[Webkit-unassigned] [Bug 169916] [GTK] Lots of css3/blending tests fail. Enable support for CSS_COMPOSITING

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Thu Jul 11 17:06:41 PDT 2019


--- Comment #7 from Carlos Alberto Lopez Perez <clopez at igalia.com> ---
(In reply to Carlos Alberto Lopez Perez from comment #6)
> I will check what's going on the bot. Please don't mark them as failing
> still.

The test bot was not fully upgraded to Debian 10. I upgraded it. Now results are a bit better.

But even after that there are some css3/blending tests failing on the bot that pass on my dev machine (running Debian 10 as well).

I think it may be related to some configuration of something that I have, which happens to be different that on the bot. Because I can reproduce what happens on the bot by simply running the tests with an empty home like in:

HOME=$(mktemp -d) Tools/Scripts/run-webkit-tests --gtk css3/blending imported/blink/css3/blending

This happens when I do that:

Regressions: Unexpected text-only failures (5)
  css3/blending/svg-blend-layer-blend.html [ Failure ]
  css3/blending/svg-blend-layer-clip-path.html [ Failure ]
  css3/blending/svg-blend-layer-filter.html [ Failure ]
  css3/blending/svg-blend-layer-mask.html [ Failure ]
  css3/blending/svg-blend-layer-opacity.html [ Failure ]

Regressions: Unexpected image-only failures (1)
  imported/blink/css3/blending/mix-blend-mode-with-squashing-layer.html [ ImageOnlyFailure ]

Which doesn't happen if I use my $HOME. The thing is that the test passing in both cases (if you actually look at what the test does and the result), but since the layers are places in different places (more margin on the left in one case) it sets it as failing.

I will rebase this tests and look forward to propose a patch for run-webkit-tests so it runs with an empty home by default always.

You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20190712/bd249e54/attachment.html>

More information about the webkit-unassigned mailing list