Jeremy Maitin-Shepard
Jeremy Maitin-Shepard
Currently, zarr v3 requires that the `fill_value` is a single scalar that applies to all elements of the array. One simple extension would be to allow the fill value to...
In the ZEP meeting on 2023-03-02, @jstriebel, @normanrz and I discussed representing sharding (ZEP 2) in the metadata as a codec rather than a storage transformer: ```json {"chunk_grid": { "name":...
It is convenient to be able to reference an array (or group or attribute) with a single string. Existing zarr v2 implementations have invented various URL syntaxes. E.g. in Neuroglancer:...
Saving this to fix after the in-flight PRs are landed: Currently the description for each array/group metadata field is indented, which makes it display as a block quote in the...
Note: I don't intend this to be given any consideration before v3 of the spec is published, I am just posting it now to keep from of it. Currently we...
In both v2 and v3 specs, attributes are stored as JSON which means they are subject to the limitations of JSON: - Infinity and NaN cannot be represented, although currently...
bfloat16 is very widely used for machine learning, much more widely used than IEEE 754-2008 float16. bfloat16 is not natively supported by NumPy, but TensorStore, JAX, and TensorFlow all add...
I would love to have support in zarr (e.g. spec version 3) for a non-zero origin, i.e. an `origin` field in the metadata file to specify a non-zero origin. In...
The current specification allows `"C"` and `"F"`. But in some cases the optimal memory layout may not match the most natural dimension order, e.g. you might want the dimensions to...
This change requires nerdfonts v2.3 or later. Conversion was done based on this table: https://ohmyposh.dev/codepoints.csv