This is the mail archive of the
guile@cygnus.com
mailing list for the Guile project.
Re: Is guile byte-code compiled?
Mikael Djurfeldt <mdj@nada.kth.se> writes:
> forcer <forcer@mindless.com> writes:
>
> > I *think* the problem is call-with-current-continuation. I gather
> > it does stack-copying in guile, while in Scheme48 they might have
> > a saner implementation (probably since they don't have to handle
> > C code), but i'm not too sure here.
>
> Well, we've found the dog. It isn't very fair to compare interpreter
> performance using call/cc as the test. :)
Yep. Especially considering guile's C-boundness, while Scheme48
is only required to handle Scheme.
I was just doing some tests with my implementation of amb, and
was annoyed about how slow guile is there. But i guess there's
not much to do :]
> You're right that Guile is forced to do stack-copying (although it
> might actually be possible to do less copying if we easily could
> determine where the current continuation and the one to call "split").
hmm..
Can't that information be added to the continuation at call/cc-time?
-forcer
--
((email . "forcer@mindless.com") (www . "http://forcix.cx/")
(irc . "forcer@#StarWars (IRCnet)") (gpg . "/other/forcer.gpg"))