Stackless Python, eventual merge?

[Greg Ewing]

Martin v. Löwis wrote:

> Depends on your notion of "danger"; I usually associate it with fear
> and uncertainty :-)


Well, I *do* experience fear and uncertainty when
I find that the normally secure and predictable
environment of Python suddenly starts crashing
on me. It takes all the fun out of things.


> If you know it can crash when you do certain things, and you document
> that limitation, there is no danger. There are already ways to crash
> Python, the dlmodule being an example in the core distribution.


The problem in this case is that it's not something
*I* do which is going to cause the crash, it's
something done by the internals of some piece of
C code that I know nothing about. Something which
is not going to be documented, because it's a
perfectly legitimate and unremarkable thing for
C code to do.

So, the only way to find out whether a particular
extension is safe to use in Stackless is to try
it. And that only gives a definite answer when
the answer is "no, it's not safe". If it doesn't
crash, I still can't be confident that it won't
crash at some time in the future.

That's why I say that I would be nervous about
using Stackless. Not to say that would stop me
from using it, but I'd feel nervous while I was
doing it! :-)


> Also, readers should notice that your example requires Stackless to be
> used:


Yes, I understand that, and I acknowledge that it
won't cause any harm to have Stackless in the core
if the Stackless stuff is switched off by default.

-- 
Greg Ewing, Computer Science Dept,
University of Canterbury,	
Christchurch, New Zealand
http://www.cosc.canterbury.ac.nz/~greg