Re: [CSS21] New Candidate Recommendation

Ara Pehlivanian wrote:
> 
> On 7/20/07, Bert Bos <bert@w3.org> wrote:
>>
>> The CSS WG published the new Candidate Recommendation (CR) for CSS
>> level 2 revision 1 with the firm intention that there won't be any more
>> working drafts.
> 
> At the risk of sounding cynical, strictly from a layman's point of
> view who works with CSS on a daily basis, may I say that the Emperor
> has no clothes? What relevance does a new Candidate Recommendation for
> CSS 2.1 have 5 years after the first draft? We're all out here in the
> trenches waiting for a decent implementation of the thing and hoping
> against hope for progress on CSS 3, and instead of the W3C moving its
> collective a** and getting on with it already, CSS 2.1 is still
> inching along 5 years on? What's the point? I mean really?

A decent set of *interoperable* implementations of the thing requires
either a clear, unambiguous, implementable spec or a good reference
implementation one can reverse-engineer. We don't have the latter, and
CSS2.1 is our best effort at the former. Unlike the old HTMLWG, the
CSSWG did not choose to abandon maintenance of its existing spec, CSS2.
I think it would be irresponsible of us to ignore problems people point
out in the spec.

CSS3 is a big project, and CSS2.1 forms the foundation for it. Ignoring
its problems means ignoring problems in CSS3. Because of the way CSS
works and the way the Web works, there is no versioning in CSS. CSS3
does not replace CSS2.1, it expands it. The HTML5 effort is doing the
same thing for HTML, but since the HTML4 spec is so inadequate, they
are rewriting it as the core part of HTML5. For us, since CSS2.1 as a
spec is in reasonable shape, it makes more sense to maintain CSS2.1 as
the core and build the CSS3 modules on top of it. This way we always
have an up-to-date, complete version of CSS for implementors to refer
to, and a place where we can integrate into the specs the solutions we
draft for problems people bring up. E.g. suppose someone brings up a
problem with the tables section. If we don't address it in CSS2.1, it
will be years before the fix makes its way into any spec: nobody is
working on CSS3 Tables right now. Little issues like the empty inlines
one Anne brought up in this last round of editing are what make or
"break the web". An author of new pages might not care, but implementors
need to consider existing content and avoid changing their behavior so
much that significant amounts of existing content renders poorly in
their implementation. If the spec is inconsistent with that requirement,
it is not fulfilling its function. Also, if the spec is ambiguous, it
is not fulfilling its function: it is not useful to authors if IE,
Mozilla, and Opera each interpret the same feature definition differently.

> Things happen really quick on the 'net... and waiting around for half
> a decade for a spec is... to put it lightly... frustrating.

One could say the same thing about implementations of said spec, about
the test suite for said spec, etc.

~fantasai

Received on Friday, 20 July 2007 23:25:42 UTC