"Stackless" XEmacs?

[David Kastrup]

Michael Sperber <sperber at deinprogramm.de> writes:

> Their web material isn't very clear, but I'm assuming that's because
> they don't use a stack representation for continuations (i.e. what
> most people just call "the stack" in traditional implementations).  In
> particular, heap-based and partially heap-based implementations of
> continuations have much lower space and time costs for things like
> threads.
>
> Will Clinger's paper contains a good survey:
>
> http://www.springerlink.com/content/h5808n962434j275/fulltext.pdf
>
> (Most Scheme implementations, as they have to support
> `call-with-current-continuation', also use an implementation strategy
> for continuations other than pure stacks.)

There is one Scheme implementation (I always forget its name, it is some
simple thing or animal) which does continuations by calling using the C
stack, but never returning, and occasionally garbage-collecting the
entire stack.  This is what I call programmer chutzpah.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum