Windows and Qt fail new editing/style/remove-nested-inline-styles.html test http://build.webkit.org/results/Windows%20Release%20(Tests)/r50133%20(5506)/editing/style/remove-nested-inline-styles-pretty-diff.html http://build.webkit.org/results/Qt%20Linux%20Release/r50134%20(3134)/editing/style/remove-nested-inline-styles-pretty-diff.html
This is due to the bug 30784. My patch for 30784 (which has been r- for various reasons) should also fix this bug.
To clarify, on very platform except Mac, executeToggleStyle will check if the entire selection has the specified style. But because of the bug 30784 (nested b'tag result in tri state of false), the selectionHasStyle returns false in Windows and Qt platforms.
So what's your suggestion for how we should resolve this? We can't leave tests red on the bots. :( Do Windows/Qt need platform-specific results? Should the "failing" section of the test be split out into a mac-only test?
(In reply to comment #3) > So what's your suggestion for how we should resolve this? We can't leave tests > red on the bots. :( > > Do Windows/Qt need platform-specific results? Should the "failing" section of > the test be split out into a mac-only test? I have fixed my patch for the bug 30784. Could you take a look? This should fix Windows build unless I misunderstood the problem and I'm not reproducing the error properly. Around L166 of EditorCommand.cpp inside executeToggleStyle, there is a if-statement that changes behavior for Mac. I have confirmed that this bug is demonstrated when I forced to run the statement in else clause, and is fixed with my patch. I'd like to confirm this on Windows but unfortunately, I don't have access to Windows machine with WebKit check out.
This regression has been fixed in http://trac.webkit.org/changeset/50172.