STINNER Victor <vstin...@redhat.com> added the comment: I created PR 7069 to remove the assertion. With this change, attached bug.py doesn't crash anymore, but it still logs warnings in verbose mode:
vstinner@apu$ ./python -v bug.py (...) PyThreadState_Clear: warning: thread still has a frame PyThreadState_Clear: warning: thread still has a frame PyThreadState_Clear: warning: thread still has a generator PyThreadState_Clear: warning: thread still has a frame PyThreadState_Clear: warning: thread still has a generator PyThreadState_Clear: warning: thread still has a frame The root issue is that Python doesn't join deamonic threads. But it's by design no? I hate daemonic threads :-) For the specific case of a fork(), it can be possible that a thread calls os.fork() while another thread is running a generator, no? -- Since this issue seems to be a regression and I have a PR fixing it, I propose to mark this issue as a release blocker. Sorry Ned :-( ---------- nosy: +ned.deily priority: critical -> release blocker _______________________________________ Python tracker <rep...@bugs.python.org> <https://bugs.python.org/issue33612> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
