xml-dev - RE: Does DTD validation work with namespaces?

[Jonathan Borden <jborden@mediaone.net>]

Richard Tobin wrote:
>
> Jonathan Borden wrote:
>
> >Actually XML names already redefine:
>
> >doctypedecl
> >elementdecl
> >[...]
>
> No it doesn't.

	Call it what you will, but XML 1.0 has productions for doctypedecl and XML
Namespaces has an, albeit slightly, different poduction for doctypedecl, to
me that is a redefinition: one adds additional restrictions, in specific the
substitution of QName for Name, but nonetheless a redefinition.

> To conform to the namespaces spec a document must meet
> certain extra constraints on these productions, but all
> namespace-compliant documents are XML 1.0 documents and their XML 1.0
> meaning is unchanged.  On the other hand a change like:
>
> > extending the syntax of EntityDecl
> > to allow the specification of an implicit prefix for an included
> > DTD
>
> would change the XML 1.0 meaning of documents.  In particular, it
> would make documents valid that were otherwise invalid.

	I'd need to see this proposed change to EntityDecl, but in principle one
would *want* certain documents which are 'invalid' under XML 1.0 to be valid
under XML1.0+Names IFF the reason they are invalid is that the namespace is
the same but the prefix has changed.

	On the other hand if you really mean to say that documents that are
currently not well formed would become well formed, that is a different
issue. On the other other hand, unless DTDs can be made to place nice with
Namespaces they are history as far as most people are concerned. So on one
hand we are playing with the 'purity' of XML 1.0, but on the other hand we
are orphaning a large chunk of XML 1.0. If a Schema spec as simple as DTDs
were on the table, it would be an easy choice.

Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org