Bug 33893 – REGRESSION(r53514?): fast/canvas/webgl/texImage2DImageDataTest.html fails on Leopard Release Bot

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Eric Seidel (no email)]

REGRESSION(r53514?): fast/canvas/webgl/texImage2DImageDataTest.html fails on Leopard Release Bot

A webgl failure seemed to start with r53514:
http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r53517%20(9580)/fast/canvas/webgl/texImage2DImageDataTest-pretty-diff.html

The build was broken, so we didn't actually get a build until r53517.

The first failure started between r53514 and r53517 inclusive.

Was added by cmarrin:
http://trac.webkit.org/browser/trunk/LayoutTests/fast/canvas/webgl/texImage2DImageDataTest.html

pol maintains this bot.  it' possible something changed on teh bot.

Maciej and I are both unable to reproduce this failure on our local machines.

[Eric Seidel (no email)]

Ahha!  webgl tests are failing on my Mac Book Pro.  They did not fail on my Mac Pro work machine however.

[Eric Seidel (no email)]

This test is in best case flakey.  It has passed on the leopard bot at least once since it started failing:
http://build.webkit.org/builders/Leopard%20Intel%20Release%20%28Tests%29/builds/9601

[Eric Seidel (no email)]

It's continuing to be flakey. :(

http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r53564%20(9628)/fast/canvas/webgl/texImage2DImageDataTest-diffs.txt

[Eric Seidel (no email)]

Created attachment 47066 [details]
Patch

[Maciej Stachowiak]

Even though the test started failing more often after my change, it seems there is evidence that the test itself is the problem. First, it sometimes fails and sometimes passes even after r53514-r53517. Second, it failed intermittently even before those revisions, though less often. Here is an example of it failing at r53494: http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r53494%20(9561)/results.html

[Maciej Stachowiak]

Comment on attachment 47066 [details]
Patch

Please cite https://bugs.webkit.org/show_bug.cgi?id=33924 instead of this bug in the Skipped comment, so this bug can be closed when the skip patch is landed and the other can remain open as the reminder to fix. With that adjustment, I'll r+.

[Eric Seidel (no email)]

Committed r53578: <http://trac.webkit.org/changeset/53578>