advent of code 2025

[terse edge]

noo i had the solution for p2 immediately but forgot to check ||if it was 2 or more parts||

instead of realizing that i went on to ||"at least 2 equal parts but they dont have to be all equal"||

my utils came in very handy but sadly im stupid

[pale stirrup]

got a solution but it's not really fast

2.5 second p2

[fallen lily]

Is it grid based

[pale stirrup]

no

||list of ranges||

[terse edge]

https://codeberg.org/eagely/adventofcode-kotlin/src/branch/main/src/main/kotlin/solutions/y2025/Day2.kt

[terse edge]

no

its easy

[terse edge]

mine runs in 400ms

buT

I have op cpu

so idk urs might be better

[pale stirrup]

its probably not

[terse edge]

what is iT

[pale stirrup]

||```kt
part2 {
var x = 0L
ranges.forEach { range ->
for (id in range) {
val idStr = id.toString()
val len = idStr.length

        for (i in 1..<idStr.length) {
            if (len % i != 0) continue
            val pattern = idStr.take(i)
            if (pattern.repeat(len / i) == idStr) {
                x += id
                break
            }
        }
    }
}
x

}

||bruteforce basically||

[terse edge]

i just did ||.chunked(i).isAllEqual()||

[pale stirrup]

one of those is a util

also spoiler

[terse edge]

womp womp

my utils are foss so theyre not utIlhs

anyone could have used my util for it 😄

[pale stirrup]

switched to your solution and it jumped to 13s

[rugged kiln]

I wonder if you can do aoc in typst, given its turing complete

[pale stirrup]

yesterday's was done in just typescript types

[quartz aurora]

Oops looks like I made an assumption that really isn't guaranteed to hold

But works on my input so whatever

[native maple]

ok time to solve d2 in elle

[native maple]

my solution is soooo cursed mao

i got both parts in like 25 mins but it was kinda slow so i redid the slow parts without string operations

@terse edge isnt it beautiful

what do we think

100ms on real input for both parts

[terse edge]

languages that dont use parens are so beautiful

[native maple]

https://github.com/acquitelol/aoc2025/blob/mistress/solutions/02/solution.le full solution

[terse edge]

https://codeberg.org/eagely/adventofcode-kotlin/src/branch/main/src/main/kotlin/solutions/y2025/Day2.kt

[native maple]

this is why rust is so beautiful?

[native maple]

now do it without string operations

[terse edge]

kotlin is so awesome i just wish it wasnt built for the jvm

[terse edge]

this is very bad looking rust idk why

but as you can see there are a lot of parens

[frigid socket]

what does that mean

[terse edge]

i can do ur *10 %10 thingy i think i did that for snafu 2022 d25 but string ops are easier and more readable

[terse edge]

dont prepend 0

just ignore that

[frigid socket]

ok

today i will do Kotlin

i liked kts

[native maple]

i mean theyre easier but also like, slower

[terse edge]

im using kotlin i dont give a shit about speed

[native maple]

i guess actually the slice part was the slow part not the digit counter

[native maple]

yeah but it matters for me i guess lol

the string op sol took like 5s this sol takes 100ms

big difference

[terse edge]

idk how long mine takes

well its 400ms

but on a 6ghz cpu

[terse edge]

prob like 2s compared to this

[native maple]

well there are some parts that make it slow for sure

[terse edge]

i dont understand any of this

only thing i understand is why the for loop is there

[frigid socket]

@terse edge is this good

[terse edge]

dont usg vscode

[native maple]

use zed

[terse edge]

for kotlin u have to use intellij

@native maple all microsoft products are made for people too lazy to configure shit

[frigid socket]

bloatware @terse edge

[terse edge]

didn't say it wasn't

[frigid socket]

Cursor

[terse edge]

but its the only thing that has a good kotlin lsp

[frigid socket]

IntelliSlop

[native maple]

||
is_twice checks for ABAB, so 44 -> true, 123123 -> true, 4343 -> true, 431 -> false

1. count the digits
2. if digit count is odd they cant possibly be the same
3. build p = 10^(d / 2)
4. split the number into x / p and x % p

this is the same theoretically as turning it into a string and splitting it in the middle and then checking whether the left and right match

is_repeat checks for whether a number is made up of k-digit blocks repeated 2 times or more

1. count the digits
2. try every k such that k * 2 <= d && d % k == 0
3. for every valid k, build p = 10^k
4. start from the right and check that every block going from right to left is the same as this block
5. if all blocks match, it is a "repeated" number, otherwise try another k
   ||

[terse edge]

yea ik the concept behind it but itd take me at least an hour to implement it

[native maple]

well i was close it took me like 30 mins ish

maybe a little more

[terse edge]

also there is a lb for the ppl that used to be an global lb more or less

[native maple]

oh that is a good point

i doubt it

[terse edge]

well it was open to everyone (now its full i didnt get in) in the aoc discord

[native maple]

oh that wasnt a question

[terse edge]

but jonathan, kroppeb, nimka, etc are there

[terse edge]

nop

[native maple]

oh

is neil doing it this year?

i didnt see any videos

[terse edge]

neil wu?

[native maple]

thistlewistle

[terse edge]

ok

oh

idk

idrc im not gonna watch content abt it

[native maple]

yea me neither

[frigid socket]

i kinda love kotlin

i should get into it more yk

i am too JSpilled

[terse edge]

use rust

[frigid socket]

vban @terse edge

wait now i cannot figure out a solution

@terse edge what if ||i divided the ID by length then length-- and divide and again and again until i inevitably get x identical sections||

okay i just misunderstood

i just have to check if it repeats twice 😭

[terse edge]

Are you using kotlin

if so theres ||string.chunked(n)||

for p2

[frigid socket]

@vague knoll my initial p1 solve was a p2 solve 😭

[vague knoll]

lol

[frigid socket]

@vague knoll i will keep doing p1 and p2 in a single function to piss off @fallen lily

[quartz aurora]

I almost always do that

[native maple]

i also do that

[frigid socket]

okay so satan is just insane

[vague knoll]

it seems like ai is not good at writing assembly

so i guess it's a pretty good option if you want a job

(I was curious how well chatgpt would do with the same challenge)

[frigid socket]

try Claude Code

[native maple]

i did it

https://github.com/acquitelol/aoc2025/blob/mistress/solutions/02/solution.ts

isnt it beautiful @terse edge

i think i can post without spoiler because i dont think this can help anybody ever

[verbal parrot]

wait im so dumb i missed this part about p1:

you can find the invalid IDs by looking for any ID which is made only of some sequence of digits repeated twice.

this makes things so much easier

[frigid socket]

i accidentally did a part 2 solve while doing part 1

😭

[terse edge]

no

no its noT

[frigid socket]

should i refer you to my therapist

[terse edge]

"isn't it beautiful"
"i think i can post without spoiler because i dont think this can help anybody ever"

[verbal parrot]

okay apparently i accidentally did p2 too

[frigid socket]

exactly

😭

[cloud iron]

dam it

[verbal parrot]

d2 is so much easier

[cloud iron]

not seen yet, but d2p1 is nice

[verbal parrot]

spinning the wheel for tomorrow

THANK GOD

wait oh my god all schools in my town are temporarily closed for the next few weeks which means… i will be free in the morning and can do aoc in the first hour (if i come up with the solution quick enough ofc)!!!!

[feral valve]

today was so easy what

[native maple]

now do it in the typescript type system

[lusty atlas]

today is funny

are you all just brute forcing it

[cloud iron]

no, i'm actually trying to find the exact stuff to only get good results
but currently i'm too low so it mean i'm probably off in my math

yep so that's the issue

[native maple]

yeah

[feral valve]

https://github.com/Tnixc/Advent-of-Code-2025/blob/main/2.ml

oh this would be way faster working with numbers yeah

[native maple]

its not that bad either

and very fast

[verbal parrot]

is my solution smart https://github.com/ficcialfaint/advent-of-code/blob/master/2025/02/02.dart

[feral valve]

fixed

yeah the github link is better

i'm really starting to like OCaml

my solution is really slow though

[cloud iron]

omg that part 2

i swear some people will use regex

[verbal parrot]

justfile spotted??

[cloud iron]

and i curse those people

[feral valve]

look inside for a cursed preprocessor

well actually cppo is the cursed preprocessor

[verbal parrot]

i’ve been tryna set up makefile for my aoc solutions repo so i don’t need to remember whatever compile command is for each day but it didn’t work so i switched to justfile and… it didn’t work either

[feral valve]

nix

could fix this

[verbal parrot]

bbbut windows users won’t be able to use it!

okay tbf they wouldn’t be able to use makefiles either

[cloud iron]

they would thanks to WSL

so this is my p1
i think i have to use a brand new strategy for p2

[native maple]

why did you jsdoc everything

[vestal owl]

NOOOOOOOOOOOOOOOOOOOOOOOO

[verbal parrot]

i love reading code in pages

[vestal owl]

i just realized danmakufu uses 32 bit floats only

fuck

[cloud iron]

because it makes vscode slightly smarter

[cloud iron]

tf

[verbal parrot]

i’m crine bro there are more comments than code

https://tenor.com/view/osamason-gif-3924553698000348200

[feral valve]

is the inference not very good?

[cloud iron]

it's fine, but adding the jsdoc takes nearly no time and makes it better

the diff isn't big, but enough to make tab suggestions better

reupload in spoilers cause yeah

[feral valve]

i've never seen that many vscode extensions holy hell

[native maple]

you could just use/.... typescript............

[vestal owl]

please censor t*pescript

anyways i cant figure out why even my obligatory assembly rewrite doesnt work

i've been staring at this for half an hour

[cloud iron]

how many times will i have to say it
typescript doesn't run in the browser

[native maple]

the humble tsc..

[lusty atlas]

what's ur cpu

my solution is way slower

my 0.37s solution

oh and this

[native maple]

m1

not sure if log10 is faster than a loop

[lusty atlas]

yes it is

i just tried

way faster

switching to / 10 loop took 0.1s longer

maybe it's just crystal being slower

[native maple]

its like 60ms slower using log10 in elle

wait

yeah about 60ms slower

oh no it is slightly faster just the conditions have changed since this morning

i now also have minecraft open

its like 10ms faster

awesome

[vestal owl]

oh my fucking god

cbt just under 1kb

[native maple]

whats the memory management model of crystal?

garbage collected?

[lusty atlas]

wait how is it faster for you, i have 7800x3d it should be way faster than m1

i ran your elle solution

[lusty atlas]

yeah lol

[native maple]

i have no idea, is your cpu arm or amd/intel?

[lusty atlas]

amd

[native maple]

also could just be things running in the background

performance is like very subjective

[native maple]

it could also be that qbe doesnt do vectorization so arm just happens to be slightly faster when neither are vectorized

i replaced the digit count with this

[lusty atlas]

py??

[native maple]

oh no thats just a second benchmark lol

[lusty atlas]

xD

[native maple]

i stole the makefile from my aoc 2024 when i compared the times

[vestal owl]

@native maple i can send my asm solution if you wanna benchmark

[native maple]

sure

[vestal owl]

you'll need fasm to compile

[lusty atlas]

tokenlagger...

[vestal owl]

or i'll just send the exe

[lusty atlas]

scary

oh you're the turkish femboy manti friend

[native maple]

i cant run fasm on macos

ill test on linux in a sec

[vestal owl]

imagine i somehow fitted an entire token stealer into 1kb

[vestal owl]

1984

also fell for the macos meme

[vestal owl]

i have no idea how everyone started calling me a femboy

it's not even just twitter i just have that kinda aura i guess

[native maple]

i pushed the log10f solution, try that see if its faster

youll have to git pull && make install-std install-runtime in the main repo tho because log10f and floorf werent in the stdlib

[vestal owl]

btw idk what you're doing but why not just keep the lengths while parsing the numbers?

that's what im doing

[native maple]

you mean like this?

its really slow idk why

[hoary mirage]

hi @native maple

[native maple]

no way manti in 🅱️encord

[hoary mirage]

I would beat you but I am in python

QWRONG

correct

[lusty atlas]

tried adding cache but it just made it way slower lmao

[native maple]

YEAH

[native maple]

so weird

[lusty atlas]

i mean you only cash the digit count

i cached everything

ig making a massive hashmap is slower than just doing the math

i wonder if a massive array would be faster

[hoary mirage]

how slow is yours noob

oh you edited rosies code

[lusty atlas]

no

mine is 0.3s for part2

[hoary mirage]

test mine

p1,p2 = 0,0
import itertools
u = {0}
with open('input.txt') as f:
    for r in f.read().strip().split(","):
        s, sl, e, el = (v for i in r.split("-") for v in (int(i), len(i)))

        for k in itertools.product("1234567890", repeat=int(len(str(e))/2)):
            if k[0] == "0":
                continue
            n = int("".join(k) * 2)
            if n >= s and n <= e:
                p1+=n

        for l in range(1,int(el/2) + 1):
            for k in itertools.product("1234567890", repeat=int(l)):
                if k[0] == "0":
                    continue
                for f in range(max(sl,2),el + 1):
                    n = int("".join(k) * (f//l))
                    if n >= s and n <= e and n not in u:
                        p2+=n
                        u.add(n)

print(p1,p2)

you have 9800x3d

[native maple]

isnt that 300ms

[hoary mirage]

yes

he saying 0.3 so it looks fast

[native maple]

mine is 80ms in elle for both parts

its not bad but can be faster

[hoary mirage]

I think I can combine p1 and p2

but it gets wrong result

[hoary mirage]

USE PYPY

NOBODY USES PYTHON3

[native maple]

USE PYPY

[lusty atlas]

don't have pypy

[fallen lily]

Nobody should use python

[hoary mirage]

INSTALL

[fallen lily]

So bad

[hoary mirage]

lc.gemini how 2 install pypy explain to @lusty atlas

[prime sphinx]

-# <:i:1263593669215256597> labsCore will be shutting down <t:1767265200:R> • Learn more

how 2 install pypy explain to <@34338357280505856…

There are a few ways to install PyPy, a fast, alternative implementation of Python. The easiest method depends on your operating system.

---

💻 Recommended Installation Methods

The quickest and most common methods are using a package manager like Homebrew (macOS), or Conda (cross-platform), or simply downloading the pre-built binaries.

1. Using Homebrew (macOS)

If you're on macOS and have Homebrew installed, this is the simplest way:

• Open your terminal.
• Run the following command to install the latest Python 3 version of PyPy:

  brew install pypy3
  

2. Using Conda (Cross-Platform)

If you use the Conda package manager (like with Anaconda or Miniconda), you can install PyPy into a new environment:

• Open your terminal/Anaconda Prompt.
• Create and activate a new environment with PyPy (e.g., using Python 3.9):

  conda create -c conda-forge -n my_pypy_env pypy python=3.9
  conda activate my_pypy_env
  

3. Using Linux Package Manager (Ubuntu/Debian)

Some Linux distributions offer PyPy through their package repositories. For Ubuntu/Debian, you can use the PyPy PPA:

• Add the PyPy PPA:

  sudo add-apt-repository ppa:pypy/ppa
  sudo apt update
  

• Install PyPy 3:

  sudo apt install pypy3
  

---

📦 Direct Download (Manual Installation)

For Windows, or if you prefer a direct download for other systems, follow these steps:

1. Download the Binary: Go to the official PyPy Download Page. Download the latest PyPy3 version for your specific operating system and architecture (e.g., Linux x86_64, Windows 64-bit).
2. Extract the Files: Unzip or untar the downloaded file (e.g., pypy3-vX.Y.Z-OS.zip or .tar.bz2) to a simple, convenient location on your machine (e.g., C:\pypy on Windows, or ~/pypy on Linux/macOS).
3. Run PyPy: PyPy is ready to run right after extraction. Navigate to the extracted folder (specifically the bin subdirectory on Linux/macOS) and run the executable.
   • Linux/macOS Example:

     cd /path/to/extracted/pypy-folder
     ./bin/pypy3
     

   • Windows Example:

     \path\to\extracted\pypy-folder\pypy3.exe
     

Optional: Add to PATH

To run the pypy3 command from any directory, you should add the directory containing the pypy3 executable (usually the bin folder) to your system's PATH environment variable.

---

✅ Verifying the Installation

After installing, you can check that PyPy is working correctly by running the interpreter and checking its version:

pypy3 --version

You should see output indicating the PyPy version and the CPython version it is compatible with (e.g., PyPy 7.3.11 with GCC 11.2.0, based on Python 3.10.13).

---

Would you like to know how to install the pip package manager for your new PyPy installation so you can install third-party libraries?

Gemini • Gemini may display inaccurate info, so double-check its responses.

[lusty atlas]

no

wtf does this do

insanity

tbh this might just be crystal skill issue

[lusty atlas]

here :)

[hoary mirage]

p1,p2 = 0,0
import itertools
u = set()
with open('input.txt') as f:
    for r in f.read().strip().split(","):
        s, sl, e, el = (v for i in r.split("-") for v in (int(i), len(i)))
        for l in range(1,int(el/2) + 1):
            for k in itertools.product("1234567890", repeat=int(l)):
                if k[0] == "0":
                    continue
                for f in range(max(sl,2),el + 1):
                    n = int("".join(k) * (f//l))
                    if n >= s and n <= e:
                        if l == f/2:
                            p1+=n
                        if n not in u:
                            p2+=n
                        u.add(n)

print(p1,p2)

optimized

try now

[hoary mirage]

creates all permutations of numbers

and checks if its in range

[lusty atlas]

hmm smart

[hoary mirage]

loooove

@native maple beaten

[lusty atlas]

asked chatgpt to translate my code to go and it runs very fast

so yes it's crystal skill issue

nvm it translated wrong

let's ask claude

claude got it

optimised rust version is 100ms

go takes 150ms

[hoary mirage]

python faster

rust bad language

[lusty atlas]

my solution is insanely dumb brute force

it checks every single number in range

i wonder why crystal is so much slower than rust though, they both use llvm backend and crystal gc shouldn't matter

[verbal parrot]

wait since when are u here

[hoary mirage]

since hour aog

[verbal parrot]

oh

[lusty atlas]

i made my solution 70ms faster by adding i64 suffix to every number

turns out otherwise it does a lot of implicit conversion

dumb that it's not smart enough to automatically do this at compile time

[verbal parrot]

what’s ur algorithm

[native maple]

sed -e 's/,/lEx\n/g;s/-/ /g;s/^$/lEx/' test.txt | dc -e '[ddd10%s110/s2Z1-l1r:Cl2d0!=C]sC[Number: ]s"0sK[s>s<0scl"nl<n[ to ]nl>ps>[l<dlCxl<Zd2/sBsL0sI[1+]silBs(lB1+s)[0s{l(s}1sJ[l{l(+;Cs%l{;Cs&0l%l&=ilJ*sJl{1+s{l{l}>:]s:l:xlJlI+sIl(1+s(l(l)>;]s;lB2*lL=;[1sIl<lK+sK]s=lI0!==lclI+scl<1+s<l<1-l>>,]s,l,x[K=]nlKpsK]sE[?xlXx]sXlXx'
``` @lusty atlas

[lusty atlas]

INSANE

[lusty atlas]

https://github.com/Vendicated/AdventOfCode/commit/d641848a093bea06c1971c1cd25b03b16ffd9de6

@native maple i have an even more optimised digit count function than the log10

[hoary mirage]

uiua2

[lusty atlas]

def digit_count(x)
  if x < 10i64
    1i64
  elsif x < 100i64
    2i64
  elsif x < 1_000i64
    3i64
  elsif x < 10_000i64
    4i64
  elsif x < 100_000i64
    5i64
  elsif x < 1_000_000i64
    6i64
  elsif x < 10_000_000i64
    7i64
  elsif x < 100_000_000i64
    8i64
  elsif x < 1_000_000_000i64
    9i64
  elsif x < 10_000_000_000i64
    10i64
  else
    11i64
  end
end

[native maple]

ok this reminds me of my solution in 2024 for a day

[hoary mirage]

same

i dont remember which day

[lusty atlas]

getting faster

[native maple]

yeah lmaooooo

day 7 2024

[hoary mirage]

STOP USING SECONDS TO MAKE IT LOOK FAST

USE MILLISECONDS

[lusty atlas]

i prefer nicer to read number

[hoary mirage]

why do yall use log

you can like get length before parsing it to int

[native maple]

hell yeah lmao

i did the bad one, then log10, then switched to this before committing because it was fastest

[hoary mirage]

djakstra day will bea early this year

so little prblems

[native maple]

what is i

the whole point is to not turn it into a string at all

[hoary mirage]

split("-")

[native maple]

oh well strings in elle are null terminated so its actually much slower to get the length of a string than to turn it into a number and use log10

[hoary mirage]

elle soo bad

[native maple]

sized strings soon

when i need sized strings i just use char[] anyway

[vestal owl]

no for twice-checking i just multiply with 10*(len/2) + 1

that's why after seeing part 2 i was like "fuck ts in out"

[vestal owl]

test my solution too baka

[vestal owl]

i will beat you with a stick

[lusty atlas]

WAIT it was fast all along i just had to enable optimisation @hoary mirage

bro

crystal doesn't optimise by default

[hoary mirage]

48MS

HOW

GUHHH DONT MAKE ME OPTIMIZE MINE MORE

YOU WILL DIE

[vestal owl]

null terminated strings
💀

[lusty atlas]

:)

[hoary mirage]

rust dev hearing about "null terminated" strings

[vestal owl]

not even that

[native maple]

just c strings

nothing special

[quartz aurora]

Null terminated strings suck, what else is new

[vestal owl]

it's just a bad idea cuz a lot of string ops require you to know the length

and also you gotta remember to add 1 to your allocations etc

[native maple]

string ops are in the stdlib you dont need to write it by hand

[vestal owl]

it's unoptimized tho

also dynamic linking ew

[lusty atlas]

why tf is it 2x as fast as rust

is this not best optimisation @native maple

[hoary mirage]

rust moment

worst language

[native maple]

yeah its slow i just havent gotten around to using sized strings beause it would require changing quite a lot of c ffi stuff

[native maple]

it is

weird

[vestal owl]

silence java dev

[vestal owl]

guess i cant complain much cuz i still havent gotten a working compiler

anyways hello world in 78 bytes

[lusty atlas]

@hoary mirage beat 42ms

[vestal owl]

i remember at 89 bytes being like "there is no way i can optimize this further"

[lusty atlas]

benchmark it

well your asm solution for today i mean

[brazen salmon]

@lusty atlas hop off ruby

[hoary mirage]

DONT U DARE CHALLENGE ME

@lusty atlas

p1,p2 = 0,0
u = set()
with open('input.txt') as f:
    for r in f.read().strip().split(","):
        s, sl, e, el = (v for i in r.split("-") for v in (int(i), len(i)))
        for l in range(1,int(el/2) + 1):
            for k in range(10**(l-1),10**l):
                for f in range(max(sl,2),el + 1):
                    n = int(str(k) * (f//l))
                    if n >= s and n <= e:
                        if l == f/2:
                            p1+=n
                        if n not in u:
                            p2+=n
                        u.add(n)
print(p1,p2)

try

[lusty atlas]

GUHHH

[hoary mirage]

NOOB

[lusty atlas]

impressive

i concede

[hoary mirage]

good

[lusty atlas]

my brute force too naive i can't win

[lusty atlas]

port to crystal / rust

and compare

[hoary mirage]

neverr

[lusty atlas]

the difference between pypy and python3 is funny

why don't they make pypy the official python version

[hoary mirage]

probably compatibility blahblahgblahg

[lusty atlas]

or yk cpython could just add jit

[hoary mirage]

I have another idea for optimization

lets see

[lusty atlas]

lmao i translated my solution to ruby and it still hasn't finished after 1 minute

slowest language award

[vestal owl]

uhhhhhhh

i'll have to write my own timer fuuuuuuuuck

[hoary mirage]

run p2

[vestal owl]

i do have a life

despite what you might believe

[hoary mirage]

those who believe

[vestal owl]

it's like 95-ish microseconds

60 msec on example

is that fast

[hoary mirage]

import math


p1,p2 = 0,0
u = set()
with open('input.txt') as f:
    for r in f.read().strip().split(","):
        s, sl, e, el = (v for i in r.split("-") for v in (int(i), len(i)))
        for l in range(1,int(el/2) + 1):
            for k in range(10**(l-1),10**l):
                for f in range(max(sl,2),el + 1):
                    d = int(math.log10(k)) + 1
                    n = k * (10**(d * (f//l)) - 1) // (10**d - 1)
                    if n >= s and n <= e:
                        if l == f/2:
                            p1+=n
                        if n not in u:
                            p2+=n
                        u.add(n)
print(p1,p2)

@lusty atlas try

[lusty atlas]

lets see

got damn

impressive

[hoary mirage]

love

I think this is max

at least w python

[lusty atlas]

didn't change much only few ms but still

the fact that you managed to squeeze out 2 extra ms

[hoary mirage]

[lusty atlas]

write the same solutuion in rust or c

[hoary mirage]

I would diee

[lusty atlas]

try crystal then it's very easy

[vestal owl]

CLOCK_MONOTONIC_COARSE is so imprecise 😭😭😭

[lusty atlas]

i'll ask claude to port your code to crystal

[hoary mirage]

yop do

[vestal owl]

i guess my best bet is CLOCK_MONOTONIC still

[vestal owl]

legalized robbery

[lusty atlas]

here's the crystal code

p1, p2 = 0_i64, 0_i64
u = Set(Int64).new
File.read("input.txt").strip.split(",").each do |r|
  parts = r.split("-")
  s, sl, e, el = parts.flat_map { |i| [i.to_i64, i.size] }
  (1..(el//2)).each do |l|
    (10_i64**(l - 1)...10_i64**l).each do |k|
      (Math.max(sl, 2)..el).each do |f|
        d = Math.log10(k).to_i + 1
        n = k * (10_i64**(d * (f//l)) - 1) // (10_i64**d - 1)
        if n >= s && n <= e
          if l == f/2
            p1 += n
          end
          if !u.includes?(n)
            p2 += n
          end
          u.add(n)
        end
      end
    end
  end
end
puts "#{p1} #{p2}"

surprising that claude got it, crystal is very niche and most ai are terrible at it

[hoary mirage]

n = k * (10**(d * (f//l)) - 1) // (10**d - 1)
``` this line is crazy ngl

I didnt write this part

Example: Say k = 123 (3 digits) and you want to repeat it 4 times to get 123123123123:

d = 3 (digits in k)
f//l = 4 (repetitions needed)
Formula: 123 * (10^(3*4) - 1) / (10^3 - 1)
= 123 * (10^12 - 1) / 999
= 123 * 999999999999 / 999
= 123 * 1001001001
= 123123123123

[lusty atlas]

did you vibecode

[hoary mirage]

nop

only that line

[lusty atlas]

this is more readable than your python version

[vestal owl]

mantı posts readable code for once and you immediately get sus for ai generated code

lmao

[hoary mirage]

he meant crystal code

the one he generated

[lusty atlas]

yeah lmao i asked him because he said he didn't write that part

who tf wrote it then

[vestal owl]

baby gronk

[fallen lily]

HELP

[lusty atlas]

okay but look how similar this code looks to python @hoary mirage learn crystal when

[vestal owl]

wasnt there another guy here that wrote in assembly

[fallen lily]

this is not my shell btw

[vestal owl]

@vague knoll hows it going

[vestal owl]

nixos mentioned lessgo

[vague knoll]

I am so traumatised i haven't started day 2 yet

[vestal owl]

damn

[hoary mirage]

brooo even discord doesnt have syntax highlihting

[vestal owl]

dw i mogged the competition anyway

[lusty atlas]

vscode has extension

it's alright just slow

kind of like rust analyser

crystal lsp is twice as slow as rust lsp

[hoary mirage]

vlang trauma

[vague knoll]

amazing

[vestal owl]

weird way to say vscope

[lusty atlas]

nah v lsp is much worse lmao

[hoary mirage]

it didnt even work

[lusty atlas]

did V get better lsp yet i should revisit V

[fallen lily]

running rustup update stable && rustup default stable fixed it

[vestal owl]

vlang has unironically some nice features that no one gives a damn about because it's more fun to meme on it

[vague knoll]

rust analyser is weird let me see errors without saving

[hoary mirage]

no one gives a damn BECAUSE THEY DONT HAVE A WORKING LANGUAGE SERVER

[lusty atlas]

bro it still doesn't have lsp

V is very cool but no LSP is dealbreaker ngl

[vestal owl]

do you need an lsp for assembly?

[fallen lily]

not really i think

[hoary mirage]

yes

[vague knoll]

i've never even heard of V

Vlang

[lusty atlas]

https://github.com/vlang/v

[fallen lily]

never met this v person

[vague knoll]

vlang should make a linter called vlang-tidy

[cloud iron]

i hate the p2, bruteforced it with regex, not happy of myself, because backtracking in regex is a sin, because not finite automata, but fuck it, it was too annoying, i was starting to do smth that was generating functions and shit

[lusty atlas]

V is funny because the way it compiles is that it translates V code to C code then compiles that code

[vague knoll]

and lsp server called vlangd

[lusty atlas]

REGEX

lmao

[cloud iron]

tbh it's literally made to be bruteforced with regex

[vestal owl]

vinter would be funnier tbh

[cloud iron]

both p1 and p2 can be solved with the simplest regex ever

[lusty atlas]

wait https://github.com/vlang/v-analyzer

[vague knoll]

ah yes

[vestal owl]

do they still do that?

UB-maxxing

[vague knoll]

that's kind of cool though

because c has a lot of compilers

[vestal owl]

vlang deez nuts

[vestal owl]

yea it's UB-maxxing

[fallen lily]

i love when i need one package but also download all of rust

undefined behaviour

[vestal owl]

undefined vendicated

[fallen lily]

libuv

hop on better timezone

[lusty atlas]

WAIT WTF

v analyer goood

[vague knoll]

why does vlang connect to discord

[lusty atlas]

it has everything @hoary mirage

time to start using v

even has formatting

[hoary mirage]

incrediblee

I will honestly stick to python this year ig

[lusty atlas]

V is rust + go fusion

combines best of rust with best of go

i wonder if V still memory leaks

[vague knoll]

perfect lang

[vestal owl]

the nested loops 😭

[vague knoll]

i like go and i like everything about rust apart from writing it

[lusty atlas]

2022 advent of code

[vestal owl]

@lusty atlas look into it expressions you'll like it

nvm

wait is v actually usable

insane

whatever it actually exists unlike jai 🙄

[lusty atlas]

yeah it seems very usable so far the lsp is responsive and good

[vestal owl]

incredbile

[lusty atlas]

bros made vlutter

[vestal owl]

god is dead

[vestal owl]

i kinda wanna beg the blowfish man for beta access but im too lazy to blow his ego up further

[fallen lily]

kinda shocked kotlin has do while

didnt think it would

[fallen lily]

solved p1 first try, p2 can't be that hard

very cursed kotlin

[fallen lily]

@terse edge rate p1

[vestal owl]

0/10

rewrite in assembly

[fallen lily]

@vague knoll did you do d2

[vague knoll]

no

I'm refactoring typescript for relaxation

[cloud iron]

0/10, bruteforced

[fallen lily]

runs fast enough

[cloud iron]

Even p2 run fast enough with regex bruteforce

But it's sadge

p1 can be done with smart algorithm to only find valid ids

[fallen lily]

REGEX

INSANE

got p2 first try

tbh d2 easier than d1

[cloud iron]

Dam

The regex is easier

If you're gonna bruteforce it anyway

[fallen lily]

if i wanted to write regex i would use javascript or c++

[cloud iron]

Well

I didn't want to do regex

[fallen lily]

when i want to write regex for fun i do vencord dev

[cloud iron]

But doing p2 while only finding valid ids and not bruteforcing is so pain i gave up

I'll do it another day

[fallen lily]

whats the non-bruteforce way

[cloud iron]

Gimme a sec

[cloud iron]

#1444763777437536368 message

My solution there

Basically it rely on the fact the invalid numbers are numbernumber

So if you can find the range of number, you just iter on it

[fallen lily]

you mean invalid?

[cloud iron]

Yes

[fallen lily]

is your p1 not bruteforced

we basically did the same things

[cloud iron]

Nope

It will only find invalid one

[fallen lily]

this is too confusing for my brain

[cloud iron]

Similar but not the same
Yours iter on all numbers in the range ans determine if they are valid
Mine determine the range of half numbers in the range then only iter on those which are the half of the invalid numbers

[fallen lily]

idgi

[cloud iron]

But you did implement an optimization

Yours is like bruteforced but better

[fallen lily]

would the worse bruteforce just be converting it into a string instead of doing math

[cloud iron]

Mine will scale in O(invalid(n)) since it doesn't matter how big the range is, i never iter on the range directly, but on the number of invalid
yours is O(n) because you iterate on all numbers in the range to filter them

[fallen lily]

i get that part

i think i will just watch a youtube video later to find out how the non-bruteforce method works

[cloud iron]

Basically yours is linear with the total of numbers covered by the ranges
Mine is linear with the total number of invalid

[fallen lily]

yes

i understand that part

[cloud iron]

I'll do my writeup if you want

[fallen lily]

nah, ill prob understand it after watching a video

[cloud iron]

Basically here's the example of how it'd work
||
So basically for 3 and 20
3 give 1 as half-bound directly compute from power of 10 cause odd
20 give 2 and 0 so 2 - 1 because 2 > 0 so 1
So the half-bounds range is {1, 1}
So when i iter on half-bounds, it generate 1 1 so 11
And indeed 11 is the only number between 3 and 20

If we take 12 and 45 by example
12 give 1 and 2 so 1 + 1 because 2 > 1 so 2
45 give 4 and 5 so 4 because 4 < 5
So half-bounds range is {2, 4}
So the numbers are 22, 33, 44
||

[frigid socket]

@fallen lily wdym math

?

[fallen lily]

not converting the number into a string

[cloud iron]

The you need to compute divisors

There's a way

[frigid socket]

insane

look at my solution

[cloud iron]

Basically 1212 is 101 * 12

[fallen lily]

@frigid socket my p1

[frigid socket]

||https://git.nin0.dev/nin0/aoc-2025/src/branch/main/2/main.kts||

[fallen lily]

you dont need bigint

you can use long

[frigid socket]

gu

[fallen lily]

if aoc ever needed numbers > 2 ^63 - 1

it would be insane

[cloud iron]

It would be funni

[fallen lily]

because so many langs have that as the limit where numbers are sane

[cloud iron]

They should have at max uint256_t

Dam you might have nerd sniped me

I want to do it the pure math way now

[frigid socket]

does long have kotlin range stuff

[frigid socket]

vai

[spring lion]

Yes, Kotlin supports ranges for Long values. You can create a LongRange using the .. operator, and also use functions like downTo and step to define progressions. These are handled by LongRange and LongProgression types.

-# Response generated by gemini-2.5-flash. AI may be incorrect or misleading.

[cloud iron]

Basically the pure math way is like this

[fallen lily]

Just look at my solution

Uses all that

[fallen lily]

That makes a bit more sense in my head

[cloud iron]

Yeah that's one step above the hybrid math string way
The math string way is more of my inner dev soul
And the pure math way is my math soul
And i stopped doing math

[lusty atlas]

i told zooter to join

[frigid socket]

Bilboofan

[fallen lily]

vai what is the equlivient of LongRange from kotlin in rust

[spring lion]

In Rust, the equivalent of Kotlin's LongRange (which represents an inclusive range of Long values) is typically achieved using the inclusive range syntax start..=end with the i64 integer type.

Here's how they compare:

• Kotlin: val myRange: LongRange = 1L..100L
• Rust: let my_range = 1i64..=100i64;

In Rust, i64 is the signed 64-bit integer type, similar to Kotlin's Long. The ..= operator creates an inclusive range, meaning both the start and end values are included in the range. These ranges can often be iterated over.

-# Response generated by gemini-2.5-flash. AI may be incorrect or misleading.

[fallen lily]

dumb ai

i asked for the type, not the syntax

[fallen lily]

vai what is the name of the type in rust

[spring lion]

In Rust, the type created by 1i64..=100i64 is std::ops::RangeInclusive<i64>.

-# Response generated by gemini-2.5-flash. AI may be incorrect or misleading.

[lusty atlas]

why not just tias

[fallen lily]

just what?

[lusty atlas]

try it and see

[fallen lily]

what does tias mean

oh

[fallen lily]

guhhh

does rust not have assignment expressions

vai does rust have assignment expressions

[spring lion]

Rust does not have assignment expressions in the way many other languages do. In Rust, variable assignment using the = operator is a statement, which means it does not return a value.

While Rust is an expression-oriented language where most constructs evaluate to a value, assignment specifically returns the unit type () (an empty tuple), and thus cannot be used for chaining assignments or within contexts expecting a meaningful return value. This design choice helps prevent common bugs where = might be accidentally used instead of == for comparison.

-# Response generated by gemini-2.5-flash. AI may be incorrect or misleading.

[fallen lily]

evil

guhhhh

you cant flatten ranges in rust

dum

vai what is the equivalent of std::iota from c++ in rust

[spring lion]

In Rust, the equivalent of std::iota from C++ can be achieved using range iterators. You can create a sequence of numbers and then collect them into a vector or other collection.

For example, to fill a Vec with numbers from 0 up to (but not including) 10, you can use:

let my_vec: Vec<i32> = (0..10).collect();
// my_vec will be [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]

If you want an inclusive range (0 to 10, including 10), you would use:

let my_vec: Vec<i32> = (0..=10).collect();
// my_vec will be [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

These range syntax (a..b for exclusive end, a..=b for inclusive end) create an iterator that yields sequential numbers, which can then be used with methods like collect() to form a collection.

-# Response generated by gemini-2.5-flash. AI may be incorrect or misleading.

[fallen lily]

collect is so weird

[terse edge]

String ops woUld be more easier

but it looks good

[fallen lily]

Tbh I didn't consider turning it into a string before I wrote that

Doing it that way is faster anyway

Did I get ghost pinged @vestal owl

EVIL

@vestal owl youuuuu

[vestal owl]

what should i do day 3 with

[fallen lily]

Cpp

[vestal owl]

excel?

[vestal owl]

boring 😭

[fallen lily]

Minecraft command blocks

Typescript types

[vestal owl]

maybe scrath again hehe

[fallen lily]

C++ templates

C++ consteval

[terse edge]

@native maple ||I tried dp before realizing dp is too slow and its just a greedy solve||

[verbal parrot]

why is today so easy

okay nevermind i thought p2 was easier

[fallen lily]

I peaked last year on day 3

[terse edge]

@native maple https://codeberg.org/eagely/adventofcode-kotlin/src/branch/main/src/main/kotlin/solutions/y2025/Day3.kt

i did dp fIrst

it worked for example

but too slow for real

[pale stirrup]

ran into some small issues on p2 but my thought process was correct out the gate

integer overflow on a long

[terse edge]

how

actuAlly I dont parse the input as numbers

i parse as list long

[pale stirrup]

idk how all i know is that using a ulong fixed it

how fast is your solution

[fallen lily]

Thanks reddit, would have gone insane tomorrow if I didn't see this

[fallen lily]

Also insane API from rust

[pale stirrup]

kotlin does first maximum

[verbal parrot]

🎉

my p2 is prolly inefficient asf but it works

[pale stirrup]

i count under 100ms as a win

[verbal parrot]

FUCK NO

it was so close to c# why c

[fallen lily]

What's so bad about c

[verbal parrot]

not bad but it will definitely be harder to do aoc in c rather than like python. at least for me. it doesn’t have lists or like literally anything else

and i don’t want to nor do i know how to manage memory in c

[fallen lily]

For this you don't have to

Just malloc and don't worry about free

You know malloc, calloc, realloc, right

[verbal parrot]

yeah

[fallen lily]

That's enough

You can use xxd to embed your input as a header file so you don't worry about file system

[verbal parrot]

i mean i don’t think there could be any issues with fs or could there?

it sounds straightforward is it not

[pale stirrup]

i think he means if you just don't know how

[fallen lily]

It's easier tbh

You just get a char array

You don't have to worry about reading it

[fallen lily]

Yes

@verbal parrot if you have a compiler that supports c23, you can use #embed

https://en.cppreference.com/w/c/preprocessor/embed

gcc 15 should have it

Clang 19 has it as well

[verbal parrot]

uhh idk i’m just using whatever gcc was in xcode

[fallen lily]

Isn't xcode like 99999 versions behind

[verbal parrot]

most likely yes

[fallen lily]

Check with -version

If not, find a way to use an up to date toolchain

( or just use xxd)

[verbal parrot]

i’ll do in five mins i’m kinda stuck in bathroom

[fallen lily]

Insave

[verbal parrot]

why specially xxd tho? will just cat not work?

[fallen lily]

Hop do people use their phones in the bathroom

[verbal parrot]

with one hand

[fallen lily]

idk

I never use it in the bathroom

It's just weird

[fallen lily]

xxd -i file.txt https://tryitands.ee

[verbal parrot]

i hate how bathroom in english can refer to both shitting and showering

i meant the first when i said im in bathroom

[pale stirrup]

it doesn't

what

[fallen lily]

No

[verbal parrot]

oh?

but your toilet is located in your bathroom and so does your shower right?

[pale stirrup]

yeah

[fallen lily]

If you say you're in the bathroom that could mean anything from cleaning yourself up to number one or number two or showering or anything. It just means you're in the room

If you say you're using the bathroom then that means you're doing number one or number two

[verbal parrot]

ohh

[fallen lily]

That's how I view it at least

[verbal parrot]

here we just say “i’m in toilet” but i think in english toilet is like 🚽

[fallen lily]

This makes no sense in English

[verbal parrot]

yeah that’s why it’s so confusing to me

[verbal parrot]

ohh thats so cool

[verbal parrot]

gcc is actually clang

[fallen lily]

faint so smart

[verbal parrot]

idgi

[hoary mirage]

whats todays problem

[vestal owl]

that was easy

wow

@hoary mirage you will love this

[hoary mirage]

nop

hate

[vestal owl]

i think i know how to solve p2 too

even p1 took like 10 seeconds 😭

i hate how slow scratch is

[hoary mirage]

your answer is too high

hell nah

oh nvm I copy pasted işnput 2 times

🥀

p2 is little tricky hm

[vestal owl]

skill issue

actually it's pretty easy

[vestal owl]

too late lmao

[vague knoll]

epic

scratch is the best language

[quartz aurora]

Today's turned out pretty nice, 145 bytes ||```py
n=[0,0]
for L in open(0):
for M in 2,12:i=-1;n[M>6]+=int(''.join(L[i:=max(range(i+1,len(L)-M+k),key=lambda i:L[i])]for k in range(M)))
print(n)

[vague knoll]

I had an idea: orthodox kotlin

basically kotlin with only java class library

(I will probably try this, also maybe trying to build kotlin with a makefile)

[hoary mirage]

WHYS IT CORRECT ON EXAMPLE

[vague knoll]

i'm doing today in go

I need a break from asm

[vestal owl]

true

[vestal owl]

same

thats why i did scratch

[hoary mirage]

broooo kys

[vestal owl]

lmaooooooo

switch to rust already

[vague knoll]

I will only write orthodox C++/kotlin or Crust

[hoary mirage]

@vestal owl tell me the solution

[hoary mirage]

hmm

idk aoc is wrong

my code is perfect

@native maple ROSIE

HHELP ME

[native maple]

hold on i haven't done it yet

i'll see if i can do it before i have to go

[native maple]

i saw something about dp from eagely

[frigid socket]

gonna do day three

mantika vibe codes

[hoary mirage]

i think its correct

however

[frigid socket]

spoiler

[hoary mirage]

YOU CANT EVEN UNDERSTAND THE CODE ANYWAY

MY VARIALBE NAMES ARE HORRIBLE

whatever change in tactics

this code only works on example

so useless

nvm

[frigid socket]

@hoary mirage mantikabubu

[hoary mirage]

WHAT NONO

I AM ANGRY RN

[verbal parrot]

@frigid socket how u ain’t done d1p2 yet

[vestal owl]

i see the problem

[frigid socket]

do you know swift

[vestal owl]

read the spoiler text i sent in the other discord

[verbal parrot]

not really why

[vestal owl]

@hoary mirage

[frigid socket]

my solution is in swift

[vague knoll]

tay

[verbal parrot]

show

[frigid socket]

https://git.nin0.dev/nin0/aoc

[verbal parrot]

you did NOT have to use regex for that 😭

[vague knoll]

regex is neever needed for anything

it improves quality of life

[verbal parrot]

sure but in this case it’s literally either the first char is L and everything else is the number or the first char is R and everything else is the number

[frigid socket]

i am learning GoLang while doing d3 @vague knoll

[verbal parrot]

@frigid socket are u picking languages randomly or are these the ones u chose by yourself

[vague knoll]

https://cdn.discordapp.com/emojis/1445723534499516416.webp?size=96

[frigid socket]

every day i decide a language

[verbal parrot]

rust when??

[frigid socket]

never

or maybe some day

[vague knoll]

rust when ?? hello guys ??? anyone using rust ??? nobody ???

[verbal parrot]

what languages are u planning to do

[verbal parrot]

yall heard of this niche language ‘rust’??

[vestal owl]

best language evet btw

(right after brainfuck)

[frigid socket]

how is this ERROR

[vestal owl]

anyways i wanna use lolcode next

[frigid socket]

i decide like the day before or same day

[vestal owl]

swift moment

[vague knoll]

go 😭

[hoary mirage]

it maeks sure you dont put unneccesary variables and bloat ur code

go sooooo good

[vague knoll]

unless swift also uses the pen island operator

[frigid socket]

it can be a WARNING

not an ERROR

[vague knoll]

so true...

but you can just do _ = varname to ignore

[frigid socket]

love

god i need to get used to redwood tree operator

[vestal owl]

didnt you say swift

[vestal owl]

duh

[vague knoll]

his day 1 solution was swift

[vestal owl]

this has to be one of the worst language design decisions ever

[vague knoll]

generally by the time you're running the code you've actually used the variable so i don't find it too annoying

[hoary mirage]

rgbcube alt

[vague knoll]

does go just not do any warnings in the compiler? and leaves that up to linting tools?

the problem with warnings being emitted as a side effect of compilation is you need to edit the code again to see them

but maybe there shouldn't be any unused errors? just leave that up to go vet or whatever?

[vestal owl]

is it false tho

https://x.com/peach2k2/status/1945082719324557726?s=20

peachey 𔐓 (@peach2k2)

zig is such an annoying brat about unused variables, there is a fork to solely add a flag to disable them

seriously wtf is wrong with you andrew

X · Jul 15, 2025, 11:27 AM

[verbal parrot]

TRUE

“error: you didn’t use return value from a function call” cuz i don’t need it????

why does it need to be an error

[vestal owl]

that does make sense for heap objects tho at least

[vague knoll]

honestly it is kind of good to explicitly ignore return values

[vestal owl]

unused variables is just fucking stupid

[vague knoll]

I've got annoyed with [[maybe_unused]] spam in C++

**I mean [[nodiscard]] spam

so nodiscard should probably be the default

¯_(ツ)_/¯

though realistically you don't need to spam it on getters like I've seen people doing

(just prefix your getters with get, oh I forget C++ devs are insane)

[frigid socket]

does GoLang not have map

insane

[verbal parrot]

idk both unused variable and unused return sounds dumb to me

[vague knoll]

it does have hashmaps

but it doesn't have Array.map

[frigid socket]

[].map()

[verbal parrot]

vai golang map

[spring lion]

In Go, a map is a built-in data structure that stores elements as unordered key-value pairs, similar to hash tables or dictionaries in other languages. Each key within a map is unique and maps to a specific value.

Declaration and Initialization

You can declare a map using the map[KeyType]ValueType syntax.

• Declaration only:

  var myMap map[string]int
  

  A newly declared map without initialization will be nil and cannot have elements added to it until it's initialized.

• Initialization with make(): The make function allocates and initializes a map, making it ready for use.

  myMap = make(map[string]int)
  // Or declare and initialize in one line
  anotherMap := make(map[string]string)
  

• Initialization with a literal:

  scores := map[string]int{"Alice": 90, "Bob": 85}
  

Working with Maps

• Adding or Updating elements: Use the map[key] = value syntax.

  myMap["key1"] = 10
  scores["Alice"] = 95 // Updates Alice's score
  

-# Response generated by gemini-2.5-flash. AI may be incorrect or misleading.

[verbal parrot]

not this dumb fuck

[vague knoll]

there are libraries you can use for that

iirc method chaning doesn't work very well in go due to automatic semicolon insertion

hello
.method()
will become
hello;
.method()

[frigid socket]

yeah i found https://pkg.go.dev/github.com/samber/lo@v1.52.0

[vague knoll]

though ig you can do
hello.
method
and that's fine

lodash style 😭

@frigid socket https://pkg.go.dev/github.com/peterje/labubu

thankks to asi you can do this

[hoary mirage]

Yes

[native maple]

ran for my bus, missed it, went to a shop, collapsed due to running too fast for too long

now i'm on the next bus i can do aoc

[vague knoll]

you're missing the riveting jolt calculation

[frigid socket]

@vague knoll

i'm actually liking golang

[vague knoll]

golang is cool

[frigid socket]

cc @lusty atlas

[native maple]

vban

[vague knoll]

https://tenor.com/view/yes-gif-22712908

[frigid socket]

whyy

[native maple]

golang? to lang is not a verb smh

[frigid socket]

@hoary mirage That's the right answer! You are one gold star closer to decorating the North Pole. [Continue to Part Two]

[vestal owl]

BOOOOOOOOOO

[frigid socket]

NixOs tag you can't be talking

[hoary mirage]

ARE YOU OKAY ROSIE

DID YOUR BLOOD SUGAR DROP

[vestal owl]

weird way to say PeakOS

[vestal owl]

most fit programmer btw

fittest

english is hard

[verbal parrot]

how tf do u do aoc in a bus

[vestal owl]

laptop?

[verbal parrot]

i can’t even use my phone in a bus how do u use a laptop

[frigid socket]

girl?????????

just collapsed and still thinking about aoc

😭

[vestal owl]

skill issue

[verbal parrot]

no it’s not there isn’t a single empty space in buses most of the time

[vestal owl]

simply wake up at 7 am

take the least traveled path

[verbal parrot]

NO

i’m so glad taxi here so cheap like from my home to school it’s only $1.67

[frigid socket]

https://tenor.com/view/bosnov-67-bosnov-67-67-meme-gif-16727368109953357722

SIX SEVENNNNNNNN

[verbal parrot]

FranzJ

nino are you tapped in??

[frigid socket]

did you do p2 @verbal parrot

[verbal parrot]

today?

[frigid socket]

yeah

[verbal parrot]

yeah

[frigid socket]

love

i am refactoring my p1 solution to support p2

[verbal parrot]

shouldn’t be that hard

lowkey why is d1p2 the hardest yet

[frigid socket]

true

[native maple]

fuck

whatever

[hoary mirage]

what happened rosie

[terse edge]

WHO PINGED ME

[native maple]

i pinged you

my code had a || and it broke the spoiler

[terse edge]

wtf thats so expensive

[native maple]

sorry

ill just post on github

[verbal parrot]

how

[terse edge]

its 30 cents for us and 22 cents for kids

in kazakhstan

[frigid socket]

fres

[terse edge]

altho its rural village

[frigid socket]

eagely is twelve

[terse edge]

lmao

[native maple]

@terse edge i did this for p1
||```rs
fn solve(string[] lines) {
total := 0;

for line in lines {
    if line.is_empty() or line.len() < 2 {
        continue;
    }

    max_prefix := line[0];
    best := 0;

    for j := 1; j < line.len(); j += 1 {
        cand := (max_prefix - '0') * 10 + (line[j] - '0');

        if cand > best { best = cand; }

        if line[j] > max_prefix {
            max_prefix = line[j];
        }
    }

    total += best;
}

return total;

}

but i wrote a generic fn after that ||handles both 2 and 12||

[terse edge]

user states their age in this message

[verbal parrot]

when i was in moscow i had to pay like $8 for a 10-15min ride

[terse edge]

ok thats moscow

not muhosransk

[native maple]

@terse edge did you do something similar
https://github.com/acquitelol/aoc2025/blob/mistress/solutions/03/solution.le

[terse edge]

working with numbers aain

https://codeberg.org/eagely/adventofcode-kotlin/src/branch/main/src/main/kotlin/solutions/y2025/Day3.kt

this is optimized solution

original sol had for loop instead of recursion

and its generic for both parts

[native maple]

thats so short wtf

[terse edge]

@native maple im not gay but i will be gaY for 3blue1bRown

Im gonna buy his merch

https://store.dftba.com/products/zeta-spiral-shirt-1

[vestal owl]

https://tenor.com/view/67-6-7-6-7-6-7-meme-gif-10920175127833152373

[terse edge]

will I gain aura if i wear this in a university

[vestal owl]

no

only if you wear the 6 7 shirt

[terse edge]

67 dId not help me understand math

3b1b dId

[vestal owl]

skill issue

[verbal parrot]

https://media0.giphy.com/media/v1.Y2lkPTc5MGI3NjExMXA4MXlhdm5qeXlqdzUzdnh0MTZzZ240bnJkZDNyMTFrbHVybm9pdCZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/8aqiXARf2MFR7Y6B69/giphy.gif

[hoary mirage]

@native maple I SOLVED IT

ON PHONE

[verbal parrot]

stop sending it without spoilers 😡😡😡😡😡😡😡😡😡😡

[frigid socket]

SPOILER

[frigid socket]

did you do p2 @hoary mirage

[hoary mirage]

Brooo not like yall can understand my code

[hoary mirage]

Yes

[native maple]

girl this code is readable

@hoary mirage more readable than my solution from yesterday

[hoary mirage]

WHAT IS THAT

[hoary mirage]

Ig true this was readable

You still need to look at it for some time

[vestal owl]

have you not seen my assembly solution

[frigid socket]

for the life of me i cannot figure out p2

@hoary mirage you should take a look

it works for p1

[native maple]

my solution in the typescript type system

[hoary mirage]

Very easy once you figure out

[frigid socket]

||

package main

import (
    "fmt"
    "os"
    "strconv"
    "strings"

    "github.com/samber/lo"
)

func check(e error) {
    if e != nil {
        panic(e)
    }
}

func getTotalJoltage(pack []int, digits int) int {
    lockedIndex := 0
    working := ""
    for i := 0; i < digits; i++ {
        highestValue := 0
        upperLimit := 0
        if i == digits - 1 {
            upperLimit = 0
        } else {
            upperLimit = 1
        }
        lowerLimit := 0
        if i == 0 {
            lowerLimit = 0
        } else {
            lowerLimit = 1
        }

        println("We are at", i + 1, "looking from", lockedIndex + lowerLimit, "to", len(pack) - upperLimit)

        for ii, battery := range pack[lockedIndex + lowerLimit:len(pack) - upperLimit] {
            if battery > highestValue {
                highestValue = battery
                lockedIndex = ii
            }
        }
        working += fmt.Sprint(highestValue)
    }

    finalValue, err := strconv.Atoi(working)
    check(err)

    return finalValue
}

func solve(path string) (int, int) {
    data, error := os.ReadFile(path)
    check(error)
    input := string(data)
    batteryPacks := lo.Map(strings.Split(input, "\n"), func(item string, _ int) []int {
        return []int(lo.Map(strings.Split(item, ""), func(vlt string, _ int) int {
            value, error := strconv.Atoi(vlt)
            check(error)
            return value
        }))
    })

    joltages1 := []int{}
    joltages2 := []int{}
    for _, pack := range batteryPacks {
        joltages1 = append(joltages1, getTotalJoltage(pack, 2))
        joltages2 = append(joltages2, getTotalJoltage(pack, 12))
    }

    part1Sum := 0
    for _, jolt  := range joltages1 {
        part1Sum += jolt
    }
    part2Sum := 0
    for _, jolt  := range joltages2 {
        part2Sum += jolt
    }

    return part1Sum, part2Sum
}

func main() {
    exampleSolve1, exampleSolve2 := solve("example.txt")
    realSolve1, realSolve2 := solve("input.txt")
    fmt.Printf("EX1: %d, EX2: %d, R1: %d, R2: %d\n", exampleSolve1, exampleSolve2, realSolve1, realSolve2)
}

||

[hoary mirage]

DONT USE 12 FOR LOOPS