W3C home > Mailing lists > Public > semantic-web@w3.org > November 2006

Re: Re[2]: AW: Content negotiation flamewar (was: Re: "Hash URIs" and content negotiation)

From: Richard Cyganiak <richard@cyganiak.de>
Date: Tue, 14 Nov 2006 13:08:14 +0100
Message-Id: <DC49BA11-78ED-4475-975B-663EF279D69D@cyganiak.de>
Cc: "'Max Voelkel'" <voelkel@fzi.de>, "'Semantic Web'" <semantic-web@w3.org>
To: "Xiaoshu Wang" <wangxiao@musc.edu>


On 14 Nov 2006, at 05:22, Xiaoshu Wang wrote:
>> As defined in the URI spec, we send only the URI
>> http://example.com/resources (of course using
>> content-negotiation). So if we are a browser, we get back
>> HTML, scroll down to "#Bob", if defined, done.
>> If  we  are a semantic web agent, we Accept:
>> application/rdf+xml and get back an RDF/XML  file.  We  parse
>>  it an look up statements about bob. We can ignore all other
>> statements which are not directly or indirectly connected to bob.
> It is interesting from the discussion that it seems that the 303  
> response
> pattern is applicable to only a subset of HTTP URIs.  HTTP URI with  
> the
> fragement identifier shouldn't and cannot comply.  Then what about  
> the HTTP
> URIs with query parameters? Should http://example.com/resources? 
> name=bob
> uses a 303 or not? I guess it would be yes?  But why?

The question mark does not really have any special meaning in HTTP  
URIs; it's just a syntactic thing.

(In practice, the part in front of the question mark often addresses  
a script, and the part behind are arguments to the script, but this  
is an implementation detail on the "other" side of the HTTP  
interface. From the Web's POV, the question mark is no different from  
the underscore or the letter "x".)

> I assume that a namespace document is an information resource. (At  
> least
> from what I just checked, the rdf namespace comes back a 200).

Yes, I think so.

> It is both
> interesting and confusing to think that a "fragment" of an information
> document can be a non-information resource. I know that I am  
> playing the
> word "fragment",

I agree. Traditionally, a fragment identified some kind of secondary  
thing, a part of the main resource. RDF breaks with that tradition by  
saying that #fragment does not identify a part of the RDF document,  
but some first-class entity in the universe of discourse. This seems  
to work pretty well, except for content negotiation.

> but somethings seems peculiar here about the demarcation of
> IR vs non-IR.  Is the distinction arbitrary, i.e., derived from the
> specifications, or is it a reflection of the real world?

The distinction is certainly not clear-cut. AWWW says: "The  
distinguishing characteristic of information resources is that all of  
their essential characteristics can be conveyed in a message."


> Xiaoshu
Received on Tuesday, 14 November 2006 12:15:22 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:44:58 UTC