Re: Treatment of RDFa in TAG Finding on Self-describing Web

First of all, thank you Steven for your patience in working through this 
with me.

Steven Pemberton wrote:

> On Tue, 09 Sep 2008 02:53:56 +0200, <noah_mendelsohn@us.ibm.com> wrote:
> 
> > As I have mentioned previously on this list, I am trying to wrap up 
the
> > TAG finding on the Self-Describing Web [1], and the major unresolved
> > question is whether there is a chain of normative specifications,
> > presumably starting with the media type registration for
> > application/xhtml+xml [2], that justifies the inference of 
> RDF triples in
> > an XHTML document that uses RDFa.
> 
> Hi Noah.
> 
> The RFC for application/xhtml+xml says
> 
>        With respect to XHTML Modularization [XHTMLMOD] and the existence
>        of XHTML based languages (referred to as XHTML family members)
>        that are not XHTML 1.0 conformant languages, it is possible that
>        'application/xhtml+xml' may be used to describe some of these
>        documents.
> 
> http://www.ietf.org/rfc/rfc3236.txt

Yes. 

> "RDFa in XHTML: Syntax and Processing" (which uses XHTML 
> Modularization) 
> says:
> 
>    XHTML+RDFa documents SHOULD be labeled with the Internet Media Type 
> "application/xhtml+xml" as defined in [RFC3236].
> 
> and further
> 
>    There SHOULD be a @version attribute on the html element 
> with the value 
> "XHTML+RDFa 1.0"
> 
> http://www.w3.org/TR/rdfa-syntax/#docconf
> 
> (Note that XHTML+RDFa is the first version of XHTML that 
> doesn't require a 
> DTD declaration to identify itself. That notwithstanding, there
> is also a 
> DTD available, since current browsers use the existence of a DTD 
> declaration to switch itself into Standards Mode).
> 
> And of course, "RDFa in XHTML: Syntax and Processing" also saysit is OK 
> to extract RDF from such documents.

OK.

> I hope this meets your requirements.

Me too :-), but for the moment I'm afraid I'm not convinced.  Let's go 
step by step.  A Web browser has in hand a URI, does an HTTP GET, and in 
return gets status code 200, Content type: application/xhmtl+xml, and an 
entity body that proves to be a match to the syntactic requirements of 
RDFA in XHTML.  The author of the browers code is aware of both the 
specifications you quote above, plus presumably the XHTML 1.0 
Recommendation and (although you've previously said it doesn't matter 
here) XHTML M12N.  His/her goal is to prove from those specifications that 
the RDFa is intended to be interpreted per the RDFa specifications (as 
opposed to being an accidental match, however unlikely, to the syntax of 
RDFa).  Going step by step:

* The media type is application/xhmtl+xml, so RFC 3236 definitely applies.

* As you note, the RFC says that "it is possible that 
application/xhtml+xml" may be used to describe XHTML family languages. OK, 
but does that say anything about which specification I consult next to 
determine which of the markup, beyond XHTML itself, is to be interpreted, 
and how?  I'm not sure how signalling that something is "possible" has 
much normative force, and even if it does, it doesn't tell me much about 
which specifications to consult next.  I could claim that I'd written a 
second specification giving a different interpretation of the RDFa markup, 
and I don't see how you'd say that one didn't apply.

* You then quote the RDFa Syntax and Processing draft: "There SHOULD be a 
@version attribute on the html element with the value 'XHTML+RDFa 1.0;". 
Let's for the moment ignore that it's a SHOULD and consider the case where 
@version is indeed set this way.  Is there something in RFC 3236 or XHTML 
1.0 to the effect that "a value in the @version attribute of the HTML 
element is a key that can be used to identify additional specification(s) 
that provide normative interpretations for markup beyond that which is 
standardized for XHTML itself?"  As best I can tell, XHTML delegates 
discussion of the <HTML> element to HTML 4.01, and that in turn says that 
the version attribute is deprecated! [1]  Is there something normative in 
either XHTML or M12N that says "go looking for a specification that seems 
to have been written by the authority responsible for the namespace used 
in the new markup?". 

So, I'm still missing some crucial connections.

To be clear, I'm very supportive of RDFa and am not looking for excuses to 
make trouble here.  I would very much like to find a clean unambiguous 
story to support the self-describing use of RDFa.  That said, my view of 
"follow your nose" is that the feel is usually:  start with specification 
X;  it well tell you that there are cases in which specification Y 
applies, and that in turn may delegate to specification Z, etc.  I'm 
missing that here.  The best I can get is that serving a wide variety of 
XHTML-family content with application/xhtml+xml is not a violation of RFC 
3236 (I.e. it's "possible").  What I'm missing is a clear chain of one 
specification pointing to the next, telling me what to read to interpret 
those enhanced formats when they show up.

I apologize if I'm being dense here, and thank you again for your patient 
help!

> 
> Best wishes,
> 
> Steven Pemberton

And to you!

Noah

[1] 
http://www.w3.org/TR/1999/REC-html401-19991224/struct/global.html#h-7.3



--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------








"Steven Pemberton" <steven.pemberton@cwi.nl>
09/09/2008 07:00 AM
 
        To:     noah_mendelsohn@us.ibm.com, www-tag@w3.org
        cc:     "Mark Baker" <distobj@acm.org>, "Ben Adida" <ben@mit.edu>
        Subject:        Re: Treatment of RDFa in TAG Finding on 
Self-describing Web


On Tue, 09 Sep 2008 02:53:56 +0200, <noah_mendelsohn@us.ibm.com> wrote:

> As I have mentioned previously on this list, I am trying to wrap up the
> TAG finding on the Self-Describing Web [1], and the major unresolved
> question is whether there is a chain of normative specifications,
> presumably starting with the media type registration for
> application/xhtml+xml [2], that justifies the inference of RDF triples 
in
> an XHTML document that uses RDFa.

Hi Noah.

The RFC for application/xhtml+xml says

       With respect to XHTML Modularization [XHTMLMOD] and the existence
       of XHTML based languages (referred to as XHTML family members)
       that are not XHTML 1.0 conformant languages, it is possible that
       'application/xhtml+xml' may be used to describe some of these
       documents.

http://www.ietf.org/rfc/rfc3236.txt

"RDFa in XHTML: Syntax and Processing" (which uses XHTML Modularization) 
says:

                 XHTML+RDFa documents SHOULD be labeled with the Internet 
Media Type 
"application/xhtml+xml" as defined in [RFC3236].

and further

                 There SHOULD be a @version attribute on the html element 
with the value 
"XHTML+RDFa 1.0"

http://www.w3.org/TR/rdfa-syntax/#docconf

(Note that XHTML+RDFa is the first version of XHTML that doesn't require a 
 
DTD declaration to identify itself. That notwithstanding, there is also a 
DTD available, since current browsers use the existence of a DTD 
declaration to switch itself into Standards Mode).

And of course, "RDFa in XHTML: Syntax and Processing" also says it is OK 
to extract RDF from such documents.

I hope this meets your requirements.

Best wishes,

Steven Pemberton

Received on Tuesday, 9 September 2008 21:52:00 UTC