Fwd: RE: [ubl-dev] SBS and Restricted Data Types

[stephen.green@systml.co.uk]

Forwarded extract from ubl-dev re question from Martin:

> Could you give an example of how to use the xpath-files when actually
> validating the document instance? Is it possible to validate that the
> sender follows the subset and make sure that he sends only what we have
> agreed upon (and no more).


----- Forwarded message from stephen.green@systml.co.uk -----
    Date: Wed, 03 May 2006 05:56:56 -0600
    From: stephen.green@systml.co.uk
Reply-To: stephen.green@systml.co.uk
 Subject: RE: [ubl-dev] SBS and Restricted Data Types
      To: ubl-dev@lists.oasis-open.org

Hi Martin,

Excellent question. Simply put I'd say you hand the subset definition
to a developer and ask him/her to produce a Schematron schema or just
write code (such as Java or C++ or perhaps Python or Perl or VB/C# or
even shell script or DOS, whatever) but there might best be a way to
allow both parties in the transaction to use the exact same validation.
That suggests a Schematron schema being a good output of the UBL TC as
is being done for codelists. We'll see how things develope for UBL 2.

Another point: although the codelist methodology suggests a 'second pass'
validation - using first XSD for the first pass and the Schematron for
a second (which depends on the success of the first for accuracy), I'd
suggest it be not just 'two pass' but 'two phase' to allow several passes
in the second phase (Schematron or equivalent for codelists, subsets and
business rules, not necessarily in that order). Alternatively, since I
believe (put me right if I'm wrong) there is an import or include mechanism
in Schematron, I wonder whether one might not just have a single Schematron
logical schema comprised of modules for codelists, subsets and business
rules that can all be used at once. The first option, perhaps with pipes of
some kind, could be the more appropriate. But if onl;y two passes are allowed
then the second could perhaps do the trick. Furthermore there is a possibility
that there may be more than one subset per message; one public like the
SBS and
another more private or specialised for trading partnerships, marketplaces,
etc - which is a strict restriction (in XSD logic) of the first subset which
is in turn a strict restriction of the schema of which the instance namespace
is derived (which could be a customisation - restriction/and or extension -
of UBL, say).

All the best

Steve