Hi, Team,
I am not a member of OMG; presumably someone on this team is. There’s a problem with the UML 2.5 specification that the Eclipse UML2 project is currently working around but which needs to be reported to OMG:
This is particularly relevant to validation of UML-RT models in Papyrus-RT. Currently, because of this problem, all state machines that redefine state machines inherited from parent capsules are flagged with validation warnings.
Could somebody with the requisite access at OMG please volunteer to raise this issue in their JIRA database? The problem, essentially, is that the StateMachine metaclass’s redefinition of the "isRedefinitionContextValid(redefinedElement : RedefinableElement) : Boolean" operation (UML 2.5 §14.5.10.6) is oddly over-constrained, requiring that the context classifier of a redefining state machine redefine the context classifier of the redefined state machine. It seems more plausible that this constraint should only require, or should also allow, that the context classifier of the redefining state machine specialize the context classifier of the redefined state machine. Otherwise, state machine redefinition can only be valid for state machines that are owned behaviours of classifiers that are nested in other classifiers (because the context classifiers are currently required to have their own valid redefinition contexts).
That is to say, an OCL formulation more like this, below, is what one would expect. In fact, the UML 2.2 specification had something very like this:
body:
redefinedElement.oclIsKindOf(StateMachine) and
let redefinedStateMachine : StateMachine = redefinedElement.oclAsType(StateMachine) in
self._'context'().allParents()->includes(redefinedStateMachine._'context'())
