> This issue has cropped up recently around here with a proposed new CPR
> system. What I really want to know is what folks are thinking about,
> here are the scenarios based on what I will loosely call schemas, by
> which I mean the definition of records.
>
> GEHR schema ==> mapped into ==> Pre-existing DBMS schema
>
> or
>
> GEHR schema ==> mapped into ==> new DBMS schema to accomodate mapping.
>
> What I think I heard the GEHR folks asking for was help with the
> latter approach for various DBMS's.
Yes, that's more or less it.
> What I head Jim ask for was the former approach.
>
> Now, it may be that the two approaches are the same, but I was thinking
They're not - legacy integration can be serious hard work, and requires a
case-by-case analysis. There are lots of theoretical and empirical approaches of
course, but the problem is not the same as just using an RDBMS as a persistence
mechanism and having the freedom to define a schema from scratch.
> that when you did a mapping from the ocean kernel to a DBMS you would be
> rather unlikely to reach a 'schema' which matched anything that
> pre-existed. For that to take place, I imagine one would need to invoke
> some kind of active (i.e. rule based) transformation applied to the
> 'schema's' and the data.
Most likely, yes.
- thomas beale
--
..............................................................
Deep Thought Informatics Pty Ltd
Information and Knowledge Systems Engineering
phone: +61 7 5439 9405
mailto:[EMAIL PROTECTED]
http://www.gehr.org
http://www.elj.com/eiffel/ebs
..............................................................
