TYP: Fix np.interp signature for scalar types
#27869
Conversation
|
@jorenham ping. |
There was a problem hiding this comment.
Thanks for submitting this! I left some notes on how to improve this. These are related to quite subtle and often-overlooked issues, so let me know if you don't understand my explanation.
But it's no surprise that you ran into them; these issues are currently all over the place in the numpy stubs, some of which I'm guilty of myself 😅. But even so, I think it's important to not repeat the mistakes of the past, otherwise it'll never get better 💪🏻.
np.interp typingnp.interp signature for scalar types
Co-authored-by: PTUsumit <2301109104@ptuniv.edu.in>
61f9844 to
bef5cf1
Compare
|
mypy_primer shows no diff: |
|
ready for review @charris |
| left: None | _FloatLike_co = ..., | ||
| right: None | _FloatLike_co = ..., | ||
| period: None | _FloatLike_co = ..., | ||
| left: _FloatLike_co | None = None, |
There was a problem hiding this comment.
Curious about this usage of None = None.
There was a problem hiding this comment.
Two things changed here:
- The
Noneshould always be the last type within a union (according to the typing spec style guide, ruff, andflak8-pyi), so I tend to correct that in the places where I change things in the stubs (not the best for the diffs, but I can't really help myself 😅). - You noted before that annotations are a form of documentation. And when documenting functions, you also document the defaults. Using
= ...in stubs was done by mypy'sstubgen(for unknown reasons), and was therefore assumed to be the way to go. But for documentation, and also for things like IDE suggestions and type-checker error messages, it's way more helpful if the actual default value is used, instead of a.... I believe the typing spec recently also started recommending doing this.
| from numpy import ( | ||
| vectorize as vectorize, |
There was a problem hiding this comment.
Remind me of why vectorize as vectorize makes sense. I've forgotten.
There was a problem hiding this comment.
In this case it doesn't, but if there's no __all__, an import spatula from spatula will implicitly export spatula as a public member of that module. It's almost always better to use __all__, but mypy's stubgen tool used to (since last month) generate stubs with such implicit exports.
|
Thanks @PTUsumit, Joren. |
Fix static interpolation handling
Description
The type hints for interp assert that the return dtype is a numpy array, but when the function is called with a scaler it gives a float. This raises some mypy errors.
#27811
Changes made on _function_base_impl.pyi and test_function_base.py