[MRG] move flake8 options from shell script to setup.cfg fixes #10067 #10080

siftikha · 2017-11-06T20:59:16Z

Pulled ignore flags from build_tools/travis/flake8_diff.sh into setup.cfg.

Left --diff and --show-source flags in build_tools/travis/flake8_diff.sh since it seemed unwise to separate them from the actual call to flake8. If they should be merged into setup.cfg as well I can do that.

fixes #10067

reasons yo

jnothman · 2017-11-06T23:42:24Z

We now need to check that flake8_diff.sh runs flake8 in a context where setup.cfg is read. Yes, not moving --diff and --show-source is the right thing to do.

…

lesteve · 2017-11-07T08:36:01Z

build_tools/travis/flake8_diff.sh

    # Examples are allowed to not have imports at top of file
-    check_files "$(echo "$MODIFIED_FILES" | grep ^examples)" \
-           --ignore $DEFAULT_IGNORED_PEP8 --ignore E402
+    check_files "$(echo "$MODIFIED_FILES" | grep ^examples)" --ignore E402


We had a problem a while ago that --ignore replaces rather than adds on top of the default, this is why we have DEFAULT_IGNORED_PEP8 env variable. You can probably use git blame and look at the commit message or PR for more details. Can you please check the behaviour?

Yeah I realized this would be the issue as I was going to bed last night. I will see if there is a way to avoid that issue.

We had a problem a while ago that --ignore replaces rather than adds on top of the default, this is why we have DEFAULT_IGNORED_PEP8 env variable. You can probably use git blame and look at the commit message or PR for more details. Can you please check the behaviour?

So I checked and this matches my expectation: --ignore flag overrides whatever is in setup.cfg, so you need to duplicate what is in setup.cfg if you want to add an additional warning to ignore. All in all I am not sure what to do with this issue/PR ...

If we had at least one non-default flake8 warning that we wanted to ignore, it would be worth to have a flake8 section in setup.cfg but it is not even the case ...

After looking into it a bit more, I think it would make the most sense to move a flake8 config file into the examples directory with the --ignore flags for that directory and just use the default --ignore flags for the rest of the project.

Can you actually have a per-folder flake8 config file? I would not think so from just looking at the flake8 doc.

afaik when flake8 checks a file like /example_proj/examples/example.py it will check for config info in /example_proj/examples then in /example_proj with the info further up in the directory structure taking priority. so if you don't have a config file in /example_proj and you have one in /example_proj/examples the file in /example_proj/examples will take precedence.

I was definitely not aware of this, nice! The problem is that the per-folder config file does not seem to be taken into account by flake8 --diff 😞

Ah the --diff thing is true. Hmm. This seems like it is probably not going to work.

lesteve · 2017-11-07T08:38:38Z

setup.cfg

@@ -38,6 +38,9 @@ artifact_indexes=
    # https://ci.appveyor.com/project/sklearn-ci/scikit-learn/
    http://windows-wheels.scikit-learn.org/

+[flake8]
+ignore=E121,E123,E126,E24,E704,W503,W504


Are they actually needed in the sense that they should correspond to the default ignore flake8 settings if I believe the comment in master?

# Default ignore PEP8 violations are from flake8 3.3.0 DEFAULT_IGNORED_PEP8=E121,E123,E126,E226,E24,E704,W503,W504

From the issue that lead me to do this pr, I think these are truly project wide settings in that they are not something that the maintainers feel are worth worrying about.

@lesteve the default ignored list can change from version to version so its better to be explicit. Additionally, the current default for the most recent version ignores e226 which is something worth enforcing according to the issue this pr was in response to.

I am not 100% convinced, I would rather trust that flake8 default is a good default. As an example: it could well be that flake8 add a completely new kind of violations that they don't want to be shown by default. Setting ignore explicitly is not the best thing to do in this case.

jnothman · 2017-11-08T22:25:47Z

The reason for putting the ignores explicitly in setup.cfg is that different versions of flake8 have had different defaults. Putting them in there means user experience will match the version of flake8 we have in CI.

siftikha · 2017-11-08T23:36:46Z

So after looking into the --diff options a bit more (thank you @lesteve ) what I've done here isn't going to work, nor is it likely to be a decent base for moving forward. So I'm going to close this pull request to avoid wasting more time on it.

jnothman · 2017-11-08T23:45:12Z

Hey, this doesn't mean the issue is irrelevant. The point of the issue is that users contributing to scikit-learn should have the same config as our CI. So we might need to duplicate these settings in build_tools and in setup.cfg, but we still want them in setup.cfg!

siftikha · 2017-11-09T00:13:13Z

I had intended to do a different pr with a .flake8 config in both the base directory and in /examples but I guess I might as well just include that in this one.

lesteve · 2017-11-09T14:28:47Z

My personal preferences:

E226 (whitespace around operator) should be ignored. z = x*x + y*y reads a lot better than its counterpart with additional spaces (same thing for x = 1/3 if you ask me)
we should only have a flake8 section in setup.cfg if we do something different from the flake8 defaults. Given the previous bullet point, I don't think we should have a flake8 section in setup.cfg.
it's fine having a .flake8 config file in examples and using this config explicitly with --config in flake8_diff.sh.

siftikha · 2017-11-09T19:18:59Z

I agree about e226, but that's something you and @jnothman should talk about since the original issue he opened mentioned wanting to enforce that.

I disagree about not including a .cfg file for flake8 though, even if it is just the defaults. The defaults can change from version to version and without a config file in the folder if a user has a system wide flake8 file, the config could accidentally be overriden. It seems better to be explicit here.

jnothman · 2017-11-09T21:28:23Z

@agramfort was quite pro-E226

agramfort · 2017-11-10T08:37:16Z

indeed I have been "formated" to respect E226 and I think that most of the code base respects it.

lesteve · 2017-11-10T09:41:35Z

I assume everybody is aware that PEP8 recommend against E226. From https://www.python.org/dev/peps/pep-0008/#other-recommendations:

If operators with different priorities are used, consider adding whitespace around the operators with the lowest priority(ies). Use your own judgment; however, never use more than one space, and always have the same amount of whitespace on both sides of a binary operator.
Yes:

i = i + 1
submitted += 1
x = x*2 - 1
hypot2 = x*x + y*y
c = (a+b) * (a-b)

No:

i=i+1
submitted +=1
x = x * 2 - 1
hypot2 = x * x + y * y
c = (a + b) * (a - b)

All in all, I think we should have flake8 warnings only for things that are uncontroversial. E226 is controversial, so we should keep ignoring it.

rth · 2017-11-10T13:02:09Z

indeed I have been "formated" to respect E226 and I think that most of the code base respects it.

There are only 74 occurrences of this linting error,

flake8 --ignore E121,E123,E126,E24,E704,W503,W504 sklearn | grep E226

which yields flake8_E226.txt. So most of the current code base does currently respect it.

Not ignoring E226 would partially revert #9123 btw.

All in all, I think we should have flake8 warnings only for things that are uncontroversial. E226 is controversial, so we should keep ignoring it.

+1

The same argument is used by pycodestyle (and flake8) to define default errors:

In the default configuration, the checks E121, E123, E126, E133, E226, E241, E242, E704 and W503 are ignored because they are not rules unanimously accepted, and PEP 8 does not enforce them.

agramfort · 2017-11-10T19:40:23Z

I won't fight for it. +0 for E226. I will just keep using it myself :)

codecov · 2017-11-14T06:30:07Z

Codecov Report

Merging #10080 into master will increase coverage by <.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #10080      +/-   ##
==========================================
+ Coverage   96.19%   96.19%   +<.01%     
==========================================
  Files         336      337       +1     
  Lines       62739    62772      +33     
==========================================
+ Hits        60353    60386      +33     
  Misses       2386     2386

Impacted Files	Coverage Δ
sklearn/cluster/_feature_agglomeration.py	`100% <0%> (ø)`	⬆️
...klearn/cluster/tests/test_feature_agglomeration.py	`100% <0%> (ø)`

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update abb43c1...db4b40a. Read the comment docs.

lesteve · 2017-11-14T10:58:35Z

@dilutedsauce I pushed some minor tweaks. With these I am fine to merge. I'll wait for the CIs and merge this one, given that there seem to be consensus, e.g. by @agramfort and @rth.

lesteve · 2017-11-14T12:25:26Z

No need to wait for AppVeyor so merging this one. Thanks a lot @dilutedsauce!

…arn#10080) Also add examples/.flake8 for examples specific flake8 configuration

salman added 2 commits November 6, 2017 15:55

sh options for flake8 moved to setup.cfg

ede7b14

Merge branch 'master' into move_sh_to_cfg

abe543f

reasons yo

lesteve reviewed Nov 7, 2017

View reviewed changes

agramfort approved these changes Nov 8, 2017

View reviewed changes

siftikha closed this Nov 8, 2017

siftikha reopened this Nov 9, 2017

salman added 2 commits November 8, 2017 19:17

.flake8 added to examples

1375079

added comments to setup.cfg

08b190c

e226 added back into ignore

b931bfe

siftikha changed the title ~~move flake8 options from shell script to setup.cfg fixes #10067~~ [MRG] move flake8 options from shell script to setup.cfg fixes #10067 Nov 14, 2017

siftikha changed the title ~~[MRG] move flake8 options from shell script to setup.cfg fixes #10067~~ move flake8 options from shell script to setup.cfg fixes #10067 Nov 14, 2017

salman added 2 commits November 14, 2017 01:34

removed messed up bayes

dbf0c99

removed messed up bayes

db4b40a

siftikha changed the title ~~move flake8 options from shell script to setup.cfg fixes #10067~~ [MRG] move flake8 options from shell script to setup.cfg fixes #10067 Nov 14, 2017

Minor tweaks

6e240e9

lesteve merged commit 62eda82 into scikit-learn:master Nov 14, 2017

maskani-moh pushed a commit to maskani-moh/scikit-learn that referenced this pull request Nov 15, 2017

[MRG] move flake8 options from flake8_diff.sh to setup.cfg (scikit-le…

4df9e26

…arn#10080) Also add examples/.flake8 for examples specific flake8 configuration

jwjohnson314 pushed a commit to jwjohnson314/scikit-learn that referenced this pull request Dec 18, 2017

[MRG] move flake8 options from flake8_diff.sh to setup.cfg (scikit-le…

8d20b80

…arn#10080) Also add examples/.flake8 for examples specific flake8 configuration

Uh oh!

[MRG] move flake8 options from shell script to setup.cfg fixes #10067 #10080

[MRG] move flake8 options from shell script to setup.cfg fixes #10067 #10080

Uh oh!

Conversation

siftikha commented Nov 6, 2017

Uh oh!

jnothman commented Nov 6, 2017 via email

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lesteve Nov 8, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jnothman commented Nov 8, 2017 via email

Uh oh!

siftikha commented Nov 8, 2017

Uh oh!

jnothman commented Nov 8, 2017

Uh oh!

siftikha commented Nov 9, 2017

Uh oh!

lesteve commented Nov 9, 2017

Uh oh!

siftikha commented Nov 9, 2017

Uh oh!

jnothman commented Nov 9, 2017 via email

Uh oh!

agramfort commented Nov 10, 2017

Uh oh!

lesteve commented Nov 10, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rth commented Nov 10, 2017

Uh oh!

agramfort commented Nov 10, 2017 via email

Uh oh!

codecov bot commented Nov 14, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

lesteve commented Nov 14, 2017

Uh oh!

lesteve commented Nov 14, 2017

Uh oh!

Uh oh!

lesteve Nov 8, 2017 •

edited

Loading

lesteve commented Nov 10, 2017 •

edited

Loading

codecov bot commented Nov 14, 2017 •

edited

Loading