One probably could contribute a few remarks to it, for instance recommending statically linking dependencies into a single uniquely identifiable shared library to help prevent undefined behaviors at runtime (when symbols are being resolved when their shared libraries are `dlopened`).

A good reference might be `pypackaging-native`, which was contributed by maintainers in the Scientific Python ecosystem and hardware vendors mostly.

See for instance: pypackaging-native.github.io/key-issues/n...

Depending on the project, you might also consider distributing it on conda-forge (the entire chain of dependencies is managed well and executables can be distributed separately if needed).

Have you considered using `auditwheel`?
github.com/pypa/auditwh...

How about competing on achieving fewer but better objectives?

Is something preventing you from adopting another mindset if this one does not suit you?

I guess its depends on the kind of opportunities one is given; it might be beneficial if it interests you keeping you in the flow, but if it ends up bothering you more than it stimulates you then accepting might be pointless.

For some, joy sparks from focusing qualitatively on fewer topics.