Right now layout test media/media-document-audio-repaint.html (fails on Chromium GPU as you see below. http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20GPU%20Mesa%20-%20chromium.org&tests=media%2Fmedia-document-audio-repaint.html http://trac.webkit.org/browser/trunk/LayoutTests/media/media-document-audio-repaint.html#L8 shows it uses layoutTestController.display(). As for now. we will skip the test on GPU. However, it would be nice if this is implemented. Please let me know if you need more information.
Adding jamesr/kbr to see if they have any ideas on implementing said DRT function I believe the common use case is to get DRT to paint over the entire screen with semi-transparent gray then trigger something that will cause a repaint, presumably testing that we're not painting too much. I'm not sure if this makes sense for GPU compositing or what the equivalent test would be.
LayoutTestController.display() is (obviously) implemented, we have a ton of tests that depend on it. It doesn't have the same incremental repaint behavior as the software path because we don't do incremental composites, we always composite the full screen. You should NOT skip the test unless it crashes. Add the correct expectations.
Thanks for the explanation! imasaki: want to double check our test expectations? the pixel result should look like the CPU version, except have a white background instead of the gray
I double checked and the pixel result good. Rebaseline committed: https://bugs.webkit.org/attachment.cgi?id=105420&action=review
sorry, my bad. The expected result does not look right. I will be working on rebaseline.
I am working on rebaseline for GPU
Rebaselines are committed. https://bugs.webkit.org/show_bug.cgi?id=67760
(In reply to comment #7) > Rebaselines are committed. https://bugs.webkit.org/show_bug.cgi?id=67760 The last rebaseline caused some text mistmatches: http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20GPU%20Mesa%20-%20chromium.org&showExpectations=true&tests=media%2Fmedia-document-audio-repaint.html So I've done another one: http://trac.webkit.org/changeset/99053
I'm rebaselining this test as a result of another change, https://bugs.webkit.org/show_bug.cgi?id=75505 . I suspect this can be closed once that one goes in.
This can now be closed.