New submission from Bradley Froehle: In Python 3.3.0, several global `PyTypeObject`s are not initialized using PyType_Ready(...).
The list of uninitialized type objects:: PyCapsule_Type PyLongRangeIter_Type PyFieldNameIter_Type PyFormatterIter_Type PySTEntry_Type PyCell_Type PyInstanceMethod_Type PyClassMethodDescr_Type PyMethodDescr_Type PyCallIter_Type PySeqIter_Type PyDictDummy_Type Each of these can be verified using gdb:: $ gdb -quiet python3.3 Reading symbols from /opt/python/3.3.0/bin/python3.3...done. (gdb) break Py_Finalize Breakpoint 1 at 0x4a40a0: file Python/pythonrun.c, line 478. (gdb) run -c "pass" Starting program: /opt/python/3.3.0/bin/python3.3 -c "pass" [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Breakpoint 1, Py_Finalize () at Python/pythonrun.c:478 478 if (!initialized) (gdb) print PyCell_Type->tp_mro $1 = (PyObject *) 0x0 (The list of uninitialized types was built by searching "info variables" in gdb to produce a list of PyTypeObjects. Those with <name>->tp_mro != NULL were treated as properly initialized. The remainder were further inspected by hand to throw out additional false positives, like those from modules which had been statically compiled into the Python binary). I'm not sure if any of these actually need fixing, but for PyCell_Type this issue already resulted in one obscure bug: #16268. ---------- components: Interpreter Core messages: 174255 nosy: bfroehle priority: normal severity: normal status: open title: Global PyTypeObjects not initialized with PyType_Ready(...) type: behavior versions: Python 3.3 _______________________________________ Python tracker <rep...@bugs.python.org> <http://bugs.python.org/issue16369> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
