in https://bugs.webkit.org/show_bug.cgi?id=18662#c20 , darin expoited this: <quoted> (...) for (Frame* frame = renderer->document()->frame(); frame; frame = frame->tree()->parent()) { if (HTMLFrameOwnerElement* ownerElement = frame->ownerElement()) rect.move(ownerElement->offsetLeft(), ownerElement->offsetTop()); } It's not correct to use the offsetLeft and offsetTop of the frame owner element's renderer because that's just the distance from the offsetParent, not the absolute position. Instead you should use one of the localToAbsolute family of functions for this. (...) </quoted> this is a real problem needs fixing
Created attachment 56618 [details] patch v1 patch does not use the localToAbsolute family, but works pretty much. Darin, do you mind having a look?
Ping review
Created attachment 57334 [details] patch v1.1 Same patch as v1, but now with an understandable ChangeLog text.
Comment on attachment 57334 [details] patch v1.1 new patch coming soon ...
Created attachment 58641 [details] (committed in r61250, reviewed by kenneth)patch v1.2 same as patch v1.1, applicable by the bots and rebased against tot.
Simon/kenneth, this should be a simple review. Can any of you help? Details in bug descritop https://bugs.webkit.org/show_bug.cgi?id=39439#c0
> Details in bug descritop https://bugs.webkit.org/show_bug.cgi?id=39439#c0 err, s/descritop/description
Comment on attachment 58641 [details] (committed in r61250, reviewed by kenneth)patch v1.2 Clearing flags on attachment: 58641 Committed r61250: <http://trac.webkit.org/changeset/61250>
Revision r61250 cherry-picked into qtwebkit-2.0 with commit 20f69610e4b79964c79d0f5382e8282fde476f28