Bug 54790 – webkitpy.common.system.executive_unittest.ExecutiveTest.test_kill_all fails on Chromium Win Release Tests bot

• Status: RESOLVED
• Resolution: WONTFIX
• Component: Tools / Tests

[Dirk Pranke]

Attaching a build log. I don't know why this test is failing; it passes for me locally. It's possible there's some sort of timing difference on the bot and we're triggering a different code path in subprocess.wait().

If we don't have an obvious explanation for this failing, I'm tempted to skip it for now to green up the python tests on win32. Thoughts?

[Dirk Pranke]

disabling this test for now to (hopefully) green up the builder.

[Dirk Pranke]

Created attachment 84162 [details]
Patch

[Dirk Pranke]

anyone have any ideas before I disable it?

[Dirk Pranke]

Committed r80048: <http://trac.webkit.org/changeset/80048>

[WebKit Review Bot]

http://trac.webkit.org/changeset/80048 might have broken Qt Linux Release
The following tests are not passing:
fast/canvas/canvas-strokePath-gradient-shadow.html

[Dirk Pranke]

Clearing ownership now that we've disabled the test, in case someone else wants to figure out why it's failing.

[Roger Fong]

Hey Dirk,
Not sure how much you remember about this problem but it seems to be happening on the Windows port too.

Here is the failure:

[167/1585] webkitpy.common.system.executive_unittest.ExecutiveTest.serial_test_kill_process failed:
  Traceback (most recent call last):
    File "/home/buildbot/slave/win-release-tests/build/Tools/Scripts/webkitpy/common/system/executive_unittest.py", line 167, in serial_test_kill_process
      self.assertEqual(process.wait(), expected_exit_code)
  AssertionError: -6 != -9

First of all it looks like the platform we use is "cygwin" not "win32".
Second -6 is SIGABRT.

It only fails on the bots...

Do you have any ideas? Thanks!

[Eric Seidel (no email)]

I suspect we just need to make the unittest more flexible for this case.  it's very likely the CYGWIN just changed the exit code returned in this kill case.

[Roger Fong]

(In reply to comment #8)
> I suspect we just need to make the unittest more flexible for this case.  it's very likely the CYGWIN just changed the exit code returned in this kill case.

Think it's reasonable for me to just add an extra case for cygwin then?

[Roger Fong]

ALthough it's a bit weird that it's changing it only on the bots and not when I run locally...

[Eric Seidel (no email)]

Seems totally reasonable to me.

[Stephen Chenney]

Test related bugs being marked WontFix. TestExpectations will contain bug if still relevant.