Skip to content

Add support for Zarr V3 #417

@ChrisJohnNOAA

Description

@ChrisJohnNOAA

Currently we support zarr through the netcdf-java, specifically the cdf-zarr module.

We want to support reading and writing of zarr v3. There's a library from the zarr community that I believe supports this:
https://github.com/zarr-developers/zarr-java

I believe the api is very similar to the current netcdf-java implementation.
The main locations that interact with zarr files currently are:
https://github.com/ERDDAP/erddap/blob/3e00aa716e42f0164173951f3bf27efee19fe80d/WEB-INF/classes/gov/noaa/pfel/erddap/dataset/EDDGridFromFiles.java
https://github.com/ERDDAP/erddap/blob/3e00aa716e42f0164173951f3bf27efee19fe80d/WEB-INF/classes/gov/noaa/pfel/erddap/dataset/EDDTableFromFiles.java
https://github.com/ERDDAP/erddap/blob/3e00aa716e42f0164173951f3bf27efee19fe80d/WEB-INF/classes/gov/noaa/pfel/coastwatch/util/FileVisitorDNLS.java

Zarr support was added into the EDD*FromFiles dataset types because it was using the existing netcdf calls to load the files. Given Zarr's differences (what's a system level file for other data types is a folder in zarr) and moving to a new library for reading the files, it might make sense to have EDD*FromZarrFiles as new dataset types.

If we do this, we should also look into how we can assist admins in migrating from the old EDD*FromFiles datasets to the new EDD*FromZarrFiles.

Metadata

Metadata

Assignees

No one assigned

    Labels

    GSoCThis project is suitable for a Google Summer of Code intern.enhancement

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions