Re: Namespace name: better to use URN or URL?

[Tim Bray <tbray@textuality.com>]

At 09:11 AM 04/09/01 +0100, David Carlisle wrote:
>Names based on the domain name system (http:// or mailto: etc) do have
>the good properties of being relatively easy to generate (relatively)
>uniquely and be memorable. 

They also have the fuzzy, intangible, but important advantage of 
being first-class citizens of the web.  When there's a Uniform
Resource Locator there is (potentially) a Resource.  I think it's
a good idea that vocabularies and their semantic elements be
a part of the fabric of the web, which they're not unless they're
addressable.  We're still working out how to use this capability -
for my ideas on the subject, see rddl.org - but it feels important.

NB this notion was first introduced to me by Dan Connolly and
TimBL, who are paid to worry about this kind of stuff.

And I agree about the egregious ugliness of UUIDs; if they
have a place anywhere in the world, it's down in the bowels
of COM objects.

>They have the bad property of tempting the
>designers of XML Schema to suggest that a schema validator might try to
>use the URI for something which famously "is not a goal" of XML
>Namespaces, but that's the price you pay, apparently.

The "not a goal" refers specifically to *schemas*.  We didn't
want either DTDs or XSDL or any other schema language making
a land-grab for the namespace-name address. -Tim