Agreed, it's not a problem of theoretical possibility. It's just that the maintainers' time can be better spent improving Python 3 than maintaining a Python 2 compatible runtime.