bpo-1054041: Exit properly after an uncaught ^C.

[gpshead]

An uncaught KeyboardInterrupt exception means the user pressed ^C and
our code did not handle it. Programs that install SIGINT handlers are
supposed to reraise the SIGINT signal to the SIG_DFL handler in order
to exit in a manner that their calling process can detect that they
died due to a Ctrl-C. https://www.cons.org/cracauer/sigint.html

After this change on POSIX systems

while true; do python -c 'import time; time.sleep(23)'; done

can be stopped via a simple Ctrl-C instead of the shell infinitely
restarting a new python process.

https://bugs.python.org/issue1054041

[gpshead]

A proper unittest is needed, along with the news entry.

Windows specific code should go in a followup PR (if better behavior is meaningful there).

[jdemeyer]

I like this idea in principle. If you replace kill(getpid(), SIGINT) by raise(SIGINT) it might actually work on Windows too.

[eryksun]

I like this idea in principle. If you replace kill(getpid(), SIGINT) by raise(SIGINT) it might actually work on Windows too.

In Windows the default behavior for C raise is simply _exit(3), which is a meaningless exit status value. The correct course of action in Windows is to exit with STATUS_CONTROL_C_EXIT (0xC000013A), which is what the default console control handler does. The CMD shell has special-cased behavior for this status value. I covered this in more detail on the tracker.

[gpshead]

Trivial to implement that status return for windows, added to this PR. waiting for buildbots to double check compilation. I still need to add unittests and news.

[gpshead]

Things appear to be working, but the unittest checking the behavior of bash fails on macOS. This is probably due to macOS's very old version of bash not enabling the SIGINT detecting behavior under "bash -i".

TODO: A bash version check and skip could be added.

[bedevere-bot]

@gpshead: Please replace # with GH- in the commit message next time. Thanks!

[gpshead]

something very strange happened on Github here. it merged my commit with the unedited subject and commit message. I was manually fixing those up and putting in a nice clean message, clicked the button and github rejected it and refreshed to show me it was merged with the mess of messages you see. :(

[pablogsal]

There is a buildbot failure that seem related to this PR:

https://bugs.python.org/issue36013