Bug 11359 – Incomplete repaint of table cell's collapsed border when changing only the cell

• Status: RESOLVED
• Resolution: FIXED
• Component: Layout and Rendering

See the attached test case, where the "table half" of the border doesn't repaint in green when you click the Test button, until you force a repaint.

Created attachment 11150 [details]
Test case

Created attachment 12057 [details]
Override repaint rect computation for table cells

Includes change log and repaint test.

[Dave Hyatt]

Comment on attachment 12057 [details]
Override repaint rect computation for table cells

It seems like you aren't dealing with the case where a table cell has overflow and collapsed borders correctly.

For example:

<td><div style="width:100px">... put something in here that overflows...</div></td>

Created attachment 12072 [details]
Override repaint rect computation for table cells

Include horizontal overflow. I don't know if vertical overflow should ever occur. I managed to get vertical overflow but I think that's a bug.

(In reply to comment #4)
> I don't know if vertical overflow should ever
> occur. I managed to get vertical overflow but I think that's a bug.
> 

I guess it could happen with an inline-block that has overflow, but currently such overflow counts as "layout overflow".

[Dave Hyatt]

I might just write this code assuming that layout vertical overflow could occur just to be paranoid, since I think maybe there are weird degenerate span cases that might cause it to happen.

[Darin Adler]

Comment on attachment 12072 [details]
Override repaint rect computation for table cells

+            if (RenderTableCell* above = table()->cellAbove(this)) {
+            left = max(left, above->borderLeft(true));
+            right = max(right, above->borderRight(true));
+            }

Indenting mistake.

+    int borderLeft() const { return borderLeft(false); }

Why overloading instead of a default value for the parameter? This seems OK, but normally I'd only overload if I didn't want the parameter value inlined.

r=me

Created attachment 12078 [details]
Override repaint rect computation for table cells

Added vertical overflow and fixed indentation. I didn't want to add the 'outer' parameter to borderLeft() in the base class. Instead, now the function that takes a parameter (and is defined for RenderTableCell only) is named borderHalfLeft().

Comment on attachment 12072 [details]
Override repaint rect computation for table cells

Removing the r+ to keep this from showing in the commit queue (I thought obsoleting alone would do it).

[Darin Adler]

Comment on attachment 12078 [details]
Override repaint rect computation for table cells

r=me

[Alexey Proskuryakov]

Committed revision 18454.

[Alexey Proskuryakov]

This has caused an assertion failure in dom/html/level2/html/HTMLCollection01.html; reverted in r18455.

Comment on attachment 12078 [details]
Override repaint rect computation for table cells

Bad, bad patch. Clearing the review flag to protect the innocent.

Created attachment 12123 [details]
Override repaint rect computation for table cells

Fixed the crash by ignoring outside borders if the table grid is dirty. Added a long comment explaining why I think this is okay.

I named the getter needsSectionRecalc despite the setter being called setNeedSectionRecalc. I intend to clean up the naming in a future cleanup patch.

[Darin Adler]

Comment on attachment 12123 [details]
Override repaint rect computation for table cells

r=me again (I read your comment, and I believe it)

[David Kilzer (:ddkilzer)]

Committed revision 18516.