I've sketched up an idea issue in
https://github.com/pantsbuild/pants/issues/9273 that proposes deprecating support for lists of interpreter constraints that are ORed in favor of using standard python packaging means of achieving the same ends. I'd love feedback on the idea mainly from the point of view of pain of conversion. Anything that can be expressed with our custom OR support can be expressed with standard `PythonRequires` metadata-style version specifiers, but converting from the one to the other will require a one-time pass over BUILD files that we likely can't fully automate. As such, getting feedback on how diverse the use of the OR feature is would be helpful.