The real issue with the Python scheduler was its algorithmic complexity. Using a faster implementation would have bought us a few extra months or maybe even year, but it would only have postponed the need for a real solution.