Return bool from operator comparison functions

[AlexWaygood]

No description provided.

[github-actions]

Diff from mypy_primer, showing the effect of this PR on open source code:

pandas (https://github.com/pandas-dev/pandas)
+ pandas/io/formats/style.py:3511: error: Unsupported left operand type for & ("object")  [operator]

[JelleZijlstra]

Returning a non-bool from these methods isn't all that uncommon; numpy arrays do it. Perhaps we could do something fancy here with generic Protocols.

(Also, unrelatedly, thanks for all the enthusiasm in improving things! I really appreciate your work here.)

[AlexWaygood]

(Also, unrelatedly, thanks for all the enthusiasm in improving things! I really appreciate your work here.)

Thanks 🙂 also, do let me know if I'm being a pain at any point!

[AlexWaygood]

Perhaps we could do something fancy here with generic Protocols.

Hmmmm. Do you have anything in mind? I actually wasn't aware of the numpy situation with these methods 😕

[JelleZijlstra]

Hmmmm. Do you have anything in mind?

Sorry I missed this. If we had only one operand, it would be something like a protocol with __eq__() -> _T, then operator.eq takes the Protocol and returns _T. But in fact we have to look at both operands, so we need something more complicated.

[AlexWaygood]

But in fact we have to look at both operands, so we need something more complicated.

Yeah. I tried doing something more ✨fancy✨ in #6448, but a lot of the false-positives seemed to arise from giving e.g. __lt__ and __gt__ different signatures. So that's quite a big constraining factor here 😕

[Akuli]

I think we should also distinguish between x < y and x.__lt__(y). An operator like < will call y.__gt__(x), if y.__gt__ exists but x.__lt__ doesn't. This was previously discussed when improving the type hint for max().

[AlexWaygood]

I think we should also distinguish between x < y and x.__lt__(y). An operator like < will call y.__gt__(x), if y.__gt__ exists but x.__lt__ doesn't. This was previously discussed when improving the type hint for max().

Just WAIT till you see my next PR