We should move towards cs{r,c}_array for the next minor release. This means a few things:
- https://anndata.readthedocs.io/en/stable/generated/anndata.settings.html#anndata.settings.use_sparse_array_on_read turn this setting on
- Inside
anndata.concat, create cs{r,c}_arrays by default (and anywhere else that they might be created)
- Update all of our tests to migrate to
csr_array for construction of AnnDatas, comparison etc.
- Push out the change on all channels to inform people of what is coming
See #2476 for where this is beginning to cause issues
We should move towards
cs{r,c}_arrayfor the next minor release. This means a few things:anndata.concat, createcs{r,c}_arrays by default (and anywhere else that they might be created)csr_arrayfor construction ofAnnDatas, comparison etc.See #2476 for where this is beginning to cause issues