[news.eclipse.modeling.mdt.uml2] Re: InstanceSpecification whose classif

["James Bruck" <jbruck@xxxxxxxxxx>]

Hi Axel,

I don't see a problem in creating an instance of an activity the way you 
describe below.   As you mention, it is legal from a UML perspective.
You may want to have a look at section 6.4 of the superstructure spec.  for 
a higher lever overview.

Cheers,
- James.


"Axel Reichwein" <reichwein@xxxxxxxxxxxxxxxxxxxx> wrote in message 
news:gaoenq$d6a$1@xxxxxxxxxxxxxxxxxxxx
> Hello,
>
> I am analysing different possibilities to map a Simulink model to a UML 
> activity model with appropriate lightweight extensions. I am using the 
> Eclipse UML2 Project to program the implementation of this Simulink-UML 
> interface.
>
> I am confronted with the problem of wanting to link actions with instances 
> of activities and not directly with activities. So instead of a 
> callBehaviorAction, a callBehaviorInstanceAction would be great :-) The 
> activities which are called from the actions share the same properties, so 
> it would be practical to have on one hand a library of activities and on 
> the other instanceSpecifications of activities, similar to classes and 
> instanceSpecifications of classes.  Surprisingly, I have not yet seen 
> examples of activity instances. Although, following the specification, 
> Activity (from BasicActivities) extends Class (from Kernel), so in theory, 
> instances could also have as classifier an activity. The Simulink concept 
> of a library of  blocks (in UML: activities) and instances of blocks in a 
> model (in UML: actions referring to activity instances) could then be 
> described in UML. As a current solution, I intend to add a stereotype to 
> the callBehaviorAction, which will refer to the instanceSpecification of 
> the activity.
>
> Any comments or other propositions are very welcome.
>
> Best regards,
> Axel