fmalita: there is a related TODO in TestExpectations assign to you. Please take a look or find a better owner.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5705d9370de375f0ec396f72ee035966424bbd54

commit 5705d9370de375f0ec396f72ee035966424bbd54
Author: Guido Urdaneta <guidou@chromium.org>
Date: Thu Aug 18 15:55:02 2016

Mark some virtual/gpu-rasterization/fast/images/color-profile-* tests as flaky on Win

TBR=fmalita@chromium.org
BUG= 638942 

Review URL: https://codereview.chromium.org/2254873005 .

Cr-Commit-Position: refs/heads/master@{#412847}

[modify] https://crrev.com/5705d9370de375f0ec396f72ee035966424bbd54/third_party/WebKit/LayoutTests/TestExpectations

 Issue 638843  has been merged into this issue.

+bsalomon

It appears that something introduced flakiness into some of the image rendering GPU layout tests on Windows Bots in the last day or so. Seems color profile related - do any Skia changes come to mind?

I can't think of anything that would introduce flakiness to these tests. AFAIK we aren't actually doing anything with the color profiles on images at this point. +brianosman

Sorry, I don't have a lot of insight either - that old TODO is there just as a reminder to re-enable a suppression.

Looking at the dashboard, it seems to me that these were failing/flaky way before ericrk's changes - e.g. color-profile-layer.html:

http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=virtual%2Fgpu-rasterization%2Ffast%2Fimages%2Fcolor-profile-layer.html

Here it is failing on Aug 16: https://uberchromegw.corp.google.com/i/chromium.webkit/builders/WebKit%20Win10/builds/13652/steps/webkit_tests/logs/stdio

Since the bots were not red before, I suspect they were all marked as flaky or failing.  I'm guessing someone rebaselined them for the other platforms and forgot to add back the win10 suppressions.  I'll see if I can find any evidence to that.

Doesn't look like they were suppressed before, but poking around at random earlier build logs I do see quite a few failures - sometimes taking 2-3 retries to pass.

So it appears they were flaky before, but not flaky enough to not pass after 3 retries.  Something happened recently that made them more flaky.

It's going to be difficult to find what caused the increase in flakiness.  I think these need to be investigated by someone familiar with color profiles, to figure why they're flaky in the first place.

Punting to ccameron, since he touched them last.  Please help reassign to the right owner.

I was sheriff yesterday and I saw an uptick in flakes after this patch https://codereview.chromium.org/2251583002/ and this rebaseline https://codereview.chromium.org/2251293002.

I reverted the patch and re-marked the tests for rebaseline https://codereview.chromium.org/2251863005/. But it looks like the patch to mark the tests as flaky (https://codereview.chromium.org/2242173007/) landed before the rebaseline ran here https://codereview.chromium.org/2254133003/, and prevented rebaseline-o-matic from rebaselining those tests.

I have reverted the patch to mark them as flaky, so that rebaseline-o-matic can process them https://codereview.chromium.org/2254303003/.

Feel free to re-mark as flaky if they still flake after the new rebaseline.

Never mind, the rebaseline has since run here https://codereview.chromium.org/2251293002.

The flakiness after rebaseline is probably caused by  bug 620126 . Flakiness dashboard is broken so we are unable to check if they are still flaky.

It's hard to tell what's going on since we haven;t run the test on windows for a long time. I'll try to rebaseline and see what the trybots have to say.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/d8ef4c33dcb9b9ff23cdcd0785618240d1019cb6

commit d8ef4c33dcb9b9ff23cdcd0785618240d1019cb6
Author: schenney <schenney@chromium.org>
Date: Thu Aug 25 18:35:31 2016

Re-enable some virtual/gpu-rasterization color profile tests

THe tests were marked as flakey after an unexplained increase in
flakiness plus some uncertainty about the baselines. So this
patch rebaselines them and re-enables since we have reason to
believe the flakiness has been addressed.

TBR=wangxianzhu@chromium.org
BUG= 638942 

Review-Url: https://codereview.chromium.org/2274573006
Cr-Commit-Position: refs/heads/master@{#414485}

[modify] https://crrev.com/d8ef4c33dcb9b9ff23cdcd0785618240d1019cb6/third_party/WebKit/LayoutTests/TestExpectations
[modify] https://crrev.com/d8ef4c33dcb9b9ff23cdcd0785618240d1019cb6/third_party/WebKit/LayoutTests/platform/win/virtual/gpu-rasterization/fast/images/color-profile-group-expected.png
[modify] https://crrev.com/d8ef4c33dcb9b9ff23cdcd0785618240d1019cb6/third_party/WebKit/LayoutTests/platform/win/virtual/gpu-rasterization/fast/images/color-profile-image-canvas-pattern-expected.png
[modify] https://crrev.com/d8ef4c33dcb9b9ff23cdcd0785618240d1019cb6/third_party/WebKit/LayoutTests/platform/win/virtual/gpu-rasterization/fast/images/color-profile-layer-expected.png
[modify] https://crrev.com/d8ef4c33dcb9b9ff23cdcd0785618240d1019cb6/third_party/WebKit/LayoutTests/platform/win/virtual/gpu-rasterization/fast/images/color-profile-svg-fill-text-expected.png

The new baselines are failing, suggesting that the try bots generate different results to build servers. Odd.

This bug seems obsolete, since I'm seeing an unrelated failure in one of those tests.