Re: URIs quack like a duck

> That identified by a URI is a Resource. If you use a URI to name the
> namespace then the namespace is the corresponding URI.
> That is how URIs identify things.

I played no part in writing the namespace spec I just read it.
It was initially surprising but in fact it works quite well and
much of XML development over the past 2 years is built on it.

The namespace name is _not_ the URI of the namespace as a resource.
The only criterion is that you pick a URI reference. It may be
an absolute URI of an existing resource, or it may be relative in
which case it may refer to many different resources depending on
context but it is not absolutely not forcedto be the URI of a resource
representing the namespace.

For all namespace processing the namespace name could have been an FPI
(which you wouldn't then have tried to dereference) the _only_ reason
ever put forward for using URI is that by using a resource that you
control you can get a unique name. But you can pick any resource that
you control: the resource doesn't have to have anything to do with the
namespace. So URI are only used as a string allocating mechanism
(you probably call that URI-abuse, but as I say, it aint my fault)

It is called the namespace _name_ if the namespace is a resource
it will also have a URI but its URI won't be its name.

I suspect that you'll say that this is horrible but it is I suspect at
the core of the disagreement and why some of the messages from the
"uri axiom" camp have seemed unintelligible to those of us using
namespaces on a daily basis. If you have the underlying assumption
that dereferencing the namespace name will return something relevant
to the namespace then you will find most namespace processing strange
and or broken.

David

Received on Saturday, 3 June 2000 17:56:46 UTC