Floats interpreted as int?

Python Help
1

Hi,

Here I see that both floats and complex numbers are interpreted as int. That does not seem like a good idea. It is not very intuitive here:

from pydantic import RootModel

class MVAWp(RootModel[float | tuple[float,float]]):
    def __str__(self):
        if isinstance(self.root, float):
            return f'{self.root:.3f}'

        low, high = self.root
        return f'{low:.3f}-{high:.3f}'

fval : float = 3.

val = MVAWp(fval)
print(val)

val = MVAWp((fval, fval))
print(val)

I would always get a type annotation error:

image
image1313×621 67.4 KB

The only way I see to make this work safely is to create my own float type.

2

You can use numbers.Real in place of float for the purpose.

3

Which typechecker are you using? It looks like mypy does not produce an error, while pyright does for the code snippet you shared. I interpret the doc section linked to just refer to function arguments, not any variable marked as float (e.g., foo(1) is acceptable for def foo(x: float): ...).

Then, I’d view mypy’s behavior as more correct, while pyright is guarding against a potential pitfall. I don’t know how pydantic works under the hood, but MVAWp(1) works even though isinstance(1, float) == False.

Ben’s suggestion still produces type errors (seems like a bug though), but outside of pydantic numbers.Real would guard against passing ints vs floats.

4

Hi,

I am using pyright and it does produce an error for me, the maintainer himself acknowledges it here