wwilson wrote:
> Gunther Schadow wrote:
> >
> > How married are you to Eiffel? It seems like a lot. I don't
>
> If the kernel supports either java api interfaces (which I believe they
> have committed to) or OMG IDL, it doesn't really matter what the
> implementation language is, except to those who would wish to change or
> modify the workings of the kernel. The OpenEMR never need see Eiffel.
> Since the kernel itself becomes opensource then a powerful force is
> unleased to keep it suitable to the tasks at hand.
"Horses for courses" as they say. This is our thinking.
> However, the part of your comment about impedance mismatch is quite
> troubling. Have you discovered that object models are critically tied
> to a particular OO language?
Put another way, the expressibility of object models (e.g. expressed in
UML) in some languages (Java, Delphi, VB etc) is limited. Starting from UML
and expressing in Eiffel or C++ is no problem. But starting in Eiffel and
building software from the model is the ultimate!
The more important impedance mismatches are between the model and derived
expressions such as IDL, ODL (for databases) and XML. The limitations are
due to e.g. the inconsistent type system, missing generic types, and lack
of assertions in the target formalisms. But in reality, there are
reasonable workarounds, and the final effect is not too bad. I have mapped
Eiffel models to ODMG-93 ODL with success in the past, and expect to have
no problems with CORBA IDL (in fact a binding has already been done by ICL
(now it's the Peerlogic ORB)) or MS COM (despite the large mistmatch). One
of the reasons the mismatch is not too dangerous is that you are mapping
APIs, not clore classes, and the API needs to be relatively flat and
simple, and funcrionally oriented.
- thomas beale
--
---------------------------------------------
Deep Thought Informatics Pty Ltd
Information and Knowledge Systems Engineering
mailto:[EMAIL PROTECTED]
http://www.gehr.org
phone: +61 7 5439 9405
---------------------------------------------
