Right now each task is kept as Pending even after completion so the executor can keep track of it, otherwise the task can not be restarted via alfad-ctl since the Executor will no longer poll the task.
If it can be assured that a task is no longer running (e.g. by tying a return value of Ready) to an update in the corresponding context, a new task could then be respawned when the "start" or "restart" action is performed over ctl.
This should not be implemeted if there is any way that a new task spawned via ctl corresponds to the same underlying task as one that is already running.
Note: it is not yet clear if removing the tasks from the executor would convey any performance benefit (might also depend on the executor implementation)
Right now each task is kept as
Pendingeven after completion so the executor can keep track of it, otherwise the task can not be restarted viaalfad-ctlsince the Executor will no longer poll the task.If it can be assured that a task is no longer running (e.g. by tying a return value of
Ready) to an update in the corresponding context, a new task could then be respawned when the "start" or "restart" action is performed over ctl.This should not be implemeted if there is any way that a new task spawned via ctl corresponds to the same underlying task as one that is already running.
Note: it is not yet clear if removing the tasks from the executor would convey any performance benefit (might also depend on the executor implementation)