Re: [CSS21] response to issue 151

On Tue, 17 Feb 2004, Jukka K. Korpela wrote:
>
> Is the heart of the matter the question whether the rules that map markup
> construct to stylistic features is (at least conceptually) part of a
> browser's default style sheet or a fictitious prefix to the page's style
> sheet?

Yes. This is the only matter, in fact. Most of the discussion around this
is largely knee-jerk reactions concerned with making political statements
about whether presentational attributes are a good or bad thing.


> If it is, I think it is questionable whether the distinction matters, or
> can realistically be implemented and maintained not only in browsers but
> also in authors' minds.

Yes, it is implementable, and implemented, and will be interoperably
implemented before CSS2.1 exits CR.

I don't think this really particularly matters much to authors, since few
authors will ever be able to tell the difference (the only way I can think
of to actually notice this would be the markup's different interactions
with a user stylesheet).


> What really matters in practice is how markup language definitions and
> browsers define and describe exactly how markup is mapped to CSS (in
> reality or conceptually). This won't be easy, but to make it more
> feasible, the logic should be as simple as possible.

HTML doesn't define presentational semantics (much). If you are trying to
say that HTML should define exact mappings of deprecated HTML
presentational syntax into CSS, then that is an issue for the HTMLWG to
examine, not the CSSWG.

I believe the SVGWG has defined this in detail for SVG, for instance.


> So I think the whole of 6.4.4 should be reduced to simple statements: If
> user agents treat markup elements and attributes in a manner that
> affects the rendering, they should do this in a documented manner. Any
> features of presentation that are describable in CSS terms should be so
> described in the documentation as CSS rules and be treated as part of
> the user agent's style sheet. The definitions of markup languages should
> describe the same way the correspondence of markup and CSS, to the
> extent that presentation describable in CSS is part of the meaning of an
> element or attribute.

That would, IMHO, be beyond the scope of 6.4.4, which is just saying
where in the cascade such a mapping should go (namely the author level
for legacy HTML content, and the UA sheet for everything else).

-- 
Ian Hickson                                      )\._.,--....,'``.    fL
U+1047E                                         /,   _.. \   _\  ;`._ ,.
http://index.hixie.ch/                         `._.-(,_..'--(,_..'`-.;.'

Received on Tuesday, 17 February 2004 10:17:49 UTC