--- /ramdisk/qt-linux-32-release-qt470/build/layout-test-results/http/tests/misc/script-async-expected.txt 2011-07-12 03:36:30.445722227 -0700 +++ /ramdisk/qt-linux-32-release-qt470/build/layout-test-results/http/tests/misc/script-async-actual.txt 2011-07-12 03:36:30.445722227 -0700 @@ -1,2 +1,2 @@ This tests for proper execution order of scripts with the async attribute https://bugs.webkit.org/show_bug.cgi?id=20710. -PASS +FAIL: Expected one of ' async external inline DOMContentLoaded slowAsync ' || ' external async inline DOMContentLoaded slowAsync ' || ' external inline async DOMContentLoaded slowAsync ' || ' external inline DOMContentLoaded async slowAsync ', Actual=' external inline DOMContentLoaded slowAsync async ' Marked as P1 critical, because it is a regression.
Peter, Markus, would one of you have time to investigate why http pipelining break the tests? I am tempted to just revert r90634 since this demonstrate Qt's Http pipelining is not ready for prime time.
@Benjamin: HTTP pipelining should not be enabled for Ajax stuff. Everything that blocks the socket (e.g. long polling, COMET, whatever) should not be pipelined.
@Benjamin: See for example bug 50644
I skip this test until you fix the bug: http://trac.webkit.org/changeset/90891
Created attachment 100659 [details] Patch
Comment on attachment 100659 [details] Patch Arr plus.
Comment on attachment 100659 [details] Patch Clearing flags on attachment: 100659 Committed r90906: <http://trac.webkit.org/changeset/90906>
All reviewed patches have been landed. Closing bug.