[Modeling] Modeling an Agent Class- composition

Dr. Hong Zhu hzhu@brookes.ac.uk
Thu, 26 Jun 2003 13:00:42 +0100

Hi, Gerd, and All,

It is so good that you agreed with me that there is a new property of the
part-whole relationship that we should look at.

> > Shared or not shared is about whether one entity can be a part of
> > many others entities. Lifetime is about whether the part will be
> > killed when the whole is killed. Am I right?
> Yes, you are.
> > If yes, then, none of the above features tells that if a part
> > quit from the whole or the whole is destroyed (it is possible
> > in dynamic classification), what happens to its roles played
> > in the whole.
> Yes, this question refers to additional properties of the
> aggregation relationship.

Thank you. This property has not been recognised before. I believe it is
important for AUML because I found it is common in role modelling.

> An aggregation, and also a compositiom,
> can have lifetime dependency (= inseparability of parts) or not.
> What is your problem? Would you like to be able to specify
> this lifetime dependency in a model? Then you can use a Boolean-
> valued tag "inseparable=true" for the part association end (as
> proposed in the paper mentioned below).
> -Gerd

The problem is that existing part-whole relations in UML is inadequate. What
I want is a more clear definition of part-whole relation in AUML and take
the new property into consideration. Explicitly defining such properties can
be a way to solve the promblem.