Advent Of Code 2023

[Piet Souris]

It is time again for the brainteasers. Enjoy!

[Mike Simmons]

Looking forward to it, thanks for the reminder!

For those not familiar with it, here's a link: https://adventofcode.com/

Lots of fun puzzles, nothing to see there for 2023 right now, but you can check out last year (or others) for more info: https://adventofcode.com/2022

There's a leaderboard or two specific to this board, that you can be added to if you ask here.

[Tim Cooke]

Day one was trickier than I was expecting. Went round a couple of failed approaches before getting it. Overlapping words were problematic.

[Junilu Lacar]

Completed Day 1 & 2 in ugly Kotlin. Must brush up on my collections operations. Part 1 on both days were pretty straightforward but the Part 2s were a little rougher than I expected.

[Tim Cooke]

I haven't had time to tackle day 2 at all yet. Looks ok just some fiddly string parsing as is the aoc way.

[Piet Souris]

Day 2 was straightforward, with the only problem (well, problem) to make something nice out of the input. Day 1B was not fully specified: should "twone" give us 22, 21 or 11? I took 21 and that worked.

[Tim Cooke]

It depends on whether you're looking for the first or the last number.

[Liutauras Vilda]

I did not understand what was done with Day6 part2.

Seeing description I was almost sure it will be one of those cases "it would run forever" unless you greatly optimise it.
I was feeling silly even trying brute-force approach, but couldn't resist - and it wasn't the case.

So far the easiest one.

[Liutauras Vilda]

I did a bit of catching up... and I am still 2 days behind (Day 10 and Day 11).

This year problems are a bit harder than previous I'd say. So takes a bit less writing and more thinking.

[Junilu Lacar]

I'm way behind: only starting Day 8 on this 18th day of Advent. But I'm learning a lot and having fun with Kotlin again. I think Day 7 has been my best solution so far. The Part 1s are pretty easy but the Part 2s have turned out to be harder that I thought they would be at first. Part of it is that I'm trying to fit Part 2 solutions in with the Part 1 solutions as best I can. I'm always looking to extract and abstract. Interestingly, there's also quite a bit of inlining and combining. It's great practice, plus the problems are fun.

[Stephan van Hulst]

I had to quit very early this year due to lots of stuff going on in my private life. I'm still intending to get on with it, but I probably won't have time for it in December.

A lot of the puzzles use the first part to put you on the wrong track: They kind of lead you in a direction that won't work for the second part, even with some tweaks.

My strategy is usually to write a very quick and dirty solution for the first part, then start from scratch to solve the second part. When I've solved the second part, I like to refactor it a bit so that the first part is a special case of the general solution.

[Junilu Lacar]

Day 9 was interesting in terms of learning. Very satisfying, too.

After reading the problem, I knew I'd be writing a recursive solution. Sure enough, I got an initial one to work with the examples pretty quickly. But when I ran it against my puzzle data, I got the wrong answer. Too high, apparently. How nice. How the heck do I find which data is screwing with my program? Good thing I thought of a test that helped me find my bug. It was kind of subtle because I didn't think of the nuance until I saw it in the test.

Tip: Just because the sum of all elements is zero, it doesn't mean all elements are zero.

When I realized what the problem was, it was a quick fix to the lambda expression. Part 1 solved.

Part 2 was ridiculously easy to solve, given the way I had set up the semantics of the program. Check this out:

fun part1() = histories.sumOf { it.extrapolateNext() }

// this is literally the only thing I had to do -- no change to any of the implementation code
fun part2() = histories.sumOf { it.reversed().extrapolateNext() }

I spent more time writing the tests for it, but not a whole lot more. Done in less than ten minutes.

Then the challenge was to find a way to use tail recursion. Took me longer than it should have but I eventually got that, too. Tail-call recursion isn't the most intuitive thing for me. Need more practice.

Overall, I learned quite a few things with the Day 9 problem. I doubt I can get caught up before Christmas Eve though. Lots of ground to make up. Anyway, it already snowed last night where I am, so I guess most of the people doing AoC got the big snow-making machine in the sky back up and running again.