The way (Rule) units are scheduled into execution is ad-hoc for Rule Units. Assuming we want to generalize the "Unit" flow of execution to a global API, having e.g.

• rule units (Drools)
• process units (jBPM)
• etc.

then the mechanism must allow for extensions.

This tasks consists in implementing a more general "Scheduler" interface, to which units can be submitted. The scheduler decides which unit has to be run next, depending on which Units are "in scope" and the actions they perform (e.g., submitting new units for run()'ing, registering a guard, etc.)