Isaac Virshup

Pointing out a very very quick version (https://github.com/theislab/ehrapy/issues/151#issuecomment-984956216) for anyone who wants something to try out while thinking about the API.

# Benchmarks ## Setup ```python from natsort import natsorted from string import ascii_letters import pandas as pd, numpy as np, anndata as ad from scipy import sparse letters = list(ascii_letters)...

I think the main issue here is `X_transformed` is an `ArrayView`, when it definitely shouldn't be. The view classes are not great, and need to be redone. I think this...

One other point: Is `NoSuchIO` a `KeyError`? Maybe it's more of a `TypeError` since it's basically dispatch?

> As it's a setting adjustable by the user, we would still need to add a config manager to anndata – unless I'm missing the point. It was a bit...

Did this ever work? I recall thinking about it when I implemented the `write_none` function, but was probably going for backwards compat then. Do you have a suggested way to...

I wouldn't like a string `None`, but we could encode a null type. E.g. `missing_el.attrs["encoding_type"] = "null"`. For now, I would say the typical way we handle this in scanpy...

@WeilerP if you wanted to look into this, I would appreciate some info on how other systems handle this. For instance `json` has `null`, but I'm not so sure about...

Thanks for the bug report, and sorry about the time it took to respond. I'm not able to reproduce with either 0.7.8 or 0.8.0. Are you still seeing this? ```python...

Could you add the notebook to the rendered docs, similar to how it's done [here](https://github.com/scverse/anndata/blob/74b63d41a9494d1d5bb78915cab9fb23169a287f/docs/benchmarks.rst)?