"exclude" config option not respected after upgrading to pep8 1.6.2 about flake8

[asottile]

In GitLab by @sigmavirus24 on Feb 16, 2015, 07:44

@IanLee1521 seems to be another regression in pep8

[asottile]

In GitLab by @IanLee1521 on Feb 16, 2015, 18:01

So I believe this is the real crux of the regression. This was changed by pep8 #368.

Namely, that if a config_file is provided by a user (or in this case passed in to pep8.StyleGuide via flake8.engine.get_style_guide()), pep8.read_config() will only read it, and not read any of the local project files PROJECT_CONFIG = ('setup.cfg', 'tox.ini', '.pep8') (nor will it read the user config either).

IMO, the config handling of pep8 could use some work overall. I'll work on a patch for this upstream.

[asottile]

In GitLab by @sigmavirus24 on Feb 22, 2015, 10:16

flake8 2.4.0 will be introducing upper caps on dependencies because issues with unexpected updates of pep8 have been a theme with version 1.6 of pep8. Upgrading to flake8 2.4.0 will fix this, but if you were looking for new functionality in pep8 1.6 you'll need to stick to 2.3.0

[asottile]

In GitLab by @blueyed on Mar 2, 2015, 14:44

@IanLee1521
There isn't an issue to track this for flake8 yet, is there?
Have you started working on it already?

[asottile]

In GitLab by @blueyed on Mar 2, 2015, 14:51

@sigmavirus24
Is the release of 2.4.0 planned for 2015-03-07 (according to 7c132d1)?

I think releasing a fixed version earlier would be appreciated, to avoid having more people run into this issue.

[asottile]

In GitLab by @sigmavirus24 on Mar 2, 2015, 14:56

@blueyed I need time to review https://gitlab.com/pycqa/flake8/merge_requests/18 and #2 for 2.4.0. If you'd like to help, then maybe I can get it delivered before the 7th. Otherwise, please be patient as I'm the only person working on this and if you look me up, I work on far more than just Flake8 and I do it all for free.

[asottile]

In GitLab by @sigmavirus24 on Mar 7, 2015, 19:25

2.4.0 is out with the version cap.

[asottile]

In GitLab by @oppianmatt on Mar 10, 2015, 06:01

This upper cap is causing our builds to break. pep8 is being downgraded to a buggy version. We are getting this error PyCQA/pycodestyle#306 which has been fixed in pep8 but flakes8 downgrades it. If we try pin it, it will now error with a pkg version error. I would say pinning an upper version in flakes8 for pep8 is the wrong thing to do. If people want to use a different version of pep8 with flakes8 they can pin it in the requirements otherwise we can't get the pep8 fixes.

[asottile]

In GitLab by @sigmavirus24 on Mar 10, 2015, 06:34

@oppianmatt alternatively, you can pin flake8 to 2.3.0. If you want flake8 to uncap pep8, talk to @IanLee1521 about getting pep8 fixed... or submit a patch to pep8 to fix the problems described here and elsewhere.

[asottile]

In GitLab by @IanLee1521 on Mar 10, 2015, 09:32

My apologies for the delay in looking into this upstream. Life got really busy the first half of this month. I will have time this coming weekend though and hope to have the patch out by the end of this weekend.

[asottile]

In GitLab by @oppianmatt on Mar 10, 2015, 10:10

I've had to pin flake8 back a version. My argument is why are you forcing us users to have a specific pep8 version? They are all buggy it seems. Now I have to have a more buggy flake8 to have a less buggy pep8. The argument to pinning pep8 down a version is that it doesn't do excludes, but should that not be the choice of the person who installs flake8? We can't use the latest version of flake8 because the version of pep8 is restricts us to is worse as in it breaks and does not run. I would rather have the choice of using the latest flake8 with my choice of pep8.

[asottile]

In GitLab by @sigmavirus24 on Mar 10, 2015, 10:18

@oppianmatt the answer is 1.5.7 breaks things for far fewer people than 1.6.x. It's Flake8's decision to use what dependencies it feels is most appropriate. The conversation is over. 2.4.0 is released. If you wanted a voice in the decision you should have involved yourself sooner.

[asottile]

In GitLab by @oppianmatt on Mar 10, 2015, 11:01

Sorry if I came off blunt, didn't mean to. I could not have been involved in the decision until after it came out as I was only alerted by the breaking of the build when the new version came out, took a while to track it down. Having something break is worse than not honoring includes/excludes and without the lower version pinned that was a choice that could be made by us, now it's taken away. Have pinned the older flake8 version as that was the only solution. Sorry for any trouble or hassle, just was wondering why we couldn't decide and why we couldn't overwrite it. But since @IanLee1521 is going to fix it soon it will be a moot point. Again, thanks for flakes8

[asottile]

In GitLab by @sigmavirus24 on Mar 17, 2015, 22:22

@oppianmatt you may be interested in https://mail.python.org/pipermail/code-quality/2015-March/000515.html

[asottile]

In GitLab by @akosiaris on Jan 15, 2016, 01:54

pep8 1.7.0 came out a few days ago, with PyCQA/pycodestyle@435d1cb in it which reportedly fixes the problem. I suppose that with this, end-users no longer have to pep8 and/or flake8 to get the desire they want, correct ? Which @oppianmatt, should fix your problem.