Re: Validating XML on the Validator - Logo please?

# Ignore this line: ikke sensitiv (not classified)

>Nothing is missing! The W3C validator only validates XML,
>not XML Schemas (yet).

Rigth, so it's not doing anything wrong (no bug), it's just not ckecking the
linked Schemas as it should (a limitation)? It's not choking on something
that would have been valid had Schemas been supported?


>Yes: maybe he's busy. I emailed him directly and he asked me 
>to post it to this list!!!

Yeah, Gerald has been pretty busy lately and the Validator has languished a
bit as a result. After a year when I couldn't, for practical (IRL) reasons,
contribute much, I'm now in the process of trying to pick up some of the
slack by submitting patches[0]. We'll see what I can manage where Schemas
are concerned.

It really hinges on finding an external parser that groks this stuff.
Reimplemeneting a full, or even partial, XML Parser is not in the cards.
However, it may be possible to work around it with the existing parser as
each individual file (the *.xml, the *.<whatever the Schema extension is>,
etc.) should be valid XML, right? That means you could just manually feed
them to a validating parser one at a time and consolidate the results.


>>This would imply directly groking XML Schemas, no?
>
>Exactly

I was afraid of that. The Validator relies on external parsers for all the
hard work, and just puts a nice (no snide comments from the peanut gallery,
thank you! ;D) interface on top of it. No XML Parser that I'm aware of
actually supports XHTML-Mod/XML-Schema as of yet.

The parser used by XSV is some kind of weird semi-Closed Source thing. I
dunno what the license terms are for the copy XSV uses, but I do know that
you can't download it, you have to ask them nicely for an
Educational/Research Licence. The W3C may be able to work out some form of
licence with them, but I sure can't; at least not and redistribute derived
works (i.e. a patch to the validator) afterwards.

It also comes with only experimental language bindings and only in Python
(ye gods, but I *hate* Python!). The only thing that makes sense without
some *major* restructuring (read: build from scratch) is something that can
be easily called from Perl (which the Validator is written in).


>>>cf http://xhtml.waptechinfo.com/extxhtml/,
>>>   http://xhtml.waptechinfo.com/exp/
>>
>>Yes, those both validate. Shouldn't they?
>>Am I missing something really obvious here?
>>Everyone off sniggering at me?
>
>*snigger*

Ah, I *knew* that wasn't just my rampant paranoia[1] talking! :-)


>you're not meant to validate them, you're supposed 
>to read them for more information! They will inform
>you a bit about XML Schemas and XHTML Families (I hope).

Ah! Right.

Actually, I'd already read them. I was watching in fascinated horror as the
"comment" thread unfolded on www-html. Dan's comment-hack pages exposed Yet
Another XHTML Bug in the Validator (see [0]), but otherwise seemed to work
fine. Those pages are actually one of my regression tests! :-)

I'll read up on this stuff and see if I can figure it out. No promises,
though. :-)


[0] - I'd planned to release something last weekend, and the
      weekend before, but Real Life kept interfering. I'm
      shooting for this weekend again, but it hinges on what
      happens IRL and the turnaround-time for patches at the
      W3C isn't exactly stellar. The good news is that I have
      ";imgonly" support, as well as experimental support for
      XML/text/etc. output, almost done. The infrastructure is
      in place (barring the discovery of insurmountable hurtles)
      so all that remains is the nitty details. The curious can
      follow the sordid process in my Diary on Avogato
      <URL:http://advogato.org/person/link/>.

[1] - In a past life I used to run Security in a 3.5K-user NT shop
      (a regional hospital). It's not a question of whether you're
      paranoid; it's a question of whether you're paranoid *enough*!
      :-|

Received on Tuesday, 29 August 2000 21:28:52 UTC