>>> b'dGVzdA==\n'.decode('base64')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "../Lib/encodings/base64_codec.py", line 20, in base64_decode
    return (base64.decodebytes(input), len(input))
  File "../Lib/base64.py", line 359, in decodebytes
    raise TypeError("expected bytes, not %s" % s.__class__.__name__)
TypeError: expected bytes, not memoryview

base64, bz2, hex, quopri, rot13, uu and zlib codecs (reintroduced recently by r86934, issue #7475) cannot be used by str.encode/bytes.decode, but with .transform() and .untransform() methods of bytes and str objects. But these methods were removed by r87176.

The last solution to use base64 codec is:

>>> import codecs
>>> codecs.lookup('base64').decode(b'YWJj\n')[0]
b'abc'
>>> codecs.lookup('base64').encode(b'YWJj\n')[0]
b'abc'

Or simply use directly the base64 module:

>>> import base64
>>> base64.decodebytes(b'YWJj\n')
b'abc'
>>> base64.encodebytes(b'abc')
b'YWJj\n'

base64, bz2, hex, quopri, rot13, uu and zlib codecs should be removed from encodings.aliases (because they introduced a confusion for Python 2 users), or removed completly (because it's easier to use directly the related module, eg. base64 or zlib).

issue10807.patch just disables hex, base64, ... codecs in aliases (so it's still possible to use they through codecs.lookup()).

That does not look like the right patch...

Ah yes :-)

Looks good, please commit.

> Looks good, please commit.

Ok, done: r87642