Re: [wsbpel] Issue - 46 - Namespace for the document fragment representing a part

[David RR Webber - XML ebusiness <Gnosis_@compuserve.com>]

Sanjiva,

This was my point earlier.

Usage scenario #1

If you have simple structures with no complex dependencies
and context switching - then XPath approach and schema is
OK - lightweight - local, static and basic.

Usage scenario #2

If you have large structures with complex interdependencies
and structural variations - schema cannot do explicit context
selection - it tells you about all possible permutations - but
not explicit ones.  That's what the OASIS CAM spec' does 
for you - its designed for that, and to give you much more - 
like lookups to external sources, and versioning.

The third option is to use XSLT.   This again works for
#1, but gets very hairy for #2.  You can look at what
the OAGIS folks did exactly in this area - and its
unmaintainable - that's why they are considering
CAM.    Other people have tried Schematron, but
again - that's OK only to a point.  BTW - CAM inherits
alot from schematron and then adds more.

Thanks, DW.
=======================================================
Message text written by "Sanjiva Weerawarana"
>
I do realize the query + assign stuff makes it possible to do some
screwy things that could make the result invalid. I think that's
the issue Yaron has raised (#11?). I personally don't think there's
much that can be done about it given the richness of schema (for
example, one may assign stuff inside an element and then maybe that
element itself never appears .. then what), but I certainly appreciate
people trying to be more rigourous about it.

Sanjiva.
<