Noissue - Fix E305 by adapting flake8-abp to use groups, run tests on Python 3.6

Noissue - Fix E305 by adapting flake8-abp to use groups, run tests on Python 3.6

[Sebastian Noack, 2017-01-12 15:55:03]

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/flake8_a...
File flake8-abp/flake8_abp.py (left):

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/flake8_a...
flake8-abp/flake8_abp.py:375: check_non_default_encoding.name =
'abp-non-default-encoding'
With flake8 3.2.0, it begun to use pycodestyle 2.2.0, introducing an E305 here,
since according to PEP-8 there should be two blank lines below the function
definition. Arguably this doesn't make much sense if the code below assigns
metadata to the same function.

However, it turned out that if we group our checkers together (by using a common
prefix separated by a dot, when registering the entry points in setup.py),
flake8 is using that prefix as name and the __version__ variable of this module,
while the attributes on the individual checkers are obsolete.

Unfortunately, this requires flake8 3, and we are still stuck with flake8 2 for
sitescripts, buildtools and cms where we also use flake8-putty. But we can
basically archive the same for flake8 2, by assigning the same name and version
in a loop, see below.

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/flake8_a...
File flake8-abp/flake8_abp.py (right):

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/flake8_a...
flake8-abp/flake8_abp.py:481: if int(flake8.__version__.split('.')[0]) < 3:
We could theoretically skip this check and assign the name and version no matter
what. However, this way it's 1. more obvious that this is a backwards
compatibility hack, 2. we make sure that the new mechanism (i.e. groups) is
already used (and tested) when using flake8 3.

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/tox.ini
File flake8-abp/tox.ini (right):

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/tox.ini#...
flake8-abp/tox.ini:2: envlist=py{27,35,36}-flake8{2,3}
For reference, as Python 3.6 besides being the current stable, still isn't
available as a package in Debian, we discussed using the
skip_missing_interpreters option, to keep the minimum test setup
straight-forward. However, this would also make any other Python version listed
here optional, and there doesn't seem to be a way to prevent that. Though you'd
still get warnings for missing interpreters, and one could argue that this might
be good enough applying some common sense. But we still couldn't agree on that
approach as it would make the current workflow/rule of addressing every issue
reported by tox more complex and therefore potentially fragile.

[Vasily Kuznetsov, 2017-01-12 16:08:51]

LGTM

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/flake8_a...
File flake8-abp/flake8_abp.py (right):

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/flake8_a...
flake8-abp/flake8_abp.py:481: if int(flake8.__version__.split('.')[0]) < 3:
On 2017/01/12 15:55:03, Sebastian Noack wrote:
> We could theoretically skip this check and assign the name and version no
matter
> what. However, this way it's 1. more obvious that this is a backwards
> compatibility hack, 2. we make sure that the new mechanism (i.e. groups) is
> already used (and tested) when using flake8 3.

Acknowledged.

[Vasily Kuznetsov, 2017-01-12 16:08:51]

LGTM

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/flake8_a...
File flake8-abp/flake8_abp.py (right):

https://codereview.adblockplus.org/29371619/diff/29371620/flake8-abp/flake8_a...
flake8-abp/flake8_abp.py:481: if int(flake8.__version__.split('.')[0]) < 3:
On 2017/01/12 15:55:03, Sebastian Noack wrote:
> We could theoretically skip this check and assign the name and version no
matter
> what. However, this way it's 1. more obvious that this is a backwards
> compatibility hack, 2. we make sure that the new mechanism (i.e. groups) is
> already used (and tested) when using flake8 3.

Acknowledged.

[Vasily Kuznetsov, 2017-01-12 16:09:56]

Funny, the mouse made a double click by accident and it seems the comment got
posted twice.