I find it a little confusing...
What if it just showed the states with input and output links? No transition events. So at runtime you'd still see the active state highlighted in the stack of states.
To see the transition events you have to open the compound.
The other thing I'm not sure about with this kind of grouping:
Don't you have to leave space in the graph for the expanded compound? So do you end up with widely spaced compound nodes? How does it work in ICE?
Thanks for the mockup!
I don't think you have to see every states and transitions on the compound. That way it'll become very difficult, specially if inside the compound complex branching is going on.
Inputs and Outputs should be the way forward. Inputs can
optionally reflect the name of the states it's going into and outputs can
optionally relfect the events/transitions it's getting out from.
You can go into the compound, make edits and come back out.
To explode a compound in ICE, we have to make space. I'll try to illustrate with some screenshots.