[Modeling] Modeling an Agent Class- register your opinion

James Odell email@jamesodell.com
Mon, 23 Jun 2003 19:10:34 -0400

On 6/23/03 6:11 PM, "Dr. Hong Zhu" indited:

> Here are some rough ideas in the form what should stay in the core.
> (1) Classes (of objects) are to be static classifications of objects;
> (2) Classes are to be single classifications of objects;
> (3) Objects are all passive.
> Are these acceptable? Please also add to this list for what you think a core
> OO model should have.

Interesting proposition.  UML 2.0 already has active objects and dynamic
classification.  So, I'm not sure that pulling these things out would enable
us to reach "core UML."   These things are part of the core of UML 2.0.
Furthermore, they are the kinds of thing we need for agents.  So, why throw
them out?

Perhaps -- as Stephen Cranefield points out -- "we should start developing a
metamodel in which the agent type metaclass *is* a new subclass of
Classifier.  If we later discover that it really can fit under Class then we
can refactor the metamodel. "  This would get us away from the OO bashing:
it get on to the task of figuring out what we do want, instead of what we
don't want.  IMO, this is a reasonable strategy at this point.

Anyone else besides Stephen, Paola, Giovanni , and I?  Or, to be balanced:
anybody with an alternate proposal for moving ahead for now?