Re: The 'resource' identified by a namespace name URI should be the namespace

David Carlisle wrote:

> > sense that the namespace spec and the W3-specified namespace names give not
> > the slightest hint that such variants of URI were ever contemplated.
>
> I (and most other people) read the namespace spec as saying any URI
> reference is allowed as a namespace name.

Allowed?  Of course!!! Expected?  Well, all the published examples I've seen use
the http scheme.

> How would you define "expected" names?

I wouldn't.  I'd put something into Namespaces Revised, probably in a non-normative
appendix, that would say how different kinds of URIs might be used in other
contexts, or not used at all.  For instance, a mailto URI wouldn't be useful for
retrieving a schema.


> > My possibly very mistaken impression (counterexamples, please speak
> > up) is that almost all of its defenders are those who were intimately involved
> > in creating it.
>
> well I wasn't involved in creating it:-)

Counterexample recognized.

> > The fundamental problem with the namespace spec is that it creates misleading
> > expectations.
>
> This leads to incorrect expectations for absolute beginners, (and I've
> never said I think it was the best possible idea) It certainly is not
> sufficient cause to change the spec now, and not sufficient cause to
> create the new type of XML document whose element names depend on
> context that the absolute proposal implies.

I'm not advocating changing the spec, only the verbiage.   Here's an example of the
kind of statement I'd like to see in a revision:

"The attribute's value is the namespace name identifying the namespace.   It must
have the form of a URI reference, although for the purposes of this specification
the namespace name is treated as an uninterpreted character string.   Other
specifications and applications may choose to attach their own interpretations to
the namespace name and to place additional requirements on its form or
interpretation.  (URI references are used in this context because they allow such
additional interpretations.)"

None of this contradicts what is in the spec now.  It presents a different way of
describing what is there now.

> > The namespace spec would work just as well if namespace names were
> > required to be serial numbers generated by some algorithm given in the
> > namespace spec itself.
>
> yes of course, but what algorithm would you have suggested they used
> (to get globally unique identifiers without use of a central registry)?

There are lots of such algorithms around, as several people have pointed out, and a
central registry wouldn't be such a bad idea anyway.

> > The namespace spec provides no statement whatsoever, except for an odd
> > disclaimer pertaining to schemas, about what kinds of URIs are appropriate,
> > what resources they might identify
> Because every URI may be used as a namespace name.
>
> > That URIs should be persistent seems to contradict the statement that it
> > doesn't matter what they identify.
>
> Not at all it means that I shouldn't (but I do) use namespace names
> starting with http://www.dcarlisle.demon.co.uk as I can't guarantee
> that those names won't be used by someone else later (if I stop
> paying demon and so lose control over dcarlisle.demon)

Hmmm.  Suppose that you don't pay up, and so someone else gets control over this
URI.   I don't see how that negatively impacts the use of that URI as a namespace
name, at least as far as the namespace spec is concerned.

Paul Abrahams

Received on Tuesday, 6 June 2000 14:14:14 UTC