Josh Moore

Oh dear. This may end up being something of a can-of-worms. All of the versions in `requirements_dev_optional.txt` are pinned, and I imagine they haven't been updated since appveyor was dropped...

Ok. I've activated dependabot (https://github.com/zarr-developers/zarr-python/pull/734) which will have us testing with newer versions, but is there any way to give guidance to users on these dependencies, @martindurant?

@MatthewBM are things working again?

I'm leaning towards closing this based on https://github.com/zarr-developers/zarr-python/pull/734 since we should be able to detect when a new version of fsspec* breaks things. But that won't necessarily help users to...

> but I have no idea about the implications of this statement. Ditto. Naively, I assume we'd need to look at a throughput metric. Are either of you up for...

> Maybe this is acceptable for a test? If it's showing any failing behavior, then yeah, it would make sense to get that into a PR so we can start...

Any chance of getting the codecov back up, @d-v-b ?

So looking back: the goal of this issue would be roughly equivalent to making FSStore the default?

> looks like CI is failing for python 3.7 due to some type annotation stuff... do we need to support 3.7 still? - Happy to hear opinions but depending on...

> _ensure_store methods of [BaseStore](https://github.com/zarr-developers/zarr-python/blob/26ff3ca554cad3c38bd831692ab4be579c77a4cb/zarr/_storage/store.py#L93) and [StoreV3](https://github.com/zarr-developers/zarr-python/blob/26ff3ca554cad3c38bd831692ab4be579c77a4cb/zarr/_storage/store.py#L267) in _storage/store.py can return None My guess is that they should throw on `None`. > https://github.com/zarr-developers/zarr-python/blob/main/zarr/storage.py#L787 Extracting the entire contents of `__init__` out...