[Modeling] Modeling an Agent Class- composition

Joaquin Peña joaquinp@us.es
Mon, 30 Jun 2003 11:47:31 +0200


Hello:

> > Thus, the description of an open agent systems, where
> > agents continuously change their role to interact with 
> others, cannot 
> > be modelled with a class-centered approach because we cannot relate 
> > them with static associations.
> > 
> > So, if you use a class approach you the option could be: 
> define a new 
> > association for dynamic interactions.
> 
> What do you mean by "static"/"dynamic" associations?
> I don't think this is a term defined in UML.
> 
> Obviously, an association's instance set (of links)
> can change at run-time (if it's not frozen by some
> constraint). So, in this sense, any association is
> dynamic.
 
I don't agree with this. But this just my opinion.

The links we establish in a class diagram are focused on the data
relations. These relations are commonly seen as static, and
commercial OO languages does not support naturally this kind of
run-time variability. That it is to say, if you code the
association in your programming language in a class, how are you
going to change them? Rewriting the class?

If we want them to change at run-time we have to use design
patterns in the model and in the implementation. Thus, in the
design pattern these association are static. We make them dynamic
decoupling one class from another.

Best Regards,

Joaquín.


> -Gerd
> 
> _______________________________________________
> Modeling mailing list
> Modeling@www.fipa.org
> http://fipa.org/mailman/listinfo/modeling
>