Thanks for the response.

Look at the case

<table style="border-collapse:collapse">
<tr>
<td style="border 4px green">1</td>
<td rowspan=2 "border 6px blue">2</td>
</tr>
<td "border 8px yellow">3</td>
</tr>
</table>

represented in ascii fixed-pitch font as:

------------------
|  1   |         |
--------    2    |
|  3   |         |
------------------

The CSS2 spec says that one half of the computed border width between cells belongs to each cell.  In resolving conflicts, a thicker border wins over a thinner border.

So the border between cell 1 and cell 2 is computed to be 6px blue.  Cell 1 gets a right border width of 3px, and cell 2 gets a left border width of 3px.

Then we run the same algorithm for the border between cell 3 and cell 2.  This time, cell 3's 8px yellow border wins.  So each cell should get a 4px border on the common edge.

Cell 2 clearly can't have a border that is both 3px and 4px wide.  That would give the content of cell 2 a non-rectangular area to lay out into which the spec does not support.  Playing games with cell 2's dimesions to make this rational seems wrong.

So what is the answer?  One answer is to assign the entire edge 8px yellow to be divided equally between all common cells, so cell 1's right border becomes 4px yellow.

Another answer is to assign cell 2's left border 4px yellow but not to change cell 1's 3px blue border that we've already computed.

Sorry, I don't see anything in the URL you reference that addresses this.  Am I missing something?
 

Steve

"L. David Baron" wrote:

>From www-style-request@w3.org  Fri Dec 11 00:25:30 1998
>From: buster@netscape.com (Steve Clark)
>Subject: CSS2 table collapsing borders -- some implementation questions

>I am implementing CSS2 table collapsing borders for Netscape's new
>"Gecko" browser.  I have some questions about border resolution, and
>particularly about how row and column spanning cells effect border
>resolution for adjacent cells.  If this topic has already been
>discussed, I would appreciate a pointer to the result.  Otherwise, I
>would like to start a dialog with any knowledgable folks familiar with
>this part of the spec.

I don't think this has been discussed before.  To try to answer what I
know about your questions, I don't see any problem with applying the
rules for border conflict resolution to row and column spanning cells
the same way they are applied to any other cells.  The conflict should
be resolved according to the rules in the spec for all borders that
could be drawn on the segment in question.  Is there something I am
missing?

The example in
http://www.w3.org/TR/REC-CSS2/tables.html#table-layers
is the only example in the spec involving rowspan or colspan.  This may
help.

David Baron

--------------------------------------------------------------------
L. David Baron         | Freshman, Harvard
dbaron@fas.harvard.edu | < http://www.fas.harvard.edu/~dbaron/ >
--------------------------------------------------------------------