Lists Home |
Date Index |
From: "Ronald Bourret" <firstname.lastname@example.org>
> The good news is that this solution is already known to work. The bad
> news is that it would cause backwards-incompatibility problems --
> documents that are both valid and namespace valid would cease to be
> valid due to a lack of xmlns attribute defaults in the DTD.
Interestingly, Part 7 of the discussions for ISO DSDL (Document Schema
Description Language) is "Namespace-aware processing with DTD syntax".
Presumably that would rubberstamp the namespace idea, since it has
proved itself to a significant user-base, and remove any objections based on
standards for ifever W3C wanted to make DTDs namespace-aware, say in XML 2.0,
when that eventuates. Part 7 has no editor yet, so I guess it is a slot indicating
that the ISO WG 34 does not want to get in the way of any good progress
rather a promise.
Another reason for namespace-aware DTDs is that then these can be regarded
as abbreviated, limited syntaxes for XML Schemas or RELAX NG. I think
James had that interesting tool which maps all uses of parameter entities to
RELAX NG. Presumably, apart from entity references, the one thing that
DTDs have that XML Schemas (and RELAX NG) do not reconstruct is
conditional sections. For conditional sections in DTDs (all that XML allows),
I think that Schematron's phase mechanism suffices: I wonder whether
phases could be grafted onto RELAX NG?
(not writing on behalf of any body)