Re: Is a namesapce a resource? - was: duck

>Tim's said "no", and I beleive him.
And I believe you (although I just looked and didn't find such a clear
statement)

What I did find him say was this:

>> All you can say is that any document [returned by dereferencing the
>> namespace URI] should be a representation of the  namespace.

which if intended as a statement of sensible practice is fine, it's
less fine if it was intended as an indication that a namespace
processor can rely on the document being returned by the namespace
URI (which can be any document accessible on the web).

The document returned by dereferencing  the namespace URI of this
xmlns="HTTP://WWW.W3.ORG/XSL/Transform" does not
describe that namespace at all.

> The question of whether there is a resource retrievable via that name
> really is a completely independent issue.

There are two questions, whether there has to be a retrievable
entity to which I think everyone agrees the answer is no.
Even if a URI scheme that allows retrieval is used.

But  I would argue that even if there is a retrievable entity there is
no necessary connection between that docuement and the namespace.
Until a week ago I would not have thought that was at all contentious.
                        
> I freely grant that this is unintuitive to those of us who think the
> purpose of a URI is to actively access a resource rather than to passively
> represent a resource.

I don't think it's a case of access/represent.
My understanding is that to name a namespace I take the URI of some
resource I control (for example my home page) If I do this then
the resource identified by the URI has no connection with the
namespace. 

But actually I can see that having a namespace aways have an absolute
URI (+ frag id) has advantages for everyone, so

"forbid"
and
"fixed-base"

do seem to be possible options to consider.

> See above; the two points are seperable. Using URI _references_, which
> opened the door to the relative syntax, was unintentional

It may not have been intended by someone, but one of the authors
clearly documented the effect of the literal interpretation in section
1 of the spec, where it is explicit that documents with different
base URI might mean that different namespace names are functionally
equivalent as URI refernces.

ie xmlns="foo" in a document and xmlns="../foo" in a document in a
subdirectory are different namespaces but refer to the same absolute
URI if viewed as URI. I could see how someone could accidentally
say "URI reference" instead of "URI" and so "unintentionally" let in
relative URI references, but I don't see how anyone can document
the behaviour of relative URI without intending or at least knowing
that they were allowed.

David

Received on Friday, 9 June 2000 17:00:33 UTC