Day 11: Monkey in the Middle
Nim: Part 1 (00:48:54, rank 4533), Part 2 (01:07:15, rank 3101)
I stuck with Nim again since I'm getting more comfortable with it. This time, parsing took quite a while as I wasn't familiar with the strscans module and ended up with some quite hacky code that I was able to simplify a bit afterward.
While testing the program out, before keeping track of the number of inspections, I ran into quite a weird issue: I was iterating over the mutable var monkeys: seq[Monkey] with a for m in monkeys:, and updating each monkey's held sequence. However, somehow those updates were sort of lost at the start of the next round. I still don't really understand why, but the intermediate fix was to switch the outer loop to for i in monkeys.len: var m = monkeys[i] indexing. However, Michal Maršálek on the Code Golf Discord gave me a couple of great tips, including that the mitems iterator can be used instead of the (implicit) items one, if you need mutable access to each item.
I'm still not 100% sure what was going on earlier but no doubt it will arise again. I should probably read the Nim manual more carefully and try to understand some of the finer points -- I've never used the ref feature for example, so there's probably some really important stuff I'm missing.
That said, the program still runs really fast.
As knew upon reading the part 2 description that some kind of clever modulo trick was probably needed, but just couldn't figure it out for nearly 20 mins. Eventually I realised that there were a finite number of monkeys, each with its own modulo value, so if I just multiplied all of them and kept taking the modulo of that when passing a "worry" value around, it would produce the correct behaviour.
(none yet)
Again, not a great result but I'm pretty happy with my performance today. Just need to learn a bit more about Nim and... well, think faster and better I guess. The scanf thing was annoying, but I eventually understood what the doc says about $* and $+:
Matches until the token following the $+ was found.
I thought this meant that $+ should be followed by another matcher which you expect to be repeated, but that's not it at all -- it literally means "match ANY characters UNTIL one that matches the next matcher or end-of-line".
This meant I could get rid of my "call scanf multiple times until one of them succeeds" code and just call it once with a single $+ at the end to match either "old" or an integer.
I can foresee this being important in future, since there's no way to match e.g. a list of integers with the default matchers, so instead we need to match all of it and manually split.
This one is slightly more computationally expensive than previous days, but not by much, coming in at 25.2 ms mean runtime for Nim and a frugal 1664 kb peak memory usage.
Benchmark 1: ./part2_nim < input
Time (mean ± σ): 25.2 ms ± 2.8 ms [User: 24.8 ms, System: 0.9 ms]
Range (min … max): 21.9 ms … 32.8 ms 102 runs
| Program | Compile time (s) | Mean runtime (ms) | Max RSS (kb) | Source bytes | Source gzipped |
|---|---|---|---|---|---|
| part2_crystal | |||||
| part2_nim |