- From: Miles, AJ \(Alistair\) <A.J.Miles@rl.ac.uk>
- Date: Fri, 24 Jun 2005 16:47:42 +0100
- To: "Ralph R. Swick" <swick@w3.org>
- Cc: <public-swbp-wg@w3.org>
Hi Ralph,
Although httpRange-14 is moot now, ...
> An important semantics of PSI is the indirection semantics.
> This includes
> (a) there _must be_ a document and (b) that document _must_ provide
> sufficient instruction to a human reader to recognize the
> intended subject.
... this is a good point, and food for VM best practices.
>
> Indirection is not an essential feature of the #fragid
> philosophy. While it
> would not be considered Best Practice (I hope) for there to be no
> "information resource" at the URI, the #fragid philosophy does not
> require such an information resource Nor does the notion of a human
> reader consuming the content of such an information resource affect
> whether the full URI is acceptable model-wise to name the intended
> subject.
What do you think of this, drafted for insertion in next SKOS Core Guide ...
* You MAY use an HTTP URI of any form to name a resource of type skos:Concept.
* If you use an HTTP URI of the form /http:[^#]*$/ (a 'slash URI', e.g. http://slash.example.com/concepts/love) to name a resource of type skos:Concept, then an HTTP GET request SHOULD return a 303 see other (redirect) response code. The agent SHOULD be redirected to an information resource that describes the original resource, that by default returns a response with content type 'application/rdf+xml', that MAY support content negotiation, and that MAY support the SPARQL HTTP protocol.
* If you use an HTTP URI of the form /http:[^#]+#[^#]+$/ (a 'hash URI', e.g. http://hash.example.com/concepts#love) to name a resource of type skos:Concept, then the resource denoted by the URI before the hash (the primary resource, e.g. http://hash.example.com/concepts) SHOULD be an information resource that describes all associated secondary resources, SHOULD respond to an HTTP GET request with a 200 response code, SHOULD return a response with content-type 'application/rdf+xml' by default, MUST NOT support content negotiation, and MAY support the SPARQL HTTP protocol.
... ? See also followup from Danbri:
http://lists.w3.org/Archives/Public/public-esw-thes/2005Jun/0044.html
... and subsequent in thread.
Cheers,
Al.
>
>
>
Received on Friday, 24 June 2005 15:47:49 UTC