Re: XHTML 2.0 - no interest in RDF/XML?

Art.Barstow@nokia.com wrote:

> > > Given the W3C's huge investment in the RDF metadata framework,
> > > why does the Metainformation Module in the 2002-08-05 WD of 
> > XHTML 2.0 
> > > not accommodate (or at least mention) the XML serialization of RDF?
> > 
> > If you can show us a feasible proposal, I would be interested in
> > looking at it.
> 
> XHTML 2.0 should include a Normative statement about embedding
> RDF/XML.

I'm afraid I wonder if you understood your own question.  You
specifically mentioned the "XML serialization of RDF", which
has been the hardest one among various RDF in (X)HTML proposals.

The issue comes down to how to embed RDF in XHTML without sacrificing
the validation.  The fact that people have been discussing this issue
for years and yet we haven't had a standard solution means the issue
is not that simple to solve.   We just need a techinically feasible
solution. 

> As others have suggested, the WG has options to consider:
> 
>  [1] http://lists.w3.org/Archives/Public/www-rdf-interest/2002Jun/0002.html

If you know how DTDs work (or don't work), you'll understand why
this isn't a viable solution.  For details, see Murray Altheim's
message, which is linked from [2]:

   http://lists.w3.org/Archives/Public/www-rdf-interest/2001Apr/0223

If the issues was that damn simple, we'd never have had this
discussion.

>  [2] http://infomesh.net/2002/rdfinhtml/

If you read through Sean's excellent document, you'll understand
why it's so difficult.

> and there's also:
> 
>  [3] http://www.w3.org/RDF/FAQ#How

You should be aware why we are so unhappy with this bogus FAQ:

    Issue faq-html-compliance: The suggested way of including RDF meta
    data in HTML is not compliant with HTML 4.01 or XHTML
      http://www.w3.org/2000/03/rdf-tracking/#faq-html-compliance

and why it's

    Currently: for discussion

and haven't had a resolution yet.

The RDF community should not underestimate the XHTML community's
desire to create valid documents.  IMHO, sacrificing the validation
of data (i.e. XHTML document) for the sake of metadata (i.e. RDF)
is an overkill.  To date, no one provided a normative DTD nor XML
Schema for RDF, which means, in particular for DTD validation, you
just can't put RDF in XHTML.

That's one of reasons why I'm reluctant to normatively impose DTD
validity in XHTML 2.0, which would effectively kill the possibility
of embedding RDF in XHTML, and why I'm investigating other
alternatives.  As Manos mentioned, with XML Schema we could skip
part of validation, and with RELAX NG, we might be able to
validate both RDF and XHTML.

However, even if we can syntactically mix them, there are other
technical issues such as "what is the meaning of the fragment
identifiers within such a document?", as Sean put it in [2].
XHTML and RDF don't share the same meaning of the fragment
identifiers, and I haven't seen a clear consensus on this issue yet.

> > We are observing "RDF in (X)HTML" discussion here and there, and
> > it seems not all people prefer to put RDF directly in XHTML.  There's
> > a couple of technical issues to be solved.
> 
> Please do not ignore this issue for the above reason ("not all people prefer
> to put RDF directly in XHTML").

Please do not interpret my message as we are ignoring this issue.
We have been observing this issue for years and trying to find
a solution.  I was just saying that embedding the "XML serialization
of RDF" *directly* into XHTML is not the only possible way to
associate RDF metadata to XHTML (as illustrated in [2]) and for
that particular method there's a couple of thorny technical issues
to be solved.

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

Received on Thursday, 15 August 2002 07:54:26 UTC