Regression in 0.660 regarding StopIteration

[boustrophedon]

I'm not really sure if this is a regression or if this is mypy picking up something new, but if so I don't understand why.

The documentation for StopIteration says that "The exception object has a single attribute value, which is given as an argument when constructing the exception, and defaults to None." I don't get an error that I'm using the optional as a non-optional in my original code with 0.650, so probably something is wrong there anyway.

def main() -> None:
    s = StopIteration("hello")
    print('value' in dir(s))
    print(s.value)

if __name__ == '__main__':
    main()

commands to reproduce:

python --version
# Python 3.7.1
python -m venv .
source bin/activate
pip install --upgrade pip # not necessary for reproduction
pip install mypy==0.650
mypy test.py
# no output
pip install mypy==0.660
mypy test.py
# test.py:4: error: "StopIteration" has no attribute "value"`

However, running python test.py shows that StopIteration definitely has a value attribute:

python test.py
# True
# hello

[JelleZijlstra]

Looks like this was introduced in python/typeshed#2533. I will submit a fix to typeshed now.

[boustrophedon]

Thanks!

[JelleZijlstra]

Fixed in typeshed now; the fix should get published with the next mypy release.

My sense is that this bug isn't bad enough to warrant an emergency release, but cc @msullivan as the release manager.