[Python-3000] Draft PEP: Module Initialization and finalization

[Tim Peters]

[Greg Ewing]
> It's been suggested before that the explicit clearing
> of a module at interpreter shutdown should be replaced
> with just letting cyclic GC do its thing. Is there any
> reason why that wouldn't work? It might lead to less of
> the unintuitive behaviour that sometimes occurs because
> of this.

Armin started moving in that direction 2.5 years ago, in patch 812369
(which is still open, but out of date).  Note that for some reason, he
still found it necessary there to None-out some module globals.