On the call:
Andrew Watson chairs Tom Rutt takes notes
There was focus on giving RMS FTF guidance on how to solve their xml schema “problem”
Problem: They have a uml domain data model, and they have generated a “user friendly” xsd from that uml model using a tool, and that is being positioned as a psm for xml
They too generated an xsd which takes all associations, and results in a containment tree rather than using xml references.
What people want to do is define uml interfaces with operation parameter types which are defined as classes in a domain specific data model expressed as a set of uml class diagrams. This cannot be done if the data model has to be cast as a mof model rather than a UML class model.
Start from UML domain application model and use QVT to transform from UML metamodel to the XSD metamodel, with a user friendly mapping could also recast the rules in mof to xmi spec which specify how to construct an xmi schema, as a qvt mapping from the CMOF metamodel to the XSD metamodel.
Pete: xmi spec will have qvt for going from cmof to xmi.
Pete: we should have in addition, a useful OMG resource which transforms from the uml metamodel to a user friendly xsd realization
Andrew: we could put that useful OMG resource somewhere on an OMG web page, perhaps the smsc wiki page
Every community of users have their own set of rules about what constitutes a user friendly XSD. However, we might be able to have a xsd form which most people find easy to use.