10.5.4. Task Priorities

Whenever an activity of type A1 ends and there is still another A1 task in the queue, the activity's resources would be re-used for the next A1 task unless there is another task (say, of type A2) with higher priority waiting for one of the resources (r1). In that case, r1 is allocated to that task, and all other resources are allocated to the next A1 task, which still has to wait for r1 becoming available again.

For example, in a pizza service, where incoming orders can be taken by pizza makers, and TakeOrder tasks have a higher priority than MakePizza tasks, when a MakePizza activity performed by a pizza maker in an oven ends while there is still another MakePizza task in the queue and there is also a TakeOrder task in the queue, the pizza maker is allocated to the TakeOrder task and the oven is allocated to the next MakePizza task, which has to wait for a pizza maker becoming available.

Algorithmically: Whenever an activity a1 of type A1 ends, collect all pairs <r, t> such that r is a resource used by a1 and t is the next task of an activity type A from the dependent activity types DAT of pool(r) with the highest priority among all DAT with task priorities higher than A1. For all these pairs <r, t>, allocate r to t. Allocate the remaining resources of a1 to the next A1 task, if there is any, otherwise release them.