RESOLVED DUPLICATE of bug 102500 104866
DOM/DOMWalk.html fails on Chromium Windows
https://bugs.webkit.org/show_bug.cgi?id=104866
Summary DOM/DOMWalk.html fails on Chromium Windows
noel gordon
Reported 2012-12-12 17:18:22 PST
Chromium win release perf bots at webkit.org are mostly always red. http://build.webkit.org/builders/Chromium%20Win%20Release%20%28Perf%29?numbuilds=100 Cause is test DOM/DOMWalk.html, which reports that time goes backwards or some such, on just about every perf test run. Running DOM/DOMWalk.html (35 of 113) runFunction returned a non-positive value: 0 runFunction returned a non-positive value: 0 runFunction returned a non-positive value: 0 runFunction returned a non-positive value: 0 runFunction returned a non-positive value: 0 runFunction returned a non-positive value: 0 FAILED Finished: 9.610000 s The same error is seen on Chromium Linux, though much less often.
Attachments
Hopefully fixes the bug (2.35 KB, patch)
2012-12-13 02:40 PST, Ryosuke Niwa
no flags
Ryosuke Niwa
Comment 1 2012-12-12 17:19:54 PST
yeah, we need to make those tests use window.performance instead.
noel gordon
Comment 2 2012-12-12 17:23:56 PST
How do I disable this test (at least until it's fixed by someone who knows this test)? I want green perf bots so I can catch real regressions.
Ryosuke Niwa
Comment 3 2012-12-12 17:26:55 PST
(In reply to comment #2) > How do I disable this test (at least until it's fixed by someone who knows this test)? I want green perf bots so I can catch real regressions. You never catch regressions on webkit perf. bots. Bots run red iff tests fail to complete as expected. It doesn’t catch perf regressions. To do that, you have to look at graphs on webkit-perf.appspot.com
noel gordon
Comment 4 2012-12-12 17:28:14 PST
/meh I want green perf bots.
noel gordon
Comment 5 2012-12-12 17:53:58 PST
(In reply to comment #4) > /meh I want green perf bots. so, how do I disable this test?
Ryosuke Niwa
Comment 6 2012-12-12 18:13:38 PST
We don't have a way of skipping a test on one platform. If you want a green bot, please either fix the test or Date.now implementation. I'll try to look into it in the next couple of days in my spare time but don't count on me.
Ryosuke Niwa
Comment 7 2012-12-13 02:40:14 PST
Created attachment 179237 [details] Hopefully fixes the bug
Ryosuke Niwa
Comment 8 2012-12-13 02:41:50 PST
*** Bug 99432 has been marked as a duplicate of this bug. ***
noel gordon
Comment 9 2012-12-13 03:46:58 PST
Comment on attachment 179237 [details] Hopefully fixes the bug View in context: https://bugs.webkit.org/attachment.cgi?id=179237&action=review > PerformanceTests/ChangeLog:3 > + [chromium perf] DOM/DOMWalk.html failures Thank you, LGTM.
Eric Seidel (no email)
Comment 10 2012-12-13 09:57:28 PST
Comment on attachment 179237 [details] Hopefully fixes the bug Fancy
WebKit Review Bot
Comment 11 2012-12-13 11:12:18 PST
Comment on attachment 179237 [details] Hopefully fixes the bug Clearing flags on attachment: 179237 Committed r137623: <http://trac.webkit.org/changeset/137623>
WebKit Review Bot
Comment 12 2012-12-13 11:12:22 PST
All reviewed patches have been landed. Closing bug.
Ryosuke Niwa
Comment 13 2012-12-19 19:14:13 PST
This test is still failing on Chromium Windows. Someone needs to debug this.
Ryosuke Niwa
Comment 14 2012-12-19 19:17:34 PST
Dominic Cooney
Comment 15 2013-01-15 00:30:51 PST
*** This bug has been marked as a duplicate of bug 102500 ***
Note You need to log in before you can comment on or make changes to this bug.