Looking at the details on the dual schemas idea

["Stephen Green" <stephen_green@seventhproject.co.uk>]

Greetings

Just further considering the idea that 
there might be a way to have both global
(UBL NDR) and local (ATG2 NDR) schema
sets which each validate the exact same
instances:- If that were achievable (albeit
with compromises to the strictness of the
local rules, which seems unpalatable) I
had the idea that a stylesheet transformation
might be a way to generate the local from 
the global schema sets. 

I started to demonstrate that the transformation
seems quite feasible BUT the main problem is
that local schemas are by nature less
modular with less room to separate out 
different namespaces. This means that it
isn't really feasible where the global schemas
are modularized such as between BBIEs and
ABIEs. UBL's design has a document in one
namespace, BBIEs within the document
structure in another namespace with another
prefix and ABIEs within the same document
in another namespace again with another
prefix again. This distinction between
namespaces and prefixes from the instance
point of view can't be represented with local
design rules without making BBIEs global, in
a separate module, and putting the document
complexType in the ABIE module. [See
example attached]. If this is done then it
should, I reckon, be a straigtforward (but not
exactly trivial) matter to transform between
NDRs. Then the main difficulty left is in 
transforming in minor versions. 

[This is where it may become necessary to keep
to a transformation from global to local rather
than local to global, due to the greater amount
of information likely to be contained in global.]

However, the main consideration, it seems to
me, is not the fact that minor versions may be
so different and require complexity in XSLT
transformations (say with XSLT 2.0) but the
fact that this is all only feasible, as far as I can
see, with a departure of the local NDR away
from all local to some local, some global with
a similar modularity to the global NDR.

All the best

Stephen Green

GlobalLocalSchemaTransformation2.zzz