Re: What are allowable property URIs?

Aaron Swartz wrote:
> Jonathan Borden <> wrote:
> >> In such a system, how would I represent any URI without a "#" in it?
> > as is. All I've done is to define a protocol for the conversion of a
> > into a URI reference.
> Exactly, but you force all the predicate URI references to contain the
> character # or _.  This doesn't fix the problem.

It doesn't 'force' anything. It simply defines a conversion of qnames to
URIs. URI references indicated by any other mechanism than a QName (e.g.
rdf:resource, rdf:about, rdf:predicate etc.) are free to include any URI
that is pleased. Since a QName doesn't directly refer to a URI in any case,
I can't see how changing the current specification from simple concatenation
of [namespace URI + localname] to [namespace URI + '#' + localname] could
cause any real problem.  We have already shown that the current behavior
works poorly or not at all for many applications -- in particular when
namespace URIs don't end in "#" or "/" -- and worse is not compatible with
other W3C recs such as XML Schema.

Rather than continue to spin our wheels on this issue, can you give a
specific example of a problem with my proposal i.e. a specific example with
specific URIs where this creates a problem **


** note: I can think of one particular issue but I think people need to
think this through.

Received on Thursday, 15 February 2001 16:48:06 UTC