STINNER Victor <vstin...@python.org> added the comment:
commit 4d96b4635aeff1b8ad41d41422ce808ce0b971c8 "bpo-39511: PyThreadState_Clear() calls on_delete (GH-18296)" "(...) The release_sentinel() function is now called when the C API is still fully working." This change is "safer" than the other one, but there is still a risk of introducing a regression. The problem is that this commit is part of a larger effort to make the Python finalization more reliable. The commit makes sense in the serie of commits pushed into the master branch, but I'm not sure that it makes sense if it's taken alone. I don't have the bandwidth right now to investigate this issue, to identify the root issue, and suggest which commits should be backported or not. ---------- _______________________________________ Python tracker <rep...@bugs.python.org> <https://bugs.python.org/issue39360> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com