Benjamin Peterson writes:

2009/9/13 Stephen J. Turnbull <stephen@xemacs.org>:

...

Benjamin Peterson writes:

 > It would actually be a huge compatibility break because finally  > statments would no longer be garunteed to execute.

Hey, kill -9 trumps any guarantee you can give, and that's what the user resorts to in the use case at hand.  This is a bad thing, even from the point of view of those who use "finally" a lot.

Naturally, users should be able to control their programs absolutely, but C-c is supposed to tell the program to clean up all its messes then exit as gracefully as possible.

Sure. But setting things up so that C-c doesn't even get through to the program doesn't help that cause. Since the OP claims it works for him, maybe the "IDEA" has merit in some more limited form, for example, by default uninstalling the Python graceful_SIGINT_handler when calling into a C module. (Probably this would have to be introduced in stages to allow extensions to update their functions as needed -- I suppose this would be a performance hit.) Then you could provide a with-signal-handler form of the standard macro for defining a Python function, with the semantics that the signal handler change gets optimized away if you specify the standard handler. And document that in this form the function needs to handle the SIGINT_signaled flag (or however it works) itself, or accept the possibility of an uninterruptible hang.