Archive extension

[kmodali]

Dear All,
I have '*.tar' files as assets and hence using the archive extension specification to create my catalog.
Subsequently I added the archive extension implementation in 'pystac' and 'xpystac' . Then I could convert the '*.tar' file from the asset object into 'xarray' dataset , as below:
archive_asset=item.assets['tape1']
archive_asset.to_dict()
{'href': 'test_1.tar',
'type': 'application/x-tar',
'archive:format': 'application/x-tar',
'archive:href': 'exp_1.zarr/',
'archive:type': 'application/vnd+zarr'}

ds_tar=xr.open_dataset(archive_asset,engine="stac")

Now, I have the scenario where the catalog search, results in a list of items and each item has '*.tar' file as asset. I would be thankful if someone could clarify the archive extension application in such scenario. Any other idea(s), suggestion(s) in this direction is kindly solicited.
Thank you.

[jsignell]

xpystac relies on the type to determine which xarray backend to use. So it doesn't know what to do with 'application/x-tar'. Do you know if xarray has a reader backend for archived zarr files? If so then it would just be a few lines of code to map from the type to the corresponding backend.

Stepping back a bit, it seems like an anti-pattern to create archives of zarr files. The more cloud-friendly approach would be to enable per-chunk compression.