[Webkit-unassigned] [Bug 74864] There is additional space not belonged to a table between the table cells

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=74864





--- Comment #14 from Arpita Bahuguna <arpitabahuguna at gmail.com>  2012-08-06 04:29:03 PST ---
(In reply to comment #12)
> > (In reply to comment #8)
> > Also, I was looking at perhaps raising another bug which removes the following FIXME from the code (in RenderTableSection::spannedRows()):
> > 
> >  // FIXME: Upper_bound might not be the correct algorithm here since it might skip empty rows, but it is
> >     // consistent with behavior in the former point based hit-testing (but inconsistent with spannedColumns).
> > 
> > I could add testcases with empty <TR>s to verify upper_bound() algorithm's behavior.
> 
> Make a test-case verifying or invalidating the problem would be great.
> 
> For this patch though, a guess the comment actually needs to be updated to remove the claim that it is inconsistant with spannedColumns, since you changed that.
> 
I agree, shall take care of updating this comment as well.

> But talking about what happens with empty-cells made me take another look at your empty-cell test-case. Since 'empty-cells' default to 'show' which means td3 and td4 renders their own border, doesn't your test-case actually demonstrating a bug when it says elementAtPoint on the borders of td3 hits td5? And a really big bug when hitting the td3 border from td4 suddenly hits td2?
> 
> Okay, so there are no borders on the table which is why it is correct. I guess that demonstrates the concern we raised about collapsed rows works correctly with collapsed cells.
>
I verified the empty-columns testcase, modifying it by adding borders for all table-cells. Even in that case our behavior remains consistent with that of other browsers (Opera, FF). The o/p on all browsers was the following:
Executing for element: td1
Node at edge-1: td1
Node at edge: 
Node at edge+1: 

Executing for element: td2
Node at edge-1: td2
Node at edge: 
Node at edge+1: 

Executing for element: td3
Node at edge-1: td3
Node at edge: 
Node at edge+1: 

Executing for element: td4
Node at edge-1: td4
Node at edge: 
Node at edge+1: 

Executing for element: td5
Node at edge-1: td5
Node at edge: 
Node at edge+1: 
So I suppose the change to upper_bound() does not affect in the spannedColumns() case. I need to verify for spannedRows() though.

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