Created attachment 217691 [details] layout test Follow up after bug https://bugs.webkit.org/show_bug.cgi?id=124647, with a more complex layout test.
Created attachment 217739 [details] patch
I just ran into the regression caused by r159626 and this patch fixed it for me. It's a fairly severe regression since it breaks all hit testing when mixing Z transformed and non-Z-transformed elements not just hover. Either r159626 should be reverted or this patch should be landed (if it's OK). Currently trunk is broken for anyone who overlays 2D controls on top of 3D content.
Comment on attachment 217739 [details] patch I would like the changelog to explain why it's ok to remove the if (m_has3DTransformedDescendant) code.
Created attachment 219312 [details] patch
Comment on attachment 219312 [details] patch View in context: https://bugs.webkit.org/attachment.cgi?id=219312&action=review > LayoutTests/transforms/3d/hit-testing/hover-rotated-with-children-negative-z.html:61 > + <p>Test passes if the hover state of a transformed div is activated even if there's a non transfrormed element behind it.</p> Typo: transfrormed
Created attachment 219402 [details] patch for landing
Created attachment 219403 [details] patch for landing
Comment on attachment 219403 [details] patch for landing Clearing flags on attachment: 219403 Committed r160699: <http://trac.webkit.org/changeset/160699>
All reviewed patches have been landed. Closing bug.