animations/combo-transform-translate+scale.html failed on Snow Leopard Debug Bot

• Component: New Bugs
• Status: RESOLVED FIXED

[Eric Seidel (no email)]

animations/combo-transform-translate+scale.html failed on Snow Leopard Debug Bot

I suspect this is just a flakey test.

http://build.webkit.org/results/SnowLeopard%20Intel%20Leaks/r53058%20(3325)/animations/combo-transform-translate+scale-diffs.txt
--- layout-test-results/animations/combo-transform-translate+scale-expected.txt	2010-01-10 19:45:06.000000000 -0800
+++ layout-test-results/animations/combo-transform-translate+scale-actual.txt	2010-01-10 19:45:06.000000000 -0800
@@ -1,2 +1,2 @@
 This test performs an animation of the translate() and scale() operators. It animates over 2 seconds. At 0.5 and 1 second it takes a snapshot and expects the result to be within range.
-PASS
+FAIL('d' was:2.71446, expected:2.5)

http://trac.webkit.org/browser/trunk/LayoutTests/animations/combo-transform-translate+scale.html

[Eric Seidel (no email)]

Failed again just now:
http://build.webkit.org/results/SnowLeopard%20Intel%20Leaks/r53248%20(3417)/animations/combo-transform-translate+scale-diffs.txt

different diff than last time:
--- layout-test-results/animations/combo-transform-translate+scale-expected.txt	2010-01-14 01:40:27.000000000 -0800
+++ layout-test-results/animations/combo-transform-translate+scale-actual.txt	2010-01-14 01:40:27.000000000 -0800
@@ -1,2 +1,2 @@
 This test performs an animation of the translate() and scale() operators. It animates over 2 seconds. At 0.5 and 1 second it takes a snapshot and expects the result to be within range.
-PASS
+FAIL('f' was:111, expected:100)

[Ojan Vafai]

Created attachment 46628 [details]
Patch

[Eric Seidel (no email)]

Comment on attachment 46628 [details]
Patch

Don't you want to start the timeouts from the startedTransform  callback?

[Ojan Vafai]

(In reply to comment #3)
> (From update of attachment 46628 [details])
> Don't you want to start the timeouts from the startedTransform  callback?

It already does. The only timeout I'm removing is the one that writes out the already calculated result and calls notifyDone. The timeouts that do measurement are still started in the webkitAnimationStart handler.

[Eric Seidel (no email)]

Comment on attachment 46628 [details]
Patch

OK.

[Ojan Vafai]

Committed r53333: <http://trac.webkit.org/changeset/53333>