[Webkit-unassigned] [Bug 94936] Incomplete rect-based hit-test result when hit-test region overlaps layers with different z-order.

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=94936





--- Comment #3 from Kevin Ellis <kevers at chromium.org>  2012-08-24 10:23:03 PST ---
(In reply to comment #1)
> I thought RenderLayer::hitTestLayer already handled that. A layer is only considered hit (and thus terminating the search) if an element fully encloses the hit-tested touch-point. 
> 
> This behaviour does create another bug of returning too many elements, since the areas that have been hit are not subtracted from the hit-test and can continue to hit elements in layers below.

Will write up a test case demonstrating the problem.  Here is a brief description of the page layout.  Have a series of absolute positioned elements each covering the full page, but with the active page having a higher z-index.  Have a left side bar that overlaps the page with a higher z-index.  The active page contains checkboxes that are closely aligned to the navigation bar.

I misinterpreted some of the diagnostic output in my original test.  When the touch overlaps a checkbox but clips the navigation bar, we see the navigation bar and active page in the hit test results, but not the checkbox.  So the results are not restricted to the highest z-index element as I originally thought; however, the search is not picking up all of the elements that overlap the hit-test area.  The hit-test scores are pretty meaningless given the size of the elements involved and it is simply pulling to the smaller one (the navigation bar) based on the relative scores.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.