tests/test_future/test_builtins.py fails with Python3.5

[sagitter]

Hi all,

the tests/test_future/test_builtins.py test fails with Python3.5:

=================================== FAILURES ===================================
___________________________ BuiltinTest.test_compile ___________________________
self = <test_future.test_builtins.BuiltinTest testMethod=test_compile>
    def test_compile(self):
        compile('print(1)\n', '', 'exec')
        bom = b'\xef\xbb\xbf'
        compile(bom + b'print(1)\n', '', 'exec')
        compile(source='pass', filename='?', mode='exec')
        compile(dont_inherit=0, filename='tmp', source='0', mode='eval')
        compile('pass', '?', dont_inherit=1, mode='exec')
        # Fails on Py2.7:
        # Was: compile(memoryview(b"text"), "name", "exec")
        self.assertRaises(TypeError, compile)
        self.assertRaises(ValueError, compile, 'print(42)\n', '<string>', 'badmode')
        self.assertRaises(ValueError, compile, 'print(42)\n', '<string>', 'single', 0xff)
        # Raises TypeError in Python < v3.5, ValueError in v3.5:
        # self.assertRaises(TypeError, compile, chr(0), 'f', 'exec')
        self.assertRaises(TypeError, compile, 'pass', '?', 'exec',
                          mode='eval', source='0', filename='tmp')
        compile('print("\xe5")\n', '', 'exec')
>       self.assertRaises(TypeError, compile, chr(0), 'f', 'exec')
E       ValueError: source code string cannot contain null bytes
tests/test_future/test_builtins.py:532: ValueError
 1 failed, 975 passed, 51 skipped, 49 xfailed, 2 xpassed, 1 warnings in 40.77 seconds 

future build log on Fedora with Python2.7/Python3.5: https://kojipkgs.fedoraproject.org//work/tasks/471/11780471/build.log

Any suggestion ?
Thanks.

[dev-zero]

it seems that python-3.5 now throws a ValueError when encountering a NULL byte in source code rather than a TypeError as in python-3.4, see also https://docs.python.org/3/library/functions.html#compile:

Changed in version 3.5: Previously, TypeError was raised when null bytes were encountered in source.

[dev-zero]

I fixed it using:

diff --git a/tests/test_future/test_builtins.py b/tests/test_future/test_builtins.py
index a6db4fd..bc2971d 100644
--- a/tests/test_future/test_builtins.py
+++ b/tests/test_future/test_builtins.py
@@ -529,7 +529,7 @@ class BuiltinTest(unittest.TestCase):
         self.assertRaises(TypeError, compile, 'pass', '?', 'exec',
                           mode='eval', source='0', filename='tmp')
         compile('print("\xe5")\n', '', 'exec')
-        self.assertRaises(TypeError, compile, chr(0), 'f', 'exec')
+        self.assertRaises((TypeError, ValueError), compile, chr(0), 'f', 'exec')
         self.assertRaises(ValueError, compile, str('a = 1'), 'f', 'bad')

         # test the optimize argument

... since it seems to be only important that it throws an exception, but not which one.

After that I get some unexpected successes:

test_imod (test_future.test_bytes.TestBytes) ... unexpected success
test_mod (test_future.test_bytes.TestBytes) ... unexpected success

Investigating those tests reveals that they are actually from py-3.5 and are supposed to succeed.

[turicas]

This issue seems related to #285 and was also found as a Debian bug, which is preventing lots of packages going from sid to testing, like: python-future, python-x2go, pyhoca-cli, remmina. So users of Debian testing cannot use Remmina (RDP client) because of this failing test. =/

Do you have any news on fixing this? Need help?