Re: URIs quack like a duck

David Carlisle wrote:

> > But I'd add that creating a new attribute would relieve much
>
> Since the only thing so far suggested to put at that URI is a schema
> file, the attribute is already specified, xsi:schemaLocatation.

The problem with that, I think, is that an element can have several
xmlns attributes, for different namespaces, attached to it.  So there's
then a question as to which namespace the xsi:schemaLocation refers
to.  (I assume we want to keep attributes order-independent.)   The new
attribute would specify both a prefix and a schema location.

By the way, adding a new attribute (let's for the moment say we call it
xmldtd:foo) provides a way of dealing with an existing troublesome
problem: validation according to DTDs.  We can then implicitly prefix
all the names in the DTD located by xmldtd:foo with the prefix -- and
behold, validation now works.

Validation according to schemas would use xmlschema:foo.  Or,
alternatively, xmlmd:foo that locates some self-describing package.

Paul Abrahams

Received on Thursday, 1 June 2000 12:33:00 UTC