[Python-Dev] minimal stackless

Vladimir Marangozov Vladimir.Marangozov@inrialpes.fr
Sat, 12 Aug 2000 21:16:17 +0200 (CEST)


I'd like to clarify my position about the mythical Stackless issue.

I would be okay to evaluate a minimal stackless implementation of the
current VM, and eventually consider it for inclusion if it doesn't slow
down the interpreter (and if it does, I don't know yet how much would be
tolerable).

However, I would be willing to do this only if such implementation is
distilled from the call/cc stuff completely.

That is, a minimal stackless implementation which gives us an equivalent
VM as we have it today with the C stack. This is what I'd like to see
first in the stackless PEP too. No mixtures with continuations & co.

The call/cc issue is "application domain" for me -- it relies on top of
the minimal stackless and would come only as an exported interface to the
control flow management of the VM. Therefore, it must be completely
optional (both in terms of lazy decision on whether it should be included
someday).

So, if such distilled, minimal stackless implementation hits the
SourceForge shelves by the next week, I, at least, will give it a try
and will report impressions. By that time, it would also be nice to see a
clear summary of the frame management ideas in the 1st draft of the PEP.

If the proponents of Stackless are ready for the challenge, give it a go
(this seems to be a required first step in the right direction anyway).

I can't offer any immediate help though, given the list of Python-related
tasks I'd like to finish (as always, done in my spare minutes) and I'll
be almost, if not completely, unavailable the last week of August.

-- 
       Vladimir MARANGOZOV          | Vladimir.Marangozov@inrialpes.fr
http://sirac.inrialpes.fr/~marangoz | tel:(+33-4)76615277 fax:76615252