RE: My action item on Moby Dec, issue 14, etc

> From: Michael Mealling [mailto:michael@neonym.net]
> Sent: 20 September 2002 17:48
> To: Williams, Stuart
> Cc: 'Michael Mealling'; Bill de hÓra; 'Jonathan Borden'; 'David
> Orchard'; 'Tim Bray'; 'Norman Walsh'; www-tag@w3.org
> Subject: Re: My action item on Moby Dec, issue 14, etc
> 
> 
> On Fri, Sep 20, 2002 at 05:42:05PM +0100, Williams, Stuart wrote:
> > > > Nonetheless if RDF held an 1-1 mappings as an axiomatic, the MT
> > > > wouldn't require the IS mapping. 
> > > 
> > > Ok, here's the issue: how can you not deductively conclude that
mappings
> > > are 1:1 exclusive given the definition of a resource as that thing
that
> > > is named by a URI?
> > 
> > That a URI maps to a single resource seems a reasonable conclusion -
> > actually it seems more like a definition or an axiom. That a resource is
> > mapped from a single URI... I'm not sure the is so clear. The mapping
may be
> > N:1 (URI->Resource).
> 
> Ok, where in the above system did equivalence get mentioned?

Don't know...  you asked how could you not deductively conclude that the
URI->Resource mappings (plural?) are "1:1 exclusive"... I think I suggested
it wasn't a conclusion I had (yet) reached and why. Did I mention
equivalence?

> If all you
> have is a URI and _nothing_ (and I do mean _nothing_) else there is no
> way of knowing that two resources are the same other than by comparing
> the URIs that are mapped to them.

Oh, I agree... and with URI alone you would only know definitely that they
were different if the mapping were 1:1 as you suggest.

Even seen it asserted in many places that a URI unambiguously identifies a
single resource. I've not seen it asserted the other way round... that a
resource is identified by a single URI. Such an assertion might be a
consequence of a definition of resource like Roy's about a resource being a
conceptual mapping over time between an identifier and a set of time varying
equivalent representations.

> An N:1 mapping of a URI to a Resource
> would require some way to determine more information about a Resource
> other than its URI and that information does not exist at this layer
> of the architecture.
> 
> If you personally need some way of talking about equivalence and multiple
> mappings and of resources and URIs then do it at another layer. Not this
one.
> And be explicit about it.....

Cool... but you still have to nail down the properties of the thing you're
layered upon.

> -MM
> 

Stuart

Received on Friday, 20 September 2002 13:54:57 UTC