Closed Bug 874108 Opened 8 years ago Closed 6 years ago

Intermittent content/events/test/test_bug656379-1.html | Releasing the mouse over the label should have unpressed (and focused) the button.

Categories

(Core :: DOM: UI Events & Focus Handling, defect)

x86
Windows 8
defect
Not set
normal

Tracking

()

RESOLVED WORKSFORME
Tracking Status
firefox24 --- disabled
firefox25 --- affected
firefox26 --- affected

People

(Reporter: RyanVM, Assigned: smaug)

Details

(Keywords: intermittent-failure, Whiteboard: [pgo])

Attachments

(2 files)

Attached image expected image from log
https://tbpl.mozilla.org/php/getParsedLog.php?id=23107841&tree=Mozilla-Inbound

WINNT 6.2 mozilla-inbound pgo test mochitest-1 on 2013-05-18 00:26:15 PDT for push a20b345ae2cb
slave: t-w864-ix-039

00:35:02     INFO -  77056 INFO TEST-START | /tests/content/events/test/test_bug656379-1.html
00:35:02     INFO -  77057 INFO TEST-PASS | /tests/content/events/test/test_bug656379-1.html | Pressed focused buttons should look different from normal buttons. - got undefined and undefined
00:35:02     INFO -  77058 INFO TEST-PASS | /tests/content/events/test/test_bug656379-1.html | Pressed focused buttons should look different from normal focused buttons. - got undefined and undefined
00:35:02     INFO -  77059 INFO TEST-PASS | /tests/content/events/test/test_bug656379-1.html | Pressing the label should have pressed the button. - got undefined and undefined
00:35:02     INFO -  77060 INFO TEST-PASS | /tests/content/events/test/test_bug656379-1.html | Moving the mouse down from the label should have unpressed the button. - expected undefined
00:35:02     INFO -  77061 INFO TEST-PASS | /tests/content/events/test/test_bug656379-1.html | Moving the mouse back on top of the label should have pressed the button. - expected undefined
00:35:03     INFO -  77062 ERROR TEST-UNEXPECTED-FAIL | /tests/content/events/test/test_bug656379-1.html | Releasing the mouse over the label should have unpressed (and focused) the button. - expected data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAC0AAAAUCAYAAAAZb7T/AAAAzUlEQVRIie2VbQ7DIAiGOQ/34Tzch/N4n3c/1g/q1DpdnFt8kyYSLDwgtoQfFH0boEULepT+BdogRKD4YUUYz5dUBpqhEWFQrgT376dj9aoaGkHBJLDbkBNBB2WQWGbPbgcoJ0br2Bv5Lye3xVA5/JyptnGmc9ClNWBCrvDY3vLuduFkqztt4sFboBNxL2ClmI3QzwQ5gArQZOdKRX0MuqYrE3X6dTwI5/ixu2w9M90FXfFzMX/L1SU4vxBifn31vXe5b6Hn14IepQU9Sg+q46CZUdLKDwAAAABJRU5ErkJggg== but got data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAC0AAAAUCAYAAAAZb7T/AAAAsElEQVRIie2UYQ6AIAiF3yU8hZfoEl7CU3Rz+6PF3shSW8kWGxuE4JeCSAYFXwP0yA6NFdPrDz0HNBR1FqAjJS0N4DJfq/UadMzfTUEvSPCVNcV30NurrJVxvjnkfSDsJmhWbQMNumZ7+nH2IXzlZttP2hN4DzTnMFitZhd0vAC4Aj2bibs1h6BNnbTWHgHHkD7V00PQrDyIQcQYurwQgWwZ02oOQU+qP/Rn0JbEJPQG/FhmuUtKXWMAAAAASUVORK5CYII=
00:35:03     INFO -  77063 INFO TEST-PASS | /tests/content/events/test/test_bug656379-1.html | Removing the label should have unpressed the button. - expected undefined
00:35:03     INFO -  77064 INFO TEST-END | /tests/content/events/test/test_bug656379-1.html | finished in 301ms
Attached image actual image from log
dbaron, can you suggest someone who can look into this?
Flags: needinfo?(dbaron)
No idea.  Did this appear when we started running tests on Windows 8 ("WINNT 6.1"), or was it running on Windows 8 successfully for a while before it started intermittently failing?

If it's the latter, I didn't see anything obvious in the days before it landed; if it's the former, maybe jimm would know?
Component: CSS Parsing and Computation → Event Handling
Flags: needinfo?(dbaron)
We've been running tests on Win8 since mid-March, so it appears that this is a legit regression.
bz fixed bug 656379 and presumably wrote this test case. Perhaps he can shed some light on this. As Boris' status says that he is away until Aug 19, pinging jst to see if he has any other suggestions for someone to start looking into this.

I also see bug 475156 and bug 482935 listed in comment 8. These bugs seem pretty different and are only listed in this one comment so this may be a red herring.
Flags: needinfo?(jst)
Flags: needinfo?(bugs)
bz could likely help out here as well, as/if needed.
Flags: needinfo?(jst)
Flags: needinfo?(bzbarsky)
I didn't actually write this testcase.  mstange did, for bug 426082.  I just copied it...  and then Matt heavily modified it for bug 539356.

In any case, is the issue perhaps that we're ending up shapshotting before we paint?  e.g. would making executeTests use requestAnimationFrame instead of executeSoon help?
Flags: needinfo?(bzbarsky) → needinfo?(matt.woodrow)
I just moved the test to run in its own window, it shouldn't have affected how it works.

We're using MozAfterPaint to ensure that we've painted before the snapshot is taken, so we should be fine there at least.
Flags: needinfo?(matt.woodrow)
These are all failures in PGO builds, whatever that means.
Whiteboard: [pgo]
Olli, can you please dig in here. We need someone to own this, I'm hoping that can be out. If not, please let me know. Thanks.
Assignee: nobody → bugs
Rather busy right now, but trying to look at this (and few other test failures) next week.
Flags: needinfo?(bugs)
Disabled on Windows for Fx24. We can always re-enable it if an upliftable fix materializes.

https://hg.mozilla.org/releases/mozilla-beta/rev/fc3969695f26
Olli, any updates here?
It is in my endless todo list. No other updates.