Re: "Re: [CSS21] rules="all" attribute specification and border-collapse property"

Patrick Garies wrote:
> Apparently, this didn't get sent to the list since I made a mistake in 
> the CC field. I've forwarded it to www-style@w3.org.
> 
> Sylvain Galineau wrote:
>> I am asking : what about IE8's rendering is less logical and/or worse
>> than that implemented by two other browser vendors ? I'm fine with you 
>> singling out IE as long as I understand what is specific to IE for the 
>> issue being reported.
> 
> I just viewed Gérard's test case [1] in the latest releases of Mozilla
> Firefox, Opera, Safari, and Windows Internet Explorer and WIE8 is the
> only browser of the four to render the table without borders around the
> table cells. However, if I turn on Compatibility View, the borders
> suddenly materialize.
> ...
> Finally, this appears to be a violation of a SHOULD in the HTML 4.01
> spec; section 11.3.1 implies that the rendering of Gérard's test case
> [1] should be identical between |border="4"| and |border="4"| with
> |rules="all"| since the former implies the latter: "Other values of
> border imply |frame="border"| and, unless otherwise specified,
> |rules="all"|". In checking all of the previously named browsers, the
> only browser that is inconsistent with this part of the spec is WIE8.
> [1] http://www.gtalbot.org/BrowserBugsSection/MSIE8Bugs/TableRulesAll-BCSeparate.html 

Sylvain,
I agree with Patrick Garies that this is wrong. Whether the border-collapse
model is used or not is, imho, out-of-scope for the spec, but there should
be borders rendered here. Probably what is happening is that your team has
defined rules="all" to be rules="cols" + rules="rows", i.e. defined the
borders to be applied to the rows and columns, rather than applied them to
the cells. Since the separate borders model cannot render borders applied
to rows and columns, they disappear when Gérard sets border-collapse: separate.
It's unfortunate that the separate borders model cannot handle borders on
rows and columns, but at least in the rules="all" case we don't need to be
limited by this problem.

~fantasai

Received on Monday, 4 May 2009 17:55:51 UTC