Update conformance test sources to initialize uninitialized variables #1804
Conversation
|
@yangdanny97, this draft of the PR looks good to me. If you think it's ready to go, I'm happy to merge it. |
|
Sounds good, thanks! |
| @@ -88,6 +88,9 @@ def int(self) -> None: # OK | |||
|
|
|||
| y: int = 0 # E: Refers to local int, which isn't a legal type expression | |||
|
|
|||
| def __init__(self, ClassC: "ClassC") -> None: | |||
There was a problem hiding this comment.
I think this is at least arguably incorrect, since ClassC in the annotation should resolve to the class variable ClassC, not the global. Mypy gives an error for this case.
There was a problem hiding this comment.
This looks to me like a bug in mypy. This is a deferred evaluation because the annotation is in quotes. The ClassC declared in the class body isn't in scope when the evaluation is performed.
There was a problem hiding this comment.
That said, this test isn't intended to check for this behavior, so we could trivially change it to avoid the mypy bug:
def __init__(self) -> None:
self.ClassC = ClassC()There was a problem hiding this comment.
@yangdanny97, if you're OK with this proposed modification, please push an update.
There was a problem hiding this comment.
@JelleZijlstra, are you still reviewing the PR, or are you done? If you're good with it, please merge.
Summary: per Eric's comment in python/typing#1804 (comment) _value_ is magic for enums, and we shouldn't give uninitialized attr errors for it Reviewed By: migeed-z Differential Revision: D59833300 fbshipit-source-id: 59391ee56aecbab8b648648d27927258b2b6d3fc
This PR adds constructors that initialize the attributes for the classes defined in each conformance test, per @erictraut comment on another PR (linked below).
#1734 (comment)
Please let me know if you'd prefer things to be initialized in a different way.
cc @stroxler