Changing the default encoding breaks test_broken_unicode (regressiontests.model_regress.tests.ModelTests)

[Raphaël Hertzog <hertzog@…>]

This bug has been reported in the Debian bug tracking system: ​http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=646634

If you add the lines below to /etc/python2.7/sitecustomize.py:

import sys
sys.setdefaultencoding('utf8')

Then you get this test failure:

FAIL: test_broken_unicode (regressiontests.model_regress.tests.ModelTests)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/root/python-django-1.3.1/tests/regressiontests/model_regress/tests.py", line 146, in
test_broken_unicode
    self.assertEqual(repr(b), "<BrokenUnicodeMethod: [Bad Unicode data]>")
AssertionError: '<BrokenUnicodeMethod: N\xc3\xa1zov: Jerry>' != '<BrokenUnicodeMethod: [Bad
Unicode data]>'

I know it's not a good idea to change the default encoding but maybe this test could be skipped if sys.getdefaultencoding() doesn't return ascii.

[Claude Paroz]

Skip if default encoding is not ascii

[Claude Paroz]

This test has been modified for Python 3 compatibility, and now it passes even when default encoding is utf-8. I think that having a default encoding different from 'ascii' or 'utf-8' is probably doomed to fail anyway in many other cases.