Re: Language = Namespace. was: How namespace names might be used

-----Original Message-----
From: Paul W. Abrahams <abrahams@valinet.com>
To: Tim Berners-Lee <timbl@w3.org>
Cc: xml-uri@w3.org <xml-uri@w3.org>
Date: Monday, June 19, 2000 5:42 PM
Subject: Re: Language = Namespace. was: How namespace names might be used


>Tim Berners-Lee wrote:
>
>> - a namesapce is identified by a URI.  (That is, if any resource is
>> identified by URI u, and a namespace is identified by URI u, then that
>> resource *is* that namespace)
>
>To pursue the question of what you mean by that: are all URIs equally
suitable
>for that purpose, assuming only that the URI is chosen so that one can be
>reasonably confident of its uniqueness?   If not, how would you distinguish
the
>suitable URIs from the unsuitable ones?

I don't mean any such assertion about the "suitability"of any particular
URI.  That is not in scope for this (xml+ns) language to define.

I would distinguish suitable ones from unsuitable ones as a user
by considering the qualities I need in my particular case.
Do I want high persistence? Do I want people to be able
to look up something I have to say about the namespace?
What namesapces do I control?  Do I want to bind the
namespace to one particular representation, or one
which can be varied in time?  Who do I want to be responsible
for ensuring the longevity and non-reuse of the identifier?

As w3c webmasterinchief, for example, I constrain namespace
names to be in http://www.w3.org/2000... (this year) because only there
do I belive we can practically guarantee the W3C persistence policy;
because I do belive that it is useful to the community that for W3C
namespaces a schema and a pointerto the spec (without predjudice to later
something more fancy)
should be available for  the auomated verifiation of the syntax of
documents,
and so on.  But this is just part of the management of the W3C
administration and policy which, like for the logo and averything else
on the site is set by lots of meetings considering social requirements.

So I wouldn't answer these questions for you, as I would
not take it upon myself to tell you which requirments were most
important for you. I would just point you back at earlier discussion
on this list about the merits of different schemes, ans at the URI specs
and at current practice for other resources.


> There certainly does not seem to be
>any agreement as to what if anything should be at the resource identified
by
>the URI, if indeed such a resource exists at all.


The resource is an abstract thing - it *is* the namespace.

There is no agreement about the best ways of representing
a namespace in an entity body, perhaps.
But we don't have to settle that before we go forward.

>Paul Abrahams
>
>

Received on Wednesday, 21 June 2000 17:00:05 UTC