Re: Status of Link header

Shane McCarron wrote:
> 
> Sorry - let me be more precise.  It appears to me that the current draft 
> is deferring to the simple relation descriptions from HTML 4.  Those 

Not really, it adopts both the values from the current IANA registry for 
Atom relations and HTML4.

> definitions are very, very old and there has been work done recently on 
> this with the W3C.  These have been recently updated and extended as 
> part of the XHTML 2 Working Group / RDFa Task Force activity defining 
> RDFa.  That activity defines a mechanism for extending the collection of 

Well, they have also been updated by the WHATWG.

I personally do not believe that a markup language related WG is the 
right place to do this, as link relations should be portable across formats.

> relationships via RDF, and also codifies the pre-defined "simple" 
> relationships such as previous.  I was only saying that if you want this 
> document to be consistent with the direction HTML is headed, you should 
> rely upon the definitions in RDFa.  Those definitions are contained in 
> the referenced vocabulary document.

I believe that the WHATWG's proposal to use a Wiki as registration 
process is a non-starter. But so is a process that requires knowledge of 
RDF.

> As to registration - I don't personally expect there to be extensions to 
> the collection of simple relationships.  If there are, it would be done 
> through the W3C processes and the vocabulary document updated.

I don't think this reflects reality (see the set of Atom link relations).

> Instead, I expect RDF vocabularies to be created that extend the 
> collection.  Since the draft Link: header document defines a 
> relationship as a "URI-Reference" I think this is entirely consistent 
> with the definition used for values of @rel in RDFa.  After all, RDF is 
> about expressing semantics via URIs.
> For more on RDFa, you might look at http://www.w3.org/TR/rdfa-syntax

I do not disagree that RDF *could* be used to do that. People who want 
"short" relation names should be able to get them through a controlled 
process; an IANA registry would fulfill that need.

People how are happy with a "long" name (a URI) should be able to mint 
them easily. If they want to use RDF to describe that relation, so be 
it, but I don't think a requirement to do so will fly.

> As an aside, RDFa syntactically uses "CURIEs" to express values for 
> @rel.  This is a lexical space concept, not a value space concept.  In 
> the value space, @rel values are URIs, so I believe this to be entirely 

URIs or IRIs?

> consistent with the view that Link: header rel parameters are 
> white-space separated URI-References.

Yes.

The concern about CURIEs is related to the fact that, despite what you 
wish, there are other people "defining" link/@rel, and it would be bad 
if the definitions disagreed on the lexical representation.

Best regards, Julian

Received on Friday, 12 September 2008 15:07:22 UTC