`N808` TypeVar names starting with a single underscore.

[Joshix-1]

I think _T_co should be an acceptable name for a covariant TypeVar. As there is no explicit test for this case it seems to me that this wasn't considered in #235
If it was intended to disallow that, then I'd like to know why and it would be good to add a test for that case.

I would prefer allowing a single underscore in front of the name of the TypeVar. But what do you think?

[sigmavirus24]

I think the more complicated we make the logic, the more likely we'll end up with unexpected problems that will make everything more and more complicated and buggy. Better to keep it simple for now

[Joshix-1]

Allowing a single under score in the beginning should be a single .removeprefix("_")

I would consider the current implementation buggy as it doesn't allow the underscore

[sigmavirus24]

You haven't explained the need for it, thus I reject the assrrtion it's a bug

[jparise]

I went back to PEP 8, and I think allowing TypeVar names to start with an underscore would be useful.

• Type Variable Names says that type variables "should normally use CapWords", which matches the Class Names language.
• Our N801 CapWords check (for class names) allows leading underscores.

[bkeryan]

I don't mean to rush you, but when do you expect a release containing this fix to be published? Knowing the timeline would help clients choose whether to wait for a patch or work around this issue with a version constraint/noqa/config option.

[jparise]

Pretty soon, I think. I was keeping an eye out for any other issues that might be reported.

[jparise]

Pretty soon, I think. I was keeping an eye out for any other issues that might be reported.

Version 0.15.1 is out now with this change.