Idea: use ruff as a replacement for flake8 + isort ?

[Tom-Willemsen]

Test for adding ruff as a replacement for flake8 + isort.

Main advantage is speed (approx 100x faster):

$ time isort --check .

real	0m0.420s
user	0m0.349s
sys	0m0.040s

$ time flake8 src/

real	0m0.859s
user	0m2.157s
sys	0m0.154s

$ time ruff .

real	0m0.011s
user	0m0.006s
sys	0m0.006s

One code change is required due to a difference in opinion between ruff and isort about constructs like:

from somewhere import thing1
from somewhere import thing2 as aliased_name
from somewhere import thing3

which ruff prefers formatted as:

from somewhere import thing1, thing3
from somewhere import thing2 as aliased_name

Other than that it is a 1:1 replacement in terms of linting rules.

This seems to be the direction a lot of modern python projects are moving in, and a better dev experience with pre-commit etc is always nice.

[codecov]

Codecov Report

Merging #890 (93649a2) into main (5e751e8) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main     #890   +/-   ##
=======================================
  Coverage   92.96%   92.96%           
=======================================
  Files          50       50           
  Lines        2417     2417           
=======================================
  Hits         2247     2247           
  Misses        170      170           

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[DominicOram]

This worked for me but required a bit of faffing to get vscode happy. Had to change Python: Select Linter to None and install the ruff extension. @dperl-dls, do you have strong opinions on this? One issue is that we're now using a different formatter to dodal.

[Tom-Willemsen]

One issue is that we're now using a different formatter to dodal.

I agree if we decide to use this we should also do an equivalent migration on dodal - I only opened the one PR initially in order to get opinions and not waste too much time if we decide we don't like this.

[DominicOram]

One issue is that we're now using a different formatter to dodal.

I agree if we decide to use this we should also do an equivalent migration on dodal - I only opened the one PR initially in order to get opinions and not waste too much time if we decide we don't like this.

In which case we should probably ask @callumforrester

[d-perl]

I don't have a very strong opinion on this but I have noticed some slowdown with flake8 so I am mildly in favour

[callumforrester]

@coretl, weren't you looking at migrating the skeleton to ruff at some point?

[coretl]

we are planning to, but haven't had time yet: DiamondLightSource/python3-pip-skeleton#142

[DominicOram]

Is a better way to do this to change the skeleton then merge it in here and dodal then?

[coretl]

@Tom-Willemsen just volunteered to do this and I said yes please...

[Tom-Willemsen]

See DiamondLightSource/python3-pip-skeleton#147

[Tom-Willemsen]

This PR is now a cherry-pick/merge of DiamondLightSource/python3-pip-skeleton#147 .

[DominicOram]

Looks good to me, thanks. Can we merge it into dodal too please?