I removed the virtual getBBox() code in SVGPathElement.cpp which was incorrect and resulted in a test breaking. In the future I will better rely on ChangeLogs, as Tim Horton's r99460 was clear about why this code was organized this way: the boundingRect() call in SVGPathElement returns the _tight_ bounds so that paths don't use the loose/fast bounding rect from m_objectBoundingBox. This test was already skipped on Chromium for other reasons (see: https://bugs.webkit.org/show_bug.cgi?id=65939) and I failed to catch this regression locally.
Created attachment 160582 [details] Roll out r126056 and r126626
Comment on attachment 160582 [details] Roll out r126056 and r126626 Clearing flags on attachment: 160582 Committed r126693: <http://trac.webkit.org/changeset/126693>
All reviewed patches have been landed. Closing bug.