Actually, we haven't discussed actual data models (or object models) to
any significant degree yet. What we have discussed are modeling frameworks
and (perhaps) methodologies. I don't think this discussion has reached a
conclusion, what we have done is put five proposals on the table and
discussed the general approach taken by each and some issues regarding what
it would take to get software components based on these approaches to
interoperate. Certainly, in the Case of VistA we have Fileman, HL7 and
various relational databases. The approaches are certainly not mutually
exclusive. Indeed, I'm not even sure they are always competitors! Both
Fileman and the relational model address data; on the other hand, HL7 is a
messaging protocol (though the protocol does have a lot to say about data,
and experience has shown that there is often something an impedance
mismatch between HL7 and files in VistA). Corbamed is (among other things)
middleware, and addresses yet another set of issues.
In my view, this discussion needs to continue, but we do need to start
moving towards a discussion of actual data models, too. Since we've been
discussing UML, it seems to make sense to me to start discussing use cases
and use th use cases as a basis for developing our object and/or data
model.
---
Gregory Woodhouse
[EMAIL PROTECTED] / http://www.wnetc.com/home.html
"An atheist staring from his attic window is often nearer to God than the
believer caught up in his own false image of God."
--Martin Buber
