[Stackless] Priorities and such like stuff...

Mike Fletcher mfletch at tpresence.com
Mon Oct 9 05:05:59 CEST 2000

Stackless Priorities:
	Figuring out if these weird errors have anything to do with
Stackless :o) 
	A Python 2.0 port of Stackless (preferably the newest code)
	Stackless next-generation (or whatever)

Micro-threads Priorities (this order may have nothing to do with reality :o)
	Break micro-threads module into the core mechanisms and the "helper"
mechanisms.  Potentially break all of the helper objects (those based on
Blockable, maybe) into their own python modules, so that only the required
code is loaded.
	Prioritisation features.
	Get a (current) micro-threads implementation "blessed" into the
Stackless distributions.
	Micro-threads reimplementation using C/C++ (at least for the
expensive global lookups and the like).  From what I can see, this will
dramatically reduce the size and complexity of the code, since everything
will already have the global interpreter lock, and can dispense with all the
weird contortions currently used to make the python code atomic all over the
place.  Don't know whether the most common blockable subclasses should be
converted to C, but I'm guessing it would be useful for Queues and RLocks at
least (I think those are the most commonly used).
	An audit of the micro-threads module by people who are
	Threading module interoperability (make micro-threads a suitable
replacement for threads to promote their adoption among the wider python
community (and provide solutions on threading-challenged operating systems))
		Shadow implementation of the threading module using
		Shadow implementation of micro-threads using threading
(lower priority)

	I use micro-threads for a fairly large research/prototype project,
would have to get the lawyers to sign off on any official proclamations...
not likely given the current level of paranoia about this particular project
(for krimminy's sake, they've put _me_ to doing C work, that's true
paranoia! (worst part is I had to advise them to do it "here is a gun, you
decide the right thing, that target on your forehead is just a
suggestion"))... closed source sucks, but at least they let me work on side

Okay, back to poking around,

Now it is time to start with discussions. For me, the most
urgent question to ask is: What do you consider the highest
priority task to be done? Upgrade to Stackless 1.2, upgrade
to Python 2.0, bugfixes, CVS access for collaborative
development, more documentation? Other proposals?

yours sincerely - chris
Stackless mailing list
Stackless at starship.python.net

More information about the Stackless mailing list