Issue 10104: test_socket failures on Debian unstable (original) (raw)

This is following r85480 (issue #7523). It seems the Debian unstable system on one of the buildbots has SOCK_NONBLOCK and SOCK_CLOEXEC, but doesn't support creating sockets using these flags:

====================================================================== ERROR: testInitNonBlocking (test.test_socket.NonBlockingTCPTests)

Traceback (most recent call last): File "/home/doko/buildarea/3.x.klose-debian-alpha/build/Lib/test/test_socket.py", line 914, in testInitNonBlocking socket.SOCK_NONBLOCK) File "/home/doko/buildarea/3.x.klose-debian-alpha/build/Lib/socket.py", line 94, in init _socket.socket.init(self, family, type, proto, fileno) socket.error: [Errno 22] Invalid argument

====================================================================== ERROR: test_SOCK_CLOEXEC (test.test_socket.CloexecConstantTest)

Traceback (most recent call last): File "/home/doko/buildarea/3.x.klose-debian-alpha/build/Lib/test/test_socket.py", line 1834, in test_SOCK_CLOEXEC socket.SOCK_STREAM | socket.SOCK_CLOEXEC) File "/home/doko/buildarea/3.x.klose-debian-alpha/build/Lib/socket.py", line 94, in init _socket.socket.init(self, family, type, proto, fileno) socket.error: [Errno 22] Invalid argument

====================================================================== ERROR: test_SOCK_NONBLOCK (test.test_socket.NonblockConstantTest)

Traceback (most recent call last): File "/home/doko/buildarea/3.x.klose-debian-alpha/build/Lib/test/test_socket.py", line 1854, in test_SOCK_NONBLOCK socket.SOCK_STREAM | socket.SOCK_NONBLOCK) File "/home/doko/buildarea/3.x.klose-debian-alpha/build/Lib/socket.py", line 94, in init _socket.socket.init(self, family, type, proto, fileno) socket.error: [Errno 22] Invalid argument

(full log at http://www.python.org/dev/buildbot/3.x/builders/alpha%20Debian%203.x/builds/258/steps/test/logs/stdio)

Thanks for the research. Given that it only happens on Debian unstable, and it only happens if you explicitly use SOCK_{CLOEXEC,NONBLOCK}, I would vote for not adding a workaround until some supported OS also exhibits the issue.