Eclipse Community Forums: Papyrus for Real Time » local transitions using/not using entry points

Help

Home

Home » Eclipse Projects » Papyrus for Real Time » local transitions using/not using entry points

Show: Today's Messages :: Show Polls :: Message Navigator

local transitions using/not using entry points [message #1776573]

Fri, 17 November 2017 19:55

Juergen Dingel

Messages: 42
Registered: January 2015

Member

Hi,

I have a question about local transitions, i.e., transitions that
go from the boundary of a composite state to a substate inside
that composite state.

I thought it did not matter if such transition emanated from
an entry point (assuming no transition goes into this entry point)
or not, i.e., whether or not in the image below the transition labeled
'ping' started from the entry point or the boundary of the composite
state.

index.php/fa/31308/0/

However, in the attached model the 'ping' transition in state 'Playing' in
Ponger is not taken if connected to the (other unconnected) entry point.

Do local transitions need to start from an entry point?

Thanks

Juergen

Attachment: localTransitions.png
(Size: 5.93KB, Downloaded 442 times)
Attachment: LocalTransitionTest.zip
(Size: 7.84KB, Downloaded 132 times)

Report message to a moderator

Re: local transitions using/not using entry points [message #1776575 is a reply to message #1776573]

Fri, 17 November 2017 19:59

Juergen Dingel

Messages: 42
Registered: January 2015

Member

Sorry, my last sentence should have read "Do local transitions need to start from the boundary of the composite state?"

Juergen

Report message to a moderator

Re: local transitions using/not using entry points [message #1776581 is a reply to message #1776573]

Fri, 17 November 2017 20:57

Ernesto Posse

Messages: 438
Registered: March 2011

Senior Member

Right. If there are no transitions whose target is the entry point, it shouldn't matter. Maybe it's a but. I'll investigate.

Report message to a moderator

Re: local transitions using/not using entry points [message #1776584 is a reply to message #1776573]

Fri, 17 November 2017 21:38

Ernesto Posse

Messages: 438
Registered: March 2011

Senior Member

I can confirm that if the transition goes from the entry point to the substate, it is not taken, but actually, the problem is kind of subtle, and I'm not sure it's a bug. If instead of using an entry point you use an exit point, it works as expected.

The idea is that Ponger is in state Playing::State2 when it receives ping, and since this state doesn't have any outgoing transitions whose trigger matches the incoming message, its containing state (Playing) is considered next. But only the "exiting" transitions of Playing are considered, and "exiting" transitions are those who leave the state or which start in one of the state's exiting points. This is because when you are taking a transition you are "leaving" the source state. In this case, you are not really leaving, but only the state's outgoing transitions are considered.

The code generator uses the following terminology (which is not "official" UML-RT terminology, but I find it useful anyway):

An outgoing transition is either a direct outogoing transition or an indirect outgoing transition. A direct outgoing transition of a state S is a transition whose source is S itself. An indirect outgoing transition of S is a transition whose source is an exit point of S.

Similarly for incoming transitions and entry points.

So in your example, when you connect the transition to the entry point, it is considered an incoming transition, not an outgoing transition, and therefore it would not become enabled when the message arrives.

Report message to a moderator