Issue #33: convert to package and include typeshed type hints #34
Conversation
Codecov Report
@@ Coverage Diff @@
## master #34 +/- ##
=======================================
Coverage 95.34% 95.34%
=======================================
Files 1 1
Lines 365 365
Branches 29 29
=======================================
Hits 348 348
Misses 14 14
Partials 3 3
Continue to review full report at Codecov.
|
contextlib2/contextlib.pyi
Outdated
| # type ignore to deal with incomplete ParamSpec support in mypy | ||
| def contextmanager(func: Callable[_P, Iterator[_T]]) -> Callable[_P, _GeneratorContextManager[_T]]: ... # type: ignore | ||
|
|
||
| if sys.version_info >= (3, 7): |
There was a problem hiding this comment.
Isn't this always defined in contextlib2?
There was a problem hiding this comment.
You could swap out all the if sys.version_info >= ...: for if True: to minimize the diff noise?
There was a problem hiding this comment.
Ah, true. And I like that suggestion for minimising the diff.
There was a problem hiding this comment.
The ABC will be a bit troublesome as you might need to import them from typing_extensions?
There was a problem hiding this comment.
I wasn't sure about that, as the typing module imports were unconditional in the typeshed version as well - it only guarded their aliasing to contextlib API entries.
I figured either mypy (et al) must be doing some shenanigans behind the scenes to let that work even on older Python versions, or else it needed to be fixed in typeshed first, and then contextlib2 could inherit that fix.
There was a problem hiding this comment.
Ah you're right, it's already in 3.6 https://docs.python.org/3.6/library/typing.html#typing.AsyncContextManager
There was a problem hiding this comment.
Heh, we both found the same info in about the same time, but picked different subthreads to update first :)
|
There are only two deprecated APIs, and I honestly suspect I should just remove them completely at this point: #31 (not in 21.6.0, though). I did enhance the CI to actually run mypy, though, and confirmed locally that it failed as expected if there was something wrong in the pyi file. So I'll merge this now, check everything over again in the morning, and assuming I don't think of anything else I missed, I'll cut the release tomorrow (And in writing that, I realised I need release notes for this change!) |
|
@graingert Ah, that would explain why mypy only complained when I introduced outright errors into the stub, and not when I only made it inconsistent with the actual implementation. |
No description provided.