[Webkit-unassigned] [Bug 84678] [GTK] spatial navigation failures

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=84678





--- Comment #3 from Zan Dobersek <zandobersek at gmail.com>  2012-06-25 09:06:51 PST ---
I debugged this a bit, specifically on the fast/spatial-navigation/snav-container-white-space.html test case.

As evident from the output, the problem is that when the test wants to move focus down from the element with id 1 to element with id 2, this element is skipped and the focus shifts to element with id 3.

It turns out that because the <a> element with id 2 is inside a <div> element[1], this, combined with enabled Cairo metrics hinting, makes the <a> elements with ids 2 and 3 intersect by a tiny rect of size (114|1) at position (8|86). With the <a> element with id 3 being on top, this makes the focus-controlling algorithm[2] mark that node as the closest node to which the focus is shifted.

The test for example passes if the parent <div> element of the <a> element with id 2 is added a padding: 2px; style or is just removed.
The test also passes in GtkLauncher.


[1] - http://trac.webkit.org/browser/trunk/LayoutTests/fast/spatial-navigation/snav-container-white-space.html#L76
[2] - http://trac.webkit.org/browser/trunk/Source/WebCore/page/FocusController.cpp#L739

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