Re: FW: [ubl] UBL master data for maintenance

[jon.bosak@sun.com]

[sue.probert@dial.pipex.com:]

| The use of spreadsheets as the 'master' of the library data
| results inevitably in errors caused by mistyping and due to the
| lack of any CCTS structural checking. This is a fact
| well-proven. Of course, spreadsheet approaches, even non-MS
| proporietary ones are always possible and will remain so.

Folks, this is not about where we keep the data model.  Our
decision (my momentariy lapse of consciousness a few days ago
notwithstanding) was to use EF as the repository for the data
model during the 1.1 production phase.  That's not in question,
and that's not the problem.  The problem is how we generate
schemas conformant to our NDRs regarding minor versions from
inputs (whatever form they may take) that express the models as
unitary wholes.  In other words, given a revised 1.1 model of
Invoice (say) that has been augmented by content additions to the
UBL 1.0 invoice, how do we then generate a 1.1 Invoice schema that
takes the form of some new parts plus an import of the 1.0 schema?
This problem is a serious one and has almost nothing to do with
what tool we're using to generate the schemas; we would still be
facing it if we were using UBLish or perl scripts or whatever
mechanism you like.

We will be discussing this in upcoming TC and SSC calls.  Please
let the non-issue of "master data model" go and focus on the
problem at hand; it's going to require attention from all the
teams to solve it.

Jon