Re: [CSS2.1] col attributes: XHTML and CSS inconsistency?

On Tue, Dec 30, 2008 at 2:46 PM, Simetrical <simetrical@gmail.com> wrote:
> On Tue, Dec 30, 2008 at 3:39 PM, Tab Atkins Jr. <jackalmage@gmail.com> wrote:
>> You say that :nth-col() would allow us to apply arbitrary properties
>> to cells.  Apparently this means that pseudoclass resolution happens
>> (or at least, *can* happen) at some time after table layout.
>
> Where "table layout" means "parsing the HTML table elements and
> determining the semantic relationship between them without considering
> any CSS rules", yes.
>
>> My
>> question is if it is possible to cause this same sort of magic to
>> happen automatically.  It doesn't depend on html tables at all - my
>> question applies to any grouping of table-column and table-cell
>> elements.
>
> The point is that this syntax would *not* apply to arbitrary
> table-column elements.  It would *only* apply to HTML <col> (or,
> presumably, equivalent syntax in other markup languages).  This is why
> it actually works, because the HTML structure is known before CSS
> starts working.
>
>> During the initial pass through the CSS engine, it doesn't
>> know what cells are in what column.  Afterwards, though, it *does*,
>> and can then cause some special table-magic inheritance.  It wouldn't
>> be any more magic than applying this pseudoclass, would it?  After
>> initial style resolution, trickle table-column styles down to their
>> corresponding table-cells.
>
> That's precisely what this proposal *doesn't* do.

Ah, that's much clearer.  I supposed that :nth-col() and/or :col()
would be for tables, not just <table>s (and any similar construct in
markup languages browsers decide to support which can unambiguously
determine column-identity purely from the markup).

If this is the case, then I withdraw any arguments I made completely.
This makes perfect sense and is unobjectionable.

~TJ