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

-----Original Message-----
From: David Carlisle <david@dcarlisle.demon.co.uk>
To: timbl@w3.org <timbl@w3.org>
Cc: xml-uri@w3.org <xml-uri@w3.org>
Date: Saturday, June 03, 2000 5:56 PM
Subject: Re: The 'resource' identified by a namespace name URI should be the
namespace


>
>> isa (x, mailbox) => not (isa(x, namespace))
>
>Finally we get somewhere!


I am so glad.

>> I can make assertions about a mailbox.
>
>Yes and if the assertions are rdf you use the mailto: URI for that.
>
>> But namespaces and mailboxes are distinct.
>
>Quite, which is why it is entirely wrong to use the namespace name
>as the URI for rdf assertions about the namespace (and why John C
>proposed an alternative URI for that use.)


You conclude the above from the fact that a namespace is not a mailbox?
Hello?  You conclude that a namespace, being identified a URI, should not
be the object of an RDF assertion?

Namespaces and mailboxes are both resources. They are distinct.
Both should be referred to by URI.

(The fact that you can tell from the URIU string that
mailto:xxx@yyy is a mailbox, by the way, is uncharacteristic of URIs.
In general, URIs do not imply anything about the type of an object.)

>Unless you are going to rewrite the entire namespace spec
>
><x xmlns="mailto:timbl@w3.org"/>
>
>is a conforming document and the namespace URI of the document element
>_is_ mailto:timbl@w3.org.

It is a "conforming document" but that doesn't mean it makes any sense,
as mailto:timbl@w3.org is my mailbox.  There are already a lot of
connotations
to it.  The identifier has been used. It is used in this message.  You can't
reuse the same identifier for something quite different.

Please distinguish between the basic syntactic constraints of the
specification
and the meaning.

 I could say that a document had
a namesapce "http://www.w3.org/2000/foo" but if I did I would not also
say that that was a panoramic picture of Sydney harbour skyline.
I could but it would lead people and machines to incorrect conclusions.
Those two things to me are both distinct and different resources.
I might admire a namespace, I might admire a photo.  I can use
a URI to refer to a namespace or a photo, without having to bolt on
an extra bit saying that I am referring to the "real URI space"or the


> If you want to make rdf assertions about
>the namespace you need to find a URI that identifies the namespace.
>The namespace name is not that.



Sigh.  The namespace identifier is not the namespace name.
You want to make a more complex world.  You want to add one level
of indirection, perhaps?  ("Anything you can do I can do meta" - quoted
by Henry Thompson, one of his colleagues).  Can we really not make
a system with one identifier for one concept?

We could have had a system which allows you in HTML to say

- this is the identifier of a picture

but life is so much simpler when we find the one absic concept and give it
one
identifier.  It is the fact that many systems and langauges can ref to the
same
identifier is what gives thwe web is power and its fundamental nature.

You can always use RDF to add lots of metadata about
related resources.  But designs in which a thing is identified differently
for different operations so not seem to be simple nor to have well-defined
properties.


Tim

Received on Wednesday, 7 June 2000 13:39:53 UTC