Re: Teaching Concurrency

[macrakis@marat.osf.fr (Stavros Macrakis)]

In article <650@ajpo.sei.cmu.edu>, griest@ajpo.sei.cmu.edu (Tom Griest) writes:
> The important thing to consider is that tasks should be used
> for operations which are really likely to be physically concurrent.

I strongly disagree.  Tasking is a structuring mechanism as much as it
is a concurrency mechanism.  There are applications of tasks where
concurrency is a happy accident.  The standard name for this is "coroutines".

Applications for coroutines include simulation and data structure
transformation (cf. in particular the Jackson method).  Modules in producer-
consumer relations (cf. Unix pipes) are a good candidate -- of course
in this case there is often the possibility of concurrency.

For many years, assembly language programmers were the only ones to
benefit from coroutine structures because no mainstream programming
languages supported them.  Ada's support for coroutines is an important
step forward.

Of course, in order to benefit fully from this structuring mechanism, the
implementation should be efficient.  As far as I know, there are no Ada
compilers today where coroutine call is comparable to subroutine call in
its cost (as it can be in principle).

	-s