Re: [xml-dev] Service Constraints

[Stephen D Green <stephengreenubl@gmail.com>]

> So I thought I'd invite members of this community to comment on
> whether you favour very open lightly constrained schema when defining
> business services (at least when expressing the data content as XSD)
> or whether you prefer to try and specify all constraints even at the
> risk of generating greater churn (aka: incompatible change).
>
> Fraser.

I would very much like to see a 'standard' pattern or set of patterns
emerging for these scenarios. The closest I've seen to something
workable is an open lightly constrained schema to which are applied
additional constraints where the open schema is a fixed standard and
the extra constraints are more flexible and adaptable. (Similar to the
pattern in databases of tables overlaid with views, though there the
additional facility is that a view can cover multiple tables and joins.)
I'd like to see improvements in XML Schema to better fit this pattern
- hence I'd like to see something developed along the lines Roger
proposed recently for more abstraction in types, such as his
crossProduct but with some way to apply use the extra constraints
even to a crossProduct and in doing so to turn abstract into concrete.
I'd especially like to see better ways to overlay the extra constraints
on top of the schema such as special facilities in Schematron and
some alternative methods such as some I've explored with the
in-progress OASIS Test Assertion Markup Language XPath profile,
but I'm not sure either of these yet quite do it.

The alternative and a very real one which is by now a well-trodden
path is to do all this outside of the markup and schema in a model
which is applied to the schema with a kind of late binding and the
standard used in many cases for business documents, as you know
of course is ISO 15000-5 CCTS, as part of the ebXML stack. Here
there are abstract, definitive constructs defined as 'core components'
and concrete types based on these in the model layer called
'business information entities'. This allows the representation of the
same constructs in both XML and other technologies such as EDI.
I'd like to see the way XML can represent CCTS models improved.
CAM is nicely positioned for this because it seems to me to be a
schema language which is CCTS-aware. Combining CAM and XML
Schema is even better. I think the whole thing is still in its infancy
and the XML technologies in this area have a long way to go. I
remember Jon Bosak once saying in an early presentation of the
proposals for the Universal Business Language something like
"surely we can do better than subsets". One day I hope we will.

Thanks for the question.

Steve

----
Stephen D Green