The patch associated with https://bugs.webkit.org/show_bug.cgi?id=55900 causes a regression on 3D CSS on Linux, Mac, and Windows. To repro on my machine, I go to http://www.webkit.org/blog-files/3d-transforms/poster-circle.html : only half of the rotating figure appears.
Looking.
FYI this patch landed after the 696 branch point, so m11 should be unaffected.
The fix is easy (I forgot to initialize m_doubleSided in LayerChromium) - but the more interesting thing is that apparently we have no layout test coverage for double sided layers. D'oh! Will construct a proper test and then post a patch.
Created attachment 85234 [details] Patch
Patch fixes the bug and adds a test. The -5000px offset on the text is ugly, but it's the only way I could figure out to put something informative in the -expected.txt without getting text into the pixel output. It'd be nice if we had a more standard way to do this sort of thing.
Comment on attachment 85234 [details] Patch Nice catch and test case.
Committed r80665: <http://trac.webkit.org/changeset/80665>