Bug 31919

Summary: REGRESSION(51418): fast/canvas/webgl/gl-object-get-calls.html is failing on the Leopard Commit Bot
Product: WebKit Reporter: Eric Seidel (no email) <eric>
Component: Tools / TestsAssignee: Nobody <webkit-unassigned>
Status: RESOLVED FIXED    
Severity: Normal CC: cmarrin, commit-queue, dglazkov, fishd, kbr, oliver
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: PC   
OS: OS X 10.5   
Attachments:
Description Flags
Patch none

Description Eric Seidel (no email) 2009-11-26 11:33:02 PST
fast/canvas/webgl/gl-object-get-calls.html is failing on the Leopard Commit Bot

https://bugs.webkit.org/show_bug.cgi?id=21554#c21
https://bugs.webkit.org/show_bug.cgi?id=31903#c4

I think it's failing 100% of the time.  I recommend we skip the test or roll out the associated change so the commit-queue can run again.

--- /tmp/layout-test-results/fast/canvas/webgl/gl-object-get-calls-expected.txt	2009-11-26 11:25:35.000000000 -0800
+++ /tmp/layout-test-results/fast/canvas/webgl/gl-object-get-calls-actual.txt	2009-11-26 11:25:35.000000000 -0800
@@ -7,7 +7,7 @@
 PASS gl.getError() is 0
 PASS gl.getError() is 0
 PASS gl.getError() is 0
-PASS gl.checkFramebufferStatus(gl.FRAMEBUFFER) is gl.FRAMEBUFFER_COMPLETE
+FAIL gl.checkFramebufferStatus(gl.FRAMEBUFFER) should be 36053. Was 36061.
 PASS gl.getFramebufferAttachmentParameter(gl.FRAMEBUFFER, gl.COLOR_ATTACHMENT0, gl.FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE) is gl.TEXTURE
 PASS gl.getFramebufferAttachmentParameter(gl.FRAMEBUFFER, gl.COLOR_ATTACHMENT0, gl.FRAMEBUFFER_ATTACHMENT_OBJECT_NAME) is non-null.
 PASS gl.getFramebufferAttachmentParameter(gl.FRAMEBUFFER, gl.COLOR_ATTACHMENT0, gl.FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL) is 0

It looks like the same failure we saw when trying to get the patch landed in the first place. :(
Comment 1 Eric Seidel (no email) 2009-11-26 11:34:39 PST
bug 21554 was the first failure we've seen like this.  All patches have failed on the commit-queue since.

I've stopped the queue for now.
Comment 2 Eric Seidel (no email) 2009-11-26 11:39:05 PST
https://bugs.webkit.org/show_bug.cgi?id=30091

http://trac.webkit.org/changeset/51418 is what started the failures.
Comment 3 Eric Seidel (no email) 2009-11-26 12:09:33 PST
I'm going to try moving the commit-queue to a different machine as a work-around from my end.
Comment 4 Kenneth Russell 2009-11-26 13:25:19 PST
I don't have commit rights; can you please comment out the failing line, regenerate the -expected.txt file and commit the result?
Comment 5 Oliver Hunt 2009-11-26 14:15:53 PST
(In reply to comment #4)
> I don't have commit rights; can you please comment out the failing line,
> regenerate the -expected.txt file and commit the result?

We need to find out why it's failing on some systems but not others, and add logic to ensure identical behaviour.
Comment 6 Kenneth Russell 2009-11-26 14:25:19 PST
(In reply to comment #5)
> (In reply to comment #4)
> > I don't have commit rights; can you please comment out the failing line,
> > regenerate the -expected.txt file and commit the result?
> 
> We need to find out why it's failing on some systems but not others, and add
> logic to ensure identical behaviour.

We need to get the commit queue reopened. There is already a bug about the unportability of this test: https://bugs.webkit.org/show_bug.cgi?id=31843 .
Comment 7 Mark Rowe (bdash) 2009-11-26 16:59:40 PST
(In reply to comment #4)
> I don't have commit rights; can you please comment out the failing line,
> regenerate the -expected.txt file and commit the result?

That sounds like it would have the effect of breaking the test on the build bots.  That makes it an unacceptable solution.
Comment 8 Kenneth Russell 2009-11-27 03:35:56 PST
Created attachment 43939 [details]
Patch

Temporarily disabled portion of test which is failing on some
machines but not others. Root cause will be investigated under
https://bugs.webkit.org/show_bug.cgi?id=31843 .

This is the correct fix for this issue. The remaining portions
of the test are both valid and valuable.
Comment 9 Eric Seidel (no email) 2009-11-27 08:09:41 PST
Comment on attachment 43939 [details]
Patch

Looks fine to me.  Thank you for the quick fix.

As noted in comment #3, I attempted to move the commit-bot to a new machine.  That move was successful.  (As you can see at http://webkit-commit-queue.appspot.com/), but this fix will make the tests pass correctly for everyone (including the old commit-bot machine if we need to move it back).

Setting cq+ too since I don't see you in committers.py yet.  Should be landed within 15 minutes or so while we return to enjoying our post-turkey haze. :)
Comment 10 WebKit Commit Bot 2009-11-27 08:21:08 PST
Comment on attachment 43939 [details]
Patch

Clearing flags on attachment: 43939

Committed r51441: <http://trac.webkit.org/changeset/51441>
Comment 11 WebKit Commit Bot 2009-11-27 08:21:13 PST
All reviewed patches have been landed.  Closing bug.