Yes, that's the principal idea, I think -- separating out the implementation details from the actual conditions that are necessary for correctness. IIRC, I've s

Yes, that's the principal idea, I think -- separating out the implementation details from the actual conditions that are necessary for correctness. IIRC, I've seen another presentation of this where the author used that freedom to implement a work-stealing quicksort across multiple threads.