Dear Wendell,

Very enlightening message.
Some further comments:

At 20:34 29/01/2004, Wendell Piez wrote:

>>In fact, since Schemas have XML syntax, and they include more information
>>than the one required to build an equivalent DTD, it is very easy to
>>generate a DTD from a Schema using XSLT.
>
>This may be true given the right schema, but I wouldn't say that writing a
>stylesheet to provide for conversion of arbitrary schemas into DTDs would
>be "very easy".

For a given type of schema and a skillful XSL programmer is not difficult
to extract a DTD out of it. Something else is to build a general purpose
tool. But for earthly digitization projects, XSL can be considered a choice
when getting a DTD from a Schema is required.

>  (And I'd be interested to see the work of Alex and his
>colleagues and how they approach it. Extreme paper! :-)

Don't be so X-treme!  :-)

Now seriously: are you talking about our humble approach to multilingual
markup?
If so, you can find information about it at:
http://www.cervantesvirtual.com/research/mm/
It works pretty well for our DL project, but is not meant to be the
ultimate solution to tagset multilingualism. Just a starter.

There are better approaches for implementing parallel multilingual tagsets.
I'm currently working on this.

However, as this was an empty field of high practical interest for us in
Spain, we developed this solution meant also to inspire other people on
using markup in their own local language, while still getting the benefits
of TEI.

>Schema constructs
>such as abstract element types and so forth, which have no DTD analogue, or
>the fact that element content models can be context dependent, or a range
>of other problems, make it very tough in the general case.

Right. It is clear that the DTD you'll get will not be able to enforce all
those constraints and enhancements that Schemas now allow, and compromise
decisions will be needed when downgrading from a complex Schema to a DTD.
But a project that decides to get a DTD downgraded from a more complex
Schema will surely be a project that does not require the complexity of
Schemas and their newer features in the first place. Otherwise, they'd
better upgrade their working XML environment to make it Schema-aware and
start using Schemas instead. This may be an alternative worth considering!

No doubt DTDs have been left being the poor cousins in the Schema family,
but still good for many purposes.

Hope this enjoyable exchange of opinions has served to answer the doubts
raised by the initial message (and not frighten potential Schema users).
Schemas are among the best things that happened lately to markup
technology. Don't you agree?

Best wishes,
Alex.-


---------------------------------------------------------
ALEJANDRO G. BIA
e-mail: [log in to unmask]

Work address:
   Head of Research and Development
   Miguel de Cervantes Digital Library
   University of Alicante (Ed. Institutos)
   Apdo. de correos 99, E-03080, Alicante, SPAIN
   Phone: 34-96-590 9567     Fax: 34-96-590 9477
   http://cervantesvirtual.com/
   http://cervantesvirtual.com/research/articles/

Teaching:
   Department of Computer Languages and Information Systems (DLSI)
   University of Alicante (Politécnica 4)
   Apdo. de correos 99, E-03080, Alicante, SPAIN
   http://www.dlsi.ua.es/
   [log in to unmask]

Home address:
   Colonia Romana 13, Edif.1, Port.2, 7-E
   Alicante, SPAIN, CP:03016

---------------------------------------------------------