http/tests/cookies/third-party-cookie-relaxing.html is failing under new-run-webkit-tests on the snow leopard bot
Example failure: http://build.webkit.org/results/SnowLeopard%20Intel%20Release%20(Tests)/r90460%20(31134)/http/tests/cookies/third-party-cookie-relaxing-pretty-diff.html --- /Volumes/Data/WebKit-BuildSlave/snowleopard-intel-release-tests/build/layout-test-results/http/tests/cookies/third-party-cookie-relaxing-expected.txt +++ /Volumes/Data/WebKit-BuildSlave/snowleopard-intel-release-tests/build/layout-test-results/http/tests/cookies/third-party-cookie-relaxing-actual.txt @@ -26,7 +26,7 @@ ALERT: Test stage 5 document.cookie is: ALERT: http://localhost:8000/cookies/resources/cookie-utility.php?queryfunction=setFooCookie ALERT: XHR response - Set the foo cookie -ALERT: Test stage 6 document.cookie is: foo=awesomevalue +ALERT: Test stage 6 document.cookie is: ALERT: Restricting to first party only cookies ALERT: http://localhost:8000/cookies/resources/cookie-utility.php?queryfunction=deleteCookies ALERT: XHR response - Deleted all cookies @@ -39,9 +39,9 @@ ALERT: Test stage 8 document.cookie is: ALERT: http://localhost:8000/cookies/resources/cookie-utility.php?queryfunction=setFooCookie ALERT: XHR response - Set the foo cookie -ALERT: Test stage 9 document.cookie is: foo=awesomevalue +ALERT: Test stage 9 document.cookie is: ALERT: Restricting to first party only cookies ALERT: http://localhost:8000/cookies/resources/cookie-utility.php?queryfunction=setFooAndBarCookie ALERT: XHR response - Set the foo and bar cookies -ALERT: Test stage 10 document.cookie is: bar=anotherawesomevalue; foo=awesomevalue +ALERT: Test stage 10 document.cookie is:
new-run-webkit-tests http/tests/cookies/third-party-cookie-relaxing.html fails when run straight on my machine. Then again, so does: old-run-webkit-tests http/tests/cookies/third-party-cookie-relaxing.html
It also fails for me with: old-run-webkit-tests http/tests/cookies and old-run-webkit-tests http I'm trying to understand how the test is supposed to work? http://trac.webkit.org/browser/trunk/LayoutTests/http/tests/cookies/third-party-cookie-relaxing.html Maybe there was a recent regression in behavior on Snow Leopard which happen to correspond with the NRWT transition? Maybe NRWT running cookie-tests in parallel manages to permanently "jam" the CF cookie store? Donno.
Qt just added support for third-party cookies, and their results seem to match SL, in at least the 6 and 9 cases: http://trac.webkit.org/browser/trunk/LayoutTests/platform/qt/http/tests/cookies/third-party-cookie-relaxing-expected.txt
Bug 63545 is another cookie-related failure which occurs on my machine. This one of course occurs on the SL bot as well: http://build.webkit.org/results/SnowLeopard%20Intel%20Release%20(Tests)/r90460%20(31134)/results.html
Even if I move my ~/Library/Cookies/Cookies.plist file out of the way, the test still fails for me.
Turns out this has nothing to do with NRWT, now that I look at the bot history (with webkit-patch failure-reason).
I don't see any changes in that range which seem particularly related. Perhaps failure-reason is confused.
The only possibly-related changes in that range are to the JIT. But I don't see how they could affect this test: http://trac.webkit.org/changeset/90237 http://trac.webkit.org/changeset/90232 I checked the results. Unless build.webkit.org/results is lying, it's definitely this range. I've not tried using git-bisect, but I think the build was broken a bunch in there, so it may not answer our question.
TestFailures says this test is flaky: r90315 (31042) (results.html): fail r90310 (31041) (results.html): pass r90308 (31040) (results.html): pass r90288 (31027) (results.html): fail r90285 (31026) (results.html): pass r90284 (31025) (results.html): fail r90280 (31024) (results.html): pass r90279 (31023) (results.html): fail r90273 (31022) (results.html): pass r90271 (31019) (results.html): fail It always fails with the same diff: http://build.webkit.org/results/SnowLeopard%20Intel%20Release%20(Tests)/r90315%20(31042)/http/tests/cookies/third-party-cookie-relaxing-pretty-diff.html
This is no longer happening. Karl removed the expectation via bug 213620.
<rdar://problem/64783252>