In general, I would think the coroutine mechanism being implemented by Lukas Stadler for the MLVM version of the hotspot JVM might be a good option; you can directly control the scheduling, although I don't think you change the mapping from one hardware thread to another. (That's probably not interesting.)

There are good results with JRuby, it would be nice to replicate with Jython - and it should be really straightforward to do that. See http://classparser.blogspot.com/

On Tue, Jul 27, 2010 at 10:05 AM, holger krekel <holger@merlinux.eu> wrote:

On Tue, Jul 27, 2010 at 17:07 +0200, Maciej Fijalkowski wrote:
> On Tue, Jul 27, 2010 at 4:36 PM, Paolo Giarrusso <p.giarrusso@gmail.com> wrote:
> > Hi all!
> >
> > I am possibly interested in doing work on this, even if not in the
> > immediate future.
>
> Well, talk is cheap. Would be great to see some work done of course.

Well, I think it can be useful to state intentions and interest. At least
for my projects i feel a difference if people express interest (even through
negative feedback or broken code) or if they are indifferent,
not saying or doing anything.

best,
holger

_______________________________________________
pypy-dev@codespeak.net
http://codespeak.net/mailman/listinfo/pypy-dev