Honour "# noqa: F401" comments

[OddBloke]

Our codebase has to work as far back as Python 3.4, which doesn't have the typing module available. As such, all of our type annotations are in strings (so they are lazily evaluated during type checking and ignored when running normally) and our typing imports are conditional, with "# noqa: F401" appended to silence the flake8 warnings.

vulture doesn't honour these comments, and it seems like an obvious win for it to do so.

(Per http://flake8.pycqa.org/en/2.5.5/warnings.html, F401 is specifically the warning for "module imported but unused"; this matches very closely with Vulture's unused import warnings.)

[OddBloke]

Looking through existing issues, I've seen that "# noqa" support has generally been rejected. Hopefully this request for just one specific form of noqa comment will be considered, as in my opinion, it would reduce the barrier to adoption substantially.

[jendrikseipp]

Since Python 3.4 has reached EOL and Vulture stopped testing for this version recently, I don't think Vulture should add workarounds that are only needed for this version. Your problem seems to be more general though. However, I don't understand why you need imports if all your type annotations are in strings. Could you please provide some example code?

[sersorrel]

Even if type hints are specified as strings, you still have to import the things used in type hints to tell the type checker what they are – it doesn't matter if you say foo: MyObject or foo: "MyObject" or foo = None # type: MyObject, if mypy doesn't know which MyObject you're referring to.

[jendrikseipp]

Thanks for the clarification! I agree that honoring F401 noqa comments makes sense. I'd be happy to review a pull request that makes this change.

[RJ722]

Regarding honoring # noqa comments, I think that once we have Vulture available as a flake-8 plugin (#161), we wouldn't need to develop/maintain this functionality ourselves. Am I thinking in the right direction?

[The-Compiler]

@RJ722 I think you're right!

[jendrikseipp]

That's true. The question is whether we want to offer this functionality to users of standalone Vulture as well. I have no strong opinion on this, but I think the decision should depend on the complexity of the patch for this functionality and how many users would like to have it.

[RJ722]

@OddBloke are you okay with changing all of # noqa: F401, V104? (V104 is issue code Vulture raises for unused imports)

[jendrikseipp]

I guess it makes sense to let Vulture honour the F401 code as well.

[jendrikseipp]

I guess it makes sense to let Vulture honour the F401 code as well.

This has been implemented in 5e7fd66.

[OddBloke]

Thanks very much!

…

On Mon., Apr. 13, 2020, 05:06 Jendrik Seipp, ***@***.***> wrote: Closed #160 <#160>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#160 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAPKEGCYFJSZD3MWYGOPW3RMLIY5ANCNFSM4HP6ARZA> .