advent-of-code

[vast jetty]

fucking amazing

ai is so NOT taking over anyone's job

[bright spindle]

Which model is that?

[vast jetty]

yeah apparently salt die has a range library

[vast jetty]

eh i dont have a subscription so whatever the normal one is

alright sure the good ones are pretty good

but this is such a stupid thing to say that i cant

MAJOR LOGIC BUG

[bright spindle]

I haven't used the free one in a long time, but the impression I've gotten from other people is that it hallucinates a lot more than the paid ones.

[lime fulcrum]

when did we have range splitting on another range in 2024 or 2023?

[lime fulcrum]

ok found it. it was 2023 day 5 part 2

[grand compass]

I keep getting pings from here for no reason bruh

[lone nexus]

WHO PINGED ME

[potent jacinth]

The guy saying that a new day of aoc is out.

A nerd.

[hollow wharf]

I fell as sleep

i had a headache

now its 5 in the morning

time to do it

[potent jacinth]

Are you in the U.S. and A.?

[hollow wharf]

A.?

[potent jacinth]

Yas.

[bright spindle]

The United States and America?

[potent jacinth]

Precisely.

[hollow wharf]

Canada

[potent jacinth]

Greatest Country in the World, after Kazakhstan.

[bright spindle]

So "Not the United States but America"

The NUSbA

[hollow wharf]

yeha

[potent jacinth]

https://tenor.com/view/coding-time-coding-gif-25155837

[hollow wharf]

first time getting one of these

tf happened here

[magic oxide]

not the $ before variable names...

[hollow wharf]

OKAY

[shy tulip]

I just froze my computer and had to hard reset it because i tried a naive solution to part 1 lol

[hollow wharf]

oh

lmao

love the speeds

[shy tulip]

I don't think I can even get an empty python file to run that fast

||```haskell
import Data.List (intersect)

parseRange :: [Char] -> [Int]
parseRange range =
let start = read $ takeWhile (/= '-') range
end = read $ tail $ dropWhile (/= '-') range
in [start .. end]

part1 :: [Int] -> [Int] -> Int
part1 ranges ingredients = length (ingredients intersect ranges)

main :: IO ()
main = do
ls <- lines <$> readFile "day5/input.txt"
let ranges = concatMap parseRange $ takeWhile (/= "") ls
let ingredients = map read $ tail $ dropWhile (/= "") ls
print $ part1 ranges ingredients```||
so simple and elegant, but absolutly kills the pc lol, it tries to load all ranges into memory

[hollow wharf]

got it

[hollow wharf]

tried to load all the ranges....

💀

[shy tulip]

It worked perfectly for the example input 💀

[hollow wharf]

haha!

[silent nymph]

in [start .. end]
🪦

[cyan sleet]

cash money

[tender swallow]

ironically the even more naive solution works just fine
||double for-loop O(nm)||

[hollow wharf]

huh?

it does?

[tender swallow]

||https://github.com/bharadwaj-raju/aoc-2025/blob/main/day5/1.py|| finishes in just 60ms

[hollow wharf]

lmaooo

[tender swallow]

it makes sense
||the number of ranges isn't really very large at all, it's just that their span is huge||

[magic oxide]

honestly don't feel like coding perl rn

gonna do it later

[cyan sleet]

last 3 days have been valuable

rubies, crystals, perls

[hollow wharf]

HAHAHA

[vast jetty]

Ayyyy

Nice

How long did they take

[hollow wharf]

i logged on around 6 so thats

25+ ish mins

[vast jetty]

Pretty gooddd

I took like 90 mins

Maybe 2 hours even

[hollow wharf]

was trivial as hell

[vast jetty]

For part 1

[hollow wharf]

lmao

[vast jetty]

I’m sorry binary search?

[hollow wharf]

thats what i used

for part 1

[vast jetty]

how

What

[hollow wharf]

@vast jetty https://discord.com/channels/267624335836053506/1446365601789509653

i got mine there

check it out

[bright spindle]

Oh, wait

wrong channel

[dense osprey]

is changing the input such that reading it is easier considered cheating?

[bright spindle]

No, why would it be?

[hollow wharf]

its not

u can do anything

with the input lmao

[dense osprey]

oh ok

i always tried avoiding it to make my program only use the "vanilla" input

[bright spindle]

Ah, ok

[hollow wharf]

part 2 was even more trivial

[vast jetty]

I mean if you feel it’s fine aide

sure

[bright spindle]

But you should spoiler your solution strategies in this channel

[hollow wharf]

yeah thats why i been deleting msgs

real quick let me do it

[vast jetty]

There are so many days where it’s literally soo easy to just

[bright spindle]

A word like ||binary search|| is a huge clue

[vast jetty]

Find replace

like a bunch of problems in 2022

[hollow wharf]

which days were even those oh

2022

lmao

[vast jetty]

Most of the regex problems we have

There was one with some weird ass format where in there’s the info

[hollow wharf]

true

[vast jetty]

I don’t quite remember what it was

[bright spindle]

Transforming data is just normal programming

[vast jetty]

But yeah

Personally I’ve never wanted to find replace since it feels against the spirit of the puzzle to ME

[hollow wharf]

also I think we are entering this algorithmic type problems now I can def see some recursive solutions coming up

[vast jetty]

That’s not to say it’s wrong tho

[hollow wharf]

in following days

[vast jetty]

There was this problem in

Hm

Yeah it was 2022 again I think

The file system one

[bright spindle]

I like AoC because you can often make it as easy or as hard for yourself as you like.

I don't think there are any canonical solutions

[vast jetty]

I saw solutions wherein people copied the input into their project manager

And it made the file system

Fair game

or 23 day 4(?)

The quadratic equation one

Wherein most people’s solution was pen and paper

Including mine

[vast jetty]

I mean

We have the weekend coming up

I’m anticipating path finding

[hollow wharf]

path finding?

graphs?

[vast jetty]

Or some other shit with BFS/DFS solutions

[bright spindle]

Jokes on the AoC puzzle guy, I have even less time for AoC on the weekend...

[vast jetty]

Yeah we often have those

[hollow wharf]

I would love those

[vast jetty]

Last year day 10 was the first path finding one I thin

I remember cuz I tried to write a BFS/DFS solution on my one

Couldn’t

Then gave up

And on a later day wrote djikstra’s

[hollow wharf]

shit

[vast jetty]

Like during the AoC but later

[hollow wharf]

I better get myself together and get in practice for it already I really wanna do on same day

[vast jetty]

Cuz I had to write djikstra’s for another day

Day 12 or smth

just copied code

Then day 19 was also pathfidnong

I couldn’t do p2 tho

Also one of these days had smth I wrote a flood fill for

16 or smth

[vast jetty]

Have you not done any other years

https://tenor.com/view/first-time-meme-movie-james-cameron-western-gif-17538858

[hollow wharf]

yes

[vast jetty]

Welcome to the club

[hollow wharf]

this also been smth

i been looking forward to each day

https://youtu.be/09_LlHjoEiY?si=oHwD01_kCPQkkqA7

https://www.youtube.com/watch?v=RBSGKlAvoiM

this is what i plan on doing

rn

after this walk

[vast jetty]

DSA is low-key pointless

I mean it’s nice to know

But kind of pointless for AoC

[hollow wharf]

i mean we been using them

[vast jetty]

Who cares that what you’re implementing is a heap

What you care is that you’re appending shit to a list and taking list[-1]

If anything look up common algorithms ig

[bright spindle]

How so?

[vast jetty]

Or honestly the best thing is just do more years

[hollow wharf]

nah i wanna learn like niche graph algorithms

just in case

[hollow wharf]

i don't have much time

[vast jetty]

I mean they’re even designed such that you don’t need formal dsa knowledge

[bright spindle]

I think at least in some cases there are DSA techniques that fit AoC problems well

[bright spindle]

No, not to solve them at all

[vast jetty]

As long as you can implement stuff

[bright spindle]

But sometimes you can make fast solutions with DSA

[hollow wharf]

yesss

[vast jetty]

Who cares if your data structure is a heap

After all in python you’re seeing that as a list anyway

Unless you use fancy data structures

[hollow summit]

if only i had knew day 3 part 2's solution was a ||"monotonic decreasing stack"|| i wouldve probably recognized it and solved it oneshot

[vast jetty]

Yeah I don’t know what that is

But I think that’s what I wrote

See what I mean?

[hollow summit]

exactly

[vast jetty]

also it’s funer like this

[hollow summit]

if you knew it beforehand you wouldve recognized it

[hollow wharf]

monotonic in a sense that its either keep on increasing or decreasing

[vast jetty]

Yeah but it’s not like you NEED them

That’s what I meant

[hollow summit]

obviously not, but it's useful to know

[vast jetty]

At the same time itd be better if you just saw some common algorithms people use

Or common things that you might not know about

Like people don’t know regex

Or fun language features that can make life ridiculously easy

Think itertools

Think cache decorator

[tender swallow]

that doesn't invalidate the performance advantages of a heap

[hollow wharf]

that makes it a suitable candidate for greedy approach

greedy was the first thing i did

as soon i saw it

[vast jetty]

Idk that’s not what it means in math

[dense osprey]

||did anyone else also use binary search for part 1 and therefore had to make the intervals disjoint (which makes part 2 trivial)?||

[vast jetty]

I think

Strictly non increasing or non decreasing

[hollow wharf]

it literally does mean the same, function keeps strictly increasing or decreasing

[vast jetty]

Gary apparently

[hollow wharf]

yh

i did that

[vast jetty]

No it’s non decreasing I think

I might be wrong tho

[hollow wharf]

literally this is what i did

making part 2 super trivial

lmaoooo

[vast jetty]

Yes

[dense osprey]

yeah

[tender swallow]

monotonic can refer to either strictly increasing or strictly decreasing
the specific direction isn't important, the point is that you stick to one

[hollow wharf]

yeah

[hollow summit]

data structures are more about how you use the memory

not what it is

[vast jetty]

My point is that you don’t need data strucure knowledge to write it yourself

[hollow summit]

and that's correct

[dense osprey]

||I was too lazy to make the intervals disjoint at first but that would have made the binary search so much more tedious and now Im glad I made them disjoint||

[hollow summit]

but to say DSA is pointless is just wrong

[hollow wharf]

i mean we use data structures all the times without knowing it, but would it be much better if we knew what we are using? yes

we can make better comments

[dense osprey]

dsa is the best part of cs

i love dsa

[vast jetty]

I guess I mean more like watching an 8 hour dsa video to prepare in the middle of AoC is kinda pointless lol

It takes a while to learn basics anyway

[hollow wharf]

isn't algorithms and optimization like the core of CS?

there is nothing else in it

[vast jetty]

Yes

[tender swallow]

oh yeah definitely
just learn what you need for each problem, that's way more fun and way more efficient learning-wise

[vast jetty]

Yes!

I def recommend solving past year problems and stuff

You always always learn smth

[dense osprey]

there are probably more efficient ways

[low condor]

I did some of 2015 this year

[vast jetty]

Yesterday I learnt that convolve is a thing

that’s cool

[low condor]

Was a very different year

[vast jetty]

Hesh

Yeah

Idk about fun tho

You can do leetcode if you hate your soul

[hollow wharf]

i don't mind going through long videos lmao also im gonna get some books on it as well

i will check local library

and online

I need to learn the best

[vast jetty]

Also Gary def learn regex

[hollow wharf]

True

[vast jetty]

I bet you we’ll have a reflex problem real soon

Regex*

[dense osprey]

i hate regex

[vast jetty]

I love regex

It’s so handy

[dense osprey]

no

[vast jetty]

It’s easy to learn

Promise

It’s so easy I’ve learnt it like 2,000 times

[vast jetty]

Wrote up a regex solution for day 2

It’s like 5 lines of code

[bright spindle]

Using DSA for AoC is useful in the sense that it helps you become more familiar with DSA techniques and make those techniques more accessible when you might need them in some other project.

Even if it's possible to solve the puzzles without them.

[vast jetty]

Last year I had the experience of like

I’m solving a problem

I do it

I go on Reddit

I see people talking about BFS vs DFS

and they’re like DFS is better

I’m like ok guys wtf is that

They explain it properly, itbeing the great community it is

im like ‘oh so what I did’

Same with the flood fill

It does feel like you’re more lost if you have noooo vocabulary to talk abojt what you’re doing

And it def takes longer

But I don’t think it mattered all that much

[bright spindle]

Also, some DSA is less obvious than others. Sure, you might implement DFS intuitively, but you're probably not gonna reinvent A* by accident.

[vast jetty]

Yeah def

[vast jetty]

Hence

Those are really handy

[naive wharf]

I'm writing a small AoC utility library, any recommendations for things to add?
I already have the "set of ranges", and 2D maps of integers/characters

[vast jetty]

2D maps?

I def recommend grid support

[naive wharf]

?

[vast jetty]

Grid parsing?

Like for day 4

Hmmm what else

[naive wharf]

that's what I'm calling 2D maps

[tender swallow]

what functionality does it provide for 2d maps?

[naive wharf]

still thinking about that. mostly just value/range set/get

[tender swallow]

get with default on out-of-bounds is extremely useful IMO

also stuff for working with neighbors and directions

[naive wharf]

oh it does that too

[naive wharf]

directions

[south musk]

part 1 today seems too easy, I'm scared of what part 2's gonna be lol

[hollow wharf]

Which day

[south musk]

day 5

today

I've solved part 1 now, part 2 doesn't seem that bad though

ok part 2 really can't be brute forced huh

[hidden musk]

just look at the ranges in the input lol

[shy tulip]

How much memory and time do you have?

[south musk]

like how much my solution takes?

[shy tulip]

No for the bruteforce 😛

[south musk]

memory, idk off the top of my head but my bruteforce doesn't take much, time... Rest of the day technically but that seems stupid

[bright spindle]

I think it's normally the case with AoC that even if the bruteforce solution doesn't work (in a reasonable timeframe), there is usually a relatively simple non-bruteforce solution that doesn't require any advanced algorithms to implement.

[cyan sleet]

if this brute force is actually stepping through all of the ranges end of the day is a slight underestimate

[white zealot]

What do you guys mean by brute force?

[cyan sleet]

a naive solution

[bright spindle]

An algorithmically simple solution that generally involves relying on computers being fast rather than being clever.

[cyan sleet]

the naive answer to "how many integers are in these ranges" of "go through all the integers and check if they are in a range" would be an example

[bright spindle]

Like, I dunno, finding primes by attempting to divide each integer with each smaller integer in turn.

[shy tulip]

Expand all ranges and count the unique numbers

[cyan sleet]

At least it's not death by old age that will get you in that case

unless you have a few petabytes of spare ram

[shy tulip]

It worked on the example input 😛

[hidden musk]

OS killed my script after about a minute. which was actually pretty surprising it took that long

[hardy relic]

I havent had time to do any aoc since day 2. I know I won't get stars but will it still tell me if the answers are correct?

[magic oxide]

you'll still get stars

there's no time limit on puzzles

you can go back and do past year's puzzles as well and get those stars

[humble copper]

day 5 is literally ||https://leetcode.com/problems/merge-intervals/||

[hardy relic]

oh ganksta

[cyan sleet]

It's a very standard problem

[humble copper]

yeahhhh

[rancid creek]

it's entirely possible that multiple puzzle sites will use the same general programming concept for a puzzle. 🙂

there was also 2023/day5

[frigid mural]

I was expecting smth of that level when I first saw ranges, part 2 really let me down

[rancid creek]

yeah, was surprised. probably going to be some harder ones coming up, maybe this weekend.

[hardy zenith]

why is there a bot that keeps pinging me?

obviously the new puzzles are posted at midnight. the notification spam is annoying.

[minor cave]

You opted into the Advent of Code role, which pings you when the puzzle releases. Not everyone is in the east coast US timezone, so some people find a ping helpful.

You can remove the role in #roles

[mossy basin]

you can also ignore the bot

[lusty rover]

https://tenor.com/view/special-ed-special-gif-14524557650897979460

[humble copper]

should i study for finals or do 2022 day 19

hmmmm

[potent jacinth]

https://tenor.com/view/horsepiv-gif-24660548

[ripe meteor]

have to catch up, got lazy yesterday

day 4 was pretty good

i hate grids

i spent some time troubleshooting and it turns out i padded the example and not the actual input 💀

all my homies hate grids

[low condor]

I had no idea which one 2022 day 19 was - and now I've looked it up

definitely do finals instead

no good can come from doing that day first

[low condor]

there's your problem ||don't pad the example - find the right data structure for the grid first||

[ripe meteor]

sounds like effort

[low condor]

way less effort than trying to deal with the problems that come afterwards

[humble copper]

its times like these where, with every bit of my heart, i wish i could very easily do per-element addition of tuples and lists

https://tenor.com/k0FYGwmsELk.gif

i see why i gave up the first time

[mossy basin]

import numpy as np

[low condor]

definitely go back to it again - when you have plenty of free time

it's a slog - but rewarding if you eventually solve it imo

[potent jacinth]

Tru

[humble copper]

a slow recursive solution actually worked for one line on the sample inputs.....

[low condor]

yeah - the sample input was never the problem on that puzzle lol

[rapid wolf]

I've had sample input correct answer wrong problems on every puzzle so far except day 1 IIRC.

[shy tulip]

Only today so far

And after seeing my mistake it was wild that the sample input worked

[low condor]

funny how that is - "sample input correct; wrong real input" has ONLY happened this year for me on day 1

[minor cave]

Rouletters, I am undecided on a language or two to add to the roulette. This is your chance for requests!

[vivid dove]

[lusty rover]

Hey, I’ve heard good things about Python

You should be mean and do something like FORTAN or COBOL

[low condor]

How about one of the days as Python 2 🤔

[hidden musk]

👁️. python 2 or 1

[buoyant perch]

I started late. Yesterday actually. I’m just staring problem #4.

[lusty rover]

It’s supposed to be esoteric
Not the language we use everyday at work

[hollow wharf]

I can suggest using racket

[lusty rover]

Haxe?

[hollow wharf]

Did u mean this as in trying to ask if that’s what I meant

Or that’s ur suggestion

[lusty rover]

Yes

[cyan sleet]

Have we done octave/matlab yet?

Some array language (Uiua/J/APL) could also be interesting, but might be too hard at this point

gnu bc is another weird-ass option
(yes, you can program in it, though idk if input parsing is doable, so might need some preprocessing)

there is R which I don't think we ever did

swift we never did I think

some sql

I wouldn't mind trying Odin

[lusty rover]

Isn’t that a website?

[lusty rover]

Speaking of single letters, D would be a good one

[cyan sleet]

it's been done

[lusty rover]

It has?

[cyan sleet]

last year

[lusty rover]

Oh

[cyan sleet]

https://odin-lang.org/

[cyan sleet]

also don't tempt Kat with fortran

(knowing Kat it will probably show up anyway)

[lusty rover]

Don’t pretend you wouldn’t love the challenge

[cyan sleet]

we've had that challenge twice 🙃

[lusty rover]

When?

[cyan sleet]

twice

[lusty rover]

Man my steel trap must be falling apart

[cyan sleet]

in 2 years

[lusty rover]

IDR

Show your code

[cyan sleet]

ye of little faith
https://github.com/algmyr/aoc-2023/blob/main/day10/main.F90
https://github.com/algmyr/aoc-2024/blob/main/day10/main.F90

[minor cave]

I am still very tempted but I feel like at least 5 of our Rouletters will crash out with fortran this late >_>

[lusty rover]

I belive you I just couldn’t remember

Ah yes

https://github.com/algmyr/aoc-2023/blob/main/day10/main.F90#L1

[marsh current]

day10/main.F90 line 1

module my_personal_aoc_hell```

[cyan sleet]

the order of languages this year has been kinda weird

ruby before crystal before perl feels so backwards

perl is proto ruby, crystal is almost ruby, ruby is

[lusty rover]

is what

[cyan sleet]

it just is

[shy tulip]

is that not a paid lang?

[cyan sleet]

octave isn't

[rapid wolf]

Flix could be interesting, tho it's mostly just a slightly cooked Haskell.

[shy tulip]

Mojo

[lusty rover]

The vaporware?

[shy tulip]

The python killler

[lusty rover]

Gotta love marketing

[cyan sleet]

I'm going to make myself a service and add a list of the languages to the readmes

oh fun, 2023 was the year I started using jj

[humble copper]

could anyone potentially give a hint for 2022 day 19 😅

normal ||dfs|| takes way too long and i can’t think of any optimizations to put on top of it

[cyan sleet]

I have some upper bound estimation in my code to reject things that are obviously bad

(I'm trying to remember the problem as I read my code :)

ah ok, so my upper bound thing is "what if I allowed doing all the actions at every turn"

and if the score from that isn't good enough I can just give up on this path

looking at some comments I did on the thread for that year

one pretty big optimization is that if you have enough robots of one kind, stop creating it

e.g. if you have more robots than you can even spend in one turn

[ivory vigil]

This is the comments in my code.

• ||stop early if we haven't built enough geode robots to reach our goal before time limit (highest seen so far)||
• ||stop early if we haven't built enough obsidian robots to reach our goal of geode robots before we need to switch to making only geode robots (as many as we need to hit our geode goal)||
• ||didn't do the stop early's exactly, just close enough to ensure super dead branches weren't explored. some dead branches still got through but not very many (eg assume we have resources to make a geode bot each minute, assume that we'll have the full output of the obsidian miners in the whole 30 minutes at the time we start building geode bots)||

(2022/19 hints on optimizing)

[cyan sleet]

I'm trying to unbreak my 2022 rust code to see how big impact some optimizations had

so the solution I have is ~10ms, without the upper bound estimate for pruning
without it is took ~2s

so ~200x speedup

I also set my thing up so that my search options skipped forward in time, no clue if that matters much for performance

i.e. my options were like

• If I wait, can I ever buy an geode bot? If so add the possibility of waiting until I can afford it and then buy it.
• If I wait, can I ever buy an obsidian bot? If so add the possibility of waiting until I can afford it and then buy it.
• ...
• Do nothing, jump to the end time.

I can imagine that cut down on the number of states I had to process, but idk how key it was to have something fast

The simple upper bound simulation helped a ton though

Like, it's very easy to write. Just simulate and ignore the fact that you can only buy one thing per turn

That way you get no branching, and you get a cheap upper bound on the best score you could reach

(good lord I had a lot of untracked stuff in my git repo at the time)

[feral hazel]

once you solve it, i remember ||https://www.youtu.be/5rb0vvJ7NCY|| being a good video on the problem

[cyan sleet]

interesting, this is using a different upper bound than I did

mine is a tighter bound, but their is faster

[hollow wharf]

are you guys ready for AOC

tonight

[half swan]

hm I might do the roulette afterward

[sullen adder]

the halfway point already

[shut trellis]

<@&518565788744024082> Good morning! Day 6 is ready to be attempted. View it online now at https://adventofcode.com/2025/day/6. Good luck!

[half swan]

OMG the person saying trash chute was right

@low condor #1446365601789509653 message

[limpid grail]

why am i jumping into the garbage chute 🥲

action packed decoration mission

can't even catch a break

oh hell naw 💀

[ivory vigil]

Did anyone else's cookies expire?

[pine tiger]

holy string parsing

[minor cave]

It depends on when you first logged into AoC after awhile this year. They should last ~4-6 weeks when they're freshly generated (which'll happen first time you visit the site after they expire/don't exist)

[ivory vigil]

I longed in 2 days before Dec 1 😄

[magic oxide]

advent of parsing the input

💀

[sullen adder]

i never want to deal with squid math again 😭

[half swan]

https://tenor.com/SxA4.gif

idek where I'm going wrong in p2

[cyan sleet]

@minor cave I'm impressed you managed to hit this task with C

[ivory vigil]

I didn't mind p2 too much

[cyan sleet]

in python?

[ivory vigil]

Yeah

[cyan sleet]

p2 is fine if I had sane string handling

but this is C

I'm in pain

[minor cave]

:)

[half swan]

(oops, forgot to hit reply, @cyan sleet )
hint: ||there is a consistent maximum amount of digits, both row-wise and column-wise, 4 for me, so if you wanted to you could waste a little bit of memory to make your life easier||

[hidden musk]

I was actually thinking that they knew

[cyan sleet]

I'm going to fix my input reading so I can actually run the same code for sample/input for p1 and p2

the number of lines differing between the cases is kinda cruel 😭

[hidden musk]

how did you know btw

[minor cave]

I didn't, roulette language I pick is entirely vibe based plus inputs from friends who have no context

[hidden musk]

uh huh 🕵️

[cyan sleet]

are we doing reverse language hardness this year?

[half swan]

okay found it

[cyan sleet]

my lsp thinks I'm writing C++

[magic oxide]

oh god

i have to reparse the input for p2?

[cyan sleet]

:)

[hidden musk]

:)

[cyan sleet]

I should try to be not dumb and separate the bulk of the parsing from computation, huh?

[magic oxide]

my parser for p1 is 70 lines...

[vast jetty]

@hollow wharf hows it go

i did part1

im trying to write regex to parse part 2

(count everything until a space)

but its hard apparently

i could just write it in python hm

idk if i should

[vast jetty]

like ||just count empty strings after a .split()||

[cyan sleet]

imma leak memory, and nobody can stop me

[vast jetty]

maybe i should and do regex later

oh never mind that doesnt even wwork anyway

alright then

[feral hazel]

got part 1 done in C
looks like i won't be pasting the code into the spoilers chat this time lol

sub 2k chars though

[hidden musk]

linux can handle it :)

[cyan sleet]

my part 1 was <60 lines

but I'm rewriting everything now

[vast jetty]

part 1 is really short no

[magic oxide]

hahahaha

[vast jetty]

oh yall are not doing python?

[feral hazel]

oo i could do something really silly with a macro

[feral hazel]

yep
different language every day
today is C
#aoc-announcements message

[vast jetty]

hmm

thats better

im so lost

[cyan sleet]

mood

[vast jetty]

i think i dont understand the problem

hm

[magic oxide]

address sanitizer time

[vast jetty]

the columns are separated by a single space?

the rest of the spaces are the left-right alignment?

how does one which space separates the columns in a given input

like

is it possible that we have overlap of columns or smth

[minor cave]

(for any further help, I'd suggest moving to #aoc-solution-hints)

[vast jetty]

or there will always be one column (in the input) of just spaces, and thats the sepator?

idk its not hints i just dont understand this problem lol

[vast jetty]

this is a fine assumption right

[minor cave]

Part of advent of code is understanding the input, I don't want to unintentionally spoil this for someone who hasn't attempted it yet

[vast jetty]

huh i dont understand the puzzle text

oh well

lemme try fucking about a bit more and if i dont get it ill go to the channel

[cyan sleet]

brain pls

bool flag = false;
for (...) {
  if (cond) flag = false;
}

[magic oxide]

very new library function

[buoyant perch]

I had to fight my IDE to make it stop deleting trailing spaces lol

I’m finally caught up. Just finished Day 6

[dense drum]

real 😭

lowk a fun problem

[cyan sleet]

ok, more general input parsing done

and p1 works still

now p2 should be a piece of cake

and done

[vast jetty]

fucking hell

so many for loops

and now i need MORE

sounds amazing

oh ffs why am i missing last col

alr minor fix

ok

holy shit fuck me

[half swan]

I am once again getting correct answer for the example but not the actual input 😭

[vast jetty]

seriously

its so annoying

im done and i think today im not gonna spend much time cleaning up code

i lowkey never wanna see this again

[potent jacinth]

LOL why y'all pythoners complain about a few spaces?

[vast jetty]

first of all my code today looks

[potent jacinth]

I haven't read it all. I want to express my sincere apologies for what I said.

[vast jetty]

which is absurd

usually it looks more wide

def not long

[potent jacinth]

My honest reaction.

[vast jetty]

its not horrible horrible

but its annoying

and also i didnt understand the problem text for the longest while

and thought i just need to ||find all characters until i hit a whitespace followed by a digit||

so i thought i can ||regex it||

until i realised i cant

like 30-40 minutess wasted?

hm

[potent jacinth]

Most sane Python developer be like.

[magic oxide]

inb4 all the functional languages are in the later days

[dense osprey]

i hate part 2

[low condor]

Looking at the graphic and that we’d be going down, I thought it rather likely 😂

[strange zinc]

augh, p2 is a nightmare. the solving is good. the parsing...

[low condor]

The parsing IS the problem

[vast jetty]

indeed

[strange zinc]

finally done with it

[hollow summit]

cephalopod math is weird

[low condor]

Advent of Code is weird

Day 7 looks kinda interesting as well

like some kind of showroom for the Monty Hall problem

Is the car behind door number 1 or 2 or is it just a goat

[magic oxide]

you make a good point

i just noticed that day 5 is like a restaurant area

[low condor]

yes and Day 4 has the rolls of paper

[magic oxide]

which fits with the day 5 theme

[low condor]

Day 1 has the snowman

etc

it's how I guessed that day 6 would have us going down the chute

[magic oxide]

cool details

[low condor]

Day 2 has the lobby - Day 3 shows all the offline elevators

and I just realised Day 4 has the massive printer as well

and it really is a big printer - it's like the size of an elevator

[half swan]

might be 3 incinerators

or 2 or 1

[low condor]

it's 3 of something. will be interesting to see what

[silent nymph]

Probably not a room though. Too small.

[low condor]

Could be a hallway of elevators

[magic oxide]

clearly it's this place
silksong spoilers

[potent jacinth]

Silksong sucks.

A rabbit possessed by Satan walking with a toothpick.

( I don't even know the gameplay )

[hollow wharf]

Is that the plot

Of the game

💀 💀

[potent jacinth]

I think so.

Look at the pictures.

There is always that satanic creature.

[hollow wharf]

Def has a toothpick

[vast jetty]

yes

shes on a quest to find the person who stole her carrots

[potent jacinth]

I have a few carrots in my fridge.

[modern terrace]

how interesting

[bright spindle]

Nope.

[naive wharf]

yay, I got to use my charmap type today

I have a theory that it's gonna get a lot harder tomorrow

[hollow wharf]

same

part 2 had me thinking kinda more than last days

[naive wharf]

we made it through the first half

[hollow wharf]

or may be i was just having a bad day

[hollow wharf]

what kinds of problems are consistent over the years

which haven't showed up yet

[naive wharf]

I haven't been following too closely

[hollow wharf]

the transpose thing we have to do today

peeps said its been like that every year

they is a transpose problem

[naive wharf]

I'll put it into the list of features for my library

[hollow wharf]

yes def

[naive wharf]

there is also always one where you get input as a long sentence

there's also the obligatory flood fill puzzle

[hollow wharf]

flood fill!

[ripe meteor]

is Day 7 the filter? any bets

[tender swallow]

<@&831776746206265384>

[worthy horizon]

!cleanban @earnest bridge

[marsh current]

:incoming_envelope: :ok_hand: applied ban to @earnest bridge permanently.

[ripe meteor]

I didn't mean literally 😭😭

[hidden musk]

kalshi for aoc topics

[hollow summit]

language roulette was C, then Zig... tomorrow HAS to be rust

[minor cave]

:)

[hidden musk]

🤨

[paper folio]

so what puzzles do y'all solve when it's not December?

[ivory vigil]

CodingQuest, Everyone Codes, Tom's Data Onion, Midges n Gnats, Codyssi

[paper folio]

wow I'l definitely check those out!

[ivory vigil]

Everyone Codes is the closest to AoC

[paper folio]

Everybody codes? Everyone Codes seems like an IT course in Vienna

[shut trellis]

<@&518565788744024082> Good morning! Day 7 is ready to be attempted. View it online now at https://adventofcode.com/2025/day/7. Good luck!

[waxen hound]

Danke Robbin!!!

[mortal bronze]

Wow... I haven't even finished the day 5 -_-

[limpid grail]

who sees a teleportation pad and just steps on it like that 😭

[lone nexus]

good noon

it's 12 PM here

[tender swallow]

ok this looks like a more interesting problem
nice

[pine tiger]

lowk choked but

allg

anyways

agnes tachnyon mentioned!!!!

[lone nexus]

agreed

[covert arrow]

aaaa I submitted the example input by accident

[pine tiger]

my condolences

[lone nexus]

this is interesting indeed

[hollow wharf]

its that time of the day

[earnest forge]

I don't understand how I'm meant to be counting beam splits

[lone nexus]

12PM here lol

[hollow wharf]

12 am

[lone nexus]

bro's cooked

[hollow wharf]

for me

[pine tiger]

i think it's ||the amount of times a ^ is hit is all||

[lone nexus]

12 hour difference eh

[hollow wharf]

is it like 12 pm like rn

[lone nexus]

well

[lone nexus]

12PM i mean by noon

[earnest forge]

Nvm I was double-counting lol

[lone nexus]

which yall should also agree

[hollow wharf]

we are literally 12 hrs apart then lmao

[lone nexus]

lmao

[lone nexus]

go to sleep 😭

[vast jetty]

shit i woke up late

ill do this in like

half an hour maybe

[hollow wharf]

day is just beginning i was talking to someone i like and now i do advent of code

🗿

[somber trench]

is there a way to connect my program input directly to where the input is stored on the website?

[earnest forge]

God damnit I did basically P2 first

[lone nexus]

🗿

[vast jetty]

sure

[hollow wharf]

there might be some extension for this

or u can create one

[vast jetty]

write an html parser

[vast jetty]

gary catch up

[hollow wharf]

If i make an extension will you guys use it

[vast jetty]

my turn to oversleep

[hollow wharf]

LOL

[vast jetty]

i forgot my alarm

[hollow wharf]

alright imma read the problem statement now

[lone nexus]

💀

alr

aight

i'll do this later

[vast jetty]

the good thing is it doesnt look too bad

the bad thing is i have no concrete idea how to do it

[pine tiger]

||p1 is simulation, p2 is dp||

[vast jetty]

the other good thing is ill worry about it after i go brush and stuff

[earnest forge]

There are a lot of libraries for that
From the top of my head, advent-of-code-data/aocd, aoc_helper (maintained by me), aoc_lube (maintained by salt-die)
I'm sure there are more

[hollow wharf]

I never did smth like this before

so this is gonna be fun and new

I understand the problem atleast now

[covert arrow]

this is surprisingly accurate lmao

[humble copper]

i was fooled

||naive recursion approach|| did not work, and neither did ||a list of all possibilities||

had to bring in 2021 methods for this 😅

[feral hazel]

does the hello world in https://ziglang.org/learn/samples/ work for any of you?
i'm getting error: root source file struct 'fs.File' has no member named 'stdout'

std.debug.print("hello world", .{}); works fine though

i'm on zig 0.14.1

[paper folio]

wow I got it done in 30 min!

[vast jetty]

whoops

[sullen adder]

the tutorial is probably for 0.15. I/O changed a lot

[hollow wharf]

which one

first or second

part

[vast jetty]

first

mate im writing code literally while brushinh

and getting my bed fixed and stuf

argh

[sullen adder]

multitasking pro

[vast jetty]

so true

alright

[vast jetty]

ill take that

[hollow summit]

oh my lord

is it yet ANOTHER recursion problem

[tender swallow]

||not necessarily||

[vast jetty]

ANOTHER recursion problem!

its in the name

the rest of aoc is gonna recursively be recursive

im confused about how to implement p2

might go get breakfast and worry about it afterwards

[hollow summit]

im giving myself a stack to work with

because thats usually what i do for "recursive" problems

[vast jetty]

well

the good news is my solution works

the bad news is it takes fucking forever

lemme see if i can memoize without fucking up

alright

thanks @cache

EH

why was this so easy lol

[feral hazel]

ah thanks
i hadn't updated for a while
compiles now on 0.15.2

[magic oxide]

yet another day of me forgetting about aoc

[vast jetty]

how could you

[magic oxide]

i got distracted theorem proving

[vast jetty]

math major?

[magic oxide]

math and cs

tho theorem proving is more CS related

as in automated proof checking

at least the thing i'm working on rn

i kinda don't feel like starting today's AoC before i finish work

[tender swallow]

finally solved 2024 day 21
nearly a year later

[hollow summit]

congrats, did the answer come to you in a dream or something

[tender swallow]

I just threw away most of my old code for it and tried out a new approach inspired by another AoC solution from that year

[hollow summit]

oh nice

[warm berry]

Yeah I am not doing part 2 today, this is not the weekend I learn pathfinding

[hollow summit]

||it's not pathfinding||

[warm berry]

Even better!

[hollow summit]

the problem says you have to ||count all possible paths||

[warm berry]

And I do not understand a way to make that happen, so I'm going to move on

[hollow summit]

everytime the solution im thinking of involves ||recursion||... i tend to almost give up

[tender swallow]

last year I was going crazy ||trying to predict which move now would result in the best move 25 layers later by heuristics, by doing simulations up to some constant further layers, constructing each layer's move sequence based on that, and other nonsense||
instead I noticed that ||you can just write down the best move for each (from, to) pair for the dpad, and use that to just construct the best move for the next layer. this alone was an efficiency improvement, but still wasn't enough though since you'd run out of time and memory trying to construct the best move 25 layers out, so I used the same strategy I used for day 11, which is that we just need the length and not the sequence 25 layers out, and we can aggressively split on As to basically cache everything||

[torn sapphire]

honestly, not the worst day today, I just had a brainfart and thought the naiive way of doing part2 was actually the clever way

[cyan sleet]

the secret is ||dynamic programming||

[torn sapphire]

that, or there is still an entirely iterative solution

[cyan sleet]

||dynamic programming|| can be iterative

[torn sapphire]

oh right, i thought ||you were just talking about making a recursive function that also uses a cache, that's just what my mind defaults to thinking dynamic programming always is :P||

[warm berry]

Yeah I don't know these terms

[cyan sleet]

😭

[warm berry]

I'm ok with not solving a day, I know I have a basic understanding of programming in general

[cyan sleet]

as a hint, you want to count the number of ways to arrive at any particular point

and counting doesn't actually require iterating over all paths

I think considering this basic case should be enough to give the overall idea

..S..
.....
..^..
.....
.^.^.
.....

annotated with counts
||```
..S.. 00100
..... 00100
..^.. 01010
..... 01010
.^.^. 10201
..... 10201

[warm heron]

Roulette language TeX when

[dense osprey]

beams cant be next to each other right?

[peak dock]

why not

[dense osprey]

ambiguity

but my input doesnt contain it anyways

[vast jetty]

i do wonder if theres a combinatorics approach

need to upgrade to pithon

[magic oxide]

welp

i procrastinated AoC way too much

guess i'll do it tmr

[vast jetty]

huh

interesting

very interesting

[magic oxide]

was going to quickly do it today jumping off of last year's roulette code, but seems like zig has moved on and it's kinda late

[vast jetty]

and makes me feel dumb for not thinking of that

HMMM

so its like literally just ||2^n|| is it?

well it cant be

hm

oh i did think of that and went 'cant be'

but clearly there should be a combinatorics solution

[naive wharf]

when counting the tachyon splits, can I just count 'splitters hit by a beam'?

[vast jetty]

yep

spoiler that

also sorry i thought you were talking about p2

[tender swallow]

wouldn't that only work if the splitters were distributed in a regular geometric pattern?

[vast jetty]

yes

but i suppose there are subproblems to be had

or something

[naive wharf]

I do think it's time to finally create an IntMap...

[naive wharf]

https://paste.pythondiscord.com/M7FA
oh wow, the easiest way to solve today's puzzle is probably to do what exactly the question specifies

[cyan sleet]

It's mostly the reader/writer generics that happened

[hidden musk]

me when it takes 30 minutes to figure out reading files

[worthy horizon]

!cleanban @vagrant bear scam

[marsh current]

:incoming_envelope: :ok_hand: applied ban to @vagrant bear permanently.

[sharp elk]

Am I the only one who's struggling with today's part 2?

[tender swallow]

not at all, I've seen many other people who had trouble
you can ask for help in #aoc-solution-hints

[sharp elk]

My solution is too slow because I'm looping over all timelines to check if it is a "^" , I'm debating shouldn't I just another language

[low condor]

Which language are you using?

[sharp elk]

python ._.

[low condor]

Language change won’t help

[tender swallow]

it's not (necessarily) a language problem, my Python solution finishes in ~65ms

[low condor]

Your approach, your algorithm

That is what needs to change

[tender swallow]

depending on how exactly you're doing this, the change might just be ||functools.cache||

[sharp elk]

I'm not saying it's a language problem I'm saying I'm lazy to change the algorithms 🤣

[tender swallow]

the problem is that an algorithm too slow to solve this is going to be too slow in any language

just by the scale of the computation

[low condor]

You’ll have to - but as said above it might or might not be a big change

I will say this - this kind of problem and the solution most go for is one of Advent of Code’s staple type of problems

You get one problem like this each year

FWIW I had to rewrite my Part 1 to Part 2

[sharp elk]

it's my first yr

[low condor]

Not a huge rewrite but still

[worthy horizon]

that said, sometimes you can get away with a stupid algorithm just because of a 100x speedup from picking a different language

Like switching from Bash to Rust

well... that might be 1000x

[ivory vigil]

I would defend bash... but you might be right.

[sharp elk]

exactly 🤣

[worthy horizon]

speaking of, did anyone solve day 4b using the most naive approach? (just doing exactly what the example walkthrough did, ||removing a bunch of paper on each iteration until you can't remove anything||)

[tender swallow]

if it's a combinatorial explosion like I assume is happening here, though, I doubt even Rust would save it

[magic oxide]

quite a lot of people did

[low condor]

Guilty as charged

[tender swallow]

that was what I did at first yeah

it was fast enough
I switched to a slightly more interesting approach later just for fun

[worthy horizon]

how long does it take in python approximately?

[inland plank]

I dont understand why the sample beam only splits 21 times
If I count the number of splitters ( ^ ) in the sample, I get 22 and not 21
What am I missing?

[low condor]

Because beams don’t fall on every splitter

[tender swallow]

not all splitters are hit by beams
one of the splitters goes unused so to speak

[sharp elk]

I did ._.

[inland plank]

Which one? Its so confusing looking at all the |'s all clumped up

[sharp elk]

Yeah It wasn't a big change and it wasn't functools.cache

[low condor]

Ah nice

[inland plank]

Is it the middle splitter in the 3rd splitter row from top?

[low condor]

5th splitter at the bottom I’d say

[sharp elk]

Just had to change list to a dict

[inland plank]

Ohhhh

gotcha

thanks I missed that

[hidden musk]

iirc, 300 ms using some numpy

[worthy horizon]

mine was 2ms, so maybe my factors aren't too far off

same naive solution

(maybe you had a harder input though)

[worthy horizon]

did you || handle the edges of the board separately to reduce branching ||?

[sharp elk]

My solution takes 2 ms

[worthy horizon]

it's a smart one though?

[sharp elk]

nope

It doesn't even use any library

[hidden musk]

nah. it's braindead simple

[worthy horizon]

do you have a link? I'm curious

[sharp elk]

I just used a dictionary to keep track of how many lasers are in a column

[worthy horizon]

oh, we're talking about day 4 part 2

[sharp elk]

nvm sorry

[worthy horizon]

I haven't touched day 7 yet, I'll probably come back to the second half of AoC next year

[sharp elk]

my day 4 was 280 ms using the simplest approach I was lazy to make it faster

[worthy horizon]

using a different language also can mean different memory footprint, for example you can better make use of memory-speed tradeoffs i you can get very small integers

though Python does have array.array and numpy

[sharp elk]

Using numpy made it slower I was amazed ._.

[worthy horizon]

perhaps counterintuitively, getting an integer out of an array.array is slower than from a list. Because you have to construct a whole new int object based on the raw numebr

maybe it's the same thing with numpy, where you chat with it a lot instead of doing vectorized operations

[bright spindle]

My p2 Python solution takes 0.8 ms.

[sharp elk]

yeah I updated mine and it takes 2 ms

[bright spindle]

But yeah, you never have to switch language for AoC

[tender swallow]

a castle, a flag, a ladder (?)

[bright spindle]

Hm, is it like a stable or something on the side there?

[vast jetty]

||cache||

[vast jetty]

i think thats what we all did

also isnt that day 5

[vast jetty]

nope its day 4

i just mistitled my post

[bright spindle]

Spoilertags

[vast jetty]

indeed

[void forge]

lmao

I've been behind, I had other things to do and I was out so I only just completed day 4

[naive wharf]

tomorrow will be the floodfill problem
I feel it in my bones

[mossy stream]

Are we the only one who solved day 2 with regex? Lmao

[naive wharf]

I maintain that just generating invalid IDs is a lot easier than checking whether a given number is an invalid ID

[tender swallow]

it was faster certainly, but idk about easier

[mossy stream]

Generating all possible invalid IDs then checking against the ones in your input?

[lime fulcrum]

not quite all. more like all within a range

[cyan sleet]

You do full scans on each iteration? If so I wrote a fun test case generator that makes your code take O(n⁴) time (assuming the board is n*n)

[woven solar]

caught up on aoc 2025, maybe I'll actually do it this year

spent an embarassing amount of time on d1p2

[low condor]

Many many people did that myself included

[lime fulcrum]

so when do we think path finding day is coming up?

[cyan sleet]

I blame spreadsheets on a phone

the ux of that is horrible

[low condor]

For me it’s definitely trying to over-optimise because I was so embarrassed to have to do it the very naive way

[cyan sleet]

ah, that wasn't even an option when doing it on a spreadsheet

or at least idk how one would do the naive way

maybe it's possible with arrayformula

[low condor]

For part 2 it would be ||breaking down each L or R instruction into 1 step ticks and checking the position at every tick ||

For Python anyway

[cyan sleet]

right

[low condor]

Not sure about spreadsheets :)

[cyan sleet]

I just don't know how one would even express that in a spreadsheet

[low condor]

Fair enough - I have no clue either

[naive wharf]

Fingers crossed

[ripe meteor]

Keeping with tradition i think the hard day should be day 18 or maybe 22

<@&831776746206265384> (its a scam)

[shy tulip]

!compban 1447363591668764672

[marsh current]

Bad argument

Could not convert "user" into UnambiguousMember or UnambiguousUser.
User "1447363591668764672" not found.

[wheat sapphire]

[shy tulip]

!compban 1356942068177702922

[marsh current]

:incoming_envelope: :ok_hand: applied ban to @tribal frigate until <t:1765494463:f> (4 days).

[hidden musk]

iirc you can make cells refer to themselves and do recursion that way. mutual recursion even

[sharp elk]

Only 12 days ._.

[ripe meteor]

😉

[hollow wharf]

It’s only 12 days

this time

[ripe meteor]

people missing my joke makes me sad

[hollow wharf]

Oh no

😭 😭 😭 😭

[hollow wharf]

Wait u meant as in overflow

Didn u

[ripe meteor]

no i meant i hope they dont have hard days

[hollow wharf]

I was way too off

💀

Ts cooked

[ivory vigil]

I'm fine without puzzles that need z3 or hard parsing hundreds of lines of assembly. But I wouldn't mind a graph traversial or A* or whatnot.

[hardy relic]

this is from aoc 2015 - how can i count ALL targets in line - on line 20?

[ivory vigil]

line.count()?

[minor cave]

Need to sleep so a helper needs to open spoiler thread. Good luck y'all

[humble copper]

i finally did 2022 day 19 part 1 after 3 years

🥲

[hidden musk]

fun language 👁️

[humble copper]

pypy is not the answer 😨

[buoyant perch]

5 minutes to go! Why am I nervous? lol

[vast jetty]

lets seee

i hope not

or maybe i hope so

cuz i can handle that

[buoyant perch]

I've been doing fine until Day 7 and then the multi timeline thing just confused the heck out of me. I still don't understand how to count a "timeline".

[vast jetty]

im expecting another another 'naive approach takes 4 petabytes of space'

[half swan]

1 timeline is one unique path the tachyon can take

[buoyant perch]

It's funny how one little problem can destroy your confidence.

[vast jetty]

you have to count all paths to all possible end points

[vast jetty]

yeah dw youlll get there

[half swan]

Think of it like a pinball machine where the pinball doesn't bounce back up

[vast jetty]

im surprised i could make it this far

[vast jetty]

like

one of those gaussian distribution thingys

i forgot hte name

[buoyant perch]

Yes, but I don't understand how to count individual timelines. I either count way too low or way too high. It's not even about developing an algorithm yet, it's about understanding the requirements.

[half swan]

Ah yeah that had me stumped for a bit

the ||recursive|| solution is kind of intuitive but I challenged myself to do it without ||recursion||

Took a couple hours but I got it

That probably should've been spoilered a lil bit

[paper folio]

are ya excited

[buoyant perch]

Like I originally thought it was just 2^#numberOfSplitters. But that's way too high. And I finally understand, that when you choose a path above, that means some splitters below don't see that particular path, so it can't be 2^#numberOfSpillters

[half swan]

2^splitters means the beam has the possibility to split at each splitter

[shut trellis]

<@&518565788744024082> Good morning! Day 8 is ready to be attempted. View it online now at https://adventofcode.com/2025/day/8. Good luck!

[buoyant perch]

Right, and I finally understand why that's not the case.

In each timeline, only one splitter at each level can be reached

[vast jetty]

oh fucking shit

[half swan]

Yes but more importantly: ||whether you can hit a splitter is determined by other splitters further up||

[vast jetty]

alright

fuck it

im writing the naive approach first

[dense drum]

fr 😭

[vast jetty]

time for ||itertools.combinations|| again...

i hope thats good enugh fro part 1

i almost fell for it

[pine tiger]

oh wow, ||mst||

[vast jetty]

ffs

even the naive approach requires work

this might be the day i clock out

alright

good problem

im gonna take a whie

[earnest forge]

Off-by-one error cost me genuinely like 10 minutes

I may be stupid

[vast jetty]

yall are done already

wow

[earnest forge]

Worst part is if my helper had worked I wouldn't have needed to write the buggy code lmao

[hidden musk]

||disjoint 👏 set 👏 union||

[hardy relic]

Slund like a protest

[vast jetty]

alr im so done

forget the problem

im fighting with fucking python rn

cuz i cant write a list comprehension apparently