[topicmapmail] XML Schema for XTM DTD
Kal Ahmed
kal@techquila.com
08 Feb 2003 17:34:53 +0000
On Thu, 2003-02-06 at 23:09, Lars Marius Garshol wrote:
>
> (Max and Daniel: please join TOPICMAPMAIL, and stop using the OASIS
> list. That list really shouldn't be used any more.)
>
> This is an attempt to summarize the thread, which has really gone all
> over the place, as threads tend to do.
>
> It seems that there is a significant number of people who want an XSDL
> schema for XTM. Some want it for political reasons, others because
> their tools require it. To me that's reason enough, and it means that
> the next XTM syntax specification should include an informative XSDL
> schema.
>
> Since this will be published by ISO SC34, which also creates RELAX-NG,
> I think we should add an informative RELAX-NG schema as well, for
> political reasons. The extra work required is slight, and I've always
> wanted to learn RELAX-NG anyway. :)
>
> As far as I can see, both schemas should constrain the XTM syntax as
> it will be as tightly as possible, but do nothing beyond that. The
> syntax itself should remain the same, and the DTD should continue
> being the normative specification of the syntax, together with the
> prose.
>
> Comments? Opinions?
>
Yes!
Later in the thread, Tony Coates wrote:
> I would certainly suggest that the way ahead for an ISO spec is to use
> RELAX NG as the normative schema language in future, rather than DTDs.
> I would also suggest that it would then be a good process to generate
> the W3C XML Schema from the RELAX NG schema automatically using James
> Clark's Trang, rather than creating the W3C XML Schema in a way that
> requires manual intervention. That should aid the quality control.
...which makes a lot of sense to me in many respects. On little nit is
that currently the XML DTD version of XTM is normative...but as I think
the momentum is away from XML DTD and toward the schema languages,
moving to a normative RELAX NG schema with informative XML DTD and W3C
Schema is the right thing to do.
Cheers,
Kal