Re: Delimited continuations

[Christopher Lemmer Webber <cwebber@dustycloud.org>]

John Wiegley writes:

>>>>>> "MH" == Michael Heerdegen <michael_heerdegen@web.de> writes:
>
> HM> Or we just don't call it continuation, which is misleading given the
> MH> suggested implementation which gives you more something like a delimited
> MH> re-execution.

The term "prompts" are often also used for delimited continuations in
Scheme-land at least: https://www.gnu.org/software/guile/manual/html_node/Prompts.html

(But they are continuations... they're delimited rather than
undelimited... that's a big advantage though because delimited
continuations compose)

> HM> Anyway, my personal opinion is to make such stuff available so that people
> HM> can experiment with it.
>
> I'll create some examples in Haskell to determine what the intended semantics
> should be, and then see how to efficiently replicate that behavior in Elisp.

May be even easier to show how delimited continuations work in schemes
like Guile or Racket, since that looks a bit more native.  In Scheme
examples you also don't need to manually do the CPS conversion... it's
done for you.

But I guess that leads to something I was unclear on: John, are you
suggesting that we make delimited continuations part of the language
(with automatic CPS conversion)?  If that's possible, it would be a big
win.  But maybe you were just demo'ing how to do delimited continuations
manually?

IMO manual CPS is cool to know how to do, but it's much nicer if your
language can do it for you :)

Go go delimited continuations!
 - Chris