development

[frosty gale]

tbf i get sensitive about badges too

i know they have absolutely zero value (except if you sell your account with a rare badge maybe) but its just an irrational sort of attachment

[deft wolf]

I don't have problem with badges personally, I have a problem with people chasing them like crazy

Badges are definitely an addition to your profile but not so important that you have to do everything to get as many of them as possible

[swift barn]

@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",
    }
    data = {"server_count": len(self.guilds)}

    try:
        # use only a single session urs was creating sessions everytime 
        if not hasattr(self, "topgg_session"):
            self.topgg_session = aiohttp.ClientSession()

        async with self.topgg_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}")

async def setup_hook(self):
    # start loop once only.
    if not self.update_topgg_stats.is_running():
        self.update_topgg_stats.start()

    await self.tree.sync()
    print("[DEBUG] setup_hook running")

then u can add like a function to close the session pool or something
try , if it works, it was a loop issue

[warped glacier]

Thanks ill try a bit later. Kinda have rebooted the bot 500 times now so ima stop

[swift barn]

piled up many requests

[frosty gale]

67th amendment of the united states says that an active government shall not infringe on the people's right to a badge (its true)

[neon leaf]

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

[swift barn]

https://tenor.com/view/67-gif-8575841764206736991

[frosty gale]

i do genuinely wonder if some of the mods in the discord dev server get paid or are they just doing it for free

i see some borderline glaze from some of them

[warped glacier]

I think there volinteer

[deft wolf]

Someone said that they are paid iirc

Also RIP faceboy, he was banned

(not for that question)

[warped glacier]

Did not fix it today but could also be on some daily limit. Ill dm you tmr if it still is

[swift barn]

yeah or worse case you actualy try on a different IP , shared IPs have a tendency to actually get rate limited by discord

rip the badge finally left

[frosty gale]

makes sense

skull emoji if they did it for free

[warped glacier]

Ma7ybe i have no way to change ip tho without risking performance so ye

[neon leaf]

@quartz kindle bad news for the ampere

the ram in my epyc is 4R, which is not compatible with ampere altra 1/2R

so will be stuck on 16gb for now :c

[quartz kindle]

wtf

[swift barn]

or in theory you can route ur requests through a proxy ?

[warped glacier]

I have too much stuff to try and proxy rn

[harsh nova]

trying to debug aoc day 4 and it just looks cool cant even be mad

figured out solving it

[lyric mountain]

bro's literally generating a minecraft world

[harsh nova]

LMAO yeah basically

[swift barn]

[queen needle]

how'd you do a visualization?

[harsh nova]

I did this

fn get_char_rep(c: char) -> char {
    if c == '@' {
        return '█';
    } else if c == 'x' {
        return '▒'
    } else {
        return '░';
    }
}

for i in &iter {
    let s: String = i.iter().map(|&b| get_char_rep(b as char)).collect();
    println!("{s}");
}

before anyone says I can just do the cast in get_char_rep. Yes but i kept it like this to spite yall

[queen needle]

Ohhh okay

[neon leaf]

gotta love stack overflows https://github.com/calagopus-rs/panel/commit/945aa6fb8fe117720625974bb4fd753fc7643849

[scenic kelp]

i see the problem

it's that you're using rust!!!!!!

[patent bramble]

whatthefuckeryisthis?!
https://support-dev.discord.com/hc/en-us/articles/10113997751447-Active-Developer-Badge

[queen needle]

day6 is not for the weak(me)

[queen needle]

https://discord.mx/i/SwNUBwreyv.png

31.34 KB (@pancake)

DiscordMX · Dec 6, 2025, 6:59 AM

happiest moment of my life

Both parts run in 6ms

And I'm not even attempting optimizations

[harsh nova]

Oh!!!!!!! I should run it using iron

[frosty gale]

is it just me or my browser or did chromium add a border around every page??

it looks terrible

ok disabling this fixed it

thank goodness

[pearl trail]

chrome now have split tab view

[deft wolf]

Indeed

[harsh nova]

Zen has rounded corners too but I kinda vibe w it

[stark kestrel]

took time

zen still superior imo

[neon leaf]

I used so many tokens that it cant even display the cost 😭

[neon leaf]

type TranslationRecord = {
    [k: string]: string | TranslationRecord;
};

type DefinedTranslations<
    O extends TranslationRecord,
    P = UnionToIntersection<PathImplObj<O>>,
> = {
    raw: O;
    paths: keyof P;
    mapping: P;

    use<K extends keyof P>(
        key: K,
        // @ts-ignore this is perfectly fine
        values: Record<GetPlaceholders<P[K]>[number], string>,
    ): string;
};

export type UnionToIntersection<U> = (
    U extends unknown
        ? (k: U) => void
        : never
) extends (k: infer I) => void
    ? I
    : never;

type PathImplObj<T> = T extends TranslationRecord
    ? {
            [K in keyof T]: T[K] extends TranslationRecord
                ? PathImplObj<T[K]> extends infer Nested
                    ? {
                            [P in keyof Nested as `${K & string}.${P & string}`]: Nested[P];
                        }
                    : never
                : {
                        [P in K]: T[K];
                    };
        }[keyof T]
    : never;

type GetPlaceholders<S extends string> =
    S extends `${string}{${infer W}}${infer RE}`
        ? [W, ...GetPlaceholders<RE>]
        : [];

how I love typescript

[ionic ibex]

https://discord.com/quests/1442944099404742667

https://discord.com/quests/1435324548827451605

https://discord.com/quests/1410358070831480904

[pearl trail]

https://tenor.com/view/thank-you-thank-you-thank-you-thank-you-thank-you-so-much-thank-you-gif-thank-you-very-much-gif-12304641100528240264

[ionic ibex]

https://discord.com/quests/1441487359450284112

[quartz kindle]

?

[pearl trail]

!

[burnt hornet]

a list of people to not trust to on Discord

• default Discord pfp's
• no pfp's

https://cdn.discordapp.com/emojis/1260962400459558954.webp?size=48&name=inse~1

[quartz kindle]

nice temps

[neon leaf]

ty

[low marten]

Turning out pretty nice 😄
Spent last night creating a bunch of branding material for around 2 hours

Here's one of the new logo

My implementation of server selection is pretty unique 😎

[acoustic bough]

looks very neat

though the sidebar/navbar kinda looks like this were the docs (just from the looks) lol

[digital swan]

i like it love a nice set of graphs

[quartz kindle]

amazing, i really like the design and the color scheme

[clever tundra]

never wouldve known this existed unless you did that, but I now know it exists, so now I am using it
only feedback I have is that I don't like the join servers for you when you login to the dashboard

[low marten]

It wont exist for much longer, i plan to remove it at around 50 members. I just need a small audience so that i can build up a strong support forum

people join > report bugs > bugs get fixed > documentation

but thanks 😎

[clever tundra]

most people see join servers for you and think scam, so just bear that in mind

and the top.gg invite link makes it so that it doesnt make a role btw

[low marten]

The bot has no major commands, it doesn't need any permissions, hence the lack of a role

[clever tundra]

other than message history?

[low marten]

no need

[clever tundra]

oh

-# website invite link makes it have a role

[low marten]

thats odd, i must have a mismatch on my dev portal haha

ill check it out now

but this is what i mean!! users discover things

my mistake, on top.gg it SHOULD have a role

for read message history

its been a long day

fixed.

[clever tundra]

[low marten]

I'll also remove autojoin for the time being

[clever tundra]

alright

[queen needle]

Did you do AoC today?

[quartz kindle]

not yet

[queen needle]

Part 2 is straight from the devil

[quartz kindle]

lmao

[quartz kindle]

this was a funny mindfuck lmao

[queen needle]

It took me a couple hours

I had to rewrite my logic from part 1, and then I used my new part 2 to make part 1 better, and then that new part 1 to make part 2 befter

[quartz kindle]

xD

took me about 40min for both parts, but second did give a little trouble

[queen needle]

How tf

I struggled because I didn't understand what had to be done

And still kind of don't

For part 2

[quartz kindle]

took me a little while to understand too

gotta learn how to read Cephalopod math

:^)

[queen needle]

function transformProblem(problem: string[]): (string | number)[] {
  const numbers = problem.slice(0, -1);
  const operator = problem.at(-1)!;

  const output: (string | number)[] = []

  for (let i = operator.length - 1; i >= 0; i--) {
    let column = "";
    for (const num of numbers) {
      column += num[i] ?? " ";
    }
    output.push(Number(column));
  }

  output.push(operator.trim());
  return output;
}```

This is the code of my part 2

core*

originally my parsing for the input into [
[...column]
[...column]
]
Etc was just for(length of line){ part = split line by " " for (number of part){ if(problems[index(number)]) push else create array for column } }

"pseudo" code lmao, I'm on mobile

[quartz kindle]

xD

[queen needle]

But for some reason the right logic for transforming wouldn't work with that input parsing system

Because I wasn't dealing with the white spaces needed

what was your solution?

[quartz kindle]

also keep in mind the last line is truncated, not of equal length as the other lines

so if you base everything off of operator.lenght, you might get wrong results

in my case the operator line was 2 characters smaller than the number lines

[queen needle]

due to whitespace?

[quartz kindle]

it has no whitespace at the end

[queen needle]

when I finally got my code working I forgot one spaxe in the example input

[quartz kindle]

it ends there

[queen needle]

Which caused me to be off by 4, that took like 30 minutes to fix

[quartz kindle]

that was one of the things that caused me to get wrong results

[queen needle]

ahh

Waiting for the amazing tim code

[quartz kindle]

i did console.log the lengths of the lines and got this: [3763, 3763, 3763, 3763, 3760]

3 characters less in the last line

[quartz kindle]

my amazing code:

function LOL6v2(input) {
    let total = 0;
    const lines = input.trim().split("\n");
    const width = Math.max(...lines.map(x => x.length));
    let n = [];
    for(let i = width-1; i > -1; i--) {
        const col = lines.map(x => x[i]);
        const op = col.pop()?.trim();
        const num = col.filter(x => x.trim()).join("");
        n.push(Number(num));
        if(op) {
            let toadd = n.shift();
            for(const xyz of n) {
                switch(op) {
                    case "+": toadd += xyz; break;
                    case "*": toadd *= xyz; break;
                }
            }
            total += toadd;
            n.length = 0;
            i--;
        }
    }
    return total;
}

[queen needle]

I saw a lot of solutions that just used eval

which I thought was pretty funny

[quartz kindle]

thats for lazy people lmao

[queen needle]

I used a reduce

[quartz kindle]

this was part 1:

function LOL6(input) {
    let total = 0;
    const lines = input.trim().split("\n");
    const ops = lines.pop().split(/\s+/);
    const numbers = lines.map(x => x.split(/\s+/));
    for(let i = 0; i < ops.length; i++) {
        const op = ops[i];
        const nums = numbers.map(x => Number(x[i]));
        let toadd = nums.shift();
        for(const num of nums) {
            switch(op) {
                case "+": toadd += num; break;
                case "*": toadd *= num; break;
            }
        }
        total += toadd;
    }
    return total;
}

pretty similar

[queen needle]

So I might be dumb

But you get the longest line and use that as your loop, but how does that get the column?

[quartz kindle]

lines.map(x => x[i])

[queen needle]

yeah but wouldn't that just be a number m

[quartz kindle]

wdym? its all numbers in that column

[queen needle]

lines I'm that case is the array of all of the numbers right?

in*

[quartz kindle]

lines is all lines, including the operators line

lines = [numbers, numbers, numbers, numbers, operators]

[queen needle]

Yeah but then how does that map for column work?

Is it a 2d arrayM

[quartz kindle]

lines.map(x => x[i])
is the same as [numbers[i], numbers[i], numbers[i], numbers[i], operators[i]]

[queen needle]

I sound stupider than I am I swear

Gimme a second

ohhhh

So like at 0 it gets the first of all of them, which is a column

I get it now

[quartz kindle]

[1234]
[8472]
[4582]
[4850]
[*___]

i = 2
3
7
8
5
_

[queen needle]

How does it turn it into the weird format that it wanted?

[quartz kindle]

thats the format they wanted, read the numbers from top to bottom

[queen needle]

yeah because you using width -1 as your start for the for loop right?

[quartz kindle]

start at the end of the lines, because they want it from right to left

and then from top to bottom on each index

[queen needle]

Ahh yeah

https://github.com/Cinnamonsroll/AdventOfCode2025/blob/main/day6/part2.ts

Making me feel like I overcomplicated it a little lmao

[quartz kindle]

xD

[queen needle]

"You pick an algorithm for part 1, and then one of the other algorithms is revealed as a bad choice for part 2. Do you switch or keep your original choice?"

Sounds about right

[quartz kindle]

xD

[queen needle]

I hope tomorrow isn't as bad

[quartz kindle]

the issue with your idea is that splitting by " " makes no sense for vertical numbers, they are not the correct boundary

[queen needle]

I just have trouble understanding what a problem asks for most of the time, not the actual implementation

[queen needle]

It works (somehow)

[quartz kindle]

you go the long way of doing corrections and transformations to fix it

[queen needle]

That method preserves whitespace (along with the line under it)

[quartz kindle]

btw did you measure performance? xD

[queen needle]

my old way for parsing was literally just

const problems = []

for(let h = 0; h < lines.length; h++){
    const line = lines[h].split(" ");
    for(let j = 0; j < line.length; j++){
        if(problems[j]) problems[j].push(line[j])
    else problems[j] = []
    }

}```

probably a little off but

[queen needle]

6ms for part 1 and 2 separate

[quartz kindle]

thats pretty good

[quartz kindle]

yeah that works for horizontal bounds, where you can split by space and get the correct number

[queen needle]

it was the only way I could think to parse just looking at it

and then I realized I could write a tokenizer

[quartz kindle]

but for vertical numbers it doesnt work

[queen needle]

Thankfully I didn't

[queen needle]

like part 2?

[quartz kindle]

ye

[queen needle]

yeah unfortunately

[clever tundra]

@low marten what timezone is most active hours based on?

[low marten]

GMT

[clever tundra]

Interesting

Oh great that's my timezone

That's nice

[quartz kindle]

holy shit today's AoC was hell

usually you can get away with many different methods and still complete it, but this time good luck lmao

[low marten]

hey again guys, i'm currently working on the backend for VC / User VC statistics, in the mean-while, i could use some help on where to actually display these stats, do i create new charts, or use existing ones, and append them with the VC info?

for example, Instead of message volume, it's activity volume, with 2 respective line charts, for 'messages' and 'minutes' minutes obviously accounting for VC

[low marten]

https://tenor.com/view/thanos-gif-25117042

[queen needle]

What was your solution

[stark kestrel]

made yesterday's today because had no time

will do today's tomorrow probably, hate that kind of day honestly

[quartz kindle]

what was yours :^)

[queen needle]

https://github.com/Cinnamonsroll/AdventOfCode2025/blob/main/day7/part1.ts

Part 1

https://github.com/Cinnamonsroll/AdventOfCode2025/blob/main/day7/part2.ts

Part 2

[quartz kindle]

yup thats the proper way to solve it

[queen needle]

What'd you do lmao??

[quartz kindle]

i tried the other ways first and got demolished hard

[quartz kindle]

this was my version, but the solution is pretty much the same

function WHATTHEFUCK(input) {
    const lines = input.trim().split("\n");
    const beams = Array(lines.length / 2).fill().map((_,i) => Array(lines[0].length).fill(0))
    beams[0][lines[0].indexOf("S")]++;
    for(let i = 2; i < lines.length; i += 2) {
        const line = lines[i];
        const b = (i - 2) / 2;
        for(let x = 0; x < beams[b].length; x++) {
            const beam = beams[b][x];
            if(beam > 0) {
                if(line[x] === "^") {
                    beams[b+1][x-1] += beam;
                    beams[b+1][x+1] += beam;   
                } else {
                    beams[b+1][x] += beam;
                }
            }
        }
    }
    return beams.at(-1).reduce((a,t) => a + t, 0);
}

my other solutions were a recursive function and a "always go left" maze solver thingy with backtracking
both would take hours/days if run until the end

[queen needle]

all though these days are getting to the point where I have issues and may not be able to solve them, so I might be tapping out soon

[quartz kindle]

for example ```js
function WHATTHEFUCK(input) {
let total = 0;
const lines = input.trim().split("\n").slice(0, 50);
go(2, lines[0].indexOf("S"));
function go(line, beam) {
if(line >= lines.length) { total++; return; }
if(lines[line][beam] === "^") {
go(line + 2, beam - 1);
go(line + 2, beam + 1);
} else {
go(line + 2, beam);
}

}
return total;

}

this takes 20 seconds if i truncate the input to 100 lines, and 200 seconds if i truncate it to 110, god knows how long it would take if i let it run for the full 140 lines

i spent hours trying to debug it because i thought it was getting stuck in an infinite loop

[queen needle]

When it turned 12am I didn't have my laptop I looked at the problem and my first thought was count the ^ and minus one, and then I realized that not every splitter splits twice and such

[quartz kindle]

but no, it just takes days if i use this method lmao

this was my other attempt (the first one actually)

function WHATTHEFUCK(input) {
    let total = 0;
    const lines = input.trim().split("\n").slice(0, 100);
    let currentLine = 2;
    let currentBeam = lines[0].indexOf("S");
    const path = [];
    while(true) {
        if(currentLine >= lines.length) {
            total++;
            const prev = path.pop();
            if(!prev) { break; }
            currentBeam = prev[0] + 1;
            currentLine = prev[1] + 2;
        } else if(lines[currentLine][currentBeam] === "^") {
            path.push([currentBeam, currentLine]);
            currentBeam--;
            currentLine += 2;
        } else {
            currentLine += 2;
        }
    }
    return total;
}

[queen needle]

I just have issues dissecting what the problem asks for, but if it's explained to me like very basic I usually can do it

[quartz kindle]

elf magic confusing you

[queen needle]

Wtf

yes

[queen needle]

This method with my input returns someone else's answer

[quartz kindle]

Lol

[queen needle]

I haven't been able to time my solution but my.giess is 1.5s

[frosty gale]

i hate tree based/backtracking algorithms so much in programming

i only had a legitimate use for one once when i needed to solve the subset sum problem for some function

[low marten]

my solution was: a little 'text > voice' switch on the Re-chart cards, i think its good enough 😭

[frosty gale]

@quartz kindle first time using protobuf wish me luck

going to use it to give the user a signed hmac ticket to the user that looks like a token but authorizes some backend action

json slow

[quartz kindle]

have you tried messagepack? but yeah give protobuf a try

or roll your own custom

if you're gonna use it for something specific and not general purpose messages, rolling your own encoding should be easy

[frosty gale]

was thinking about it but im using rust which im not too great at so wouldnt know how to (safely) deserialize a struct and serialize it back again

[frosty gale]

doesnt messagepack also include field names? for this protobuf is perfect because the data is creates/serializes will only ever be handled by the application itself so it already knows all the fields

[frosty gale]

also it includes potentially variable string lengths so if i wanted to implement my own id have to handle that case which could be a whole rabbit hole

[quartz kindle]

for variable string lengths just utf8 encode xd

[quartz kindle]

the spec does, but there are implementations that offer options to remember/cache fields

idk if the rust implementation does that

protobuf should be fine, the only thing i dont like about it is that its schema based

[frosty gale]

yeah i dont like that either but in this case the library im looking at supports just defining the protobuf structs as regular language structs so i dont have to use the schemas

although im not a fan of the extra dependency, maybe ill just stick to json which is already in my stack and since its a tiny object protobuf likely wont have any noticeable performance improvement anyway

but then the waste of using json, such pain, those 10 cpu cycles i could save

[quartz kindle]

xD

[neon leaf]

i can recommend bincode

if its rust-to-rust

doesnt include field names

only pure bytes, and is serde compatible

[lyric mountain]

aight time to catch up

[frosty gale]

supports serde too :o

looks a bit lighter than protobuf

looks pretty good, will probably use that, thanks

[lyric mountain]

oooooo @queen needle now I see what the drawing is

it's the map of the facility

[queen needle]

Oh yeah

[queen needle]

Today was hard

Likely won't be able to do tomorrow, I know I'm nearing my end of being able to do aoc

[lyric mountain]

funny to write a visualization of day 7

[queen needle]

Day 7 had the best visualization I think

Well day 1 someone made an awesome visualization

[queen needle]

My friend about today's part 2

[quartz kindle]

nah there is always a way

[stark kestrel]

Day 7 part 2 seems oddly easy... 🤔

Time to see if the works

[queen needle]

Yeah i almost couldn't find the way today

[lyric mountain]

day 7 part 2 is more difficult than I expected

[deft wolf]

Krypton and KuuHaKu right now

[lyric mountain]

I'm prolly just approaching it wrong

[stark kestrel]

had to read today's part 1 for a long time lmfao

before i finally understood

i think i understood at least

[queen needle]

I solved it and I still don't understand it

[stark kestrel]

[queen needle]

I just realized day 7 is just a variant of pascals triangle

Also I just heard of kruskals algorithm, and realized I naively implemented it for today's solutions

[wheat mesa]

Kruskal my goat

I haven't looked at day 7, is it graph theory?

I'm assuming something about minimal spanning trees

Literally just took an exam on that stuff 30 minutes ago

[queen needle]

Day 8 is

day 7 idk what the like CS term for it is

[lyric mountain]

yeah figured a few moments ago it was pascals

[stark kestrel]

feeling the same after solving part 1

google gave me graph shit, i implemented 1-to-1 as wikipedia writes on the article

[stark kestrel]

part 2 was kinda easier

sometimes they have mercy after a more painful part 1

[quartz kindle]

today's part one was annoying, i had to re-read it a bunch of times to understand what they actually want

part 2 seems to be even more annoying,i dont understand it at all

[queen needle]

Me neither

But the code works

[quartz kindle]

how did you code it if you didnt understand xD

[queen needle]

kinda just kept trying things

But also my part 1 and 2 are very similar

Thankfully

[quartz kindle]

holy shit finally

this legit took all day

[queen needle]

No way you just finished

@quartz kindle let's see the code

[quartz kindle]

i couldnt figure out what the hell they wanted me to do

[quartz kindle]

part 2:

function STUPIDELVES(input) {
    const lines = input.trim().split("\n").map(x => x.split(",").map(Number));
    const distances = [];
    const groups = new Map();
    for(let i = 0; i < lines.length; i++) {
        for(let i2 = i + 1; i2 < lines.length; i2++) {
            const item1 = lines[i];
            const item2 = lines[i2];
            const dx = item2[0] - item1[0];
            const dy = item2[1] - item1[1];
            const dz = item2[2] - item1[2];
            const dist = Math.sqrt(dx ** 2 + dy ** 2 + dz ** 2);
            distances.push([dist, item1, item2]);
        }
    }
    distances.sort((a,b) => a[0]-b[0]);
    let nextid = 1;
    let last;
    for(const pair of distances) {
        const item1 = pair[1];
        const item2 = pair[2];
        if(item1[3] && item2[3]) {
            if(item1[3] !== item2[3]) {
                const old = groups.get(item2[3]);
                groups.delete(item2[3]);
                groups.get(item1[3]).push(...old);
                old.forEach(x => x[3] = item1[3]);
                if(groups.size === 1) last = [item1, item2];
            }
        } else if(item1[3]) {
            item2[3] = item1[3];
            groups.get(item1[3]).push(item2);
            if(groups.size === 1) last = [item1, item2];
        } else if(item2[3]) {
            item1[3] = item2[3];
            groups.get(item2[3]).push(item1);
            if(groups.size === 1) last = [item1, item2];
        } else {
            const id = nextid++;
            groups.set(id, [item1, item2]);
            item1[3] = id;
            item2[3] = id;
        }
    };
    return last[0][0] * last[1][0];

[queen needle]

oh wow

[quartz kindle]

zero optimization went into this, i just wanted it to work

[queen needle]

https://github.com/Cinnamonsroll/AdventOfCode2025/blob/main/day8/part2.ts

Similar solutions

[quartz kindle]

ye

[queen needle]

I don't think there were many unique solutions for today tho

[queen needle]

That's the issue I have, dissecting the provlems

[quartz kindle]

i asked chatgpt to help me with understanding what they want

didnt help much

actually made it worse and sent me in wrong directions

lmao

[queen needle]

Lmaoooo

[quartz kindle]

this was part 1 lmao:

function STUPIDELVES(input) {
    const lines = input.trim().split("\n").map(x => x.split(",").map(Number));
    const distances = [];
    const groups = new Map();
    for(let i = 0; i < lines.length; i++) {
        for(let i2 = i + 1; i2 < lines.length; i2++) {
            const item1 = lines[i];
            const item2 = lines[i2];
            const dx = item2[0] - item1[0];
            const dy = item2[1] - item1[1];
            const dz = item2[2] - item1[2];
            const dist = Math.sqrt(dx ** 2 + dy ** 2 + dz ** 2);
            distances.push([dist, item1, item2]);
        }
    }
    distances.sort((a,b) => a[0]-b[0]);
    distances.length = 1000;
    for(const pair of distances) {
        const item1 = pair[1];
        const item2 = pair[2];
        if(item1[3] && item2[3]) {
            if(item1[3] !== item2[3]) {
                const old = groups.get(item2[3]);
                groups.set(item2[3], []);
                groups.get(item1[3]).push(...old);
                old.forEach(x => x[3] = item1[3]);
            }
        } else if(item1[3]) {
            item2[3] = item1[3];
            groups.get(item1[3]).push(item2);
        } else if(item2[3]) {
            item1[3] = item2[3];
            groups.get(item2[3]).push(item1);
        } else {
            const id = groups.size + 1;
            groups.set(id, [item1, item2]);
            item1[3] = id;
            item2[3] = id;
        }
    }
    return [...groups.values()].map(x => x.length).sort((a,b) => b-a).slice(0, 3).reduce((a,t) => a*t, 1)
}

also, i did day 1 yesterday which i hadnt done before

[queen needle]

for (let t = 0; t <= 1000; t++) {
  const { a, b } = edges[t] as Edge;
  unite(a, b);

  if (t === 1000) {
    const componentSize = new Map<number, number>();
    for (let v = 0; v < N; v++) {
      const root = find(v);
      componentSize.set(root, (componentSize.get(root) ?? 0) + 1);
    }

    const sizes = Array.from(componentSize.values()).sort((a, b) => a - b);
    const result = sizes.at(-1)! * sizes.at(-2)! * sizes.at(-3)!;

    console.log(result);
  }
}```

only difference In my part 1 vs part 2

[queen needle]

Why do you set the length?

Or more so why does setting the length do anything

[quartz kindle]

a = [1,2,3,4,5];
a.length = 2; // a = [1,2]

faster than a.slice(0, 2)

lmao

[queen needle]

That's so stupid

what was youe day 1 solution?

[quartz kindle]

some annoying shit, i kept getting stuck because i was accidentally counting one zero too much

let me find lol

[queen needle]

Bros gotta hit the up arrow in the js console

[quartz kindle]

part 1:

function Day1Lol(input) {
    const lines = input.trim().split("\n");
    let dial = 50;
    let total = 0;
    for(const n of lines) {
        const num = Number(n.slice(1));
        const dir = n[0];
        if(dir === "R") {
            dial = ((dial + num) % 100 + 100) % 100;
        } else {
            dial = ((dial - num) % 100 + 100) % 100;
        }
        if(dial === 0) total++;
    }
    return total;
}

part 2:

function Day1Lol(input) {
    const lines = input.trim().split("\n");
    let dial = 50;
    let total = 0;
    for(const n of lines) {
        const num = Number(n.slice(1));
        const dir = n[0];
        if(dir === "R") {
            const r = dial + num;
            total +=  Math.floor(r / 100);
            dial = r % 100;
        } else {
            const r = (dial === 0 && num > 0 ? 100 : dial) - num;
            total += -Math.floor(r / 100);
            dial = (r % 100 + 100) % 100;
            if(dial === 0) total++;
        }
    }
    return total;
}

[quartz kindle]

lmao thats exactly what i was doing

[queen needle]

Standard solution

Looks nice

[quartz kindle]

i did this while debugging why it was not working:

function Day1Lol(input) {
    const lines = input.trim().split("\n");
    let dial = 50;
    let total = 0;
    for(const n of lines) {
        const num = Number(n.slice(1));
        const dir = n[0];
        if(dir === "R") {
            dial += num;
            while(dial > 99) { total++; dial -= 100; }
        } else {
            if(dial === 0 && num > 0) { dial = 100; }
            dial -= num;
            while(dial < 0) { total++; dial += 100; }
            if(dial === 0) total++
        }
    }
    return total;
}

[queen needle]

Bro wtf

I mean if it helped you get to the solution

my debugging looks like 100 console logs

[quartz kindle]

not really but i did eventually figure it out

[quartz kindle]

my debugging = putting console logs inside loops and letting it log 273982749 times

[queen needle]

Maybe if you used an ise

Ide

[quartz kindle]

maybe

[queen needle]

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);```

Pretty basic part 1

part 2 I just added another loop

for (let i = 0; i < value; i++) {
    dial = (dial + sign + 100) % 100;
    if (dial === 0) password++;
  }```

[quartz kindle]

nicenice

[queen needle]

So are you all caught up for this year?

[quartz kindle]

did all 8 days so far yes

[queen needle]

Are you going to finish it out

@lyric mountain did you ever get today's/yesterdays?

[quartz kindle]

idk, 4 days remaining

you think its gonna be too hard?

[crimson vapor]

how bad are the days so far

[queen needle]

Yeah only 4 more days, how much more difficult can it get

[crimson vapor]

are they like standard DSA or more involved

[queen needle]

Idk, I think you'll be able to solve it as long as you can dissect it, I personally think I'll fall short a little but

[queen needle]

Today was just like a copy paste kruskals algorithm

[quartz kindle]

idk i wanted to pull my hair out a few times the past 2 days

[crimson vapor]

yeah im lowkey not tryna do MST anymore

[queen needle]

I'm just struggling because I have never taken a CS course so idk any of these fancy algorithms and things

[quartz kindle]

me neither

[queen needle]

That's fair, maybe we can help each other with solutions lmao

[crimson vapor]

might be worth grabbing and reading CLRS

you can skip a lot of the proofs but the explanations about graph and tree algorithms are pretty thorough

[queen needle]

I mean it'd be nice knowledge to have, but I just program for fun and make simple websites and such I don't do much, the only algorithm I've ever had to make is minimax lmfao

[crimson vapor]

not planning on doing SWE?

[quartz kindle]

i hate trees in general

[queen needle]

nah, I plan to be a french teacher

I just like programming for fun, I think if I pursued a job in it I would grow to hate it

[crimson vapor]

quite possibly plus a french teacher would be more rewarding

[queen needle]

only tree I like is AST

[crimson vapor]

I find it fun and interesting which is exactly why I'm gonna go into software

forever grateful that my prof skipped red-black trees and b trees

[queen needle]

what would a red black tree even be used for?

[crimson vapor]

no clue

kruskals is DSUF where you greedily add the lightest edge lengths that make a change yeah?

[queen needle]

https://discord.mx/i/RGgM2nImRq.png

46.15 KB (@pancake)

DiscordMX · Dec 9, 2025, 1:39 AM

[queen needle]

yes

[crimson vapor]

im locked in

[lyric mountain]

Day 7 I still need to figure out part 2

Day 8 didn't start yet

[queen needle]

What are you having trouble with in part 2?

[lyric mountain]

Finding the correct way to deal with it

Tried using a pascal pyramid but didn't test yet, gonna do it tomorrow

[patent bramble]

not even 2 weeks after hitting 100 guilds and I'm coming up hot on I passed 200

[queen needle]

Today's AOC isn't hard logically, but my part 2 takes forever to run that idek if it works

[queen needle]

@quartz kindle when you do it im going to need you to bust out optimizations

i run in like 2s

[lyric mountain]

ughhh ok day7 is a pita

[lyric mountain]

@queen needle I dont get wtf I'm supposed to do on day 8

[quartz kindle]

what are you stuck on? im stuck on day 9 part 2 right now xd

[lyric mountain]

I'm stuck in trying to understand what I have to do

as in, the puzzle itself

[quartz kindle]

ah yes i got stuck on it for hours too

part 1 or part 2?

[lyric mountain]

no tim, u didnt understand

I didnt get the puzzle itself

[queen needle]

I have to do a assignment for class, if you dont get it by the time im done i can try and help

[lyric mountain]

didnt even start coding, I just dont know what I have to do

ik I need to connect the closest boxes, but then it mentions separate groups but what's the threshold for separation?

do I have limited cord? the cords have a limited length?

[quartz kindle]

its basically like this:
A is closest to B therefore A-B
C is closest to B thereforce C-B, but since A-B now you have A-B + B-C = A-B-C
then X is closest to Y, so now you have A-B-C X-Y

[queen needle]

polygons

[lyric mountain]

ugh, ok I think I get it now

[queen needle]

@quartz kindle

[quartz kindle]

lmao

[queen needle]

how you doing on it

[quartz kindle]

playing league rn

i'll get back to it later

i have a general idea of what i want to do, just implementing it is annoying

[queen needle]

real

[stark kestrel]

welcome to the club

took me like half an hour to even understand what it wants from me

[lyric mountain]

I must be doing smth wrong cuz I get 2x4, 3x2 and 7x1 circuits after 10 connections

the 5-box group only appears at 11th

[stark kestrel]

i guess today's part 2 is complex because part 1 was

let positions: Vec<Point2> = aoc_rs::input::read_file(aoc_rs::input::get_path(9, test))
        .iter()
        .map(|l| aoc_rs::regex::coord(l).unwrap())
        .collect();

    positions
        .iter()
        .flat_map(|a| positions.iter().map(move |b| a.covered_area(b)))
        .max()
        .unwrap_or(0)

[queen needle]

yea

my part 1 is 31 lines total

my part 2 is 85 lines total

[stark kestrel]

okay yeah this sounds like a pain

[lyric mountain]

oh found the issue, nvm

[stark kestrel]

are the red tiles guaranteed to be in the same order?

[queen needle]

Yes

[stark kestrel]

like, 1 links to 2,2 links to 3,3 links to 4, etc.

[queen needle]

well

[stark kestrel]

makes it less painful

[queen needle]

Yes

[stark kestrel]

because it just says that the last and first are linked

[queen needle]

The red tiles are not affected from part 1 to part 2

[queen needle]

well you only need two corners

[stark kestrel]

In your list, every red tile is connected to the red tile before and after it by a straight line of green tiles.
ah nvm

[queen needle]

Oh yeah

i was confused as to what part you were referring

[stark kestrel]

less painful then

yeah part 1 is just computing areas of every possibility

unless there's some nerdy math solution

[queen needle]

The hardest part of part 2 is time complexity

[stark kestrel]

https://media.discordapp.net/attachments/986906791373856810/1233659037237907548/lala.gif

[queen needle]

I haven't seen one yet from the solutions thread

[queen needle]

my first approach ran in 25s

[stark kestrel]

cargo build --release

[queen needle]

my current one runs in about 2.5s

i don't know any further optmizations

maybe using ASM

i even saw like 10s+ c++ solutions

[stark kestrel]

sum progress

[queen needle]

awesome

then just count interiors

[tired zealot]

I want to make so people can open a thread ticket, but like if I ping someone there they get added…

[stark kestrel]

yippie

[stark kestrel]

probably over-engineering it

i now do part 1, but check if any tile is not X or #

[queen needle]

ahh

[queen needle]

is this not normal thread behavior?

[queen needle]

is your solution done then?

[stark kestrel]

not yet, sounds like for a grid of 100000-100000 it takes some time

[queen needle]

oh yeah

a long time

[stark kestrel]

i guess it's not a smart idea

[queen needle]

i thought it was really funny that the example is like: 11,7 9,10
and then the first line was just under 100 thousand

[stark kestrel]

always these grids & trees problems are a pain

[queen needle]

you can still use part 1

to help with lookup

[tired zealot]

Yes but how do servers do it then?

[stark kestrel]

yeah but making the grid so far takes over a min

then ctrl+c-ed

🧌

[queen needle]

LMAO

gotta start optimizing early

[stark kestrel]

yeah i printed the grid, not too smart

[quartz kindle]

i tried making a 2d array grid

crashed from OOM

[stark kestrel]

time to check with my input 🧌

[queen needle]

https://kishimotostudios.com/articles/aabb_collision/

Someone used this method

[stark kestrel]

got one, but didn't work

way too low

[queen needle]

Going to implement that it's so much faster

[stark kestrel]

probably because of the starting tile i choose to fill from

[queen needle]

oh wow

this solution runs so much faster

way less math and logic

https://discord.mx/v/sBFfr9f5td.mp4

[stark kestrel]

still getting some value that is too low 🧌

[wheat mesa]

Haven’t even looked at the problem, but if someone is using AABB then this sounds like a good use case for quadtrees

[stark kestrel]

https://tenor.com/view/d9luxe-ants-gif-10684512841626566565

okay time to ctrl+c

[queen needle]

LMAO

[queen needle]

ooh ill have to look those up

[prime cliff]

Woot got members, invites and bans all done

[queen needle]

Making awesome progress

[patent bramble]

Is there any special/proper way to ensure a guild bot can DM specific members when needed? I don't know if Discord ever fixed the "allow DMs from others" to exclude bots and allow them by default.

[lament rock]

Try it and catch the exception if it fails

Optionally cache if you can or cannot DM that user for a specific amount of time

[patent bramble]

eh if it's going to be hit or miss I'm not even going to bother tbh.

[lament rock]

Good plan. DMing users from a guild channel isn't great. Let the user DM your bot unless you are generating something private

[queen needle]

Had to give up on day 10 part 2

😭

[slim abyss]

anyone know how to add vote linked to a command?

[next kite]

Same question

[deft wolf]

What do you mean by "vote linked to a command"? Like someone need to vote before they can use a command?

[pearl trail]

lmao

fucked up so bad

[lyric mountain]

aight, done with day 8 at last

issue was on groovy's pow operator

[quartz kindle]

lmao

[lyric mountain]

this is some picasso shit

it's funny that day 9 was made for java

there's a literal suite of poly processing in stdlib

[vivid fulcrum]

how do you all have the energy to do aoc... I come from work and the last thing I want to do is spend even more time on the computer 🤣

[queen needle]

Almost every part 2 for today uses a library called z3

I really want to try and get a solution without an external lib

[queen needle]

https://discord.mx/i/pLUJ899XFl.png

5.95 KB (@pancake)

DiscordMX · Dec 10, 2025, 4:18 PM

it runs!!!

takes less than a second, no external libraries

[honest notch]

To add to this, ephemeral messages are really good if you wanna avoid dms altogether as theyre like private messages within a server

[quartz kindle]

i feel you, its time consuming and exhausting, i have no energy left for anything else if i spent my day on it

[queen needle]

I don't have anything else, it keeps me busy and entertained for a little bit

Or for today a lot a bit

[lyric mountain]

hell yeah, visualization

[queen needle]

Day 8?

[lyric mountain]

9

oh, my input was a circle

[queen needle]

Yeah

that's why I was confused

[queen needle]

Most inputs are a circle for day 9

[lyric mountain]

but yep, algo still worked

[queen needle]

How did you create the visualization?

[lyric mountain]

funny that they added an indent JUST to fuck up with boundary detection

[lyric mountain]

I built the boundary poly then used jframe to create a window I could draw to

[queen needle]

Ahhh okay

[lyric mountain]

java things

[queen needle]

My thought was just converting each range to a point system and scaling it, and then just drawing pixels for all ~400 lines/~800 points

[lyric mountain]

think I'll commit with the visualization

[lament rock]

Only limited to app commands unfortunately

[lyric mountain]

bro I'm starting to hate elves

[queen needle]

LMAO

me reading todays

[lyric mountain]

@solemn latch heard u wanted money

[solemn latch]

I always want money

[hollow briar]

FREE MONEY?

[lyric mountain]

just finished reading it, this one gonna be nasty

[queen needle]

yeah

my part 1 has a lot of bitwise operations

[lyric mountain]

didn't start it yet, but that puzzle concept itself is complicated even without code

[queen needle]

very

even just the parsing is a huge step

even though it's just regex

[stark kestrel]

i don't understand today's part 2

@queen needle am i understanding right that the [] is irrelevant for part 2?

like you start like before, at 0 everywhere and you need to hit the {} by pressing the least buttons where each button press add 1 to the item

[queen needle]

Yes

[stark kestrel]

so the same as part 1, but instead of hitting 1 for the # - you gotta hit the values in {}

[queen needle]

yes

It's system of equations

systems

[stark kestrel]

fair fair

thanks 🫡

[queen needle]

Good luck

[lyric mountain]

brute force through part 1 go brrrrr

[frosty gale]

kind of same except im not too big into code challenges

last time i did a few was when i did google foobar a few years ago

prefer working on my own stuff and just bashing out features

[patent bramble]

I don't know why but my Discord webhook for APPLICATION_AUTHORIZED keeps double/triple firing at times when added to guilds. Anyone else that uses this webhook have that happening? If it were a minute or two apart I'd assume that they kicked and re-added, but these are a few seconds apart when it happens, and it's happened to me before too even so idk..

One happened earlier today with a 6 second difference. I find it hard that the user added the bot twice that fast, or that it's happening that often...

[spark flint]

probably not sending a 200 back quick enough

[patent bramble]

I mean I think half a second is pretty quick lol

[queen needle]

Day 11 is scary easy

[stark kestrel]

Both parts?

Quickly skimmed through part 1 description and seems fairly easy

But just like the recent part 1

[young orbit]

Hey everyone, I’m running into an issue with Discord.py and could use some help.
I’m trying to log global avatar changes, but on_member_update() only gives me the guild avatar, not the global one. I tried using on_user_update(), but that event doesn’t include any guild context, so I can’t tie the change back to a specific server for logging.
Does anyone know the correct way to detect when a user updates their global avatar and still access the guild information? I’ve been stuck on this for a while and can’t find a reliable method.
Any help would be greatly appreciated!

[stark kestrel]

That's a limitation

Keep some cache or something to know in what guild the user is in I guess

[young orbit]

yes thx im working on it

[neon leaf]

if anyone needs more compatible valkey/postgres docker images, Ive made some for amd64, arm64, ppc64le, s390x and riscv64
https://github.com/calagopus-rs/valkey
https://github.com/calagopus-rs/pgautoupgrade (still working on this one rn)

[lyric mountain]

@queen needle bro wtf I had the answer to part 1 right in my face this whole time

still quite bruteforcing, but bro how come I never thought abt using bitwise to generate non-repeating unique combinations that progress naturally from smallest to largest combos

now onto part 2

[lyric mountain]

yep, I'm clueless on this one

[queen needle]

I saw people for java using Z3

[lyric mountain]

Yeah sadly I'll have to resort to libs

[vivid fulcrum]

have you confirmed it's discord that's sending the double requests?

it sounds like you may have somehow set up multiple listeners

[lyric mountain]

aight, done, hate that it was mostly a math puzzle instead of a logical one

[lyric mountain]

the first part was hard but doable with some logic

part 2 wasn't a programming puzzle, that shit was uni-level linear algebra

[queen needle]

part 2 took me so long because I was adamant on not using a library

[queen needle]

real

[lyric mountain]

so was I, not proud of the solution to that one

well, lets see wtf the elves are doing at 11th floor

[queen needle]

dude 11 is so easy

Makes me scared for tomorrow

but if this works like normal advent of code tomorrow only has one part

because the last day is usually part 1 and then part 2 is you have all the previous stars

[lyric mountain]

imo

[lyric mountain]

this reminds me of this

[queen needle]

Solve day 11 first

Genuinely felt like a day 3 esque problem

[lyric mountain]

doing it atm

[queen needle]

Day 10 type shit

Googled, looked at solutions, and still took like 6 hours

[queen needle]

I pretty much agree with this ranking though

[lyric mountain]

hm ok, day 11 p1 is basically node traversal

[queen needle]

yes

[lyric mountain]

ummm, are loops possible?

as in, a path that leads back to itself

[queen needle]

i don't think so, my code doesn't worry about that

Or well I memoize

Yeah I memoize already seen things so I think a loop is possible?

[lyric mountain]

cuz going svr -> out is taking a hella long time using dfs

[queen needle]

Do you cache?

[lyric mountain]

no

[queen needle]

try adding caching

[lyric mountain]

cache what, exactly

[queen needle]

const k = d + "|" + (f ? 1 : 0) + "|" + (c ? 1 : 0);
  if (k in memo) return memo[k] as number;```

That makes no sense out of context

[lyric mountain]

indeed it doesnt

[stark kestrel]

i did cache it still took ages lmao

[queen needle]

mine ran in like 5ms 💀

[queen needle]

https://github.com/Cinnamonsroll/AdventOfCode2025/blob/main/day11/part2.ts

[stark kestrel]

probs the way i cache

[lyric mountain]

like, I'm passing down a tuple on every jump, which contains the name + both conditions

[queen needle]

It won't let me copy code pn mobile

[lyric mountain]

pathing from "you" goes pretty fast, but I'm starting to suspect "svr" has a loop

[queen needle]

I know some people had issues with it from the sub reddit

[stark kestrel]

there

fixed my cache

was like last year's day 19

[queen needle]

idk I'm just scared for tomorrow

Linear algebra into basic graph theory makes me think tomorrow is going to be something devastating

[stark kestrel]

gained 0.4ms because i used a hashset rather than a vec

i always forget hashset exists

@solemn latch bro is a bot

way too fast + always just sends that

diff of all 3 messages was less than 900ms for the first and 400ms for the second

[stark kestrel]

won't get any lower

[stark kestrel]

definitely

[queen needle]

But like I did everyday up to it so I gotta do tomorrow

Tomorrow we're going to have to make a esolang

[lyric mountain]

yeah, memo worked

now to wait for the final boss

but yk, today might be helpful to get some more performance for my bot's dungeon crawler

[wheat mesa]

I haven’t even got day 6 yet 😭 been so busy with finals I haven’t had the time

[lyric mountain]

I need to path through nodes to find which are accessible to the player, gonna use memo now

[lyric mountain]

dw, day 10 will take 5x longer than the entirety of days 1..9

[patent bramble]

Multiple listeners for a web endpoint? I doubt that's the case. Although Cloudflare Workers does run in multiple datacenters all around, no reason it would be triggering twice unless they received it twice..?

[vivid fulcrum]

it could happen anywhere in the communication layers, you should verify if discord actually sent two identical requests

if not, it may be a cloudflare worker quirk

[patent bramble]

I know they'll retry if no response is received within 5 seconds, but it's very intermittent when this happens.

[frosty gale]

cloudflare workers wouldnt duplicate requests, only way is if you receive 2x requests

strange issue though do you have any logging setup in the worker to see if it is indeed 2x different requests?

[neon leaf]

https://react.dev/blog/2025/12/11/denial-of-service-and-source-code-exposure-in-react-server-components

2 more vunerabilities

[queen needle]

🥀🥀

[rustic nova]

i wanted to do holidays

wtf

meh "Source Code Exposure" not that too deep

[shell tundra]

dbrand cards (free)

[prime cliff]

🙂

[lament rock]

DoS as well as shell access in such a short timespan isnt looking good for react

Glad I never adopted

[rustic nova]

glad im sane

[prime cliff]

😄 DMs and Groups

[queen needle]

https://discord.mx/i/WbIRTgbEWb.png

108.56 KB (@pancake)

DiscordMX · Dec 12, 2025, 5:19 AM

[queen needle]

What all do you have left

[prime cliff]

Well i have barely worked on message stuff at all or per-channel permissions

[upper kite]

I build Discord bots, latest one builds out a whole server, channels, roles, sorts permissions, welcomes people anti raid etc etc usese ai and asks quesions with free input option and pretty much does it all, now i have no idea where to get people to test this, its mainly for brand new servers and hopefully helps those with not too much knowledge on how to setup servers... well... set em up haha

but yea if anyone here wants to test ? or has suggestions for me on where i can get people who want to test that would be super helpful

[rustic nova]

didnt veld or oliy build something like that too

and that got spammed?

[prime cliff]

Dont you mean they tried using Guilded?
And it got bot spammed so they had to lock it

Wanna try a demo of it Aurel?

[stark kestrel]

Nah

There was a chat app

[prime cliff]

There was hu?? 😔

[stark kestrel]

https://github.com/veld-chat

web client isn't there apparently

[pearl trail]

good ol days

[lyric mountain]

that's dangerously close to discord

[lyric mountain]

at last, aoc done

[humble gyro]

OH GOD

that's a fire logo tho

@prime cliff do u want my logo, I will pass the torch

[humble gyro]

web = desktop client btw

btw you guys can't see this

[stark kestrel]

ah didn't look the code

fair enough

[queen needle]

Woooooo

[pearl trail]

hell nah 100 hours 💀

[neon leaf]

what are you doing

[pearl trail]

"fixing" my hard drive by filling zeroes to all sectors

*5tb hdd

[neon leaf]

100 hours is fine then

[pearl trail]

oh really?

[neon leaf]

i have seen worse on ssds

23h for 1tb

(dram-less)

[pearl trail]

oh, fair enough

the remaining time now become 113 hours xD

guess i'll wait till tomorrow to see if the remaining time still adds up or not

[vivid fulcrum]

dotnet supremacy

[prime cliff]

Funny you should mention that my chat app backend is C# too and the client is C# Avalonia :p

[vivid fulcrum]

how is avalonia? I've been meaning to try building UI in dotnet, been eyeing Uno

[prime cliff]

Avalonia is a bit more mature than Uno and i had issues installing Uno due to some .NET SDK stuff.
If you need a more complex app then Avalonia is much better but both work either way.

[warm surge]

dann

dam*

damn*

wtf iphone

[lyric mountain]

wont that destroy the hdd lifetime tho?

[pearl trail]

hmmmm, some sources on internet says doing that will tell the hdd’s firmware which sectors are bad and mark them.. you might be right, my hdd started “failing” when i tried to recover 2tb of lost data because windows installer formats the whole disk

[lyric mountain]

If it's indeed writing to the hdd then it'll burn the sectors very quickly

Tho hdds are more resistant than ssds in that matter

Did u try checking the drive's smart report?

It usually reports any failure detected

[quartz kindle]

in certain cases it is a good thing to zero out an hdd, there are cases where you may be able to "correct" bad sectors by repeatedly attempting to write to them

but in other cases it helps wear it down even further

the hdd's firmware will indeed only mark bad sectors when write+read fails multiple times, and doing that preemtively just offsets the problem so it doesnt get stuck retrying them during normal operation

[lyric mountain]

Just run the hdd through a strong magnet :v

Guaranteed fast zeroing

[quartz kindle]

xD

fast one-way zeroing

[pearl trail]

yeah i tried, smart report already showing 0% health, tells me to backup immediately because the hdd will fail in few hours, it said

[lyric mountain]

0% health lmao

[pearl trail]

but then, before i leave it to sleep, i tried using manufacturer's software , andd holy god it's back

[quartz kindle]

yeah 0% = trash bin

[pearl trail]

the health went from 0% to 50%

so i guess yeah, zeroing won't always fix the issue

that complete drive test took 7 hours

[quartz kindle]

looks like it doesnt have any actual realocated sectors, but it does have pending ones, those are gonna cause slowness and freezeing until the firmware decides to actually reallocate them

[pearl trail]

wdym it doesn't have any actual realocated sectors? the no. 5 smart says the value is 200,,

[quartz kindle]

depends from drive to drive, and the meaning of the values needs to be interpreted by the firmware, but from what i remember 200 is the initial value

140 is the threshold

and the value goes down as it gets worse

if it goes below 140 = bad

[pearl trail]

ohhh, what about pending sectors? it should be 200 too?

[quartz kindle]

ye

[pearl trail]

ohhh i see

thank you guys

[quartz kindle]

back in the day i used to play with HDDRegenerator

this is what they claim:

Almost 60% of all hard drives damaged with bad sectors have an incorrectly magnetized disk surface. We have developed an algorithm which is used to repair damaged disk surfaces. This technology is hardware independent, it supports many types of drives and repairs damage that even low-level disk formatting cannot repair. As a result, previously unreadable information will be restored. Because of the way the repair is made, the existing information on the disk drive will not be affected!

[pearl trail]

oh damn it's paid xD

but does it really work for you?

[quartz kindle]

last time i used it as like 15 years ago xDDD

and it as also a pirated copy

in some disks it did help a bit, in others didnt make a difference

[pearl trail]

xDD

[quartz kindle]

afaik they just do a very aggressive cycle of read+write+read multiple times on every sector

[pearl trail]

i can feel that will take 100++ hours

or even more 💀

[prime cliff]

Electricity go brrrrrr

[quartz kindle]

yup

i would leave it running for 24-36 hours on 80gb hdds

[patent bramble]

For user agent, are URL/versionNumber supposed to be something specific? Mine? Discords?
https://discord.com/developers/docs/reference#user-agent

Also wild that when I tried to markdown embed that same link it got blocked lmao

[lament rock]

https://github.com/DasWolke/SnowTransfer/blob/master/src/RequestHandler.ts#L398

[humble gyro]

holy shit wolke

[patent bramble]

So just a URL to your bot and your bot's version number? That's what I was assuming, but I never got blocked for not providing this (yet) lmao.

[lament rock]

I've been maintaining his Discord stuff for a while now

[humble gyro]

goat

how has he been

haven't spoken to him in YEARS

[lament rock]

Realistically, you could use anything, but if you use your link and your lib grows large, then they provide stats for how many use your lib and allegedly get a section in the official ddevs server

[lament rock]

I haven't talked to him in a while. He included me as an employee in a company he owned for building out apps for clients, but that server with other employees has been dead for a while. He occasionally pops around when I ask for some guidance. But he seems to be doing alright and moderately successful

If you have a connection to him, he'd probably value reaching out to him

I see he is a mutual

[patent bramble]

Oh, yeah that's irrelevant. This is all mostly custom coded lmao

O well let them have some stats ig lmao

[eternal osprey]

hey guys

[deft wolf]

Hey

[neon leaf]

https://crates.io/crates/compact_str

wtf

this may be magic

[frosty gale]

doesnt rust already perform small string optimization on strings?

that sounds basically like it

[neon leaf]

i dont think so for dynamic data

[frosty gale]

from 2018

We guarantee that we will never perform this optimization for Vec: https://doc.rust-lang.org/std/vec/struct.Vec.html#guarantees. I don’t think the same is explicitly documented for String, but String is simply a wrapper around Vec so it still holds there at least in practice.
thats odd, i was under the impression it did

its basically expected to happen for strings in many languages because it offers massive speedups when dealing with strings

my day is officially ruined because of that

[neon leaf]

afaik its to simplify the std so unsafe code is safer

imagine trying to do raw pointer shit if every vec was optimized to nirvana

well time for me to switch most of my strings to this

[frosty gale]

yeah i guess that does make sense in theory, but is it really worth having every string equal an allocation?

[frosty gale]

same, in fact ill do it now

[neon leaf]

keyword most

if i know a string has a 95% chance of being longer than 24 bytes, no point in changing

man i was just trying to find the vec based kvmap crate i forgor the name of

[frosty gale]

although, this crate says it can store 24 bytes max on the stack, but thats the size of the string struct itself? where is it storing a size/boolean that determines whether this is heap allocated or stack allocated? since the struct itself will act as a union to achieve the stack storage, unless it does something else

[neon leaf]

very clever

but yeah only 23 chars ig

[frosty gale]

scammed!!

knew there was a trick to this magic

i want my 1 byte refund

[neon leaf]

oh my days

this is fucking genius

@quartz kindle

[frosty gale]

that is very clever

[neon leaf]

beautiful

actually ill just do deref itll implement most stuff already

[frosty gale]

this crate is basically a free rust performance boost considering it doesnt already do this for strings

[neon leaf]

yes

and finally fixes my 1.5gb unused heap problem

it also lowers string allocation rate

from 2x to 1.5x

nice addition imo

[frosty gale]

although this only works really well in rust because it validates that any bytes are utf 8 before they are stored in a string, in a lang like c++ a string is literally just a bunch of bytes which could be any value from 0 to 255, so unless you enforce that same runtime safety check for bytes that rust does for utf8, you cant achieve that otherwise

youd have to sacrifice one or more bytes on the stack for tracking the state/length otherwise

so youre sort of trading performance for compactness in this case, but rust already verifies all strings are utf8 so in the rust world youre technically not paying anything extra

but still incredibly clever way to leverage the fact that strings are guaranteed to be utf8 in rust

[neon leaf]

it even provides a format function

oh my

such a good crate

[frosty gale]

bro is orgasming

ez port

[neon leaf]

i sadly have to manually rewrite a ton

worth it tho

[frosty gale]

why though

[neon leaf]

utoipa

nvm i have genius idea

ill make a submodule

and implement utoipa for it myself

the beauty of open source

[frosty gale]

ah

[neon leaf]

sad that there is still no real alternative to utoipa

it feels outdated and broken

[frosty gale]

uh oh

i hope theres a feature that can be enabled

[neon leaf]

--features serde

[frosty gale]

nice

[neon leaf]

also has sqlx stuff

[frosty gale]

although doesnt seem to support rusqlite sadly

[neon leaf]

perfect

[neon leaf]

.as_str()?

ah wait fromsql

rip

[frosty gale]

although may be worth porting to another sqlite3 lib? been thinking about it for a while since i eventually want to support other dbs as well not just sqlite

[neon leaf]

sqlx, trust

[frosty gale]

rusqlite is fast but a bit annoying to work with

[frosty gale]

lol

i have one of my ci/cd checks run on a very low powered x86 oracle server simply because i pity it, this is the only check it can run without shitting itself and still it takes 5-10 minutes to finish

might just kill that server alltogether and move that final check to run on my arm server but might come in handy some day for an x86 only project

[prime cliff]

Pog audit logs and webhooks

[crimson vapor]

we have discord at home

[pearl trail]

peak electrical

sandal is the only safety equipment

[misty bone]

Hey

[neon leaf]

🙏

[dire goblet]

https://tenor.com/view/pfft-trying-not-to-laugh-laughing-holding-laugh-covering-laugh-gif-10899551814881068905

[frosty gale]

let bro build his chat client 😭 🙏

[misty bone]

Hello

[crimson vapor]

I’m playing guys

Real ones remember veld chat

[humble gyro]

realest shit

[queen needle]

Real

veld chat come back when?

[low marten]

'for no particular reason' 😭

[prime cliff]

We have veld chat at home

[low marten]

I'm thinking of parallel hosting soon, Flask or Quart 🤔

[queen needle]

10/10

[humble gyro]

wait this is kinda goated

[small tangle]

Dont forget to add the Nitro paywalls

[swift barn]

and warp +

smh

[prime cliff]

Already got a bunch of stuff added yea, gonna work on mobile compatibile UI now 😅 i can do android adb

[humble gyro]

woah builderb is adding warp x nitro?

man lunar chat is gonna be epic

[prime cliff]

You want to see a live demo of it Veld 👀

[humble gyro]

yes please

[prime cliff]

DM 🙂

[queen needle]

Any chance I could 👀

[prime cliff]

Sent 🙂

[queen needle]

Awesome thank you

[humble gyro]

very cool

[queen needle]

I agree

[prime cliff]

Thanks ♥️ it's actually my first time using avalonia since 2 months ago so i need to learn how to design it better.
I also want to add more customized features like full user profiles (how guilded works), calendar type channel, more split permissions and other stuff.

Maybe even have like customized component channels for bots a bit like slack

[marsh tapir]

@prime cliff dm?

[prime cliff]

Why

[marsh tapir]

I have to ask something

[prime cliff]

Ask it here pls

@marsh tapir ^

[marsh tapir]

How we can deploy our bot after coding

[prime cliff]

And how am i supposed to know what language or tools you're are using?

[wheat mesa]

You need a host

[marsh tapir]

How

[marsh tapir]

Python

[prime cliff]

You can either buy a server ($3 month) or use a free hosting service (not recommended) such as:
https://render.com/
Pella.app, bot-hosting.net and WispByte

[marsh tapir]

Thnx

[woven comet]

For hosting I'd recommend northflank- Thats what I use for a bot thats in 50+ servers atm. They can be a-tad expensive. But they have a "free plan" for like 3 months and they aint on a fix rate. Its all depending on how much the bot is used.

Example for me its around $15 per month for a bot in 50 servers

so if its a 1 bot you might even be able to get away with free

but even as free they are good

[warm surge]

damn veld moment

[prime cliff]

$15 for 50 servers seems a bit excessive 🙃

[pearl trail]

just get contabo since you don’t really care about performance 🗿

[woven comet]

That also includes built-in health checks, Database and storage.

For host on its own its like $5-7

[spark flint]

Still overpriced

[prime cliff]

Eh i mean if they want extra stuff then sure it's worth it

[spark flint]

Hetzner €3.49 for 4gb ram and 40gb ssd

Install database on there

Health checks can be done for free with https://betteruptime.com

[wheat mesa]

Yeah this is super expensive. For that price you could buy a raspberry pi for the price of 4 months of hosting and be able to host pretty much forever

[prime cliff]

I wouldn't say that now with ram prices xD

[brazen rose]

topgg devs, How do you verify that a user is actually the owner of a bot before adding it? Is this manual? If not, how do you do this in a programmatic way?

I'm desperate for a solution to check the owner applications programmatically.

[deft wolf]

They don't afaik

[brazen rose]

jeez

[deft wolf]

I never tried it because I didn't see the reason for it but I saw some people complaining that their bot was added by someone else

[brazen rose]

YO I actually found a way. Application owned emojis are a way.

it includes the user field and I can verify the asset as well

• https://discord.com/developers/docs/resources/emoji#emoji-object
• https://discord.com/developers/docs/resources/emoji#get-application-emoji

[lament rock]

The user object refers to the bot user

[brazen rose]

I know no bot that allows users to upload to application emojis since there are only 2k

[brazen rose]

?

the last sentence

[deft wolf]

So it could be a bot user

[lament rock]

Alright. I know a few people upload them programmatically

[brazen rose]

if it way uploaded via a command, then yes.

[brazen rose]

well for verification you have to go to the dev portal and do it manually.

that is a neat way I think.

[lament rock]

What if a team is already at the emoji limit and they're unwilling to upload a new emoji

Would you support an existing emoji

[brazen rose]

nevermind, there is a better way with oauth2

[brazen rose]

if the user matches, yes

also, 2k emojis is a VERY high limit

its an acceptable edge case

[lament rock]

builds are a different thing from bots

the app devs would have to submit builds to Discord to show on the now dead Discord games library

[brazen rose]

no, this scope allows you to fetch a route and get a status code apparently which indicates owenership

[lament rock]

[brazen rose]

I'm searching the docs right now xD

nvm

[deft wolf]

Asking on ddevs > searching docs

[warped glacier]

I made a antiraid tool!

[brazen rose]

I did many things lol
I now made a discussion on GH
https://github.com/discord/discord-api-docs/discussions/8035

[frosty gale]

lmao

a system where you invite a topgg bot and your bot into a server and make it output a specific string would be a cool verification method

albeit complex so maybe just in the bots bio

adding dyno to my profile as we speak

[deft wolf]

You can't access bio via API afaik

[brazen rose]

yes, sadly

[thorn shell]

Heya guys quick question:

if (config.topggToken) {
        try {
          const poster = AutoPoster(config.topggToken, client);

          poster.on("posted", (stats) => {
            console.log(
              `📊 Posted stats to Top.gg | ${stats.serverCount} servers`
            );
          });

          poster.on("error", (err) => {
            console.error("❌ Top.gg AutoPoster error:", err);
          });

          console.log("✅ Top.gg AutoPoster initialized");
        } catch (error) {
          console.error(
            "❌ Failed to initialize Top.gg AutoPoster:",
            error.message
          );
        }
      } else {
        console.warn(
          "⚠️ Top.gg token not found in config - AutoPoster disabled"
        );
      }

returns:
❌ Failed to initialize Top.gg AutoPoster: Unsupported client on my raspberrypi5 server
But it fully works in windows10

[wheat mesa]

Are you sure your client is initialized? I imagine that if client is null it would be problematic

[thorn shell]

Found the issue
My node packages had a conflict and for some reason it didn't load correctly
I reinstalled packages and got it to work

$ rm -rf node_modules package-lock.json
$ npm install

[queen needle]

Probably stupid but I just learned a fun way to swap integers without a variable being a middle man:
https://discord.mx/i/pVtv7gl2UA.png

51.68 KB (@pancake)

DiscordMX · Dec 15, 2025, 9:42 PM

[crimson vapor]

Yeah that’s xor for you

[queen needle]

yeah i knew like a^b^b == a but i never correlated that

because im used to python or js where i just do, a, b = b, a

[scenic kelp]

it is indeed stupid

a temp is generally faster

[queen needle]

faster like code execution wise?

id suspect this to be faster since it's just doing bit operations but I don't know much about C languages

[scenic kelp]

yes

the xor trick messes up CPU pipelining

[queen needle]

i would be lying if I said I knew what that meant

[neon leaf]

afaik its also just straight up more instructions anyway

[scenic kelp]

and realistically if it were faster than swapping then it would be implemented by optimizing compilers

[queen needle]

ah yeah that's true

[scenic kelp]

my general advice is that you should only focus on optimizations at the macro level until you know you need them at the micro level

focus on structuring your code in a sensible way

[lament rock]

I knew I wasn't crazy -_-

[prime cliff]

🙂

[queen needle]

yoo awesome

[prime cliff]

I think embeds should suffice for now time to work on message replies next with a max of 3 🙂

[queen needle]

bawesome

[pearl trail]

https://www.threads.com/@githubprojects/post/DSS_0HYkrQG?xmt=AQF0JD5lS5v9rTXwKLpwC13tnMi0sb9XPTws3FC8uI21AvCbauuKliD9w-gqat8tB6Qf5uI8&slof=1 lmao this is funny

actually a good idea xD

[prime cliff]

Just round robin your website so half the people land on the fake error page

[frosty gale]

you dont need to fake one, real ones happen often anyways

[frosty gale]

these look fine at first glance but on second glance its cursed

[harsh nova]

I really like the vscode one

[frosty gale]

the go one 😭