Re: "Converting DTDs (and DTD developers) to RELAX NG schemas" paper

Daniel


  Since it's far far easier to validate the semantic of a Relax-NG
  Schemas it seems to me that doing the opposite way (i.e. having
  the Relax-NG master and relying on automatic generation for the XSD)
  is more likely to garantee the semantic of both schemas,


That's what we're doing in the OpenMath 2 specification
(http://www.openmath.org/standard/om20)

Writing the schema (in rnc) and deriving rng, xsd and dtd as needed.

There is though a question that has come up with regards choosing a
normative format.

In practice the schema is authored in the compact syntax. I have been
arguing that that is the one that should be normative (and in the body
of the OpenMath specification) as it is
a) readable and
b) the actual source.
However some have argued that the RNG XML syntax should be the normative
version as it is more standard (in particular further down the ISO
standardisation tracks).

What is the current feeling amongst Relax MG practitioners, is it OK to use
the compact syntax as a normative specification of a grammar?

David

________________________________________________________________________
This e-mail has been scanned for all viruses by Star Internet. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________