See attached testcase. It looks like converting buttons to new-flexbox made it behave differently with respect to floats.
Created attachment 190409 [details] testcase Firefox renders like old-webkit
What revision did this regress in? (i.e. qualify the "now" in the title).
It was bug 109994 aka r143643, on 2013-02-21 12:58:04
Note, this also affects twitter.com (the "New Tweet" button in the topright corner)
Created attachment 190601 [details] Patch
This patch does fix the problem, but I'd appreciate opinions on whether this is the right fix. I could instead add isRenderButton() to the if. Come to think of it, I should also add a test to css3/flexbox that doesn't use a button. New patch coming...
Created attachment 190602 [details] Patch
*** Bug 111241 has been marked as a duplicate of this bug. ***
Created attachment 191286 [details] Testcase for floats inside a flexitem
OK, including new-flexbox in this if is correct because flex boxes establish a formatting context which means they should expand to include floats.
Comment on attachment 190602 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=190602&action=review > Source/WebCore/ChangeLog:13 > + Also include new flexbox as a renderer that needs to enclose > + overhanging floats. Might be worth an extra sentence here to mention that the spec says that flexboxes establish a block formatting context.
Created attachment 191306 [details] Patch
*** Bug 111263 has been marked as a duplicate of this bug. ***
Comment on attachment 191306 [details] Patch Clearing flags on attachment: 191306 Committed r144706: <http://trac.webkit.org/changeset/144706>
All reviewed patches have been landed. Closing bug.