IMHO: I would like us to back port all python 3.x features we can into
stackless 2.8 and keep one version to rule them all. This split
version path I think was a bad choice in hindsight.  TBH I have never
really understood what the differences were between 2.x and 3.x and
why they are not compatible. (so i could be blowing smoke based on
faulty info)

Having two streams is confusing to new programmers and it seems like
there is a bunch of overhead to support two streams that could be
better spent on adding new features.

I would like to see Stackless forge ahead into support for micro
threads running on multiple cores so that a stackless programs become
the gold standard for python on multi core machines. possibly a way to
have tasklets flagged as thread safe (no global var access) and only
communicate via cross process channels.

Then Stackles could manage the additional hardware threads as a worker
pool and schedule tasklets to those threads as needed. (BUT-TON of
pain in the details I would think)

The Idea of Copy on channel communication between processes sounds
like a good first step.

Or make the jump and say that stackless will be 3x based from now on
and no more support for 2x will be done. This option i think would
alienate a lot of developers though.

Just stop with the two streams of python/stackless.

- rob

On 1/7/14, Martijn Faassen <[email protected]> wrote:
> Hi there,
>
> On 01/07/2014 04:33 PM, Christian Tismer wrote:
>
>> Yes, that is one possibility, when stackless is only considered as a
>> work-around.
>
> I think that should be a realistic consideration though. It might just
> happen that people come to Stackless only interested in the 2.8 features
> and not in the Stackless features, and it's better to be prepared.
>
>> On the other hand, we might also think of continuing the stackless series
>> on the version 3 line.
>> So when people upgrade from stackless 2.8 to stackless 3.4, the
>> additional
>> futures would be clearly supported ;-)
>
> Sure, but I think it'd be better to aim for maximum compatibility for
> the start. Adding things to __future__ would make it impossible for me
> to write polyglot code that runs on Stackless 2.8 and Python 3.x, and I
> think that's a worthwhile goal.
>
> Regards,
>
> Martijn
>
>
>
> _______________________________________________
> Stackless mailing list
> [email protected]
> http://www.stackless.com/mailman/listinfo/stackless
>


-- 
Life: Bah, I will worry about it when it is over.

_______________________________________________
Stackless mailing list
[email protected]
http://www.stackless.com/mailman/listinfo/stackless

Reply via email to