Subactivities

[molla]

When I create a sub-activity, it creates an embedded activity diagram for the sub-activity, so that when I double-click on the sub-activity, it jumps to the relevant diagram.

This is fine, especially to describe a hierarchical structure of processes and/or activities that all have their own diagrams.

The problem occurs when you want to use these in a diagram. If you use the "simple link" to refer tothe process/activity, then this double-clicking will be active. But, if you insert some links between these elements in that diagram, then the activities/processes will permanently get that link. (e.g. You use a "flow" between two activities in a diagram) Namely, if I use two activities - that are linked in a diagram - in a new diagram, the link will automatically appear.

If I want to link activities/processes differently according to which diagram I use them in,  then I have to use instances - namely "Actions" as it is now possible in Version 4.50. However, double-clicking on Actions will not make you jump to the relevant diagram for the activity from which the Action is derived. I am not sure if there is any guidance in UML 2.0 documentation for this, or maybe this is an EA topic only.

[KP]

Double-clicking on a composite element in EA will take you to its child diagram. Actions can't be composite elements because they are atomic elements (UML 2.0 describes them as "the fundamental unit of behavior specification").

I don't know how this affects your problem. You said

if I use two activities - that are linked in a diagram - in a new diagram, the link will automatically appear.

You can right-click on the connector and select  "Set Visibility | Hide Connector". It's reasonable for connectors to have different visibility in different diagrams. Does this help?

[molla]

The following is a simpler description of my problem:

I am using EA to build a Business Process Model. My process descriptions consist of Activity diagrams where I show the activities that make up the process and the flow between them. Since activities can be "re-used" in other process descriptions - they are commonly used - if I use flow links between the activities, these appear in all such diagrams that these activities take place in. However, the flows between these activities can be different in different process diagrams. To solve this problem, I used an instance of an activity (which is now an Action, for some reason). I want to show that different instances of a standard activity can show different relations (e.g. flow) when they are parts of different processes. (Did I say I was going to give a "simple explanation? )

[thomaskilian]

You have probably reached the limits of UML(and EA) in respect of BPM. The only solution I see is to duplicate the activity in question and use it according to what context it is needed. A trace between these "instances" could assist. Not really helpful  :-/

[Kevin Brennan]

Actually, his problem is in EA, not UML. I've run into this myself.

Basically, if I understand correctly, molla wants to use the same elements on different diagrams and draw a different set of connections between them on each diagram. EA doesn't really support this, unless you do what he did--create instances if the items and link the instances, not the originals.

I don't think UML requires this, although to be fair the way EA does it is, I think, more in keeping with the intent of OO modeling.

[molla]

Yes, Kevin has described it the way I encountered the problem. I really need to look at the UML 2.0 definition to understand the limits a bit better.

[Eve]

It is possible to show connections between elements in one (or more) diagrams and not in others.

If you right click on a connector (conection) there is a sub-menu "Set Visibility" that allows you to either hide the connector on the current diagram or on others that it appears in.  (If you hide it on others a manual diagram refresh appears to be needed)

Simon