CI: isort ci

[alimcmaster1]

• passes git diff upstream/master -u -- "*.py" | flake8 --diff

asv_bench and doc dir where added to the isort skip section. We should actually check this directory in CI.

Checking doc dir is more effort that its worth ( barely any .py files ) so let's just leave this.

[codecov]

Codecov Report

Merging #24746 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master   #24746   +/-   ##
=======================================
  Coverage   92.38%   92.38%           
=======================================
  Files         166      166           
  Lines       52358    52358           
=======================================
  Hits        48373    48373           
  Misses       3985     3985

Flag	Coverage Δ
#multiple	90.81% <ø> (ø)	⬆️
#single	43.07% <ø> (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 33f91d8...8d47066. Read the comment docs.

[codecov]

Codecov Report

Merging #24746 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master   #24746   +/-   ##
=======================================
  Coverage   92.38%   92.38%           
=======================================
  Files         166      166           
  Lines       52363    52363           
=======================================
  Hits        48376    48376           
  Misses       3987     3987

Flag	Coverage Δ
#multiple	90.81% <ø> (ø)	⬆️
#single	42.91% <ø> (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 6c84005...fa214cd. Read the comment docs.

[gfyoung]

cc @datapythonista

[datapythonista]

I understand the point of the PR, and this makes things more consistent, but I'm not a big fan of what we did with isort. So making the list of excludes even longer, seems the opposite direction we should take.

Not a problem with getting this merged, but would be nice to see a solution of a very strict and complex and not-human friendly validation of half of the code base, and no validation in the rest.

[datapythonista]

Sorry, just saw that almost all the code has been isorted, didn't see that before. I guess this makes more sense then.

[alimcmaster1]

hey thanks for the comments @datapythonista , how come you are not a fan of isort out of interest? - In my mind we previously had an undocumented set of rules for what import order should look like ( by dependency structure ) - so now we don't need to battle on PRs about what that order should be, I just run one command and don't have to worry/often IDE's like to change the import order too ( "optimize imports" in pycharm for example ).

I will follow up after this PR to isort the asv_bench directory.

[datapythonista]

afaik none of the pandas devs is able to sort the imports properly unless using isort. I'm happy we respect pep8 and we have the imports sorted by stdlib, third-party and pandas. But the rest doesn't add any value, but extra steps and CI problems even for files with few imports.

[jreback]

using isort is much better than the alternative - random changes due to import ordering and re-ordering

[jreback]

can you merge master

[jreback]

thanks @alimcmaster1