RE: Information resources

 -----Original Message----- 
 From: ext Tim Berners-Lee [mailto:timbl@w3.org] 
 Sent: Thu 7/31/2003 10:40 PM 
 To: Dan Connolly 
 Cc: Tim Bray; WWW-Tag 
 Subject: Re: Information resources
 

    While URIs can, in general, be used to refer to any sort of
    resource, the case of an *information resource*, a resource
    which is information, is particularly
    relevant to the structure of the Web.  Of these the Web protocols
    can provide "representations": expressions in some data format,
    with accompanying metadata.
 

Why representations particularly of "information resources"? How are representations of "information resources" any different than representations of any other kind of resource? Why would such resources be any more relevant to the Web?

A representation of the city of Paris, in the form of an HTML instance, is just as relevant to the web, IMO, than a representation of a web page, in the form of an HTML instances. There is no functional distinction insofar as how the web works. Both the web page and the city of Paris are resources denoted by web-resolvable URIs, and their representations correspond to HTML instances (both may have other representations which do not correspond to HTML instances). Why would an "information resource" such as a web page have any special status over the city of Paris insofar as the architecture of the web is concerned? It shouldn't.

Rather than the nature of the resources denoted, surely what is particularly relevant to the architecture of the web is

(1) URIs which are web-resolvable to representations, and 

(2) encodings for representations which provide for the (indirect) expression of  links to other representations by reference to other resources by means of URIs which are web-resolvable to representations (whew, that's as short at I could get it ;-)

What the URIs actually denote is irrelevant to the functionality of the web. And that irrelevance should be a cornerstone of the web architecture and a principle point of collaboration between web and SW functionality.

How URIs are web-resolvable to representations, and how representations embody links to other representations, are certainly relevant to the realization of the web, and to that end there are specifications such as for HTTP, HTML, etc. 

But we need not restrict the kinds of resources which may be denoted by web-resolvable URIs to define the fundamental architecture of the web. To do so is unnecessary, unwarranted, and damaging to the present and future utlity of the web.

Regards,

Patrick

Received on Monday, 4 August 2003 05:24:49 UTC