Understand the self-destructing nature of Enum._ignore_

[kstauffer]

Similar to _order_ (which mypy already handles), _ignore_ is deleted by EnumMeta. Currently, (#12121) mypy thinks it's an error to set _ignore_ in an Enum subclass because 'Cannot override writable attribute "_ignore_" with a final one.' This PR both:

1. Makes mypy ok with setting _ignore_ in the class body; and
2. Makes it an error to later access the (deleted) _ignore_ attribute ('Type[E] has no attribute "_ignore_"')

The PR adds a test that accessing _ignore_ after the class definition is flagged as an error. I wanted to test that (1) above has been fixed, but I couldn't figure out how to make a failing test. So I went "out of band" from the unit test themselves; I created a simple test file:

from enum import Enum
class E(Enum):
    _ignore_ = '' # how to assert that old mypy "Cannot override writable"
E._ignore_

And checked that before this PR, mypy raised an error on line 3 but not line 4, and after this PR line 3 was ok but line 4 was now in error.

This PR does not cause mypy itself to ignore the attributes. That is, after this definition:

from enum import Enum
class E(Enum):
    _ignore_ = "X"
    X = 1

mypy still thinks E.X exists. I suppose mypy could parse the value of _ignore_ when it's a literal , but I don't have plans to look into that.

[sobolevn]

Thank you!

Blocking

• Let's add a constant

Non-blocking

I think if we are adding _ignore_ support, we must be sure that this case works correctly:

class Some(Enum):
   a = 1
   b = 2
   _ignore_ = ('a',)

reveal_type(Some.a)
reveal_type(Some.b)

However, I think that it can be addressed in a separate PR.
But, we need at least an new issue for it.

[kstauffer]

However, I think that it can be addressed in a separate PR. But, we need at least an new issue for it.

Created issue #12157.

[github-actions]

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

[sobolevn]

Thank you!

[kstauffer]

Thank you too!

…

On Fri, Feb 11, 2022 at 01:31 Nikita Sobolev ***@***.***> wrote: ***@***.**** approved this pull request. Thank you! — Reply to this email directly, view it on GitHub <#12128 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADLDC2XQARTD2IYL5XB7PMTU2SUNHANCNFSM5NS7LXFQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[hauntsaninja]

Looks good, thank you for this!