The object `None` works but the type `NoneType` doesn't work

[hyperkai]

I used the object None, then error doesn't occur but the readability isn't good not to differentiate the type None and the default value None as shown below:

*Memo:

• mypy 1.18.2
• Windows 11

                   # Cannot differenciate
                   # ↓↓↓↓   ↓↓↓↓
def func(name: str | None = None) -> str:
    return f'My name is {name}.'

So, I used the type NoneType to improve the readability to differentiate the type NoneType and the default value None but the error occurs as shown below:

from types import NoneType

                   # Can differenciate
                   # ↓↓↓↓↓↓↓↓   ↓↓↓↓
def func(name: str | NoneType = None) -> str:
    return f'My name is {x}.'

$ mypy test.py
test.py:5: error: NoneType should not be used as a type, please use None instead [valid-type]
Found 1 error in 1 file (checked 1 source file)

Actually, it's not reasonable that the object None works but the type NoneType doesn't work.

[wyattscarpenter]

I think the rule that tells you to use None instead of NoneType is deliberate, but might date back to an era of mypy when NoneType was broken. I haven't seen any good reason to disallow it since. So I think it should be fine to use NoneType, and mypy should allow it.

[emmatyping]

IIRC this was introduced because having two ways of spelling type(None) is confusing. I strongly recommend using None for both the type and value.

[A5rocks]

I think this is just mypy-as-a-linter. I hope --disable-error-code=valid-type works? IMO not a bug.

I guess maybe this would be better under a different error code?

[wyattscarpenter]

For historical context, which I've now bothered to track down, this diagnostic was introduced on purpose in #11288, and another motivation is also explained in #13139.

As I say, I don't see any real reason to disallow NoneType. I don't think it's that confusing to document that None is a typing shorthand for NoneType, given that Python already does that; nor does allowing NoneType seem likely to me to lead to invalid code. Incidentally, you can also write Literal[None], these days, which creates an equivalent type.

That said, I agree that this is just mypy-as-a-linter, and I imagine that some people like this lint rule because they really like just having None, and OP just has the opposite preference. Making it its own error code so it can be easily disabled seems like a great solution.