A5rocks
Alright. Does this happen when you run just `python -X dev`? (This should drop you into a REPL) Does it happen if you try `python -X dev -c "import cffi"`?...
Did you ever figure this out or did it fix itself?
I'll close this, let me know if you get this again/feel free to reopen.
`hasattr(select, "kqueue")` must be False for that to be False fwiw
Yeah, because eventlet patches out select.kqueue. If you import trio first things should work... we should probably make the error message better.
Would this as an error message work better?: "unsupported platform or the IO primitives trio relies on are patched out" if either eventlet or gevent are detected in sys.modules?
No update because it's not something we can really solve. The error message is better at least? (I guess if we felt like it we could implement an event manager...
As of https://github.com/python/mypy/pull/17392 the overlapping is gone. However, in the time since, this has gained an LSP error. I still think the issue can be closed?
Minimized: ```py def some[T]() -> None: x: list[T] reveal_type(x) ```
No, I think this *should* work with PEP 695 syntax. The user is explicitly binding the type variable in the function and mypy should respect that intent even if there's...