Re: FragIds in semantic web (ACTION-543)

On Mon, May 9, 2011 at 7:42 AM, Graham Klyne <GK-lists@ninebynine.org> wrote:
> It's arguable that any syntax that claims to represent RDF makes implicit
> reference to http://www.w3.org/TR/rdf-concepts/#section-fragID.

This text is nonnormative and just explains what happens in the case
that there is an application/rdf+xml representation - which we already
knew by reading RFC 2396 or 3986 together with 3870. What we've been
discussing is what happens if there *isn't* an application/rdf+xml
representation, in which case this section doesn't apply, if I read it
correctly. (E.g. if the only representation is application/xml.)

i.e. "the same thing that it does in an application/rdf+xml
representation of the resource" only makes sense if there is such a
representation.

The passage suggests (but does not say) that the way you understand a
URI in RDF may be different from the way you understand it in IETF
"identification"-land. E.g. it says nothing about consistency between
representations, which is a potential divergence with IETF. And of
course nowhere does any RDF spec come out and say, normatively, that
URIs in RDF mean what they "identify" according to IETF, so having the
two namespaces diverge is at least excusable, if not necessarily
desirable.

I think some of the confusion may be because there are two independent
questions. One is, when you're trying to understand some RDF, how do
you take 3986 and all the media type registrations into account?  This
is not what the 'MIME and the Web" report needs to address, since RDF
Concepts (etc.) can take care of itself and gets to decide whether RDF
uses the IETF/AWWW URI namespace or not. The interpretation of RDF is
outside the scope of 'MIME and the Web'. The question rather is when
is RDF supposed to influence what fragment identifiers "identify" in
the IETF/AWWW URI namespace? - where this is asked outside the context
of RDF.

Jonathan

> #g
> --
>
> Jonathan Rees wrote:
>>
>> For what it's worth - and backing up your point that RDF seems to be
>> doing its own thing with fragids independent of what 3986 says - of
>> the many RDF serialization format registrations either completed or in
>> progress, the only one that says anything about fragids is
>> application/rdf+xml (RFC 3870).
>>
>> Turtle and N3 are registered:
>> http://www.iana.org/assignments/media-types/text/turtle
>> http://www.iana.org/assignments/media-types/text/n3
>> but say nothing about fragids.
>>
>> OWL Manchester, Functional, and XML are submitted and in progress, and
>> say nothing. And of course, as I said, text/html and application/xml
>> also say nothing.
>>
>> This state of affairs reinforces the RDFa view that updating the
>> registrations is either unnecessary or unimportant. If Turtle doesn't
>> do it, why should RDFa do it? (ACTION-509) Maybe this is right, I
>> don't know.
>>
>> If we end up deciding this is important, we might consider updating
>> the W3C guide on registering media types
>> http://www.w3.org/2002/06/registering-mediatype
>> to urge consideration of fragids, since this document is very likely
>> what the authors of all the above registrations consulted. That
>> doesn't solve the problem in general, but it would be a start.
>>
>> (thanks for help from Eric P, Ivan H, and Sandro H)
>>
>> Jonathan
>>
>
>
>

Received on Monday, 9 May 2011 13:41:32 UTC