Akshay Subramaniam

There's another issue with GPU support and numcodecs that can potentially be solved with an interface change that I'd like to bring up. Typically, we would want to schedule work...

There are two issues with having the size be part of the chunk header (either like in the LZ4 codec or like in zstd): 1. Not all codecs currently support...

Has there been any development on this front? Being able to move type bound procedures into the python class would be very useful!

I'm running into an issue using `tag = "v1.1.2"` where the commands are not found. This used to work for me with `v1.0.0` but reverting to that now doesn't seem...

Would propose renaming this issue to `Generalized NDArray support` to disambiguate between zarr arrays and the underlying ndarray datastructures used. ## Proposal for adding generalized ndarray support to zarr -...

Thanks for the feedback @normanrz. I think we should be able to do this in pure python but would need to add cupy as a dependency to zarr. That can...

@madsbk The output-by-caller policy is definitely very useful especially from a copy optimization and a memory allocator standpoint. The one con that you called out is where I am stuck...

@jhamman You're right, `get_partial_values` should work just fine in theory, but there are a couple of issues: 1. The current batched codec pipeline doesn't use `get_partial_values` and dispatches the `get`...