Patrick Stickler wrote: > > > In my example, a request for application/rdf+xml could be 303- > > redirected. The target of the redirect can still negotiate further, > > i.e. dereferencing it may yield RDF or N3 or some other format. If > > the original request URI wants, it could also perform the > > 303-redirect if the client only Accepts text/rdf+n3, or any similar > > description format. > > You seem to be presuming that no other representation exists for that > resource. Why would 303 be used in the above case, rather than 415? > I'm talking about a specific use-case where the Accept header *only* contains applicaton/rdf+xml, in which case the XHTML and HTML variants which do exist SHOULD NOT be returned -- I'm not presuming they don't exist. If an RDF representation does exist, it would be served using a 200 response, not a 303. I don't think 415 applies here, we are talking about GET requests with no entity body, not PUT or POST requests with a media type unknown to the server. > > If a GET on the query URI in question normally returns an HTML > instance, and my agent asks for text/n3 and no N3 is available via > that URI, why would the server send a 303 response? > It wouldn't, unless the server knows of a related resource which is N3. If the server knows of no related resource in that media-type, then the response is 406. The 303 only indicates that *maybe* there's a representation (of some other resource) that interests the client. -EricReceived on Monday, 14 April 2008 22:22:45 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 12 September 2008 07:02:20 GMT