[Python-Dev] Re: Python 2.0 and Stackless

[Just van Rossum]

At 11:49 AM -0400 06-08-2000, Barry A. Warsaw wrote:
>It's okay if there are some peripheral modules that are available to
>CPython but not JPython (include Python .NET here too), and vice
>versa.  That'll just be the nature of things.  But whatever basic
>language features Stackless allows one to do /from Python/ must be
>documented.

The things stackless offers are no different from:

- os.open()
- os.popen()
- os.system()
- os.fork()
- threading (!!!)

These things are all doable /from Python/, yet their non-portability seems
hardly an issue for the Python Standard Library.

>That's the only way we'll be able to one of these things:
>
>- support the feature a different way in a different implementation
>- agree that the feature is part of the Python language definition,
>  but possibly not (yet) supported by all implementations.

Honest (but possibly stupid) question: are extension modules part of the
language definition?

>- define the feature as implementation dependent (so people writing
>  portable code know to avoid those features).

It's an optional extension module, so this should be obvious. (As it
happens, it depends on a new and improved implementation of ceval.c, but
this is really beside the point.)

Just

PS: thanks to everybody who kept CC-ing me in this thread; it's much
appreciated as I'm not on python-dev.