Bug 69624 – [chromium] Rebaseline layout test zoom-animator-scale-test2.html for win, mac

• Status: RESOLVED
• Resolution: FIXED
• Component: Layout and Rendering

[W. James MacLean]

When https://bugs.webkit.org/show_bug.cgi?id=68035 lands, will need baselines for the new test on win and mac. These can be taken off the bots.

[Kent Tamura]

The results of WIndows and Mac look wrong.
http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20GPU%20Mesa%20-%20chromium.org&tests=zoom-animator-scale-test2.html&showExpectations=true&showLargeExpectations=true

[W. James MacLean]

Created attachment 110902 [details]
Patch

[Kenneth Russell]

Comment on attachment 110902 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=110902&action=review

> LayoutTests/platform/chromium/test_expectations.txt:2717
> +BUGWK69624 WIN GPU : platform/chromium/compositing/zoom-animator-scale-test2.html = IMAGE

Did you intend to leave the GPU modifier here given that the new baseline you're providing is for chromium-gpu-win? You may well have; just confirming.

[W. James MacLean]

(In reply to comment #3)
> (From update of attachment 110902 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=110902&action=review
> 
> > LayoutTests/platform/chromium/test_expectations.txt:2717
> > +BUGWK69624 WIN GPU : platform/chromium/compositing/zoom-animator-scale-test2.html = IMAGE
> 
> Did you intend to leave the GPU modifier here given that the new baseline you're providing is for chromium-gpu-win? You may well have; just confirming.

Yes, as (strangely) the test is passing on Win WebKit Dbg(2) but not on Win GPU, Win7 GPU and Vista GPU, so I'm still experimenting a bit.

[Kent Tamura]

Comment on attachment 110902 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=110902&action=review

> LayoutTests/platform/chromium/test_expectations.txt:2718
> +BUGWK70046 SKIP MAC : platform/chromium/compositing/zoom-animator-scale-test2.html = IMAGE+TEXT

Do not skip a valid test. SKIP means buildbots don't produce results.

[W. James MacLean]

(In reply to comment #5)
> (From update of attachment 110902 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=110902&action=review
> 
> > LayoutTests/platform/chromium/test_expectations.txt:2718
> > +BUGWK70046 SKIP MAC : platform/chromium/compositing/zoom-animator-scale-test2.html = IMAGE+TEXT
> 
> Do not skip a valid test. SKIP means buildbots don't produce results.

But this isn't a valid test for Mac, not until someone undertakes the work described in https://bugs.webkit.org/show_bug.cgi?id=70046 to add zoom animation to ScrollAnimatorChromiumMac. Until then we know the test will always fail, so is it worth running it?

[Kent Tamura]

Comment on attachment 110902 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=110902&action=review

>>> LayoutTests/platform/chromium/test_expectations.txt:2718
>>> +BUGWK70046 SKIP MAC : platform/chromium/compositing/zoom-animator-scale-test2.html = IMAGE+TEXT
>> 
>> Do not skip a valid test. SKIP means buildbots don't produce results.
> 
> But this isn't a valid test for Mac, not until someone undertakes the work described in https://bugs.webkit.org/show_bug.cgi?id=70046 to add zoom animation to ScrollAnimatorChromiumMac. Until then we know the test will always fail, so is it worth running it?

Running this test on Mac is helpful.  Now we expect IMAGE+TEXT, and it means the test doesn't crash. We would find a regression If this test crashed.

[W. James MacLean]

Created attachment 111017 [details]
Patch

[W. James MacLean]

(In reply to comment #7)
> (From update of attachment 110902 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=110902&action=review
> 
> >>> LayoutTests/platform/chromium/test_expectations.txt:2718
> >>> +BUGWK70046 SKIP MAC : platform/chromium/compositing/zoom-animator-scale-test2.html = IMAGE+TEXT
> >> 
> >> Do not skip a valid test. SKIP means buildbots don't produce results.
> > 
> > But this isn't a valid test for Mac, not until someone undertakes the work described in https://bugs.webkit.org/show_bug.cgi?id=70046 to add zoom animation to ScrollAnimatorChromiumMac. Until then we know the test will always fail, so is it worth running it?
> 
> Running this test on Mac is helpful.  Now we expect IMAGE+TEXT, and it means the test doesn't crash. We would find a regression If this test crashed.

Ahh, understood. Have revised accordingly.

[WebKit Review Bot]

Comment on attachment 111017 [details]
Patch

Clearing flags on attachment: 111017

Committed r97588: <http://trac.webkit.org/changeset/97588>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[W. James MacLean]

Created attachment 111655 [details]
Patch

[W. James MacLean]

This test is still flaky on WIN GPU (although better than before, when it was 100% fail). I've marked the test as SLOW since longest runtimes are getting up in the 5+ second range, and I've modified the internal timeout to be longer.

I've re-built Chrome on a Win 7 machine, and found this test passes 100% of the time on a dedicated machine, so I suspect it's an issue with the bots running the test slowly.

Once we get the test more stable, we can experiment with decreasing times if desired.

[Kenneth Russell]

Comment on attachment 111655 [details]
Patch

OK. In general though it's better to file a new bug than reuse an existing one for tracking purposes.

[W. James MacLean]

(In reply to comment #14)
> (From update of attachment 111655 [details])
> OK. In general though it's better to file a new bug than reuse an existing one for tracking purposes.

Thanks, will file a new bug next time.

[WebKit Review Bot]

Comment on attachment 111655 [details]
Patch

Clearing flags on attachment: 111655

Committed r97916: <http://trac.webkit.org/changeset/97916>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.