RE: Is There a Problem? (was: The Progress of CSS)

On Tue, 5 Jul 2005, Mark Birbeck wrote:
> > 
> > The CSS working group agrees, and indeed many years ago decided to 
> > split CSS into multiple modules to aleviate this problem.
> 
> But the modules need to be able to move through the process 
> independently of each other.

They can. (And are.)


> Also, a crucial part of my argument is that these smaller specs wouldn't 
> necessarily be maintained by the CSS WG; many small specs maintained by 
> the same people as before doesn't buy you anything in the nimble stakes.

Indeed. Web Fonts is maintained by the SVG group. Speech is maintained by 
someone who joined the CSSWG exclusively to do Speech. And so forth.

We're doing exactly what you're asking for. :-)


> > However, we still have to maintain our existing specs. If people raise 
> > issues on CSS2 (as they have been doing) then we're not going to 
> > ignore them, as that would just mean CSS2 was useless as a 
> > specification.
> 
> I understand, and that is a problem. But it shouldn't hold up the 
> future.

Well, it doesn't hold it up by design. It's a resource problem.

Maintaining existing specifications is a higher priority than making up 
new stuff. This is as it should be -- indeed maintaining existing specs is 
the most critical work any working group can do, since the issues raised 
on those specs are raised because the work is being implemented and used; 
by ignoring comments, one would be ensuring that interoperability of the 
existing specs will be poor.

Unfortunately, if a working group has X units of worktime per week, and 
dealing with maintaining existing specs takes X units of worktime per 
week, there is no time left for new work.

In the case of the CSSWG, maintaining CSS 2.1 takes all our resources. 
(There are times where we get more issues raised per week than we can 
actually resolve per week!) We often have no time left for CSS3. Despite 
this, some members of the WG are spending some of their own time working 
on CSS3 drafts anyway, beyond their official "20%" CSSWG time commitment.


> To pick an example, it surely can't be right that the XForms spec is 
> still waiting for a formal document that contains the xf:repeat 
> pseudo-elements and the various Model Item Property pseudo-classes. They 
> are not very complicated, and could easily have been produced on their 
> own.

What is holding up that draft going past CR and on to REC is simple: there 
is no test suite. I'm sure the CSSWG would be quite happy to have someone 
(e.g., you) come in and take section 4 of the CSS3 UI module and create an 
independent CSS3 module out of it, and push that through CR -- but it 
would get stuck at the same place it is stuck now, because there is no 
test suite.

Section 4 of CSS3 UI needs roughly 150 tests, and then two implementations 
which pass all those tests, before it can get past CR.

Nobody in the CSSWG has the time to write those tests at the moment. 
Personally my time is spent dealing with the CSS2.1 test suite (464 tests 
in our unpublished copy at the moment, 928 if you count the XHTML and HTML 
versions separately) and the Selectors test suite (303 tests, 786 if you 
count the XHTML, HTML, and XML variants separately).

You are of course welcome to contribute tests to the test suite. It would 
directly help in bringing the draft to REC.


> But the thing I think is really missing is that we don't have a way of 
> 'automatically' generating properties in a way that would save a lot of 
> work. For example, the SSML specification [1] could easily have produced 
> a parallel document that brought the CSS speech properties right 
> up-to-date.

I'm not sure what you mean by "'automatically' generating properties". 
I'm sure the SSML group would have been welcome to write a CSS3 module if 
they were so inclined.


> I see no need for the CSS WG to 'own' the resulting documents 
> themselves.

I agree.


> (Those writing other standards can add XPointer schemes, XPath 
> functions, schema data types, XML namespaces, and so on, without 
> consulting some other Working Group. Yet I can't add a simple styling 
> property in anything less than three years.)

I don't see how this is the case. The SVG group seems to be quite happily 
adding CSS properties left right and centre (somewhat to the concern of 
the CSS working group, actually! Not all the additions seemed to make 
sense. I believe the CSS and SVG groups have now reached a compromise on 
which properties make sense in CSS and which don't).

There is nothing stopping a working group from working with the CSS 
working group on adding features to CSS.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Tuesday, 5 July 2005 13:36:40 UTC