🪅-progaming

[placid cape]

use tailwind

[median root]

sober has no aarch64 build sad

[hoary sluice]

doesnt bedrock work natively on linux

[placid cape]

it doesnt

[jade stone]

Like Minecraft bedrock

[placid cape]

because the app uses UWP on windwos

and they're stupid

[jade stone]

Imsane

[placid cape]

they use UWP to avoid cracking on computer but you can still crack it pretty easily

[leaden crater]

@valid jetty I'll try table later, for now I'm stuck with dd iso inside a dd iso and deserted instructions

[valid jetty]

i dont like it lol

[hoary sluice]

whats a dd iso

[valid jetty]

i have my own abstraction for styles

[hoary sluice]

of course you do

[leaden crater]

cd but dd

dd if=mint.iso of=/dev/sda whilst dev sda has arch iso

very not fun to happen

[valid jetty]

const { merge, styles } = createStyleSheet({
    container: {
        display: 'flex',
        alignItems: 'center',
        justifyContent: 'center'
    },

    button: {
        marginBottom: '4em',
        padding: '2em'
    }
});

export function MyComponent() {
    return <div style={styles.container}>
        <button style={merge(s => [s.button, { color: '#EEE' })}>Button 1</button>
        <button style={merge(s => [s.button, { color: '#DDD' })}>Button 2</button>
        <button style={merge(s => [s.button, { color: '#CCC' })}>Button 2</button>
    </div>;
}

this

[hoary sluice]

whats wrong with that

were you booted into the arch so

iso

[valid jetty]

its extremely simple but so powerful

type StyleSheet = Record<string, React.CSSProperties>;

export const mergeStyles = (...styles: React.CSSProperties[]) => styles.reduce((pre, cur) => ({ ...pre, ...cur }), {});
export const createStyleSheet = <T extends StyleSheet>(sheet: T) => ({
    styles: sheet,
    merge: (callback: (sheet: T) => React.CSSProperties[]) => mergeStyles(...callback(sheet))
});

and the styles are typed and eveything

[lavish frigate]

are you using react for this...

[leaden crater]

i had to get gparted to not fuck up my nvme so i had to install another distro over arch that has pre installed gparted so ya

[valid jetty]

yes because im the most comfortable with it

[lavish frigate]

😭 😭 😭 😭 😭

[valid jetty]

this has some complicated state

[hoary sluice]

why not just install gparted inside the iso

also doesnt the arch iso have gparted

pretty sure it does

[leaden crater]

can't

[leaden crater]

nop

[hoary sluice]

why not

why not use fdisk

[leaden crater]

gparted is more safe

I'm gonna lose it if i lose my windows install

i stayed like 40 mins to fix mirrors anyways

the iso was pretty much fucked and i didn't want to download again

[valid jetty]

• when width and height change, the matrix is resized but it keeps all the existing values in it
• when editing the matrix, values go into a seperate scratch matrix which overrides the actual matrix when you exit edit mode
• when you perform an iteration of the algorithm, the previous state of the matrix is stored in a set so you can undo the operation
• all of that is statefully and automatically synced with localStorage via a hook i wrote, and it acts exactly like normal state
• its passed through context too, to all the components that need each piece

its pretty complicated lol

im sure i could do that with something like svelte

and i was tempted

but i dont have time

[lavish frigate]

i wasnt suggesting svelte 😭

[placid cape]

jquery

[valid jetty]

i have no idea how i would do this in another framework lmao

import { useState, useLayoutEffect } from "react";

export const useStorageValue = <T>(
  key: string,
  cb: (x: string) => T,
  def: string | null = null,
  formatter?: (x: T) => string,
) => {
  const [value, setValue] = useState<T>(cb(localStorage.getItem(key) ?? def!));

  useLayoutEffect(() => {
    localStorage.setItem(key, (formatter ?? String)(value));
    // eslint-disable-next-line react-hooks/exhaustive-deps
  }, [value]);

  return [value, setValue] as const;
};

const rows = useStorageValue("rows", Number, "4");
const cols = useStorageValue("cols", Number, "10");

const oldRows = useStorageValue("rows", Number, "4");
const oldCols = useStorageValue("cols", Number, "10");

const editing = useState(false);

const matrix = useStorageValue<number[][]>(
  "matrix",
  JSON.parse,
  "[]",
  JSON.stringify,
);

const scratchMatrix = useStorageValue<number[][]>(
  "matrix",
  JSON.parse,
  "[]",
  JSON.stringify,
);

const prevStates = useStorageValue<Set<number[][]>>(
  "prevStates",
  (x) => new Set(JSON.parse(x) as number[][][]),
  "[]",
  (x) => JSON.stringify(Array.from(x)),
);

[lavish frigate]

i mostly asked cuz i dont really see how react makes it so much easier i wouldve suggested vanilla since we arent in 2006 anymore and vanilla is powerful enough

[hoary sluice]

why not use dioxus

[lavish frigate]

🔥

[lavish frigate]

im planning on making something in vanilla js ts with rust backend soon

[valid jetty]

vanilla is viable but again this is complicated state which can be expressed more beautifully and concisely in react

[hoary sluice]

why not use dioxus

[placid cape]

nanostores

[lavish frigate]

cuz its 1 page

[placid cape]

https://github.com/nanostores/nanostores

[lavish frigate]

and i dont need components

or anything like that

[lavish frigate]

also better user experience since fast and only having a minified js file is better than a giant 5 million gb bundle with react or dioxus and all that garbage

[placid cape]

astro 🔥

[valid jetty]

for the record this deploys a single react file lol

[lavish frigate]

i dont need 50 files on my system all used to create one file either

[valid jetty]

its 188kb ungzipped

[lavish frigate]

who the hell thought of all this garbage?

[valid jetty]

which is high but whatever

[lavish frigate]

i love that rsbuild has an x on the icon (it works perfectly fine)

i despise the current js ecosystem is so much everything for nothing

[valid jetty]

yeah i agree with that

thats a lot of files

i was gonna try something new but the simplex tableau is a complicated thing

if i got it wrong it wouldve been hell to debug, and it was already hell enough to debug the state even in react

i might rewrite it

@hoary sluice remember this https://unofficial-d25-2026.netlify.app/

[hoary sluice]

did you ever get it to work

you tried making a really hard aoc and instead made a really easy typing speed test

[valid jetty]

😭

[formal belfry]

workspaces moment

[valid jetty]

i gave up iirc

maybe i got it to work

[valid jetty]

@hoary sluice @leaden crater i think its fully finished !!!!

i finished a hobby project in a day thats a record for me

[hoary sluice]

didnt you write ichigo in a day

[hoary sluice]

what is this evil magic

[valid jetty]

2 days if you count just the parser and compiler

otherwise 3 days if you count the lexer too

[valid jetty]

Linear programming is a technique for finding the minimum (or maximum) of a linear function of a set of continuous variables subject to linear equality and inequality constraints.

[hoary sluice]

we should configure nvim together

[valid jetty]

nvchad blobcatlightpink~4

[hoary sluice]

nvchad isnt perfect

[valid jetty]

works well enough and i still use zed mostly anyway xd

[hoary sluice]

ive read this so many times it doesnt help

[valid jetty]

only time i use nvim is when im sshed into my thinkpad

[valid jetty]

basically

imagine you have a company

the company makes like.. paints A and B

[hoary sluice]

what's a paint

[valid jetty]

to make paint A you need 35 of ingredient X and 42 of ingredient Y and it makes you $25

to make paint B you need 38 of ingredient X and 33 of ingredient Y and it makes $22

you have like 2000 of X and 1500 of Y

how much paint do you make of each kind such that you maximize your profit

[hoary sluice]

aoc claw machine problem

[valid jetty]

its basically that but on a larger scale (3+ constraints)

at which point you cant graphically solve it by intersection anymore

[hoary sluice]

can i have unlimited constraints

[valid jetty]

not really that was just systems of equations

[valid jetty]

yeah thats kinda the point of the simplex algorithm

to extend it to as many constraints as needed

[hoary sluice]

but isnt this also a system of equations

[valid jetty]

i mean yeah but not really because theyre inequalities not equations

they become equations when you introduce slack variables

but then you have more variables than equations so you cant solve them like systems of equations

[valid jetty]

thats where the tableau comes in ^^^

[hoary sluice]

i mean aoc was you have product A and product B, getting A takes a x steps and b y steps and getting b takes c x steps and d y steps

right

isnt that the same

[valid jetty]

yeah but that was solveable directly

[hoary sluice]

how do i use it

[valid jetty]

try to like see how the inequalities map to the tableau

i recommend watching a video about this if you really wanna learn

its hard to explain without going into a lott of detail

[hoary sluice]

this looks useful for aoc

[valid jetty]

i mean it might be

but its a bit evil to get a question on this

[hoary sluice]

theres usually multiples solutions for aoc problems

[valid jetty]

this is essentially my solver impl

export function pivot(
  matrix: number[][],
  row: number,
  col: number,
  rows: number,
) {
  const newMatrix = matrix.map((row) => [...row]);
  const p = newMatrix[row][col];
  newMatrix[row] = newMatrix[row].map((val) => val / p);

  for (let i = 0; i < rows; ++i) {
    if (i === row) continue;

    const m = newMatrix[i][col];
    newMatrix[i] = newMatrix[i].map((val, j) => val - newMatrix[row][j] * m);
  }

  return newMatrix;
}

export function autoPivot(matrix: number[][], rows: number, cols: number) {
  // Ignore the RHS row when finding the pivot column
  const c = matrix[0].findIndex(
    (x) => x === Math.min(...matrix[0].slice(0, matrix[0].length - 1)),
  );

  if (c === -1) return;

  // Find the row with the smallest ratio
  const ratios = matrix.slice(1).map((x, i) => {
    const pivot = x[c];
    return pivot > 0 ? matrix[i + 1][cols - 1] / pivot : Infinity;
  });

  const r = ratios.findIndex((x) => x === Math.min(...ratios));
  if (r === -1) return;

  return pivot(matrix, r + 1, c, rows);
}

export const isOptimal = (matrix: number[][]) =>
  matrix[0].slice(0, matrix[0].length - 1).every((x) => x >= 0);

const solve = () => {
  const solvePrevStates: number[][][] = [...prevStates];
  let solveMatrix = matrix.map((row) => [...row]);

  while (!isOptimal(solveMatrix)) {
    const matrix = autoPivot(solveMatrix, rows, cols);
    if (!matrix) break;

    solvePrevStates.push(solveMatrix);
    solveMatrix = matrix;
  }

  setPrevStates(setWithDedupe(solvePrevStates, deepEqual));
  setMatrix(solveMatrix);
};

its not that big but probably too big to write in the middle of an aoc

especially considering you have to actually create the tableau

[hoary sluice]

library

[valid jetty]

i guess lol

i wouldnt recommend just refactoring into kt and putting into a lib tho

it wont be helpful at all if you dont know how to use it, like you wont even catch the cases where you "could" use it

[hoary sluice]

return LP().solve("M = m - 5").ineq("m <= 9p2 + 6p3 + 11p4").ineq(/* etc */)

[hoary sluice]

im obv gonna learn how it works beforp making it

[valid jetty]

lmao good

[valid jetty]

that sounds like pain to parse

[hoary sluice]

i only have to write the parser once

[valid jetty]

ACTUALLY

you might find this useful https://rosiepie.notion.site/Power-Rule-e358c822af0d4136b90906ca708b8880?pvs=74

[leaden crater]

project sekai???

[valid jetty]

or well https://bit.ly/cute-power-rule

[hoary sluice]

how are you pulling random articles you wrote out of your ass

[valid jetty]

LMAO

[hoary sluice]

is power rule

f = x^2
f' = 2x

[valid jetty]

yes

but it only works if the power is a constant and the base is the variable you’re differentiating with respect to

otherwise you have to use chain rule or something

[hoary sluice]

i didnt know they had names i thought u kinda just did that

[valid jetty]

lol nope there’s like 5

constant rule, power rule, product rule, quotient rule, chain rule

[solar thistle]

typical low-level developer

[hoary sluice]

disappear, power -> mul, product stays constant, quotient -> ln, inner derivative * outer derivative?

[valid jetty]

product and quotient not quite

[leaden crater]

all these rules when all you need is a calculator tsch

[solar thistle]

one of the rules forbid calculators

time to get murdered by grammar police

[valid jetty]

d/dx[f(x) * g(x)] = (f’x)(g(x) + (f(x))(g’x)

[hazy pine]

i thought this was pro gaming not math :(

[hoary sluice]

i think we learned that and i frgoro

[valid jetty]

d/dx[f(x) / g(x)] = ((f’x)(g(x) - (f(x))(g’x))/(g^2(x))

[hoary sluice]

lisp

[valid jetty]

chain rule is d/dx[f(g(x))] = f’(g(x)) * g’x

[leaden crater]

ok whats the integral of the 7th derivative of angular momentum of an orbit with respect to the 2nd derivative of time in Gyrs

[hoary sluice]

most likely 0

[solar thistle]

OOM error killing host process

[valid jetty]

statistically pi or pi/2 and i don’t like it

[hoary sluice]

youre effectively taking the 6th derivative of some formula, theres a pretty good chance its gonna be 0

[valid jetty]

oh true

[leaden crater]

nop, eccentricity + gravity -> not zero

[hoary sluice]

and the 2nd derivative of time is also 0 and integrating with respect to 0

[leaden crater]

i think this fits https://en.wikipedia.org/wiki/Perturbation_theory

[solar thistle]

stop being big brain guys my dumbahh can't comprehend

[leaden crater]

2nd derivative of time is acceleration

[hoary sluice]

no its not

thats the 2nd derivative of position wrt time

d''x/dt''

d''t/dt'' is 0

unless youre in a black hole or something stupid like that

where time is like a polynomial or smth

[leaden crater]

7 dimensional time

[hoary sluice]

u mean 3 dimensional

cause we need d''

[leaden crater]

nop

12 dimensional everything

[hoary sluice]

have fun integrating with respect to 0

@valid jetty whats the 7th derivative of x with respect to 0

[leaden crater]

have fun not integrating

i like to math

[hoary sluice]

have fun taking a disrespectful derivative (derivative w.r.t. 0 get it haha)

[median root]

alr lets play my fav game, where are the 42 errors in my code comming from after restarting my lsp?

[median root]

ah

[hoary sluice]

me and u

(ur the cat)

(stop programming so much and work out)

[valid jetty]

WHAT

evil,,

[hoary sluice]

there is no approximate solution to an undefined equation

[supple whale]

rookie shit, 3 way development is where its at

[valid jetty]

nice start to april i guess

its not gonna last tho because i have about a month until my end of year exams so i have to focus on those instead

ive been recommended to do 6 hours of revision a day every day for the next month

[lavish frigate]

https://vxtwitter.com/HSVSphere/status/1908089924584083544

HSVSphere (@[e-mail])

Vibe coders, visualized:

QRT: buitengebieden
Birds are fed by their parents in their infancy. When the time comes to feed themselves, there can be some confusion when the food does not go into their mouth by itself..

▶ Play video

vxTwitter / fixvx · Apr 4, 2025, 9:31 AM

[dense sand]

Should i switch from make to meson/cmake

[dense sand]

@supple whale do you have somewhere the args i should use when compiling with emcc

[dense sand]

i thought that 1024kib would be enough to run java program 😭

[fleet cedar]

Lol, you think 1mb is enough for anything?

[dense sand]

wait mb i meant 1024 bytes

💀

well the heap filled up instantly, but i didnt run gc

[fleet cedar]

Have fun with that

[hoary sluice]

@valid jetty https://youtu.be/lMJYGvdNd04

how did i not know abt this

[supple whale]

I posted them before

[simple sapphire]

april fools !

[valid jetty]

this is cool but have you seen ci”

jump to the nearest “, clear everything until the next “, and enter insert mode

[hoary sluice]

of course

this is trivial

[dense sand]

im thinking if i should rewrite my heap again and instead of having preallocated block i should just use the platform's malloc implementation

any tips on deciding on this?

[valid jetty]

@leaden crater @hoary sluice YOU RUINED IT I HAVE ROMANIAN ON MY おすすめ NOW

[leaden crater]

[hoary sluice]

romanian math edits are crazy

[leaden crater]

is there even a solution besides a=b=c=1

[dense sand]

what's wrong with my gpt 😭

[fleet cedar]

Yeah gpt sometimes grows an attitude

Much more fun than its default "personality"

[lavish frigate]

exposing yourselves as ai loving vibe coders...

[dense sand]

i just need someone to discuss the pros and cons of some stuff sometimes... i dont really use it to code anything

[lavish frigate]

mhm

leak entire chatgpt history NOW!

[valid jetty]

yeah i do that too sometimes

[dense sand]

ive never deleted the chats with gpt

so like

i could

[valid jetty]

however isnt that just a modern version of the coding duck lol

[dense sand]

yea

[valid jetty]

i use my plushies instead for that purpose

[dense sand]

i was actually thinking of making a big 3d printed duck with rpi and speaker inside wired to openai api

[lavish frigate]

we shouldve trained chatgpt on stackoverflow correctly

[dense sand]

its still just a glorified search engine

[lavish frigate]

we as a society NEED it to sometimes just reply with "You actually broke rule 5 which is no duplicate questions so i cannot help with your request 🤓"

[valid jetty]

"hey chatgpt how do i solve this problem"

"fuck you i cant answer that, somebody already asked me this question at some point in the past with similar context, just ask them what their answer was lol"

[fleet cedar]

Search engine and copyright violation engine

[nimble bone]

@valid jetty is a vibe coder

[lavish frigate]

"Can't you just Google that yourself?"

[valid jetty]

LMAO

@nimble bone hiiiii do you know how to use this https://acquitelol.github.io/simplex/

[nimble bone]

nop im stupit

[valid jetty]

im trying to find ANYONE who knows what this even is

the only people who know are people who do further maths at my school

!!! (its so pointless)

[dense sand]

i fucking hate vibe coders, in our class we had to write a test to try out tdd and then exchange it with other group to write the implementations. we received gpt generated code which didnt even compile mainly because they were passing strings to enums everywhere for some reason

when i confronted them about it they were like "well its not our problem, you have to work with it"

like man with this attitude you wont be getting paid for this

i ended up just rewriting everything from the ground up

[valid jetty]

people who vibe code are like

developed ipad kids

ipad kids who grew up

@nimble bone i have banana milk 😋 😋 😋 😋

[nimble bone]

how

give

[valid jetty]

nooo

mine

:3

[leaden crater]

@valid jetty

[valid jetty]

hiii

[leaden crater]

hiiii

can you help me with a problem i saw, i tried both calcualtor, photomath, wolfram and ||ai|| and none got it right

maybe im stupid and there doesnt exist a solution at all

the logs arent multiplied so i think its impossible

[valid jetty]

what is your goal

a simplification or a decimal approximation

[leaden crater]

simplification

[deep mulch]

@valid jetty are you a vibe coder

@valid jetty code a procedural tree generator with elle

[fleet cedar]

Be a procedural tree generator

||Also known as a gardener||

[dense sand]

In real world, no tree is binary

[valid jetty]

@deep mulch ```py
def parity(n):
return "eovdedn"[n % 2 :: 2];

odd even

print(parity(5), parity(8))

[deep mulch]

NO

add js-like type coercion @valid jetty

[wispy kernel]

What is the correct answer?

[leaden crater]

@deep mulch

IDENTIFICATION DIVISION.
PROGRAM-ID. COLLATZ.
DATA DIVISION.
WORKING-STORAGE SECTION.
01 NUM PIC 9(10).
01 TEMP PIC 9(10).
PROCEDURE DIVISION.
MAIN-LOGIC.
    DISPLAY "ENTER A NUMBER: "
    ACCEPT NUM
    PERFORM UNTIL NUM = 1
        DIVIDE NUM BY 2 GIVING TEMP REMAINDER TEMP
        IF TEMP = 0
            COMPUTE NUM = NUM / 2
        ELSE
            COMPUTE NUM = NUM * 3 + 1
        END-IF
        DISPLAY NUM
    END-PERFORM
    STOP RUN.
    

[valid jetty]

[deep mulch]

mm Fortran

[leaden crater]

WHAT

[deep mulch]

I love Fortran

[leaden crater]

its not fortan

[deep mulch]

oh

cobol

[leaden crater]

catbol

[deep mulch]

or forth

@leaden crater

[shrewd canopy]

questionable dolfies

[hazy pine]

is that unico

[runic sundial]

@royal nymph

[hazy pine]

vencord rust rewrite when

[ionic lake]

today

[mellow phoenix]

less than a year ()

[supple whale]

man, imagine if discord actually let us make 3rd party clients

wait actually, does is the new discord game integration available to every1?

cant u just use it to make a fake game, and ur own simple discord client, in rust?

[shrewd canopy]

it lets you if you match all requests with requests from official clients 1:1

[shrewd canopy]

yeah you just fill form and they give u access

however its oauth2 and has pretty limited functionality

[supple whale]

@hazy pine welp, vencord rust time!!!!

[supple whale]

i assume, read dms, write dms and read friends list?

and thats it?

[hazy pine]

i am not talented enough for this

theres alot of 3rd party client we'll be finee

what if we just make vesktop but with tauri, would that be rust enough

[supple whale]

yeah but they are bannable

[shrewd canopy]

read list of guilds, their channels, roles, your friend/blocked users/ignored and friend request list, update your status/rich presence etc

[shrewd canopy]

nah not really lol

[supple whale]

ay, thats quite a lot

[hazy pine]

but if its an exact 1:1 recreation of the regular client, then likee

is it really a problem

[supple whale]

every single person i know which tried their own non-web based discord client got banned

[shrewd canopy]

then they didn't bother to mitigate antiabuse systems

if you send an appeal and explicitly say you used third party client then you'll unbanned

[supple whale]

yeah i think the game integration thing is x100 safer

[fleet cedar]

I know one guy who's been running his own for years

[supple whale]

and easier

[shrewd canopy]

ya oauth2 almost has essentially no antiabuse systems

[supple whale]

https://github.com/electron/electron/issues/41578

https://cdn.discordapp.com/emojis/765237394869190666.webp?size=48&name=deadinside~1

fucking half the thing stops rendering

[winged mantle]

I made a cursed solution for logger lazy evaluation

(if level is hidden ?. returns undefined)

(also allows checking whether level is shown with if (logger.info)

[royal nymph]

lmao

[supple whale]

wouldnt it be better to simply change the function to a noop instead of console.info when the level is disabled?

and not have the extra markup?

[winged mantle]

it's so that it only evaluates the message expression if the level is enabled

in case you call a really expensive function for some reason

before i had it execute the message if it was a function

but i realised that would be an ideal weak point for rce

[supple whale]

fair

[winged mantle]

looks good at least

[supple whale]

ah yes escape codes

i fucking hate it

i recommend the debug package if u want to find more codes and how to find which are supported

but it makes my blood boil, because while its nice in console, when logged to a text file its fucking awul

[winged mantle]

I think i will implement logging to file separately

[supple whale]

i log all my console messages to a logs.txt file, and opening that makes it awful

[winged mantle]

which will be fun

maybe i can also detect if escape codes are supported

[supple whale]

i did it via console, because its legit just piping stdio into a file

so its so laughably easy its crazy

[winged mantle]

but i don't think piping into a file works with stderr anyway

[supple whale]

https://www.npmjs.com/package/supports-color

cant beat chalk for it

:^)

[winged mantle]

if i'm actually being sensible and using libs i might as well find a logger lib

although none of them looked like much fun to me for some reason

[supple whale]

https://www.npmjs.com/package/debug

by far the BEST logging lib i used

but it has issues when you use electron logger

since you cant easily just disable or change colors for browser

[winged mantle]

thanks but i think i'll see how painful it is to write to file first

you'd probably want it to archive the file every time it's midnight and gzip it

[supple whale]

what are u running?

node/deno?

[winged mantle]

node

i have not gzipped a file before

[supple whale]

thats just a utility lib for logging to console.log

that lets u enable/disable different debugs

[winged mantle]

i saw it supports multiple streams in a quick skim

[supple whale]

so u can do DEBUG=chat:* and it debugs for example chat:message chat:event

[winged mantle]

kinda useful

[supple whale]

ya do

import Debug from 'debug'

const debug = Debug('chat:message')

debug('emitted message: %s', message)

[winged mantle]

my solution was just to try not to litter the log with too many messages

[supple whale]

and it just outputs to console.log

and then you pipe console.log aka stdio to a file yourself

which is piss, like 4 LOC

import { createWriteStream } from 'node:fs'

process.stdout.pipe(createWriteStream('/log.txt', { flags: 'a' }))

[winged mantle]

i heard that console.log is blocking

is this true

[supple whale]

well yes ofc

its a sync function

[winged mantle]

but does it wait for the syscall 😭

[supple whale]

no, its a stream

[winged mantle]

yeah that would be weird

[supple whale]

the initial console.log call is sync

but then its passed to a stream, which is event loop based

[winged mantle]

i assume it uses a buffer

[supple whale]

from what i understand

[winged mantle]

but i would've thought the actual syscall would be on another thread

[supple whale]

no, its on "the same thread"

or the same process

its kinda weird with how nodejs does native bindings

because its the same thread, but its async from the v8 vm itself...

and its fucky

the less you know the better for your sanity

[winged mantle]

sounds about right

[dense sand]

What the hell is there even a good auth library in js?

Better auth works not works

Authjs just sucks

I dont want to switch to supabase duh

Do i just roll my own auth

[hoary sluice]

just use dioxus

[dense sand]

I need something compatible with nextjs

If i were to use other lang for backend id just use java

[hoary sluice]

dioxus is fullstack

[dense sand]

Im not using rust because its a group project

[hoary sluice]

perfect, convert your group as well

[dense sand]

They are vibe coders tho

[leaden crater]

w.. what..

[fleet cedar]

Assassinate them

[runic sundial]

fucking love llvm

i love having to work around

CACHING

and CONSTANT FOLDING

[lavish frigate]

large language... v... machine-learning (ai)

[runic sundial]

I've had 2 memory related issues with C++ thus far, most of what's fucked with this demented language is how people are like

"DAMN BRO, BUT WHAT IF WE MADE IT FASTER?????"

Shit lang where I guess it's common practice to optimize for the sake of optimization

But now my tests are passing again

[fleet cedar]

Isn't that gep ub?

[runic sundial]

ub?

unpossible

@fleet cedar what the hell is ub

[fleet cedar]

Undefined behavior

[runic sundial]

in what way?

[fleet cedar]

InBoundsGEP requires the offset to be in bounds

[runic sundial]

troll~2

Well it is, it is in bounds once it reaches the verifier

but it actually doesn't seem to give a shit if it's the first index that's out of whack

It's the ones after that where it actually validates, since it would need to calc the machine offset

Either way. The pain here is that I'm re-mapping GEP offsets.

And this fucker has been giving me a bad time

Where it will gleefully alias instances of GEPOperator

[sage smelt]

how do i make a plugin click on a message's button using the button's customid?

[leaden crater]

is this see plus plus

[sage smelt]

for more context, i get the messages using MessageStore.getMessages(channel.id);

[shrewd canopy]

whats that for

[hollow patio]

How evil can you possibly be
Me:

[shrewd canopy]

i love

(i want zod but for python)

[balmy lintel]

isnt that kinda selfbot behaviour

[winged mantle]

why does thei nvite endpoint for bots only show friend invites if you remove the bot prefix of the token

[supple whale]

i rly need to get better at using abortsignal

esp for like destroying classes and shit

[winged mantle]

javascript destructor when

[fleet cedar]

I've never seen a GC language with proper dtors

[winged mantle]

does anyone know

hiddenphox uses fetch manually

hf!inviteinfo aABKp5XD

[pearl stag]

Invite Info: `aABKp5XD` (Friend)

Expires

<t:1744634980>

Inviter

TheKodeToad (@thekodetoad) (706152404072267788)

​

Avatar

[winged mantle]

where is discord staff in this server...

[valid jetty]

this is interesting

#define __DEFER_INTERNAL__(f, v) \
    auto void f(char*); \
    char v __attribute__((__cleanup__(f))); \
    auto void f(char*)
#define __DEFER_WITH_COUNTER__(x) \
    __DEFER_INTERNAL__(__f##x##__, __v##x##__)
#define __DEFER__(N) __DEFER_WITH_COUNTER__(N)
#define defer __DEFER__(__COUNTER__)

int main() {
    int *x = malloc(6 * sizeof(int));
    defer { free(x); };

    return 0;
}

[supple whale]

already exists

using

[winged mantle]

uh

[supple whale]

which calls a destrictor when an object exits scope/wants to be GC'ed

[winged mantle]

does it really work like in c#

i thought it worked like

using(console) { log("h") }

[supple whale]

https://www.typescriptlang.org/docs/handbook/release-notes/typescript-5-2.html

[winged mantle]

and was very discouraged

[supple whale]

class TempFile implements Disposable {
    #path: string;
    #handle: number;
    constructor(path: string) {
        this.#path = path;
        this.#handle = fs.openSync(path, "w+");
    }
    // other methods
    [Symbol.dispose]() {
        // Close the file and delete it.
        fs.closeSync(this.#handle);
        fs.unlinkSync(this.#path);
    }
}
export function doSomeWork() {
    using file = new TempFile(".some_temp_file");
    // use file...
    if (someCondition()) {
        // do some more work...
        return;
    }
}

[winged mantle]

oh

is this new

[supple whale]

its not implemented yet

https://github.com/tc39/proposal-explicit-resource-management

[winged mantle]

ah this is the one

kotlin simulator

[supple whale]

wtf is that

[winged mantle]

kotlin inspired

[supple whale]

awful!

[supple whale]

i mean

i think it was created before kontil went open source

[deep mulch]

delete

[winged mantle]

i was just taking a chance to bash kotlin

[supple whale]

beats java tbf

[deep mulch]

no one should write Kotlin code like that

[winged mantle]

nah "i program java how did how did you know" [gigachad gif]

[deep mulch]

with is intended for changing the context of a scope

[winged mantle]

i mean it adds good things to java

i just don't like some of the other things

the only time i loved kotlin was in a dream

[deep mulch]

kode tode

[winged mantle]

that is completely true

iirc

i don't remember the dream now

but i remember posting about it

[deep mulch]

tode.kode

[shrewd canopy]

pretty boring in c++ tbh

[deep mulch]

delete

[hoary sluice]

@valid jetty i rizzed up my economic teacher to give me an A if i do checkup on friday

im gonna graduate with a perfect gpa now

[valid jetty]

lmao insane

[deep mulch]

@valid jetty hiii

[hoary sluice]

@valid jetty https://youtu.be/RmqgyRsCfJs

[jade stone]

guhhhh disjoint unions so insane in typescript

https://stackoverflow.com/questions/54908193/refining-typescript-disjoint-union

[balmy lintel]

what the hell is that

[balmy lintel]

ai thumbnail..

[leaden crater]

@valid jetty how hard do you think it would it be to make a mahjong discord bot

[valid jetty]

jai has this

i should put this in elle

[valid jetty]

i meannn isn’t a mahjong clone pretty hard in general

[winged mantle]

do you really need to separately fetch response message with discord api

i guess this is why people don't do buttons based on message ID

I made a lookup with message id, button id

[winged mantle]

so 1358876721079910431, next_page could retrieve data about the command and then use it to respond

i guess uuids is better?

[deep mulch]

Jorp!

@valid jetty later I will add SSTV encoding/decoding to my bot

[valid jetty]

rewrite it in rust

[deep mulch]

nop

[shrewd canopy]

a interaction response message? they recently added a way to immediately get message right after responding, eg for dpy: https://discordpy.readthedocs.io/en/stable/interactions/api.html#discord.InteractionResponse.send_message

[leaden crater]

dont think so

[winged mantle]

the message is optional and is not present for me

[shrewd canopy]

Are you responding with message update?

[winged mantle]

I am responding by creating a message

but it should be present
** Only present if type is either CHANNEL_MESSAGE_WITH_SOURCE or UPDATE_MESSAGE.

time to monkey patch fetch to work out what oceanic is doing again

[supple whale]

has any1 ever overwrote types in a typescript extended class?

class HashSet<T extends {id?: string}> extends Map<string, T> {
  constructor (iterable: Iterable<T> = []) {
    super()
    for (const o of iterable) this.add(o)
  }

  _getId (o: T) {
    return o.id
      ? o.id
      : JSON.stringify(o, Object.keys(o).sort())
  }

  has (o: T): boolean {
    return super.has(this._getId(o))
  }

  add (o: T) {
    super.set(this._getId(o), o)
  }

  delete (o: T): boolean {
    return super.delete(this._getId(o))
  }
}

its crying about

[lavish frigate]

Map<string, T> the string is the key type and T the value type
when you are doing has (o: T) for example you are trying to use the value type when it expects the key type, you also dont copy the method parameters correctly

class HashSet<T extends {id?: string}> extends Map<string, T> {
  constructor (iterable: Iterable<T> = []) {
    super()
    for (const o of iterable) this.add(o)
  }

  has (key: string): boolean {
    return super.has(this._getId(o))
  }

  ...
}

[shrewd canopy]

it somewhat works :)

[leaden crater]

@hoary sluice https://users.math.msu.edu/users/gnagy/teaching/ode.pdf See Chapter 4

[valid jetty]

wait until you heard about 2nd order pdes

❤️

[leaden crater]

pdes?

[valid jetty]

partial differential equations

[leaden crater]

ohhh

[valid jetty]

with partial derivatives

have u done this thing yet

[leaden crater]

ordinary differential equations by vladimir i. arnold i believe is a pretty good book for its stuff

[valid jetty]

https://en.wikipedia.org/wiki/Solid_of_revolution

[leaden crater]

nop, not at school,,

tho im more advanced that what is school

[valid jetty]

school is designed for the slowest person

[leaden crater]

i think i learned abt integrals in 6th grade myself

[valid jetty]

integrals in 6th grade????

werent you like 11

[leaden crater]

yea from youtube

uhhh

[valid jetty]

wtf 😭

[leaden crater]

i was 13 then

i still dont know how to solve integrals that are complex

i learned abt log, complex numbers + complex plane, derivatives, integrals, lambert w , hyperoperators, transcendental numbers, etc a few years ago,,, but i have to revise it

blackpenredpen, 3b1b, vsauce, etc were pretty interesting to watch

i did a lot during quarantine but not integrals, my mom said she wouldn't help me because "im too young"

now i've been interested in literature recently not much math..

[pine prawn]

That actually reminds me alot of myself. Are you by chance ASD?

[leaden crater]

asd?

[frosty obsidian]

autistic

[pine prawn]

Autism spectrum disorder

Cause what you are describing is a really common symptom :3

[leaden crater]

nop probably im not

[pine prawn]

Why does that sound so much worse ;-;

[frosty obsidian]

it doesn't

[leaden crater]

my grades at school are stagnating at 50% almost failing anyways, math didnt help

[pine prawn]

At what subjects?

[leaden crater]

everything

only math, english, history is higher than 80%

[pine prawn]

Is (secondary) language stressful or hard for ya?

[leaden crater]

everything is

i have so many unfinished things because i really dont know where to start from

[pine prawn]

I feel like im looking at a mirror

[pine prawn]

Thats a symptom in both adhd and asd.. like you are basically describing me

[leaden crater]

i was supposed to make project abt smth in german literature, didnt do it for 3 months, only at the deadline and even then from 11pm onwards

[pine prawn]

Oh you got german too?

[leaden crater]

yea

3 languages

[pine prawn]

.. yeah we are clones

[leaden crater]

at least im happy that i got to pass b2 exam

i hardly knew anything before hand

[pine prawn]

Do you find busywork difficult?

[leaden crater]

i was typing without thinking and then i passed, i dont think at any essay, i feel like im putting words from a table

[leaden crater]

busywork?

[pine prawn]

Repeated similar tasks meant mostly to keep you busy

For me, 30 questions with the same formula is just hell. I just make a program to do it for me after the 3rd question

[leaden crater]

yea kinda

[valid jetty]

lmao in lockdown i was playing minecraft

i only started getting interested in that stuff AFTER lockdown

[leaden crater]

if i have to use same thing i just do like

• // - or Let {thing} be ()1 , {thing} be ()2 etc and use those as replacement

[leaden crater]

i made a castle on the school laptop during lockdown

:3

probably is not there anymore, but it was so boring to watch all useless studies

so when we had to go to school i just used the laptop for "work"

[pine prawn]

Mhm! If i understand the concept then why manually calculate it

[leaden crater]

i used to abbreviate essays

same word replaced by ()1 , ()2, etc

or sentences

etc

arrows

[pine prawn]

Oh? So you struggle with long essays?

[leaden crater]

yea but im getting better

when i have to write essay for literature i make weird sketches with many arrows and tree-like structures and it just works

[pine prawn]

So you need things structured how you like it?

[leaden crater]

yea

i could do an essay on anything if i had the struct

[pine prawn]

Is hand placement awkward?

[leaden crater]

eh?

[pine prawn]

Dino hands?

[leaden crater]

i just write

no

[pine prawn]

I mean like in general

[leaden crater]

normal hands..

[pine prawn]

Alrightt

Im trying to not sound rude but, id recommend speaking with a professional :D

[leaden crater]

nop

[pine prawn]

Cause to me, it genuinely sounds like you might just have some form of ASD

[pine prawn]

Oke

[leaden crater]

i dont need professionals, if i dont know it it stays like that

@pine prawn planes or trains

[valid jetty]

tbh i probably have some form of thing

i get hyperfocused on advanced math and computer science but do terribly in stuff like biology and chemistry

or like stuff i don’t pick up any of what im taught in school

like how i got 9/33 in physics and 30/30 in cs

lmao

[leaden crater]

bio and chem so bad..

[valid jetty]

bio and chem were my worst enemy in school

[leaden crater]

i cant get how people learn biology its literally 10000 pages of learning

[valid jetty]

it’s just pure memorization

[leaden crater]

i cant memorize biology..

at all

[valid jetty]

lmao yeah

[leaden crater]

for chem currently we have organic chem i told the teacher sorry, cant do it

[pine prawn]

Do i have to choose-?

[leaden crater]

yes.

[lavish frigate]

Trains obviously

[pine prawn]

Fuck no, memorisation sucks

[valid jetty]

i pick trains because most of my friends are trains

[leaden crater]

nownow

train plane

[lavish frigate]

Who turned your friends into trains

[valid jetty]

the liberal agenda..

[pine prawn]

Rules and algorithms are very good doe

[lavish frigate]

WOKE!

[pine prawn]

Trains..

[valid jetty]

intuitive knowledge is the best…..

[pine prawn]

I mean.. if it makes sense it makes sense :3

[deep mulch]

insanity

[leaden crater]

i wish sky planes existed and we had to drive out from the atmosphere

[valid jetty]

learning the roots or base of something and then it making sense directly to built upon it and make out more knowledge is the best way to learn

cs is literally EXACTLY that and that’s why i love it

[deep mulch]

Rosie is a savant

[leaden crater]

i want to make my own universe

[deep mulch]

@leaden crater let's make one

[pine prawn]

Ikr! Either it needs to have rules and predictable algorithms, or it should be intuitive in a way that makes experimentation fun and semi predictable

[leaden crater]

impossible

[pine prawn]

Well-

[valid jetty]

yeah ^^^^ that’s EXACRLY why cs is so good

[pine prawn]

Ikrrr

[leaden crater]

not even with the most complex VR humanity possess, the feeling wont be the same

[valid jetty]

you just fuck around and find out and learn in the process

[pine prawn]

Oh you were talking about vr?

[valid jetty]

i wonder if there’s a way to make a completely DIY computer

[leaden crater]

yea

[leaden crater]

proabably

[deep mulch]

@valid jetty program an ASIC

[pine prawn]

I thought you meant a literal, physical, mini universe

[deep mulch]

or no

FPGA

[valid jetty]

like you manually mine out the materials, make the cpu and other parts from scratch, then write your own language and then kernel and then stdlib and then OS

[leaden crater]

a friend of mine works with vhdl so probs possible

[deep mulch]

@valid jetty build a slayer excitor circuit

[leaden crater]

possible

i would do it

[valid jetty]

possible but does it exist

[leaden crater]

i will do it

[valid jetty]

do it.

[deep mulch]

@leaden crater is simulated on my pc

[valid jetty]

@deep mulch i’m an ai

[deep mulch]

I know

[pine prawn]

I mean.. how do you think modern pc’s were first made~?

[leaden crater]

i dont understand how antique people didnt have the knowledge of today. like...what

[valid jetty]

i mean yeah but in this society..

[deep mulch]

I force sand to think

[leaden crater]

i want to make my own sword

[pine prawn]

Humans are pretty darn intelligent

[pine prawn]

Ooo, forgery!

Just dont make a katana

[deep mulch]

how

[leaden crater]

machete

[pine prawn]

Good choice

[leaden crater]

but probs katana too, i know the process from veritasium

[pine prawn]

Thats fair

[leaden crater]

and the documentation is public

[pine prawn]

If you make a katana make sure to give it the edgiest name you can think of

If it can make you cringe after hearing one third of the name, then its good

[leaden crater]

mysword0

[pine prawn]

Programming reference, very fun!

Should have done mysword0.dat

[leaden crater]

https://www.youtube.com/watch?v=Tt6WQYtefXA

[deep mulch]

@leaden crater I eat sand

[leaden crater]

sandman

[pine prawn]

Crunchy

[leaden crater]

sword

[pine prawn]

Mercury-lead alloy :D

[leaden crater]

https://www.youtube.com/watch?v=8hnu5DspBso
https://www.youtube.com/watch?v=eukDccnTjJ8
https://www.youtube.com/watch?v=_zhdsLsEdyY

yup

best metal

i really want to show people what i watch on youtube but most of them never actually care for my videos

despite putting a lot of effort in searching in history for it

[pine prawn]

I mean.. u literally watch what i watch

[leaden crater]

really?

[pine prawn]

Seems like it

Labcoatzz?

[leaden crater]

didnt know abt that chanel

[pine prawn]

Plasma channel?

[leaden crater]

plasma?

[pine prawn]

You should check em out!

[leaden crater]

im insane

[pine prawn]

Oooo, drone shots?

[leaden crater]

no

cabview

[pine prawn]

Sounds relaxing

Might check that out

[leaden crater]

i also watched demo ranch until a few days ago

[pine prawn]

Oh i remember watching that like.. 2 years ago?

Heres a less serious and not irl suggestion: GrayStillPlays

Also have you watched nilered?

[leaden crater]

already watched all his universe sandbox 2 + old gta 5 videos

[pine prawn]

He is awesome

Though im gonna sleep now

Gn!

[leaden crater]

theres also integza

gn

[pine prawn]

Ooo yeah!

[valid jetty]

@hoary sluice @deep mulch is it a good idea to get rid of this feature

use std/io;

fn deref(i32 *`ptr`) @manual -> i32 {
    `%res =w loadsw %ptr`;
    `ret %res`;
}

fn add_one(i8 `val`) @manual -> i8 {
    `%res =w add %val, 1`;
    `ret %res`;
}

fn `identity`(i32 `val`) @manual -> i32 {
    `ret %val`;
}

fn main() {
    let a = 123;

    $assert(deref(&a) == 123, nil);
    $assert(add_one(a) == 124, nil);
    $assert(identity(a) == 123, nil);
    $assert(`identity`(a) == 123, nil);

    // The function name can be represented without exact literals too in this case
    $assert(`identity`(a) == identity(a), nil);

    io::println("All `exact literal` tests have passed!".color("green").reset());
}

it would simplify the compiler quite a lot to get rid of the whole inline IR thing, there are a lot of compiler hacks to bypass it and make this work

but it makes the language more interesting

its not even very useful anymore for anything

[deep mulch]

yes

looks very cursed

[shrewd canopy]

This might be useful for tests
Like in Kotlin you can do:

@Test
fun `it works`() {
  assertEquals(sum(42, 42), 84)
}

[valid jetty]

why would it be useful

[jade stone]

What do the backticks do

[valid jetty]

they define an "exact literal"

when its in the case of a top level statement, it acts like an IR instruction

when in the case of a function decl or call, it acts like a raw string

[jade stone]

Oh, is that not source code

[valid jetty]

so you can do yeah this

[valid jetty]

it is

An exact literal is Elle's way of implementing inline IR into the language. This basically means that you can write intermediate language code directly in Elle which compiles without any type, size, scope, or name context.

You can create an "exact literal" by wrapping the inline IR with "`" on both sides of the expression, and ensuring you include a semicolon at the end.

You can also use the manual return directive, which states that Elle should NOT include an automatic return if the function does not return anything by default. You can do this by adding the @manual attribute to your function.
@jade stone

[jade stone]

Why not just use a function called it with a lambda argument

[deep mulch]

what does that mean

[jade stone]

it ("does a thing") {
doThing();
}

[deep mulch]

hate

[jade stone]

@deep mulch rate horrorcode

nmd: if (importUses?.uses.length === 1) {
    const loc = importUses.uses[0].location;
    const call = findParrent(loc, isCallExpression);

    if (!call || call.arguments.length !== 1 || call.arguments[0] !== loc)
        break nmd;

    // ensure the call is `n.n(...)`
    const funcExpr = call.expression;

    // ensure something like `foo.bar`
    if (!isPropertyAccessExpression(funcExpr)
        || !isIdentifier(funcExpr.name)
        || !isIdentifier(funcExpr.expression))
        break nmd;
    // ensure the first part is wreq
    if (!this.isUseOf(funcExpr.expression, this.wreq)
        || funcExpr.name.text !== "n")
        break nmd;

    const decl = findParrent(funcExpr, isVariableDeclaration)?.name;

    if (!decl || !isIdentifier(decl))
        break nmd;

    this.vars.get(decl)
        ?.uses
        ?.map((x) => x.location.parent)
        .filter(isCallExpression)
        .map((calledUse): Range[] | undefined => {
            if (exportName === WebpackAstParser.SYM_CJS_DEFAULT) {
                // TODO: handle default exports other than just functions
                return isCallExpression(calledUse.parent)
                    ? [this.makeRangeFromAstNode(calledUse)]
                    : undefined;
            } else if (typeof exportName === "string") {
                const expr = findParrent(calledUse, isPropertyAccessExpression);

                if (!(!!expr && expr.expression === calledUse && expr.name.text === exportName))
                    return undefined;

                return [this.makeRangeFromAstNode(expr.name)];
            }
            throw new Error("Invalid exportName");
        })
        .filter((x) => x !== undefined)
        .forEach((use) => {
            const final = use.at(-1);

            if (!final)
                throw new Error("Final is undefined, this should have been filtered out by the previous line as there should be no empty arrays");

            uses.push(final);
        });
}

[valid jetty]

you can break if statements ?????

[jade stone]

yop

[deep mulch]

yea

[valid jetty]

glorified goto

[deep mulch]

rosie insane

[jade stone]

yeah

i use a labled block in my actual code, just moved the label to the if for indentation

you can label any statment iirc

https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Statements/label

[valid jetty]

wait ANY statement??? so i can just have like IIFE behavior in my functions

like the reason i want

(() => {
    const x = 1;
    if (!x) return;
})()
``` is so i can break out with return

does this mean i can do

foo: {
    const x = 1;
    if (!x) break foo;
}

[jade stone]

that is valid

[valid jetty]

wtf

best feature

[jade stone]

thats why im doing it

this logic need to have early returns

and its only used once

so i dont want to extract it to its own function

[valid jetty]

yeah i would wrap that in an IIFE

thats amazing

[jade stone]

you could label the if statment itself, but that just looks a lot worse and wouldnt keep locals in their own scope (not like there are any locals in that scope anyway)

[royal nymph]

do {
  const x = 1;
  if (!x) break;
} while (false);

labels overrated

[winged mantle]

why use goto when you can use cometo

(copy and paste the code from where you wanted to goto to where you called it)

[full spire]

Encoded SSTV audio with Martin 1 mode

Error

Unfortunately an error occurred during command processing

Error

Unfortunately an error occurred during command processing

[deep mulch]

bruh

[full spire]

Encoded SSTV audio with Martin 1 mode

[full spire]

Decoded SSTV image

[jade stone]

@deep mulch me when my horrorcode works first try

[deep mulch]

guhh

[jade stone]

oops

wrong link

https://github.com/sadan4/VencordCompanion/commit/6189bd9360ae0cfed7537dbe12698d5ad28c05e0#diff-0e63e0f9f3d1d4f963e63b856e3cd983f42d49afb8e55b5ece01c8d95bd6bc20

[deep mulch]

@jade stone why is sstv so poorly documented

[jade stone]

whats that

[deep mulch]

https://en.wikipedia.org/wiki/Slow-scan_television

[jade stone]

insane

[deep mulch]

watch this

[full spire]

Error

Unfortunately an error occurred during command processing

[deep mulch]

why didnt it copy

what

i found a vencord bug

[jade stone]

[deep mulch]

copy link does not copy

[jade stone]

oh, yeah i ran into that yesterday

[full spire]

Encoded SSTV audio with Martin 1 mode

Decoded SSTV image

[jade stone]

crunchy

[deep mulch]

guhhh somethings wrong with the decoder

its niot supposed to be that crunchy

[jade stone]

@deep mulch are you on vesktop, web or stable

[deep mulch]

vesktop

see

[jade stone]

stream of images

[deep mulch]

yes

[uneven gate]

I've been making a plugin with a setInterval but it breaks after my computer goes to sleep

doing some research I could fix this by using powerMonitor

import { powerMonitor } from "electron";
...
start() {
    createInterval();
    powerMonitor.addListener('suspend', () => clearInterval(interval));
    powerMonitor.addListener('resume', createInterval);
}

It yells at me saying

Cannot import electron in browser code. You need to use a native.ts file

Is there any docs for how to use a native.ts file in vencord? I tried looking at how other plugins do it but I'm new here and cant read them

[dense sand]

wow

[jade stone]

is that jvm native code

[dense sand]

its my own jvm native code

[jade stone]

insane

[dense sand]

i honestly didnt know how else id write it lol

well i actually got an idea

im thinking of opensourcing it so people can laught at my trash C code

[uneven gate]

why can't you just do something like this? am I stupid?

NATIVE(java_lang_Class_getPrimitiveClass) {
  char *name = ToString(vm, (ObjectRegion*)ToAddress((struct Variable*)argv[0]));

  char *primitiveName = NULL

  if (strcmp(name, "int") == 0) {
    primitiveName = "java/lang/Integer";
  }
  if (strcmp(name, "long") == 0) {
    primitiveName = "java/lang/Long";
  }
  if (strcmp(name, "short") == 0) {
    primitiveName = "java/lang/Short";
  }
  if (strcmp(name, "boolean") == 0) {
    primitiveName = "java/lang/Boolean";
  }
  if (strcmp(name, "byte") == 0) {
    primitiveName = "java/lang/Byte";
  }
  if (strcmp(name, "char") == 0) {
    primitiveName = "java/lang/Character";
  }
  if (strcmp(name, "float") == 0) {
    primitiveName = "java/lang/Float";
  }
  if (strcmp(name, "double") == 0) {
    primitiveName = "java/lang/Double";
  }
  if (strcmp(name, "void") == 0) {
    primitiveName = "java/lang/Void";
  }
  
  free(name);
  return CreateVariable(STACK_ELEMENT_IS_ADDR | STACK_ELEMENT_LONG, primitiveName);
}

[dense sand]

no you are not stupid i realized it too and rewrote it already lmao

[leaden crater]

why so many strcmp..

[uneven gate]

Can I see the rewrite?

[dense sand]

its literally the same as yours...

[leaden crater]

hell descends on earth with strcmp

[uneven gate]

@dense sand why are you writing a jvm in c?

[dense sand]

i was bored

[uneven gate]

real

[dense sand]

it was kinda my dream always

[uneven gate]

.. to make a jvm?

[dense sand]

yes lol

i want to make it wasm compatible so i can run this on web

[uneven gate]

you should do the BEAM next (30 years)

[jade stone]

what else would you run it on

[uneven gate]

java

[dense sand]

jvm in java

peak

[uneven gate]

If it hasn't been done yet I'd be surprised actually

[leaden crater]

why not just

NATIVE(java_lang_Class_getPrimitiveClass) {
  char *name = ToString(vm, (ObjectRegion*)ToAddress((struct Variable*)argv[0]));
  
  char *primitiveName = 
    (strcmp(name, "int") == 0) ? "java/lang/Integer" :
    (strcmp(name, "long") == 0) ? "java/lang/Long" :
    (strcmp(name, "short") == 0) ? "java/lang/Short" :
    (strcmp(name, "boolean") == 0) ? "java/lang/Boolean" :
    (strcmp(name, "byte") == 0) ? "java/lang/Byte" :
    (strcmp(name, "char") == 0) ? "java/lang/Character" :
    (strcmp(name, "float") == 0) ? "java/lang/Float" :
    (strcmp(name, "double") == 0) ? "java/lang/Double" :
    (strcmp(name, "void") == 0) ? "java/lang/Void" : NULL;
  
  free(name);
  return CreateVariable(STACK_ELEMENT_IS_ADDR | STACK_ELEMENT_LONG, primitiveName);
}
``` or something, this should work no?

[uneven gate]

I HATE TERNARIES

[leaden crater]

they're pretty good

[jade stone]

this is ugly

[uneven gate]

Ternaries should only ever be 1 line imo

[leaden crater]

nop

[dense sand]

Teavm

[leaden crater]

do you want an implementation with goto?

[uneven gate]

any more and its a disgrace

[dense sand]

Ugh mb i thougjt i was replying to a different thing

Teavm is jvm compatible for web

[uneven gate]

why can't switches just be expressions

[jade stone]

i swear im going to have a 2000 line file just for parsing webpack ASTs by the time im done with my vencord extension

[leaden crater]

because you're using too many lines for something that can be written shorter using ternary and also looks pretty

or you can just write a switch case anyways

[uneven gate]

I think multiline/nested ternaries are a mistake
react kinda solidified this viewpoint for me

[uneven gate]

I far prefer v-if from vue over what react does with ternaries

[jade stone]

yea, id rather have an IIFE / labled statment over nested ternaries

[leaden crater]

insane

[uneven gate]

I hate how many times I've done IIFEs in js for random stuff like this.

[leaden crater]

1 billion line ternary operators for every number

[jade stone]

you should see discords code, they have some pretty cursed IIFEs

[leaden crater]

is discord open source even

[jade stone]

no, im talking about the minified code that the clients get

[leaden crater]

otherwise it's just weird binary translated

[uneven gate]

I have delved into discord's code to find some FluxDispatcher things and did not like the minified react experience

[jade stone]

minified react gets very easy to read once you have a bit of experience

but the most annoying thing for me was not being able to see where things were in the source

so i made go to definition and list references for the minified code

[uneven gate]

let primitiveName: Option<&str> = match name {
    "int" => Some("java/lang/Integer"),
    "long" => Some("java/lang/Long"),
    "short" => Some("java/lang/Short"),
    "boolean" => Some("java/lang/Boolean"),
    "byte" => Some("java/lang/Byte"),
    "char" => Some("java/lang/Character"),
    "float" => Some("java/lang/Float"),
    "double" => Some("java/lang/Double"),
    "void" => Some("java/lang/Void"),
    _ => None,
};

Actually nvm it doesn't look any better

[hoary sluice]

how will it make it simpler

dont you parse the entire function body as ir

[hoary sluice]

dont need it anymore hopefully will never need it again

[hoary sluice]

wait until you hear about 2nd order digital filters (theyre awesome )

rosie whats the convolved impulse respones of two identical RC filters with the impulse response 1/τ * e^(-t/τ)

[deep mulch]

@leaden crater hii

[leaden crater]

@deep mulch hiii

[deep mulch]

@jade stone boo

[jade stone]

scary

[deep mulch]

should calling init more than once just silently skip or throw error

[leaden crater]

@deep mulch

[deep mulch]

nini

[leaden crater]

do you speak java or c

[deep mulch]

yes

[leaden crater]

both?

[deep mulch]

yws

[dense sand]

does anyone know how can i intellij tab this properly so the values are prettily under each other

[lavish frigate]

😭

[dense sand]

i srsly dont know 😭 please help

[leaden crater]

why do you... define

[dense sand]

Yes

I define

[leaden crater]

why

[hoary sluice]

warning

actually probably depends on the system

[dense sand]

What else should i do

Its for my parsing library

[leaden crater]

can't you just use the numbers

[fleet cedar]

You should use consts

[dense sand]

Its 250 instructions

[fleet cedar]

It's 250 words that are now 100% synonymous with those numbers in every context

Namespacing please

[dense sand]

I use raw C

No namespaces

[fleet cedar]

Ew

Why

[dense sand]

I like C

[leaden crater]

crazy

[ornate quiver]

if you're using raw C then at least prefix the opcode swith OP_

typically raw C means you prefix stuff to names

[valid jetty]

no

you can insert IR instructions anywhere in a function

@manual just tells the compiler not to insert a manual return statement if the function returns before via IR (as that’s not tracked by the compiler)

there are at least 6 hacks throughout the whole project to allow that to work

[formal belfry]

a

[valid jetty]

ok ive restricted the dialect of what you can say with exact literals

you can no longer do inline IR with them

but you can still do this

use std/io;

fn `add.works`() {
    $assert(42 + 42 == 84, nil);
}

fn `mul.works`() {
    $assert(42 * 2 == 84, nil);
}

fn main() {
    `add.works`();
    `mul.works`();

    io::println("All `exact literal` tests have passed!".color("green").reset());
}

[fleet cedar]

My decompiler also has that exact syntax for raw identifiers

It's used in exactly one place in the decompiled codebase

A call to debug.`TestScenarioflagSet( GF1_EV_02_32_03_END )`()

[valid jetty]

lmao insane

i wonder, how do you handle spaces in something like this

[fleet cedar]

I don't see why spaces would be different from any other letter

[valid jetty]

because my raw identifiers just put the characters exactly as they are into the IR

[fleet cedar]

Maybe if the ir itself gets confused, that'd cause trouble

[valid jetty]

which, as expected, causes an IR error if theres spaces

[fleet cedar]

Guess that's fair

That's what mangling is for isn't it

[valid jetty]

but i dont wanna transform the raw identifier because then its no longer a raw identifier so idk

[valid jetty]

i tried to specifically refrain from mangling so elle code is compatible with the C abi

[fleet cedar]

Only mangle things that need mangling?

[valid jetty]

i mean i guess so hm

[fleet cedar]

I need to make my lexer not allow rawing keywords though

[leaden crater]

what does ` do

[dense sand]

Why such a mess? I got 3/(cos²(3x-π/4))

[leaden crater]

try using wolfram alpha

even if it doesn't say steps

[hoary sluice]

cause u can probably multiply out the ^2 and simplify it

[pine prawn]

Is that?

What language is that..?

[valid jetty]

its a language im working on called elle

@deep mulch I FINALLY ADDED ELSE IF 😭

after all this time

this

use std/libc/io;

fn main() {
    x := 4;

    if x == 1 {
        io::cprintf("hi x = 1\n");
    } else if x == 2 {
        io::cprintf("no way x = 2\n");
    } else if x == 3 {
        io::cprintf("omg x = 3\n");
    } else {
        io::cprintf("whaa x = %d\n", x);
    }
}
``` when compiled with `--noalloc --nostd --nofmt --nosm` compiles into
```ts
type :ElleEnv = { l, l, l }
type :ElleMeta = { l, l, w, l, l, w, w }
data $.54 = { b "hi x = 1\n", b 0 }
data $.57 = { b "no way x = 2\n", b 0 }
data $.60 = { b "omg x = 3\n", b 0 }
data $.62 = { b "whaa x = %d\n", b 0 }
export function w $main() {
@start
    %x.addr.51 =l alloc8 4
    storew 4, %x.addr.51
    %x.50 =w loadw %x.addr.51
    %tmp.53 =w ceqw %x.50, 1
    jnz %tmp.53, @ift.53, @iff.52
@ift.53
    %tmp.55 =w call $printf(l $.54, ...)
    jmp @end.52
@iff.52
    %x.50 =w loadw %x.addr.51
    %tmp.56 =w ceqw %x.50, 2
    jnz %tmp.56, @elift.55.0, @eliff.55.0
@elift.55.0
    %tmp.58 =w call $printf(l $.57, ...)
    jmp @end.52
@eliff.55.0
    %x.50 =w loadw %x.addr.51
    %tmp.59 =w ceqw %x.50, 3
    jnz %tmp.59, @elift.58.1, @eliff.58.1
@elift.58.1
    %tmp.61 =w call $printf(l $.60, ...)
    jmp @end.52
@eliff.58.1
    %x.50 =w loadw %x.addr.51
    %tmp.63 =w call $printf(l $.62, ..., w %x.50)
    jmp @end.52
@end.52
    ret 0
}

[leaden crater]

me

[valid jetty]

close enough

[leaden crater]

:3

[fleet cedar]

Isn't else if kinda trivial to add

[valid jetty]

well i mean it was sorta easy after messing with it for like 20 mins

its easy on paper but implementing it is kinda annoying

[fleet cedar]

Nah, just need to add one more clause in the parser after else

[valid jetty]

my parser isnt a traditional recursive descent parser

fn parse_if_statement(&mut self) -> AstNode {
    self.advance();

    let tokens = self.yield_tokens_with_delimiters(vec![TokenKind::LeftCurlyBrace]);
    let expression = Statement::new(tokens, 0, &self.body, self.shared).parse().0;

    self.expect_tokens(vec![TokenKind::LeftCurlyBrace]);
    self.advance();

    let body = self.yield_block(false);

    let mut elifs: Vec<(Box<AstNode>, Vec<AstNode>)> = vec![];
    let mut else_body: Vec<AstNode> = vec![];

    loop {
        if self.current_token().kind == TokenKind::Else {
            self.advance();

            if self.current_token().kind == TokenKind::If {
                self.advance();

                let tokens = self.yield_tokens_with_delimiters(vec![TokenKind::LeftCurlyBrace]);
                let elif_condition =
                    Statement::new(tokens, 0, &self.body, self.shared).parse().0;

                self.expect_tokens(vec![TokenKind::LeftCurlyBrace]);
                self.advance();

                let elif_body = self.yield_block(false);
                elifs.push((Box::new(elif_condition), elif_body));
            } else {
                self.expect_tokens(vec![TokenKind::LeftCurlyBrace]);
                self.advance();

                else_body = self.yield_block(false);
                break;
            }
        } else {
            break;
        }
    }

    self.position -= 1;

    AstNode::IfStatement(IfStatement {
        condition: Box::new(expression),
        body,
        elifs,
        else_body,
        location: self.current_token().location,
    })
}
``` and also the parser isnt the hard part its the codegen thats hard

making everything jump to the next thing properly was annoying

[fleet cedar]

How

It's literally just an if-else with another if in the else branch

And that looks like a pretty standard recdesc to me anyway

[valid jetty]

the operator precedence parsing is completely different

[valid jetty]

take a look

its not hard but its easy to get confused in my head

[fleet cedar]

I mean sure if you want to represent it as ```rs
struct IfStmt {
expr: Expr,
then: Block,
elseifs: Vec<(Expr, Block)>,
else_: Option<Block>
}

[valid jetty]

its ```rs
#[derive(Debug, Clone, Eq, PartialEq)]
pub struct IfStatement {
pub condition: Box<AstNode>,
pub body: Vec<AstNode>,
pub elifs: Vec<(Box<AstNode>, Vec<AstNode>)>,
pub else_body: Vec<AstNode>,
pub location: Rc<Location>,
}

how else can you represent it

i thought i did an interesting thing by making the else_body not an Option<T>, because then having an empty else {} will still make the body be empty and so the compiler will skip the else branch altogether

[fleet cedar]

I'd just go with a ```rs
struct IfStatement {
expr: Expr,
then: Block,
else_: Else,
}
enum Else {
None,
ElseIf(IfStatement),
Else(Block),
}

Or just have it always be a Block, which has 0, 1, or N elements depending on style

[valid jetty]

hmmm actually a good idea

i might do that thank you

but it would be a Vec<Else> surely

[fleet cedar]

Why?

The inner IfStatement has its own else

[valid jetty]

oh its recursive i didnt notice that lol

this really looks like a parser representation of

if true {
    ...
} else {
    if true {
        ...
    }
}
``` lmao

beautiful

[fleet cedar]

Yep

[valid jetty]

for the longest time thats basically what i was doing anyway

but its more of a syntax hack than a language feature

fn char::unicode_length(char self) {
    if (self & 0b10000000) == 0 {
        return 1;
    } else { if (self & 0b11100000) == 0b11000000 {
        return 2;
    } else { if (self & 0b11110000) == 0b11100000 {
        return 3;
    } else { if (self & 0b11111000) == 0b11110000 {
        return 4;
    }}}}
}

[fleet cedar]

Sure you'll end up with multiple consecutive labels at the end, but if that causes issues for your backend then said backend sucks

[valid jetty]

nope i dont think it would

[dense sand]

Binary integers arent a C thing until C23, thats wild

[valid jetty]

is this a sane way to test this???

fn main() {
    `if.works`();
    `if.else.works`();

    `if.else_if.picks.if`();
    `if.else_if.picks.else_if`();

    `if.else_if.else.picks.if`();
    `if.else_if.else.picks.else_if`();
    `if.else_if.else.picks.else`();

    `if.else_if.1.else_if.2.picks.if`();
    `if.else_if.1.else_if.2.picks.else_if.1`();
    `if.else_if.1.else_if.2.picks.else_if.2`();

    `if.else_if.1.else_if.2.else.picks.if`();
    `if.else_if.1.else_if.2.else.picks.else_if.1`();
    `if.else_if.1.else_if.2.else.picks.else_if.2`();
    `if.else_if.1.else_if.2.else.picks.else`();

    $println("All `if statement` tests have passed!".color("green").reset());
}

im glad i even wrote a test for this because it turns out i was handling the label wrong when theres a bunch of else-ifs but no else

[dense sand]

how beatiful
https://github.com/KoblizekXD/classparse/blob/cf1aaa505004e00c82febae8d6707d38ded70652/src/utils.c#L189

[pearl stag]

utils.c: Line 189

switch (opcode) {

[fleet cedar]

Such is the life of a c

[dense sand]

i mean what else could i do in C++