BPMN2 — Reuse of processes in orchestration and choreography

Modelers should be able to define processes independently of whether
they will be reused in processes or choreographies. Currently the
information coming in or out of a process is either through data flow or
through messages, I understand it. The modeler must commit to one of
these in advance, or define two copies of the process, one with
messages, the other with data flow. Suzette brought this up separately
and says a solution was agreed that Mariano will write up. Didn't see
any issue from her on this, sorry if this is a duplicate.

In section 10.3.2 Executions Semantics for Data, add the following text at the end of the section:
—
In the case of Throw and Catch Events, given their nature, the execution semantics for data is different.
When a ThrowEvent is activated, all DataInputAssociations of the event are executed, filling the DataInputs of the Event. Finally DataInputs are then copied to the
elements thrown by the event (Messages, Signal, etc). Since there are no InputSets defined for Events, the execution will never wait.
When a CatchEvent is activated, DataOutputs of the event are filled with the element that triggered the event. Then all DataOutputAssociations of the event are
executed. There are no OutputSets defined for Events.
To allow invoking a Process from both a CallActivity and via MessageFlow, the StartEvent and EndEvent support an additional case.
In the case of a StartEvent, the DataInputs of the enclosing process are available as targets to the DataOutputAssociations of the Event. This way the Process
DataInputs can be filled using the elements that triggered the StartEvent.
In the case of a EndEvent, the DataOutputs of the enclosing process are available as sources to the DataInputAssociations of the Event. This way the resulting elements
of the EndEvent can use the Process DataOutputs as sources.
Disposition: Resolved