- From: Pierre-Antoine Champin <swlists-040405@champin.net>
- Date: Tue, 27 Oct 2009 10:01:45 +0100
- To: Toby Inkster <tai@g5n.co.uk>
- CC: Raphaël Troncy <Raphael.Troncy@cwi.nl>, Semantic Web <semantic-web@w3.org>
Hi all, Le 27/10/2009 00:24, Toby Inkster a écrit : > On Mon, 2009-10-26 at 23:06 +0100, Raphaël Troncy wrote: >> We had a conversation with Tim Berners Lee during this workshop that >> pretty much agrees with what Toby just wrote below ... while I was >> arguing than nobody has formally defined what is the 'sameness' of two >> representations of a resource. The accessibility community has defined >> the notion of "equivalent" when the two representations both fulfill >> the same function or purpose upon presentation to the user, and in an >> accessibility context, it is fine to say that a text is another >> representation of an audio resource ... > > My personal answer is that two responses are "the same enough" if you, > as a publisher, would be happy to publish them under the same URI > without any explicit way of referring to them individually. If you, as > the publisher, would be satisfied never being sure which representation > a consumer will get, then they're OK to share a URI. Note that deciding to serve different representation under the same URI does not mean that we do not also provide a specific URI for each representation. For this, the Content-Location HTTP header is your friend [1]; it can provide the most specific URI of the retrieved representation. E.g. GET /some_resource HTTP/1.1 Host: example.com Accept: application/rdf+xml ... would give 200 OK Content-type: application/rdf+xml Content-Location: /some_resource.rdf ... while GET /some_resource HTTP/1.1 Host: example.com Accept: image/* ... would give 200 OK Content-type: image/png Content-Location: /some_resource.png ... pa [1] although unfortunately browsers do not implement it correctly; see http://jigsaw.w3.org/HTTP/CL/ . However, as long as the fetched URI and the content location are in the same "folder", everything should be ok.
Received on Tuesday, 27 October 2009 09:02:20 UTC