[Stackless] Explicit yield/break to Preemptive Stackless Loop

[Jeff Senn]

On Nov 9, 2006, at 7:04 AM, Christian Tismer wrote:

> Richard Tew wrote:
>> What say you, Jeff and Christian?  There is no reason we shouldn't  
>> just
>> remove this slp_initial_tstate check, right?
>
> No, I'm missing some convincing consideration why this
> is right. We need a concept how multiple execution
> chains in multiple threads are supposed to play together,
> or if we better want to avoid this and allow for single,
> thread-based tasklets only, for instance.
>
> I have no clue, yet where this should go and need more input.

Well... The feature I am using is to run a scheduler in the non-main
(non-first) thread.  And I think that is very stable.  As for
running schedulers in many threads at the same time, I have not done  
much testing,
but it seems like we would want that to work (even though the
advantages are minimal due to the GIL).

When this came up before I did some testing about running the scheduler
in one thread and then starting up helper threads to feed channels
and this also seemed to work well.

So... I'm with Christian in not being *sure* that it will work, but
also I see no reason why it shouldn't in theory.  I guess I propose
we go ahead as if it should work, and start fixing "bugs" and see
how hard it is... but we will definitely need Christian's expertise
to help...

-Jas



_______________________________________________
Stackless mailing list
Stackless at stackless.com
http://www.stackless.com/mailman/listinfo/stackless