@quartz kindle its alive!!
#development
neon leaf
lyric mountain
def file = new File('values.txt')
file.text.split(',')*.split('-')
.collect { it*.toInteger() as IntRange }
.collect {
[it, it.findAll { (it as String) =~ /^(.+)\1$/ }]
}
.each { range, vals -> println "For $range.from..$range.to: $vals" }
quartz kindle
<@180112943612952577> its alive!!
such a beast deserves way more ram
feed it properly will ya
quartz kindle
xD
quartz kindle
LOl
neon leaf
im also waiting for nvme prices to chill out again
500€ for 2x4TB before, now its like 620€
not as bad as ram but still
neon leaf
mmm
lyric mountain
oh wait, ur supposed to sum the invalid ids
def vals = new File('values.txt').text
def totalWrongIds = vals.split(',')*.split('-')
.collect { it*.toBigInteger() as NumberRange }
.flatten { it.findAll { it =~ /^(.+)\1$/ } }
.sum()
println totalWrongIds
crashes with my user input 🗿
the hell did u input lmao
stark kestrel
some quite big numbers
Caught: java.lang.NumberFormatException: For input string: "9393862801"
java.lang.NumberFormatException: For input string: "9393862801"
queen needle
not supposed to share your inputs
lyric mountain
ah yes, integer issues
queen needle
11-22,95-115,998-1012,1188511880-1188511890,222220-222224, 1698522-1698528,446443-446449,38593856-38593862,565653-565659, 824824821-824824827,2121212118-2121212124
This is the example input
stark kestrel
the example input works
it just prints the ids themselves instead of summing them up
queen needle
in groovy what is *.
lyric mountain
runs the following function on every entry
queen needle
Ah yeah so js map instead of like for each
lyric mountain
it's shorthand for map yeah
lyric mountain
thank god java is stable as always
bro, how
queen needle
Oh groovy is just a java subset?
lyric mountain
superset
queen needle
Or superset whatever it's called
lyric mountain
yea, any java code works on groovy
queen needle
that's cool
lyric mountain
the println wasnt even necessary tbh
lyric mountain
aight, did day 1 to just for completion
def input = new File('../inputs/day1.txt').readLines()
int zeros, dial = 50
(input =~ /([LR])(\d+)/).each { _, dir, num ->
def val = num as int
dial = ((dial + (dir == 'R' ? val : -val)) + 100) % 100
if (dial == 0) zeros++
}
println zeros
My exact solution lmaoo
import * as fs from "fs";
type Line = `${"L" | "R"}${number}`;
const inputFile = "input.txt";
const fileContent = fs.readFileSync(inputFile, "utf-8");
const lines = fileContent.trim().split("\n") as Line[];
let dial = 50;
let password = 0;
for (const line of lines) {
const direction = line[0];
const value = parseInt(line.slice(1));
const sign = direction === "L" ? -1 : 1;
dial = (dial + sign * value + 100) % 100;
if (dial === 0) password++;
}
console.log(password);```
prime cliff
Gonna do permissions stuff differently with 4 categories 🙂
clever tundra
@neon leaf how much do you know about CF tunnels
pearl trail
ask the question, i might be able to help
wheat mesa
same
clever tundra
I've got it all configured but I keep getting an I/O error.
I'm pretty sure:
- DNS not cause
- ports not blocked
- internet is good and stable
But when I visit the site I get error 1003 and my logs are all just I/o errors about port 7844
If you need any more specific info lmk, no guarantees how fast I'll be to reply thi
pearl trail
If you need any more specific info lmk, no guarantees how fast I'll be to reply ...
you sure can curl it through the server right? then i assume you already configured it on "published application routes" ?
patent bramble
<@745619551865012274> how much do you know about CF tunnels
Cloudflare also has a discord if you want to get further help with things there. I'd definitely clarify a lot more though. What type of service are you running, what's the goal, etc. You can provide clarity without sharing sensitive information.
Ice heard discord has been known to "lax API restrictions for larger bots". Does anyone here happen to know what all that entails? My main thought ATM is the 50 req/sec per bot. Don't want to have to worry about that being an issue (way down the road ofc), but gotta know so I can plan better.
pearl trail
Wdym
like, what are you trying to do here? tunneling local port to be accessible through domain?
quartz kindle
google gave me extra access to gemini 3 pro because i have the 2tb google drive plan, time to test their new antigravity code editor
lyric mountain
aight, day 3 done too
println input
.collect { it.split('') as List<String> }
*.with {
def head = it.dropRight(1).max()
return head + it.drop(it.indexOf(head) + 1).max() as int
}
.sum()
the one that wiped someone's disk?
quartz kindle
the one that wiped someone's disk?
lmao
queen needle
```groovy
println input
.collect { it.split('') as List<String> }
*.with...
how the hell
queen needle
How is it so small
lyric mountain
oh it's simple, u dont need to compare every number against eachother, simply get the largest number in the row (that's not the last index) then do it again but starting from the previous index + 1
queen needle
oh my god
im a little dumb
for (const line of lines) {
let arrayOfNumbers = line.split("").map(Number);
let currentLargestJoltage = 0;
for (let i = 0; i < arrayOfNumbers.length; i++) {
const number = arrayOfNumbers.at(i)!;
const mutatedArray = arrayOfNumbers.slice(i + 1);
for (let j = 0; j < mutatedArray.length; j++) {
const compareNumber = mutatedArray.at(j)!;
if (!compareNumber) continue;
const finalNumber = number * 10 + compareNumber;
if (finalNumber > currentLargestJoltage) {
currentLargestJoltage = finalNumber;
}
}
}
joltageSum += currentLargestJoltage;
}```nah the important part is whether u got the correct answer
queen needle
How did you do day3 part 2?
queen needle
When you solve the first one, and put in your answer, it opens up part 2
lyric mountain
oh I didn't see it
queen needle
oh it's simple, u dont need to compare every number against eachother, simply ge...
for (const line of lines) {
let arrayOfNumbers = line.split("").map(Number);
const largest = Math.max(...arrayOfNumbers.slice(0, -1));
joltageSum += (largest * 10 + Math.max(...arrayOfNumbers.slice(arrayOfNumbers.indexOf(largest) + 1)));
}``` i tried that and just get a little bit higher than my correct answernevermind i fixed it, for some reason i have to do line.trim()
lyric mountain
hm, for the second part I can use the same code I think
just need to add support for nth values
queen needle
i would love to see what you come up with
i struggled a lot
but eventually got there
although i am confused ```ts
import * as fs from "fs";
type Line = ${number};
const inputFile = "input.txt";
const fileContent = fs.readFileSync(inputFile, "utf-8");
const lines = fileContent.trim().split("\n") as Line[];
let joltageSum = 0;
for (const line of lines) {
let arrayOfNumbers = line.trim().split("").map(Number);
const largest = Math.max(...arrayOfNumbers.slice(0, -1));
joltageSum += (largest * 10 + Math.max(...arrayOfNumbers.slice(arrayOfNumbers.indexOf(largest) + 1)));
}
console.log(joltageSum);
``` returns 16927 if i remove .trim() it returns 16946 and there is no empty space in my input.txt so im confused as to how this happens
clever tundra
like, what are you trying to do here? tunneling local port to be accessible thro...
Yes, but so that the (my) IP doesn't get publicly shown
swift barn
Yes, but so that the (my) IP doesn't get publicly shown
If it returns http 404 that’s normal thing for cf tunnelling
pearl trail
Yes, but so that the (my) IP doesn't get publicly shown
have you configured something like this?
clever tundra
If it returns http 404 that’s normal thing for cf tunnelling
Nope
clever tundra
have you configured something like this?
Let me check
clever tundra
have you configured something like this?
Is that zero trust or normal
clever tundra
mine is a bit different to yours
ahhhh i found it
aaaa nevermind it already existed
yeah
lyric mountain
i would love to see what you come up with
int digits = 12
println input
.collect { it.split('') as List<String> }
*.with {
def prio = it.toUnique().toSorted().reversed()
def queue = it
return (0..<digits)*.with {i ->
for (p in prio) {
int v = queue.dropRight(digits - i - 1).indexOf(p)
if (v > -1) {
queue = queue.drop(v + 1)
return p
}
}
throw "Reached end of input" as Throwable
}.join() as long
}
.sum()
basically u make a list of priorities, then scan the digits excluding n - 1 indexes from the tail
queen needle
for (const line of lines) {
const arrayOfNumbers = line.trim().split("").map(Number)
let currentLargestJoltage = 0;
let maxId = -1;
const poweredBatteries = 12;
for (let i = 0; i < poweredBatteries; i++) {
maxId++;
if (maxId >= arrayOfNumbers.length) break;
let max = arrayOfNumbers.at(maxId)!;
let upperBound = arrayOfNumbers.length - (poweredBatteries - i - 1);
for (let j = maxId + 1; j < upperBound; j++) {
if (arrayOfNumbers.at(j)! > max) {
max = arrayOfNumbers.at(j)!;
maxId = j;
}
}
currentLargestJoltage = currentLargestJoltage * 10 + max;
}
joltageSum += currentLargestJoltage;
}```and every time u find the value u shorten the head so u always start from the same place
clever tundra
have you configured something like this?
yes
queen needle
and every time u find the value u shorten the head so u always start from the sa...
That's a nice approach, are you going to keep doing it every day, your solutions are very nice
lyric mountain
probably, cant stop at 4 stars 
queen needle
good
lyric mountain
That's a nice approach, are you going to keep doing it every day, your solutions...
tyyy
queen needle
Groovy is so small compared to java
lyric mountain
yea, I love that lang ngl
it's very sugary tho, like, nearly everything has a "groovy way"
queen needle
make a github repo with all your solutions
lyric mountain
I did
queen needle
Ohh i thought I was following you on github but i wasn't
pearl trail
yes
weird that it shows healthy on the dashboard,, do you use host-installed cloudflared or through docker?
quartz kindle
you guys doing advent of code? i always forget that it exists lol
catching up now, did day 2 and 3, didnt do day 1 yet
anyways thought id share mine too xD
day 3 part 1:
input.split("\n").reduce((a,t) => {
const first = Math.max(...t.slice(0, -1));
const second = Math.max(...t.slice(t.indexOf(first) + 1));
return a + Number(`${first}${second}`);
}, 0);
day 3 part 2:
input.split("\n").reduce((a,t) => {
const out = [Math.max(...t.slice(0, -11))];
for(let i = 0, n = t.indexOf(out.at(-1)) + 1; i < 11; i++) {
const s = t.slice(n, -(10-i) || undefined);
const d = Math.max(...s);
n = t.indexOf(d, n) + 1;
out.push(d);
}
console.log(t, out)
return a + Number(out.join(""));
}, 0);
first option, note its not currently running tho since i didnt want to flood my pi with errors
twas degraded yesterday
when i try to run it i use:
cloudflared tunnel --protocol http2 run dashboard
if i remove --protocol http2 i get:
error="already connected to this server, trying another address"
if i go back to http2 i get:
2025-12-03T17:27:23Z ERR Unable to establish connection with Cloudflare edge error="DialContext error: dial tcp 198.censor:7844: i/o timeout" connIndex=0 event=0 ip=198.censor 2025-12-03T17:27:23Z ERR Serve tunnel error error="DialContext error: dial tcp 198.4censor:7844: i/o timeout" connIndex=0 event=0 ip=198.censor 2025-12-03T17:27:23Z INF Retrying connection in up to 2s connIndex=0 event=0 ip=198.censor
sometimes i get a mix of both errors
queen needle
anyways thought id share mine too xD
day 3 part 1:
```js
input.split("\n").redu...
I am glad im not the only one who forgot to just do first * 10 + second and just made a template string wtih them and wrapped it in number
quartz kindle
I am glad im not the only one who forgot to just do `first * 10 + second` and ju...
lmao, could do that but we are not really going for performance this time
just whatever gets the job done and its not overcomplicated
my shit for day 2 was absolute garbage that took some solid 10 seconds to run
quartz kindle
day 2 part 2:
n = 0;
"4487-9581,755745207-755766099,954895848-955063124,4358832-4497315,15-47,1-12,9198808-9258771,657981-762275,6256098346-6256303872,142-282,13092529-13179528,96201296-96341879,19767340-19916378,2809036-2830862,335850-499986,172437-315144,764434-793133,910543-1082670,2142179-2279203,6649545-6713098,6464587849-6464677024,858399-904491,1328-4021,72798-159206,89777719-90005812,91891792-91938279,314-963,48-130,527903-594370,24240-60212".split(",").forEach(x => {
const [start, end] = x.split("-").map(x => Number(x));
for(let i = start; i < end + 1; i++) {
const str = i.toString();
const z = str.length;
const divisors = [...Array(z + 1).keys()].slice(1).filter(i => z % i === 0);
for(const div of divisors) {
const match = str.match(new RegExp(`.{1,${div}}`, 'g'));
if(match.length > 1 && new Set(match).size === 1) { n += i; break; }
}
}
});
console.log(n)
lmao
i just hardcoded the input
queen needle
interesting
import * as fs from "fs";
type Line = `${number}-${number}`;
const inputFile = "input.txt";
const fileContent = fs.readFileSync(inputFile, "utf-8");
const lines = fileContent.trim().split(",") as Line[];
let totalWrongIds = 0;
for (const line of lines) {
const [lowerRange, upperRange] = line.split("-").map(Number) as [
number,
number
];
for (let i = 0; i <= upperRange - lowerRange; i++) {
const id = String(lowerRange + i);
const doubled = id + id;
const substring = doubled.slice(1, -1);
if (substring.includes(id)) {
totalWrongIds += parseInt(id);
continue;
}
}
}
console.log(totalWrongIds);```this was mine
pearl trail
when i try to run it i use:
`cloudflared tunnel --protocol http2 run dashboard`
waht why does that look so complicated. doesn’t you only need to put token given when creating connector? basically just cloudflared service install the-token
or you avoid this method?
pearl trail
either way regardless of what I do i always get error 1033
that’s what happen when cloudflare is unable to reach your server
idk why the connector status is healthy, but you might want to redo installing the connector
clever tundra
well i tried again removing --protocol http and get a mix of i/o timeout, already connected, and about 2 registered connection
clever tundra
idk why the connector status is healthy, but you might want to redo installing t...
alright, so just new connector?
pearl trail
that works too, make sure that you use this one for simplicity
then you will never need to touch your server, you manage everything on the cf's dashboard
clever tundra
yeah im just deleting the old tunnel and starting fresh lmao
lyric mountain
damn day 1 part 2 had hands
int zeros, dial = 50
(input =~ /([LR])(\d+)/).each { _, dir, num ->
int val = num as int
if (val > 99) {
zeros += val / 100 as int
val %= 100
}
int rot = dial + val * (dir == 'R' ? 1 : -1)
if (rot < 0) {
rot += 100
if (dial != 0) zeros++
}
dial = (rot + 100) % 100
zeros += dial ? (rot.abs() / 100 as int) : 1
}
println zeros
yeah i just did another loop instead of doing it the mathmatical way
queen needle
thats for part 1 or 2?
part 2
queen needle
although i am confused ```ts
import * as fs from "fs";
type Line = `${number}`;...
also, any idea @quartz kindle
quartz kindle
although i am confused ```ts
import * as fs from "fs";
type Line = `${number}`;...
which trim causes the change?
lyric mountain
bro, day 2 part 2 was so silly lmao
all I had to do is add ONE character
regex is op
queen needle
which trim causes the change?
let arrayOfNumbers = line.trim().split("").map(Number);
lyric mountain
from /^(.+)\1$/ to /^(.+)\1+$/
queen needle
Yeah i wish I woud have done regex after seeing everyone elses solutions
quartz kindle
`let arrayOfNumbers = line.trim().split("").map(Number);`
if you remove that trim your result changes?
the only way that would happen is if somehow your file has spaces in it for some reason
i tested your code in mine and there was no difference with or without trims
queen needle
hmm
quartz kindle
ah yes, if your file has \n\r newlines because windows, then yeah
queen needle
what's odd is that for part 1 of today it works with or without change, but if I use my part 2 solution but target 2 digits instead of 12, it needs trim
quartz kindle
```ts
import * as fs from "fs";
type Line = `${number}-${number}`;
const input...
btw this solution is so much better than whatever the fuck i did yesterday lmao
queen needle
You should make a most performant solution for each one
quartz kindle
lmao
quartz kindle
they are telling me to run
queen needle
run as in run the most performant code
lyric mountain
what's odd is that for part 1 of today it works with or without change, but if I...
for 2 digits, you'd get 2 + 3
for 3 digits, you'd get 2 + 3 + 0
queen needle
so should I split my input into lines my \r\n
quartz kindle
just hardcode the string
queen needle
well what's also weird is this \r\n issue only appears with certain solutions
quartz kindle
icba setting up a code editor for function prototyping lmao
lyric mountain
queen needle
I just don't understand how 1 needs the trim and one doesn't
both ouput the correct answer
lyric mountain
try console.loging the number arrays
also use the example so u dont get printed a huge array
quartz kindle
in the second one you dont actually traverse the entire line
you just filter and extract high numbers from it
so the zeroes that are appended from the \r doesnt affect anything
queen needle
but part 2 is the one that needs teh trim?
quartz kindle
oh i thought you meant part 1 needs the trim
queen needle
did you mean part 1
quartz kindle
I just don't understand how 1 needs the trim and one doesn't
i read that as 1 needs the trim, as in part 1 needs it lmao
queen needle
because printing the array in part 1, I see what you mean ts [ 9, 8, 7, 6, 5, 4, 3, 2, 1, 1, 1, 1, 1, 1, 1, 0 ] [ 8, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 9, 0 ] [ 2, 3, 4, 2, 3, 4, 2, 3, 4, 2, 3, 4, 2, 7, 8, 0 ] [ 8, 1, 8, 1, 8, 1, 9, 1, 1, 1, 1, 2, 1, 1, 1 ]
lyric mountain
for both parts u can use the same code, as in, your solution for part 2 must work for any n < length
queen needle
ahh i see now
is there anyway to do it without doing \r\n without having to have an anonymous async function?
queen needle
so just .split(/\n/g)
queen needle
Well i want to try and find a global solution that works no matter the input
because i use the same boilerplate day to day
lyric mountain
\n should work with regex
queen needle
Wait but why does the .trim have to be on the line when i already trim the file contents
lyric mountain
cuz ur reading the whole file
lyric mountain
nvm u said ends with an s
dire valve
guys where can i generate an api key
queen needle
you meant like this right, const lines = fileContent.trim().split(/\n/g) as Line[];?
lyric mountain
guys where can i generate an api key
u need ur bot to be approved on topgg first
dire valve
u need ur bot to be approved on topgg first
how do i do that
lyric mountain
submit said bot
dire valve
ok ty
yo im sry to ask but where do i do that 😭
no option to create anything here
lyric mountain
you meant like this right, `const lines = fileContent.trim().split(/\n/g) as Lin...
yes, u dont need to trim the content if u dont put an empty line in the file
lyric mountain
yo im sry to ask but where do i do that 😭
dire valve
ty
dire valve
https://support.top.gg/hc/en-us/articles/23135162935708-How-to-Add-Your-Bot
Bro i dont want a bot i want just the api to scrape list of discord servers
is that allowed?
lyric mountain
no
lyric mountain
well, even with access to the api you'd not be able to list the servers
since there's no endpoint for that afaik
lyric mountain
odd
queen needle
thank you I never would have thought \r being the issue
lyric mountain
yw
patent bramble
Does anyone here know if "Embed Links" is needed for ephemeral responses? My guess is no since it's only a response and less "advertisement"
lyric mountain
embed links afaik is only necessary to send embeds
patent bramble
embeds wouldn't be links though
embeds are just messages with style
links direct users to sites
lyric mountain
no no, you see, even without that perm u can send links size they're text
when I say "embeds" I mean this block thing
patent bramble
or you mean the link preview embed yeah
lyric mountain
without that perm u can send the link, but the embed wont show
patent bramble
oop I best check that lmao
might be an issue
o shit u rite
I need to update that lmao
lyric mountain
it's confusing af the naming of that perm
patent bramble
should just be embeds tbh
lyric mountain
or SEND_EMBEDS
patent bramble
tru
I love how we're supposed to account for the new BYPASS_SLOWMODE but it's not even an option for installing apps
or are they exempt from slowmode?
Note: This primarily affects users, as bots are not affected by slowmode restrictions.
ok nvm
stark kestrel
```ts
for (const line of lines) {
let arrayOfNumbers = line.split("").map(Numb...
Part 1 was fun
Part 2 made me rewrite part 1 lmao
queen needle
Lmaooo
stark kestrel
I also got the max then the max after the first max for part 1
queen needle
I've seen that for most people, they wrote one, and then wrote two so it was dynamic
stark kestrel
Part 2 I was like 
fn get_max_joltage(battery: &Vec<usize>, n: usize) -> usize {
let mut stack = Vec::with_capacity(n);
let mut possible_removals = battery.len() - n;
for &row in battery {
while possible_removals > 0 && !stack.is_empty() && *stack.last().unwrap() < row {
stack.pop();
possible_removals -= 1;
}
stack.push(row);
}
stack.truncate(n);
stack.iter().fold(0, |value, &elem| value * 10 + elem)
}
ended up with that
queen needle
Today makes me miss the global leaderboard
you would have seen the people who wrote part 1 dynamic vs those who didn't
stark kestrel
and i first read wrong, the word column triggered me
so i read the columns as being the battery
stark kestrel
you would have seen the people who wrote part 1 dynamic vs those who didn't
yeah
queen needle
```rs
fn get_max_joltage(battery: &Vec<usize>, n: usize) -> usize {
let mut ...
Part 1/2?
stark kestrel
n=2 for part 1
n=12 for 2
fn part_one(test: bool) -> usize {
let columns = aoc_rs::input::read_file_grid_usize(aoc_rs::input::get_path(3, test));
columns.iter().map(|col| get_max_joltage(col, 2)).sum()
}
fn part_two(test: bool) -> usize {
let columns = aoc_rs::input::read_file_grid_usize(aoc_rs::input::get_path(3, test));
columns.iter().map(|col| get_max_joltage(col, 12)).sum()
}
usually I stay up until midnight to do them but last night I passed out early, and I feel like if I stayed up and saw part 2 I would have just gone to bed
queen needle
well I used to them in my 1st hour during school but now I have college where only 2 in person classes in the week, both later in the afternoon
Last year I got to the leaderboard a couple of times so like I'll never top that
wheat mesa
you would have seen the people who wrote part 1 dynamic vs those who didn't
you would've seen AI bots at the top lol
queen needle
you would've seen AI bots at the top lol
Yeah that's part of why it was removed but still miss it
prime cliff
🙂 got permissions mapped out
queen needle
Awesome
shell tundra
🙂 got permissions mapped out
this is so cool, what is this, ur own chat app?
prime cliff
Yup
shell tundra
🔥
warm surge
🙂 got permissions mapped out
builder
prime cliff
Oh hi you've been released from shawshank
warm surge
Oh hi you've been released from shawshank
funny
patent bramble
Does Discord have any built in insights for apps to track growth/engagement? 👀
pearl trail
i am not sure about that
patent bramble
Is there a list of the error codes returned by the API for 4XX errors? For instance 670006 is missing guild feature (specifically gradient, if it's that specific).
patent bramble
https://gist.github.com/Dziurwa14/de2498e5ee28d2089f095aa037957cbb
Nice, thanks. Kinda odd to copy from the site to a repo but I guess data retention if anything. Assuming discord food isn't official lol
spark flint
yeah its unoffficial
.food is for user API errors
but its. a big list of every known API error regardless of bot / user
patent bramble
nice
patent bramble
.food is for user API errors
Like everywhere? Or just Discord?
spark flint
just discord
patent bramble
ah lol gotcha
queen needle
Today's advent of code was hard lmao
lyric mountain
oooo I once coded smth similar to today's puzzle
queen needle
oooo I once coded smth similar to today's puzzle
You shouldn't show the inputs day to day on GitHub
lyric mountain
aren't the inputs individually generated?
queen needle
queen needle
aren't the inputs individually generated?
Yes, but I think he doesn't want people to show them because then they could get reverse engineered or something
queen needle
oooo I once coded smth similar to today's puzzle
Minesweeper is as close as I've done lol
stark kestrel
aight I'll gitignore them
I did
# AoC input files that are not test
/**/input/*
!/**/input/*_test.txt
so i push the test input and use that for the tests in code
repo is currently private either way but yeah
gonna see what day 4 looks like tonight
queen needle
Day 4 is fun
lyric mountain
aight, done
this one ended longer than usual
def input = getClass().getResource('/day4.txt').readLines()
boolean COUNT_ONLY = false
def mat = input.flatten { it.split('') as List<String> } as List<String>
int cols = input.first.length()
int rows = input.size()
int total = 0
boolean loop = true
while (loop) {
loop = false
for (i in mat.indices) {
if (mat[i] == '@') {
def (x, y) = [i % cols, i / cols as int]
int count = (Math.max(0, y - 1)..Math.min(y + 1, rows - 1)).collect {
int col = cols * it
return mat.subList(
col + Math.max(0, x - 1),
col + Math.min(x + 2, rows)
).count { it == '@' }
}.sum() as int
if (count <= 4) {
total++
if (!COUNT_ONLY) {
mat[i] = '.'
loop = true
}
}
}
}
}
println total
lol the second one was just a tiny change in the code
part 1:
n = 0;
cells = document.body.innerText.split("\n").map(x => x.split(""));
grid = [cells.length, Math.max(...cells.map(x => x.length))];
for(let y = 0; y < grid[0]; y++) {
for(let x = 0; x < grid[1]; x++) {
const current = cells[y][x];
const adjacent = [cells[y-1]?.[x-1],cells[y-1]?.[x],cells[y-1]?.[x+1],cells[y]?.[x-1],cells[y]?.[x+1],cells[y+1]?.[x-1],cells[y+1]?.[x],cells[y+1]?.[x+1]];
if(current === "@" && adjacent.filter(x => x === "@").length < 4) {
n++;
}
}
}
console.log(n)
part 2:
n = 0;
cells = document.body.innerText.split("\n").map(x => x.split(""));
grid = [cells.length, Math.max(...cells.map(x => x.length))];
while(true) {
const prev = n;
for(let y = 0; y < grid[0]; y++) {
for(let x = 0; x < grid[1]; x++) {
const current = cells[y][x];
const adjacent = [cells[y-1]?.[x-1],cells[y-1]?.[x],cells[y-1]?.[x+1],cells[y]?.[x-1],cells[y]?.[x+1],cells[y+1]?.[x-1],cells[y+1]?.[x],cells[y+1]?.[x+1]];
if(current === "@" && adjacent.filter(x => x === "@").length < 4) {
n++;
cells[y][x] = ".";
}
}
}
if(n === prev) { break; }
}
console.log(n)
I was stumped on the second one for sometime, and then I realized I just need to change my function to return a count, and put it all in a while loop
queen needle
part 1:
```js
n = 0;
cells = document.body.innerText.split("\n").map(x => x.spli...
The grid length doesn't change
quartz kindle
things would be totally different if we went for performance, but since we dont care about performance, we just brute force xd
queen needle
you don't need to find the maximum
quartz kindle
The grid length doesn't change
ye i figured but i still left it there
queen needle
things would be totally different if we went for performance, but since we dont ...
I don't know what the most performant solution would look like
I saw some people talking about it, but didn't see anyone post there's that was
quartz kindle
you need to do what you can to keep it in a single loop, for example on each find only backtrack enough to check the cells that could be affected by the swap and then proceed from where you left off
there are also tricks to improve adjacent checking so you dont need to check all 8 cells
queen needle
how so?
Also I screwed myself over for so long, I used .at instead of [index] and that caused it to grab the very last row on the first one because .at handles negatives and it took me way too long to realize and fix it
queen needle
Pfft I didn't know that
I used .at because function syntax is nicer imo and I assumed it had some performance things under the hood
quartz kindle
general rule in js is new things are always more convenient but slower than old things
queen needle
that's not fun
quartz kindle
then with time they improve the new things so that they are only 1% slower instead of 200% slower
i dont think there is a single new thing in js that was introduced to make something faster
queen needle
there are also tricks to improve adjacent checking so you dont need to check all...
The only way is not checking ones they don't exist like not checking above if there isn't and such
quartz kindle
The only way is not checking ones they don't exist like not checking above if th...
well for one you can check only until you are sure something can or cannot be
if out of 8 fields, the first 4 alreadt have a roll, you can skip the rest
if out of 8 fields, 5 are empty, you can skip the last 3
otherwise you can also try a minesweeper style of marking the number of adjacents on each cell so you dont need to check certain neighbours again
these are all more complex to do, but performance-wise they would be quite effective
queen needle
I saw lots of people using queues today
queen needle
otherwise you can also try a minesweeper style of marking the number of adjacent...
when I read today's the first thing that came to mind was finding my neighbors function from minesweeper I made like 2 years ago
I also originally dynamically made the adjacent cells array
quartz kindle
generally, advanced algorithms all look for doing as much as they can in a single pass and avoid multiple passes as much as possible, so it involves a lot of going back and forth with indexes and stuff, remembering positions/pointers, resuming where it left of, etc...
this goes back to the old ways of seeking HDDs, it would be much faster to work with nearby data than to go back to the beginning and scan the entire thing again
queen needle
I wish I found an efficient algorithm for today
I just check every character, check the 8 adjacent, and then move on to the next character
quartz kindle
yeah thats what i did too
its the fastest way to prototype a working function / solve the problem
lyric mountain
Do you not do any grid handling
I prefer to handle 1d arrays when working with grids
queen needle
well I mean it's the only way I could really think to do it, only micro optimization I know how to do would be to only check possible cells so don't check above if it isn't there
queen needle
I prefer to handle 1d arrays when working with grids
Interesting
I wish I could optimize my day 2
It takes like 1.5s
lyric mountain
I used the scan method
queen needle
Scan method?
lyric mountain
as in, do a pass -> if anything is removed -> repeat
so I do a full scan to remove any valid slots, and if anything is removed, I scan again to check if any can be removed now
the loop only ends when no slots are removed
this also lets me use the same func for both parts, since it wont loop again
btw done removing the inputs from git history
twilit knot
Chat what I'm supposed to do here? I just want my bot to have working cmds and a command button type thing on the profile, My bot have like 9 working slash cmds
lyric mountain
mine takes abt 1.25s and 37 full scans
ran the entire task 10 times beforehand to let jit do its thing
lyric mountain
Chat what I'm supposed to do here? I just want my bot to have working cmds and a...
you upload the entire slash cmd json (the one u sent to discord to register them)
quartz kindle
mine takes abt 1.25s and 37 full scans
mine did 46 passes in 85ms
lyric mountain
wtf
twilit knot
you upload the entire slash cmd json (the one u sent to discord to register them...
Wait I coded the bot in python and I just like defined the cmds and all in the py file and it works as perfectly, does it count as registered and Im supposed to paste that in box right?
lyric mountain
the string is hardcoded
I didn't time the init part, only the looping
lyric mountain
but well, u also hardcoded the cell accesses, that's prolly faster than extracting views
lyric mountain
Wait I coded the bot in python and I just like defined the cmds and all in the p...
there's probably a way to extract the commands json in d.py
twilit knot
there's probably a way to extract the commands json in d.py
Oh alright
quartz kindle
nvm strings are not mutable
lyric mountain
just realized, I might be able to reduce the amount of scans by jumping to the first subview index, instead of index 0
queen needle
nvm strings are not mutable
I kept having this issue
That's why I have this mess
I need to create a timing system for each of these
twilit knot
there's probably a way to extract the commands json in d.py
Did this but it still don't show up on my bot profile let alone clickable buttons
lyric mountain
Did this but it still don't show up on my bot profile let alone clickable button...
it wont show in ur bot profile cuz topgg isn't discord
slash commands show in your bot profile when they're used frequently
twilit knot
it wont show in ur bot profile cuz topgg isn't discord
How this bot got it then? I just want it to be like this
twilit knot
slash commands show in your bot profile when they're used frequently
Ohh
Makes sense
quartz kindle
@lyric mountain@queen needle
from ~85ms to ~15ms (can still be further improved)
function THEJOB(input) {
let n = 0;
const grid = input.split("\n").map(x => x.split(""));
const size = [grid.length, grid[0].length];
const marks = [];
for(let y = 0; y < size[0]; y++) {
for(let x = 0; x < size[1]; x++) {
CHECK(y, x);
}
}
while(marks.length) {
const m = [...marks];
marks.length = 0;
for(const mark of m) {
for(let i = 0; i < 9; i++) {
const ny = mark[0] - 1 + Math.floor(i / 3);
const nx = mark[1] - 1 + (i % 3);
CHECK(ny, nx);
}
}
}
function CHECK(y, x) {
const current = grid[y]?.[x];
if(!current) { return; }
const adjacent = [
grid[y-1]?.[x-1], grid[y-1]?.[x], grid[y-1]?.[x+1],
grid[y]?.[x-1], grid[y]?.[x+1],
grid[y+1]?.[x-1], grid[y+1]?.[x], grid[y+1]?.[x+1]
];
const l = adjacent.filter(x => x === "@").length;
if(current === "@" && l < 4) {
n++;
grid[y][x] = ".";
marks.push([y,x]);
}
}
return n;
}
one full scan, remember all the changes, then followup scans only check adjacents of changed cells
lyric mountain
one full scan, remember all the changes, then followup scans only check adjacent...
yeah this sped up a lot on my code too
now it's 1.12s
oh wait, I forgot to skip the full scan
nvm, didn't matter
prime trellis
why everyone still using old components for embeds?
quartz kindle
@queen needle@lyric mountain down to 4ms
function THEJOB(input) {
let n = 0;
const grid = input.split("\n").map(x => x.split(""));
const size = [grid.length, grid[0].length];
let marks = [];
for(let y = 0; y < size[0]; y++) {
for(let x = 0; x < size[1]; x++) {
CHECK(y, x);
}
}
while(marks.length) {
const length = marks.length;
for(let m = 0; m < length; m++) {
const mark = marks[m];
const y = mark >> 8;
const x = mark & 255;
CHECK(y-1, x-1);
CHECK(y-1, x);
CHECK(y-1, x+1);
CHECK(y, x-1);
CHECK(y, x+1);
CHECK(y+1, x-1);
CHECK(y+1, x);
CHECK(y+1, x+1);
}
marks = marks.slice(length);
}
function CHECK(y, x) {
const gy = grid[y];
if(gy) {
const gx = gy[x];
if(gx === "@") {
let brazil = 0;
const gyminus = grid[y-1];
const gyplus = grid[y+1];
if(gy[x-1] === "@") {brazil++}
if(gy[x+1] === "@") {brazil++}
if(gyminus) {
if(gyminus[x-1] === "@") {brazil++}
if(gyminus[x] === "@") {brazil++}
if(gyminus[x+1] === "@") {brazil++}
}
if(gyplus) {
if(gyplus[x-1] === "@") {brazil++}
if(gyplus[x] === "@") {brazil++}
if(gyplus[x+1] === "@") {brazil++}
}
if(brazil < 4) {
n++;
gy[x] = ".";
marks.push((y << 8) + x);
}
}
}
}
return n;
}
I genuinely do not understand my code or the algorithm you two describe enough to speed my solution up
lyric mountain
ah cmon
queen needle
Bro literally wtf
quartz kindle
caching variables as much as possible, avoiding short loops, avoiding dead/unneeded code, avoiding unnecessary array allocations
queen needle
brazil++ is peak
lyric mountain
I genuinely do not understand my code or the algorithm you two describe enough t...
when you remove an index, save the nearby indexes to a list
next time u iterate, run over these instead
prime trellis
```js
function THEJOB(input) {
let n = 0;
const grid = input.split("\n")...
u wanna help with that?
quartz kindle
bonus, store y x coordinates as bitfield xD
quartz kindle
u wanna help with that?
feel free to further improve it
lyric mountain
dont see u doing that
prime trellis
check dm @quartz kindle
@quartz kindle:
VM67:170 Result: 9446 @ removed
VM67:171 Time: 4.10 ms
VM67:172 Speed: 9756.1 cells/ms
VM67:175
v2:
VM67:180 Result: 9446 @ removed
VM67:181 Time: 3.00 ms
VM67:182 Speed: 13333.3 cells/ms
quartz kindle
show code?
queen needle
AI can't compare to Tim
prime trellis
asking AI to do it?
lol
queen needle
AI can't compare to Tim
-# yet™
neon leaf
quartz kindle
im asking chatgpt to improve mine and see how it does
prime trellis
im using deepseek
quartz kindle
so far it produced 1 wrong code, and 1 slower code
prime trellis
30ms faster than u now
quartz kindle
lmao
prime trellis
let me see opus
quartz kindle
show then
queen needle
I Wanna see
prime trellis
i think opus can do better
queen needle
I don't see many improvements left in his code of any at all
neon leaf
rewrite it in rust ofc
quartz kindle
xD
queen needle
oh true true
prime trellis
XD
queen needle
rust in the browser console
quartz kindle
so far every solution chatgpt gave me was slower or equivalent to mine
but tbh this is too small of a job to get meaningful results
you would need a much larger input to get a better assessment
queen needle
make your own input
queen needle
input * (input.length ** input.length)
quartz kindle
input.repeat(100)
quartz kindle
xD
queen needle
time complexity from hell
quartz kindle
chances are 12 for loops are faster than 90% of other methods out there
prime trellis
ultra = opus
quartz kindle
ultra = opus
is the code actually being tested or is AI just giving you those numbers?
because AI doesnt actually run the code
it just gives you random benchmark results
queen needle
And is it giving the right answer still
quartz kindle
cool, can you show the code? wanna test here too
prime trellis
he skip u
quartz kindle
lmao
prime trellis
queen needle
send it in a https://sourceb.in
prime trellis
with test everything i just copy that
prime trellis
I wanna look but I'm on mobt
quartz kindle
nice
queen needle
Don't generate a grid, use the grid from aoc
prime trellis
Don't generate a grid, use the grid from aoc
idc for real i just put into deepseek to see
just go into deepseek and test it too
sometimes is good
queen needle
..@@.@@@@.
@@@.@.@.@@
@@@@@.@.@@
@.@@@@..@.
@@.@@@@.@@
.@@@@@@@.@
.@.@.@.@@@
@.@@@.@@@@
.@@@@@@@@.
@.@.@@@.@.```Test the fastest solution with this input
quartz kindle
a lot of the changes are not actually meaningful, what actually matters is that it flattened the 2d array into a single continuous block and then used multiplication to simulate rows and colummns
prime trellis
Test the fastest solution with this input
let me test it
quartz kindle
for example instead of
abc
xyz
123
it stored as
abcxyz123
and then it gets rows and columns like this
cell = block[rowindex * rowsize + colindex]
queen needle
a lot of the changes are not actually meaningful, what actually matters is that ...
That's an interesting approach
queen needle
for example instead of
abc
xyz
123
it stored as
abcxyz123
and then it gets ro...
well you gotta keep opTIMizing, can't get beat by AI
prime trellis
```
..@@.@@@@.
@@@.@.@.@@
@@@@@.@.@@
@.@@@@..@.
@@.@@@@.@@
.@@@@@@@.@
.@.@.@.@@@...
tim 0.10 deepseek 0.100
neon leaf
thats the same value
prime trellis
yeah
prime trellis
?????
prime trellis
prime trellis
I just want to know if it gets the right answer 😭
ok 1sec
VM40:198
:mag: TESTING ORIGINAL VERSION...
VM40:203 Result: 43 @ removed
VM40:204 Time: 0.10 ms
VM40:205 Speed: 1000.0 cells/ms
VM40:208
:zap: TESTING ULTRA FAST VERSION...
VM40:213 Result: 43 @ removed
VM40:214 Time: 0.100 ms
VM40:215 Speed: 1000.0 cells/ms
VM40:218
:bar_chart: PERFORMANCE COMPARISON:
VM40:219 Results match: :white_check_mark: YES
VM40:220 Speedup: 1.0x faster
VM40:221 Time saved: 0.00 ms
VM40:224
:chart_with_upwards_trend: GRID ANALYSIS:
VM40:233 Size: 10×10 = 100 cells
VM40:234 @ symbols: 71 (71%)
VM40:235 . symbols: 29 (29%)
VM40:238
:dart: GRID AFTER FIRST REMOVAL WAVE:
VM40:273
After first removal wave (only @ with <4 neighbors):
VM40:274 .......@..
.@@.@.@.@@
@@@@@...@@
@.@@@@..@.
.@.@@@@.@.
.@@@@@@@.@
.@.@.@.@@@
..@@@.@@@@
.@@@@@@@@.
....@@@...
VM40:277
:abacus: SUMMARY:
VM40:279 Initial @ count: 71
VM40:280 Total removed @: 43
VM40:281 Remaining @: 28
VM40:282 @ after first wave: 58
VM40:283 Additional removed in propagation: 30
VM40:285
==================================================
VM40:286 :checkered_flag: TEST COMPLETE```Speed: 1000.0 cells/ms
prime trellis
a lot of the changes are not actually meaningful, what actually matters is that ...
and yes
prime cliff
is the code actually being tested or is AI just giving you those numbers?
Not only that but AI can also hallucinate and make up any random shit you give it even things like benchmarks or "performance" improvements
prime trellis
any feedback?
prime cliff
Do you really need to add "secure oauth 2.0" xD like what does 2.0 mean?
quartz kindle
yeah it often halucinates about what improves performance and what doesnt. for example it will say "this change massively improves performance" and it actually makes it worse.
AI is good at finding good algorithms, but not as good at microoptimizing
neon leaf
cuz the standard is oauth2 ig
prime cliff
Regular users don't need to know about what oauth2 is though
queen needle
Why have a pill button in a slightly rounded container, and why does the button have a slight border but transparent background? It makes the button look odd
prime trellis
Do you really need to add "secure oauth 2.0" xD like what does 2.0 mean?
version from my pc...
i making few login page
and pushing it into community
queen needle
Are you making it or is AI 🤔🤔🤔
prime trellis
front opus
pancake and tim why u are mad on ai?
tell me
u dont like it?
helps alot btw
changed @prime cliff all happy rn
prime cliff
I would just get rid of the text tbh
queen needle
^
prime trellis
yea?
queen needle
The standard for login seperator components is to provide Oauth methods after you already have a login/sign up screen
queen needle
If it's just one login method you do not really need a seperator
prime cliff
Only you and other developers will know what "oauth" is and at that point they already know what oauth is unless you're building a developer oriented app
queen needle
and why separate the button and the text about the button?
prime cliff
The seperator and text is better imo design wise
queen needle
The seperator and text is better imo design wise
That's fair, I mean it kind of works in this scenario but also the button being so large but providing so little, being a pill with a transparent background, and the text being so far away, it all just creates empty space and some noise when it isn't needed
prime trellis
okay @queen needle
u wanna give me feedback about 1 more thing?
or o @prime cliff what u think about that?
queen needle
I mean I can but my feedback is very biased @prime cliff is a very good developer who would be more help
prime cliff
I'm not that good ;-;
queen needle
Better than me at least
prime trellis
or o <@190590364871032834> what u think about that?
i dont like Instant Server Setup but idk
prime trellis
I'm not that good ;-;
go ahead 😄
prime trellis
Better than me at least
go ahead pancake u are good ❤️
queen needle
I'll have to wait until after class the video is too small on mobile
prime cliff
or o <@190590364871032834> what u think about that?
Nice but is it mobile compatible :p
prime trellis
Nice but is it mobile compatible :p
ofc
prime trellis
Nice but is it mobile compatible :p
queen needle
Nice but is it mobile compatible :p
Imagine
prime trellis
everything is compatible with mobile version...
queen needle
I don't think you're supposed to say stuff like that in here (idk the rules) but you can always dm me
lyric mountain
aight @queen needle @quartz kindle I'm back
managed to reduce it to 5ms :v
but the code got big af cuz I went raw java
queen needle
Bro went raw
queen needle
so what I'm hearing is when I get home I'm going to have to add timing to mine and do a bunch of optimization s
lyric mountain
anyway imma just send it to git
queen needle
hm, how do u?
queen needle
Literally what the fuck
I still don't understand how
I have some ideas to improve part 1
But that's it
lyric mountain
it applied as much optimization as I knew on it
oh actually
the time in the prints might actually be wrong
cuz I'm reading the file inside the task
queen needle
I literally know like one thing I can optimize on mine that's it
I'm going to include file reading into the time for mine
queen needle
The only optimization I know I can do is handle the checks normally and don't use an array and loop through it
lyric mountain
things u can do to shave processing time off it
- use the lowest array type you have (faster creation)
- save changed indexes to a secondary array, iterate over it from the second pass onwards
- use bitwise whenever possible
also try to stay low on array accesses
if u need to use the access index more than once make a var for it
queen needle
I'll work on implementing those when I get home
stark kestrel
today's part 2 is what i feared
part one was fun as always
fn part_one(test: bool) -> usize {
let adj = aoc_rs::parsing::get_adjacent_8_grid_char(aoc_rs::input::read_file_grid_char(
aoc_rs::input::get_path(4, test),
));
adj.get(&'@')
.unwrap_or(&Vec::new())
.iter()
.filter(|neighbors| neighbors.iter().filter(|v| **v == Some('@')).count() < 4)
.count()
}
gonna work on part 2
queen needle
part one i got 5ms
queen needle
okay got 2ms
queen needle
2ms for part 1
lyric mountain
maybe if I used in-memory changes or partitioned the input
queen needle
part2 im stuck at 16ms
lyric mountain
oh
queen needle
can't break 16ms 🙁
lyric mountain
link still the same right?
stark kestrel
both parts
queen needle
bruh
stark kestrel
rust my beloved
lyric mountain
both parts
aint that 20ms?
stark kestrel
it is
lyric mountain
https://sourceb.in/oizOHBQkrO
use direct array accesses instead of loop
stark kestrel
i just do it for the answer honestly
but it's quite fast so yeah
and if i do it a second time it gets quite stable to 5ms likely because of some optimizations of the OS
lyric mountain
best I could get was 2ms (for part 2)
queen needle
i couldn't pass 11ms
I'm trying to find one from a previous year where i had a stupidly fast one
lyric mountain
u can try profilling it, if js has smth like that
to find out what's taking too long
neon leaf
written in rust ofc its nice
frosty gale
ive gotten sick of rust recently so im going to try go instead
neon leaf
but cold starting the bins every time doesnt sound very effective
stark kestrel
go is nice because it teaches you principles that you must use in rust
lyric mountain
wait, this doesnt make sense lul (time is big cuz it's also counting the 1000 warmup runs)
why tf is the first taking 10x as long
frosty gale
if thats js its probably because it does a lot of hidden optimizations under the hood like hidden caches and precomputations to make future access/operations faster
lyric mountain
it's java
lyric mountain
lmao
frosty gale
maybe jit is warming up idk
lyric mountain
perhaps, dont think I can profile only the last run tho
oh wait, I can just not warmup
frosty gale
or it could just be the profiler thats doing something thats making it take so long
small tangle
what jdk are u using? 25 has some improvements regarding warmup time
wheat mesa
ngl day 4 was easier than the other days lol
been using python for this, relatively convenient
I figure I should get better at doing it for interviews anyways
lyric mountain
what jdk are u using? 25 has some improvements regarding warmup time
25
I'm trying to see if it's possible to reduce to 1ms
neon leaf
ty github
neon leaf
indeed
atleast they run decently fast
unlike my windows builds that take 15min+ with cached deps
stark kestrel
normal windows behavior
lyric mountain
ok, not possible, 2ms is the rock bottom
covert gale
i still gotta start day 4
i took a peak earlier today and it seems pretty easy
one of the previous aoc's had something similar
warm surge
wsp @covert gale
covert gale
hello
wheat mesa
i took a peak earlier today and it seems pretty easy
Yeah honestly it was easier than days 1, 2, and 3
queen needle
Another stupid easy day
queen needle
const ranges = new Set(`3-5
10-14
16-20
12-18`.split("\n").map(x => x.split("-")).map(x => ({lower: Number(x[0]), upper: Number(x[1]) + 1})).map(x => Array.from({length: x.upper - x.lower}, (_, i) => i + x.lower)).flat()).size;
ranges;```doesn't work on real input due to array sizes
the solution i actually used is a lot better
I'll be interested to see the optimizations done
frosty gale
cloudflare is down AGAIN lmao
shopify, downdetector and LinkedIn bye bye
atp they're just playing an early April fools joke
deft wolf
The fact that downdetector is down is very ironic 
neon leaf
works for me, tho it has been 10min so idk
pearl trail
yeah, not down just latency spike
on some region
*at least the sites on cf not down
frosty gale
it did cause a temporary outage of many sites
so basically an outage
(source: all my work related sites went down and the countless news articles)
lyric mountain
oh, today's the first day where I cant use the same code for both parts
cuz part 2 is completely backwards to part 1
neon leaf
A change made to how Cloudflare's Web Application Firewall parses requests caused Cloudflare's network to be unavailable for several minutes this morning. This was not an attack; the change was deployed by our team to help mitigate the industry-wide vulnerability disclosed this week in React Server Components. We will share more information as we have it today.
typical react moment
prime trellis
u saw this guys?
pearl trail
oh, so must be some only on specific datacenter(?) im in asia
pearl trail
u saw this guys?
yeah #development message
frosty gale
> A change made to how Cloudflare's Web Application Firewall parses requests cau...
they keep mentioning this was not an attack on every incident but ironically that's worse 😭
frosty gale
u saw this guys?
atp for my nextjs project I haven't touched for a year I get at least one CVE alert from github every month so doesn't surprise me
prime trellis
atp for my nextjs project I haven't touched for a year I get at least one CVE al...
haha yea
queen needle
This one was a pretty well known leetcode medium
Oh I didn't know that, that's cool
queen needle
cuz part 2 is completely backwards to part 1
yeah but atleast part 2 isn't too much of a change
i could reuse like 70% of the code
lyric mountain
Ngl I'm finding this one harder than the past 4
queen needle
how??!
lyric mountain
idk, never had to face smth similar
queen needle
part 1 or part 2?
wheat mesa
I think it’s harder than the past 4 for sure
queen needle
```ts
const ranges = new Set(`3-5
10-14
16-20
12-18`.split("\n").map(x => x.spli...
I mean you could always do this horrible good way
generate an entire array for all the ranges, remove duplicates, get the length
wheat mesa
It’s too slow for how large the ranges are
There is an O(n) solution for merging the ranges instead of needing to iterate over every single number within the ranges
lyric mountain
O(n) would mean for n ranges you're iterating n times
if you're expanding the ranges then it's much larger than that
queen needle
would it be O(n^3) if i go over the number-number once to turn it into {upper:..., lower:...} and then go through it again to merge, and then the last time to add to the fresh ingredients?
lyric mountain
no idea how to do big-O when the expanded inputs vary in magnitude
wheat mesa
Yeah this wouldn’t be an easy comparison
queen needle
I assume my way can be optimized I just don't know how
wheat mesa
Expanding the ranges is incredibly slow in comparison to merging them
wheat mesa
That’s the “correct” O(n) solution for merging the ranges
If you were doing the naive way of throwing every number in a range into a set then it would be infinitely slower
queen needle
oh no no
i did that as a joke because I got bored
that solution still takes about a second though
stark kestrel
the advantage of having read the input first for part 1 made me solve part 2 within seconds
made the merging directly
queen needle
If you make the merging directly could you still do part 1 with the merges?
stark kestrel
yeah
items
.into_iter()
.filter(|&i| {
let x = i as isize;
merged_ranges.iter().any(|&(l, r)| x >= l && x <= r)
})
.count()
part 2
merged_ranges
.iter()
.map(|&(l, r)| (r - l + 1) as usize)
.sum()
dang
stark kestrel
my merging is likely not the cleanest
fn read_and_merge_ranges(test: bool) -> (Vec<(isize, isize)>, Vec<usize>) {
let sections = aoc_rs::input::read_file_in_sections(aoc_rs::input::get_path(5, test), "\n\n");
let mut ranges = aoc_rs::regex::ranges(§ions[0]);
ranges.sort();
let items = sections[1]
.lines()
.map(|line| line.parse().unwrap())
.collect();
let mut merged = Vec::with_capacity(ranges.len());
for (l, r) in ranges {
if let Some((_, last_r)) = merged.last_mut() {
if l <= *last_r {
*last_r = (*last_r).max(r);
continue;
}
}
merged.push((l, r));
}
(merged, items)
}
made the capacity of the merged vec the len of the ranges, as it can't go higher
i took my first range, saw it was like 5 trillion of diff
so i was like, yeah ain't making a vec of those items 🤣
would've done the lazy technique otherwise
add all items and slap a .contains
queen needle
lmaoo
stark kestrel
always reminds me of the rocks/fish problem of previous years
queen needle
which problem?
stark kestrel
let me find them again
stark kestrel
which problem?
https://adventofcode.com/2021/day/6 - the fish one
https://adventofcode.com/2024/day/11 - the rock one
queen needle
let { readFileSync } = require("node:fs"), { EOL } = require("node:os"), ageCounts = Array(9).fill(0);
const fishies = readFileSync("../input.txt", "utf-8").split(",").map(Number);
for (let fish of fishies) {
ageCounts[fish]++;
}
for (let i = 0; i < 80; i++) {
ageCounts[(i + 7) % 9] += ageCounts[i % 9]
}
console.log(ageCounts.reduce((a, b) => a + b));```AoC in 2021 😭 unc status achieved
queen needle
my beautiful solution for 2021 day 6
stark kestrel
fishies
i did 2021 & 2022 in go
2023 skipped because military
2024 & 2025 rust because learning it
and pre-2021 in python
wheat mesa
I did some of 2024 in rust (or maybe it was 23, can’t remember)
queen needle
https://adventofcode.com/2021/day/6 - the fish one
https://adventofcode.com/202...
DiscordMX
stark kestrel
iirc 19&20
stark kestrel
queen needle
last year i tried to do as many languages as i could
for python i tried to also do a oneline version everyday
wheat mesa
I’ve been putting off python practice for years
It’s lowkey been biting me in the ass for interviews though, figured I might as well practice LC-style questions with it
I use Java in my interviews but I’m realizing how much time I’m wasting
queen needle
lmao
queen needle
<:poggythumbsup:819012393031696395>
literally no clue how to do day11 of 2024 wtf
the general problem just looks like ```
stones = []
loop(25){
stones = blink(stones)
}
stones.length
yes
stark kestrel
the general problem just looks like ```
stones = []
loop(25){
stones = blink(sto...
yeah
pub fn execute(test: bool) {
println!("{}", blink(25, test));
println!("{}", blink(75, test));
}
quartz kindle
because i decided to do some stupid linked list shenanigans instead of doing something remotely normal
but oh well, it takes 0.4ms to run /shrug
queen needle
yeah
```rs
pub fn execute(test: bool) {
println!("{}", blink(25, test));
...
could you help me understand the rules of "blinking"?
If the stone is engraved with a number that has an even number of digits, it is replaced by two stones. The left half of the digits are engraved on the new left stone, and the right half of the digits are engraved on the new right stone. (The new numbers don't keep extra leading zeroes: 1000 would become stones 10 and 0.)```
I don't get what it means by 1000 becoming 10 and 0, and then also in the example it shows 99 becoming two 9's? but that isn't in any of the rulesIf it's even, split - so
1000 becomes 10 00
99 becomes 9 9
33 becomes 3 3
queen needle
yeah but, 99 isn't even?
stark kestrel
The "is even" is for the number of digits in the number
queen needle
oh
quartz kindle
part 1:
function UGANDA(input) {
const breakpoint = input.indexOf("\n\n");
const ranges = input.slice(0, breakpoint).split("\n").map(x => x.split("-"));
const values = input.slice(breakpoint + 2).split("\n");
let n = 0;
for(const val of values) {
const num = Number(val);
if(ranges.find(x => num > Number(x[0]) && num < Number(x[1]))) {
n++;
}
}
return n;
}
part 2:
function UGANDA2(input) {
const breakpoint = input.indexOf("\n\n");
const ranges = input.slice(0, breakpoint).split("\n").map(x => x.split("-").map(Number));
const list = { start: ranges[0][0], end: ranges[0][1], prev: null, next: null };
let first = list;
for(let i = 1; i < ranges.length; i++) {
const [start, end] = ranges[i];
let current = list;
if(start < current.start) {
while(current.prev && current.prev.start > start) {
current = current.prev;
}
const node = { start, end, prev: current.prev, next: current };
if(current.prev) {
current.prev.next = node;
}
current.prev = node;
if(!node.prev) {
first = node;
}
} else {
while(current.next && current.next.start < start) {
current = current.next;
}
const node = { start, end, prev: current, next: current.next };
if(current.next) {
current.next.prev = node;
}
current.next = node;
}
}
let n = 0;
let start = first.start;
let end = first.end;
while(first.next) {
first = first.next;
if(first.start > end) {
n += end - start + 1;
start = first.start;
end = first.end;
} else if(first.end > end) {
end = first.end;
}
}
return n + (end-start+1);
}
If the stone is engraved with a number that has an even number of digits, it is replaced by two stones. [...]
queen needle
okay i might be bad at reading lmao
stark kestrel
iirc i got super confused for that day while reading it at the beginning
sometimes i just read half the words lmao
queen needle
part 2:
```js
function UGANDA2(input) {
const breakpoint = input.indexOf("\n...
i understand so little of this
quartz kindle
the general idea is sort using a linked list and then traverse the sorted list
stark kestrel
bro got matrixed by linked lists
quartz kindle
initially i wanted to run the mergining while sorting, but it got too complex and i gave up
this is basically just a glorified version of sorting an array lol
i have no idea why the fuck i decided to do it this way
im a fucking idiot
queen needle
i mean hats off to you, i've seen no one else do it that way
quartz kindle
this works just as well, if not better:
function UGANDA2(input) {
const breakpoint = input.indexOf("\n\n");
const ranges = input.slice(0, breakpoint).split("\n").map(x => x.split("-").map(Number));
const sorted = ranges.toSorted((a,b) => a[0] - b[0]);
let n = 0;
let start = sorted[0][0];
let end = sorted[0][1];
for(let i = 1; i < sorted.length; i++) {
const [s,e] = sorted[i];
if(s > end) {
n += end - start + 1;
start = s;
end = e;
} else if(e > end) {
end = e;
}
}
return n + (end-start+1);
}
fml
spent 3 hours for nothing
queen needle
is the time the same?
queen needle
how tf
import * as fs from "fs";
const inputFile = "input.txt";
const fileContent = fs.readFileSync(inputFile, "utf-8");
let stones = fileContent.split(" ").map(Number);
function blink(arr: number[]): number[] {
const out: number[] = [];
for (const stone of arr) {
if (stone === 0) {
out.push(1);
continue;
}
const digits = Math.floor(Math.log10(stone)) + 1;
if (digits % 2 === 0) {
const half = digits / 2;
const div = 10 ** half;
const left = Math.floor(stone / div);
const right = stone % div;
out.push(left, right);
} else {
out.push(stone * 2024);
}
}
return out;
}
for (let i = 0; i < 25; i++) {
stones = blink(stones);
}
console.log(stones.length);
2024 day 11 part 1 solution
i mean also part2
considering you just change one number
oh nevermind, it doesn't work for part 2
queen needle
2024 day 11 part 1 solution
import * as fs from "fs";
const inputFile = "input.txt";
const fileContent = fs.readFileSync(inputFile, "utf-8");
let counts = new Map<number, number>();
for (const n of fileContent.split(" ").map(Number)) {
counts.set(n, (counts.get(n) || 0) + 1);
}
function step(map: Map<number, number>): Map<number, number> {
const next = new Map<number, number>();
for (const [stone, count] of map) {
if (stone === 0) {
next.set(1, (next.get(1) || 0) + count);
continue;
}
const digits = Math.floor(Math.log10(stone)) + 1;
if (digits % 2 === 0) {
const half = digits / 2;
const div = 10 ** half;
const left = Math.floor(stone / div);
const right = stone % div;
next.set(left, (next.get(left) || 0) + count);
next.set(right, (next.get(right) || 0) + count);
} else {
const v = stone * 2024;
next.set(v, (next.get(v) || 0) + count);
}
}
return next;
}
for (let i = 0; i < 75; i++) counts = step(counts);
let total = 0;
for (const v of counts.values()) total += v;
console.log(total);
part 2
just changed it to a map
quartz kindle
your's shouldnt be much slower, just add the timers after reading the file from disk
stark kestrel
just changed it to a map
yeah that's the thing
similar to today's problem imo
as well as the fishes
queen needle
for the fishes i didn't have to change the data structure
array still worked
day 5
Part 1: 14ms
Part 2: 9ms
I swear it ran slower last night
maybe the code got scared it was being timed and ran faster
quartz kindle
lmao
queen needle
Well also i think it was running fast, it just takes a little bit to show in the terminal
lyric mountain
aight, done it
lyric mountain
no, still groovy
def input = getClass().getResource('/day5.txt').readLines()
def map = new TreeMap<Tuple2<Long, Long>, Integer>((a, b) -> a.v1 <=> b.v1 ?: a.v2 <=> b.v2)
def ranges = input.inject(map) {
tot, v ->
if (!v.empty) {
def (a, b) = v.split('-').collect { it as long }
if (b != null) {
tot[Tuple.tuple(a, b)] = 0
} else {
for (p in tot.keySet()) {
if (a >= p.v1 && a <= p.v2) {
tot[p] += 1
break
}
}
}
}
return tot
}
def ids = ranges.keySet().toList()
def merged = new ArrayList<Tuple2<Long, Long>>()
for (id in ids) {
def (aMin, aMax) = id
if (!merged || merged.last.v2 < aMin) {
merged << id
continue
}
def (bMin, bMax) = merged.last
merged[-1] = Tuple.tuple(bMin, Math.max(aMax, bMax))
}
println "Fresh: ${ranges.values().sum()}"
println "IDs: ${merged.sum { it.v2 - it.v1 + 1 }}"
lmao
stark kestrel
spaceship operator in groovy
turbofish operator in rust
sounds like based languages
queen needle
turbofish?
stark kestrel
::<>
lyric mountain
we also have elvis ?:
quartz kindle
lmao
lyric mountain
and churros =~
stark kestrel
and churros `=~`
lyric mountain
queen needle
gotta speed that up
lyric mountain
eh, aint spending more time on that one :v
queen needle
that's fair lmao
stark kestrel
I didn't hyperfine mine, gonna do it later, probably not super fast though
quartz kindle
its including file loading probably
lyric mountain
yep
queen needle
ahh
lyric mountain
but file access doesnt change much in java/groovy, they get cached
what's slowing it down is groovy itself, it aint exactly used for speed since all that sugar adds extra processing
quartz kindle
the algorithm itself shouldnt take more than 10ms
quartz kindle
adding a while(true){} counts as implementing it wrong lmao
stark kestrel
realized that for searching the range that matches for part 1 you can likely do a binary search
can be faster by some ms 🤷♂️
queen needle
adding a `while(true){}` counts as implementing it wrong lmao
you saying this made me scared thinking I did it
quartz kindle
lmao
lyric mountain
ok, made the optimized version
gonna benchmark now
@quartz kindle @queen needle can I get any faster?
lyric mountain
ns / 1e6
queen needle
good news: i got 0.2ms
bad news: now the answer is wrong
keen stag
halo
queen needle
Alright i got 0.36ms using c++
i can't do faster
and then a friend of mine immediately got 0.24ms using zig
need someone who actually knows c++
wheres brain
lyric mountain
I think the only way I can get faster with java is by building a native image
or using native arrays
queen needle
try
quartz kindle
bro using zig lmao
queen needle
bro using zig lmao
If it's good enough for bun it's good enough for AoC
quartz kindle
If it's good enough for bun it's good enough for AoC
not saying its bad, just unusual, rarely ever see someone using it
queen needle
He's using AOC to learn it
queen needle
That was my thought
They haven't released any specific year merch unfortunately
I usually buy the hoodie with the design
deft wolf
Discord is removing active dev badge 
deft wolf
From the ddevs faq channel
deft wolf
Not for too long probably 
swift barn
From the ddevs faq channel
warped glacier
Any idea why this could be causing a discord ratelimit on boot
@tasks.loop(minutes=30)
async def update_topgg_stats(self):
await self.wait_until_ready()
if not self.topgg_token:
return
url = f"https://top.gg/api/bots/{self.user.id}/stats"
headers = {
"Authorization": self.topgg_token,
"Content-Type": "application/json",
}
try:
data = {"server_count": len(self.guilds)}
async with aiohttp.ClientSession() as session:
async with session.post(url, headers=headers, json=data) as resp:
if resp.status == 200:
print(f"[TOP.GG] Posted server count: {data['server_count']}")
else:
error = await resp.text()
print(f"[TOP.GG ERROR] Status {resp.status}: {error}")
except Exception as e:
print(f"[TOP.GG EXCEPTION] {e}")```This is the only place its called
async def setup_hook(self):
self.update_topgg_stats.start()
await self.tree.sync()
print("[DEBUG] setup_hook running")```And if i remove it i have no issues
harsh nova
I still dont understand how that code would be doing that. self.guild should(?) not make and calls to d API. I dont know what wait_unti\l_ready() does under the hood but my guess it it just waits for the discord ready event which should also not make calls to d API.
My very very loosely formed idea is that some part of that code crashes and it tries to restart, crashed again, and enters a loop? That might cause the repeated auth api calls to issue ratelimits but it seems like you are doing proper error handling
stark kestrel
how many times did you start the bot
warm surge
People are arguing about the badges
lmao
warped glacier
how many times did you start the bot
1
stark kestrel
you sync on every start/setup, only thing that can rate limit
warped glacier
you sync on every start/setup, only thing that can rate limit
I boot it 1 every maybe 2hrs or so. So it shouldent be a boot issue
warm surge
im watching it rn @deft wolf
stark kestrel
I boot it 1 every maybe 2hrs or so. So it shouldent be a boot issue
Why not
You have a limited number of syncs per day
stark kestrel
Sync only when needed
warped glacier
You have a limited number of syncs per day
Im talking when im working on it
stark kestrel
Make it a command or something
warped glacier
Like today i hadent touched it in 3 weeks till now and immediantly got that ratelimit error
stark kestrel
Running on your pc i suppose
warped glacier
Where else would it be
warped glacier
You think i can afford a VPS for a bot that is free all around?
stark kestrel
Lots of people can
Since you didn't mention anything, I'm trying to isolate the issue
Seeing your responses I'll pass
warped glacier
I want to know what there jobs are then Bc i barely make enough to live.
Sidetracked
warped glacier
I still dont understand how that code would be doing that. self.guild should(?) ...
Your right there. Wait is ssry a sleep function till discord returns "Startup sucess" or smth like that. And the lens does technically call on discord as its pulling the server count live but that is 1 time.
harsh nova
Your right there. Wait is ssry a sleep function till discord returns "Startup su...
I'm just guessing. I dont write python code and have never used whatever library you are using. self.guilds feels to me like something that should not call an API as it's a property and not a function
warped glacier
I'm just guessing. I dont write python code and have never used whatever library...
Your fine and your honestly right its just odd what it is doing. I would ask dpy but there kinda a bunch of ***** over ther (In nice words. Banned me for asking for help)