Code qa refresh

[pp-mo]

Replaces #564

Don't understand why this was needed, since the branch location is the same.
Seems like GitHub got confused somehow by my adding a commit + then removing it

[ukmo-ccbunney]

As per your other PR, this looks good.
Remember to raise an issue for the ignored numpydoc tests.

[ukmo-ccbunney]

Actually - jsut spotted one thing:

ruff-pre-commit hook is now version 0.7.4, you've got it as 0.7.3

Can you update?

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 89.60%. Comparing base (9425d19) to head (f8bdc8d).
Report is 33 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #568   +/-   ##
=======================================
  Coverage   89.60%   89.60%           
=======================================
  Files           8        8           
  Lines        2473     2473           
  Branches      420      420           
=======================================
  Hits         2216     2216           
  Misses        159      159           
  Partials       98       98           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[pp-mo]

ruff-pre-commit hook is now version 0.7.4, you've got it as 0.7.3

Just added in a re-ordering of the pre-commit hooks to match the current template order.

[ukmo-ccbunney]

ruff-pre-commit hook is now version 0.7.4, you've got it as 0.7.3

Just added in a re-ordering of the pre-commit hooks to match the current template order.

You are still running on 0.7.3 for Ruff though.

[pp-mo]

You are still running on 0.7.3 for Ruff though.

Sorry, missed including that somehow!

[ukmo-ccbunney]

I wonder if rather than setting strict=false in your [tools.mypy] section, it would be better to set strict=true and capture all the failing error codes?

For iris-grib, this would require the following ignores in the pyproject.toml:

strict = true 
disable_error_code =  [
    "arg-type",
    "assignment",
    "attr-defined",
    "index",
    "misc",
    "name-defined",
    "no-untyped-call",
    "no-untyped-def",
    "override",
    "type-arg",
    "union-attr",
    "var-annotated",
]

These can then all be added to a new issue documenting the failing mypy tests for iris-grib?

What do you think?

[pp-mo]

I wonder if rather than setting strict=false in your [tools.mypy] section, it would be better to set strict=true and capture all the failing error codes?
... What do you think?

Well, here's one way of looking at what that does ...

$ mypy . --no-color-output | grep -oE "\[[^[]*\]*\]$" errs.txt | sort | uniq -c
      9 [arg-type]
      8 [assignment]
    198 [attr-defined]
      5 [index]
     15 [misc]
     12 [name-defined]
   1421 [no-untyped-call]
    964 [no-untyped-def]
      3 [override]
      5 [type-arg]
     27 [union-attr]
     19 [var-annotated]

[pp-mo]

I wonder if rather than setting strict=false in your [tools.mypy] section, it would be better to set strict=true and capture all the failing error codes?

Done, I think.
See f8bdc8d

I don't know why the numbers don't seem to add up.
Presumably the settings are not independent so that, if all those checks are turned back "on", other errors emerge which rely on more than one of them being "on"

[ukmo-ccbunney]

OK. I think we're good to go.

[pp-mo]

Brilliant, thanks @ukmo-ccbunney !!

To complete this work :

• Small follow-up coming to rationalise the toml content : Update ruff controls #572
• some issues for future removal of "ignores"
  • ruff Address disabled Ruff checks #573
  • mypy Address disabled MyPy checks #570
  • numpydocs Address disabled numpydocs-validation checks #569
  • repo-review Address disabled repo-review checks #574