Re: [xml-dev]Changing Namespaces Between Specification Versions

[rjelliffe@allette.com.au]

David Orchard wrote:

> Personally, I like the idea of using the namespace name to indicate
> incompatible changes and version #s to indicate compatible changes.
>
> The issue of multi-namespaces and appropriate version #s is very
> difficult.  What is the version of a doc that contains n different
> namespaces, each which may be versioned compatibly and retain their
> namespaces?  Do we have to have a version # that is a composite of the
> union of all the namespaced items that may be present in the document?

The problem is not "difficult": it is to a great extent artificial, and
caused by basing our schemas on single namespaces, rather than having a
separate notion of "document type" which SGML of course had. I.e. caused
by XML Schemas as a direct function of its namespace-limited design.

In the absence of a concept of document type, there is no reliable
(first-class) way to specify "we use that standard vocabulary, but only
this subset".  Or "We use these standard vocabularies, but only these
versions".  All that has to be carried out of band, which is a euphemism
for by magic or by humans, which is a recipe for non-scalability and
disuse and irrelevance.

ISO NVDL could be used, for example, as a layer on top of XSD to provide a
kind of facility for identifying the specific schemas to be used for
different namespaces, but like XML Catalogs I think it would be more
effective in some kind of ZIP package to allow all the schemas to be
bundled.

Cheers
Rick Jelliffe