Re: URIs quack like a duck

On Sun, Jun 04, 2000 at 04:02:21PM -0400, Clark C. Evans wrote:
> On Sun, 4 Jun 100, John Cowan wrote:
> > Clark C. Evans scripsit:
> > >  a) one de-references the function (if possible)
> > >     and compares the resources directly.
> > 
> > As I have repeatedly shown on this list, resources
> > *cannot* be compared directly, only the entity bodies that happen
> > to represent them at a particular moment.
> 
> Let's not go there as the ground is too slippery.  *evil grin*
> 
> > >  b) an injective identification function is named.
> > 
> > What about direct assertions?  We can't in practice dereference the
> > names "Hesperus" and "Phosphorus" to determine if they identify
> > the same thing (because the planet Venus is not a network-retrievable
> > resource), nor can we, in the Real World, limit ourselves to just
> > one of these names.
> > 
> > But we can look in a database that informs us that Hesperus is Phosphorus,
> > an *a posteriori* truth (it is neither trivial to claim it, nor absurd
> > to doubt it).
> > 
> > So I add to your list:
> > 
> >    c) metadata informs us that the two URIs name the same resource.
> 
> I think that this is a higher level migration or 
> re-writing layer (converting one document from 
> namespace X into a document from namespace Y).  
> 
> That the current spec says *unique* but allows for
> *non-unique* URIs is the core problem here, let us
> solve it.  Putting this metadata system as a requirement
> for an implementation of the namespace specificition
> would be a step backwards.

That's the point. You don't want to have to have such a metadata system.
Therefore you are left with only comparing either the Resource or the
Identifier. You can't compare the Resource because it may have policy
along with it. You left with comparing Identifiers and nothing else.
Thus you are only left with the ability to say that two Resource
are only ever equal if they have the same Identifiers...

> For deterministic processes we need *unique* and 
> *persistent*  names; it *is* what the spec says.
> 
> Yes?

Sure. URIs are unique (and injective, surjective). They just aren't
persistent...

-MM

-- 
--------------------------------------------------------------------------------
Michael Mealling	|      Vote Libertarian!       | www.rwhois.net/michael
Sr. Research Engineer   |   www.ga.lp.org/gwinnett     | ICQ#:         14198821
Network Solutions	|          www.lp.org          |  michaelm@netsol.com

Received on Monday, 5 June 2000 10:27:00 UTC