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 works for now.",
+      "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 we need this constructor. The `trustedIssuers` is a `Collection` and the `trustedIssuer` is a `Predicate`. I don't see how they are related.",
+      "reference": "Will you please update the copyright message to now include `2021`?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `&data[offsets[index]]` and `&data[offsets[index + 1] - 1]`.",
+      "reference": "Potential subscript out of range."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "please undo this change. you didn't really meant to make it, right ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work for now.",
+      "reference": "To check! are these millseconds or seconds :thinking: The docs aren't clear"
+    },
+    {
+      "prediction": "I think this should be `if num > int64(u.ctx.MaxUploadParts) || num > int64(MaxUploadParts) || num > int64(u.ctx.MaxUploadSize) / int64(u.ctx.PartSize) {`",
+      "reference": "Why was this moved?"
+    },
+    {
+      "prediction": "Why do we need this?",
+      "reference": "What are these object with suffix Expansion?"
+    },
+    {
+      "prediction": "I think this should be `true` by default. I don't think we want to allow unprotected transactions by default.",
+      "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": "I don't think you need to use `using` here. `typedef` is fine.",
+      "reference": "I'm kind of surprised this doesn't require `typename`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be able to use `bison --yacc` instead of `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 don't think we need this. We can just use `BlockAuthorAccount` and `EnodeAccount` as the same account.",
+      "reference": "can we explain to users in the docs what happens if they leave the field blank?"
+    },
+    {
+      "prediction": "Please add a comma after `user_data=None`",
+      "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 don't think this change is necessary. `File.separator` is always a single character, so `replaceAll` and `replace` should have the same effect.",
+      "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": "`ProductNotFoundException` should be renamed to `ProductGoneException`",
+      "reference": "It could be nice to be explicit about this change in upgrade notes, what do you think?"
+    },
+    {
+      "prediction": "I think this should be `companion.provider.authProviderName` instead of `companion.provider.authProvider`",
+      "reference": "spotted a bug."
+    },
+    {
+      "prediction": "I think this should be `(int) (Math.random()*(rowsPerAnimal/matchingRowsPerAnimal))`",
+      "reference": "We were always using 0000000 prefix"
+    },
+    {
+      "prediction": "Why do we need to check `gJobMgr` here?",
+      "reference": "Here we also need to determine if gJobMgr has called the init function."
+    },
+    {
+      "prediction": "Style/CaseEndAlignment: Align the keyword with the previous line.",
+      "reference": "Out of curiosity, why is `condition.any?` also necessary here?"
+    },
+    {
+      "prediction": "`std::shared_ptr<HostManager> hm(new HostManager(space));`",
+      "reference": "Always use `std::make_shared` whenever possible, it will save you one memory allocation."
+    },
+    {
+      "prediction": "I think this should be `v0.3.0` or `v0.4.0`",
+      "reference": "Update to real tag before pull."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the health endpoint to show up in prometheus.",
+      "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 don't think we need to implement `ICoreLogger` here.",
+      "reference": "Adding ICoreLogger here lets us pull this into Core as a strategy"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "why did you change this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the formatter.",
+      "reference": "This is invalid syntax on 1.8."
+    },
+    {
+      "prediction": "I'm not sure this is a good change. I think we want to keep the old behavior. The old behavior is that the `[O:2]` and `[N:3]` atoms are attached to the `[C:1]` atom. The new behavior is that the `[O:2]` and `[N:3]` atoms are not attached to the `[C:1]` atom.",
+      "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 think this should be `uint64` instead of `Uint64`",
+      "reference": "Is this required for tuple encoding? This is confusing."
+    },
+    {
+      "prediction": "nit: I think we can remove the `FlowPreparer` from `FlowRunnerManager` since it is only used in `FlowRunner`.",
+      "reference": "Does it need to be a member variable? It is currently only used in one method."
+    },
+    {
+      "prediction": "I think this should be `apiClient.sendSyncPlayCommand(sessionId, 'ListGroups')`",
+      "reference": "@MrTimscampi doesn't this need an update to apiclient?.."
+    },
+    {
+      "prediction": "I think this should be `#if defined(X86_64)` instead of `#if defined(X86_64) || defined(X64)`.",
+      "reference": "This is a nop: prob best for history to not change the line"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should just return an error here.",
+      "reference": "Can you delete this whole file now?"
+    },
+    {
+      "prediction": "`ArrPassing` -> `ArrayPassing`",
+      "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 think this should be `recyclerAdapter.setKeepSorted(UserPreferences.isQueueKeepSorted());` instead of `recyclerAdapter.setKeepSorted(UserPreferences.isQueueKeepSorted());`",
+      "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.SiteRunning`",
+      "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": "nit: I think we can remove the `// 14)` here.",
+      "reference": "Are you sure `v3EndpointIDToContainerName` and `v3EndpointIDToTask` are saved in the state file, can you verify that?"
+    },
+    {
+      "prediction": "I think this mutex is unnecessary. I don't think we have any concurrent access to the `cryptoSetupServer`.",
+      "reference": "This mutex is never used. Should it be, if users can now make calls into the crypto setup?"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "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 don't think this is the right way to do it. If the exchange rate is `None` it means that the currency is not supported, so we should show an error message to the user.",
+      "reference": "Should this be `None` instead?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. If the key is already in the source, we should not add it again.",
+      "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 don't think we need to add `save`, `delete`, `list` and `prune` here.",
+      "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": "I think this should be `getActiveFlows` instead of `getRunningFlows`.",
+      "reference": "why returning a string instead of a list?"
+    },
+    {
+      "prediction": "I don't think we should use the `nameof(IntegrationIds.AdoNet)` here. The `nameof(IntegrationIds.AdoNet)` is used in the `AdoNetConstants` class, which is used in the `AdoNetCommandIntegration` class. The `AdoNetCommandIntegration` class is used in the `NpgsqlCommandIntegration` class. The `NpgsqlCommandIntegration` class is used in the `NpgsqlConstants` class. The `NpgsqlConstants` class is used in the `NpgsqlCommandIntegration` class. The `NpgsqlCommandIntegration` class is used in the `NpgsqlConstants` class. The `NpgsqlConstants` class is used in the `NpgsqlCommandIntegration` class. The `NpgsqlCommandIntegration` class is used in the `NpgsqlConstants` class. The `NpgsqlConstants` class is used in the `NpgsqlCommandIntegration` class. The `NpgsqlCommandIntegration` class is used in the `NpgsqlConstants` class. The `NpgsqlConstants` class is used in the `NpgsqlCommandIntegration` class. The `NpgsqlCommandIntegration` class is used in the `NpgsqlConstants` class. The `NpgsqlConstants` class is used",
+      "reference": "Was there a reason for changing all these integration names?   Are they actually used for anything besides information in logs?"
+    },
+    {
+      "prediction": "I don't think this needs to be synchronized.",
+      "reference": "we should not resurrect this method that we just removed"
+    },
+    {
+      "prediction": "This should be `return self.get_screenshot_as_file(filename)`",
+      "reference": "This needs to be just return self.get_screenshot_as_file(filename)"
+    },
+    {
+      "prediction": "I don't think we need to check `readyState !== 0` here. `readyState === 0` means that the request has not been sent yet, and `readyState === 4` means that the request has been sent and the response has been received.",
+      "reference": "Shouldn't this return false here?"
+    },
+    {
+      "prediction": "I think we should make this a `uint32` instead of a `*uint32` so that we don't have to check for `nil` all over the place.",
+      "reference": "Should have json tag"
+    },
+    {
+      "prediction": "I think this should be `static` instead of `STATIC`.",
+      "reference": "Is this needed after all? Looks like the struct was re-defined inside the test file."
+    },
+    {
+      "prediction": "I think this should be `RemoveReachableFiles removeReachableFiles(String metadataLocation)`",
+      "reference": "I think the name of the method should match the name of the action: `removeReachableFiles`."
+    },
+    {
+      "prediction": "I think this should be `Manage consent types`",
+      "reference": "can you make the C in Consent lower case to match the other ones in this list?"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "fix the group"
+    },
+    {
+      "prediction": "I think we should use `anyMatch` instead of `count > 0`",
+      "reference": "Prefer `Stream.anyMatch` instead of iterating over all slots."
+    },
+    {
+      "prediction": "Why is this change necessary?",
+      "reference": "we have `logging.TestingLog()` that should be used for that purpose."
+    },
+    {
+      "prediction": "I think this should be `if (i < src_len - 2 && src[i] == '%' && isxdigit(* (const unsigned char *) (src + i + 1)) && isxdigit(* (const unsigned char *) (src + i + 2))) {`",
+      "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 %s 2> /dev/null` instead of `pidof %s 2>&1`",
+      "reference": "Why change it?"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "nb: this change is unnecessary"
+    },
+    {
+      "prediction": "I think we should rename this to `PublicURL` or `PublicLinkURL` to make it more clear that it's a URL.",
+      "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": "Should this be `omitempty` as well?",
+      "reference": "I don't think the Internal API types need the json annotations. They are never serialised, right?"
+    },
+    {
+      "prediction": "I think we should rename this to `ReportApplicationDeploymentCompleted` to make it more clear that it's about the application deployment and not the pipeline deployment.",
+      "reference": "Maybe this was added incidentally. We don't need this function in the interface."
+    },
+    {
+      "prediction": "I think this should be `strings.HasSuffix(v, \"/\") && !strings.HasSuffix(t, \"/\")`",
+      "reference": "Is this a safe/reliable way to split the URL?"
+    },
+    {
+      "prediction": "This should be moved to the end of the method.",
+      "reference": "do we also need to profile the time to getConnection()?"
+    },
+    {
+      "prediction": "I think this should be `errors=\"replace\"`.",
+      "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 think this should be `legacy-holds-list` for consistency with the other legacy routes.",
+      "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 think this should be `Infof` instead of `Noticef`.",
+      "reference": "I think if yo use %s then you do not need reason.String() and can just do reason."
+    },
+    {
+      "prediction": "I think this should be `wf-task-timeout`",
+      "reference": "Having \"wf\" prefix doesn't make sense anymore. Please remove."
+    },
+    {
+      "prediction": "I think we can remove this now that it's in the helpers file.",
+      "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` and `DEFAULT_MAX_MEMORY` instead of `DEFAULT_CPU_LIMIT` and `DEFAULT_MEMORY_LIMIT`.",
+      "reference": "Default max cpu should be 8 and memory 64GB"
+    },
+    {
+      "prediction": "I don't think we need to mention `itemsize` here.",
+      "reference": "@itholic, can you remove `nbytes = unsupported_property('nbytes')` at `_MissingPandasLikeSeries`?"
+    },
+    {
+      "prediction": "I think this test should be updated to check that `TaskCPUMemLimit` is set to `true`.",
+      "reference": "What happened to the assertion on `conf.TaskCPUMemLimit`?"
+    },
+    {
+      "prediction": "I think this should be `createRef: () => ({ current: null })`",
+      "reference": "Just curious: Is using `Object` faster than a literal `{}`?"
+    },
+    {
+      "prediction": "I think this should be `specPath.Child(\"duration\").Child(\"duration\")` and `specPath.Child(\"renewBefore\").Child(\"duration\")`",
+      "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": "Why is this removed?",
+      "reference": "Should we remove the `Aliases` field as well?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in `onResume` or `onStart` of the `VoiceSearchActivity`.",
+      "reference": "what's the effect of turning these two off? is there an issue on file for context?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the only thing that makes sense. We want to test the latest beta version of gcloud with the latest release of the wrapper, and the latest GA version of gcloud with the latest release of the wrapper. I don't think it makes sense to test the latest beta version of gcloud with the latest beta version of the wrapper, or the latest GA version of gcloud with the latest beta version of the wrapper. I don't think it makes sense to test the latest beta version of gcloud with the latest GA version of the wrapper, or the latest GA version of gcloud with the latest GA version of the wrapper. I don't think it makes sense to test the latest beta version of gcloud with the latest release of the wrapper, or the latest GA version of gcloud with the latest release of the wrapper. I don't think it makes sense to test the latest beta version of gcloud with the latest beta version of the wrapper, or the latest GA version of gcloud with the latest beta version of the wrapper. I don't think it makes sense to test the latest beta version of gcloud with the latest GA version of the wrapper, or the latest GA version of",
+      "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 think this should be IEnumerable<KeyValuePair<string, string>>.",
+      "reference": "Are we renaming this too in a follow up PR?"
+    },
+    {
+      "prediction": "This should be `sa` not `ar`",
+      "reference": "The indentation on this is off by one space to the left"
+    },
+    {
+      "prediction": "nit: extra space before `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": "This import is not needed.",
+      "reference": "Is order really irrelevant here?"
+    },
+    {
+      "prediction": "This import is not required.",
+      "reference": "I think this import will be unused now. If it is unused remove it."
+    },
+    {
+      "prediction": "I think this should be called `NewStatelessRwTx` or `NewStatelessRwTxWithKV` or something like that.",
+      "reference": "Let's change to tx"
+    },
+    {
+      "prediction": "I think this should be 0.05 or 0.01. 0.10 is a lot.",
+      "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\") != \"\"`",
+      "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": "Style/TrailingCommaInHashLiteral: Avoid comma after the last item of a hash.",
+      "reference": "Style/SymbolArray: Use %i or %I for an array of symbols."
+    },
+    {
+      "prediction": "I think this should be `assert.NoError(checkDockerCompose())` instead of `assert.NoError(checkDockerCompose)`.",
+      "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 this should be done in the `OAuth20RestTemplate` class, not here.",
+      "reference": "I'm slightly concerned about this because doesn't ORCID use XML instead of JSON?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to handle this, but it seems to work.",
+      "reference": "brackets around if conditions isn't really pythonic."
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "reference": "It turn out that when we close server it closed underlying listener itself, so this line always generated and error."
+    },
+    {
+      "prediction": "Please revert these changes. They are not needed.",
+      "reference": "Looks like this is for debugging? Maybe just remove."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 `m_indirectKernelMem.reserve(0x1440);` instead of `for (auto& usm_mem : m_indirectKernelMem) { usm_mem.reset(*m_queue, m_instance_id); // TODO 0x1440= 5184, arbitrary, larger than largest encountered kernel. usm_mem.reserve(0x1440); }`",
+      "reference": "I'm not quite sure that we preallocate memory. Do you have a good reason for that?"
+    },
+    {
+      "prediction": "I don't think we need to lock here.",
+      "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's better to use `HostAddr targetLeader = singleReplica_ ? src_ : nebula::kRandomPeer;` instead of `HostAddr targetLeader = singleReplica_ ? src_ : kRandomPeer;`",
+      "reference": "if the targetLeader is src_ itself, it is really need call transLeader function?"
+    },
+    {
+      "prediction": "I think we can remove the `method` parameter from this method. It's only used in `getAndSavePagedResponseTypeName` and `getAndSavePagedResponseTypeNameAsync`, and both of those methods have the `method` parameter.",
+      "reference": "Can we remove this featureConfig since you removed in L98 (assuming it is not used else where)"
+    },
+    {
+      "prediction": "nit: remove unused `using System.Collections.Generic;`",
+      "reference": "namespaces in this file need a sorting"
+    },
+    {
+      "prediction": "I think this should be `if c.Wireguard.EncryptHostTraffic && c.KubernetesProvider == config.ProviderAKS && c.RouteSource == \"WorkloadIPs\"`",
+      "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 `maxSegmentsPerCell=maxSegmentsPerCell or 255, maxSynapsesPerSegment=maxSynapsesPerSegment or 255`",
+      "reference": "`get` has an optional second parameter that is the default if the key isn't found."
+    },
+    {
+      "prediction": "Shouldn't this be `echo \"<id>$user->id</id>\\n\";`?",
+      "reference": "Please also indent the new response line as was done to the existing line above."
+    },
+    {
+      "prediction": "This should be `if (this->GetTarget() == this) {`",
+      "reference": "Shouldn't need this-> here."
+    },
+    {
+      "prediction": "I think this should be `MESSAGE_TRADE_REQUEST` 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'm not sure why this is needed. The `CreateGraphDetections` method is already in the `if (split[1] == typeof(MassErrorGraphController).Name)` block.",
+      "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": "Why did you remove the default export?",
+      "reference": "I know we're deleting this soon but I thought I'd fix it anyway :smile:"
+    },
+    {
+      "prediction": "I think you can use `args.get(0).getType()` instead of `ASTHelpers.getType(tree.getArguments().get(0))`",
+      "reference": "I like this, makes it clearer what scope the continuation is in"
+    },
+    {
+      "prediction": "Shouldn't this be `A1`?",
+      "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 think we can remove the `tmpdir` and `OnLoadCommit` arguments here.",
+      "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 think we can remove this interface.",
+      "reference": "Similarly, this has only one implementing class"
+    },
+    {
+      "prediction": "I don't think we need to set the `Content-Type` header here. It's only used when sending a POST request to the server, and we don't do that here.",
+      "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` instead of `?PhpParser\\NodeAbstract`",
+      "reference": "I must have forgotten that one in previous PR"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [101/80]",
+      "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 we can remove the `m.Runtask.Spec.Task` and `m.Values` from the `AsTemplatedBytes` function.",
+      "reference": "Can this be `BuilderForYAMLObject`"
+    },
+    {
+      "prediction": "I don't think this is necessary. The `pool.close()` should be called in the `finally` block of the `test_multiprocessing_threadpool` function.",
+      "reference": "better use `with mp.Pool(...) as pool:` here"
+    },
+    {
+      "prediction": "Shouldn't this be `Response.Status.INTERNAL_SERVER_ERROR`?",
+      "reference": "Since this is a config problem, BAD_REQUEST is probably not the right response."
+    },
+    {
+      "prediction": "I think this should be `if dur <= 30*time.Second {`",
+      "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 we can remove the `this_name == that_name` check here, since we already have `this_column == that_col` in the `if` statement.",
+      "reference": "can you rename `that_col` to `that_column`?"
+    },
+    {
+      "prediction": "I think this should be `components` instead of `componenets`.",
+      "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": "If kv_ is nullptr, we should not update the last update time.",
+      "reference": "Each task finished will result in the meta info updated?"
+    },
+    {
+      "prediction": "Shouldn't this be `trim()` instead of `replace( /\\s/g, '' )`?",
+      "reference": "Quick follow-up @aaemnnosttv, why this and not `.trim()`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to go about it, but it's the only way I could get the tests to pass.",
+      "reference": "I don't think we should have file-wide exceptions for this rule, let's annotate the individual instances."
+    },
+    {
+      "prediction": "Why is this here?",
+      "reference": "@raprasad you plan to take this \"if true\" out, right?"
+    },
+    {
+      "prediction": "I think this should be `os.path.expandvars(path)` instead of `os.path.expanduser(path)`. `os.path.expanduser` only expands `~` and `~user`, while `os.path.expandvars` expands all environment variables.",
+      "reference": "Could you adjust this to `with open(path, \"rb\"):` here and below? We should make sure that we close all files."
+    },
+    {
+      "prediction": "nit: move this to the top of the file.",
+      "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 think this should be `applyPreflightCorsResponse(resp, req)` so that we have access to the `Origin` header in the `applyPreflightCorsResponse` function.",
+      "reference": "Why conditional check is done? As i understand, later both `applyPreflightCorsResponse()` and `allowAllCorsActions()` does the same"
+    },
+    {
+      "prediction": "I don't think this is the right place to put this warning. I think it should be in the `OnUnregister` method.",
+      "reference": "has not yet exit?"
+    },
+    {
+      "prediction": "I don't think we should deprecate these fields. They are still used in the wild.",
+      "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.finer` instead of `finest`. `finest` is the lowest level of logging, and I don't think we want to log this at that level. `finer` is the next lowest level, and I think it's appropriate for this level of logging.",
+      "reference": "We chose `info` in the test code for obvious reasons. Changing to `finest` makes debugging harder and noisier."
+    },
+    {
+      "prediction": "Nit: I think it would be better to rename `codeFlowLocations` to `stackLocations` and `stackLocations` to `codeFlowLocations`.",
+      "reference": "This is just a rename..."
+    },
+    {
+      "prediction": "Shouldn't this return an error instead of an empty string?",
+      "reference": "Should probably panic here. I think hashers are contracted not to return errors (and returning \"\" doesn't handle the error)"
+    },
+    {
+      "prediction": "I think this should be `files.Directory` as well.",
+      "reference": "Aside: this is a confusing name for an interface that contains multiple addresses."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we want to return the exit code from the command that was run, not the exit code of the `bolt task run` command itself.",
+      "reference": "We expect to use Bolt::CommandResult for scripts as well?"
+    },
+    {
+      "prediction": "I think we can remove the `superclass.method(:next_runnable_index_for)` and just pass `superclass` as the first argument to `Metadata::ExampleGroupHash.create`.",
+      "reference": "same thing as above inre commas and args?"
+    },
+    {
+      "prediction": "I think this should default to `true`",
+      "reference": "Should we validate that this is between 0 and 100?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do.",
+      "reference": "`HardKiller` doing `os.Exit()` without proper shutting down other dependencies. I think there should be a better approach for this."
+    },
+    {
+      "prediction": "I think it would be nice to have a `wlr_drm_format_set_add` that takes a `const uint32_t` instead of a `uint32_t` so we don't have to cast here.",
+      "reference": "Why isnt the return value checked?"
+    },
+    {
+      "prediction": "I think this should be removed.",
+      "reference": "remove this debug"
+    },
+    {
+      "prediction": "Shouldn't this be `m_child_layers.end()` instead of `m_child_layers.begin()`?",
+      "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 think it would be better to use `endswith` instead of `==`.",
+      "reference": "what is this change about?"
+    },
+    {
+      "prediction": "I don't think this is needed anymore.",
+      "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 `VerifyTrustDomainMemberID` instead of `VerifySameTrustDomain` since it verifies that `id` is a member of `td` and not that `id` and `td` are the same.",
+      "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 we can remove this `MinTimes(0).MaxTimes(1)` as it is already covered by the `CreateContainerExec(gomock.Any(), gomock.Any(), gomock.Any(), gomock.Any()).Return(&types.IDResponse{ID: containerID}, nil).MinTimes(0).MaxTimes(1)` above.",
+      "reference": "hmm.. this is Hacky, the test will succeed in Windows even if you remove the calls to these, right?"
+    },
+    {
+      "prediction": "I don't think we need to do this. We can just pass in the ruby version as a job parameter.",
+      "reference": "should this block be in the file `05_install_ruby.rb`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to be the only way to get the recorder to work.",
+      "reference": "To match outbounds, let's just call this `Unary: recorder`, `Oneway: ...`."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should be using `hash.update(string.encode('utf-8'))` instead.",
+      "reference": "`string.encode('utf-8')` is more common but I guess this is mostly cosmetic :)"
+    },
+    {
+      "prediction": "I think we should keep the `uri` field.",
+      "reference": "Consider logging `r.Url.Path`? Either way, ship it!"
+    },
+    {
+      "prediction": "I think this should be `minTealVersion := logic.ComputeMinTealVersion(contextHdr, group)`",
+      "reference": "this probably should be done lazely only if logic/app call txn in the group. Or even done in LogicSigSanityCheck?"
+    },
+    {
+      "prediction": "nit: I think it's better to set it to `CONTINUE_AS_NEW_INITIATOR_UNSPECIFIED` instead of `CONTINUE_AS_NEW_INITIATOR_WORKFLOW` here, and set it to `CONTINUE_AS_NEW_INITIATOR_WORKFLOW` in `addWorkflowExecutionStartedEventForContinueAsNew` and `addWorkflowExecutionStartedEventForContinueAsNewWithInitiator`.",
+      "reference": "Please move this to `common/enums/defaults.go`."
+    },
+    {
+      "prediction": "This import is not used.",
+      "reference": "In general, if you find unalphabetized imports, you should alphabetize them. Fine for now though."
+    },
+    {
+      "prediction": "This should be `duraton` -> `duration`",
+      "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'm not sure this is the best way to handle this, but I think it's the best we can do for now. The problem is that we don't know what the units are for `explicitIonMobility` and `explicitCompensationVoltage`. If `explicitIonMobilityUnits` is `compensation_V`, then `explicitIonMobility` and `explicitCompensationVoltage` should be the same value. If `explicitIonMobilityUnits` is not `compensation_V`, then `explicitIonMobility` and `explicitCompensationVoltage` should be different values. The problem is that we don't know what the units are for `explicitIonMobility` and `explicitCompensationVoltage`, so we don't know if they should be the same or different values.",
+      "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 way to do this. I think it would be better to have a function that takes a `key` and returns a list of strings. Then you can join them with spaces in the `bulk_send` function.",
+      "reference": "Eliminate this function; handled by mappings and analyzer"
+    },
+    {
+      "prediction": "Shouldn't this be `$datasource->getRecordRoute($id)`?",
+      "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 don't think we should be using `finest` here. `finer` seems more appropriate.",
+      "reference": "This is in test code: understanding what we're doing is important in this context."
+    },
+    {
+      "prediction": "I think this should be `import llvm \"llvm.org/llvm/bindings/go/llvm\"`",
+      "reference": "This change (and a few similar ones below) are not related to rpi3 support, and should be removed."
+    },
+    {
+      "prediction": "I think we can remove this line.",
+      "reference": "Hmm, why did we come to need the `,` at the end?"
+    },
+    {
+      "prediction": "I don't think this is correct. `query._fields[paths[i]]` will always be `true` if `query._fields` contains `paths[i]` because `query._fields[paths[i]]` is set to `true` by `query.select(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 should be a `map[string]*Statement` where the key is the name of the statement.",
+      "reference": "What is this? I'm a bit unclear why the parser would have a list of statements in it."
+    },
+    {
+      "prediction": "I think this should be `prometheus_operator_enforced_sample_limit_total` to be consistent with the other metrics.",
+      "reference": "I believe this is Prometheus name, wonder if this is descriptive enough of a label name? @nrchakradhar @simonpasquier wdyt?"
+    },
+    {
+      "prediction": "I think this should be `false` by default. I don't think we want to drop VXLAN and IP-in-IP traffic from workloads by default.",
+      "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 don't think we need to pass `removeSmallerNonces` to `RemoveTransaction` method. We can just call `RemoveTransaction(hash, blockNumber, true)` in `RemovePendingTransaction` and `RemoveTransaction(hash, blockNumber, false)` in `RemoveConfirmedTransaction`. What do you think?",
+      "reference": "void RemoveTransactions(Address sander, long removeBelowThisNonce) and separate these two calls"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "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.",
+      "reference": "nit: should not need this."
+    },
+    {
+      "prediction": "I think this should be in `ProjectProperties` instead of `Constants`.",
+      "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 think this should be `isinstance(x, (list, tuple))`",
+      "reference": "It would be nice if we have some docs explaining this. Core luigi should avoid having hard to understand code."
+    },
+    {
+      "prediction": "I don't think we need the `ignoreCase` parameter here.",
+      "reference": "extract a method to call from a test"
+    },
+    {
+      "prediction": "I think this should be `private final ExecutorService collectorExecutor;` and `collectorExecutor = Executors.newFixedThreadPool(1, new DefaultSolrThreadFactory(\"collectorExecutor\"));`",
+      "reference": "nitpick: regular order is private than final."
+    },
+    {
+      "prediction": "I think this should be `missing-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 think these should be in alphabetical order, so LINQ, SIMD, Span.",
+      "reference": ">public const string LINQ = \"LINQ\"; [](start = 8, length = 34) Are there duplicated benchmarks here? #Closed"
+    },
+    {
+      "prediction": "I don't think we need to pass `options` to this method. We can just use `options[:pe_ver]` and `options[:pe_ver_win]` directly in the method.",
+      "reference": "Probably worth taking the opportunity to add yardocs to this method now. Esp. curious about what the options hash accepts."
+    },
+    {
+      "prediction": "I think this should be `if (!view && layer->current.keyboard_interactive) {`",
+      "reference": "This is Very Meh . We want to get rid of `role_data`, and it's an internal field."
+    },
+    {
+      "prediction": "I don't think we need to call `tryCall` here.",
+      "reference": "`await this.driver().loadExtension(path);` should work"
+    },
+    {
+      "prediction": "I'm not sure why this line was removed.",
+      "reference": "Is any of this needed anymore?"
+    },
+    {
+      "prediction": "I think this should be `#if defined(__x86_64__)` instead of `#if __SIZEOF_POINTER__ == 8`.",
+      "reference": "Can you add a mention here and in `ostree-repo.c` that the holes were found with `pahole`?"
+    },
+    {
+      "prediction": "I don't think we should deprecate `get_search_results` in favor of `search_results`. I think we should deprecate `search_results` in favor of `get_search_results`.",
+      "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": "I'm not sure this is the right thing to do here. I think we should be able to raise an assertion and have it fail the test, but I don't think we should be able to raise an exception and have it fail the test. What do you think?",
+      "reference": "I believe that you only need to rescue Beaker::DSL::Assertions, as they include Minitest::Assertions."
+    },
+    {
+      "prediction": "I think this should be `activity.ActivityTraceFlags &= ~ActivityTraceFlags.Recorded;`",
+      "reference": "Do we need `&=` or `=` is sufficient?"
+    },
+    {
+      "prediction": "I think this should be `$settings['config_sync_directory'] = '{{ joinPath $config.SitePath $config.SyncDir }}';` instead of `$settings['config_sync_directory'] = '{{ joinPath $config.SitePath $config.SyncDir }}';`",
+      "reference": "Let's keep both of these here. It should work on most any version of Drupal 8 then true?"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "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": "I don't think we need to change the copyright here.",
+      "reference": "Q: are we supposed to change this for files that already exist?"
+    },
+    {
+      "prediction": "I think this should be `selection_specs=[(DynamicMap, {})]`",
+      "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": "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.<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.<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.<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.<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",
+      "reference": "is this test for the case above? seems to cover a numeric rather than empty val?"
+    },
+    {
+      "prediction": "I think this should be `response.content = v`",
+      "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 = false;` instead of `bool dynamo_exited_synched_and_cleaned = false;`",
+      "reference": "_and_synched seems to be more consistent w/ exited_and_cleaned"
+    },
+    {
+      "prediction": "Why is this needed? I don't see any netcoreapp3.1 specific code in this file.",
+      "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 don't think you need to cast to `Long` here. `parameters.get(\"type\")` is already a `Long`.",
+      "reference": "should use Number instead of Long"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Had no idea that method existed!"
+    },
+    {
+      "prediction": "I don't understand why this is needed. `tabBar()` returns a `QTabBar` instance, which has a `width` attribute.",
+      "reference": "I wonder if we shouldn't just do `w.show()` here, which causes Qt to correctly calculate the sizes."
+    },
+    {
+      "prediction": "I think we can remove the `object_store.key_for_paths(paths)` and just use `object_store.key_for_path(paths[0])`",
+      "reference": "may be nice to have helper method to generate s3 paths rather than the minor code dup"
+    },
+    {
+      "prediction": "I don't think we need to fetch the private config here. We only need to fetch the private config if the user is not logged in.",
+      "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 `if c1.ClientAuth != tls.NoClientCert && c2.ClientAuth != tls.NoClientCert && (c1.ClientCAs != c2.ClientCAs || !StringSliceEqual(cfg1.ClientCerts, cfg2.ClientCerts)) {`",
+      "reference": "This line is getting a little long; let's move the conditions to at least two lines (maybe three)."
+    },
+    {
+      "prediction": "This is not needed.",
+      "reference": "Don't use relative paths. src directory is already on the include search path"
+    },
+    {
+      "prediction": "Style/HashSyntax: Use the new Ruby 1.9 hash syntax.",
+      "reference": "Line is too long. [86/80]"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "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": "This should be `if (dma_h == NULL) {`",
+      "reference": "Empty space at end."
+    },
+    {
+      "prediction": "I think this should be `@project` instead of `@bolt_project`",
+      "reference": "I think similar to line 176 here we'll want to call `detect_project_conflict` after overriding this."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "How about `const { promises: fs } = require('fs')` and then replacing `fs.promises.` with `fs.`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should close the body before we read from it.",
+      "reference": "This ignores errors closing the body. Not likely to happen, but no reason to take the risk either."
+    },
+    {
+      "prediction": "I don't think we need this change.",
+      "reference": "Why was this change needed?"
+    },
+    {
+      "prediction": "I think this should be `ARG_IMPORT_PEAK_BOUNDARIES_FILE` to be consistent with the other `ARG_IMPORT_*_FILE` arguments.",
+      "reference": "Needs a description added to CommandArgsUsage.resx"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do it, but it seems to work.",
+      "reference": "The line is a bit too long. That's why the test currently fails. Please break it into two lines."
+    },
+    {
+      "prediction": "I think we should use `anyMatch` instead of `count > 0`",
+      "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 an `int` instead of a `uint8`. I don't think we have any ADCs with more than 256 channels.",
+      "reference": "Due to this change, src/examples/adc needs to be modified"
+    },
+    {
+      "prediction": "I think this should be called `singlePointer` instead of `singlePtr` to be consistent with the rest of the codebase.",
+      "reference": "`s/singlePointer/singlePtr` or please rename the type to `singlePointer`."
+    },
+    {
+      "prediction": "I think this should be `1000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",
+      "reference": "Do we need it? If right now it's toml"
+    },
+    {
+      "prediction": "I think this should be `if (component && component._onResolve && !component._onResolveCalled)`",
+      "reference": "You are never calling oldUnmount, this could lead to a plugin chain failing."
+    },
+    {
+      "prediction": "`null` can be omitted.",
+      "reference": "What does setting partitionInfo to null mean exactly?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `CDSStream` class that has a `length` attribute and a `data` attribute. The `data` attribute would be a `dict` of `Column` objects, and the `length` attribute would be the length of the `Column` objects. The `CDSStream` class would have a `stream` method that takes a `dict` of `Column` objects and updates the `data` attribute. The `CDSStream` class would also have a `length` attribute that is the length of the `Column` objects. The `CDSStream` class would also have a `length` attribute that is the length of the `Column` objects. The `CDSStream` class would also have a `length` attribute that is the length of the `Column` objects. The `CDSStream` class would also have a `length` attribute that is the length of the `Column` objects. The `CDSStream` class would also have a `length` attribute that is the length of the `Column` objects. The `CDSStream` class would also have a `length` attribute that is",
+      "reference": "Not sure you need the initial ``not_updated`` as ``any([])`` evaluates to false."
+    },
+    {
+      "prediction": "I think this should be `published_at = updated_at` instead of `published_at = created_at`.",
+      "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": "Line is too long. [82/80]",
+      "reference": "This doesn't apply to all developers"
+    },
+    {
+      "prediction": "This should be `if ('doctrine_orm' == $config['db_driver'] && isset($bundles['FOSRestBundle']) && isset($bundles['NelmioApiDocBundle']) && isset($bundles['JMSSerializerBundle'])) {`",
+      "reference": "I think `serializer.xml` can be moved into condition below too"
+    },
+    {
+      "prediction": "I think it would be better to have this in a separate file, e.g. `README.md` in the `benchmarks` folder, so that it doesn't clutter the `Program.cs` file.",
+      "reference": "Probably put this in a simple README.md file?"
+    },
+    {
+      "prediction": "I think this should be an enum, not a flags enum.",
+      "reference": "Super Cool bitwise   Minor: Most likely you don't need `System.Collections.Generic`, `System.Linq` & `System.Threading.Tasks`"
+    },
+    {
+      "prediction": "I don't think we need to pass `appId` to this method. We can get it from `Thread.currentThread().getContextClassLoader().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent().getParent",
+      "reference": "If we just return the TCC, maybe we need to update the method name for it."
+    },
+    {
+      "prediction": "I don't think we should remove the method group name here.",
+      "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 this should be 0.5",
+      "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 `signingName = \"ecs\"` instead of `signingName = \"ecs\"`",
+      "reference": "Why this way?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be using `t.TempDir()` instead of `os.Getenv(\"TEST_TMPDIR\")` here.",
+      "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": "I'm not sure this is the right place to do this. I think it should be done at the point where the path is being constructed.",
+      "reference": "This modification is not correct. Suggections: 1. upgread guava to 30.0-jre will fix this issue"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it makes sense to return an error here.",
+      "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 think we can remove the `@Override` here.",
+      "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 don't think this is correct. `filterExpression()` is not the same as `filterExpressions`. `filterExpressions` is the list of filter expressions, while `filterExpression()` is a single expression that combines all the filter expressions.",
+      "reference": "We had a wrong predicate before: we compared a list to an expression."
+    },
+    {
+      "prediction": "I think this should be `group:65534:--x\\ngroup:65535:--x\\n`",
+      "reference": "Name copy-pasted from above. \"empty\"?"
+    },
+    {
+      "prediction": "Why do you delete this line?",
+      "reference": "May I ask why change this?"
+    },
+    {
+      "prediction": "I'm not sure this is correct. I think we need to add `len += value.Tags?.Count ?? 0;` and `len += value.Metrics?.Count ?? 0;`",
+      "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 don't think this is a good idea to have a config item for this. It's not something that should be configurable by the user. It's something that should be configurable by the dev.",
+      "reference": "The description and name is unclear. Typo in 'which'."
+    },
+    {
+      "prediction": "I don't think we should be adding this to the public API. It's not part of the W3C spec, and it's not part of the Selenium spec.",
+      "reference": "Files should have an extra line at the end of them."
+    },
+    {
+      "prediction": "I think this should be `void proc_set_num_simd_saved(int num)`.",
+      "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": "I think this should be `hash_table_size(d->files)` instead of `d->completed_files`.",
+      "reference": "This stat may fail, right? I think dag_file_exists does not actually check the file is there."
+    },
+    {
+      "prediction": "I think we can remove the `{}` here.",
+      "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 is the only place that needs to be updated to use the new `Edges()` method.",
+      "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 thing to do, but it seems to work for me.",
+      "reference": "Can you add an `else:` case to handle an invalid length? That will prevent weird errors."
+    },
+    {
+      "prediction": "I don't think we should remove the constructor that takes a `JspParser` as parameter.",
+      "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": "Why is this no longer needed?",
+      "reference": "Where did this functionality move to?"
+    },
+    {
+      "prediction": "Can we use `long minVal, long maxVal, int regionNum` instead of `long minVal, long maxVal, long regionNum`?",
+      "reference": "maybe should firstly check whether current tidb support `split table region`?"
+    },
+    {
+      "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's a better way to do this.",
+      "reference": "Please use the same (plural) name for the new component since it renders multiple notifications."
+    },
+    {
+      "prediction": "I think this should be an error, not an info.",
+      "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 throws Exception here.",
+      "reference": "why did you took `Exception` out? Is this method only throwing runtime exceptions?"
+    },
+    {
+      "prediction": "nitpick: I think it would be better to say \"generalizes to the case where labels are probabilistic (i.e. numbers between 0 and 1)\" instead of \"generalizes to the case where labels are probabilistic (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers between 0 and 1) (i.e. numbers",
+      "reference": "please revert all of these unrelated whitespace changes"
+    },
+    {
+      "prediction": "I think we should rename this to `EncryptionProtocol` or `EncryptionMethod` or something like that. `EncryptionType` is a bit too generic.",
+      "reference": "PGP_INLINE is clearer IMHO"
+    },
+    {
+      "prediction": "remove empty line",
+      "reference": "Please group imports"
+    },
+    {
+      "prediction": "I think we should keep this at 5. I don't think we need more than 5 threads.",
+      "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 think this should be `(dilation_h * (kernel_h - 1) + 1) / stride_h + 1`",
+      "reference": "Let's fix the extra spaces above."
+    },
+    {
+      "prediction": "This should be `List<? extends WebElement> findElements(By by);`",
+      "reference": "should a similar change be done for findElement?"
+    },
+    {
+      "prediction": "I think it would be nice to have a `NewWithKeys` function that takes a map of keys as an argument. This would allow us to create a keystore with a set of keys without having to call `AddKey` for each key.",
+      "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": "I think this should be `duckv1.CloudEventOverrides` instead of `duckv1.SourceSpec`.",
+      "reference": "I think this needs a json inline tag here"
+    },
+    {
+      "prediction": "I think this should be `dict:dict()` instead of `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 don't think this is correct. `once` is only used in `Start()`, and `Start()` is only called in `NewRoundRobinList()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which is only called in `NewRoundRobinPeerPicker()`, which",
+      "reference": "should we change/wrap the error?"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "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": "Line is too long. [85/80]",
+      "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 you need the `xrange` here, since you're not using `i` in the body of the loop.",
+      "reference": "Is this a bugfix?"
+    },
+    {
+      "prediction": "Why are we removing this?",
+      "reference": "Is this a relevant change?"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "@dansanduleac what do you think of this one? dots were aligned and no longer are"
+    },
+    {
+      "prediction": "Why is this being removed?",
+      "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 should be done in the `PluginBase` constructor, or at least in the `PluginBase` `init()` method.",
+      "reference": "The _OverviewPlugin_ is forcefully enabled without the option to disable it, using _PluginDescription.alwaysEnabled_. That should also work for this plugin?"
+    },
+    {
+      "prediction": "I don't think this should be in this class. It's not a search field, it's a filter field.",
+      "reference": "Can you facet on this new externalStatus Solr field?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 this is the right thing to do here. If the plan doesn't have a `return` statement, then it should be returning `nil` by default. If the plan does have a `return` statement, then it should be returning the value of the `return` statement. If the plan doesn't have a `return` statement, then it should be returning `nil` by default. If the plan does have a `return` statement, then it should be returning the value of the `return` statement. If the plan doesn't have a `return` statement, then it should be returning `nil` by default. If the plan does have a `return` statement, then it should be returning the value of the `return` statement. If the plan doesn't have a `return` statement, then it should be returning `nil` by default. If the plan does have a `return` statement, then it should be returning the value of the `return` statement. If the plan doesn't have a `return` statement, then it should be returning `nil` by default. If the plan does have a `return` statement, then it should be returning the value of the `return` statement. If 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 don't think this is necessary.",
+      "reference": "Try using using `.setDismissOnTouch(true)` instead"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to update the span. I think it should be done in the `finally` block, after the `scope?.Span.SetException(ex);`",
+      "reference": "Zach, is `UpdateSpan` of concern too? I see that it is called from `ExecuteAsyncInternal`."
+    },
+    {
+      "prediction": "nit: I think we can remove the `final` here.",
+      "reference": "Nit: its scope can be reduced to package default"
+    },
+    {
+      "prediction": "Should this be `ApplicationScheduler`?",
+      "reference": "Why is this get;set;?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. It should be done in `StatefulReader::matched_writer_add` and `StatefulReader::matched_writer_remove` after the call to `getRTPSParticipant()->createSenderResources(wdata.endpoint.remoteLocatorList, false);` and `getRTPSParticipant()->deleteSenderResources(wdata.endpoint.remoteLocatorList, false);` respectively.",
+      "reference": "Should we also create sender resources for unicast locators? Should we make a similar change on StatelessReader?"
+    },
+    {
+      "prediction": "I don't think we should change the error message here. It's a bit more specific to IIS, and I don't think we want to make it more generic than it already is.",
+      "reference": "Can we use `dotnet` instead of `net`?"
+    },
+    {
+      "prediction": "I don't think we need to add this method to the `ITracer` interface. It's not part of the public API.",
+      "reference": "Should this be `ForceFlushAsync`? Given we have the `async` suffix elsewhere."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Is this change intentional?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `nativeOffset != -1` check and just call `SetProperty` unconditionally.",
+      "reference": "This could be an integer property if we supported explicitly setting non-string-valued properties."
+    },
+    {
+      "prediction": "I don't think we should change the visibility of this constant.",
+      "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": "Is this a bug fix? If so, can you add a test for it?",
+      "reference": "why add `return` here?"
+    },
+    {
+      "prediction": "I think it would be better to use `Bytes.FromHexString` instead of `0xf8c61953f3b94a91aefe611e61dd74fe26aa5c969d9f29b7e063e6169171a772`",
+      "reference": "put this hash in variable"
+    },
+    {
+      "prediction": "nit: remove empty line",
+      "reference": "Why not use `kbfsblock.ZeroRefNonce`?"
+    },
+    {
+      "prediction": "I don't think we need to change the name of the namedtuple.",
+      "reference": "what mypy warning does this address? I'm confused by this change."
+    },
+    {
+      "prediction": "This should be `if (attacked == this) { return; }` instead of `if (!attacked || attacked == this) { return; }`",
+      "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'm not sure if this is the best way to do this, but it works for now.",
+      "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 don't think this is the right place for this script.",
+      "reference": "Leave this out and move the script to the scripts folder."
+    },
+    {
+      "prediction": "I think this should be `nameOrHandle` instead of `handle`.",
+      "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": "Should this be `IDuplexPipe Transport { get; }`?",
+      "reference": "TransferFormat.Text in general purpose connection interface? :/"
+    },
+    {
+      "prediction": "I don't think we need to use `self` here. `attr_accessor :executable, :missing_text` should be enough.",
+      "reference": "Is this `@missing_text` ?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it should be done in the `processRouteInfo` method of the `client` struct.",
+      "reference": "This work for both sides?"
+    },
+    {
+      "prediction": "I don't think we need to import `map` from `lodash` anymore, since we're using `lodash-es` now.",
+      "reference": "`lodash` shouldn't be grouped under WordPress dependencies"
+    },
+    {
+      "prediction": "This test is not testing anything. It's just calling the function and not asserting anything about the result.",
+      "reference": "Include the standard parameterized error tests as well."
+    },
+    {
+      "prediction": "This should be `middleware.RepoAssignment()` instead of `middleware.RepoRef()`.",
+      "reference": "All lowercase in route rule."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 think this should be `rangeToDisplay = rangeToDisplay > 24 ? 3 : rangeToDisplay;`",
+      "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 think this should be `SetHead(context.Context, types.TipSetKey) error`",
+      "reference": "We still need to figure out a way to get rid of this write method. Is this still necessary?"
+    },
+    {
+      "prediction": "I think this should be `self.assertItemsEqual(classifier.getPartitionIdPerPattern(), [433, 213, np.inf, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433, 413, 433,",
+      "reference": "`assertItemsEqual` compares the sequences after sorting them, is that intended?"
+    },
+    {
+      "prediction": "`File.expand_path?(path)` should be `File.expand_path(path)`",
+      "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 `serviceDefinition.Host?.EqualsIgnoreCase(AzureHostSignature) == true || serviceDefinition.Extensions.ContainsKey(\"x-ms-parameterized-host\")`",
+      "reference": "what's the goal of this?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to put this. I think it should be in the `node` package.",
+      "reference": "why is this here? We already have a message pool in the node."
+    },
+    {
+      "prediction": "why is this changed?",
+      "reference": "Another mess after all renames and enum conversions. Doesn't break anything though."
+    },
+    {
+      "prediction": "I think this should be `storage_strict_json = True` instead of `storage_strict_json: True`",
+      "reference": "I don't think it's necessary to change this for the history plugin tests?"
+    },
+    {
+      "prediction": "I don't think this change is necessary. The `fill_local_ip` function is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCPv4Transport::init`, which is only called from `TCP",
+      "reference": "In general, I suggest using the `kind()` getter inhereted from `TransportInterface`"
+    },
+    {
+      "prediction": "I don't think we should remove these attributes. They are there for a reason. The `ThreadAgnostic` and `ThreadSafe` attributes are used by the `AsyncWrapperLayoutRenderer` to determine if the layout is safe to use with the `AsyncWrapperLayoutRenderer`. The `AppDomainFixedOutput` attribute is used by the `AsyncWrapperLayoutRenderer` to determine if the layout is safe to use with the `AsyncWrapperLayoutRenderer` when the `AsyncWrapperLayoutRenderer` is used with the `AsyncWrapperTarget`.",
+      "reference": "Is the removal of `[AppDomainFixedOutput]` on purpose? (looks like a copy-paste...euh..remove error)"
+    },
+    {
+      "prediction": "`AWS IAM` -> `AWS Identity and Access Management`",
+      "reference": "AWS Identity and Access Management (IAM)"
+    },
+    {
+      "prediction": "I think this should be `PublicKey` instead of `public_key`",
+      "reference": "gofmt everything pls"
+    },
+    {
+      "prediction": "`+kubebuilder:printcolumn:name=\"duration\",type=string,JSONPath=`.spec.duration`,description=\"the duration of the chaos action\"`",
+      "reference": "ditto, We do not need add this feature for PodXXXChaos"
+    },
+    {
+      "prediction": "I think this should be `lowerCamel(lowerCamelApiTypeName)` instead of `lowerCamel(lowerCamel(lowerCamelApiTypeName))`.",
+      "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 thing to do here. I think it should be `from.CanCombineWith(to) && !to.CanCombineWith(from)`, but I'm not 100% sure.",
+      "reference": "Can you explain this change to me?"
+    },
+    {
+      "prediction": "I think this should be `../artifacts/storageclass-1r.yaml` instead of `../artifacts/single-replica.yaml`",
+      "reference": "let us rename the artifact to `storageclass-single-replica.yaml`"
+    },
+    {
+      "prediction": "Should this be `Microsoft.AspNetCore.Server.Kestrel.Transport.Libuv`?",
+      "reference": "Nit: I should have done this before, but rename the file to LibuvFunctions.cs. Make sure no other files are misnamed."
+    },
+    {
+      "prediction": "I don't think we need to reserve this for LogCodec. We can just use it in JavaBinCodec.",
+      "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 this class is used anywhere.",
+      "reference": "If only Gatling uses it, can we skip introducing new entity for now? Looks like overkill..."
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "rename -> info ?"
+    },
+    {
+      "prediction": "I think this should be `upstream_status_when_all` instead of `upstream-status-when-all`",
+      "reference": "Remove the `config_path=dict(...)`, that is automatic!"
+    },
+    {
+      "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's a better way to get all the iframes on the page. I tried `document.querySelectorAll('iframe, frame')` but that doesn't seem to work.",
+      "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 it's better to use `model.getConfig().put(key, value);` instead of `model.setConfig(config);`",
+      "reference": "As this method is just set the internal cofigModel, it's better to use setConfigs instead of addConfigs."
+    },
+    {
+      "prediction": "I don't think this is a good idea. We should be able to load an image from STDIN in less than 10 minutes.",
+      "reference": "Is this line needed?"
+    },
+    {
+      "prediction": "nit: I think we can remove this import.",
+      "reference": "nit: import formatter"
+    },
+    {
+      "prediction": "I think this should be `isNonNullableReturnType = isNonNullableReturnType && !(_operation.Responses[key].Extensions?.Get<bool>(\"x-nullable\") ?? true);`",
+      "reference": "can you use LINQ's `All` instead? Should be a side-effects free one liner with that."
+    },
+    {
+      "prediction": "I don't think we need to do this. spawn.h should be available on 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": "I don't think we need to implement `IsPaused` for `fakeTwoPhaseChaos`.",
+      "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 named `hasManageHighPriorityFlowsPermission`",
+      "reference": "This method is validating permission for hp flow. Can you please rename this method to convey the same?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "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`",
+      "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 don't think this is correct. If the import path is `foo/bar` and the import name is `foo`, then the import statement should be `import foo \"foo/bar\"`, not just `import \"foo/bar\"`.",
+      "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 `if full_parent_name.startswith('composite_root/root/'):`",
+      "reference": "Just to clarify, should this be 'composite_root/root/' or 'composite_root/'?"
+    },
+    {
+      "prediction": "I think this should be `status.SubscriptionID = \"\"` instead of `status.SubscriptionID = nil`",
+      "reference": "Any reason not to move it to after line 281? These are all pubsub related fields IIUC."
+    },
+    {
+      "prediction": "I don't think it's necessary to mention that zero is not considered negative.",
+      "reference": "Checks if value is a negative `Number` primitive or object. Zero is considered neither positive nor negative."
+    },
+    {
+      "prediction": "Why not use `r.URL.Scheme`?",
+      "reference": "Just a nit, but remove the empty line here. :)"
+    },
+    {
+      "prediction": "I think this should be `if v2 != nil {` instead of `if v2, ok := p.ext.Rules.Apply(trimmed) {`",
+      "reference": "A `GoPackageInfo` details message should be added to the `CompilationUnit_FileInput` to ensure the import path remains the same."
+    },
+    {
+      "prediction": "This import is no longer needed.",
+      "reference": "Please don't create new fields with the `m` prefix. I'd rather mix than introduce more of these."
+    },
+    {
+      "prediction": "I think this should be `cpuCircuitBreakerThresholdPct` instead of `cpuCircuitBreakerThreshold`.",
+      "reference": "Can we please simplify these names? they are awfully verbose and repeating the parts that are already unique and obvious."
+    },
+    {
+      "prediction": "I think this should be `alreadyVisible || singleBccRecipientFromAccount`",
+      "reference": "Flipped the logic here, was this actually wrong before?"
+    },
+    {
+      "prediction": "nit: I think this should be `int res = 0;` instead of `res = 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` or `./gradlew :core: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 we should avoid importing `github.com/algorand/go-algorand/data/basics` in this package.",
+      "reference": "Group w/ algorand imports"
+    },
+    {
+      "prediction": "Please remove unused imports.",
+      "reference": "I don't think this import is being used."
+    },
+    {
+      "prediction": "Why do we need to return a slice here? I don't think it's possible to have more than one root for a pattern.",
+      "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": "`$config['cdn']['fallback']['primary'] ?? $config['cdn']['fallback']['master']` should be `$config['cdn']['fallback']['primary'] ?? $config['cdn']['fallback']['master'] ?? null`",
+      "reference": "You did not check this code"
+    },
+    {
+      "prediction": "I think this should be `len(types) > 0 && len(data) == 0`",
+      "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": "Please add a comma at the end of the line.",
+      "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": "Why is this test removed?",
+      "reference": "Is it intentional that we removed this annotation?"
+    },
+    {
+      "prediction": "I think this should be `ServiceDefinitionDocumentState.Both`",
+      "reference": "should be general?"
+    },
+    {
+      "prediction": "Why do we need to check for `LowestInsertedHeader?.Number ?? 0` instead of just `LowestInsertedHeader.Number`?",
+      "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 think we can remove this class and just use `FakeServer` directly.",
+      "reference": "Is a FakeApp class needed? Can test use a Mockito mock instance instead?"
+    },
+    {
+      "prediction": "nit: I'd prefer to see this as `is_constant_value_data` instead of `is_constant` to avoid confusion with `is_constant` in the `Variable` class.",
+      "reference": "We should make this an enum."
+    },
+    {
+      "prediction": "Why is this no longer needed?",
+      "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['yaxis'] = plot.yaxis[0]`",
+      "reference": "Minor unrelated fix that snuck in."
+    },
+    {
+      "prediction": "This import is no longer needed.",
+      "reference": "Took the opportunity to fix the import ordering."
+    },
+    {
+      "prediction": "Shouldn't this be `elif isinstance(wavelength, (int, str)):`?",
+      "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 think this should be 15ms.",
+      "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": "nit: I think this should be `github.com/GoogleCloudPlatform/go-gcp-dev/pkg/gcp`",
+      "reference": "nit: add a blank line under this."
+    },
+    {
+      "prediction": "I think this should be `--chain-id` instead of `--chainID` to be consistent with the rest of the flags.",
+      "reference": "no need to add this flag query the endpoint set-up to determine the chainID"
+    },
+    {
+      "prediction": "I think this should be `if (node.querySelector('input[type=\"submit\"], img[type=\"submit\"], button[type=\"submit\"]')) { return true; }`",
+      "reference": "Here is a few scenario - What if there is a submit button with in a form, but is always disabled?"
+    },
+    {
+      "prediction": "I think this should be `Weh TYPE_W, OPSZ_half_16_vex32_evex64`",
+      "reference": "wrong code: should be `Wh_e`"
+    },
+    {
+      "prediction": "I think this should be `static vString *label = vStringNew ();` instead of `static vString *label = NULL;` and `vStringClear (label);` instead of `vStringNewOrClear (label);`.",
+      "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 don't think we need this constant. We can just use `BuildIDOSEnv` directly.",
+      "reference": "BuildIDOSEnvVarName or similar, otherwise it sounds it's actually holding actual value of env var"
+    },
+    {
+      "prediction": "I think this should be `current_app.logger.info('Spark consumer has started!')` instead of `current_app.logger.info('Spark consumer has started!')`",
+      "reference": "I like this standard \"container has started\" message. Should we have a \"container exited because FOO\" message as well?"
+    },
+    {
+      "prediction": "I don't think this import is 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 don't think we need to expose `instanceRegister` in the global scope.",
+      "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 `Error performing %s operation: the disk cache has been closed.`",
+      "reference": "No trailing punctuation in the error message."
+    },
+    {
+      "prediction": "This should be moved to the finally block.",
+      "reference": "tested? seems like it could break"
+    },
+    {
+      "prediction": "I think this method should be `private static` since it's not used outside of this class.",
+      "reference": "This have to be public because the renamed o.a.l.backward_index.TestBackwardsCompatibility refers this."
+    },
+    {
+      "prediction": "I think this should be `consensus[consensusTestUnupgradedProtocol] = consensus[consensusTestFastUpgrade(protocol.ConsensusCurrentVersion)]`",
+      "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 if this is the right place to do this, but I think we should also check for `AWS_WEB_IDENTITY_TOKEN_FILE` and `AWS_ROLE_ARN` in the shared config.",
+      "reference": "Should this also validate that the RoleArn env var is provided, or just let the creds fail?"
+    },
+    {
+      "prediction": "I don't think we need to import `opentracing` here.",
+      "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 it would be good to add a test for the case where the new name is not hashable.",
+      "reference": "Shall we also add `ks.Series([1, 2, 3], name=[\"0\", \"1\"])`?"
+    },
+    {
+      "prediction": "I think this should be `perm_addNodesToWhitelist` instead of `perm_addNodesToAllowlist`",
+      "reference": "Can this be updated?"
+    },
+    {
+      "prediction": "I don't think this is a good idea. If you want to add a new format, you have to add it to the `formats` array in the service definition. If you want to change a format, you have to change it in the `formats` array in the service definition. If you want to remove a format, you have to remove it from the `formats` array in the service definition. If you want to change the quality of a format, you have to change it in the `formats` array in the service definition. If you want to change the format of a format, you have to change it in the `formats` array in the service definition. If you want to change the height of a format, you have to change it in the `formats` array in the service definition. If you want to change the constraint of a format, you have to change it in the `formats` array in the service definition. If you want to change the resizer of a format, you have to change it in the `formats` array in the service definition. If you want to change the quality of a format, you have to change it in the `formats` array in the service definition. If you want to change the format of a format",
+      "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 don't think we need to lock the container here. The container is already locked in `connectInterfaceToOVSAsync`.",
+      "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": "I think it would be better to use `implode(', ', $knownPositionsNames)` instead of `implode('\", \"', $knownPositionsNames)`.",
+      "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`",
+      "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": "I think this should be `SupportYAMLTags: false` by default. I don't think we want to support YAML tags by default.",
+      "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'm not sure this is the right thing to do. I think it should be `executorService.shutdown();` and then `executorService.awaitTermination(10, TimeUnit.SECONDS);` and then `executorService.shutdownNow();`",
+      "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": "nit: I think it would be better to pass `HiveMetastoreTest.class` instead of `HiveMetastoreTest.class.getClassLoader()`.",
+      "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": "`protocol.StateReader` -> `StateReader`",
+      "reference": "I think it is weird to store sr in protocol struct"
+    },
+    {
+      "prediction": "I think this should be `if (CmpCommon::getDefault(TRAF_RELOAD_NATABLE_CACHE) == DF_ON)`",
+      "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 don't think this is the right place to put this.",
+      "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 or self.rasterized or self.datashaded:`",
+      "reference": "SyntaxError I think"
+    },
+    {
+      "prediction": "Why is this disabled?",
+      "reference": "I've filed an internal spec issue to add these missing VUs."
+    },
+    {
+      "prediction": "This import is not used.",
+      "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's better to keep the imports in alphabetical order.",
+      "reference": "In iceberg, we usually don't use `*` to import package, it's more clear to import the specify package one by one."
+    },
+    {
+      "prediction": "I don't think this import is 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'm not sure this is the right thing to do here. I think it should be `timeoutControl.StartTimingWrite(minDataRate, maxBufferedBytes);` and `timeoutControl.BytesWritten(minDataRate, maxBufferedBytes);` should be removed.",
+      "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'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Good call. Makes we wonder if we should do this with the ARIA props."
+    },
+    {
+      "prediction": "I don't think we should be importing `java.sql.*`",
+      "reference": "Please do not use wildcard imports."
+    },
+    {
+      "prediction": "I think this should be `L_INFO` or `L_DEBUG`. `L_DEBUG` is too spammy.",
+      "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 don't understand why this was removed.",
+      "reference": "Should there be an equivalent check or is it ok to remove this comparison?"
+    },
+    {
+      "prediction": "I think we should revert this change.",
+      "reference": "sorry, why this change?"
+    },
+    {
+      "prediction": "`NamedList<?>` should be `NamedList<Object>` to be consistent with the rest of the codebase.",
+      "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": "Please remove unused use statements.",
+      "reference": "Be careful, the base model is imported here !"
+    },
+    {
+      "prediction": "I think this should be `[\"eth\", \"tg\", \"debug\"]`",
+      "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": "Align the parameters of a method call if they span more than one line.<br>Tab detected.",
+      "reference": "What is the plan when we deploy the Crawler VM project and eliminate the Clump model?"
+    },
+    {
+      "prediction": "I don't think we need to remove these fields. They are used to determine the role of the cluster in the cluster set.",
+      "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.Compiler == \"gc\" || runtime.Compiler == \"gccgo\"`",
+      "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": "Shouldn't this be `fields + ' ' + this.schema.options.discriminatorKey`?",
+      "reference": "fields may be an object."
+    },
+    {
+      "prediction": "nit: I think we can remove the empty line here.",
+      "reference": "i'm not confident these are the right defaults... anybody have input?"
+    },
+    {
+      "prediction": "I think this should be `DiskGroups` to be consistent with the rest of the codebase.",
+      "reference": "DiskGroupList can be better name"
+    },
+    {
+      "prediction": "nit: I think we can remove the `gatewayCfg` parameter from this function and pass the `metricsHandler` directly to the `ServeHTTP` method.",
+      "reference": "nit: lets leave gateway options at the end of the func signature."
+    },
+    {
+      "prediction": "Is there a reason to use `apply()` instead of `commit()`?",
+      "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 think this should be `queryResult.ErrorMessage != nil && *queryResult.ErrorMessage != \"\"`",
+      "reference": "Took me almost 4 hours to find this."
+    },
+    {
+      "prediction": "Please remove the extra blank line.",
+      "reference": "There should be only a single blank line between imports and `__all__`."
+    },
+    {
+      "prediction": "I'm not sure why this is changing.",
+      "reference": "Doc size increased by 1 since _root_ is also returned by queries."
+    },
+    {
+      "prediction": "I don't understand why this is necessary. If `clmin == clmax` then `lidx == uidx` and `cmap[lidx:uidx]` will return an empty array.",
+      "reference": "Now the return value has changed (or at least now includes `clip`) it might be worth updating the docstring..."
+    },
+    {
+      "prediction": "I think this should be `sf.NewWorkingSet(nil)` instead of `sf.NewWorkingSet(nil, nil)`.",
+      "reference": "shadow: declaration of \"err\" shadows declaration at line 410 (from `govet`)"
+    },
+    {
+      "prediction": "I think this should be `data-' . $attr . '=\"' . htmlspecialchars($value, ENT_QUOTES, 'UTF-8') . '\"`",
+      "reference": "Should we be escaping $attr and/or $value here? Seems like htmlspecialchars() might be in order to be on the safe side."
+    },
+    {
+      "prediction": "This should be `args << \" #{quiet_o}\" if quiet_o`",
+      "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 don't think we need this anymore.",
+      "reference": "These variables are already created in constants. Please reuse those."
+    },
+    {
+      "prediction": "Shouldn't this be `query_args.append('versionId=%s' % quote(self.version_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 think this should be `return nil, err`",
+      "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 this should be `Bolt::Error`",
+      "reference": "Looks like this isn't used in this file?"
+    },
+    {
+      "prediction": "`hidden_name` should be `self.name` if `self.name != 'index'` and `level_0` if `self.name == 'index'`.",
+      "reference": "why hidden name? How about `index_name`?"
+    },
+    {
+      "prediction": "I think this should be `hiveScanMode = CmpCommon::getDefaultInt(HIVE_SCAN_SPECIAL_MODE);`",
+      "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": "`NewAnyTag(\"shard-context\", shard)` -> `NewAnyTag(\"shard\", shard)`",
+      "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 we need this file.",
+      "reference": "Why do you have to disable all of these checks?"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to call this.",
+      "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 what you're trying to test here. If you want to test that the initial spill of aflags doesn't use a slot, then you need to make sure that the initial spill of aflags does use a slot, and then make sure that the second spill of aflags doesn't use the same slot. I don't see that happening here.",
+      "reference": "In this test, we want xax to be dead, so that aflags are not spilled to a slot."
+    },
+    {
+      "prediction": "I don't think we should enable this by default. I think it should be disabled 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": "Line is too long. [87/80]",
+      "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": "`address` is already a string, no need to call `toString()` on it.",
+      "reference": "you can use `toString(true)` that will work for BTC and remove the prefix for BCH."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it should be done here. The reason is that we don't want to leave caret mode when switching tabs, but we do want to leave caret mode when switching tabs if `persist_mode_on_change` is set to `True`.",
+      "reference": "This line is too long now - please break it after the comma."
+    },
+    {
+      "prediction": "I think this should be `ResponseHeaders() (Headers, error)`.",
+      "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": "I think it would be better to throw an exception if the cart is empty.",
+      "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 this is a bug. `daysFromTimestamp` should be used for `TIMESTAMP` and `daysFromDate` should be used for `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": "nit: I think it's better to rename this test to `TestSendX509SVIDResponse`",
+      "reference": "Tests for JWT handler functionality?"
+    },
+    {
+      "prediction": "I'm not sure why this was removed. It seems like it should still be there.",
+      "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 don't think we need to test this in a separate test class. We can just add it to the `DefaultBucketViewTest` class.",
+      "reference": "Should we move that information in the capability itself?"
+    },
+    {
+      "prediction": "Should this be `_auto_indexable_1d = True`?",
+      "reference": "Is this needed?"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "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": "Why do we need this? I don't see it being used anywhere.",
+      "reference": "BlockTree alredy has a polluted API, this one seems very internal"
+    },
+    {
+      "prediction": "I think this should be `bolt_inventory['features'] = ['puppet-agent', 'bolt']`",
+      "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 don't think this is the right thing to do here. I think we should just return `null` and let the caller decide what to do in this case.",
+      "reference": "This bit I wasn't really sure about."
+    },
+    {
+      "prediction": "I think this should be `parseChecks(checks, locale.checks);`",
+      "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` instead of `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'm not sure about this change. I think it's better to have the window in the presenter, not in the setup. The presenter should be able to get the window from the setup, but not the other way around.",
+      "reference": "`_applicationDelegate` and `_window` are already set in the call for `: this (applicationDelegate, window)`, no need to assigning them again here"
+    },
+    {
+      "prediction": "I think this should be `props.selector(context)` instead of `props.selector(props.selector(context))`.",
+      "reference": "We might not have a selector prop in the consumer"
+    },
+    {
+      "prediction": "`_blockTree.Head?.Header == _blockTree.Genesis` should be `(_blockTree.Head?.IsGenesis ?? false)`",
+      "reference": "IMO == true is more readable than ?? false"
+    },
+    {
+      "prediction": "I'm not sure why this was changed, but I think it should be `bson/lib/bson/parser/utils` instead of `bson/lib/parser/utils`.",
+      "reference": "feel free when editing sections like this to introduce modern features like object destructuring."
+    },
+    {
+      "prediction": "Please remove this line, it's not needed.",
+      "reference": "I think the correct pragma is `# -*- coding: utf-8 -*-"
+    },
+    {
+      "prediction": "nitpick: I'd remove the `# noqa: F401`",
+      "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 a `sync.RWMutex` instead of a `sync.Mutex`.",
+      "reference": "I think we need to clear this out on logout, and update it on login, right?"
+    },
+    {
+      "prediction": "I think this should be `glnx_throw_prefix (error, \"Invalid NULL 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 don't think we need to change the name of this constant.",
+      "reference": "Is this header used by other outside service?"
+    },
+    {
+      "prediction": "I think this test should be moved to `test_provisioning.py` and renamed to something like `test_no_work_prov_unpacker`.",
+      "reference": "that's fine by me to have simple test for provisioning check"
+    },
+    {
+      "prediction": "I don't think we need to pass the cron expression to the BasicTimeChecker. The cron expression is only used by the CronTimeChecker.",
+      "reference": "These two methods seem to be identical except the names. Any idea why two methods are needed?"
+    },
+    {
+      "prediction": "I think this should be `config.INFLUX_DB` instead of `config.INFLUX_TEST_DB`.",
+      "reference": "I don't see much difference between this and `create_db` module. How about having a function with a `database` argument?"
+    },
+    {
+      "prediction": "I don't think we need to pass the service name to `getDataErrorComponent` and `getNoDataComponent` anymore.",
+      "reference": "See above, no need to pass the module name."
+    },
+    {
+      "prediction": "I think this should be \"validation sets\" instead of \"validation set\" since there can be multiple validation sets.",
+      "reference": "LightGBM supports multiple validation sets, so please leave `sets`."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "you don't need to create a enum for this, java allows you to switch on a string ;) (since java 7?)"
+    },
+    {
+      "prediction": "I think this test should be removed.",
+      "reference": "What is this obscure syntax anyway? o.O"
+    },
+    {
+      "prediction": "I think this should be `self._algorithm` instead of `self._sfdr`.",
+      "reference": "I think we should have one for the CLA Classifier, and KNNClassifier as well. And how about RecordSensor?"
+    },
+    {
+      "prediction": "Please add the license header to this file.",
+      "reference": "Missing new line."
+    },
+    {
+      "prediction": "I think this should be `model.schema.options.strictPopulate` instead of `model.schema.base.options.strictPopulate`.",
+      "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 it's better to use `Builders.Build.A.Block.TestObject` instead of `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": "I think this should be `gsub(/(?<!\\\\)\\\\/, '\\\\e')` so that backslashes that are not preceded by another backslash are escaped.",
+      "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 think this should be `Selector: \"default == ''\"` instead of `Tag: \"default\"`.",
+      "reference": "`has(default)` i think is the preferred way of doing this"
+    },
+    {
+      "prediction": "I think we can remove the `DataFile` type here as well.",
+      "reference": "Does this mean we can only do rewrites if there are no `DeleteFile`'s?"
+    },
+    {
+      "prediction": "nit: I think it would be more readable to split this into two lines.",
+      "reference": "Actually, these are fine; we're using 110 as the max line length. (it's also set in the .pylintrc)"
+    },
+    {
+      "prediction": "Please remove this extra blank 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 don't think we need to close the stream here.",
+      "reference": "guessing this method could/should close it but I didn't see where it was getting called."
+    },
+    {
+      "prediction": "I think this should be `EXECUTION_SOURCE_PARAM` instead of `EXECUTION_SOURCE`.",
+      "reference": "Could you please use `camelCase` in the property name to follow the naming convention used in the rest of the properties?"
+    },
+    {
+      "prediction": "I don't think we need the space at the end of this string.",
+      "reference": "Can we revert this change? looks like an extra space was added at the end"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to get the namespace from. I think it should be in the `RubyPackageMetadata` class.",
+      "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 think this should be `violations_{}'.format(model_name)`",
+      "reference": "Remove the pylint disable and remove the .format() from the string. The tablename is just 'violations'"
+    },
+    {
+      "prediction": "Style/RedundantSelf: Redundant self detected.<br>Metrics/LineLength: Line is too long. [123/120]",
+      "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 it should be done in the `Restart` method.",
+      "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 don't think this is correct. If you have 2 cpus per model, and 2 models, you should have 4 parallel readers.",
+      "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 this should be `UNEXPECTED_ALERT_BEHAVIOUR` instead of `UNEXPECTED_ALERT_BEHAVIOUR`.",
+      "reference": "you can change the string value, but not the code constant"
+    },
+    {
+      "prediction": "I don't think this is a valid test case. `split1checkupstagehandshakeupstreamerranterredcaperred2y9e2w` is a valid bech32 string.",
+      "reference": "File is not `goimports`-ed"
+    },
+    {
+      "prediction": "I think we can just remove this line for now.",
+      "reference": "Is this Todo necessary?"
+    },
+    {
+      "prediction": "I'm not sure about this change. If `value` is not a dictionary, then `value.get('message', None)` will raise a `TypeError`.",
+      "reference": "I would recommend to leave out `None` because None is already the default."
+    },
+    {
+      "prediction": "I'm not sure this is the right place for these flags. I think they should be defined in the `cmd` package.",
+      "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 if this is the right thing to do, but I think we need to make sure that if the send task throws an exception, it gets propagated to the caller of `StartAsync` so that it can be logged.",
+      "reference": "Nit: rename to sendError."
+    },
+    {
+      "prediction": "I think this should be done in `setup.py` instead of here.",
+      "reference": "`pylint` wants to know why this `import` isn't at the top of the file; and it's kinda right"
+    },
+    {
+      "prediction": "I think this should be `trackName` instead of `url`",
+      "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": "Why is this change necessary?",
+      "reference": "what was the issue here? why did you have to directly set?"
+    },
+    {
+      "prediction": "I think this should be `if len(country) == 0 { country = \"Unknown\" }` instead of `if len(country) == 0 { country = \"Unknown\" }`",
+      "reference": "I'm confusied - Isn't `countryString` left empty if `len(country)` != 0?"
+    },
+    {
+      "prediction": "This should be `QOverload<int>::of(&QSignalMapper::mapped).of(this, &DesktopSwitch::setDesktop);`",
+      "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 this is the right place to call this. I think it should be in the `defer func() { ... }()` that calls `defer stopGPGAgent(installChroot)`.",
+      "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": "This import is not used.",
+      "reference": "this file is model generated by protobuf, just for compatible test it's better not change it."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but I think it's better than having a bunch of `if` statements in the `track` method.",
+      "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": "Style/TrivialAccessInstanceVariable: Use an instance variable directly instead of using attr_reader.",
+      "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": "I don't think this change is correct. `data_arr_device` is a `sycl::buffer`, not a `la::matrix`, so it doesn't have a `get_data` method.",
+      "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 `true` by default.",
+      "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'm not sure this is the right thing to do, but I don't have a better idea.",
+      "reference": "Arguably this is actually still valuable for perf, maybe it should just be a warning?"
+    },
+    {
+      "prediction": "I don't think we need to do this here. The `TopologyDriver.AddPeer` call will add the peer to the addressbook, so we don't need to do it twice.",
+      "reference": "didn't way say that the topology driver will actually insert the entries to the address book?"
+    },
+    {
+      "prediction": "I'm not sure about this change. I think it should be `oldParentVNode != null && oldParentVNode._children || 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 this should be `--filename-suffix SUFFIX` and `--filename-ext EXTENSION`",
+      "reference": "also in the `.md` please."
+    },
+    {
+      "prediction": "Line is too long. [81/80]<br>Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "RHEL is an acronym and is usually all caps. Why are you adding a redhat regex matcher?"
+    },
+    {
+      "prediction": "This class is no longer needed.",
+      "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'm not sure this is the right place for these constants. I think they should be in the `manifest` package.",
+      "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 think this should be `self._pin = False`",
+      "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 think we can remove this constant. It's only used in one place.",
+      "reference": "Nit: `Headless` is one word, not two, and so doesn't need camelcasing in this way."
+    },
+    {
+      "prediction": "I think this should be `const runCommand = (data, keepalive, callback) => {` instead of `const runCommand = (data, keepalive, callback) => {`",
+      "reference": "Why not import audit as well?"
+    },
+    {
+      "prediction": "`$categoryData->parent = $this->getReference(self::CATEGORY_ELECTRONICS);`",
+      "reference": "OK, i think there will be more of those..."
+    },
+    {
+      "prediction": "This should be `\u041e\u0432\u043e\u0438 \u0430\u043a\u0440\u0435\u0434\u0438\u0442\u0438\u0432\u0438 \u043d\u0435 \u0441\u0435 \u0441\u043e\u0432\u043f\u0430\u0453\u0430\u0430\u0442 \u0441\u043e \u043d\u0430\u0448\u0438\u0442\u0435 \u0437\u0430\u043f\u0438\u0441\u0438.`",
+      "reference": "Can you add : 'password' => 'The provided password is incorrect.', ?"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Extraneous blank line."
+    },
+    {
+      "prediction": "`virtualTree` is not used in this function, so it can be removed.",
+      "reference": "nitpick: use `const virtualTree`, I know it was `let` earlier."
+    },
+    {
+      "prediction": "I'm not sure this is true. I don't think we have any code that listens to `mdFlushListener` and caches the new version.",
+      "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); }` instead of `if (!FileHelper.IsRootedPath(source)) { source = FileHelper.CombinePath(source); }`",
+      "reference": "We need tests for this"
+    },
+    {
+      "prediction": "I think we can remove the `WithGRPCDialOption(grpc.WithBlock())` now that we're using `WithInsecure()`.",
+      "reference": "NewExporter takes in two arguments: a configuration for the metrics connection and one for the traces"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think we should be able to get the current time from the player and use that to calculate the percentage to seek to.",
+      "reference": "Why not use currentPlayer instead of playbackManager.getCurrentPlayer()?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it works for now.",
+      "reference": "Avoid comma after the last parameter of a method call, unless each item is on its own line."
+    },
+    {
+      "prediction": "I think this is a bit confusing. I think it would be better to have a single `DataDir` that is used both for checkpointing and for agent-managed containers.",
+      "reference": "You should remove the outdated sentence."
+    },
+    {
+      "prediction": "I think this should be renamed to `STATE_KEY_FIRST_TIME_EMPTY_SUBJECT`",
+      "reference": "Please restore the empty line after the group of `STATE_*` constants. This visual separation makes the code easier to read."
+    },
+    {
+      "prediction": "I think this should be `RestoreFiles: removed ${fileIDs.length} files that finished uploading`",
+      "reference": "Unsure why this is `RestoreFiles` now vs `GoldenRetriever`, thought we deprecated the first?"
+    },
+    {
+      "prediction": "I'm not sure why we need to subtract 2 here.",
+      "reference": "Not sure what this is. Bad merge?"
+    },
+    {
+      "prediction": "nit: I think this should be `if err != nil { return err }`",
+      "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 we should add a test for this in this PR as well.",
+      "reference": "Can we remove this   ?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": " create testing votes\n```\nReview comment?",
+      "reference": "line is 195 characters (from `lll`)"
+    },
+    {
+      "prediction": "I don't think you need to import `k8s.io/kubernetes/pkg/util/slice` anymore.",
+      "reference": "could not import k8s.io/kubernetes/pkg/util/slice (invalid package name: \"\") (from `typecheck`)"
+    },
+    {
+      "prediction": "nit: can you remove the extra line here?",
+      "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.defineProperty(realmConstructor.Object.prototype, \"toJSON\", { value: function (_, cache = new Map()) { return this.map((item, index) => item.toJSON ? item.toJSON(index.toString(), cache) : item); }, writable: true, configurable: true });`",
+      "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 `lobGlobals = new ExLobGlobals(nullptr);`",
+      "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 think this should be `MvxModalPresentationAttribute` instead of `MvxModalViewPresentationAttribute`",
+      "reference": "Don't make it sealed"
+    },
+    {
+      "prediction": "I think this should be `a.limits.DeepCopy()`",
+      "reference": "Did we miss that for jwt before?"
+    },
+    {
+      "prediction": "I think this should be `sam.SERCOM0_USART` instead of `sam.SERCOM1_USART`.",
+      "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 this should be `sa` instead of `ar`",
+      "reference": "Spaces for indentation, not tabs"
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I think we can get rid of `binding_to_global_index_range_map_` and `global_start_to_index_map_` and just use `global_index_range_` and `bindings_` instead.",
+      "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'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "You should do: AssertEx.AreEqual(importPeptideSearchDlg.CurrentPage, ImportPeptideSearchDlg.Pages.match_modifications_page);"
+    },
+    {
+      "prediction": "why do we need to check `existSchema.getSummary() != null` here?",
+      "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 `>= 11 && <= 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 this is the right place to do this. I think it should be done in `proxy_test.go` instead.",
+      "reference": "I guess we can start by putting the `const StatusContextCancelled = 499` at the bottom of this file"
+    },
+    {
+      "prediction": "nit: I think it would be better to use `types.EmptySignature` instead of `nil` here.",
+      "reference": "Blocking: populate a non-empty value and check it below."
+    },
+    {
+      "prediction": "I don't think we need to create a new permission for each of these actions. We can just use the existing `ADMIN` permission.",
+      "reference": "What is this number value for? How is it used?"
+    },
+    {
+      "prediction": "I think this should be `assert: true, echo: true, retry: true, timing: true`",
+      "reference": "This needs an end keyword"
+    },
+    {
+      "prediction": "I think this should be `internal record DependencyTelemetryData : IEquatable<DependencyTelemetryData>` and override `Equals` and `GetHashCode` to use `Name` and `Version`.",
+      "reference": "I think this is the first record type in Datadog.Trace!"
+    },
+    {
+      "prediction": "Why is this change necessary? I don't think it's related to this PR.",
+      "reference": "Nit: missing newline above."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but it seems like a good place to me. I don't think we want to wait for the proxy to start before doing this.",
+      "reference": "Since it's a one-off, worth putting an info log before and after."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "I'm sorry, but the indentation is still wrong. This should be indented by one tab."
+    },
+    {
+      "prediction": "I think this should be `self._sdf.schema.fields` instead of `self._metadata.column_fields`.",
+      "reference": "can we add an example?"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this check. I think it should be done in the `Identify` call, and then the `Identify` call should return an error if the behavior is `CHAT_SKIP`, and then the caller can decide what to do with that error.",
+      "reference": "We usually avoid blank lines at the start of functions."
+    },
+    {
+      "prediction": "This file should be removed.",
+      "reference": "I dont think this line is needed"
+    },
+    {
+      "prediction": "I think this should be an error, not a fatal error.",
+      "reference": "We should not panic here !!! We return error i.e. 500 http code"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "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 we should have a common place to define these constants, so that we don't have to define them in multiple places.",
+      "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": "`dict(` -> `dict(`",
+      "reference": "Should not switch to `lvis_v1_instance` here because that base config uses ClassBalancedDataset to oversample the data."
+    },
+    {
+      "prediction": "Can you add a test for a table with only one row?",
+      "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 = IsFeatureFlagEnabled(ConfigurationKeys.FeatureFlags.AspnetRouteTemplateResourceNamesEnabled) || IsFeatureFlagEnabled(ConfigurationKeys.FeatureFlags.AspnetRouteTemplateResourceNamesEnabled);`",
+      "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 we should rename this to `SkipIID` or `SkipInstanceIdentityDocument` or something like that. `NoIID` is a bit confusing.",
+      "reference": "Would \"DisableIID\" be a better name? For example, we used DisableMetrics, ImageCleanupDisabled, ... etc. for other similar fields."
+    },
+    {
+      "prediction": "I don't think this is the right fix. The problem is that we don't know how many instructions will be in the next block, so we can't know when we've reached the limit. I think the right fix is to remove the `- 1` from the `max_bb_instrs` option.",
+      "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": "nit: I think it would be better to add `BOOLEAN` to the list of types that do not need to be converted to/from C strings.",
+      "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 `Blockly.WorkspaceSvg.prototype.isRightSide = function(e) {`",
+      "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 this should be `config:\"portrange;local\"`",
+      "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 this change is necessary.",
+      "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.outs.getsockname()[0])`",
+      "reference": "I suspect L2Socket needs a similar change (line 502) for the case where `iface` is None (line 479)"
+    },
+    {
+      "prediction": "I think we can remove this file since we don't have any customizations for this model in the admin interface.",
+      "reference": "Was this all that was needed to add CRUD operations for the LicensePermission model? :-)"
+    },
+    {
+      "prediction": "Shouldn't this be `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 think it would be good to have a JIRA to track this.",
+      "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 don't think you need to set the `MergePolicy` here. The `RandomIndexWriter` already uses a `MockRandomMergePolicy` by default.",
+      "reference": "Nice -- this forced `merge-on-getReader/commit` to be used more often in this test?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. I think we should just have a `Precision` field on the `Asset` type, and set it to `1000000000000000000` by default. Then we can just check `asset.Precision == 1000000000000000000`.",
+      "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": "I think this should be `if (!\\in_array($sortFieldParts[0], $aliases, true)) {`",
+      "reference": "where `$aliases` is defined ? @javiereguiluz"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we want to count the number of successful requests, not the number of identities that match the selectors.",
+      "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 think this should return an empty list of events, not an unimplemented error.",
+      "reference": "`ctx` is unused in ListEvents"
+    },
+    {
+      "prediction": "This should be `4.2.0` now.",
+      "reference": "We've had that test on iOS for years."
+    },
+    {
+      "prediction": "I think it would be better to have this as an option in the `BenchmarkConfig` class, so that we don't have to hardcode it here.",
+      "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 it would be better to raise an error if both `new_tab` and `new_window` are `False`.",
+      "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 30s or 60s. 10s is way too short.",
+      "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 think this should be `if (shiftAmountInt >= 256 || shiftAmountInt < 0 || shiftAmountInt > 255) {`",
+      "reference": "what happens if it's == 0"
+    },
+    {
+      "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 `run_as` method in the `Bolt::Shell::PowerShell` class.",
+      "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 don't think we need to update the import here.",
+      "reference": "Did we need this alias? Or is it just a holdout from development?"
+    },
+    {
+      "prediction": "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.<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.<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.<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.<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",
+      "reference": "This should have failed before."
+    },
+    {
+      "prediction": "I think we can remove the `RegexOptions.Compiled` here.",
+      "reference": "Slower construction, faster matching. I wonder if this will be noticeable in the relenv?"
+    },
+    {
+      "prediction": "I think this should be `<< throughput_publisher_.subscribers_ << \"/\" << throughput_publisher_.data_discovery_count_ << C_DEF << std::endl;`",
+      "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 <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": "This import is not used. Please remove it.",
+      "reference": "These tests actually go to the server."
+    },
+    {
+      "prediction": "nit: `onCompletedCalled = true;` should be `onCompleted = Task.Run(() => onCompletedCalled = true);`",
+      "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 think this should be `if (component.constructor.getDerivedStateFromProps)` instead of `if (typeof component.constructor.getDerivedStateFromProps === 'undefined')`",
+      "reference": "I'd be open to loosening this check if it can help offset the size."
+    },
+    {
+      "prediction": "I don't think we need to link to the user guide here. We can just say `See also Transform and apply a function`.",
+      "reference": "I'm wondering which we should use, stable or latest?"
+    },
+    {
+      "prediction": "I don't think we need to change the base class for this. We can just add a new method in the base class and override it in the subclass.",
+      "reference": "possibly extending EncodingType enum?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think it should be `driver.find_element(:id, 'iframe_page_heading').should be_kind_of(WebDriver::Element)` instead of `driver.find_element(:id, 'iframe_page_heading').should be_kind_of(Selenium::WebDriver::Element)`",
+      "reference": "I tested it only in Firefox (`./go //rb:firefox-test`)"
+    },
+    {
+      "prediction": "I don't think we need to implement `IProviderContainer<TracerProvider>` here.",
+      "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 this should be `frontend/express/public/core/events/javascripts/countly.events.models.js` and `frontend/express/public/core/events/javascripts/countly.events.views.js` instead of `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 think this should be `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 think this should be `self._bump_and_store_timestamp(collection_id, parent_id, record, modified_field=modified_field, last_modified=last_modified)`",
+      "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 is this removed?",
+      "reference": "why removed the MemoryDiagnoser attribute?"
+    },
+    {
+      "prediction": "I don't think this is correct. `localFieldGetters[0].call(doc, doc[localField]);` should be `localFieldGetters[0].call(doc, doc[localFieldPath.path]);`",
+      "reference": "Why do this instead of `localFieldPath.applyGetters(doc[localField], doc)` ?"
+    },
+    {
+      "prediction": "This test is no longer needed, as the `isomorphism_kind::induced` case is tested in `test_subgraph_isomorphism.cpp`.",
+      "reference": "Just for a check. Do you really want to delete this?"
+    },
+    {
+      "prediction": "I think we don't need to import `github.com/iotexproject/iotex-core/blockchain/action` here.",
+      "reference": "github.com/golang/protobuf/proto group in next group"
+    },
+    {
+      "prediction": "I think this should be `paych` instead of `paymentChannel`",
+      "reference": "Nit: what about just `pay`? Is there some other subcommand that would conflict with?"
+    },
+    {
+      "prediction": "I think it would be better to throw a `CoercingParseLiteralException` with a message like `Address value '${inputValue}' is not prefixed with '0x'`.",
+      "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": "Please remove the extra `}` at the end of the line.",
+      "reference": "How about use `protected` ?"
+    },
+    {
+      "prediction": "I don't think we need this special case. We should be able to store the root hash in the same way as any other hash.",
+      "reference": "Should it not be `common.HashLength*2 + common.IncarnationLength` ?"
+    },
+    {
+      "prediction": "Please remove the extra space after `node`.",
+      "reference": "those change are done by save plugin."
+    },
+    {
+      "prediction": "I think this should be `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 think we can remove this interface, it's not used anywhere.",
+      "reference": "This is an appsec concept right? Should it live in the Appsec namespace?"
+    },
+    {
+      "prediction": "I think this should be `machine.I2C0.WriteRegister(0x09, 'Z', version)`",
+      "reference": "I'd personally recommend having the demo check for error, unless it's worthless."
+    },
+    {
+      "prediction": "I don't think we should use `finer` here. `finest` would be more appropriate.",
+      "reference": "Since this is in a test, I imagine that the choice of `info` level was deliberate."
+    },
+    {
+      "prediction": "`toArray` is not used in this file.",
+      "reference": "good call swapping these names! makes more sense this way i think"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "Maybe leaving this out will save some bytes as in let `let isHydrating = replaceNode === null`"
+    },
+    {
+      "prediction": "I think this should be `GenNamespaceImports(cur_name_space_);`",
+      "reference": "Let's manually create one extra whitespace line here (code_ += \"\")"
+    },
+    {
+      "prediction": "Why do we need to construct a copy of the bucket pool?",
+      "reference": "nit: copy or clone"
+    },
+    {
+      "prediction": "I think this should be `return d.fields.MatchFold(revField) != nil || d.m != nil && d.m[revField] != nil`",
+      "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 we need to import `run` here.",
+      "reference": "Is this to avoid collisions?"
+    },
+    {
+      "prediction": "I think this should be `EnableProcessBlockStats` instead of `EnableAssembleStats`.",
+      "reference": "nit: a better name for this variable would be enableLogStats, but it's beyond the scope of your change."
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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 think this should be `logrus.StandardLogger().Out != os.Stdout`",
+      "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 right place to put this, but I think it should be in the `imports` namespace.",
+      "reference": "This is my favorite part of this PR. Being able to `import` 3rd party tools will hugely improve our workflows. :heart:"
+    },
+    {
+      "prediction": "I'm not sure about the name of this rule. Maybe `UnsynchronizedStaticMessageDigest` or `UnsynchronizedStaticMessageDigestRule` would be better.",
+      "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 it's better to return `null` instead of `Keccak.Zero`",
+      "reference": "Keccak.Zero should not be used to mean null"
+    },
+    {
+      "prediction": "I think it would be good to have a test case for `null` bounds as well.",
+      "reference": "perhaps add a couple more edge cases."
+    },
+    {
+      "prediction": "I don't think we need to explain what a bit_flag is here.",
+      "reference": "Is this blank line needed, or typo?"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I don't have a better idea.",
+      "reference": "This was a major PitA to find: it was causing some huge goroutine leaks and unfinished prefetches."
+    },
+    {
+      "prediction": "Use 2 (not 0) spaces for indenting an expression spanning multiple lines.",
+      "reference": "you can use Enum#flat_map here."
+    },
+    {
+      "prediction": "I don't think this is correct. The `PrintWriter` is not closed in this method.",
+      "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": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Put a comma after the last parameter of a multiline method call."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do, but I think it's the only way to get this to work.",
+      "reference": "It seems like this is the only way to access javascript `static` methods from Java."
+    },
+    {
+      "prediction": "I'm not sure if this is a good idea, but I think we can remove the `@private` tag from this file.",
+      "reference": "Why was this removed here?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "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 thing to do, but I think we should be using `project` instead of `select` here. The reason is that `select` is used to remove fields from the schema, while `project` is used to add fields to the schema. In this case, we want to add fields to the schema, so we should use `project`.",
+      "reference": "I agree with this because it is the opposite of `GetProjectedIds` used above."
+    },
+    {
+      "prediction": "I think this should be `ebsOptimized = True` instead of `ebsOptimized,`.",
+      "reference": "add 'sriovNetSupport' also to this list"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to be the only way to prevent duplicates from being added to the accountsInOrder list.",
+      "reference": "Code style issue: `if` body is not wrapped in braces."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "Call requires API level 11 (we are on 10)"
+    },
+    {
+      "prediction": "why do we need to make it read-only here?",
+      "reference": "do we need to pass the DB if it is the same"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should be able to create a chain for a gcOp.",
+      "reference": "Might be a good idea to rename the variable so it doesn't shadow the type."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "Again I'd prefer if this is done in JavaScript rather than in SQL."
+    },
+    {
+      "prediction": "nit: remove the extra space after `for`",
+      "reference": "there is still `2` spaces after `a` here.. :P"
+    },
+    {
+      "prediction": "I think this should be `if m.Device == \"bind\" && m.Destination == \"/dev\" && m.Source == \"/dev\"`",
+      "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 don't think we need a feature flag for this. We can just make it the default.",
+      "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 think this should be `receiver` instead of `provider`",
+      "reference": "Maybe `consumerID, receiverID, issuerID`, at least it's a convension in overal repo"
+    },
+    {
+      "prediction": "I think it would be nice to have a link to the blob.Open docs here.",
+      "reference": "Where does the auth go?"
+    },
+    {
+      "prediction": "I think this should be `if payloadGas == 0 || (math.MaxUint64-baseIntrinsicGas)/payloadGas < payloadSize {` instead of `if payloadGas == 0 || (math.MaxUint64-baseIntrinsicGas)/payloadGas < payloadSize {`",
+      "reference": "0 < minimum gas price, `ErrUnderpriced` is more proper"
+    },
+    {
+      "prediction": "I think this should be `TOKEN_EXP_MINUTES` instead of `TOKEN_EXP_DEFAULT`.",
+      "reference": "these should be alphabetized"
+    },
+    {
+      "prediction": "I think we can remove the `typeWidth` parameter from this method. It is only used in `readBatchOfDictionaryEncodedLongs` and `readBatchOfDictionaryEncodedFloats` and both of these methods are only called from `readBatchOfDictionaryEncodedLongs` and `readBatchOfDictionaryEncodedFloats` respectively.",
+      "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 think this should be `if cs.dirty.Load() == 3 {`",
+      "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 use `BaseModule` instead of `Module` ?",
+      "reference": "Base model is imported here"
+    },
+    {
+      "prediction": "remove unused import",
+      "reference": "import but not used"
+    },
+    {
+      "prediction": "I think we can remove the `nullabilityHolder` parameter from `readBatchOfDictionaryEncodedLongs` and `readBatchOfDictionaryEncodedFloats` as well.",
+      "reference": "What about the call in `case PACKED` just below? Does that also need to use the `typeWidth`?"
+    },
+    {
+      "prediction": "I think this should be `swarm.EncryptedReferenceSize * 2`",
+      "reference": "not sure i like this change. The encryption package does not need to know about references"
+    },
+    {
+      "prediction": "I don't think this is the right way to do this. You should be able to use `$(shell llvm-config-7 --version)` to get the version of the llvm installed on the system.",
+      "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 `if _, ok := config.Consensus[protocol.ConsensusCurrentVersion]; !ok {`",
+      "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 don't think this change is correct. `luigi.run(['--local-scheduler', '--no-lock'] + args)` should be equivalent to `luigi.run(args + ['--local-scheduler', '--no-lock'])`.",
+      "reference": "Is there a reason for the order swap here?"
+    },
+    {
+      "prediction": "I think we should be throwing a `FileNotFoundException` instead of an `InvalidOperationException` here.",
+      "reference": "Where will this exception get caught ?"
+    },
+    {
+      "prediction": "This should be `*keystore.KeyStore` instead of `keystore.KeyStore`",
+      "reference": "Why this field is made public?"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to pass the nodata value as an argument to the `read_vgrid_value` function. That way, the nodata value can be different for different grids.",
+      "reference": "Is the `vmultiplier` only used here for checking if a grid value is nodata?"
+    },
+    {
+      "prediction": "I think this should be `cert-manager.io` instead of just `cert-manager`",
+      "reference": "Is this deliberate? If so, why?"
+    },
+    {
+      "prediction": "I think this should be `https://testssl.sh/openssl-rfc.mapping.html`",
+      "reference": "both URL works. I corrected it anyway."
+    },
+    {
+      "prediction": "I don't think we need to make this `volatile` since it's only set in the constructor.",
+      "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 don't think this is correct. The level is a number between 0 and 8 inclusively.",
+      "reference": "The callback get the log level as a number."
+    },
+    {
+      "prediction": "I think this should be `>=2.2.0, <2.2.1, !=2.2.0.1`",
+      "reference": "Is there any issue with 2.2.3? If not this should be `<2.3`."
+    },
+    {
+      "prediction": "I think we should keep this test.",
+      "reference": "Shall we keep this test?"
+    },
+    {
+      "prediction": "Please use `GeneralUtility::makeInstance()` instead of `t3lib_div::makeInstance()`.",
+      "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 method should be `isSameCatalog(DataType other) { return this.getClass().equals(other.getClass()); }`",
+      "reference": "how about rename to shouldNarrowDataTypeTo?"
+    },
+    {
+      "prediction": "I think this should be `Store.current_backend = 'bokeh'` instead of `Store.current_backend = 'bokeh'`.",
+      "reference": "Why not check that the one available renderer is 'bokeh'?"
+    },
+    {
+      "prediction": "I think this should be `var ErrOverflow = fmt.Errorf(\"bundler reached buffered byte limit: %w\", bundler.ErrOverflow)`",
+      "reference": "Can you import `bundler.ErrOverflow` rather than redefine it here?"
+    },
+    {
+      "prediction": "I think this should be `resourceNameType = protoParser.getResourceReference(parameterField.getProtoField());` instead of `resourceNameType = protoParser.getResourceType(parameterField.getProtoField());`",
+      "reference": "This is just a renaming of the function getResourceType() to getResourceReference()"
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer single-quoted strings when you don't need string interpolation or special symbols.",
+      "reference": "Seems like this might need to allow for `\"`s to be used here?"
+    },
+    {
+      "prediction": "This should be `auto combatPtr = g_luaEnvironment.getCombatObject(g_luaEnvironment.lastCombatId);`",
+      "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 don't think we need this anymore.",
+      "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 we should keep these around for now, but we can move them to `dagster.core.sensors` in a follow-up PR.",
+      "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": "Why is this public?",
+      "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 this is the right thing to do here. I think we need to be able to distinguish between null and invalid values.",
+      "reference": "Is there a way to make it so `values.New(nil)` works instead of adding a new function?"
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think we should use `Sonata\\MediaBundle\\Form\\Type\\ApiDoctrineMediaType` as the parent type, and then override the `getBlockPrefix` method to return `sonata_media_api_form_doctrine_media`.",
+      "reference": "> when requirement of Symfony **will be** >= 2.8 Same for others."
+    },
+    {
+      "prediction": "I think this should be `roles/iam.serviceAccountTokenCreator` instead of `'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 remove unused imports.",
+      "reference": "Please remove the unused imports :)"
+    },
+    {
+      "prediction": "Align the elements of a hash literal if they span more than one line.<br>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": "I'm not sure if we want to enable pprof by default. I think we should only enable it in dev mode.",
+      "reference": "nit: this should be further down w/ the rest of the github imports"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to be the only way to get the amount in dollars instead of cents.",
+      "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 think this should be `if (stream == nullptr || stream != stream->getCtx()->_defaultStream && stream->getCtx()->_defaultStream != nullptr)`",
+      "reference": "Formatting here and all below: missing space before open bracket for the compound statement."
+    },
+    {
+      "prediction": "I don't think we need to remove this import.",
+      "reference": "Avoid using old mortbay utilities for converting JSON. Uses existing Solr Utils to convert from JSON string."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do here. I think we should be using `executeCommand` in the `execute` method of `CommandOperationV2` and then have `CommandDirect` extend `CommandOperationV2` instead of `CommandOperationV2` extending `CommandDirect`.",
+      "reference": "I actually think this wrapping is great, but let's call it `RunCommandOperation`?"
+    },
+    {
+      "prediction": "I think this should be `if what == \"*\" && handler.GenericErrorPage != \"\" { return hadBlock, c.Err(\"Expecting a numeric status code or '*', got '\" + what + \"'\") }`",
+      "reference": "Maybe it is better to add checking for status code duplications? It can prevent some stupid copy-paste user errors."
+    },
+    {
+      "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 don't think this is correct. The mangling is done in the prologue, not the epilogue.",
+      "reference": "This lea should match instr_check_xsp_mangling and be undone. I don't think i#3307 applies here."
+    },
+    {
+      "prediction": "I think this should be `ipld.StateIpldStore` instead of `ipld.StateIpldStore`.",
+      "reference": "I like the idea of this being an interface as it makes the test setup a bit easier, thoughts?"
+    },
+    {
+      "prediction": "I'm not sure why this is necessary. The `pubkey_hexes` list is already sorted, and `pubkey_deriv_info` is a dictionary, so the order of the keys is guaranteed to be the same as the order of the values.",
+      "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 a `ValueError` rather than an `Exception`.",
+      "reference": "Use a more granular error type. I would recommend `ValueError` in this case."
+    },
+    {
+      "prediction": "This should be `if (!g_events->eventPlayerOnTradeRequest(tradePartner, player, tradeItem)) { return; }`",
+      "reference": "A space after each argument, and an empty line above this if-statement."
+    },
+    {
+      "prediction": "I don't think this is necessary. `isReactComponent` is only used by `React.isValidElement`, which we don't use.",
+      "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 `void AddScopeListener(IScopeListener listener);` and `void RemoveScopeListener(IScopeListener listener);` instead of `void AddScopeListener(Action<Scope> listener);` and `void RemoveScopeListener(Action<Scope> listener);`",
+      "reference": "Maybe rename to `RegisterListener` or `RegisterSubscriber`?"
+    },
+    {
+      "prediction": "`refs` is not used in this function, so it can be removed.",
+      "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": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "We should get rid of this big array and think about a better way to check this."
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think it would be better to do this in the `MvxViewModelInstanceRequest` class itself. The `MvxViewModelInstanceRequest` class has a `ViewModelInstance` property, so it should be able to take care of caching the `ViewModelInstance` itself.",
+      "reference": "Should we cache the result of `Mvx.Resolve<IMvxChildViewModelCache>()` to avoid a lookup each time?"
+    },
+    {
+      "prediction": "I think this should be an error, not a warning.",
+      "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 think this should be `mux` instead of `httpMux`",
+      "reference": "let's just fold this inline in mux.go, no need for the indirection to this new file"
+    },
+    {
+      "prediction": "This import is not used.",
+      "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 this should be `assertEquals(22, tokens.size());`",
+      "reference": "There's a unnecessary System.out."
+    },
+    {
+      "prediction": "I think this should be `if c.enableProxy && gatewayConfig.IPv6 != nil {`",
+      "reference": "Should we add the flows only when NodePort is enabled?"
+    },
+    {
+      "prediction": "I think this should be `is_release_build=yes`",
+      "reference": "ITYM to flip this one..."
+    },
+    {
+      "prediction": "I think we can remove the `eslint:recommended` as it's already included in `plugin:@typescript-eslint/recommended`",
+      "reference": "What's the difference between typescript-eslint/eslint-recommended and typescript-eslint/recommended? I really can't get it."
+    },
+    {
+      "prediction": "I'm not sure I like the name \"AttackTimeout\". I think it would be better to call it something like \"MitigationTimeout\" or \"SuspiciousPacketTimeout\" or \"SuspiciousPacketMitigationTimeout\" or \"SuspiciousPacketMitigationDuration\" or \"SuspiciousPacketMitigationTime\" or \"SuspiciousPacketMitigationDelay\" or \"SuspiciousPacketMitigationWait\" or \"SuspiciousPacketMitigationWaitTime\" or \"SuspiciousPacketMitigationWaitDuration\" or \"SuspiciousPacketMitigationWaitDelay\" or \"SuspiciousPacketMitigationWaitTime\" or \"SuspiciousPacketMitigationWaitDuration\" or \"SuspiciousPacketMitigationWaitDelay\" or \"SuspiciousPacketMitigationWaitTime\" or \"SuspiciousPacketMitigationWaitDuration\" or \"SuspiciousPacketMitigationWaitDelay\" or \"SuspiciousPacketMitigationWaitTime\" or \"SuspiciousPacketMitigationWaitDuration\" or \"SuspiciousPacketMitigationWaitDelay\" or \"SuspiciousPacketMitigationWaitTime\" or \"SuspiciousPacketMitigationWaitDuration\" or \"SuspiciousPacketMitigationWaitDelay\" or \"SuspiciousPacketMitigationWaitTime\" or \"SuspiciousPacketMitigationWaitDuration\" or \"SuspiciousPacketMit",
+      "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": "I think we should keep the copyright line.",
+      "reference": "Why is this line deleted?"
+    },
+    {
+      "prediction": "I think this should be `node.store.block_count (transaction).number () >= node.ledger.bootstrap_weight_max_blocks` instead of `node.store.block_count (transaction).sum () >= node.ledger.bootstrap_weight_max_blocks`",
+      "reference": "minor type `boostrap`"
+    },
+    {
+      "prediction": "nit: remove unused import",
+      "reference": "Nit: please don't separate imports into groups."
+    },
+    {
+      "prediction": "I don't think we need to set the selected element to `null` here. I think we can just call `getControl().getOnAllCategorySelection().run();` and be done with it.",
+      "reference": "Thinking if it would be nice to have something like `setNothingSelected`."
+    },
+    {
+      "prediction": "I think this should be `Infof` instead of `V(logf.DebugLevel).Info`",
+      "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": "why not use `folly::toAppend`?",
+      "reference": "Generally, if cast from int to char*, you could cast directly. Because no errors should happen."
+    },
+    {
+      "prediction": "I think we should make this configurable in `azkaban.properties` as well.",
+      "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 `yaml:\"logging,omitempty,flow\"`",
+      "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": "I think it's better to return `void` here.",
+      "reference": "Would it make sense to also add `setMaxParallelism` in addition to this to match the Flink API?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to call `closeBillingWindow` but it seems to be the only way to close the modal.",
+      "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 think we can remove the `!c._pendingSuspensionCount++` check and just do `c.setState({ _suspended: (c._detachOnNextRender = c._vnode._children[0]) });`",
+      "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'm not sure this is the right thing to do here. I think we should just let the `os.path.exists` call raise the `UnicodeEncodeError` and let it bubble up to the caller.",
+      "reference": "I think it's a good idea to log this as debug, the same way the other branch gets logged."
+    },
+    {
+      "prediction": "I think this should be ``a.get_jsonld(\"oaf\")``",
+      "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": "I don't think we need this constant. We can just use `\"system.guest\"` directly.",
+      "reference": "keeping this as an unexported constant, as importing the role pkg creates a cyclic dependency."
+    },
+    {
+      "prediction": "I think this should be `if a.Doc.RevisionOn(c.RevisionField()) && c.opts.RevisionField != \"\" {`",
+      "reference": "Why `c.RevisionField()` here but `c.opts.RevisionField` just below?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "nit: this file diff seems unrelated and probably should have been a separate PR"
+    },
+    {
+      "prediction": "I think this should be `const 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 it's a good idea to log it here. It's not an error, it's just a debug info.",
+      "reference": "Is this from debugging?"
+    },
+    {
+      "prediction": "I think this test is not needed anymore.",
+      "reference": "Its betet use MockedObject"
+    },
+    {
+      "prediction": "I think this should be `DeleteByTLF` instead of `Delete`, since it's still deleting by TLF.",
+      "reference": "\"including\"? Since it doesn't take any parameters, I'm not sure how it can update anything else..."
+    },
+    {
+      "prediction": "I think this should be changed to \"Azure Resource Management only supports HTTPS scheme.\"",
+      "reference": "nit: I'm assuming you have verified this is what the error message should read like"
+    },
+    {
+      "prediction": "`sys.exit` returns `None`, so this should be `sys.exit(main(sys.argv[1:]))`",
+      "reference": "Why do you need that? (real question)"
+    },
+    {
+      "prediction": "This change is unrelated to this PR. Please revert it.",
+      "reference": "If I understand the patch correctly the `else` is no longer required."
+    },
+    {
+      "prediction": "`none` is not a valid value for `blacklistTypeStr`. It should be `none` or `worker` or `master`.",
+      "reference": "Properly describe valid flag values here: `neither` should be replaced with `none` as they parsed below."
+    },
+    {
+      "prediction": "I think we can remove this try-catch block. If we can't remove the executor from the database, it's not the end of the world. The executor will be removed from the database next time the server restarts.",
+      "reference": "Can we escalate this to `warn`. It is a major event."
+    },
+    {
+      "prediction": "I think these should be 22 and 23, respectively.",
+      "reference": "Please move these up to line 86 so the numbers are in order."
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "Add a docstring to contain the paper link."
+    },
+    {
+      "prediction": "I think this should be `targ := \"--target=\" + config.Target.Triple`",
+      "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 thing to do, but I don't have a strong opinion either way.",
+      "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": "why do we need this?",
+      "reference": "you could probably use `const` for both?"
+    },
+    {
+      "prediction": "I don't think we need to export this function. It's only used in the `matches` function below.",
+      "reference": "This looks like an \"ES Module\" to me. I do not understand this TODO."
+    },
+    {
+      "prediction": "nit: `final` is not needed here.",
+      "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": "This should be `lua_pushlstring(L, player->getIP().to_string().c_str(), player->getIP().to_string().length());`",
+      "reference": "There's `pushString` to use with C++ strings."
+    },
+    {
+      "prediction": "I think this should be `Eventually(getCIDRBlockRules(felix, \"ip6tables-save\"), \"8s\", \"0.5s\").Should(BeEmpty())`",
+      "reference": "qq: Should this include the iptables6-save sim. to the inverse checks above?"
+    },
+    {
+      "prediction": "I don't think we need to wake up the async thread here. The async thread will be woken up when the writer is added to the userWriterList.",
+      "reference": "What about async user writers? We could be checking `param.mode == ASYNCHRONOUS_WRITER`"
+    },
+    {
+      "prediction": "I don't think you can use `getAnnotation` here, because `parent` is an `ASTClassOrInterfaceDeclaration` and not an `ASTClassOrInterfaceType`.",
+      "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 `Interceptors(...func(http.Handler) http.Handler) InboundOption` instead of `Interceptor(func(http.Handler) http.Handler) InboundOption`.",
+      "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 think this should be `static final String ELEVATE_DOCS_WITHOUT_MATCHING_Q = QueryElevationParams.ELEVATE_DOCS_WITHOUT_MATCHING_Q;`",
+      "reference": "We should not use a static import to be consistent with other existing use of the QueryElevationParams."
+    },
+    {
+      "prediction": "I think this should be `memset_s(&null_gbs_info, sizeof(null_gbs_info), 0, sizeof(null_gbs_info));`",
+      "reference": "No need to check return value?"
+    },
+    {
+      "prediction": "Why not use `importlib_resources` on 3.7-3.8 as well?",
+      "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 we can just use `len(antreaSourceStatsElementList)` instead of `len(antreaSourceStatsElementList) + len(antreaDestinationStatsElementList) + len(antreaLabelsElementList)`, since `antreaDestinationStatsElementList` and `antreaLabelsElementList` are always empty.",
+      "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": "I don't think we need this variable. We can just use `MCXT_NUM_SIMD_SLOTS` instead of `num_simd_saved_abs`.",
+      "reference": "Same thing: `num_simd_registers`. Ditto below."
+    },
+    {
+      "prediction": "This should be `2.3.0.dev0`",
+      "reference": "Should be 2.1.1"
+    },
+    {
+      "prediction": "I think it would be better to pass the `required_subject` as an argument to the `assertions` function, instead of having it as a separate parameter.",
+      "reference": "nitpick: `required_subject` doesn't seem to be used here"
+    },
+    {
+      "prediction": "Please remove this empty line.",
+      "reference": "We can revert this to reduce the diff of the PR."
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this, but I think it should be done in `__init__` rather than `_get_presentationType`. This is because `presentationType` is used in `__init__` to determine whether to use `UIA` or `UIAContent` as the base class. If `presentationType` is set to `self.presType_layout` in `_get_presentationType`, then `UIAContent` will be used as the base class, which means that `presentationType` will never be set to `self.presType_content` in `__init__`, which means that `presentationType` will never be set to `self.presType_layout` in `_get_presentationType`, which means that `presentationType` will never be set to `self.presType_content` in `__init__`, which means that `presentationType` will never be set to `self.presType_layout` in `_get_presentationType`, which means that `presentationType` will never be set to `self.presType_content` in `__init__`, which means that `presentationType` will never be set to `self.presType_layout` in `_get_presentationType`, which means that `",
+      "reference": "Extraneous blank line."
+    },
+    {
+      "prediction": "I think this should be `codechecker_lib.analyzers.result_handler_plist_to_db`",
+      "reference": "Is this the same import as in the line 20?"
+    },
+    {
+      "prediction": "Line is too long. [86/80]",
+      "reference": "why those additional checks after published? template.version should always be present and greater than zero"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this. I think this should be done in a separate PR.",
+      "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": "I think this should be `Automatically sends crash reports to Brave`",
+      "reference": "I merged already so need another PR, but I think this needs to be at the bottom."
+    },
+    {
+      "prediction": "Please remove the trailing space.",
+      "reference": "space in key string?"
+    },
+    {
+      "prediction": "I think this should be `if b.localMnt && len(b.nsOptions) > 0 {` so that we don't add `-l` if there are no namespace options.",
+      "reference": "The commends of this function should be updated"
+    },
+    {
+      "prediction": "I think it's better to add `reg_decoded_bbox` as a parameter of `AnchorHead` and pass it to `BBoxHead`.",
+      "reference": "The docstring is outdated."
+    },
+    {
+      "prediction": "I don't think we need to make these `internal` since they are only used in this file.",
+      "reference": "Could rename to `ServiceVersionKey` for consistency with the suggestion to rename the TracerSetting."
+    },
+    {
+      "prediction": "I think this should be `overwrite.lower() == 'y'`",
+      "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 `getDefaultSettings` and not `setDefaultSettings`",
+      "reference": "The `MediaBlockService` commit must be remove"
+    },
+    {
+      "prediction": "nit: I think it would be better to use `len(rules_engine.rule_book.resource_rules_map) == len(rre.SUPPORTED_RETENTION_RES_TYPES)` instead of `len(rre.SUPPORTED_RETENTION_RES_TYPES) == len(rules_engine.rule_book.resource_rules_map)`.",
+      "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 think this should be `Uppy v3`?",
+      "reference": "@Murderlon Should it just say `some future version`?"
+    },
+    {
+      "prediction": "Why do we need to assign cspcObj to cspcGot here?",
+      "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 don't think this is the right way to do it. We should have a separate endpoint for getting address details.",
+      "reference": "invalid operation: exp (variable of type *Service) has no field or method getAddressDetails (from `typecheck`)"
+    },
+    {
+      "prediction": "I don't think you need to set the config here, right?",
+      "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 what you mean by \"it isn't goroutine safe\". `t` is not goroutine safe, but `syncErr` is.",
+      "reference": "If you feel like it, may as well apply all the suggestions below to this test too."
+    },
+    {
+      "prediction": "I think this should be `==` instead of `===`",
+      "reference": "do we really need a relation or can we just assume INTERSECTS?"
+    },
+    {
+      "prediction": "I don't think we need to change the visibility of this class.",
+      "reference": "Is this a `MetricKey` ?"
+    },
+    {
+      "prediction": "I think this should be `public static final long DEFAULT_SCHEDULE_END_EPOCH_TIME = 2524608000000000000L;`",
+      "reference": "what's your rationale to pick up this date?"
+    },
+    {
+      "prediction": "nit: I think we can remove the `fscommon` prefix here.",
+      "reference": "NIT: I'd have preferred that the new package be called \"utils\", but that's not a big deal."
+    },
+    {
+      "prediction": "I don't think we need to mention AWS KMS CMK here.",
+      "reference": "Simple Storage **Service**"
+    },
+    {
+      "prediction": "I think it would be better to move this logic into the `SolrConnection` class. The `SolrConnection::commit()` method should only commit if the `SolrConfiguration::getEnableCommits()` returns `true`.",
+      "reference": "I think, retrieving the setting could also be done outside the loop. What do you think?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to do this. I think we should do this in the `TransactionValidator` itself. The `TransactionPool` should not know anything about the private state.",
+      "reference": "This TODO isn't related to this change. We should remove it."
+    },
+    {
+      "prediction": "I think this should be `logger.finer` instead of `finest`. `finest` is the lowest level of logging, and I don't think we want to log this at that level. `finer` is the next lowest level, and I think it's appropriate for this level of logging.",
+      "reference": "We chose `info` in the test code for obvious reasons. Changing to `finest` makes debugging harder and noisier."
+    },
+    {
+      "prediction": "Shouldn't this be `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 think this should be `env.Printf(\"Selectors     : %s\\n\", strings.Join(agent.Selectors, \", \"))`",
+      "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 (playerOutfit && playerOutfit->lookMount != 0) {`",
+      "reference": "Remove this line please."
+    },
+    {
+      "prediction": "I think this should be a `ConcurrentHashMap` instead of a `HashMap`.",
+      "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'm not sure if this is the best way to do this, but it seems to be the only way to get access to the `ariaRoles` and `ariaAttrs` objects.",
+      "reference": "I agree with Stephen it's better to invoke getStandards() only once."
+    },
+    {
+      "prediction": "`irregular_dims` should be a set, not a list.",
+      "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": "I don't think we need to log the creation of the container.",
+      "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": "I think this should be `td:nth-child(n+2)` so that it applies to all columns except the first one.",
+      "reference": "this makes sure there is some padding space around the size column"
+    },
+    {
+      "prediction": "I think this should be `#include <Util/LuaUtil.h>`",
+      "reference": "Put these includes in alphabetical order"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "reference": "Can this just be an else?"
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "Merge this line with Line7"
+    },
+    {
+      "prediction": "Line is too long. [101/80]",
+      "reference": "`clearfix` is needed to stop the `float:right` content overflowing the list item container."
+    },
+    {
+      "prediction": "I think this should be `[AspNetCoreMvc3Benchmark]` instead of `[AspNetCoreBenchmark]`.",
+      "reference": "Can this be put on the assembly?"
+    },
+    {
+      "prediction": "I think this should be `!alwaysOk || isAllHealthy(results)`.",
+      "reference": "Could we rename the parameter to `overrideStatusCode`? The name `alwaysOk` implies that the health check result would be always healthy."
+    },
+    {
+      "prediction": "I don't think we need to expose the name of the schedpolicy in the API.",
+      "reference": "\"/schedpolicy\" is repeated, i would make it a constant and use it here and cluster client.go"
+    },
+    {
+      "prediction": "Style/RedundantKernelSrand: Redundant Kernel#srand call detected.",
+      "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 don't think this is the right place to do this. I think it should be done in the `__init__` method of the `AddonsManager` class.",
+      "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": "nit: I think it would be better to call it `expireMetadata` instead of `expireTableMetadata`",
+      "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 don't think we need this. `helpers.FullClose` is just a convenience function that calls `Close` and then waits for the stream to be fully closed. I think we can just call `Close` on the stream directly.",
+      "reference": "I do not think that this type (and its constructor) have to be exported."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the caller of this 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 don't think this is necessary. We can just use `acme.ChallengeProvider` directly.",
+      "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": "Why do you need to add `daal::algorithms::adaboost::quality_metric_set::` here?",
+      "reference": "Why interface1 ? It should be in inner so."
+    },
+    {
+      "prediction": "I don't think we need to include the `auth_token` here.",
+      "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 best way to handle this. I think it would be better to have a `if (tbase == t_base_type::TYPE_STRING && type->is_binary() && allocate) { ... } else { ... }`",
+      "reference": "Why is that test different to line 4022 (allocate is not tested there)? Shouldn't that be consistent?"
+    },
+    {
+      "prediction": "I think this should be `if (rc == -1) { step_ = HANDLE_ERROR; } else if (rc == 0) { step_ = INSERT_CLOSE; } else if (rc == 1) { step_ = CREATE_MUTATIONS; } else { step_ = HANDLE_ERROR; }`",
+      "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 think we can remove the `+ 1` here.",
+      "reference": "Don't see any more TODOs here.."
+    },
+    {
+      "prediction": "I think this should be `strings.TrimPrefix(dockerImage, \"s3://\")` instead of `strings.Split(dockerImage, \"/\")[len(slice)-1]`",
+      "reference": "Can you extract this out to a constant?"
+    },
+    {
+      "prediction": "I don't think we need to load the log module here.",
+      "reference": "any testcase for saargs, and tidyargs argument processing?"
+    },
+    {
+      "prediction": "I think this should be `private const string PlatformNotSupportedExceptionMessage = \"Platform not supported\";` so that it can't be changed at runtime.",
+      "reference": "@ericstj do you think it is worth combining these two? With the presence of the message meaning it is enabled?"
+    },
+    {
+      "prediction": "I think this should be `/dev/xvda1` instead of `/dev/xvda`.",
+      "reference": "why not use instance.RootDeviceName ?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "not needed, ADIOS2 also needs to check for subfiles. ADIOS2 tries to remove dependency on serial `fstream`."
+    },
+    {
+      "prediction": "I think this should be `throw new Error(`No extension support for ${path} in react-native-sqlite-storage`);`",
+      "reference": "no need to wrap in new Promise - you can simply throw the exception"
+    },
+    {
+      "prediction": "I think this test should be moved to the `tls_test.go` file, as it's testing the `tls` package, not the `server` package.",
+      "reference": "Isn't this testing an implicit config, not a manual config?"
+    },
+    {
+      "prediction": "I think this is a bit confusing. I think it would be better to have a `map[string]map[string]string` where the first key is the chain name, and the second key is the rule hash, and the value is the full rule.",
+      "reference": "I don't think this is used any more, please remove."
+    },
+    {
+      "prediction": "Why do we need to convert the node key to an identity here?",
+      "reference": "shouldn't we name abbreviations up-cased? `myId` -> `myID`?"
+    },
+    {
+      "prediction": "I think this should be `node[ATTR_KEY].ref = null;` instead of `delete node[ATTR_KEY].ref;`",
+      "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 think this should be `t.Fatal`",
+      "reference": "Oof. I may have missed this while reviewing #3805"
+    },
+    {
+      "prediction": "I don't think you need to include `fstream` and `stdexcept` here.",
+      "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": "Shouldn't this be `export const ReactComponent = () => <div />;`?",
+      "reference": "Should we maybe return `<svg />` instead?"
+    },
+    {
+      "prediction": "I think this should be `const load_plugins = require('../lib/plugin-loader').load_plugins;`",
+      "reference": "Please use camelCase in new code"
+    },
+    {
+      "prediction": "I think this should be `place` instead of `places`.",
+      "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 don't think this is a good idea. If you want to use `noConflict()` you can do it in your own bundle.",
+      "reference": "This was just my laziness and I didn't copy the first line. Just remove it together with one empty line. Thanks"
+    },
+    {
+      "prediction": "I think it's better to use `Shopsys\\FrameworkBundle\\Model\\Product\\ProductDataFactoryInterface` instead of `Shopsys\\ShopBundle\\Model\\Product\\ProductDataFactory`",
+      "reference": "Why did you decide to change CartItemTest but you didnt change QueryBuilderWithRowManipulatorDataSourceTest?"
+    },
+    {
+      "prediction": "I think this should be `if pollForKID == keybase1.KID(\"\") { return cachedUserInfo, nil }`",
+      "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:build go1.16 && integration`",
+      "reference": "Should this tag addition for generated files be handled explicitly in `private/model/cli/gen-api/main.go`"
+    },
+    {
+      "prediction": "I think this should be `!K9.getQuietTimeNotificationEnabled()` instead of `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": "`APP_DOMAIN` is only set in production, so this will fail in other environments. I think we should set `APP_DOMAIN` to `learn.thoughtbot.com` in `config/environments/development.rb` and `config/environments/test.rb`.",
+      "reference": "Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping."
+    },
+    {
+      "prediction": "I think this should be `source_host.include? 'github.com'` instead of `source_host.exclude? 'github.com'`",
+      "reference": "cool, here (L146 && L148) maybe (repo_private? || force_private?) && !installation? to prevent double call of installation?"
+    },
+    {
+      "prediction": "I think this should be `Failed(\"Failed to GetActiveApp: %v\", 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 think this should be a warning, not an error.",
+      "reference": "Same: DO NOT USE IN PRODUCTION. Yes we should shout ;)"
+    },
+    {
+      "prediction": "Is there a reason to increase this from 3000ms to 5000ms?",
+      "reference": "This change may be hiding more errors, if we continue to see more."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. If the connection has a feature, it should be used. If the connection doesn't have a feature, then the stream should not have one either.",
+      "reference": "Beware of exposing the underlying features directly. Any mutable fields should be reset per request."
+    },
+    {
+      "prediction": "I'm not sure if this is the right fix, but it seems to work for me.",
+      "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 think this should be `EC.element_to_be_clickable(By.ID, \"disabled-button\")` instead of `EC.element_to_be_unclickable(By.ID, \"disabled-button\")`",
+      "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 `public FileAppender<T> appender() { 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": "I'm not sure why this was changed, but I think it should be `2021-01-18`",
+      "reference": "the test was failing likely due to locale issue. had to increment the date by 1 to make it pass"
+    },
+    {
+      "prediction": "I don't think this should be a warning, it should be an error.",
+      "reference": "The changes below here can also be undone."
+    },
+    {
+      "prediction": "`beehiveContext` -> `beehive_context`",
+      "reference": "Instead of importing context besides beehivecontext, I'd suggest include golang context functionalities into beehivecontext. To keep channelq only rely on beehive."
+    },
+    {
+      "prediction": "I don't think we need this class to be public.",
+      "reference": "I'm not sure about this one. Are you sure we don't need to initialize this fragment anywhere else?"
+    },
+    {
+      "prediction": "I don't think this is the right place to add the treatment to the history",
+      "reference": "You didn't set a source here, so it will probably fail."
+    },
+    {
+      "prediction": "I think this should be `transporttest.NewNoDepsMatcher(t)` instead of `transport.NoDeps`.",
+      "reference": "Might be easier to just do `handler{Handler: rpcHandler}` since zero-value of `Deps` is now valid."
+    },
+    {
+      "prediction": "Style/StringLiterals: Prefer double-quoted strings unless you need single quotes to avoid extra backslashes for escaping.",
+      "reference": "Could you please add docs for this method?"
+    },
+    {
+      "prediction": "I don't think this is needed.",
+      "reference": "we should probably `@Deprecated` this implementation to be removed in PMD 7.0.0"
+    },
+    {
+      "prediction": "I think this should be `target.options['shell'] == 'powershell'` instead of `target.options['login-shell'] == 'powershell'`",
+      "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 why this change is necessary.",
+      "reference": "Do something similar to what was done in JsonTests (can you derive from JsonTests?)"
+    },
+    {
+      "prediction": "I don't think we need to ignore the type here. `__path__` is a `List[str]` and `pkgutil.extend_path` returns a `List[str]` as well.",
+      "reference": "Yes, not much we can do here."
+    },
+    {
+      "prediction": "nit: I think we can remove this import.",
+      "reference": "Why do we need this?"
+    },
+    {
+      "prediction": "I think this should be `bitcore.networks.testnet` instead of `bitcore.networks.livenet`",
+      "reference": "Would prefer to see an actual script here."
+    },
+    {
+      "prediction": "I think this should be `from.toPath().toString()`.",
+      "reference": "seems reasonable to also want to include the 'to' location?"
+    },
+    {
+      "prediction": "Please remove this line.",
+      "reference": "This change is unnecessary."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be the caller's responsibility to call `unselectAll()` before calling `setOnMouseClicked()`.",
+      "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 don't think we need to make this configurable. If we want to add more annotations in the future, we can just add them here.",
+      "reference": "What about Secrets, are we not also trying to cover those?"
+    },
+    {
+      "prediction": "(style) line over 80 characters",
+      "reference": "(commit messge): how about > update internal commit API for RFC 11"
+    },
+    {
+      "prediction": "I think this should be `request.ConstantWaiterDelay(10 * time.Second)`",
+      "reference": "This will now wait for 2.5 hours, not 90 minutes, if the waiter delay is 5s"
+    },
+    {
+      "prediction": "I think this should be `if (!event || !event.nativeEvent) return`",
+      "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'm not sure this is the right thing to do, but it seems to be the only way to get the `etcdctl snapshot save` command to work.",
+      "reference": "We're passing the same thing twice?"
+    },
+    {
+      "prediction": "I think this should be `base64.StdEncoding.EncodeToString([]byte(\"user-data\"))`",
+      "reference": "Shouldn't the value in Bootstrap.Data be just a plain string?"
+    },
+    {
+      "prediction": "I think this should be `SamplingPriority.UserReject` instead of `SamplingPriority.AutoReject`",
+      "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": "`/readyz` should be `/healthz/ready`",
+      "reference": "The const can also be used in cloud/pkg/cloudhub/servers/httpserver/server.go L46-47?"
+    },
+    {
+      "prediction": "I think this should be `===` instead of `==`.",
+      "reference": "Use a constant for `procedures_prototype`."
+    },
+    {
+      "prediction": "`arg!r` should be `arg!a`",
+      "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 thing to do here. We don't want to set SA_RESTORER if it wasn't set by the caller.",
+      "reference": "style nit: missing leading '*'"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should have a `OperationTracer.NO_TRACING` instance that we can reuse.",
+      "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 if this is the right thing to do, but I think we need to keep the SPI0_SCK_PIN, SPI0_MOSI_PIN, and SPI0_MISO_PIN variables around for compatibility.",
+      "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": "Layout/TrailingWhitespace: Trailing whitespace detected.<br>Metrics/LineLength: Line is too long. [121/120]",
+      "reference": "May be useful to have the reason here, though clearly not pressing."
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but it seems to work for me.",
+      "reference": "These were inverted before"
+    },
+    {
+      "prediction": "I think this should be `return Collections.emptyMap();` instead of `return new HashMap<>();`",
+      "reference": "DB is implemented in PR 951"
+    },
+    {
+      "prediction": "I don't think these are used anywhere.",
+      "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": "I think this should be `Copyright (C) 2006-2021 NV Access Limited, Manish Agrawal, Derek Riemer, Babbage B.V.`",
+      "reference": "Could you revisit the full header and add appropriate spaces after the hashes?"
+    },
+    {
+      "prediction": "I think this should be `codeEditorHint` or `codeEditorHelpText` or something like that. `codeHint` sounds too generic.",
+      "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 this down to 2 seconds.",
+      "reference": "this was flapping"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to do this, but it's the only place I could find where the height of the keyboard is calculated.",
+      "reference": "Please add a check to see if SafeAreaInsets is available. It isn't on all iOS versions."
+    },
+    {
+      "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.<br>",
+      "reference": "interpolation here was causing the translation to not get picked up"
+    },
+    {
+      "prediction": "I don't think we need to unregister the writer from the AsyncWriterThread here. The AsyncWriterThread will be destroyed when the RTPSParticipant is destroyed, so the writer will be automatically unregistered.",
+      "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` by default.",
+      "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 `Match().OutInterface(ifaceMatch).IPVSConnection()` instead of `Match().OutInterface(ifaceMatch).IPVSConnection()`.",
+      "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'm not sure why this is needed, but it seems like it could be replaced with `return int(math.Max(float64(x), float64(y)))`",
+      "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": "nit: I think we can use `forwarding-all-apis` instead of `all-apis-forwarding` to make it more consistent with `forwarding-selected-apis`",
+      "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 don't think we need to add a new framework id for this. I think we can just use `FrameworkPrometheus` and `FrameworkM3` here.",
+      "reference": "I wouldn't add this. Just completely ignore config if custom reporter is not `nil` in server options."
+    },
+    {
+      "prediction": "I don't think this is the right place to load the cluster id. I think it should be loaded in the `MetaClient`'s `waitForMetadReady` method.",
+      "reference": "What's the meaning about the code?"
+    },
+    {
+      "prediction": "I think this should be `loadingUrl.getHost().equals(bridge.getHost())` instead of `loadingUrl.getHost().contains(bridge.getHost())`",
+      "reference": "Would this incorrectly trigger if `loadingUrl.getHost() = \"something.app\"` and `bridge.getHost() = \"app\"`?"
+    },
+    {
+      "prediction": "I think it's better to put this in the docstring of `loss_bbox` instead of `reg_decoded_bbox`.",
+      "reference": "Note generally it -> It"
+    },
+    {
+      "prediction": "I think we can remove the `force` parameter from the `createIndexSchema` method.",
+      "reference": "Does this mean that every call to `configSet.getIndexSchema()` will create a new schema object (if shared schema is not enabled)?"
+    },
+    {
+      "prediction": "Please remove the empty line.",
+      "reference": "License header should use the Apache one."
+    },
+    {
+      "prediction": "I don't think we should be panicking here. I think we should just return an empty list of matches.",
+      "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 think this should be `if config.OnPrem.Enabled() && config.AWSRegion == \"\" { return nil, errors.New(\"AWS_DEFAULT_REGION has to be set when running on-premises\") }`",
+      "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.version` instead of `self.__version`.",
+      "reference": "maybe naming `version` to `package_version` or something similar can be more descriptive"
+    },
+    {
+      "prediction": "I don't think we need to make this protected.",
+      "reference": "Why was this changed?"
+    },
+    {
+      "prediction": "I think this should be `!amd64` instead of `!arm64`",
+      "reference": "Should not have snuck in here right? This is a golang 1.17 feature, wii definitely want to update these"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [121/120]",
+      "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": "I don't think this is the right place to log the warning. The warning should be logged when the incremental analysis is enabled, not when it is disabled.",
+      "reference": "this warning shouldn't be produced if we configured a cache either"
+    },
+    {
+      "prediction": "Why is this needed?",
+      "reference": "what if the collection is already there?"
+    },
+    {
+      "prediction": "I think we should keep the link to `org.openqa.selenium.Screenshot`",
+      "reference": "instead of removing can you reference org.openqa.selenium.TakesScreenshot ?"
+    },
+    {
+      "prediction": "Please remove empty lines at the end of the file.",
+      "reference": "Please revert unrelated changes"
+    },
+    {
+      "prediction": "I think it would be better to use `jQuery.i18n.map[\"common.company\"]` instead of `jQuery.i18n.map[key]`",
+      "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 change is necessary. The `jsonTypeDTO.getName().replaceAll(\"[^a-zA-Z0-9]\", \"\")` is used to generate the `id` of the type. The `id` of the type is used as the name of the directory where the type is stored. The name of the directory should not contain any special characters, so the `replaceAll(\"[^a-zA-Z0-9]\", \"\")` is necessary. The `jsonTypeDTO.getName().replaceAll(\"[^a-zA-Z0-9_]\", \"\")` is used to generate the `name` of the type. The `name` of the type is used as the name of the file where the type is stored. The name of the file should not contain any special characters, so the `replaceAll(\"[^a-zA-Z0-9_]\", \"\")` is necessary. So, the `replaceAll(\"[^a-zA-Z0-9]\", \"\")` is necessary for both the `id` and the `name` of the type. So, the `replaceAll(\"[^a-zA-Z0-9_]\", \"\")` is necessary for both the `id` and the `name` of the type. So, the `replaceAll(\"[^a-zA-Z0-9_",
+      "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": "nit: I think we can remove the `from dagster.core.definitions.events import AssetKey` line since it's not used in this file.",
+      "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 we can remove this import.",
+      "reference": "No need for blank line here. This block will then be reordered alphabatically."
+    },
+    {
+      "prediction": "I think this should be `local-cluster` or `local-cluster-name` or something like that to make it clear that this is the name of the local cluster.",
+      "reference": "let's call it `in-cluster` instead of `local`. less chance of confusion."
+    },
+    {
+      "prediction": "I don't think we need to store the previous ref here. It's not used anywhere.",
+      "reference": "Please use camelCase instead of underscore_case"
+    },
+    {
+      "prediction": "I think this should be `trim_to_bytes(f\"{info}\\n{body}\", ELASTIC_LIMIT_BYTES)`",
+      "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 `SCHEMA` instead of `FIELDS`.",
+      "reference": "Is this used?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we should be able to remove the `#if NET45 || NET46` from this file, and just use `#if NET452 || NET46`.",
+      "reference": "Should we use `#if NETFRAMEWORK` for consistency with the other projects?"
+    },
+    {
+      "prediction": "I think this should be `return decoded, err` instead of just `return`",
+      "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 to remove this method. It's used in the tests.",
+      "reference": "line 26 can be removed i think > applicationId: or('config.unsplashAPI.applicationId', 'settings.unsplash.applicationId'),"
+    },
+    {
+      "prediction": "I don't think we need to change this line.",
+      "reference": "To generalize for possible future new enum options."
+    },
+    {
+      "prediction": "I don't think this is the right way to do it. I think you should use `os.path.dirname(os.path.realpath(__file__))` instead of `os.environ.get(\"PWD\", os.getcwd())` because `os.environ.get(\"PWD\", os.getcwd())` will always return the current working directory and not the directory where the script is located.",
+      "reference": "Shall we remove this variable at the end of UTScapy execution?"
+    },
+    {
+      "prediction": "nit: I think it would be better to throw an exception if the service config is empty.",
+      "reference": "What is this checking for? Why is returning the list of interfaces the correct behaviour in this case?"
+    },
+    {
+      "prediction": "I don't think we need this one. We can just use `FastSync` for HMB sync as well.",
+      "reference": "Please name it correctly as Beam ;)"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we should only retry if the token is still valid. If the token is not valid, we should not retry.",
+      "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": "Shouldn't this be `ServiceDefinitionDocumentState.Composed` by default?",
+      "reference": "any reason to not have defaults here like you had before? (ARM)?"
+    },
+    {
+      "prediction": "I'm not sure if this is correct, but I think it should be `page_info.limit_value` instead of `@response.limit_value`.",
+      "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 don't think this change is necessary.",
+      "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 don't think this is true. If you return a Portal from a component, it will be treated like any other component and will be diffed normally.",
+      "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 don't think we need to clone `hooks` here. `clone` is only used in `cloneModel` which is only used in `createModel` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose.model` which is only used in `mongoose",
+      "reference": "The more correct way of doing this is `this.s.hooks.clone()` but either way works. Thanks for finding this :+1:"
+    },
+    {
+      "prediction": "nit: I think we can remove the `selfNS` parameter and just use `selfNS == \"\"` in the `if` statement below.",
+      "reference": "Can we update bool to an enum instead of a string? This helps future extensibility"
+    },
+    {
+      "prediction": "Align the elements of a hash literal if they span more than one line.<br>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": "This id is not used anywhere, can we remove it?",
+      "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": "Is this change intentional? If so, can you explain why?",
+      "reference": "Why was this assert removed? Shouldn't we check if the file is in the zip?"
+    },
+    {
+      "prediction": "I think this should be `if _, ok := spc.Labels[string(apis.OpenEBSUpgradeKey)]; ok && spc.Labels[string(apis.OpenEBSUpgradeKey)] == \"true\"`",
+      "reference": "This logic should be handled at informer handle functions i.e. AddFunc, UpdateFunc, DeleteFunc"
+    },
+    {
+      "prediction": "I think this should be `return apisv1alpha1.PoolType(poolSpec.PoolConfig.DefaultRaidGroupType) == apisv1alpha1.PoolStriped || apisv1alpha1.PoolType(poolSpec.RaidGroups[0].Type) == apisv1alpha1.PoolStriped`",
+      "reference": "S1008: should use 'return <expr>' instead of 'if <expr> { return <bool> }; return <bool>' (from `gosimple`)"
+    },
+    {
+      "prediction": "I think this should be `const defaultTitle = \"foo: bar\"`",
+      "reference": "Just to remove branching, use `defaultTitle` explicitly in the test cases. (Is this gofmt'd?)"
+    },
+    {
+      "prediction": "I think we can remove the `SampleType.IN_CODE` argument here.",
+      "reference": "I don't think you use this variable anywhere. (You call `SampleTransformer.newBuilder()` below)"
+    },
+    {
+      "prediction": "(style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use tabs where possible (style) code indent should use",
+      "reference": "The digest is in binary and variable size (the digest size is result if > 0). So we can't use xxstrdup."
+    },
+    {
+      "prediction": "I think it would be better to have `addEvent(String name, Map<String, AttributeValue> attributes)` and `addEvent(String name, long timestamp, Map<String, AttributeValue> attributes)`",
+      "reference": "We don't need this additional method."
+    },
+    {
+      "prediction": "I don't think we need to add this to the ClientConfig interface.",
+      "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 don't think this is the right thing to do. The default QoS should not be used in this case.",
+      "reference": "Why not using the constant `PARTICIPANT_QOS_DEFAULT` here?"
+    },
+    {
+      "prediction": "I think this should be `tm_implementation` to be consistent with the rest of the codebase.",
+      "reference": "Shouldn't this be `tm_cpp` to match `temporalImp`? (There are multiple CPP implementations, so 'cpp' is ambiguous.)"
+    },
+    {
+      "prediction": "nitpick: I think it would be better to use `File::PATH_SEPARATOR` instead of `File::SEPARATOR` here.",
+      "reference": "Need to remove this empty line for rubocop to be happy."
+    },
+    {
+      "prediction": "Why are these commented out?",
+      "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 this should be `else if (os.platform() === 'win32' && mongoClient.s.options.autoEncryption.cryptdConnectionString == null)`",
+      "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 should be called in the child thread, not the parent thread.",
+      "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 think it would be better to have a default value of `SpanIdGenerator.ThreadInstance.CreateNew()` instead of `null`",
+      "reference": "Under what circumstance would we have a span id already? Is this for testing purposes?"
+    },
+    {
+      "prediction": "I'm not sure this is correct. The semaphore could have been signaled by another queue, and then waited on by this queue.",
+      "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 `lookup failed: context canceled`",
+      "reference": "just clarifying the error messages a bit."
+    },
+    {
+      "prediction": "I think it would be better to have a `TelemetryRequestType` enum instead of a `TelemetryRequestTypes` static class.",
+      "reference": "Have those (and other data like conf...) shared across tracers?"
+    },
+    {
+      "prediction": "This file is empty, can we remove it?",
+      "reference": "could we add this file only when we need one?"
+    },
+    {
+      "prediction": "I think this should be `ACMEDNS01ResolvConfPath` to be consistent with the other fields.",
+      "reference": "This may not even be needed."
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think we want to include these in the build hash, but not in the runtime hash. The reason is that we want to be able to cache the runtime hash, but we don't want to cache the build hash. If we don't include these in the build hash, then we won't be able to cache the build hash, and we won't be able to cache the runtime hash. I think we want to include these in the build hash, but not in the runtime hash. The reason is that we want to be able to cache the runtime hash, but we don't want to cache the build hash. If we don't include these in the build hash, then we won't be able to cache the build hash, and we won't be able to cache the runtime hash. I think we want to include these in the build hash, but not in the runtime hash. The reason is that we want to be able to cache the runtime hash, but we don't want to cache the build hash. If we don't include these in the build hash, then we won't be able to cache the build hash, and we won't be able to cache the runtime hash.",
+      "reference": "Capitalisation seems inconsistent here?"
+    },
+    {
+      "prediction": "I think this should be `ILogger<T>` where `T` is the type of the object that is being logged.",
+      "reference": "maybe better Error(string text) and Error(Exception ex) separately?"
+    },
+    {
+      "prediction": "I think this should be `lsp` 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": "This should be `if (target && target->getZone() == ZONE_PVP) { return; }`",
+      "reference": "Trailing tab, remove it in another PR."
+    },
+    {
+      "prediction": "I think this should be `w.tlMgr.signalFatalProblem(w.tlMgr.taskQueueID, err)`",
+      "reference": "what about merging the error checking logic to within `errShouldUnload`?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right place to put this, but I think it would be good to add a `published?` scope to the `Article` model and use that instead of `by_published` here.",
+      "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 don't think this is the right place to put this check. I think it should be in the `pca::run` function. The reason for this is that the `pca::run` function is the only place where we have access to the `policy` object.",
+      "reference": "Should it really be done on test side?"
+    },
+    {
+      "prediction": "I'm not sure I understand this. Why do we need to remove the execution credentials in cleanupTask?",
+      "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 we should pass args here, as it is used in the flow exporter config.",
+      "reference": "is there a reason why this is not called from inside `validate`?"
+    },
+    {
+      "prediction": "I think we can remove the `rwTimeout` from the `clientServer` struct and pass it to the `connect` method.",
+      "reference": ":+1: I much prefer this style for functions with more than a few arguments."
+    },
+    {
+      "prediction": "I don't think this change is necessary.",
+      "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 removed?",
+      "reference": "Why did you remove this line?"
+    },
+    {
+      "prediction": "I think this should be \"Invalid column indices\" or \"Invalid indices for columns\" or something like that.",
+      "reference": "Very unclear error message. Remember that these are messages that your users see. Please be more specific"
+    },
+    {
+      "prediction": "I don't think we need to hardcode the OS and architecture here. We can get it from the VMSS spec.",
+      "reference": "Can we add these when we need them"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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 thing to do here, but it seems to work for me.",
+      "reference": "Would it make sense to log a debug warning here?"
+    },
+    {
+      "prediction": "I don't think we should remove this check. If the address is not in the keystore, we should tell the user and not proceed.",
+      "reference": "\"in the keystore\", but nevermind"
+    },
+    {
+      "prediction": "I think this should be `if (list->empty () || node.peers.total_weight () < node.config.online_weight_minimum.number ())`",
+      "reference": "This should be done in a separate PR."
+    },
+    {
+      "prediction": "I don't think this is correct. The `Text.standardize` method is used to remove accents and other special characters from the string. If you want to ignore case, you should use the `String.equalsIgnoreCase` method instead of `String.equals`.",
+      "reference": "Rather than this maybe `Text.sanitize` would be better here"
+    },
+    {
+      "prediction": "Why is this false? Is it because we don't know if the node is a validator or not? If that's the case, I think it would be better to have a `IsValidator` property in the `Node` class and set it to `false` by default.",
+      "reference": "lot of places setting default 'false' to static value - noise in PR"
+    },
+    {
+      "prediction": "I don't think we need to store the original image width and height in the loop result.",
+      "reference": "I think you should use `$event->getOriginalFileUrl()` instead of `$sourceFilePath` here."
+    },
+    {
+      "prediction": "nit: I think it's better to use `api_quota_configs = notifier_configs.get('api_quota', {})` so that we don't have to check if `api_quota_configs` is `None` later.",
+      "reference": "nit: It's a matter of taste, but it would be tighter if this is called `api_quota`."
+    },
+    {
+      "prediction": "I think we can remove the import for `GasCalculator` as it's not used anymore.",
+      "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 think this should be `iteritems()` instead of `items()`. `items()` returns a list of tuples, while `iteritems()` returns an iterator over tuples.",
+      "reference": "doubt this matters..."
+    },
+    {
+      "prediction": "I don't think we should import `scapy.consts` in `scapy.all`.",
+      "reference": "Gets updated versions of `LOOPBACK_INTERFACE`, `LOOPBACK_NAME` when importing scapy."
+    },
+    {
+      "prediction": "I think we should use `spec.isUnpartitioned()` instead of `Objects.equal(spec, 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": "nit: I think it's better to have `workflowDeleteManager` as a field of `historyEngineImpl` instead of passing it to `newTransferQueueTaskExecutorBase`.",
+      "reference": "looks like it's not used? Do we plan to use it in the future?"
+    },
+    {
+      "prediction": "I think we should keep the `Config` name here.",
+      "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 don't think you need to import `http_error` here.",
+      "reference": "nitpick: superfluous parenthesis"
+    },
+    {
+      "prediction": "I think this should be `span.SetMetric(Tags.Measured, true);`",
+      "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": "Why do we need to clear the stateDB here?",
+      "reference": "The position change of this line may be a hard fork."
+    },
+    {
+      "prediction": "I don't think this is a good idea. I think it should be `if (id > 0 && id < 30000) {`",
+      "reference": "id is unsigned"
+    },
+    {
+      "prediction": "I don't think we need to add this to the docstring.",
+      "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": "nit: can you add a blank line between the parameters?",
+      "reference": "Looks like this didn't need to change."
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the `__init__` method of the `config_parser` class.",
+      "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": "This should be `int64` instead of `int`.",
+      "reference": "I think @sergeybykov converted (wanted to convert) all `shardID` to `int32` from `int64`. Please check with him."
+    },
+    {
+      "prediction": "I think this should be `std::max<int32_t>(creature->getMaxHealth(), 1)` instead of `std::max(creature->getMaxHealth(), 1)`",
+      "reference": "This will only fix client side (so showing it in-game) server still gonna hold invalid values I guess?"
+    },
+    {
+      "prediction": "This should be `msg.format(class_name=class_name, extras=extras, apply_args=apply_args)`",
+      "reference": "Shouldn't this be `cls.param.warning`?"
+    },
+    {
+      "prediction": "I'm not sure why this was removed, but I think it should be kept.",
+      "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 `address[0] in forbidden_hosts`.",
+      "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 this should be `protected` instead of `public`.",
+      "reference": "Why not just access `ops` directly like before?"
+    },
+    {
+      "prediction": "Why is this removed?",
+      "reference": "Is there even a need to keep the bool?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right way to do this, but it seems to be the only way to get this to work on Windows.",
+      "reference": "wrap to stay under 80 characters"
+    },
+    {
+      "prediction": "I don't think this is necessary.",
+      "reference": "`return NULL` is more clear than `throw` at least in my reading, and consistent with #3362."
+    },
+    {
+      "prediction": "Why do we need to remove this?",
+      "reference": "Both of these are important. Without it, locust won't find my libraries in the cloud."
+    },
+    {
+      "prediction": "`ListWorkloads` should be renamed to `ListServices` to be consistent with the rest of the codebase.",
+      "reference": "This should remain as `ListServices`"
+    },
+    {
+      "prediction": "I'm not sure if we need this package at all. I think we can just use `v2` as the package name in the `v2` directory.",
+      "reference": "If no tests, remove this file."
+    },
+    {
+      "prediction": "I think this should be `gsuite-service-account-key-file` instead of `groups-service-account-key-file`.",
+      "reference": "This will require changes to the docs. I suggest searching the gh-pages branch for the previous variable name."
+    },
+    {
+      "prediction": "This should be os.path.join(globalVars.appDir, \"lib\", buildVersion.version, \"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 `EnableInMem` to be consistent with the rest of the config.",
+      "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. [117/80]",
+      "reference": "I'll need to pull this out of here since we're sharing this in other tests. Same for `procurement_spec.rb`"
+    },
+    {
+      "prediction": "Metrics/LineLength: Line is too long. [83/80]",
+      "reference": "what about `observer` as a var name here (and below) rather than `obs` ? -- would be clearer, imo!"
+    },
+    {
+      "prediction": "I'm not sure this is the right thing to do. I think it should be `except KeyError as e:`",
+      "reference": "could you change this to instead of being a `try.. except` be `message = message.get('message')`"
+    },
+    {
+      "prediction": "Why is this change needed?",
+      "reference": "Add space between License and nameapace"
+    },
+    {
+      "prediction": "Style/SymbolArray: Use %i or %I for an array of symbols.",
+      "reference": "should we have a `before_filter` for auth here?"
+    },
+    {
+      "prediction": "nit: I think it would be good to add a log message here as well.",
+      "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": "Metrics/LineLength: Line is too long. [101/80]",
+      "reference": "Can we use record.update here?"
+    },
+    {
+      "prediction": "I think this should be `rw.response.Arg3Writer` instead of `rw.response.Arg2Writer`.",
+      "reference": "Shouldn't the exact case option matter here?"
+    },
+    {
+      "prediction": "`luigi.uti.Derived` -> `luigi.util.Derived`",
+      "reference": "Errr, why haven't we noticed this until now?"
+    },
+    {
+      "prediction": "I don't think we need to throw an exception here. We can just log a warning and ignore the previous value.",
+      "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": "This should be 1100000.",
+      "reference": "Please only increment by 1."
+    },
+    {
+      "prediction": "I think this should be `org.joda.time.Duration` instead of `org.joda.time.Period`.",
+      "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 think this should be `preventWheelOnOverlays` or something like that. `preventWheel` sounds like it's going to prevent all wheel events, not just on overlays.",
+      "reference": "Please check if is it possible to mark it as private. What's more please add this to the TypeScript definition file."
+    },
+    {
+      "prediction": "I think this should be `std::toupper` instead of `::toupper`.",
+      "reference": "Maybe move this to `util.h` ?"
+    },
+    {
+      "prediction": "This should be `#include <flatbuffers/grpc.h>`",
+      "reference": "oops. this won't work will it"
+    },
+    {
+      "prediction": "I don't think this is right. If the handle is a single team, then the TLF ID must be a single team. If the handle is a single user, then the TLF ID must be a single user. If the handle is a mixed team, then the TLF ID must be a mixed team. If the handle is a mixed user, then the TLF ID must be a mixed user. If the handle is a mixed team/user, then the TLF ID must be a mixed team/user. If the handle is a mixed user/team, then the TLF ID must be a mixed user/team. If the handle is a mixed team/user/team, then the TLF ID must be a mixed team/user/team. If the handle is a mixed user/team/user, then the TLF ID must be a mixed user/team/user. If the handle is a mixed team/user/team/user, then the TLF ID must be a mixed team/user/team/user. If the handle is a mixed user/team/user/team, then the TLF ID must be a mixed user/team/user/team. If the handle is a mixed team/user/team/user/team, then the TLF ID must be a mixed team/user/team/user/team. If the handle is a mixed user",
+      "reference": "@strib I assumed this is what you meant; but let me know if I'm wrong!"
+    },
+    {
+      "prediction": "I think it should be `command` instead of `decision`.",
+      "reference": "I believe the error has changed"
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this. I think it would be better to have a `#constraints` method that returns an array of constraints. Then you can check if the array is empty or not.",
+      "reference": "Lint/UnneededDisable: Unnecessary disabling of Naming/PredicateName."
+    },
+    {
+      "prediction": "Line is too long. [84/80]",
+      "reference": "Should this be removed now?"
+    },
+    {
+      "prediction": "I think this should be called `streamingType`.",
+      "reference": "streamHandle sounds vague to me. Can you be more specific?"
+    },
+    {
+      "prediction": "I'm not sure if this is the right thing to do, but I think we need to merge the `cursorState` options with the `cursorOptions` because the `cursorState` contains the `batchSize` and the `cursorOptions` contains the `maxAwaitTimeMS`.",
+      "reference": "Does everything on cursorState belong in the command options?"
+    },
+    {
+      "prediction": "I think this should be `for d.NextArg() || d.NextBlock(0) || d.NextBlock(1) {`",
+      "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": "nit: I think you can use `bucket` instead of `bucketRegex` in the `fmt.Sprintf` above.",
+      "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 `ModelToViewTransformer<ApiModel>`",
+      "reference": "I think `ApiModelT` would be clearer than `T`"
+    },
+    {
+      "prediction": "I don't think you need to update the year here.",
+      "reference": "Shouldn't this be 2019?"
+    },
+    {
+      "prediction": "Can we rename this to `CONTAINERIZED_FLOW_FAILED` to make it more generic?",
+      "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": "Shouldn't this be `nodeName !== 'DT' && nodeName !== 'DD' && !role`?",
+      "reference": "This should allow `role=definition` and `role=term`, possibly also `role=list`?"
+    },
+    {
+      "prediction": "I think we should use `Boolean.valueOf(value)` instead of `Boolean.parseBoolean(value)`. `Boolean.parseBoolean(value)` returns `true` if `value` is not `null` and `Boolean.valueOf(value)` returns `true` if `value` is not `null` and `value.equalsIgnoreCase(\"true\")`.",
+      "reference": "Should this use `boolean` instead of `Boolean`?"
+    },
+    {
+      "prediction": "Why is this test case removed?",
+      "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 don't think this should be a public method.",
+      "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": "nit: `blogSvidPath` -> `blogCertPath`",
+      "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": "I think it's better to use `reader.Value.ToString()` instead of `reader.Value?.ToString()`. `reader.Value` can be `null` only if `reader.TokenType` is `JsonToken.Null`, and in this case `reader.Value.ToString()` will return `\"null\"`, which is not what we want.",
+      "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 if this is the best way to do it, but it seems to be the only way to make it work.",
+      "reference": "Shouldn't this `TEXT` also be dependent on the platform, then?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do this, but it seems to work.",
+      "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": "Why do we need to include tbb/task.h here? It is not used in this file.",
+      "reference": "If tbb.h is included, there is no big sense to include other TBB headers."
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it seems to work.",
+      "reference": "Let's use the same `S3_HOSTNAME_PATTERN` as above here."
+    },
+    {
+      "prediction": "This should be removed.",
+      "reference": "This shouldn't be here."
+    },
+    {
+      "prediction": "I'm not sure this is the best way to do this, but it works for now.",
+      "reference": "A `case` statement may be simpler here."
+    },
+    {
+      "prediction": "I think this should be `curTagCount + newTagCount > domainTagsLimit - remainingTagsToInsert`",
+      "reference": "why are we adding the current tag count to new tag count ?"
+    },
+    {
+      "prediction": "I don't think this is correct. The `dh` option should not be set to `none` by default. It should be set to `none` only if the user explicitly sets the `dh` option to `none`. If the user does not set the `dh` option to `none`, then the `dh` option should not be set to `none` by default.",
+      "reference": "Are you writing \"none\" to file content? no good"
+    },
+    {
+      "prediction": "I don't think we need to change the constructor. We can keep it as it is and just add the setters.",
+      "reference": "Should we allow `string|string[]` here?"
+    },
+    {
+      "prediction": "I'm not sure if this is the best way to do it, but it works for now.",
+      "reference": "Is there a reason for logging `error.message` specifically, maybe log the whole error object?"
+    },
+    {
+      "prediction": "I think we need to add this to `AspNetCoreInstrumentation` as well.",
+      "reference": "Why is this needed?"
+    },
+    {
+      "prediction": "I don't think we need to do anything here. The container is already running, so we don't need to attach to it.",
+      "reference": "Nothing. We only expect container-connected events when going via the plugin, not the proxy."
+    },
+    {
+      "prediction": "nit: `let rootId = 0;` -> `let rootId = 1;`",
+      "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 think this is a bug in the test. I think it should be 2.",
+      "reference": "this seems a bit wrong no? why is this change needed?"
+    },
+    {
+      "prediction": "I think this should be `less than or equal to`",
+      "reference": "Still confusing. Why don't we return something like `'Fee too low: expected X but found Y'`?"
+    },
+    {
+      "prediction": "I think this should be `json:\"-\"` so that it doesn't show up in the JSON output.",
+      "reference": "Should be from lowercase `json:\"ports\"` the same is defined in MarshalJSON"
+    },
+    {
+      "prediction": "This should be `sns-ruby-example-create-subscription.rb demonstrates how to create an Amazon Simple Notification Services (SNS) subscription using the AWS SDK for Ruby.`",
+      "reference": "Simple Notification **Service** (singular)"
+    },
+    {
+      "prediction": "I don't think this is the right place to do this. I think it should be done in the constructor of `NashornEngine`.",
+      "reference": "This code should be executed in NashornEngineFactory"
+    },
+    {
+      "prediction": "I'm not sure this is the right place to seed the random number generator.",
+      "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 `false` for the next major release.",
+      "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