- From: Williams, Stuart (HP Labs, Bristol) <skw@hp.com>
- Date: Thu, 7 Jun 2007 11:28:21 +0100
- To: "Tim Berners-Lee" <timbl@w3.org>
- Cc: <www-tag@w3.org>
Hello Tim, At our F2F last week we discussed the possible significance of a chain of 30x redirects arising from an intial attempt to access a non-information resource, Resource(A) ie. get <A> -> 303: Location <r.1> get <r.1> -> 303: Location <r.2> ... get <r.n> -> 303: Location <r.n+1> ... get <r.k> -> 200: Rep(r.k) IIRC this discussion arose while exploring the suggestion that the first 303 response might be used to infer that target of the first redirection, Resource(r.1) in the case above, is an information resource. I think that your view was that Resource(r.1) should provide information about Resource(A), Resource(r.2) about Resource(r.1), ..., Resource(r.n) about Resource(r.n-1) and so on such that the information provided in the representation Rep(r.k) contains information about (possibly amongst other things) Resource(r.k-1). Can you clarify whether you thing Resource(r.k) should provide information about Resource(r.k-1) ONLY or about Resource(r.k-1) AMONGST-OTHER-THINGS (which potentially includes Resource(A))? FWIW, I agree that for pragmatic reasons it would be good for such chains to be short, say just one redirection (though of course some would prefer zero). However, I would prefer a relaxed understanding of the signficance of the redirection such that it is regarded as no-more than a suggestion of a place to look for information about a subject (Resource(A)). Whether such information is actually available at the redirection target (Resource(r.1)) cannot be determined without looking. Looking may involve further redirections. I'd be interested in your thoughts. I'm trying to give some thoughts as to how this (short chains) would be expressed as a good practice. BTW: I think that there may also be a principle that arises directly from the httpRange-14 resolution [1] which could be expressed either as: a) "Non-information resources cannot have webarch:representations." (factual) or b) "Do not deploy webarch:representations for non-information resources." (imperative) which is consistent with your response to John Cowan [2] Thanks Stuart (expecting flak) -- [1] http://lists.w3.org/Archives/Public/www-tag/2005Jun/0039.html [2] http://lists.w3.org/Archives/Public/www-tag/2007Jun/0024.html -- Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks RG12 1HN Registered No: 690597 England
Received on Thursday, 7 June 2007 10:29:34 UTC