Conversation
Test Results 2 files - 2 2 suites - 2 24s ⏱️ -32s For more details on these failures, see this check. Results for commit d74acfb. ± Comparison against base commit b73a3b0. ♻️ This comment has been updated with latest results. |
Code CoverageDetailsDiff against mainResults for commit: d74acfb Minimum allowed coverage is ♻️ This comment has been updated with latest results |
6b4ae98 to
ab0855e
Compare
|
Currently, we only have DanubeHis and no other provider supporting Zarr, so it is difficult to properly test this part. I will continue once we add another provider that supports Zarr. The Zarr endpoint still needs to be implemented, along with the related tests. |
|
I do not understand why fsspec, aiohttp and requests dependencies are needed here. |
This PR improves the way Zarr stores are exposed and accessed through the /data endpoint, allowing Zarr data to be accessed in a cloud-native way.
The endpoint
/data/{federation_backend}/{collection_id}/{item_id}/zarrexposes the Zarr store as an HTTP filesystem, follows the standard Zarr + fsspec access pattern. Clients can access the data remotely using common libraries such as zarr, xarray, or dask without downloading the full store.Exemple usage:
Human-readable index:
A dedicated endpoint:
/data/{federation_backend}/{collection_id}/{item_id}/zarr/indexreturns a JSON listing of the files contained in the Zarr store and their paths.File access inside the Zarr store:
Individual files can be accessed via:
/data/{federation_backend}/{collection_id}/{item_id}/zarr/{path_to_file}, where path_to_file can be found in the human-readable index.