Nico Schlömer

> I remember large chunks of scipy that were sloppy about this, but I think a motivated user could fix them one by one. I'm happy to report that said...

@mattip Release note added.

> It would be good to have a test in core/tests/test_deprecations.py. It is not super important to be in that file, but it keeps deprecations in one place and also...

If there's anything I can do to advance this, let me know.

@asmeurer [came up with an interesting example.](https://mail.python.org/pipermail/numpy-discussion/2021-September/082076.html) It goes something like ```python import numpy as np a = np.random.rand(2, 3) a[0] = [1, 2, 3] # ok a[0] = np.array([1,...

> I was aware of it, and always considered this part of the deprecation here. I am not sure why you see it as different from your example of arr1d[0]...

Nah, just scalars for now. I think that's already an improvement over the current behavior. I think it'd make sense to have the multidim case follow in a second PR.

I sure am! Not sure what exactly needs to be done though. Perhaps another run on SciPy, Pandas, etc. to see if anything has crept up in the meantime.

I'm not sure about `abs()`. It depends on where you're coming from mathematically. If you're looking at the RMS as a [generalized mean](https://en.wikipedia.org/wiki/Generalized_mean) with p=2, then `abs` is wrong. After...

@ilayn : > RMS is not a generalized mean in the sense given in Wiki page I don't understand this. It's exactly what's in the Wiki page for p=2, isn't...