Lists Home |
Date Index |
One doesn't need to change anything.
A DTD can be interpreted to provide the subset of the information set
permitted by REC-xmls-1-20010502, Appendix D.
As such a DTD is a permissible alternative to the XML-encoding for schemas.
Which means it can, as it stands, be used as the basis to generate the
PSVI for a subset of documents. Those which don't entail models or data
types which cannot be expressed in a DTD.
CL-XML does that for namespace aware DTDs. Actually, it doesn't care
whether namespaces are specified in the document or not. In a documents
without namespaces all names are simply in the infamous "null" namespace.
The resulting post-validation document model analogous to the PSVI.*
Which makes it a "minimally conformant schema aware processor". At least
according to REC-xmls-1-20010502, 2.4. Hey.
[* It doesn't retain the specified details about the validation results
and some of the information would need to be generated from the model,
but these issues are not to the point.]
Jonathan Borden wrote:
> Since it doesn't change the syntax, nor rules of XML 1.0, this _could_ be
> done perhaps as an update to XMLNS, but without modifying XML 1.0.
> Define "Namespace Valid" as distinct from "Valid" just as XML Schema defines
> "Schema Valid" without changing XML 1.0
> That is to say, a proper "non validating parser" would be enabled to signal
> "namespace validity", (since DTDs still deal with entities etc, I am not
> sure this could be defined as a property of the 'infoset' but rather the XML
> I don't see a problem, what am I missing?