Re: Clarifying what a URL identifies (Four Uses of a URL)

On Tue, 2003-01-21 at 16:27, Sandro Hawke wrote:
> > > On the one hand you're saying that a URI identifies a Resource across 
> > > contexts. 
> > 
> > No. I'm saying that URIs and Resource exist regardless of contexts.
> 
> You said "URIs identify Resources".  I read that as claiming that
> there exists a special IETF-sanctioned function which maps each URI
> string to a "Resource", in the web-protocol world in general or at
> least HTTP in particular.  Are you claiming that?

No. The term 'Resource' is not being used in some network
accessible/protocol sense. A 'Resource' is that concept that is
identified by a 'URI'. Simply that and nothing else. The 'function' you
speak is the identity function as described in RFC 2396.

> Even the most narrow HTTP-only form of that claim appears false, since
> TimBL and RoyF cowrote (with others) the relevant RFCs and yet don't
> agree on httpRange-14!  If they can't agree on the range of that
> function, that must mean their conceptualizations of the supposed
> function are quite different.  If the writers of the standard can't
> agree, I suggest that means the standard does not actually bear
> meaningfully on the issue.

Again, that may be the case inside the context of which the 'http:' URI
scheme is most applied but that's a problem with either the 'http:' URI
scheme or the context. Its not a problem with URIs in general.


> The whole notion of "Resources" is a red herring in Web Architecture.
> It's vague enough that people can read into whatever they want, with
> no real consequences for conflicting interpretations.

That is by design....

> I bet I can
> rewrite any web architecture text to be clearer for any particular
> technical/computing audience without using the word "resource".  (But
> the stakes will have to be high to be worth my while and pay off the
> people I'm supposed to be working for.  :-) And I restrict my bet to
> English.)

It might be clearer for any particular audience but will it still be
uniformly applicable regardless of any context its used? URIs get used
in routers for crying out loud. No http, no document, no BASE, no XML,
no nothing. They have to be universally applicable and without
additional baggage for the exact same reason that IP doesn't gaurantee
delivery: some things don't need it. TCP handles gauranteed delivery.

-MM

Received on Tuesday, 21 January 2003 20:57:09 UTC