Fork: `xarray-schema` → `xarrera`. Maintaining an updated & functional version.

[javgat]

Hi everyone,

I’m reaching out because I’ve forked xarray-schema and am continuing development under the name xarrera (github.com/javgat/xarrera). The original project has been inactive for a while and is becoming hard to use due to outdated dependencies (e.g., pkg_resources) and unfinished features.

I know there was talk about eventually integrating this kind of validation into pandera, but that’s not yet available, and in the meantime having a working, up‑to‑date tool is valuable for those of us using xarray.

What I’ve done:

• Forked xarray-schema including the changes made after xarray-schema v0.0.3 release
• Set xarrera with updated dependencies and a clean state.
• Finished CoordsSchema validation while maintaining dimension consistency, based on the information on issues and pull requests.
• Made the package available on PyPI and conda‑forge.
• All changes are stated at the CHANGELOG.md.
• Referenced original authors and contributors in the README and in the CHANGELOG. I’ll be happy to reference specific contributions more explicitly if desired.

My plan:
Keep the project simple, functional, and updated. I’m not aiming to add heavy new features, but I’d like to:

• Maintain compatibility with modern Python and xarray.
• Merge small, useful improvements that the community wants.
• Eventually, when pandera (or another project) covers this functionality, we can deprecate or archive xarrera.

I’d love your input:

• Would anyone be interested in co‑maintaining or moving the project to a community organization?
• What do you think about the roadmap? Any features you’d really like to see preserved or added?
• If you opened an issue or PR in xarray-schema that hasn’t been addressed, I’m happy to look at porting it over.

Thanks to everyone who worked on xarray-schema (especially @jhamman, @andersy005, @paiforsyth, and others). I hope this fork can keep the spirit alive until a more integrated solution appears.

Cheers,
Javier (@javgat)

[javgat]

Mentioning people that may be interested:
@jhamman, @andersy005, @paiforsyth, @orianac, @bmwilly, @quentinblampey, @afrendeiro, @jbweston, @jsteward-orion, @strixcuriosus, @juliet29, @AdamTheisen, @jbusecke, @giovp, @ljstrnadiii, @kmuehlbauer, @garciampred, @rsokl, @calgray

[bmwilly]

@javgat note that pandera support is impending unionai-oss/pandera#705 (comment)

[javgat]

@bmwilly I see. Thanks for letting me know! It looks more developed than I thought. I'll switch to pandera when xarray support is ready. I see that it will require Python>=3.10 and xarray>=2024.10, which is fine, but it might be useful to have an up-to-date and functional (although simpler) version for Python 3.9 and older xarray versions. (At least, it is for me).

[calgray]

The pandera feature looks promising, quite glad schemas for xarray is getting much needed attention.

My main concern of introducing xarray into pandera would be the difficultly in supporting backwards compatibility for multiple pandas and xarray breaking changes. Not impossible, just hope this doesn't mean pandera is regularly delayed in supporting a range of new and old xarray releases due to it not being a primary interest to pandera maintainers.

I'd be happy though for xarrera to strive for a bleeding edge of xarray and schema features, and pandera there to define what it means for the xarrera API to be pandera-like/inspired.