Re: [Gauche-devel] Implementation of Futures

[Shiro K. <shi...@gm...>]

Hi Glenn,
Your feeling is correct.  'future' spawns a new pthread to compute the
expression, and "future-get" calls thread-join! to get the result.
If you create a bunch of futures, they all run in parallel. If you
print something during the computation, the order you'll see the output is
indeterministic---the output happens _before_ future-get anyway.
If you make check-trees-of-depth return a value, and collect teh values of
future-get in a list, you'll see the list is always consistent.
Does this answer your question?

--shiro





On Wed, Feb 8, 2023 at 8:44 PM Glenn Takanishi <gl...@ne...>
wrote:

> 2023-2-8
>
> Hello!
>
> Just curious about how futures is implemented.  A short description would
> suffice with respect to some code I was playing with (it's attached to this
> email just for fun).
>
> The Gauche documentation on futures says that "future" calls code in a
> separate thread (I assume a Posix pthread).  So I get the feeling that a
> "future" is like "pthread_create" in C.  And that "future-get" can be
> likened to "pthread_join".  The structure of the programs between Scheme
> and C look alike.
>
> But the output returns after "future-get" is random.  The output result
> from pthread_join is usally not.  This is not a problem for me.  I'm just
> curious as to why this happens.  The behavior using Guile scheme is similar
> with random output results from "touch".
>
> This question might seem weird to you, and I apologize if so.  I'm just
> curious.  Thanks.
>
> Glenn
>
>
> _______________________________________________
> Gauche-devel mailing list
> Gau...@li...
> https://lists.sourceforge.net/lists/listinfo/gauche-devel
>