>
> 1. control of extension and restriction
> using the XSD derivation rules
> is enforced by compliant tools
That is not an advantage of substitution groups. O
> 2. the XSD derivation provides a trail
> - a trace back to the types
> from which derivations were made
> (helps with audit requirements, etc) 2a. compliant tools
> provide a better,
> richer experience to developers
But the trail is the library itself as resident in the schemas from the
different versions and the artifacts stored in the registry as UML
profiles.
>
> 3. this is becoming a more ubiquitous
> way to extend/restrict - audit
> rules tend to expect financial software
> to follow such ubiquitous
> software designs and methodologies
> (in my limited experience)
Please provide specific audit requirements that are satisfied by
substitution groups. It seems to me that you are saying that if I use
substitution groups, then I am extending the types. Therefore I need
specific audit rules that address this.
>
> 4. people are wanting XML to deliver
> on the eXtensible bit (see ubl-dev)
The extensible bit is in the vocabulary and how customizers use that
vocabulary and our normative schema. Not in how we create our normative
schema.
> 5. software vendors are increasingly
> investing in substitution group
> support (hence 2a above and more)
And Ford is building new cars, so does that mean I need to buy one?
>
> 6. this delivers on UBL's original
> promise of polymorphic inheritance
> (albeit at a price of a single rule
> about sg's being dropped or qualified)
I am still not convinced that this is a key aspect of UBL. There were
two people who supported this concept - and they are no longer with us.
>
> 7. this buys into the benefits of using
> global rather than local
Sorry - I completely disagree here. Our discussions of global were
centered around managing by both types and elements.
>
> 8. this best fits the OO development
> practises (see JAXB)
Not sure I understand this one. Our OO development is related to our
modeling at the data level - not how we create the types.
> I could go on...
Please do. So far I don't see any specific technical reason that
necessitates using substitution groups.
Mark
