Re: Code list schema

[jon.bosak@sun.com]

[brs@itst.dk:]

| for political reasons I think it best not to have any validation
| done via schematron that could theoretically be done using XSD.

Schematron is an international standard (ISO/IEC 19757 - DSDL
Document Schema Definition Language - Part 3: Rule-based
validation - Schematron) and therefore just as "official" as
anything that W3C has published.  It appears to be reasonably well
supported in standard tools at this point, too.

The idea here is to provide an out-of-the-box two-stage validation
environment, the first pass being XSD schema validation and the
second being Schematron (SCH) schema validation.  If we can do
this in a way that is no more trouble for users than XSD
validation alone (and this is my chief concern), then it seems to
me that we have provided a huge value to the users, because now
they can augment the Schematron schemas we give them in the
release package to provide a vastly richer rules-checking
capability than they could achieve through XSD validation alone.

The key here will be ease of use; if we can describe how to set up
a two-stage validation environment in a simple README using
software freely downloadable from the web, then I think we will
have greatly increased the usability and attractiveness of the UBL
2.0 package.  For maximum ease of use, we should not only provide
XSLT to generate Schematron for any arbitrary code list instance
conforming to our standard code list schema, but also provide
generated Schematron for the code lists actually referenced in UBL
2.0 and show how to assemble all this into an environment that
"just works" (in the same way that the XSD schemas in 1.0 "just
work").

I haven't used Schematron, so I'm making some big assumptions here
about the feasability of what I've just described.

Jon