RE: XSchema Spec, Section 3, Draft 1 (Namespaces)

Simon St.Laurent (SimonStL@classic.msn.com)
Wed, 1 Jul 98 18:56:54 UT

>it's actually only the start. each time one wishes to bind a prefix to a
>region of a namespace, well, one is going to have to include a namespace pi.
>yes, that's once per document. don't look at me. don't even look at the wd's
>authors. it's inherent in the encoding problem.
>and yes, since a namespace pi does not permit docmentation, an element makes
>sense to perform that task. that's one extra.

And again, since the XSchema processor is already going to have to handle the
task of expanding namespace references in the name attributes of ElementDecl
elements, the PI is redundant and doesn't accomplish anything. The XSchema
processor is in charge of that task. We could choose to use the
less-accomplished PI instead of the XSC:Namespace element, but I see no
benefits to doing that.

If XSchema can do a better job for itself than the PI, I have no qualms about
letting XSchema handle a task, as I had no qualms about doing a better job
than DTDs currently can do.

I think we're going in circles here, and no one else is joining in. If you
want to include the extra PI, I don't think it'll do any damage. If anyone
would like to join the discussion, I'd very much like to hear other opinions
that might lead us out of this circle.

Simon St.Laurent
Dynamic HTML: A Primer / XML: A Primer / Cookies