Got some great feedback on <https://racket.discourse.group/t/racket-set-performance/479?u=badkins|this Discourse thread re: Racket set performance> - which brought the time for a Racket port of a Python program down from 10 seconds to 4.1 seconds, and eventually to 1.38 seconds which is just slightly faster than the Python version. However, the Python version is using a tuple as the set key, and my last tweak used a number as the key which allowed switching from hasheqv to hasheq, so I’m not sure it’s entirely fair :) I was quite surprised at the difference between equal?, eqv? and eq? for hash/set performance.