STINNER Victor <vstin...@redhat.com> added the comment:
Using "write.setsockopt(socket.SOL_SOCKET, socket.SO_SNDBUF, 0)", I ran a stress test on test_socket() in 3 different terminals and I failed to reproduce the bug, whereas the same test (in 3 terminals) usually reproduces the bug in less than 30 seconds. Using "write.setsockopt(socket.SOL_SOCKET, socket.SO_SNDBUF, 1)", I still reproduce the bug. It seems like Windows behaves differently with SO_SNDBUF=0 as if the socket has really no kernel buffer and so respects user preference for SO_SNDBUF=0 ;-) ---------- _______________________________________ Python tracker <rep...@bugs.python.org> <https://bugs.python.org/issue34130> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
