Re: Information resources

On Mon, 2003-07-28 at 19:29, Tim Bray wrote:
> Dan Connolly wrote:
> 
> >> Other resources named 
> >>by URIs may exist entirely apart from the Web.
> > 
> > 
> > Hmm... that seems contradictory; I'd prefer to leave
> > "entirely apart from the Web" out of it...
> 
> Actually, how about "may exist independently of the Web".  I.e. if the 
> Web goes away, so does http://www.tbray.org/misc/Tim, but the ISBN 
> doesn't, nor would the person identified by foaf#Dan.
> 
> > Hmm... I thought what distinguished information resources from
> > resources in general is that only information resources have
> > representations (in the webarch sense).
> 
> I totally disagree.  My examples include the RDF namespace and the FOAF 
> thing, both of which have representations,

Actually, no, there aren't any widely deployed
web protocols for getting a representation of
  http://www.w3.org/1999/02/22-rdf-syntax-ns#
You can only GET a representation of
  http://www.w3.org/1999/02/22-rdf-syntax-ns

I already explained about the FOAF thing; you seem
to have skipped past that part of my message or something:

>> e.g. take http://example.com/foaf#Dan none of the
>> deployed/specified infrastructure provides a way get a
>> representation of that resource; you can only get a representation
>> of http://example.com/foaf

>  and in fact there are systems 
> that will take that urn:isbn: thing and give you something useful.

If there are, then it's an information resource (by my
suggested definition).

(Using the same URI for a web page and a physical
book seems iffy to me, but my suggested text
doesn't legislate against it.)

>   I 
> can't imagine anything in the universe for which you could decree "there 
> can be no representations".

I'm not suggesting we go there. What I'm suggesting is,
again:

  While URIs can, in general, be used to refer to any sort of
  resource, the case of an *information resource*, that is, one
  for which Web protocols provide representations, is particularly
  relevant to the structure of the Web ...

Reasonable people may disagree (and evidently do!) about
whether a galaxy is something you can represent (in
the webarch sense) with some bits and a MIME type; I'm not
suggesting we take a firm position on that. But I'd like
to introduce this term (and ground it in protocols, for now)
in hopes that it will help the community
talk about the situation. We seem to make up
names for it repeatedly...

> An  information resource is something that is primarily information. 
> That's all (I think). -Tim


I thought the way I made the distinction
would appeal to you, since it's a distinction
that's grounded in running code and bits on the wire.

It's intended to be responsive to comments from programmers
such as...

  I'd say that a URI was "Web resolvable" if one could build a
  HTTP proxy to resolve it.
  -- http://www.w3.org/mid/20030719230728.G4241@www.markbaker.ca

and

[[
b) Deployed code doesn't support fragment identifiers as first-class
objects
-- I can't ask an HTTP proxy about them, I can't query an HTTP server about
them, etc.
]]
  -- http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0405.html


It's also responsive to more philosophical comments such as...

[[
Fragments are not first-class
resources; not even when they consistently identify semantics across
multiple representations.
]]
 -- http://lists.w3.org/Archives/Public/www-tag/2002Jul/0178.html


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Tuesday, 29 July 2003 00:36:53 UTC