Re: How namespace names might be used

At 08:56 PM 10/06/00 -0400, Michael Mealling wrote:
>On Sat, Jun 10, 2000 at 03:29:13PM -0700, Tim Bray wrote:
>> 
>> Level 1 idea: use the URI to retrieve the semantic resource.  I think this
>> is more or less what TimBL and Dan are in favor of.  ...
>
>But I think you are missing a rather fundamental aspect of what Tim, Dan
>and myself are saying: the URI identifies the semantic resource not
>a particular representation of it. Its not a question of the URI not
>allowing you to get at those different semantic parts. That function
>is left to the protocol that allows you to ask questions (methods)
>about that URI. I.e. for the URI 'foo' give me the human readable
>version, the RDF, the XML Schema, runnable code, etc. Those are all
>representations of the same set of semantics that are identified by
>teh URI.

There are two things wrong with this argument.  First of all, whereas
such a protocol could be imagined, it doesn't currently exist.  I think that
such a thing is more or less what my option 3 suggested; but it can't
possibly depend on direct-retrieval URI semantics.   Second, it is *not*
the case that stylesheets, RDF, schemas, etc, are "representations of the
same set of semantics".  The semantics of display and of full-text-search
and of purchase-transaction-execution and so on are not on the same planet.
In the real world, the set of semantics potentially relevant to a chunk
of XML spreads messily across the intellectual and engineering universes,
and dereferencing a single URI ain't gonna get you there.

As for content-negotiation, I admit to being not up to date on the latest
& greatest developments.  I do worry that if it grows to complex & baroque,
it will undermine the wonderful directness and simplicity that are central
to the excellent design of the HTTP protocol. -Tim

Received on Sunday, 11 June 2000 13:24:29 UTC