General tip about AoC problems: a favorite “part 2” tactic is to expand the problem in a way that makes inefficient solutions intractable. For instance, the solution for my day 5 part 2 was ~15 million so locations would likely be LARGE

There’s another one later on this year where the straightforward “generate and count” approach fails even more spectacularly - with my input, a counting approach would have taken quite a while to count to 15 trillion!

IMO this is the best part of AoC, since usually it takes months / years of “production” use to find out an algorithm is inefficient.

You can usually get some benefit from changing Enum to Stream calls as the Stream functions are lazy and do not allocate all that memory up front with each call.

Yeah when the ranges are massive even lazy evaluation is going to eat up memory. Like others have said, thinking in terms other than evaluating each item is key for this sort of problem.