RE: httpRange-14

 

 -----Original Message----- 
 From: ext John Black [mailto:JohnBlack@deltek.com] 
 Sent: Sat 8/2/2003 9:02 PM 
 To: Bill de hÓra; Tim Berners-Lee 
 Cc: Roy T. Fielding; Norman Walsh; Public W3C 
 Subject: RE: httpRange-14
 
 



 It is true, HTTP-URIs used as names, rather than web retrievers,
 could be used to name anything.  The meaning of names is with the
 namer(s). 
 
 But we can agree that HTTP-URIs, when used as names, will name those
 very things that are retrieved by the Web.  This is possible and makes
 a huge difference.  If all (or most) of us agree that HTTP-URIs name
 the things retrieved, then suddenly RDF now has a huge set of things
 to talk about.  

  

But this does not work in practice, because http: URIs can resolve to more than one thing (representation). At best, you could use them to name collections of representations, or access points, or containers, but not "the things that are retrievable".

The apparent ambiguity between what the URI denotes, and what is returned from a GET, can be easily solved by the server stating in the response the URI denoting the actual entity returned, which it should be doing anyway.

Then, folks (and SW agents) are free to talk about the resource denoted by a URI, or a representation of that resource (also a resource) denoted by some other URI, etc. and one can make relations between resources and their representations explicit in descriptions, such that when one does an MGET on some URI http://example.com one might get back as part of its description

   <http://example.com <http://example.com> >
    :hasRepresentation <http://example.com/index.html <http://example.com/index.html> > ;
    :hasRepresentation <http://example.com/index.xml <http://example.com/index.xml> > ;
    :hasRepresentation <http://example.com/orgChart.jpg <http://example.com/orgChart.jpg> > ;
    :hasRepresentation <http://example.com/companyInfo.rdf <http://example.com/companyInfo.rdf> > .
etc.

Patrick

Received on Monday, 4 August 2003 06:30:01 UTC