Codecov Report

Attention: Patch coverage is 95.09804% with 5 lines in your changes missing coverage. Please review.

Project coverage is 99.61%. Comparing base (c4c8ce4) to head (e47c52e).

@@            Coverage Diff             @@
##             main    #3047      +/-   ##
==========================================
- Coverage   99.62%   99.61%   -0.02%     
==========================================
  Files         122      122              
  Lines       18405    18397       -8     
  Branches     1226     1227       +1     
==========================================
- Hits        18336    18326      -10     
- Misses         47       48       +1     
- Partials       22       23       +1     

Not sure if this is worth it, it just causes a lot of churn indenting/dedenting things for no real reason. The elif technically is redundant if the previous if always returns, but it makes it more clear that they're mutually exclusive, and might cause bugs in future if someone removes one of the returns. This also causes symmetric if-else checks to become non-symmetric now, which might be confusing to the reader.

I agree with @TeamSpen210 that this doesn't really help anything. It doesn't make things clearer (unlike flake8-commas) and removes semantics. Looking through the error codes listed on https://pypi.org/project/flake8-return/ (I'm sure ruff implements them smarter, ofc):

• R501 conflicts with mypy nvm, I was confusing this with implicit None return w/ a return type like None | int. I guess this is useful for standardization reasons, but I don't really like it.
• R502/R503 is just type checking
• R504 is potentially useful, but in general assigning to a variable is useful because it allows us to show a nicer name in tracebacks. (maybe we could just enable this? I'm not too sure)
• R505/R506/R507/R508 remove semantics