"Literal[5]" cannot be assigned to type "ToJustInt"

[osamimi]

├── scipy v1.15.1
├── scipy-stubs v1.15.1.0

This call

    coeffs = scipy.signal.firls(
        numtaps=5,
        bands=np.array(scipy_bands),
        desired=np.array(scipy_desired),
        weight=np.array(scipy_weights),
        fs=1.0,
    )

produces the following:

Argument of type "Literal[5]" cannot be assigned to parameter "numtaps" of type "ToJustInt" in function "firls"
  Type "Literal[5]" cannot be assigned to type "ToJustInt"
    "Literal[5]" is incompatible with protocol "JustInt"
      "__class__" is an incompatible type
        Type "() -> type[Self@object]" cannot be assigned to type "() -> type[int]"
          Function return type "type[Self@object]" is incompatible with type "type[int]"
    "Literal[5]" is incompatible with "integer[Any]"

I don't understand why this doesn't type check?

[jorenham]

What (pyright?) version are you using?

[jorenham]

I checked with the latest pyright, basedpyright, and mypy, but I can't reproduce it.

import numpy as np
import scipy

coeffs = scipy.signal.firls(
    numtaps=5,
    bands=np.array([.0]),
    desired=np.array([.0]),
    weight=np.array([.0]),
    fs=1.0,
)

$ uv run pyright issue_420.py
0 errors, 0 warnings, 0 informations 
$ uv run basedpyright issue_420.py
0 errors, 0 warnings, 0 notes
$ uv run mypy issue_420.py
Success: no issues found in 1 source file

It would also help if you could share the particular config that you're using.

[osamimi]

I was seeing this in vscode with pylance extention 2024.12.100 which seems to use pyright 1.1.390 and the setting typeCheckingMode = "standard"

But I tried command-line pyright 1.1.390 and 1.1.392 and that doesn't show any errors. Not sure how to explain this...

[jorenham]

Hmm that's indeed strange... As far as I'm aware, pylance directly uses pyright 🤔

[jorenham]

Perhaps they use different typeshed versions? Because I recently fixed a bug there that was causing issues with these Just* types in pyright.

Anyway, you could try upgrading to the pre-release version of pylance, and see if that helps.

[jorenham]

I just tried with Pylance 2024.12.100 in standard more, but I'm still not able to reproduce it. Did you try reinstalling Pylance?