This is a word game. You can't define away the problem of backwards compatibility by saying "well it wasn't us who changed in a backwards incompatible way, it was the underlying language we subset".

You especially can't do this when the parent made the claim that PNaCL has exactly this issue.