[webkit-reviews] review denied: [Bug 85849] Incorrect rect-based hit-test result for replaced elements : [Attachment 141817] Patch

[bugzilla-daemon at webkit.org]

Dave Hyatt <hyatt at apple.com> has denied Raymes Khoury <raymes at google.com>'s
request for review:
Bug 85849: Incorrect rect-based hit-test result for replaced elements
https://bugs.webkit.org/show_bug.cgi?id=85849

Attachment 141817: Patch
https://bugs.webkit.org/attachment.cgi?id=141817&action=review

------- Additional Comments from Dave Hyatt <hyatt at apple.com>
This patch is closer, but now it will do the wrong thing with images/plug-ins,
etc. when the mouse actually is inside where the enclosing inlines would be.

This is a performance optimization essentially to avoid creating line boxes for
inline flows that are detected to match some enclosing parent. What this means
is the answer you give on a hit test has to be identical to the answer that
would have been given if the performance optimization were not in effect.

What is checked in currently is wrong, and all the patches that have been
produced are also wrong. However, the latest patch is probably the "least
wrong" of all of them, so in that sense it's an improvement. I don't mind
landing this change, but it needs to be accompanied by a follow-up bug and
FIXMEs in the code, so that we can actually fix this the right way in the
future.

The right way to fix the bug is that you have to actually determine where the
culled inline is (we have code elsewhere that does this sort of thing, e.g.,the
rect/quad collection methods in RenderInline) and do an accurate test for
intersection.