Hmm, I can understand the reasons for different widths in the last test case, and I believe this will be fixed by bug 94604, but it's odd that the widths of the blue-100 and green-10 images would be affected by this change. I'll look to see if there's any mac specific code in the image layout, but I don't have a mac so my debugging will be limited to code inspection. Ojan, do you have any idea why the widths of these image would be affected by this change?

*** Bug 94675 has been marked as a duplicate of this bug. ***

I don't think this is platform-specific. Instead, I think you're additions to the test exposed a race condition on the images actually loading. Notice that the height of the image that's supposed to stretch is also wrong. It's clearly not the C++ change that caused the failure. Also, it consistently passes on Chromium Linux release and is flaky on Chromium Linux debug: http://test-results.appspot.com/dashboards/flakiness_dashboard.html#showExpectations=true&tests=css3%2Fflexbox%2Fflexitem.html I'm not really sure where the race is though. We don't check the dimensions of the flex items until the body's onload event, which AFAIK should happen after all the images/iframes have loaded.

css3/flexbox/flexitem.html has been skipped in r126337 (https://bugs.webkit.org/show_bug.cgi?id=94723). Please revert r126337 when this issue is resolved. Thanks.

I can't get this to fail locally for me with a Chromium Linux debug build. :(

This test now passed for GTK and Chromium release. It fails or is flaky on all the Chromium debug bots. Seem like we should try unskipping it for other ports.

Specifically, we should unskip for mac and mark the Chromium failure as debug only. Qt and EFL were already skipping this test for other reasons, so I think we should leave those alone.

Created attachment 161285 [details] Patch

Committed r127050: <http://trac.webkit.org/changeset/127050>

(In reply to comment #9) > Committed r127050: <http://trac.webkit.org/changeset/127050> All the Mac platforms are now reporting this css3/flexbox/flexitem-stretch-image.html as an unexpected pass: http://build.webkit.org/results/Apple%20MountainLion%20Debug%20WK1%20(Tests)/r127166%20(374)/results.html http://build.webkit.org/results/Apple%20MountainLion%20Release%20WK1%20(Tests)/r127171%20(433)/results.html http://build.webkit.org/results/Apple%20MountainLion%20Debug%20WK2%20(Tests)/r127166%20(447)/results.html http://build.webkit.org/results/Apple%20MountainLion%20Release%20WK2%20(Tests)/r127171%20(439)/results.html http://build.webkit.org/results/Apple%20Lion%20Debug%20WK1%20(Tests)/r127166%20(2386)/results.html http://build.webkit.org/results/Apple%20Lion%20Release%20WK1%20(Tests)/r127171%20(3007)/results.html http://build.webkit.org/results/Apple%20Lion%20Debug%20WK2%20(Tests)/r127167%20(2977)/results.html http://build.webkit.org/results/Apple%20Lion%20Release%20WK2%20(Tests)/r127165%20(2770)/results.html I am going to remove it from platform/mac/TestExpectations

(In reply to comment #10) > (In reply to comment #9) > > Committed r127050: <http://trac.webkit.org/changeset/127050> > > All the Mac platforms are now reporting this css3/flexbox/flexitem-stretch-image.html as an unexpected pass: > http://build.webkit.org/results/Apple%20MountainLion%20Debug%20WK1%20(Tests)/r127166%20(374)/results.html > http://build.webkit.org/results/Apple%20MountainLion%20Release%20WK1%20(Tests)/r127171%20(433)/results.html > http://build.webkit.org/results/Apple%20MountainLion%20Debug%20WK2%20(Tests)/r127166%20(447)/results.html > http://build.webkit.org/results/Apple%20MountainLion%20Release%20WK2%20(Tests)/r127171%20(439)/results.html > http://build.webkit.org/results/Apple%20Lion%20Debug%20WK1%20(Tests)/r127166%20(2386)/results.html > http://build.webkit.org/results/Apple%20Lion%20Release%20WK1%20(Tests)/r127171%20(3007)/results.html > http://build.webkit.org/results/Apple%20Lion%20Debug%20WK2%20(Tests)/r127167%20(2977)/results.html > http://build.webkit.org/results/Apple%20Lion%20Release%20WK2%20(Tests)/r127165%20(2770)/results.html > > I am going to remove it from platform/mac/TestExpectations Sounds good. There's clearly a subtle race condition here. This same patch caused it to start failing on the Chromium Mac debug bots, which is what we were hoping for since the point of this test was to expose this race.