[Stackless] Advice about September Stackless Python Talk

[Andrew Francis]

Hi Folks:

I will be giving a Stackless Python to the Montreal Python User's Group in mid-September. The talk won't be the same as the one I gave at Pycon 2008. For starters, I want to demo stuff. The talk should be 45 minutes so I have more time to develop ideas. I am starting to design the talk. The audience will range from beginners to Python interpreter implementors. 

I am trying to decide on what themes to stress. I am thinking that Stackless Python's cooperative scheduling model and channels, leads to a clutterless, Pythonic style of concurrent programming. The main reason to programme with Stackless Python is used because it is easy, not necessarily for performance.

Besides describing the features of Stackless Python and why Stackless Python matters, I was thinking of adding the following sections.

1) A why is Stackless Python "Stackless" (A Stackless Python under the hood section. Maybe a comparison of Stackless tasklets and generators may be mentioned. [This I have to do more research]

2) The future of Stackless: PyPy (get into tasklets, coroutines, greenlets) [I can get a chance to play more with PyPy]

3) Pickling.

What I would take out of the Pycon 2008 talk:

1) Discussion of my WS-BPEL project - that just confuses people.
2) Nitty-gritty details of a networking solution.

Since networking is a big topic, and there are different specific approaches I want to find what is common. I will argue:

due to the blocking problem - Reactors and the Reactor pattern
is naturally used

due to channels, most solutions use: Active Object. In turn, Active Object uses many other design patterns (proxy).

So in the process, I can explain techniques from integration with Twisted to drop-in socket replacements.

That said, any suggestions?

Cheers,
Andrew