Maximilian Roos

https://github.com/dtolnay/serde-yaml/issues/363

I can see that perspective — @snth?

I'll archive by default, and then we can unarchive if @snth would prefer to keep it open (hope that's reasonable, didn't want to have things waiting for you @snth, totally...

Thanks for the issue @etienneschalk . Could you clarify the issue? What is the difference between the current output and the output you'd want?

Ah, thanks a lot for explaining. That does look weird. Generally `pytest-accept` will ignore any `...` and just replace the whole docstring. So it's unclear why it fails here. Would...

Thanks for the excellent example. I spent some time looking into this. Unfortunately, I think it's a problem with python's stdlib. Specifically in this example... ```python def negate(value: bool) ->...

This seems to be fixed! Thank you. I have another large issue, I'll open a new issue

That's a cool example, thanks @frazane . ...but unfortunately it looks `fastmath` this does: > Allow optimizations for floating-point arithmetic that assume that arguments and results are not NaNs or...

> wouldn't this then be a safer option for our use case? (I still get the same performance gains in the micro benchmarks above AND the results are also the...

> I guess a more careful inspection of the generated asm might reveal whether nan checks are actually done. But the fact that the `np.allclose` evaluates to True is a...