[gme-users] RE: [udm-users] BON2 vs. UDM

Miklos Maroti mmaroti at isis.vanderbilt.edu
Tue Apr 13 18:55:16 CDT 2004


Dear All,

First, I did not wished to participate in this emotional discussion,
but then I realized that this is much more fun than writing proposals.
I worked on both GME/BON and UDM extensively, although I do/try not to 
follow the code and API changes very closely. But here are my thoughts:

UDM is more generic than GME/BON, and GME/BON stores and provides
more information than UDM; it would be a bad design if the two were 
more closely tied to each other. If UDM and BON were united, there
would be a lot of pressure to include existing features of GME/BON
to UDM, such as access to registry entries, location, etc. If UDM 
is really a viable tool-integration framework than these differences 
should not be pushed down to UDM but be expressed by modeling 
transformations (e.g. GReAT) and made the transformations automatic.
Dependency and compatibility on UDM would also tie down the evolution 
of GME.

Yes, it is very unfortunate that BON2 is so hard to use (because of
reference counting). And yes, the two programming frameworks are
very similar, although in the details quite different. We need an
easy to use framework to exploit the full potential of GME. This was
traditionally the C++ BON. Consider writing your next interpreter in
Python or Java! :)

Miklos


More information about the gme-users mailing list