Bump mypy to 1.4.1

[AlexWaygood]

No description provided.

[AlexWaygood]

With the latest mypy update, this test fails... but only with --python-version 3.11 or --python-version 3.12:

typeshed/test_cases/stdlib/builtins/check_dict.py

Lines 52 to 54 in c402107

	i3: Iterable[str] = ["a.b"]
	i4: Iterable[bytes] = [b"a.b"]
	assert_type(dict(string.split(".") for string in i3), Dict[str, str])

I can reproduce this locally with mypy 1.4 installed:

typeshed>python tests/regr_test.py stdlib --platform win32 --python-version 3.7 3.8 3.9 3.10 3.11 3.12
Running mypy --platform win32 --python-version 3.7 on the standard library test cases... success
Running mypy --platform win32 --python-version 3.8 on the standard library test cases... success
Running mypy --platform win32 --python-version 3.9 on the standard library test cases... success
Running mypy --platform win32 --python-version 3.10 on the standard library test cases... success
Running mypy --platform win32 --python-version 3.11 on the standard library test cases... failure

test_cases\stdlib\builtins\check_dict.py:54: error:
Expression is of type "dict[Any, Any]", not "dict[str, str]"  [assert-type]
    assert_type(dict(string.split(".") for string in i3), Dict[str, str])
    ^

Running mypy --platform win32 --python-version 3.12 on the standard library test cases... failure

test_cases\stdlib\builtins\check_dict.py:54: error:
Expression is of type "dict[Any, Any]", not "dict[str, str]"  [assert-type]
    assert_type(dict(string.split(".") for string in i3), Dict[str, str])
    ^


Test completed with errors

Anybody have any idea why this might be?? I'm struggling to reproduce the issue in mypy playground: https://mypy-play.net/?mypy=latest&python=3.11&gist=7a359709f41eb7a8a26cc8e201dbae79

[AlexWaygood]

I bisected using an editable install of mypy to find the mypy commit that broke this test:

66602fcde3468d7a284aa89cb280d448f454e07b is the first bad commit
commit 66602fcde3468d7a284aa89cb280d448f454e07b
Author: Shantanu <[email protected]>
Date:   Tue May 2 06:51:13 2023 -0700

    Fix spurious errors on builtins.open (#15161)

    Fixes #14796

 mypy/semanal_main.py | 1 +
 1 file changed, 1 insertion(+)

Cc. @hauntsaninja

[AlexWaygood]

Luckily this seems to be a typeshed-specific problem; most users of dict won't be affected here. The regression is only reproducible with --custom-typeshed-dir <typeshed_clone>. Here's the results I get with mypy 1.4.0 installed:

typeshed>mypy -c "reveal_type(dict(string.split('.') for string in ['a.b']))"
<string>:1: note: Revealed type is "builtins.dict[builtins.str, builtins.str]"
Success: no issues found in 1 source file

typeshed>mypy -c "reveal_type(dict(string.split('.') for string in ['a.b']))" --custom-typeshed-dir .
<string>:1: note: Revealed type is "builtins.dict[Any, Any]"
Success: no issues found in 1 source file

This is presumably because typeshed main and mypy's vendored copy of typeshed now have slightly different versions of builtins.pyi, as mypy reverts the change where we started using LiteralString in builtins whenever it does a typeshed sync. (Mypy does this due to the fact that it doesn't yet support LiteralString).

[hauntsaninja]

I'll revert that mypy PR :-/

[github-actions]

According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉

[hauntsaninja]

Nice, thank you!

[AlexWaygood]

Nice, thank you!

Thank you for the 1.4.1 release!