[Modeling] Comments on interaction diagrams

Marc-Philippe Huget M.P.Huget@csc.liv.ac.uk
Wed, 12 Mar 2003 14:12:52 +0000

Hi all,

James Odell wrote:

> Yes, this is a chicken-and-the-egg type of problem.  Doing a metamodel
> without a diagram is problematic; developing a diagram without a metamodel
> make things impractical.  So, we have started with the diagram as a possible
> way to express agent interactions.  Once we find that it provides a good way
> of communicating interactions, we then need to define the metamodel.  If we
> start too early with the metamodel and want to make changes to the diagram,
> then we could be creating double work.  However, we can certainly start on
> the basic foundations of Interaction diagrams now.  The UML 2.0 metamodel is
> 99.99% completed.  So, we can start by reusing some of that to see if it
> holds up under the agent approach.  If not, we will have a *lot* of work to
> do.  But, in any case, --IMO -- we should make sure the Interaction Diagram
> now reflects what we want to express, before spending too much time on the
> metamodel.
> Anybody else have similar or differing opinions? Perhaps Paola and Hong
> could start by moving over the AUML 2.0 metamodel to see how well it works?

I totally agree with Lin and Jim, if we begin to think hard about the metamodel,
we will have headache and we won't be sure that the metamodel fits what we want,
and only diagrams will give us the result, let's continue to work on interaction
diagrams, when the diagram will be stable, we will move to metamodel. Well,
about examples, I agree that we need that, I will add several in the draft, in a
near future I guess.


Marc-Philippe Huget

Agent Applications, Research and Technology Group
Department of Computer Science
University of Liverpool
Chadwick Building, Peach Street
L69 7ZF Liverpool
United Kingdom

email: mph@csc.liv.ac.uk