Python 1.6 The balanced language

Tim Peters tim_one at email.msn.com
Wed Sep 13 06:27:35 CEST 2000


[discussion of coroutines in Icon, Jcon and JPython snipped]

[Neel Krishnaswami]
> If I understand things correctly (which I very well may not), most
> Java implementations have "green threads" by default -- ie, threads
> that don't map 1-for-1 to OS threads. So I don't see any fundamental
> reason why these threads should be any slower than, say, microthreads
> for Stackless.
>
> This would imply a different implementation strategy for J-Stackless
> vis-a-vis C-Stackless, but so what? Am I misunderstanding something
> basic?

If you missed anything, it was that the msg you replied to talked only about
the implementation of coroutines.  C-Stackless is much more powerful than
just *that*.  So that coroutines were implemented easily in Java by Jcon
doesn't mean all of what C-Stackless can do is implemented easily in Java.
Indeed, I'm not sure any port of Scheme to the JVM yet has managed to get
full continuations working, and *full* C-Stackless is no easier than that.
As for speed, since Java and Jcon actually exist, get out your stopwatch and
try it.

too-old-to-fall-for-*that*-one-anymore<wink>-ly y'rs  - tim






More information about the Python-list mailing list