Re: TB16 Re: Comments on arch doc draft

Joshua Allen wrote:

>
> It is the old "You can't dereference a car" debate.  If we say that the
> namespace name is the URI for the namespace, it can't also be the URI
> for the web page.  Because then how would you know what a particular
> assertion was talking about?
>
> For example, the assertion
> "http://www.soapware.org/ns/foo dc:creator joe@blogs.com"
>
> Are we saying that Joe created the namespace, or the web page?  Which is
> it?


It depends, for example:

<http://www.soapware.org/ns/foo> rdf:type xml:Namespace .
<http://www.soapware.org/ns/foo> dc:creator <joe@blogs.com> .

is different from:

<http://www.soapware.org/ns/foo> rdf:type doc:Document .
<http://www.soapware.org/ns/foo> dc:creator <joe@blogs.com> .

that is to say that URIs and URI references can be treated opaquely. What
URIs identify is a whole 'nother issue.

>
> A URI is supposed to be an identifier, and if we say that a URI is the
> identifier for any arbitrary number of things, we sabotage the entire
> purpose of URIs.

Yes but that is an entirely different argument, for example one can say

xml:Namespace owl:disjointWith doc:Document .

to assert that XML Namespaces and documents are disjoint, i.e. the URI
identifies one or the other.


One of the main arguments against the use of http:
> URLs is that this causes confusion and overloads the concept of
> identity.

I don't follow how "http:" causes, adds to, or helps solve this problem.

> If it is OK to say that an http URL can simultaneously
> identify a namespace and a web page, what's to stop someone from saying
> that the same URL is also a car and a butterfly?
>

see above e.g.

ex:car owl:disjointWith ex:butterfly .

Jonathan

Received on Tuesday, 2 July 2002 10:31:54 UTC