A006 and A002 both trigger for lambda's

[jaap3]

This example triggers two violations (with preview enabled), is this intentional?

lambda id: id + 1

$ ruff check --isolated --preview --select A test.py

test.py:1:8: A002 Argument `id` is shadowing a Python builtin
  |
1 | lambda id: id + 1
  |        ^^ A002
  |

test.py:1:8: A006 Lambda argument `id` is shadowing a Python builtin
  |
1 | lambda id: id + 1
  |        ^^ A006
  |

$ ruff --version
ruff 0.7.2

[dylwil3]

Nice catch! It looks like flake8-builtins distinguishes between (named) functions and lambdas. We should match that behavior and maybe update the documentation on A002 and A006.

My only hesitation is that this would be a breaking change for a stable rule - so if people were running the checker with preview off and catching this shadowing on lambdas, they would no longer catch it if we matched flake8-builtins.

Not sure the best protocol for handling this - @MichaReiser / @AlexWaygood ?

[AlexWaygood]

Great question. When we expand the scope of a stable rule, we generally put the new behaviour in preview mode for a couple of months before promoting it to stable. When we reduce the scope of a rule, it's more of a judgement call.

I can see the argument that this might be a breaking change for some people -- but it definitely feels like a bug to me to have two rules complain about the exact same error. So I'd be inclined to ship the behaviour change straight into stable for this one, so that users can benefit from the bugfix even if they haven't selected preview = true (most don't).