Re: pressing question about containermembershipproperty syntax

At 03:42 PM 12/21/02 -0500, Dan Brickley wrote:

>* Brian McBride <bwm@hplb.hpl.hp.com> [2002-12-21 11:18+0000]
> >
> > My take on this:
> >
> > The namespace
> >
> >   http://www.w3.org/1999/02/22-rdf-syntax-ns#
> >
> > is controlled by W3C.  We can say that it does contain the name _:1 and it
> > does not contain the name _:01.  I have never seen it suggested before 
> that
> > _:01 was legal.  I suggest that we make it clear in the vocabulary 
> document
> > that it is not.
>
>In 1999, W3C published a recommendation that was all about the namespace
>called "http://www.w3.org/1999/02/22-rdf-syntax-ns#". At that time W3C 
>gave no hint
>that this namespace contained things named 'nodeID', 'datatype' etc. The 
>spec was
>somewhat elusive on whether one could expect in the future for revelations 
>about
>such things to be made. In 2001-3 we've revealed our view that there are 
>other
>occupants of this namespace, ie. nodeID, datatype etc.
>
>I believe we need to decide whether to 'seal up' this namespace with the 
>publication of
>our specs, or whether to allow the possibility that further occupants of 
>the ns may
>be subsequently 'revealed'. Perhaps we can do this incrementally, by saying
>that the ns remains potentially open for future additions, but that we 
>definitively
>assert that it doesn't contain anything with a name beginning with '_0'.
>
>In other words, we should decide whether we're claiming the RDF specs 
>provide an
>exhaustive account of the things that are named in that namespace.

I think it's a reasonable and pragmatic approach to say that it doesn't 
contain names beginning "_0" (for reasons of possible confusion of 
container vocab.) without saying that the namespace is sealed for all time.

Looking beyond that case, I suggest that it's probably appropriate to say 
as little as necessary about the future population of this (or any) 
namespace.  Maybe it is more important to say what should be the 
expectations of any namespace used for RDF vocabulary (a kind of "cool 
URIrefs don't change" exhortation, maybe)?

#g


-------------------
Graham Klyne
<GK@NineByNine.org>

Received on Sunday, 22 December 2002 05:47:49 UTC