What I am wondering is whether or not Maarten thinks that the inefficient stack access problem is somehow related to multitasking. I'd say that it isn't -- that it's a problem whether you have multitasking or not.
when not multitasking, after the 'invention' of the call tree allocation of function parametres, I think we can put the 'stack inefficieny' issue to rest.
I agree that if you go beyond the simplest "task switching" i.e. the workloop, the stack DOES become an issue, but with no 'dynamic' task switching and call tree allocation, I see no "stack inefficiency".

Erik