bpo-46475: Add typing.Never and typing.assert_never #30842
Conversation
JelleZijlstra
requested review from
Fidget-Spinner and
gvanrossum
as code owners
There was a problem hiding this comment.
I think we should iterate this until @Fidget-Spinner approves it and then he can merge it. I think the idea is good, and IIRC typing-sig is in agreement. (Though are there folks at Pyre or Google who don't read it but want issues instead? Sigh. :-( )
|
Thanks @gvanrossum! I think I addressed all previous comments, so from my perspective this is ready. https://mail.python.org/archives/list/typing-sig@python.org/thread/CGIYKGMF6XLDX3F2JNRCCE7HC6K2XLZ2/ is the typing-sig thread. Nobody opposed it. |
|
Should I wait for the tests or are they hanging? |
|
They just started, I had to fix a merge conflict. It's going to conflict with the Self change too, which is why I pinged you on that one first. I think this is the easiest way forward:
|
| def int_or_str(arg: int | str) -> None: | ||
| match arg: | ||
| case int(): | ||
| print("It's an int") | ||
| case str(): | ||
| print("It's a str") | ||
| case _: | ||
| assert_never(arg) | ||
|
|
||
| If a type checker finds that a call to assert_never() is | ||
| reachable, it will emit an error. |
There was a problem hiding this comment.
This example is much better than the examples in the main doc. :-) And so is the sentence following the example.
There was a problem hiding this comment.
That sentence is also in the docs for assert_never, but not Never. I'm pushing some changes to improve the Never docs according to your feedback.
Honestly I'm a bit confused about the relationship between the docstring and the RST docs. I end up copying from one to the other.
There was a problem hiding this comment.
Usually the docstring is shorter, since it is often displayed by tools that have limited output space (it's really annoying to get a scroll bar in a tooltip in VS Code :-).
|
Thanks for your feedback! I pushed a round of docs changes. |
|
Thank you for the feedback! |
Backport of python/cpython#30842, with additional tests from @sobolevn's python/cpython#31222.
Backport of python/cpython#30842, with additional tests from @sobolevn's python/cpython#31222.
Backport of python/cpython#30842, with additional tests from @sobolevn's python/cpython#31222.
Backport of python/cpython#30842, with additional tests from @sobolevn's python/cpython#31222.
Backport of python/cpython#30842, with additional tests from @sobolevn's python/cpython#31222.
Another pending proposal from typing-sig.
Neverprimitive that is semantically identical toNoReturn, but documented explicitly as a bottom type. I chose to make it a separate object instead of a pure alias forNoReturnso that dynamic checkers can distinguish between the two and so that we can document them separately.assert_never()function to statically assert that code is unreachable. Based on @davidfstr's feedback, the error raised by the function does not include the argument.https://bugs.python.org/issue46475