DTD catalogues (Was: XHTML 1.1 validator rejects xml:space attributes)

Hi Shane, all

On 11 oct. 07, at 11:18, Shane McCarron wrote:
> Independent of this issue.... it is certainly reasonable to have the  
> validator use locally cached versions of things referenced by  
> documents being validated.

Yes, I agree. For better or worse though, the sgml parser used by the  
validator is using a catalogue solution and I suspect adding a caching  
capability to it would be a little hard.

>> Are these DTDs changing so often that it would be a burden for the  
>> XHTML WG to drop a mail to www-validator saying "the DTD for foo  
>> has been updated, please update your catalogs"? Remember, readers  
>> on this list include not just developers for the W3C's validator,  
>> but pretty much all of them.
> It is surely not a burden.  We manage a great number of DTDs, and  
> many of those are under active development right now.  I am not sure  
> how often you can handle updates, nor when it would be appropriate  
> to update something as public as the validator.

As a matter of fact, I think we will have a small update of the  
validator soon: just a couple of bug fixes that are due - overdue -  
for inclusion in the stable distribution. I was wondering if that  
would be a good time to update the sgml-lib used by and distributed  
with the markup validator to use the latest version of dtd and modules  
de xhtml modularization, 1.1, basic and print? I attach a bulky diff  
between the version of DTDs in http://www.w3.org/MarkUp/DTD/ and the  
ones currently in the validator. Could we work together on this update?

>> That said, if a DTD is changing a lot for a period of time, we  
>> could remove it temporarily, upon your request, from the  
>> validator's catalog, but that should not be systematic.
> I would be happy to work with you on developing a process that makes  
> sense for all groups; surely this cannot be unique to XHTML.

I think XHTML is pretty much the only technology still changing/ 
developing its DTDs. But indeed, a process to handle the evolution of  
schemas and their inclusion in our tools would be extremely useful for  
now and for the future. Our current process,rather crude, implies that  
a spec goes to CR, PR, REC and then stays there - not the reality any  
more. The process was also made for a  time when validation was  
provided after technologies went to REC, which we now want to change  
to allow validation since the early days - for early adoption.

The way I have found to handle this is to add "awareness" of the  
document types early on, but only add the schemas to catalogue at a  
later stage, when they become stable. This, however, has not worked  
well with XHTML mod and friends since these have been re-worked on  
since REC.

Any help making the process more agile - maybe, as I wrote earlier,  
simply removing files from the catalogue when they go from stable to  
"under work" status - would be welcome.
-- 
olivier