My day 17: https://github.com/jorispz/aoc-2019/blob/master/src/commonMain/kotlin/day17/p17.kt I went low-tech and did part 2 by hand using Excel, given the small world map it seemed a lot quicker than creating an algorithm

Hey guys, I'm back on day 7, part 1 (yeah, a bit behind :-)) -- I was happy because (seemingly) my existing computer from day 5 was sufficient for day 7 without any changes. However, upon testing, I see my code works on the specific phases given, but as far as I can tell those are not the max thruster values of those programs. Can someone confirm for me what the output of program

3,15,3,16,1002,16,10,16,1,16,15,15,4,15,99,0,0

should be with phases

4,3,2,2,0

? It should be less than

43210

according to the text, but I get

43220

.

My solution, based on coroutines and Flow: https://github.com/jakubgwozdz/AdventOfCode/blob/master/src/advent2019/day17/Route.kt most difficult part was algorithmically finding the routine and functions A, B, C. I did it by testing all possible subpaths and adding some optimizations: • function A is 1 up to 5 first elements of path • function C is 1 up to 5 last elements of path • function B is any 0 up to 5 elements of path • main routine is 10 calls long, starts with A, ends with C, and has 0 to 8 calls of A,B,C inside And then I perform a loop that takes every combination of A,B,C and main and tests if it results in previously found scaffolding route I wonder if I could do it less brutey forcey

For day 7, part 2, I used coroutines with `Channel`s to send I/O around the amps. I have one coroutine for each amp, as well as the system I/O (initial inputs, amp E output). However, I get indeterministic results if I run the coroutines multi-threaded -- using a single-threaded dispatcher works fine. Did anyone else have a similar approach and run into this?

I lost 50 points due to my compiler freezing twice

I'm back from traveling and managed to do Day 19. I'll catch up on 15 through 18 over the weekend. 🙂

I had a mistake in my input for Day20. One portal has 4 points instead of two. To earn the star I had to find a mistake and correct the input manually 😞

Wow, I don't get Day 20, Part 2 at all.

GRRRR. I walked my dogs and came up with the answer - my input.txt was corrupted! Redownloading and running again worked fine. No idea how that happened.

Went back and finished Day 17 instead of 22P2. I just went manual (drew my maze, figured out the path, programmed it by hand). I guess there's probably a way to generalize that, but I'm probably not going to bother considering that one was 5 days ago and nobody is going to read that post anyway. 😄

Finally managed to finish Day 22. Had to learn some new material to do it myself. It was quite an interesting experience. Some spoilers in the thread. https://github.com/Rtchaik/AoC2019/blob/master/src/advent2019/day22/solution22.kt

Day24: https://github.com/Rtchaik/AoC2019/blob/master/src/advent2019/day24/solution24.kt

Day 25 was done partly manually and the last stage automated: https://github.com/Rtchaik/AoC2019/blob/master/src/advent2019/day25/solution25.kt

Ok I might be a week or something late to the party, but I wanted to see if JVM solutions can be easily ported to the browser, and lo and behold, they can. Almost copy-paste 🙂 here, enjoy my awesome frontend skills at https://jakubgwozdz.github.io/JSTry104/cryostasis.html (last puzzle partial solution)

I said last year at the end that I wasn't going to blog about my solutions again, but I probably will. Last year I was in temporary housing, so my work setup was essentially a TV tray and a crappy chair in the side of my bedroom, so not ideal. And I had some travel so I fell behind. But this year, I'm finally in a house with a dedicated office (in Raleigh, NC, if anybody is close!) and since covid is here, I won't be going anywhere. So I guess why not? I probably won't change much about how I do it - solve the puzzle, optimize for clarity instead of speed, attempt to explain it all with a blog post, ideally day-of.

I can't include the text of the problem in my posts any more, I got called out on Twitter by one of the AoC people (which I thought was rude, they could have emailed me and I would have removed them all). I think that's a bit of a bummer, it makes the solution harder to connect to the problem, but I can't do much about that as I don't want to get into a "fair use" argument with them.

I promised myself I would branch out this year and use AoC to learn some other language, but doing it in Kotlin is just too attractive.... So once more I will be attempting to do AoC 2020 in Kotlin common code, and run my solution on the JVM, in Node, and using a native executable on Windows. For no particular reason I'll be using GraalVM for running the JVM part, and will probably see what happens when I run the JS version in Graal as well.

Does anyone remember how the two different ordering algorithms work in the leaderboard? Or at least how they worked last year 🤔

2^-1 🔥 (We will never reach the release if we continue like this)

My day 1 solution in common code: https://github.com/jorispz/aoc-2020/blob/master/src/commonMain/kotlin/day1/p01.kt I had a lot of frustrating infrastructure issues. My existing IntelliJ run configuration somehow got corrupted, which took me ages to figure out. Then, my native build didn't work, exited quietly without throwing any kind of error or stacktrace, just an exit code which hinted at a null pointer. The solution isn't particularly nice or clean, but it's been enough for today 🙂 Below the timing measurements for JVM (OpenJDK11), NodeJS and native. (The times were drastically improved by some preprocessing on the input which you can find on line 9 in the source)

That was a fun start. I'll post mine later, but it's a nested set of

mapNotNull

and

first()

calls. 🙂

[Thread] Links to my solution and post.

My 1ms solution (plus a dumb and slightly less dumb): https://github.com/bjonnh/advent_of_code/blob/master/src/main/kotlin/y2020/day01/main.kt

ok I can wait

319th on part 1, 177th on part 2. code in thread:

how’d you solve it?

My day 2 solution here, timings per platform below. Nothing particularly fancy, quite pleased with how extension functions help with readability.

| JVM (OpenJDK 11) | 9.1 ±  14.7  | `71, 16, 11,  7,  7,  6,  5,  5,  5,  8,  4,  5,  2,  3,  3,  4,  2,  2,  4,  4` |
| Node JS          | 17.4 ± 18.6  | `90, 43, 25, 20, 12, 10,  8,  8,  9,  8,  9,  8,  8,  8, 19, 11, 11, 11,  9, 11` |
| Native           | 39.1 ±  7.5  | `36, 36, 37, 34, 34, 36, 34, 45, 41, 68, 34, 42, 36, 35, 42, 34, 38, 35, 41, 34` |

aaand - back to day1 - I believe I managed to do O(n^2) in part2 and kind-of O(n) in part1 (kind-of, because it requires input to be sorted, so O(nlogn) effectively). testable version on https://jakubgwozdz.github.io/advent-of-code-2020/day01/ - I wonder if I missed any edge case

Does anyone know of a resource for generating/getting very large valid puzzle inputs?

I try to avoid regexps due to their lower performance