Re: Validating XHTML with embedded RDF

Sorry, I was traveling for an HTML WG FtF meeting and couldn't read
e-mail regularly.

Joseph Reagle <reagle@w3.org> wrote:

> > And when it comes to embedding
> > RDF/XML in XHTML, I have no hope to work out a DTD-based solution.
> 
> Do you mean technically, or from an ecology point of view?

Both.

> If I understood 
> Sean's proposal in [a] it appearred possible to write a RDF-in-XHTML DTD 
> Modularization.

Note that I was specifically talking about RDF/XML, as defined by
the RDF/XML Syntax Specification.  It is apparent that the current
RDF/XML syntax is not fully expressible by DTD nor XML Schema, and
the RDF Core WG seems to have rejected the XML Schema WG's comments
to develop a single canonical syntax which could be more XML Schema
or DTD friendly [13] as "beyond the scope of RDF Core's current
charter".  So, I see no chance to define a generic RDF/XML DTD, in
which case there's no chance to define a generic RDF/XML-in-XHTML DTD.

Of course, if we agree to restrict the syntax of the XML serialization
of RDF, we could define something, as Sean illustrated.  But it's not
RDF/XML as currently defined.

> It's unsavory, from an ecology point of view, because one 
> doesn't want to do this for every XML combination. However, in the shorter 
> term, this task force needn't solve that larger problem. Could the W3C 
> issue a RDF+XHTML DTD modularization while we also worked on a solution to 
> the larger problem?
> 
> [a] http://infomesh.net/2002/rdfinhtml/#embedAndValidate

That largely depends on the RDF Core WG.  The HTML WG can offer help
to combine DTD modules, once an RDF DTD module becomes available.
However, the HTML WG is not the right group to decide which sort of
(restricted) serialization format for RDF is appropriate.  We need
input from RDF folks.

> Do you mean DSDL VCSL would provide a better solution to the "semantics" 
> problem? If so, I've looked at [a] and don't follow. 
> 
> [a] http://www.y12.doe.gov/sgml/sc34/document/0363.htm

No, it won't solve the "semantics" problem, either.  The scope of DSDL
VCSL is validation only.  But DSDL VCSL could give us possibility to
use different kind of schema languages for different "islands" of
a document, e.g. RELAX NG for XHTML 2 part and RDF Schema for RDF part,
as opposed to an XML Schema-based approach, which requires all schemata
need to be written in XML Schema.

James Clark's recent Namespace Routing Language (NRL) [14] proposal is
an another input to DSDL VCSL, and a sample implementation in Jing now
supports RELAX NG, XML Schema and Schematron, which is a good step
forward.  I'll show you NRL-based approach in a separate message.

> > If you look at the latest XHTML 2.0 draft, you'll notice that
> > the Metainformation Module now allows nesting of the meta element [12].
> > This is a step forward to allow RDF/XML-like encoding of metadata
> > through the meta element, as Micah/Steven proposed.
> 
> I looked at [12] and wasn't sure what you meant, and searched for "recurs" 
> and "nest" and didn't see the relevant text?

Look at the abstract definition.  The content model is now something like
this, which means a meta element can include one or more meta elements
(note: this is not DTD-compatible syntax).

    ( ( PCDATA | Inline )* | meta+ )

"meta and RDF" section will be updated when we (presumably this TF)
decide an appropriate way to serialize RDF through this syntax.
But remember, this is for XHTML 2.0.  There's no plan to do something
similar for earlier versions.

> If that's an unrelated problem, all the better (keeps things simple), but my 
> understanding was that the XHTML community would resist XML Schema 
> proposals, akin to what you've already proposed [5] and stated "For XHTML 
> 2, I have hope". So I'm trying to understand how well grounded that hope is 
> and if there's technical objections to it. <smile/>
> 
> > [5] http://www.w3.org/People/mimasa/test/schemas/SCHEMA/xhtml1-loose.xsd

It's not an unrelated problem, but a separate problem.  Most things are
more or less related, and we cannot solve all the problems at once.
We don't necessarily have to satisfy everyone by only one solution.

Technically it is still possible to "use" character entities together
with XML Schema or other schema languages, independent from a particular
document type definition, as noted in "An XML Core WG View" [15].

For XHTML 2, no way to get away from XML Schema.  We've also adopted
RELAX NG as well, and there might be a good chance to give up DTD, as
I'm pretty sure that there're quite a few aspects of XHTML 2.0 which
DTD cannot express, and having a DTD for XHTML 2.0 would be rather
harmful than helpful.  Even if we (well, *I*, as an editor) provide
something, I seriously doubt its normative status.  You might still
have some possibility to use character entities, but that should be
separated from the document type definition.

[13] http://lists.w3.org/Archives/Public/www-rdf-comments/2003JanMar/0489
[14] http://www.thaiopensource.com/relaxng/nrl.html
[15] http://www.w3.org/XML/Core/2002/10/charents-20021023

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

Received on Saturday, 21 June 2003 00:44:14 UTC