W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2009

Re: Review of new HTTPbis text for 303 See Other

From: Pat Hayes <phayes@ihmc.us>
Date: Tue, 4 Aug 2009 10:30:36 -0500
Cc: Julian Reschke <julian.reschke@gmx.de>, "www-tag@w3.org WG" <www-tag@w3.org>, HTTP Working Group <ietf-http-wg@w3.org>
Message-Id: <7CA9202A-6F6B-4B2F-956D-570C17CA2ED9@ihmc.us>
To: Tim Berners-Lee <timbl@w3.org>

On Aug 2, 2009, at 2:54 PM, Tim Berners-Lee wrote:

>
> On 2009-08 -02, at 09:54, Pat Hayes wrote:
>
>>
>> On Jul 31, 2009, at 3:41 PM, Julian Reschke wrote:
>>
>>> Pat Hayes wrote:
>>>> ...
>>>> hope I do not misrepresent anyone here.) Apparently, therefore,  
>>>> two people both quite expert in reading the HTTP spec do not  
>>>> interpret the phrase "requested resource" in the same way,  
>>>> leaving me and I suspect others in a state of complete  
>>>> confusion. ...
>>>
>>> We have multiple issues open with respect to cleaning up that  
>>> terminology. I would recommend to delay any new discussion until  
>>> we're done with that (which *should* be the -08 versions of the  
>>> drafts).
>>
>> OK, great. Sorry if my insistence on this point has caused needless  
>> feather ruffling. For the record, let me outline the case that  
>> raises the central issue. No need to reply.
>>
>> 1.   http://example.com:8080/people/richard_cyganiak  identifies  
>> Richard Cygniak, the actual human being.
>> 2.   A GET request on http://example.com:8080/people/ 
>> richard_cyganiak  resolves to some HTTP endpoint (server, whatever;  
>> I'm not sure of the right terminology here. I mean, the network  
>> entity which functionally handles the request and emits an HTTP  
>> response, and whose behavior is governed by the HTTP specs. I  
>> gather that this may not be identical with the http:resource  
>> (information resource) associated with it: the latter may lie just  
>> behind the endpoint, which constitutes an "interface" to it. (?))
>> 3.  There is an information resource, R, at this endpoint. (Again,  
>> I'm not sure of the right terminology. By R I mean a thing which  
>> has a transmittable representation in the sense of the HTTP specs,  
>> so that it is possible for the endpoint to send a 200-coded  
>> response to a GET request with a URI which identifies R.)
>> 4. To emphasize, this means that there are two resources in the  
>> picture: Richard Cygniak, who is not an information resource or an  
>> http:resource (with the current wording, ie a network object or  
>> service) and has no transmittable represetnation; and R, which  
>> falls under both categories, and does have a transmittable  
>> representation. The URI identifies the first and resolves to (an  
>> endpoint interfaced to) the second.
>>
>> OK, so now the questions that need to be resolved are (at least :-):
>>
>> A. Is this possible? (If not, how is it to be prohibited, since  
>> owners of URIs can, it seems, set up such a situation.)
>
> No.

Well, it is *possible*. I guess you are saying, it should be deprecated.

> Peer pressure from people whose system can't use the server.

Hmm, but why would they not be able to use it? Seems to me that if we  
can get our various storys straight, then this possibility is not only  
workable but might be quite useful. Imagine there is some elaborate  
Web ontology linked data service thing which uses 303 redirection on a  
whole range of URIs it treats as denoting external entities, a thing  
along the lines of DBpedia. But it also has a URI of its own, one that  
identifies it, and to which it responds with a nicely designed,  
informative web page explaining its history and how to use it and so  
forth. LIke DBpedia, in fact. Seems to me that we should be able to  
say that this thing is a resource, and that it can be described in  
just this way. Yes, it has a 200-codable representation of itself,  
which it can deliver when you GET the appropriate URI, but it also  
handles a large number of other URIs using 303 redirection, conformant  
with http-range-14. One thing, one resource (or maybe one HTTP  
endpoint/server), does all of this. Why not? It seems more natural to  
say that about DBpedia than to have to say that DBpedia is not one  
thing but thousands of different things, one for each URI it redirects.

If we prohibit the situation I outlined, then any http endpoint which  
performs a 303 redirect must be 'invisible' : it cannot have a URI of  
its own, so to speak, one that delivers information about it with a  
200 code attached. And this seems both a needless restriction, and one  
that might be a real problem under some circumstances.

Pat

>
>> B. Under these circumstances, is the "requested resource" R, or is  
>> it Richard Cygniak? (If the former, what is the relationship, if  
>> any, between the 'requested' resource and the 'identified' resource?)
>
> Richard.
> The 'requested' thing is presumably means that denoted by the URI in  
> the GET request.
>
>> C. Does http-range-14 require that the endpoint emit a 303 response  
>> under these circumstances? (If the answer is no, then some  
>> explanation is needed.)
>
> Yes.
>
>> Pat
>>
>>
>
>

------------------------------------------------------------
IHMC                                     (850)434 8903 or (650)494 3973
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Tuesday, 4 August 2009 15:31:46 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:08 GMT