Re: Using urn:publicid: for namespaces

> 1) The name assigned to a concept within a schema (which is
> unique)
> 2) A location of the definition of the concept

Where 2) is, of course, just a variant of 1). I argued about this
recently with Aaron Swartz, who uses some of his pages at
logicerror.com as predicates. I asked him, what if I wanted to talk
about some of those pages?

Now, he and I both know the situation; that "what a URI identifies" is
a resource, not the representation of a resource that is returned on
dereferencing it. But this leads to screwyness, in that people
naturally want to do things like:-

<http://logicerror.com/likes> rdfs:isDefinedBy
<http://logicerror.com/likes> .

rather than the "correct":-

<http://logicerror.com/likes> rdfs:isDefinedBy
   [ :dereferencedFrom <http://logicerror.com/likes>; date
"2001-08-10" ] .

This is a mess.

TimBL even appears to believe (according to [1]) that HTTP URLs can
only identify generic documents, and therefore all namespaces in HTTP
space must end in a "#". This is of course incorrect: an HTTP URL
doesn't necessarily identify only documents, as the range of HTTP
response codes in RFC 2616 will tell you. Perhaps HTTP could do with a
"this is information that defines the resource you are looking for"
response code, but there you go...

Yes, there are a number of problems, and no, people don't seem to be
decided about how to resolve them. I still think that namespaces
ending in "#" are a decent option, notwithstanding the fact that what
URI References denote is not explaned in all that much detail in RFC
2396, and that it is media type dependant... but URI use is
contextualized, so I don't think it's as much of a problem as people
make it out to be.

The Dublin Core namespace "works", and that ends with a "/". The RDF
namespace "works", and that ends with a "#". The bigger problems are
namespace persistence, and bad management thereof.

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
:Sean :hasHomepage <http://purl.org/net/sbp/> .

Received on Monday, 13 August 2001 15:46:34 UTC