comment stringlengths 24 1.38k | label stringclasses 2
values |
|---|---|
> I think it's cleaner. We do not have to re-implement and maintain different versions start-needed-srv in each script. Just one "vdsm_init_common.sh --start-needed-srv" to rule them all.
Here we disagree.
But we repeat our-selves... and not progressing, so we can stop doing that... and progress... as whatever we hav... | negative |
I think I'll cherry-pick, rebase and add the bootblock.elf and coreboot.ram fixes in #750.. makefile half-baked output. | positive |
I wonder why jetbrains repeats this code in all their bundles. | negative |
Having nightmares about empty lines kidnapping you in an alien ship and performing colonoscopy? | negative |
indent.
But how about defining
skipontest() {
if [ -z "${TESTING_ENV}" ] ; then
"$@"
else:
echo skipping "$@"
}
It should be added to ovirt_store_config (in case the test happens to be run on ovirt-node). | positive |
Since all TNG constants in the header are prefixed with TNG I think it could be nice to just call this variable API_VERSION and then use TNG_API_VERSION in tng_io.cpp and switch the names in version.h.in. | positive |
Thanks a lot for your review.
What do you think about "in the adapted class"?
Anyway, I'm going to bother some of the guru to hear what they think about it... | positive |
OK. I'll check if [1] still works with asiidoctor before pursuing it any further then.
[1] https://gerrit-review.googlesource.com/#/c/49451/ | positive |
if in some places function returns something then it should return always something | positive |
Just a pre-emptive note, I put these TODO's in the code, because they are separate refactoring issues, which would be better handled in a separate bug. | positive |
Because it is used below in g_list_foreach, where gpointer is mandated. | positive |
We probably should ensure the project is clean before doing this, otherwise a user might lose uncommitted changes they care about:
project = Project(
manifest = self.manifest,
name = path,
remote = RemoteSpec('origin'),
gitdir = os.path.join(self.manifest.topdir, path, '.git'),
worktree = os.path... | positive |
Field length expected here is 2, but field below is FT_INT32.
Is this an unsigned short (FT_UINT16) ? | positive |
Can I call GetMatrix w/out having called SetDataFrameObject? Add a bried usage description here. | positive |
@BearND You're probably referring to the idea of returning the normalized location of the face (instead of the absolute coordinates in pixels)... I think we can take care of this in a subsequent patch; and it's not immediately clear whether it will simpli | positive |
if a comment is behind the field it describes, doxygen needs "///<" (with a symbolized left arrow) to realize it correctly | positive |
Done
P.S: I was confused between EBADF and EBADFD. Then I tried to be consistent as I see libgfapi returning EBADFD in several other glfs_* methods when glfs_resolve_fd() fails. | positive |
Hi Bob,
I think the setting of screen resolution is not important as it makes no different when I changed it to smaller or larger size therefore I chose the medium size (1024x768).
It's hard to figure out what date of rendering the map and it does not really affect to the embed code I will remove it from the code. | positive |
How about that one? Is it because you won't be able to automatically change call sites? | positive |
Why checking for null pointer here? delete handles null pointers just fine, you only add overhead here. | negative |
The doxygen documentation will also be a bit better if you add
\related ArrayRef
to these functions (and similarly for the ConstArrayRef ones). | positive |
sounds good. will update once the dependence:
http://review.couchbase.org/#/c/33453/ is merged. | positive |
I agree if the file exist we should skip the conversion and print a information message | positive |
Not all build tools use -j. Thus the simpler approach of directly setting build_cmd="make -j2"/"jom -j4" might be better. | positive |
I'm not clear if "${workspace_loc:gitiles-parent}" mean "$WORKSPACE/gitiles-parent" or if it means "Path to the project named "gitiles-parent".
When I changed the on-disk location (via the rename above) of my gitiles checkout (under my /src/workspaces directory) to /src/workspaces/gitiles-parent then I got to the next... | negative |
I don't see this as needing more prominence here, so much as the increase in docs is *an entire second proposal* wedged into this one. I'd vote for merging as is, and making the docs process a standalone follow-up. | negative |
There is no VDSM action here, only database - just make the command transactive, and save yourself all this hassle. | negative |
This condition is always true. I assume the condition you meant was n != ptrSize. Though I see you removed this condition altogether in CL 9704 (rebase failure?). | positive |
I have been told in the past some compilers don't initialize all elements to 0 with 'foo[n]={0};' - I've never found such a compiler, so maybe it's an urban legend. :)
memset() yup shoulda done that. Ironically I looked for a memset equivalent in glib, so I don't know why I didn't use it. | positive |
Why is this rounding necessary? Just a guess, clock skew? | negative |
I don't think most of these are necessary. The dissector is (and should be) organized to just define functions before they are used. | negative |
Adding to Megumi's nit: Consider initializing the value to true, and possibly collapsing these two methods down to one with an argument | positive |
I was mentioning the blank line. Please don't remove tk.WithExtra. | positive |
More complete would be:
Þöøôðýøõ ÷ðóру÷úø �ÿø�úð óþрþôþò, �þõôøýõýøõ üõôûõýþ | positive |
Patch Set 1:
Alon,
Regardless of discussion we have in the bug #996021
calling .disconnect() when sdk instance is no longer
needed is a good change, i think it can confidently
go in. | positive |
git commit --amend' and everything up until the last Change-Id line can be removed. | positive |
You're preforming this check multiple times.
it's better to create a separate function, in case we will need to add a new "localhost" name/ip | positive |
If non-empty, go into reverse mode where the buildlet dials the coordinator instead of listening for connections. The value is a comma-separated list of modes, e.g. .... | positive |
As remember filtering name can be related to expert information - this give you more filtering options. For example "btle.crc.incorrect" and "Incorrect CRC" | positive |
Can you clarify what is meant by "ListBox => PreNotify() ..."? That doesn't make things any clearer... | negative |
Would it make it easier to override if we return a List? That way if someone wants to add an adapter, they could simply add to the list instead of copying the array.
This method is used in an enhanced for loop, so a list would work just as well as an Array | positive |
there should be
} else {
rv = get_env_plugin_info(pi);
In this case the lcb_cntl user will get correct effective type | positive |
I do not understand... I am not trying to fix the world now...
all I need is to ask vdsm what is its id without having to deal with handling the /etc/vdsm/vdsm.id
are the above valid to this task?
the new code of vdsm-reg will call vdsm-tool vdsm-id --force and get id no matter if it is uuid or generated from anywhe... | negative |
Not sure about this method. It seems to make sense to me the way Dan explained it but I'm not comfortable giving this change a +2 since I didn't write any of the original code and may not be aware of all of the issues. | negative |
Maybe "findSubClasses" would be a better name for these methods? | positive |
You shouldn't check "_doc.length.bodylen == 0 && (iterator->opt & FDB_ITR_NO_DELETES)" again because we already checked it above. | positive |
don't remove this comment here! the require once is obsolete only | negative |
This check was removed along with the corresponding GenFramebuffers and GenRenderbuffers calls because they are redundant(see CheckFBOResources bellow) and the allocated buffers are leaked. This is one of the bugs that my patch fixes. In my patch the buffers are allocated only once(with error checking) in the AllocateF... | positive |
actually, if we get rid of nIndex, we would have to change a lot of code, because in this one, the Insert method returns the index of the inserted/already existing element. | positive |
I don't get it. How are those unused locks got rid of? | negative |
So that means to also change comparisons with STRING_NOTFOUND (which is defined to 0xFFFF), lastIndexOf() returns -1 if not found, so the condition here can be simplified to
if (nPos > 0) | positive |
- Should be like below:
textDirection = SWT.AUTO_TEXT_DIRECTION; | positive |
It is pretty inconsitent. Some have very short descriptions (like this one) others only point to the jdk doc (which is what I will do).
Big TODO update all url pointers. | negative |
Cosmetic: According to last changes related to tvb_captured_length and tvb_reported_length it is good point to start thinking about "reported" case (btsnoop format provide both frame length). In "reported" case we can have only part of frame: for example first 2 bytes. This dissector will fail at this line, so nothing ... | positive |
Yep, I've merged the two into |supportedSignatureAlgorithms|. | positive |
Agree displayable is not ideal. Looking for something that means "not internal". How about getNonInternalFieldModifiers()?
Given they're hidden APIs / internal comments I suggest not wasting any time on this: my point is the comment is not as clear to me as I think it could be.
Feel free to act on the suggestion or i... | positive |
Isn't this a dict_get failed rather than a dict_set? | negative |
how are we guaranteeing that when we invoke this clean up there are no threads consuming these RPCs? | positive |
The size here is now just the size of the struct, hence the names (old/new) will not be a part of the dict.
Further, dict_set_bin, takes in a pointer, that it strores in the xattr/dict and later is freed when this dict is destroyed. In the case of passing a pointer which is a structure containing another pointer, the ... | positive |
Patch Set 3: I would prefer that you didn't submit this
(2 inline comments)
I don't see the need for this change. If you built with the ebuild unmodified, you'd get an image.bin which would be what you want. As far as I could tell looking at the new function, it's identical to the old one except that you're leaving o... | negative |
But what we are using is oslo version. I think this should be fixed in oslo. I submitted a patch to oslo https://review.openstack.org/56211 | positive |
commit.hashCode() should be sufficient. A commit can only have one author. | positive |
No, this is here by purpose to document that this is a static method, as that is not otherwise evident without looking at the header. | negative |
Trying to swap the layout when obj1 == obj2 should just "return 0" (i.e. be no-op). There should also be a check that the two objects have the same uid and gid, and that the UID doing the swap is either owner or has mdd_capable(uc, CAP_FOWNER) since this would ensure the process isn't trying to steal data (i.e. the pr... | positive |
line wrap before throws so the line isn't over 80 columns | positive |
If I understand this right the buildr process is able to load Java classes dynamically and invoke them with reflection? That does it make it slightly easier to do some complex steps. | positive |
I don't think we should let people modify a branch that they don't have access to, even if it is for a commit that they are replacing that used to be in another branch. Locked down is locked down, right?
I see how it might be annoying to have to redo a change if you used the wrong branch the first time, though... Perh... | negative |
Is there a function for this query already? or should we make one for future?
It is just that this is quite a big block of code, and would be more readable factored out | positive |
> So at very least you should only check/parse patch set part, if ref starts with REFS_CHANGES, if at all.
It already does this, startIndex checks for the prefix. | positive |
Just cleaning, this regex remove empty string concatenation: + ("")
This can happen when we start a new group of concatenation (see appendConcatOperation()) and there is nothing to print afterwards. In this case the string ends with a concatenation of an empty string | positive |
provide any output provided by a successful command. | positive |
nitpick: add a space between "if" and its opening pare | positive |
This will cause a double free. The previous GF_REF_PUT() will already call auth_cache_entry_free() (where entry is free'd) if auth_cache_add() has failed. | positive |
It was intended to be Modes from the start, but could not be fixed before without braking BC.
Should probably go as a separate commit. | positive |
That will unfortunately only work if there are no 4-core Intel Xeons that work in dual-socket configuration because we could confuse a workstation that has two such processors with an 8-core Xeon E5 in single-socket configuration. | negative |
My only worry here is that future c or assembly versions of memcmp16 might use x14, x15. I have used a more conservative approach in my tree:
1852 .Ldo_memcmp16:
1853 /* save lr, and return value if comparison equal */
1854 stp x0, x30, [sp, #-16]!
1855
1856 /* set-up args for __memcmp16 */
1857 mov ... | positive |
You cannot design an API for logging. We should take out the "command" argument and have generic errors. It's easy looking at vdsm.log to discover what was the command executed before this failure. | negative |
For some reason I thought that the overflow behavior was unnecessary, now I see that it is used in case of multiple dexes. I reverted the change. | negative |
Because of presence of data-parsoid, this is not a new block level image. Title or test needs editing. | positive |
This could impact the development context when doing a partial compilation, is it noticable?
We should re-use the filename (micro-op). | positive |
The user might still want to disable it...
Unlike the other options, this one does not need a pirq_table.c or mptable.c file be sitting around...
Generally dropping these tables from the build is almost never worth it, since most OSes expect something there. | positive |
Patch Set 1: Looks good to me, but someone else must approve
(2 inline comments)
LGTM with a couple of small optional suggestions. I agree that using more realistic sites to consume memory would be nice, but I think getting something up and running soon would be valuable as well. | positive |
[info] Unable to get class information for @throws tag 'UnableToCompleteException'. | positive |
I can do that in an extra commit, here I only wanted to add the possibility to access the created proto_item. But if you want me to, I can add the change to this commit also. | positive |
This should read:
This survey tests the local metadata performance using the echo_client to drive the MDD layer to perform operations. It is run with multiple threads (to simulate MDT service threads) locally on the MDS node, and does not need Lustre clients in order to run. | positive |
I took out the cast. That was causing the warning I think. | positive |
I did not make comments on all of these before because it was the same comment every time. Unless there is a good documented reason, these should all be set to ignore. | negative |
It probably is, but what happens when somebody upgrades the libvirt? Do we require vdsm restart as well? Or will it suddenly start working? | positive |
Oh, you want it really obvious what type the variable is :) | positive |
If this place is fixed to contain fd->flags, should the other places(anon fd) also be fixed to use fd->flags? | positive |
"""create a vdsm.config.config clone, modified by tunables
tunables is a list of (section, key, val) tuples""" | positive |
For consistency, this could be named "storeInHttpSession". | positive |
Add:
When you copy a manifest object, the new object is a "normal" object -- not a copy of the
manifest. Instead it is a concatenation of all the segment objects. This means that you cannot
copy objects larger than 5GB in size.
All metadata is preserved during the object copy. If you specify metadata ... | positive |
how about use default wikitext namespace (see getDefaultWikitextNS below) or allow specifying namespace as an argument? | positive |
good catch... I should blame autocomplete :P | positive |
please explain what part of mdd_object_create() you need to reuse() (except mdo_create()) | negative |
This is a refactoring from my side that sneaked in when I split the commit. I'll revert this asap since it adds nothing to this change. | positive |
The three tests above that checks namespaces should probably grouped together. They seems to be copy pasted and just change which setting is gathered.
You could uses a dataProvider too. That will build an array of tests cases and pass the case to a test method. Anyway, you should extract the foreach / assert in a new... | positive |
(I think you could indent this line another four spaces if you want.) | positive |
if that's a uintptr_t, then you again don't need this double cast. | negative |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 2