Interest in supporting multiple backends?

[henryiii]

With a lot of new PEP 517 backends available now, would there be any interest in generalizing this to check arbitrary backends? While it might not be able to print nice instructions for all of them, it could still check to see if the git contents matched the SDist. This came up in the context of flit (See pypa/flit#522 (comment) for example). Just curious if it's something you've thought about or would be interested in, or would be possible. Didn't see "backend' in a quick search through issues.

[mgedmin]

I think it makes sense for check-manifest to be able to compare static manifests in various formats (MANIFEST.in, whatever properties flit needs in pyproject.toml) against version control system file lists.

I don't want to promise anything -- my time/enthusiasm is very limited these days.

I can see some steps I'd want to do first, like split the 1000+ line check_manifest.py into a package with a bunch of modules. The most worrying thing is that I'm not sure what to do with the check_manifest() function. It's already very long and complex.

[henryiii]

FYI, I built a tool around what I was thinking - it builds the SDist, then compares, without any knowledge about the specific build backend. https://github.com/henryiii/check-sdist This approach can't give suggestions about what to change, but isn't backend specific.

So maybe check-manifest still could slowly gain specific support for some of the most popular backends? Poetry, hatching, flit-core, and pdm-backend are the main ones.