Philipp A.

anndata-rs seems to handle only int32 and int64: https://github.com/kaizhang/anndata-rs/blob/df51c561e1efc453696fb2ae6c1a567e812f73a2/pyanndata/src/data/array.rs#L90 but that should be quick to remedy: https://github.com/kaizhang/anndata-rs/pull/18

> How do we ensure that we don't mess it up The difference between `const obj = { field: undefined }` and `const obj = {}` is that `field in...

Yeah, it’s not that complicated. Imagine some API takes a parameter `obj`, where it’s valid that `obj[key] === undefined`. Now if none of material-ui’s APIs where this applies acts differently...

I think strings are at least the primary reason for object dtypes to appear. Sadly there’s no variable width unicode string dtype that would make things clearer. But you’re of...

Yeah it’s true, R factors are fiddly. I’ve oscillated a few times between “strings everywhere” and “factors wherever useful”. Currently I’m in the latter mode, since `facet_wrap` respects factor level...

I’d definitely prefer the native API whenever it’s possible, since it’ll never run into issues like “the iframe is too small and now the button/text field UI is invisible”. So...

We could make it an option: By default clean up, and if the user would rather have a partially written file because of a deadline, let them have the option.

> The issue here is that it could have failed while writing any element inside the object, so I don't want to make promises about what can be recovered if...

I think #1991 doesn’t address this issue at all, instead it raises an issue if somehow a template file has an empty name _before_ substitution, which probably can’t even happen...

I definitely agree with the first paragraph. Especially with the upcoming `.mode` things get pretty long: ```py # as is now do_things_with(adata.mode['foo'].layer['bar']) # with .-access do_things_with(adata.mode.foo.layer.bar) ``` I’m not sure...