Dependency autocurator is not maintained, only supports Linux and Python <=3.8

[chengzhuzhang]

The E3SM team is using esg-publisher for data publication to ESGF. However, we found it that there is conflict in maintaining esg-publisher dependencies with E3SM's own libraries in one conda environment. The current resolution is to maintain a separate environment for data publication. The main problem is that autocurator is not maintained, only supports Linux and Python <=3.8, details are documented by @tomvothecoder here. E3SM-Project/datasm#143. I'm posting here to raise awareness and hopefully we can find a solution. Thank you!

[sashakames]

@chengzhuzhang @tomvothecoder Have you upgraded to v5.2.0 yet? This supports xarray.
See https://esg-publisher.readthedocs.io

[tomvothecoder]

Thank you Sasha, that makes sense to me. E3SM is fortunate to have a few people who can and is willing to manage conda-forge feedstocks.

No big deal on the documentation since it still works fine.

[sashakames]

If conda-forge would much preferred method for you and there's time on your/E3SM to create the feedstock, I'd be very happy to accept the PR for the esg-publisher side.

[tomvothecoder]

The pip solution works fine for the datasm repo right now. If we ever need esgcet on conda-forge and it is not available on there by then, it is a good opportunity for @TonyB9000 to try creating and maintaining a feedstock repo (he's the main user of esgcet).

[TonyB9000]

@tomvothecoder @sashakames Agreed! I know that Tom has provided extensive instructions on this - but until one actually works through the process - it remains mysterious.

("he's the main user of esgcet" - I had no idea!)
("I opened a PR in the datasm repo for @TonyB9000 to start experimenting." - ???)

Bad git-practice on my part - I process repo almost entirely from the acme1-console side, and only visit the git web stuff when I PR.

[tomvothecoder]

("he's the main user of esgcet" - I had no idea!)

I should have been more specific and mean from the E3SM Publication Team (you, me and Jill).

("I opened a PR in the datasm repo for @TonyB9000 to start experimenting." - ???)

I hyperlinked the PR in that original comment.