pycodestyle 2.3.1 py36 egg in PyPI has "v2.4.0" code

[ginoledesma]

Hi, All

I'm not sure who maintains the publishing of the pycodestyle module in PyPI, but I've been running into issues very similar to #741 , where pytest-flake8 was pulling in v2.3.1 of pycodestyle yet failing to run with exactly the same stack trace as issue #741. We were getting inconsistent results depending on the way we were setting up our environment — pip vs setuptools.

It seems like a build of 2.3.1 at commit 8f3aebd was published as a py36 egg to PyPI, but this is different from the tagged release at https://github.com/PyCQA/pycodestyle/tree/2.3.1.

$ curl -O https://files.pythonhosted.org/packages/9a/4c/40238ba394be5c2d3d794f072695331dd3e2111351e2820e45352eefc9e9/pycodestyle-2.3.1-py2.7.egg
$ curl -O https://files.pythonhosted.org/packages/e4/81/78fe51eb4038d1388b7217dd63770b0f428370207125047312886c923b26/pycodestyle-2.3.1-py2.py3-none-any.whl
$ curl -O https://files.pythonhosted.org/packages/2a/ad/19df03d320cbf2b5e32444bdc9e52f212d2c2a8e9202137c68f780c7e6f4/pycodestyle-2.3.1-py3.6.egg
$ curl -O https://files.pythonhosted.org/packages/e1/88/0e2cbf412bd849ea6f1af1f97882add46a374f4ba1d2aea39353609150ad/pycodestyle-2.3.1.tar.gz

$ mkdir py27-egg py36-egg wheel tar

$ unzip pycodestyle-2.3.1-py2.py3-none-any.whl -d wheel/
$ unzip pycodestyle-2.3.1-py2.7.egg -d py27-egg/
$ unzip pycodestyle-2.3.1-py3.6.egg -d py36-egg/
$ tar -C tar/ -zxvf pycodestyle-2.3.1.tar.gz 

$ find . -type f -name pycodestyle.py -exec shasum {} ';'
1ffe96ff7a0e4f3602569c7a8136d035d4c3c52d  ./wheel/pycodestyle.py
59c6e99db5e1f02b95b56b1502f647b724a38700  ./py36-egg/pycodestyle.py
1ffe96ff7a0e4f3602569c7a8136d035d4c3c52d  ./tar/pycodestyle-2.3.1/pycodestyle.py
17c3428ec0c4c240ebca7a72a43d05a8bd71d6ab  ./py27-egg/pycodestyle.py

The python 2.7 egg seems to be working ok (haven't fully tested), but it's python 3.6 egg that's broken. If for some reason setuptools/pip picks up the egg version of the package (instead of wheel/source), and they happen to be running python 3.6+, then they could run into this problem as well.

[sigmavirus24]

Thanks for pointing those out. Those eggs shouldn't have been there. They shouldn't be installable any longer and shouldn't cause further problems.