[
Lists Home |
Date Index |
Thread Index
]
> Or is it just a bad idea to have multiple grammars with the same
> namespace and avoid xs:redefine.
Yes, it's a bad idea to have multiple grammars with the same
namespace.
And it's a good idea to avoid xs:redefine - for exactly this reason.
Actually, there really are many good reasons why one wants multiple grammars
for the same namespace, whether they are constructed using redefine or not.
Two good use cases for this are
(a) different validation rules apply to the same document at different
stages of a workflow or pipeline
(b) different validation rules apply to different messages that contain
projections of the same business object
I know several projects that are really struggling with these issues.
I'm coming to the conclusion that there's a need for a concept of a named
schema, which is a collection of named schema components. There's no clash
if two different schemas contain different definitions of the same name.
Quite how one integrates that into schema-aware XSLT processing, so that you
can transform an input object conforming to type S1.T into an output object
conforming to type S2.T (where S1,S2 are the schema names, and T is the type
name), I don't yet know!
Michael Kay
|
