This adds O(reviewers+cc) RPC to Gerrit before upload,
though done concurrently in up to 10 threads. This information isn't
used yet, but will be in follow up CLs.
R=ehmaldonado
Bug: 877717
Change-Id: I2374f249ee874a71089244309e50e8e88a3dee7d
Reviewed-on: https://chromium-review.googlesource.com/c/1242847
Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
Commit-Queue: Andrii Shyshkalov <tandrii@chromium.org>
Print the message asking the user to upload gerrit logs for git-cl upload
failures after the exception.
Also, initialize the error logger only for git-cl upload, so that the
headers are not overwritten unless git-cl upload is executed again.
Bug: 881860
Change-Id: I8e63432eecca5253e56096e3ad0218eec4058e12
Reviewed-on: https://chromium-review.googlesource.com/c/1277426
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
And ask the user to report the printed headers to the bug.
Bug: 881860
Change-Id: I0a6cc0d90b4a721f72cd7eee4cba59934facbb1e
Reviewed-on: https://chromium-review.googlesource.com/c/1266837
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
This CL basically replaces "skip if is_gce" with "skip if default auth method
is not gitcookies".
This presumable makes 'git cl' work on LUCI in a consistent manner. Before, it
worked only if the LUCI bot happened to also be GCE bot.
R=tandrii@chromium.org
BUG=891755
Change-Id: I2caa219a4082438a5e026e728bfb62f46a0c80fd
Reviewed-on: https://chromium-review.googlesource.com/c/1260053
Commit-Queue: Vadim Shtayura <vadimsh@chromium.org>
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
Parses git diff for changed python files in a method similar to
clang-diff and feeds the resulting line ranges into yapf.
Also sets the default style to the yapf file included in depot tools
by searching parent directories of each changed file to find a yapf
style config (.style.yapf). If none is found the default style
file will be the chromium .style.yapf included in depot tools.
Note: Even if line ranges are specified, yapf will fix indentation
issues for the entire file.
This is intended see https://github.com/google/yapf/issues/499
This may cause some issues if git cl format is run on a file
with lots of indentation issues or on a file or when run on a
third_party file that is formatted with pep8 and does not include
a .style.yapf and may make many more changes then the user expects.
Still undecided on whether this should be turned on by default but
if not I think the non --full support is a positive change anyways.
Bug:846432
Change-Id: Ib85797f4a8e1021870901ff465ec10f7e70deb87
Reviewed-on: https://chromium-review.googlesource.com/1249642
Commit-Queue: Aiden Benner <abenner@google.com>
Reviewed-by: Nodir Turakulov <nodir@chromium.org>
Reviewed-by: agrieve <agrieve@chromium.org>
This removes 1 RPC when tbr-ing second or later patchset.
Since tests cover TBR only in first upload, there are
no test changes. Since adding test is real PITA here,
keep them as is.
R=ehmaldonado
Change-Id: I90801b06727ce2131a703d6cd6c24ffad798bc8a
Reviewed-on: https://chromium-review.googlesource.com/1242212
Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
Commit-Queue: Andrii Shyshkalov <tandrii@chromium.org>
Print "git cl comments" json to stdout when the output file is "-"
Change-Id: I91021317e52019a0d499cccc4f40f062127b300e
Reviewed-on: https://chromium-review.googlesource.com/1228094
Auto-Submit: Leszek Swirski <leszeks@chromium.org>
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>
Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
Running git cl issue -r in a tree with dozens/hundreds of branches
could be a bit slow since the code parsed .git/config twice per
local branch. By parsing it only once and then looking things up
in a hashtable it becomes much faster. (For my local tree with 700
branches 13 seconds became 0.3 seconds)
Bug: 880734
Change-Id: I67f45de32fb7f2cc5960174e59f3476ef3021a3a
Reviewed-on: https://chromium-review.googlesource.com/1206352
Commit-Queue: Andrii Shyshkalov <tandrii@chromium.org>
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
`git cl split` currently runs a cq dry run for every uploaded CL. This
has overloaded our infrastructure a few times in the past. This CL changes
the command to not dry run by default, and adds a flag to enable this. The
flag has a warning about doing this, and tells the user to email
infra-dev@chromium.org if they're going to be generating >~10 CLs.
Bug: 878117
Change-Id: Ic865c09b188b8d4f202785f4763f7b7b8910c9cf
Reviewed-on: https://chromium-review.googlesource.com/1191927
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
Commit-Queue: Stephen Martinis <martiniss@chromium.org>
This fixes Gerrit project detection based on remote URL,
accounting for potential 'a/' prefix in the URL path component,
which isn't part of the Gerrit project name.
R=ehmaldonado, mmoss
Bug: 876964, 876910
Change-Id: I473ae8c6c9e0f2034b350901abd67db151e0a3d3
Reviewed-on: https://chromium-review.googlesource.com/1187573
Reviewed-by: Michael Moss <mmoss@chromium.org>
Commit-Queue: Andrii Shyshkalov <tandrii@chromium.org>
Sadly, this makes maintaining one test handling covering
exceptional circumstance very complex, so it was removed.
I've decoupled test that ensures that GetRemoteUrl works from
upload codepath.
R=ehmaldonado@chromium.org
Bug: 876910
Change-Id: I39de410c72d893e73492d5c3fc8f60a6ebc4f11f
Reviewed-on: https://chromium-review.googlesource.com/1186142
Reviewed-by: Ryan Tseng <hinoka@chromium.org>
Commit-Queue: Andrii Shyshkalov <tandrii@chromium.org>
As it is right now, when a function is collecting metrics it prints
the notice and calls sys.exit() to ensure the notice is the last thing
that is printed.
git-cl split has to call git-cl upload multiple times, but once it has
been called once, it exits, so only the first cl is uploaded.
This separates metrics collection from notice printing, so that the
function that is collecting metrics behaves like a function that isn't.
It also makes sure we don't collect metrics for multiple functions at
the same time.
Bug: 868280
Change-Id: Ic58ebe7d19e09ed85fa8b0af76dcbf608ee4c9bc
Reviewed-on: https://chromium-review.googlesource.com/1153503
Reviewed-by: Aaron Gable <agable@chromium.org>
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>
Adds '.ts' to list of extensions to be handled by clang-format.
Change-Id: Ia00e59ba3ab617ebf248cfcf6df07dd98cdc14ce
Reviewed-on: https://chromium-review.googlesource.com/1120325
Reviewed-by: Ryan Tseng <hinoka@chromium.org>
Commit-Queue: Deepanjan Roy <dproy@chromium.org>
If there are no builds in build search response, the response is {}.
Handle this case.
R=qyearsley
Bug: 849361
Change-Id: I9f0472e140698a1d1c5b98b301bc562ac457785b
Reviewed-on: https://chromium-review.googlesource.com/1085899
Reviewed-by: Quinten Yearsley <qyearsley@chromium.org>
Commit-Queue: Nodir Turakulov <nodir@chromium.org>
Make `git cl format --python -full --dry-run` return non-zero exit code if
files are not well formatted.
Make `git cl format --python -full --diff` print diff, rather than new file
contents.
Change-Id: I666d360051563898d3508d776817bf9a61433cb0
Reviewed-on: https://chromium-review.googlesource.com/1081340
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
Commit-Queue: Nodir Turakulov <nodir@chromium.org>
When using non-squashed uploads, there usually isn't an associated
issue with a branch. Thus setting WIP when the issue tag wasn't
present would continually set WIP even on trivial rebases. Not
adding the WIP tag when uploading allows the user to set WIP manually,
while not forcing the WIP flag after every new upload.
Also updates the unit tests that use no-squash.
Bug: 767439
Change-Id: I990a82139fefe1a0c5c7b149d39045cf985539b7
Reviewed-on: https://chromium-review.googlesource.com/1033187
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Aaron Gable <agable@chromium.org>
It is used by CMDland but it was not declared and is causing failures.
TBR=agable@chromium.org
Bug: 832566
Change-Id: Ia8ba5df458f9a39e415b7a9495dc338def0b16a5
Reviewed-on: https://chromium-review.googlesource.com/1013056
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>
Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>
Reviewed-by: Michael Moss <mmoss@chromium.org>
Currently all tests in a PRESUBMIT.py file are run in parallel, but not
all tests across PRESUBMIT.py files.
This introduces a flag that will allow presubmit to run all tests across
PRESUBMIT files in parallel.
Bug: 819774
Change-Id: Idd3046cb3c85e9c28932a9789ba7b207a01d9f99
Reviewed-on: https://chromium-review.googlesource.com/994241
Reviewed-by: Aaron Gable <agable@chromium.org>
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>
Since no one can upload or land changes from Rietveld anymore,
PRESUBMIT and its support files no longer need to be able to
support it.
Bug: 770408
Change-Id: I4ca6391291e7b0755c78af453c3d006ad3666a17
Reviewed-on: https://chromium-review.googlesource.com/991053
Reviewed-by: Robbie Iannucci <iannucci@chromium.org>
Commit-Queue: Aaron Gable <agable@chromium.org>
This reverts commit f3eed0016e.
Reason for revert: Some PRESUBMIT checks still expect
CheckRietveldTryJobExecution to exist.
Original change's description:
> Remove Rietveld support from presubmit
>
> Since no one can upload or land changes from Rietveld anymore,
> PRESUBMIT and its support files no longer need to be able to
> support it.
>
> R=tandrii@chromium.org
>
> Bug: 770408
> Change-Id: I749092b66fdca16d5cef77e8cefc905aa5375b50
> Reviewed-on: https://chromium-review.googlesource.com/693380
> Commit-Queue: Aaron Gable <agable@chromium.org>
> Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
TBR=agable@chromium.org,tandrii@chromium.org
Change-Id: I72e29bd8a9739406f29190adbeb7eb7718ed21cd
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: 770408
Reviewed-on: https://chromium-review.googlesource.com/991012
Commit-Queue: Aaron Gable <agable@chromium.org>
Reviewed-by: Aaron Gable <agable@chromium.org>
git_cl.py uses auth.py to authenticate to cr-buildbucket,
but when it does so, it caches the token under a key for the
current codereview host, not the buildbucket hostname.
This is confusing, but is the desired behavior, so that users
can use different credentials (e.g. chromium and google) when
communicating with buildbucket about CLs on different gerrit
hosts (e.g. chromium and chrome-internal).
To prevent other people from getting confused like I did,
add some documentation to this effect.
Change-Id: If16896d15423bbdecf9624393773ac01d0d16a66
Reviewed-on: https://chromium-review.googlesource.com/990674
Commit-Queue: Aaron Gable <agable@chromium.org>
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
Reviewed-by: Nodir Turakulov <nodir@chromium.org>
Since no one can upload or land changes from Rietveld anymore,
PRESUBMIT and its support files no longer need to be able to
support it.
R=tandrii@chromium.org
Bug: 770408
Change-Id: I749092b66fdca16d5cef77e8cefc905aa5375b50
Reviewed-on: https://chromium-review.googlesource.com/693380
Commit-Queue: Aaron Gable <agable@chromium.org>
Reviewed-by: Andrii Shyshkalov <tandrii@chromium.org>
Over last two weeks of running tryjobs on CLs of others, I found
this very useful. This CL also simplifies code.
R=agable@chromium.org
Change-Id: I5c7ce0d311b1ca024b92227dbec54e5197205c62
Reviewed-on: https://chromium-review.googlesource.com/930742
Reviewed-by: Aaron Gable <agable@chromium.org>
Commit-Queue: Andrii Shyshkalov <tandrii@chromium.org>
Andrii Shyshkalov
Edward Lemur
Vadim Shtayura
Aiden Benner
Sergiy Byelozyorov
Leszek Swirski
Daniel Bratell
Stephen Martinis
erikchen
Deepanjan Roy
Sidney San Martín
Nodir Turakulov
Jamie Madill
Olivier Robin
Aaron Gable
Ravi Mistry
Robert Iannucci