The patch also needs a versionchanged in the docs to say that unicode names are supported too. http://bugs.python.org/review/14462/diff/7181/Lib/test/test_re.py File Lib/test/test_re.py (right): http://bugs.python.org/review/14462/diff/7181/Lib/test/test_re.py#newcode197 Lib/test/test_re.py:197: re.compile('(?P<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>x)(?P=𝔘𝔫𝔦𝔠𝔬𝔡𝔢)(?(𝔘𝔫𝔦𝔠𝔬𝔡𝔢)y)') Writing non-BMP chars directly in the source is a bit risky -- some editors might not deal well with it, and in the past we had bugs with pyc files too. http://bugs.python.org/review/14462/diff/7181/Lib/test/test_re.py#newcode198 Lib/test/test_re.py:198: self.assertRaises(re.error, re.compile, '(?P<©>x)') Maybe add a comment that mentions why this fails (or something like "# check that valid identifiers works" at the top). http://bugs.python.org/review/14462/diff/7181/Lib/test/test_re.py#newcode203 Lib/test/test_re.py:203: self.assertRaises(re.error, re.sub, '(?P<a>x)', r'\g<©>', 'xx') Do these tests cover all the cases you changed in re.py? http://bugs.python.org/review/14462/