Talk:Active object

I think this task is flawed: as my note at the bottom of the E example shows, its result can be achieved in a simpler and more efficient way by not using any continuing activity at all.

I propose that the task be modified to either:

• discuss only dependence on time, and not "a task that updates the object's state". This would also allow e.g. functional-reactive systems to meet the criteria, which I think would be good.
• be something which, unlike the current one, actually requires such an active process. One way that comes to mind to do this would be to make the integrated input be requested by the integrator each timeslice (polling), rather than put in by an external event.

After having written them, I like both of these options; but it seems unreasonable to have two slightly different "integrator" tasks. Opinions? --Kevin Reid 01:40, 4 November 2008 (UTC)