[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Scheme-reports] Promises that force themselves
Okay, so my understanding is that:
> If no value has been computed for the promise,
> then a value is computed and returned. The value
> of the promise must be cached (or "memoized") so
> that if it is forced a second time, the previously
> computed value is returned.
> If no value has been cached for the promise,
> the delayed expression is invoked. Once that
> expression returns normally for the first time,
> its returned value must be cached (or
> "memoized") so that if it is forced afterwards,
> the value computed from the first return is
erk, the prose sucks, I suppose. I guess it needs to be improved.
So basically, this can be said to mean that in a multithreaded
environment, because the promise's value gets cached only once the
expression returns for the *first* time, the delayed expression can be
executed in two separate threads simultaneously (i.e. an
implementation that allows that is R7RS-compliant).
So if my understanding is correct, I think I'd like to propose a
*proper* purely-functional promise as a R7RS-large library with the
assurance that the delayed expression will only evaluate once, in one
thread, with the possible drawback that promises that force themselves
have unspecified behavior.
Anyway, what would the following yield?
(define x 0)
(if (= x 5)
(set! x (+ x 1))
(set! x (+ x 1))
(force p) => 10 (maybe?)
(force p) => ????
So: in case of a recursion, does it return the same value as the first
returning external (force p) - which yields 10 - or the same value as
the first returning internal (force p) - which (I think) should yield
You know, I don't really know of a good use in applications of
re-entrant promises. Laziness is hard enough with purity, having
mutation is a recipe for complexity.
I think that making re-entrant promises as unspecified behavior will
not actually affect *actual* applications/libraries, but I don't have
any actual data backing that idea. I might try to hack a G-machine
implementation onto an actual application of laziness in Scheme if I
have time, ha.
Scheme-reports mailing list