fast/loader/submit-form-while-parsing-2.html fails intermittently (17 times during 200 builds) on the Qt-WK2 bot with this error: --- /ramdisk/qt-linux-32-release-webkit2/build/layout-test-results/fast/loader/submit-form-while-parsing-2-expected.txt +++ /ramdisk/qt-linux-32-release-webkit2/build/layout-test-results/fast/loader/submit-form-while-parsing-2-actual.txt @@ -1,3 +1,2 @@ -PASSED: If we successfully got here without an assertion or crash, all is well. +FAIL: Timed out waiting for notifyDone to be called -
You can easily reproduce this bug: $ Tools/Scripts/old-run-webkit-tests -2 fast/loader/submit-form-while-parsing-2.html --iterations 100 Testing 1 test cases 100 times. fast/loader . fast/loader/submit-form-while-parsing-2.html -> failed ............... fast/loader/submit-form-while-parsing-2.html -> failed ... fast/loader/submit-form-while-parsing-2.html -> failed . fast/loader/submit-form-while-parsing-2.html -> failed ............. fast/loader/submit-form-while-parsing-2.html -> failed .......... fast/loader/submit-form-while-parsing-2.html -> failed .. fast/loader/submit-form-while-parsing-2.html -> failed . fast/loader/submit-form-while-parsing-2.html -> failed ....... fast/loader/submit-form-while-parsing-2.html -> failed ........................... fast/loader/submit-form-while-parsing-2.html -> failed .............. fast/loader/submit-form-while-parsing-2.html -> failed . fast/loader/submit-form-while-parsing-2.html -> failed .... fast/loader/submit-form-while-parsing-2.html -> failed . 80.61s total testing time 87 test cases (87%) succeeded 13 test cases (13%) had incorrect layout
I skipped it to make Qt-wk2 bot happier: http://trac.webkit.org/changeset/104561
I'm seeing this fail on Mac frequently too (and can reproduce locally).
This test loads a subframe that immediately submits a form, and also starts a zero-delay timer that starts another navigation. In WK1, the timer doesn't get a chance to fire, as form submission happens first. In WK2, the timer fires and starts a navigation. The URL it uses is incorrect (there should be no "resources/" part), so instead of successful load, we get an error. But there is a race between navigation and form submission - when submission happens first, we get a load event and pass, and when navigation happens first, we get a timeout. It's easy to fix the timeout by fixing the URL, but we'll still have a race between navigation and form submission. I'm not sure if it's a well defined situation.
Fixed the URL in r128665 to make the test a bit more reasonable. Not yet sure what to do about the race.
Qt port is gone but the platform specific expectation of 'Mac-wk2' is on different bug. I am marking this duplicate of it to other bug. *** This bug has been marked as a duplicate of bug 105952 ***