webkitpy: Remove --run-chunk and --run-part.

webkitpy: Remove --run-chunk and --run-part.

LayoutTests supports '--shard-index' and '--total-shards' from gtest
API (and the environment variables) which work in less surpising manner.

Chunking previously would pad the last chunk out with tests from the
first chunk. This causes some tests to end up being run *twice* which
prevents clean merging of the output results.

BUG=524758
Review-Url: https://codereview.chromium.org/2686303003
Cr-Commit-Position: refs/heads/master@{#450045}
Committed: https://chromium.googlesource.com/chromium/src/+/e6219e6d4d7a63635d3c451bb36a90d82f4c17a9

[mithro, 2017-02-10 23:55:54]

Description was changed from

==========
webkitpy: Remove chunk padding.

Chunking previously would pad the last chunk out with tests from the
first chunk. This causes some tests to end up being run *twice* which
prevents clean merging of the output results.

BUG=524758
==========

to

==========
webkitpy: Remove chunk padding which causes duplicate test runs.

Chunking previously would pad the last chunk out with tests from the
first chunk. This causes some tests to end up being run *twice* which
prevents clean merging of the output results.

BUG=524758
==========

[mithro, 2017-02-10 23:55:54]

tansell@chromium.org changed reviewers:
+ jeffcarp@chromium.org, qyearsley@chromium.org

[mithro, 2017-02-10 23:56:09]

The CQ bit was checked by tansell@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-02-10 23:56:46]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-11 00:59:38]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-02-11 00:59:38]

Dry run: Try jobs failed on following builders:
  win_chromium_rel_ng on master.tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_...)

[mithro, 2017-02-11 02:59:09]

Description was changed from

==========
webkitpy: Remove chunk padding which causes duplicate test runs.

Chunking previously would pad the last chunk out with tests from the
first chunk. This causes some tests to end up being run *twice* which
prevents clean merging of the output results.

BUG=524758
==========

to

==========
webkitpy: Remove --run-chunk and --run-part.

LayoutTests supports '--shard-index' and '--total-shards' from gtest
API (and the environment variables) which work in less surpising manner.

Chunking previously would pad the last chunk out with tests from the
first chunk. This causes some tests to end up being run *twice* which
prevents clean merging of the output results.

BUG=524758
==========

[mithro, 2017-02-11 02:59:09]

tansell@chromium.org changed reviewers:
+ dpranke@chromium.org

[mithro, 2017-02-11 03:00:01]

The CQ bit was checked by tansell@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-02-11 03:00:21]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[jeffcarp, 2017-02-11 03:17:18]

lgtm

[Dirk Pranke, 2017-02-11 03:41:31]

lgtm as-is, though see my comment for the suggestion that maybe it'd be better
to mirror gtests, assuming my understanding of what gtests do is correct.

https://codereview.chromium.org/2686303003/diff/20001/third_party/WebKit/Tool...
File
third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/controllers/layout_test_finder.py
(right):

https://codereview.chromium.org/2686303003/diff/20001/third_party/WebKit/Tool...
third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/controllers/layout_test_finder.py:195:
def _split_into_chunks(test_names, index, count):
I believe gtests do the rough equivalent of:

    def _split_into_chunks(test_names, index, count):
        tests = [t for t, cur_index
                 in enumerate(test_names)
                 if cur_index % count == index] 
        return tests, list(set(test_names) - set(tests))

Perhaps we should just do that instead? It could expose yet more flakiness, but
it seems like it'd be more consistent.

[commit-bot: I haz the power, 2017-02-11 04:23:46]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-02-11 04:23:47]

Dry run: This issue passed the CQ dry run.

[mithro, 2017-02-13 18:27:04]

https://codereview.chromium.org/2686303003/diff/20001/third_party/WebKit/Tool...
File
third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/controllers/layout_test_finder.py
(right):

https://codereview.chromium.org/2686303003/diff/20001/third_party/WebKit/Tool...
third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/controllers/layout_test_finder.py:195:
def _split_into_chunks(test_names, index, count):
On 2017/02/11 03:41:31, Dirk Pranke wrote:
> I believe gtests do the rough equivalent of:
> 
>     def _split_into_chunks(test_names, index, count):
>         tests = [t for t, cur_index
>                  in enumerate(test_names)
>                  if cur_index % count == index] 
>         return tests, list(set(test_names) - set(tests))
> 
> Perhaps we should just do that instead? It could expose yet more flakiness,
but
> it seems like it'd be more consistent.

Your correct that gtest actually does something like you suggest, however it
doesn't *guarantee* any sharding method, so I think we should do our method for
now.

When running on swarming we will be shuffling the tests anyway.

[mithro, 2017-02-13 18:27:19]

The CQ bit was checked by tansell@chromium.org

[commit-bot: I haz the power, 2017-02-13 18:28:01]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-13 19:44:56]

CQ is committing da patch.

Bot data: {"patchset_id": 20001, "attempt_start_ts": 1487010439056190,
"parent_rev": "0731bc293ca970d89fda47bb325185fe0b4aab20", "commit_rev":
"e6219e6d4d7a63635d3c451bb36a90d82f4c17a9"}

[commit-bot: I haz the power, 2017-02-13 19:45:56]

Description was changed from

==========
webkitpy: Remove --run-chunk and --run-part.

LayoutTests supports '--shard-index' and '--total-shards' from gtest
API (and the environment variables) which work in less surpising manner.

Chunking previously would pad the last chunk out with tests from the
first chunk. This causes some tests to end up being run *twice* which
prevents clean merging of the output results.

BUG=524758
==========

to

==========
webkitpy: Remove --run-chunk and --run-part.

LayoutTests supports '--shard-index' and '--total-shards' from gtest
API (and the environment variables) which work in less surpising manner.

Chunking previously would pad the last chunk out with tests from the
first chunk. This causes some tests to end up being run *twice* which
prevents clean merging of the output results.

BUG=524758

Review-Url: https://codereview.chromium.org/2686303003
Cr-Commit-Position: refs/heads/master@{#450045}
Committed:
https://chromium.googlesource.com/chromium/src/+/e6219e6d4d7a63635d3c451bb36a...
==========

[commit-bot: I haz the power, 2017-02-13 19:46:00]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/chromium/src/+/e6219e6d4d7a63635d3c451bb36a...