New submission from STINNER Victor: Since my changeset a5efd5021ca1, the Python test suite starts to fail randomly. Running test_asyncio modifies sys.exc_info(): it is not (None, None, None) after the execution of test_asyncio. The problem comes from test_cancel_make_subprocess_transport_exec() of Lib/test/test_asyncio/test_subprocess.py.
I tried to write a simple script which does not depend on Python to reproduce the issue. See attached excinfo_bug2.py script. Output: --- exc_info after except (<class '__main__.MyException'>, MyException(), <traceback object at 0x7f09201ad7c8>) exc_info at exit (<class '__main__.MyException'>, MyException(), <traceback object at 0x7f09201abd08>) --- exc_info is supposed to be (None, None, None), at least at exit. I will try to write an even simpler script to identify the bug. ---------- files: excinfo_bug2.py messages: 235033 nosy: haypo priority: normal severity: normal status: open title: gnerator bug with exception: tstate->exc_value is not cleared after an except block versions: Python 3.4, Python 3.5 Added file: http://bugs.python.org/file37922/excinfo_bug2.py _______________________________________ Python tracker <rep...@bugs.python.org> <http://bugs.python.org/issue23353> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
