Isaac Virshup

> Where is the changelog? `release-notes/0.10.0.md`

It's very easy to add, but adding more types to serialize adds more work to other implementations. So I'd like to be conservative about this. E.g. if we really need...

In `anndata.compat` there are dummy classes defined for this. Basically replace `zarr.Array` and `zarr.Group` with `anndata.compat.ZarrArray` and `anndata.compat.ZarrGroup`

Encountering ordered dicts is kinda common, but idk that means we need to support round-tripping them in anndata. Like, maybe we could convert them to dicts on write. But then...

Hey @jhamman! I think it's pretty close to becoming a priority. Figuring out how heavy of a lift sparse arrays will be is the main thing here. Could you point...

I think this has been discussed offline, but I'm not totally convinced of this feature. At the moment, this is largely adding functionality equivalent to: ```python pd.DataFrame(adata.obsm[key], index=adata.obs_names) ``` which...

As an update, this should now be possible using dask backed arrays. I think we need to consider what exactly we want to get here. I suspect we may want...

> you'd like to let Dask handle the concatenation. Is this correct? And for this we would need to find a way to load data into dask without copies. Sorry...

> To clarify, we don't necessarily need a lazy intermediate object, right? These aren't necessary no. It could be useful to have an object interface for the on-disk store, but...

This has been implemented for anndata 0.10 with: * [`ad.experimental.concat_on_disk`](https://anndata.readthedocs.io/en/latest/generated/anndata.experimental.concat_on_disk.html) * as well as support for dask sparse arrays, as demonstrated in this [`gist`](https://gist.github.com/ivirshup/3fbe634b648304978ea77469b5d88961) Waiting on some usage docs to...