[Python-Dev] Re: Re: Stackless Python

Gordon McMillan gmcm at hypernet.com
Wed Jun 2 14:22:41 EDT 2004


Terry Reedy wrote:

[trying Stackless]
> Among other things, I looked to see if there were currently a free trial
> period for Eve, as there sometines is for online games.  If I were to try
> it, and it ran smoothly for some time without lags and crashes (which I
> have experience of), then that would suggest that the tools (including
> Stackless) and use thereof were adequate to the task.  If not, that
> something (though not necessarily Stackless) would be shown lacking.

Let me pop my head up for a second.

You won't see Stackless do anything (from a user's perspective) that can't be 
done without Stackless[1]. Stackless (like generators) just makes some things 
much easier.

Basically, any state machine(s) with complex internal state (and relatively 
simple events) will be much more straightforward to write in Stackless. Put 
another way (if you don't think in terms of "state machines"), if you've got 
callbacks onSomething(...) where a lot of the code is concerned with where 
you are in a complex interchange, it will be much easier in Stackless.

I can't speak to the new Stackless, but in 2000 I did a proprietary, 
commercial app for Avid that used Stackless (and lots of C extensions). It 
went through Avid's QA, so it would not have been released if it crashed.
The Mac OS8/9 port was difficult, but not because of Stackless.

-- Gordon

[1] Though perhaps you can do more / quicker, because of more efficient memory 
usage than an implementation using threads.



More information about the Python-Dev mailing list