[Stackless] General "stackless changes to the python app" question

[Michael Brian Bentley]

I've skimmed the various sources of information for stackless. I 
generally understand what stackless provides, but I'm not clear 
enough on what changes to the python source are made to convert 
python into stackless python. Is there a brief description, something 
other than a diff, that describes the changes? Either way I'll 
probably do a diff and start reading the relevant inline comments -- 
if any.

I was reading the blog discussion about GIL this morning, while 
reading the python reference manual for 2.5.1 (full disclosure: note 
that I have yet to write a line of python code), and I decided that I 
didn't know if stackless improves on the single task part of a single 
task multiple thread model.

I have a project I would like to tackle that depends on being able to 
run many threads over time. (Likely well short of an EVE's worth of 
threads, since I'm bound to normal laptop and desktop hardware.) I 
imagine that some of it could be handled by running as many 
interpreters as there are processors. I guess I could also run as 
many interpreters as there are desired tasks.

I'm looking forward to doing the experiment in stackless.

Thanks very much,

-Mike Bentley