An implementation of the idea expressed here:

http://mail.python.org/pipermail/python-dev/2002-June/025067.html

This exposes the clearing of the intern dictionary to
Python via gc.release_interns().  Patch includes doc
updates and a test case.

Logged In: YES 
user_id=6380

Um, that's totally unsafe. See
http://mail.python.org/pipermail/python-dev/2002-June/025071.html

Logged In: YES 
user_id=12800

Yes, I know why this isn't right.  Notes:

- we probably want to incref the string pointed to by
ob_intern so that the reference is counted in the interned
string's refcount

- when the referent is freed, we'll need to decref the ob_intern

this addresses

http://mail.python.org/pipermail/python-dev/2002-June/025071.html

Logged In: YES 
user_id=12800

Guido also suggests to move this sys instead of gc (although
gc isn't totally out of the question).  Maybe Neil has an
opinion one way or the other?

Logged In: YES 
user_id=6380

Barry, are you going to do an implementation of what we
decided on? If not, maybe unassign this and let someone else
submit one.

Logged In: YES 
user_id=12800

I'm planning to, yes.  I may not get to it until early next
week though, so if someone is motivated to do it before
then, let me know (or steal the assignment).

Logged In: YES 
user_id=21627

Could not this get exposed as str.some_static_method?

Logged In: YES 
user_id=12800

I'm closing this as rejected since others have more time and
energy to do this (i.e. murder intern strings :) right.