On current chromium builds, we're seeing EventHandler::hitTestResultAtPoint get the wrong location when inside an iframe that is scrolled (behaves as if the iframe scroll position was at the top).
I suspect r127457 <http://trac.webkit.org/changeset/127457>, testing that now and working on a layout test.
Yep, works fine if I revert r127457. I'll upload a layout test that demonstrates the issue. Allen, do you have time to take this? We're branching for Chrome M23 this weekend, so we'd like a fix (or revert) this week.
That makes sense. The new code does not adjust for the scrollOffset in FrameView(). It should be simple to fix though by adding frameView()->scrollOffset() to the point passed to the child frame.
It would be nice also add a test for it.
(In reply to comment #1)
> Yep, works fine if I revert r127457. I'll upload a layout test that demonstrates the issue. Allen, do you have time to take this? We're branching for Chrome M23 this weekend, so we'd like a fix (or revert) this week.
I will fix it tomorrow if that is okay with you?
Thanks Allan. It looks simple enough, I'll take a crack at uploading a fix for you to review. If I don't have the CL in good enough shape tomorrow, then given the urgency, I'll gladly hand it off to you.
Created attachment 163722 [details]
Looks like I'm not going to get this done today - I'm a little confused about the co-ordinate systems and the right place to apply the scroll offset, will need to dig deeper to understand it fully. So I'll just reassign to Allan.
I've included a layout test that demonstrates the problem (I've verified it fails in ToT, but passes with r127457 reverted).
Created attachment 163816 [details]
Comment on attachment 163816 [details]
Attachment 163816 [details] did not pass chromium-ews (chromium-xvfb):
New failing tests:
Created attachment 163834 [details]
Do not use FrameView locations since it is relative to parent frameview, instead calculate the location based on border and padding.
Thanks Allen. I've verified this patch fixes the original issue as well.
Comment on attachment 163834 [details]
Clearing flags on attachment: 163834
Committed r128462: <http://trac.webkit.org/changeset/128462>
All reviewed patches have been landed. Closing bug.
Please include the revision that regressed in the bug title next time. That helps find out what a change broke without having to do trips to bugzilla.