By that logic, why even spend effort migrating from a known-working implementation to one which is known to have outstanding bugs that there isn't enough bandwidth to fix?