Re: Proposal: Simply specify a XHTML lax, or XHTML+RDF Namespace and Schema

Joseph Reagle <reagle@w3.org> wrote:

> To put a point on Mimasa's work with a potentially rash proposal: why not 
> just specify a XHTML+RDF or XHTML-lax namespace and schema and ask people 
> to use that? Problem solved!?
> 
> Mimasa has already provided the XHTML-lax schema, I even have it running as 
> an (temporary) experimental validator at:
>   http://policy.w3.org/xsd-xhtml1-validate

Good.  Actually with Namespace Routing Language (NRL) [1], we could do
a bit better than "lax" validation.  I put together a couple of
XHTML+RDF schemata in NRL:

   XHTML 1.0 Strict + RDF
     http://www.w3.org/People/mimasa/test/schemas/nrl/xhtml1s-rdf.nrl

   XHTML 1.0 Transitional + RDF
     http://www.w3.org/People/mimasa/test/schemas/nrl/xhtml1t-rdf.nrl

   XHTML 1.0 Frameset + RDF
     http://www.w3.org/People/mimasa/test/schemas/nrl/xhtml1f-rdf.nrl

   XHTML 2.0 + RDF
     http://www.w3.org/People/mimasa/test/schemas/nrl/xhtml2-rdf.nrl

XHTML 1.0 + RDF schemata use XML Schema for validating XHTML part and
RELAX NG for RDF/XML part.  In this case the XHTML schema is a "closed"
schema as opposed to "open" schema I proposed earlier.  As such, you
cannot put anything other than RDF in XHTML.

On the other hand, inside RDF, syntax is checked against the RELAX NG
schema for RDF/XML as described in Appendix A of the RDF/XML Syntax
Specification but RDF can contain elements from arbitrary namespaces,
so any namespace is allowed inside RDF.

XHTML 2.0 + RDF schema uses RELAX NG for both XHTML 2.0 and RDF, and
XHTML 2.0 schema is also "closed".  If it's really necesssary, an NRL
schema could be written to accept arbitrary namespaces in addition to
RDF/XML.

Unless the RDF Core WG is willing to develop schema(ta) for RDF/XML in
addition to the non-normative RELAX NG version as currently described,
I don't think we can do more, for the purpose of validating XHTML with
embedded RDF/XML.

> What objection would the HTML or RDF communities likely raise? If none, is 
> it just a matter of assigning the action to do it, and to find the 
> resources to maintain a validator service?

NRL is an input to ISO/IEC DSDL Part 4 (a.k.a. VCSL) and I'm reasonably
confident that VCSL would allow us to do something similar.  I'm happy
to maintain those schemata and keep an eye for the development of DSDL,
and provide VCSL version when it becomes an International Standard.
Whether we want a specific validator service for it is another question
and I cannot answer to that question unilaterally.

Apart from XHTML+RDF issue, a validator service based on Jing [2] would
be an interesting idea, as now that it supports RELAX NG, XML Schema,
Schematron and NRL, so it provides greater opportunity for validation.

[1] http://www.thaiopensource.com/relaxng/nrl.html
[2] http://www.thaiopensource.com/relaxng/jing.html

Regards,
-- 
Masayasu Ishikawa / mimasa@w3.org
W3C - World Wide Web Consortium

Received on Saturday, 21 June 2003 01:30:51 UTC