<Advent of Code 2024 day 22> (no spoilers) :thread...

# advent-of-code

[Advent of Code 2023 day 22]

Advent of Code 2024 day 22 (no spoilers) 🧵

[Dan Fingal-Surma]

Works on the test input, wrong answer on the real input... 🧐 🧐 🧐 (Part 2). Gotta take a break for now

[bj0]

trying to think of ways to choose/limit sequences on p2

[Dan Fingal-Surma]

I'll say what I'm doing in the spoilers thread

[bj0]

i tried one thing but it didn't work

[Max Thiele]

Do you use Long as a result type?

[Dan Fingal-Surma]

The answer is the approach was sound but there was a bug that the test input didn't tickle.

[Dan Fingal-Surma]

I don't think you need to use Long as every step you mod by 2^24. However I started with Long so I have not verified for real.

[Max Thiele]

Yeah for the rng Int is fine but I initially also used Int to sum up the values, which works for the test input but not for the real one

[bj0]

got it!

[bj0]

i defaulted to Long, but didn't check if its necessary

[Dan Fingal-Surma]

Checked, Int can overflow on * 2048 = shl 11

[Dan Fingal-Surma]

because 24 + 11 > 31 🙂

[ephemient]

yeah but it doesn't matter because the overflow is irrelevant after the prune

[Jaap Beetstra]

What the h*** am I reading?

[Dan Fingal-Surma]

I get negative numbers as a result of the overflow

[Dan Fingal-Surma]

Ah because of % instead of mod

[Dan Fingal-Surma]

I get a performance penalty switching to Int/mod from Long/%