https://tbpl.mozilla.org/php/getParsedLog.php?id=8468888&tree=Mozilla-Inbound 1800 INFO TEST-START | chrome://mochitests/content/chrome/content/xul/content/test/test_bug330705-2.xul 1801 INFO TEST-PASS | chrome://mochitests/content/chrome/content/xul/content/test/test_bug330705-2.xul | The first box element is still focused after blur() has been called on the second box element 1802 INFO TEST-END | chrome://mochitests/content/chrome/content/xul/content/test/test_bug330705-2.xul | finished in 131ms 1803 INFO TEST-START | chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html 1804 INFO TEST-PASS | chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html | The first box element is still focused after blur() has been called on the second box element 1805 INFO TEST-PASS | chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html | The first box element is still focused after blur() has been called on the second box element 1806 INFO TEST-END | chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html | finished in 27ms 1807 INFO TEST-START | chrome://mochitests/content/chrome/content/xul/document/test/test_bug199692.xul 1808 INFO TEST-UNEXPECTED-FAIL | chrome://mochitests/content/chrome/content/xul/document/test/test_bug199692.xul | finished in a non-clean fashion (in chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html) 1809 INFO TEST-END | chrome://mochitests/content/chrome/content/xul/document/test/test_bug199692.xul | finished in 23ms Sort of a puzzling log, but I think what that's saying is "test_bug330705-2.xul started, "finished", continued running its setTimeout while test_bug398289.html thought it was the one that was running, since that's 330705-2 output in 398289, then called finish() while test_bug199692.xul was running, making bug 668728 think it was test_bug398289.html at fault while it was really another victim." Or maybe I don't understand either the log or bug 668728.

These are the interesting parts of the log: 1800 INFO TEST-START | chrome://mochitests/content/chrome/content/xul/content/test/test_bug330705-2.xul 1801 INFO TEST-PASS | chrome://mochitests/content/chrome/content/xul/content/test/test_bug330705-2.xul | The first box element is still focused after blur() has been called on the second box element 1802 INFO TEST-END | chrome://mochitests/content/chrome/content/xul/content/test/test_bug330705-2.xul | finished in 130ms 1803 INFO TEST-START | chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html 1804 INFO TEST-PASS | chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html | The first box element is still focused after blur() has been called on the second box element 1805 INFO TEST-PASS | chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html | The first box element is still focused after blur() has been called on the second box element 1806 INFO TEST-END | chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html | finished in 28ms 1807 INFO TEST-START | chrome://mochitests/content/chrome/content/xul/document/test/test_bug199692.xul 1808 INFO TEST-UNEXPECTED-FAIL | chrome://mochitests/content/chrome/content/xul/document/test/test_bug199692.xul | finished in a non-clean fashion (in chrome://mochitests/content/chrome/content/xul/content/test/test_bug398289.html) 1809 INFO TEST-END | chrome://mochitests/content/chrome/content/xul/document/test/test_bug199692.xul | finished in 23ms Note that the message from test_bug330705-2.xul is printed twice after this test has finished. This means that SimpleTest.finish would be called twice, and presumably the call to SimpleTest.finish from test_bug398289.html gets executed in the context of test_bug199692.xul, and this is caught by my check. So, effectively, test_bug330705-2.xul is causing its next two tests to be skipped some of the times. The only way that this could happen is if the focus handler is called multiple times. I have no idea why that would happen, but it _is_ happening some of the times. The simplest way to fix this problem is to remove the focus handler after the first time it's executed.

Comment on attachment 587717 [details] [diff] [review] Patch (v1) r=me