Bug 58690 – new-run-webkit-tests: save stderr into the layout-test-results dir

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Dirk Pranke]

new-run-webkit-tests: save stderr into the layout-test-results dir

[Dirk Pranke]

Created attachment 89846 [details]
Patch

[Dirk Pranke]

Splitting the saving of the error text out from the display of it (which I'll let Ojan fix in the original bug).

[David Levin]

Comment on attachment 89846 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=89846&action=review

> Tools/Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:341
>          tests_run = get_tests_run(['failures/expected/keybaord.html'], tests_included=True, flatten_batches=True)

btw, typo: keybaord

> Tools/Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:348
> +                          'stuff going to stderr')

What makes the file contain "stuff going to stderr"?

[Dirk Pranke]

(In reply to comment #3)
> (From update of attachment 89846 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=89846&action=review
> 
> > Tools/Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:341
> >          tests_run = get_tests_run(['failures/expected/keybaord.html'], tests_included=True, flatten_batches=True)
> 
> btw, typo: keybaord
>

Interesting. I'll fix that in a different bug.
 
> > Tools/Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:348
> > +                          'stuff going to stderr')
> 
> What makes the file contain "stuff going to stderr"?

That is done at port/test.py:147, which has custom logic to fake output going to stderr for that particular test. That code has been there forever.

[Ojan Vafai]

Comment on attachment 89846 [details]
Patch

Does this mean we're writing stderr into the test and into the stderr.txt? We should only be doing the latter. Fine with that being part of this patch or a separate patch though.

[Dirk Pranke]

H(In reply to comment #5)
> (From update of attachment 89846 [details])
> Does this mean we're writing stderr into the test and into the stderr.txt? We should only be doing the latter. Fine with that being part of this patch or a separate patch though.

Hm. Good point, I had forgotten that the chromium ports were doing that. I'll fix that in a separate patch.

[Dirk Pranke]

Committed r84112: <http://trac.webkit.org/changeset/84112>

[Dirk Pranke]

grr. I don't know why I spaced and committed this yesterday w/o updating the baselines.

This was rolled back in r84127 / bug 58690.

[Dirk Pranke]

double-grr: comment #8 applies to bug 58708, not this bug. This patch was fine and didn't need to be rolled out.