You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Enhance SOM reachability analysis to handle distributed systems #2945
The current SOM reachability analysis assumes a non-distributed system. In this context, non-distributed means SOM transitions happen as soon as a trigger event is emitted by a component. The management of SOMs and their transitions can be conceptualized as implemented by a single mode manager component. This mode manager receives the trigger events and (a) activates and deactivates components and (b) updates component modes according to the modes and transitions in the instance model. In this situation, mode transition trigger events do not travel over buses but are processed immediately on creation. The connections that transport trigger events in the model are only used to indicate which transitions are executed in response to a trigger event.
In contrast, a distributed system needs multiple independent mode managers with each being responsible for the modes of a subset of the system's components (mode domain). Each modal component must be assigned to exactly one mode manager. In this situation, trigger events travel between individual mode domains over buses, which can introduce - in the worst case - arbitrary delays and reordering of trigger events.
Activity