eellison

> What would the category split look like? Having a different error message for function redefinition than other kinds of variable redefinition? `Redefinition of unused decorated function` or `Redefinition of...

> I think you're assuming that pyflakes knows far more than it does hencee my "basically identical cases" comment. They're very nearly identical to pyflakes if I'm understanding this thread...

> pyflakes currently special cases `typing.overload` because it knows that very specific special case -- it can't realistically discern between the decorators where the function is intentionally unused afterwards and...

```python class C: @property def an_attr(self): return 0 @property def an_attr(self): return 1 ``` Would have the error message: `Redefinition of unused decorated function`, and be in a different category...

> This still doesn't seem done Yea, it's not done sorry, I just pushed our comments in case it was super high pri for someone to take. have been spotty...

Going to address comments in the follow up PR, CI green 🚀

@pytorchbot merge

Is this fixed ?

Less than double what ? can we get a perf run of this before landing.

> The mm template is only used in a few OSS models, so I'd expect the general compile-time impact is low. But yeah good to have a perf run before...