RE: [office] Clarification re: DSDL and W3C Schema

[Sean McGrath <sean.mcgrath@propylon.com>]

I agree with John. There is a lot to be said for making the normative
expression of the constraints independent of schema-language du-jour.
Some BNF-ish notation along with narrative text is one possibilty.
This will of course be imperfect, true expression of the specification
can only be seen through running code. However, I think this bears
consideration as putting one particular schema language "on top"
to generate the others mechanically risks:

1) creating an impression that one particular schema language is superior
to the others (lets not start that debate!)

2) mechanically generated schemas will always be yucky compared
to hand crafted ones - creating second class citizens of the generated
notations.

regards,
Sean McGrath
http://seanmcgrath.blogspot.com


At 21:04 08/01/2003 +0000, John Chelsom wrote:
>One reflection on the DTD/Relax/W3C Schema debate:
>
>We seem to be assuming that we will need to perform an automated (and
>probably imperfect) transformation from a normative syntax into others that
>people may want to use in their own implementations. We could, of course,
>use any notation we like for the normative expression of the specification,
>so long as it unambiguously defines the XML structures in the open office
>format. We could then publish implementations as DTD/Relax/W3C Schema,etc
>with comments where necessary on how each syntax does (or does not) support
>what we have defined.
>
>As far as I can see there is no necessity to define an automated
>transformation from our specification to any given format. Our problem is
>(almost) the same one that has been faced by activities such as UBL and HL7
>(for those of you interested in a far bigger manifestation of the problem in
>the healthcare domain). The problem is that XML schema languages (I think
>all of them, but I someone may want to correct me) are not rich enough to
>specify the logical information models required in those specifications.
>
>The only thing that we definitely need to do is make sure our specification
>is complete and unambiguous in itself. We can then add examples of
>syntax-specific implementations to help people implement their own
>applications, but these aren't normative and (almost certainly) won't be
>generated 100% automtically from the specification itself.
>
>Regards,
>John
>
>***********************************************
>CSW Group Ltd
>4240 Nash Court
>Oxford Business Park South
>Oxford
>OX4 2RU
>Tel: +44/0 1865 337400  Fax: +44/0 1865 337433
>Web: http://www.csw.co.uk
>
>Delivering value to our customers by deploying innovative, web-based,
>products and solutions for the integration of processes, data and knowledge
>***********************************************
>CSW's XML Summer School 27th - 31st July 2003
>Wadham College, Oxford
>Further information: http://www.xmlsummerschool.com
>***********************************************
>Legal Disclaimer: http://www.csw.co.uk/disclaimer.htm
>***********************************************

http://seanmcgrath.blogspot.com