Another implication of this is that its impossible for 2 mutually incompatible copies of the same package to exist in the same environment. If packageA needs nu

Another implication of this is that its impossible for 2 mutually incompatible copies of the same package to exist in the same environment. If packageA needs numpy 1.20 and packageB needs numpy 1.21, you're stuck.