RESOLVED FIXED 67359
empty-urls.html seems to have erratic output. 
https://bugs.webkit.org/show_bug.cgi?id=67359
Summary empty-urls.html seems to have erratic output.
David Levin
Reported 2011-08-31 22:35:27 PDT
Full name: http/tests/loading/empty-urls.html Flakiness examples: http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=http%2Ftests%2Floading%2Fempty-urls.html&showExpectations=true
Attachments
patch (2.34 KB, patch)
2011-09-01 09:49 PDT, Nate Chapin 		no flags
DetailsFormatted DiffDiff
yet another attempt (6.68 KB, patch)
2011-10-27 16:35 PDT, Nate Chapin 		no flags
DetailsFormatted DiffDiff
Patch for landing (10.02 KB, patch)
2011-10-28 09:35 PDT, Nate Chapin 		no flags
DetailsFormatted DiffDiff
Show Obsolete (2) View All Add attachment proposed patch, testcase, etc.
Nate Chapin
Comment 1 2011-09-01 08:43:24 PDT
The failure I'm seeing on the dashboard appears to be based on whether the load event finishes before notifyDone() is called. Will try to stabilize that.
Nate Chapin
Comment 2 2011-09-01 09:49:15 PDT
Created attachment 105980 [details] patch
WebKit Review Bot
Comment 3 2011-09-01 11:36:14 PDT
Comment on attachment 105980 [details] patch Clearing flags on attachment: 105980 Committed r94320: <http://trac.webkit.org/changeset/94320>
WebKit Review Bot
Comment 4 2011-09-01 11:36:18 PDT
All reviewed patches have been landed. Closing bug.
David Levin
Comment 5 2011-10-25 18:17:08 PDT
This doesn't appear to have been fixed for two reasons: 1. The expectations file never had this line removed. 2. The test is still flaky: http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=http%2Ftests%2Floading%2Fempty-urls.html
Nate Chapin
Comment 6 2011-10-27 13:47:28 PDT
(In reply to comment #5) > This doesn't appear to have been fixed for two reasons: > 1. The expectations file never had this line removed. > 2. The test is still flaky: http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=http%2Ftests%2Floading%2Fempty-urls.html I'm stumped on how to resolve this. It's flaky because the order of the frame load callbacks for the main frame (specifically, didFinishLoadForFrame and didHandleOnloadEventsForFrame) is non-deterministic. I'm not seeing a good way to turn those outputs off, nor how to force the ordering to be deterministic. Suggestions welcome.
Adam Barth
Comment 7 2011-10-27 13:48:26 PDT
Comment on attachment 105980 [details] patch I think you can turn them off by moving the test out of the "loading" directory.
Nate Chapin
Comment 8 2011-10-27 14:26:17 PDT
(In reply to comment #7) > (From update of attachment 105980 [details]) > I think you can turn them off by moving the test out of the "loading" .... That did it. Where on earth is that defined/documented?
Nate Chapin
Comment 9 2011-10-27 16:35:34 PDT
Created attachment 112780 [details] yet another attempt
David Levin
Comment 10 2011-10-27 16:43:36 PDT
(In reply to comment #9) > Created an attachment (id=112780) [details] > yet another attempt You're going to need to resolve conflicts to get a cq+ that will work.
Nate Chapin
Comment 11 2011-10-28 09:35:25 PDT
Created attachment 112875 [details] Patch for landing
WebKit Review Bot
Comment 12 2011-10-28 09:58:50 PDT
Comment on attachment 112875 [details] Patch for landing Clearing flags on attachment: 112875 Committed r98732: <http://trac.webkit.org/changeset/98732>
WebKit Review Bot
Comment 13 2011-10-28 09:58:55 PDT
All reviewed patches have been landed. Closing bug.
Note You need to log in before you can comment on or make changes to this bug.