Optimising the backrefs membership operation resulted in a 11x speedup in benchmarks, but that seems irrelevant as it was an O(n^(n - 1)) operation (IIRC, I am too lazy to figure this out again), but is now O(1) where 'n' is the number of immutable objects.
Optimising the backrefs membership operation resulted in a 11x speedup in benchmarks, but that seems irrelevant as it was an O(n^(n - 1)) operation (IIRC, I am too lazy to figure this out again), but is now O(1) where 'n' is the number of immutable objects.