RE: Relationship between regions and transitions



RE: Relationship between regions and transitions

From: Waldin, Earl <Earl.Waldin_at_paranor.ch>
Date: Tue, 12 Dec 2006 10:08:46 +0100
Message-ID: <1DE41A26EA92CA4FB831A9449AB0841E5DC19F@srv-par2.paranor.ch>
 

-----Original Message-----
From: puml-list-request@cs.york.ac.uk
[mailto:puml-list-request@cs.york.ac.uk] On Behalf Of Alessandro Folli
Sent: Monday, 11 December, 2006 21:43
To: puml-list@cs.york.ac.uk
Subject: Re: Relationship between regions and transitions



Thank you very much for the detailed answer. 
 
You're welcome
 
I think it's a good idea to consider that a transition between two
states is owned by the "closest" region. 
 
It certainly seems reasonable. 
 
I have exported a State Machine Diagram from Visual Paradigm to XMI and
I have noticed that the XMI doesn't care about the relationship between
regions and transitions. Looking at the XMI specification I didn't find
anything referring to it.  
 
Be careful to distinguish between an export of the diagram and an export
of the model, even though they are both parts of the same file. The
diagram part is not necessarily one-to-one with the model. There may be
no need to talk about the ownership relation between regions and
transitions within the  diagram part. The diagram part contains
references to model elements in the model part. So it's realy only the
model part of an export that corresponds directly to the UML metamodel.
 
The XMI spec wouldn't say anything about this anyway. XMI just says how
to serialize a model given the specification of a metamodel written in
MOF. From the viewpoint of XMI, the UML metamodel written in MOF is only
a structural model. So from XMI's viewpoint, transitions need only be
owned by regions. It doesn't care which region owns which transition.
With no additional constraints in the UML specification on the ownership
of transitions, every toolbuilder is free to choose how he wants to do
it. So the XMI export from Visual Paradigm more likely just represents
how that tool handles this issue in its internal implementation.
Unfortunately, this could cause problems with model exchange. For
example, if one tool puts all the transitions in the topmost region and
another puts them in the least common ancestor region, then there could
be trouble exporting/importing models between these two tools. We ran
into a problem in this area.
 
Anyway, as you suggested, it's always possible to find the closest
region of a transition just searching the common ancestor of the two
states.
 
Thank you,
 
  Alessandro Folli
 
 
Received on Tue 12 Dec 2006 - 09:08:52 GMT