Am Sonntag, den 05.02.2006, 14:09 -0500 schrieb Stephen Brackin:
> Xavier Leroy showed me how to set things up so that an OCaml
> computation is interrupted after a pre-set time limit to ask the user
> how long to continue. Now Iÿd like to do the same thing for two
> parallel computations that Iÿd like to have share the time available
> more-or-less evenly. (Iÿm writing OCaml code for automatically proving
> results in the HOL Light theorem prover, and I have a situation where
> the code needs to prove that a value is positive or prove that itÿs
> negative, but the code doesnÿt know which, if either, it will be able
> to prove.) Can this be done? If so, then whatÿs the easiest way to do
> it? As a special case of this question, for an expression of the form
> ´let x,y = f(args1),g(args2)¡, how does the system allocate time
> between the computation of x and y? (If itÿs more-or-less equally,
> then that solves my problem; if one of x or y is computed first, then
> the other, then Iÿve still got a problem.)

Yes, you have. O'Caml does not parallelize computations, it does just
one thing after the other (here in an unspecified order).

If your computations do not mutate any global value (i.e. are really
side-effect free), you can spawn two threads and let the operating
system schedule how the CPU is shared by the computations. However: I
said REALLY side-effect free, and that has to be valid not only for your
own code but also for all functions you call (e.g. consider functions
with hidden caches that pretend they are pure but actually aren't).

Otherwise I do not see any practical solution for your problem. 

Gerd

-- 
------------------------------------------------------------
Gerd Stolpmann * Viktoriastr. 45 * 64293 Darmstadt * Germany 
gerd@gerd-stolpmann.de          http://www.gerd-stolpmann.de
Telefon: 06151/153855                  Telefax: 06151/997714
------------------------------------------------------------