Logged In: YES 
user_id=6380

In the sake of stability for Python 2.3's accelerated
release schedule, I'm postponing this until after 2.3.

I'm also skeptical that it ca be absolutely correct.
What if there is Python code of the form

    for i in some_function(): ...

where some_function() is a C extension that at some
point invokes range(), directly from C. Then when
range() peeks in the opcode stream, it would believe
that it was being called in the place of some_function().

So maybe I should just reject it as unsafe?