🪅-progaming

[ornate quiver]

not on window either
all the references are enclosed

main webpack bundle is loaded through the main property of package.json in an asar package

[hoary sluice]

bro homerow mods are annoying sometimes, i tried pasting the aoc answer but to do that i have to hold N and press V (since N is ctrl) and i did it too fast and submitted 'nv' as my answer

aoc needs to not put u on cooldown if u submit an answer with a letter in it when its supposed to be ints

[ornate quiver]

what I did is rename the asar bundle and made a small loader to run code before loading the actual asar
kinda like vencord used to do it

[royal nymph]

still does it that way

can you send the whole file that is in

also why is it not minified lmao

gitkraken insane

is gitkraken foss

[ornate quiver]

it was
I formatted

[royal nymph]

you also renamed __webpack_module_cache__?

[ornate quiver]

nop that part wasnt

funny

it's like half minified

[dense sand]

Guys what do you think would be faster:

1. parse javadoc data out of the html pages that are generated
2. parse source java files and extract the javadoc comments

Faster in terms of actual speed

[ornate quiver]

webpack loader at the end

[royal nymph]

13mb

horror

[ornate quiver]

yeah

[royal nymph]

bro dropped entire gitkraken code

[calm ruin]

14MB

okay so

webpack_Require is not on global right

[ornate quiver]

no

[calm ruin]

also is it gitkraken desktop

[ornate quiver]

yes

[calm ruin]

okay wait

[ornate quiver]

I couldn't find anything leaking to global

[calm ruin]

300mn

[royal nymph]

I'll take a look soon

not rn

[calm ruin]

can you open devtools

[ornate quiver]

not by default

[calm ruin]

how do I enbale

[ornate quiver]

you can open by replacing forceDevtools: and make it true

in main bundle

[ornate quiver]

i think I might've done it in this one here

[royal nymph]

you just need to register a global keybind with electron

and then call browserWindow.openDevTools()

or well you would want to use the toggle function

[ornate quiver]

im not sure what they're doing
but it doesn't matter anyways because main bundle is loaded in main process
renderer loads a different bundle so devtools are useless

i need to patch main bundle

[calm ruin]

why th do they have 3 app asar 😭

[ornate quiver]

it's so cursed

[hoary sluice]

@calm ruin u would love 2017 aoc

[calm ruin]

nop not now

[calm ruin]

bfable

[hoary sluice]

runs in 2 seconds on kotlin

36ms optimized rust

[royal nymph]

you can still open devtools

https://www.electronjs.org/docs/latest/tutorial/debugging-main-process

open chrome://inspect in a chromium browser

[ornate quiver]

o

[royal nymph]

and launch gitkraken with --inspect

then you can connect to it from the devtools in the browser

if you want to debug before the app starts, you can use --inspect-brk instead which will immediately break

[ornate quiver]

neat
I will try

[calm ruin]

yeah that worked

I was gonna unpack default_app.asar modify and repack

but this is easier

[royal nymph]

I mean that's also very easy

[calm ruin]

but it only shows node

no browser

[royal nymph]

yes that's the point

rusher needs to debug main process

[calm ruin]

oh

[ornate quiver]

default app is ignored

that's electron default

[calm ruin]

wth does it even use then

[royal nymph]

if you want to debug renderer / preload, you can just do this

[calm ruin]

its file structure so confusing

[ornate quiver]

app.asar

[calm ruin]

app.asar only had unpacked stuff

[ornate quiver]

there's a 7zip plugin for asar
it makes editing so much easier

[ornate quiver]

no it's packed

[dense sand]

That sounds like some arab name

[calm ruin]

I mean it had the stuff thats already in app.asar.unpacked

[ornate quiver]

correct

[ionic lake]

do you still want

[dense sand]

Sure

[royal nymph]

eh

its so easy

[ornate quiver]

.unpacked is literally just for git

[ionic lake]

lmao no

[dense sand]

I don't want plain black bg

[royal nymph]

pnpm add --global @electron/asar

# unpack
asar e app.asar app

# then edit files in app folder

# pack
asar p app app.asar

[dense sand]

And i dont want to fuck with shaders

[ionic lake]

i guess you want something like this https://bg.ibelick.com

[dense sand]

Yeesss!!! Exactly!!!!!

[royal nymph]

this is incredibly easy to do

[dense sand]

Thankssss

[royal nymph]

you can pollute global prototypes to intercept this assignment

https://github.com/Vendicated/WebpackGrabber

this does exactly what u need

[ionic lake]

you're welcome lmk if you need anything

[calm ruin]

@ornate quiver I managed to get it

tho that amd0 might change

but you can probably do same thing with c

[royal nymph]

ur so slow

i already sent

[calm ruin]

BAD

I WAS TRYING TO MAKE STUPID DEBUGGER WORK

export function waitForDeclaration(obj, prop, callback: (obj,prop)=>{}) {
    Object.defineProperty(obj, prop, {
        set(v) {
            Object.defineProperty(obj, prop, {
                value: v,
                writable: false,
                configurable: false,
                ...Object.getOwnPropertyDescriptors(v),
            });

            callback(obj, prop);
        },
        configurable: true,
    });
}
``` @royal nymph what do you think about my hlorious code

[royal nymph]

https://github.com/Vendicated/Vencord/blob/main/src/shared/onceDefined.ts#L30-L47

[pearl stag]

onceDefined.ts: Lines 30-47

export function onceDefined<T extends object, P extends LiteralUnion<keyof T, PropertyKey>>(
    target: T, property: P, callback: (v: P extends keyof T ? T[P] : any) => void
): void {
    const propertyAsAny = property as any;

    if (property in target)
        return void callback(target[propertyAsAny]);

    Object.defineProperty(target, property, {
        set(v) {
            delete target[propertyAsAny];
            target[propertyAsAny] = v;
            callback(v);
        },
        configurable: true,
        enumerable: false
    });
}

[calm ruin]

STEALER

[royal nymph]

trol

[calm ruin]

well we wrote same code lmao

except you have insane typescript definitions

[hoary sluice]

disgusting

[hoary sluice]

this is not a real programming language

[calm ruin]

well it lets you do cursed stuff

[fleet cedar]

Why does it take a callback rather than returning a promise

[calm ruin]

I at least like its that part

async bad

also cant promise also introduce some delay

this needs to call instantly

while blocking main thread

[fleet cedar]

Hm, that's a fair point

Resolving a promise only schedules the then to be run at the runtime's convenience

[calm ruin]

yeah thats bad

[ornate quiver]

oh neat

alright i will try

[calm ruin]

why did you ignore me

😢

[ornate quiver]

i saw yours too

idk what to use yet

https://github.com/Vendicated/WebpackGrabber/blob/bfefd7b907dc9c95c827279a048e73562e9bb3bc/WebpackGrabber.user.js#L15-L33

[pearl stag]

WebpackGrabber.user.js: Lines 15-33

function extractPrivateCache(wreq) {
    let cache = null;
    const sym = Symbol("wpgrabber.extract");

    Object.defineProperty(Object.prototype, sym, {
        get() {
            cache = this;
            return { exports: {} };
        },
        set() { },
        configurable: true,
    })

    wreq(sym);
    delete Object.prototype[sym];
    if (cache) delete cache[sym];

    return cache;
}

[ornate quiver]

how does this work

[fleet cedar]

Just evil magic, nothing special

Define a magic symbol that does funky things when looked up, then trick webpack into running that lookup function

Thereby getting the object it's looked up on

[ornate quiver]

oh neat

[royal nymph]

you dont need that

you only need it for when the cache isnt exposed at all

it's usually available on wreq.c as is the case with gitkraken

but very rarely it isn't

[ornate quiver]

oh

[royal nymph]

it's to get a reference to the cache if it's only a local variable

[ornate quiver]

can i make webpack cache a proxy then or something
to intercept new modules being added

[royal nymph]

what exactly are u trying to do

same as vencord?

regex patches?

or do you want to monkey patch

[ornate quiver]

nop
i just need to monkeypatch the result of two functions from different modules

find by props

[calm ruin]

https://github.com/Vendicated/JsPosed use

[royal nymph]

WebpackGrabber has everything

[royal nymph]

lmao u dont need anything like that even

if u just want to patch 2 functions

[calm ruin]

WRONG

prototype definitions so bad

[ornate quiver]

i dont see anything that lets me intercept new modules as they are added

[royal nymph]

why would u need to do that

if u just wanna monkeypatch

[calm ruin]

waitForModule doesnt exist on webpackgrabber yes

[ornate quiver]

because if i do it too late then it doesnt matter

[royal nymph]

oh okay

[ornate quiver]

wont work

[royal nymph]

well

.c has the cache which is the finished exports

.m has the module source

BRO

fuck this bot lmao

[calm ruin]

[royal nymph]

anyway you need to patch .m

[calm ruin]

can you patch its .push method to capture new modules

[ornate quiver]

it doesnt seem that wreq.m exists

original module list is hidden

[royal nymph]

oh yes

it's just a local variable

[royal nymph]

you do need this then 🥳

very slight variation tho

[ornate quiver]

its not an array, but an object

[calm ruin]

then proxy 🙏

[ornate quiver]

this is so confusing

how

[royal nymph]

let me explain

this is the webpack require function

[calm ruin]

its all because you confuse rusher with dumb symbol

if you put skibi-i instead of symbol he wouldnt get confused

[royal nymph]

whenever we require q, it

1. obtains cache[q] to look up the module in the cache
2. if we found cached exports it just returns that
3. if not, it creates a new entry on cache[q]
4. then it calls modules[q] to initialise the module (require call) and puts the return value (the exports) in the cache

that function is meant to extract the private cache variable

so what it does is create a symbol (unique key)
then it declares a getter & setter for our symbol on the object prototype (so every object has this getter and setter)
then it calls require with our symbol
require looks up our symbol in the cache and thus invokes our getter
now our getter function has a reference to the cache

[calm ruin]

basically this lookup calls getter with "this" as cache

[royal nymph]

but you don't want the cache (wreq.c), you need the modules (wreq.m - not actually exported in this scenario)

so you need to slightly alter that function to your needs

[ornate quiver]

oh i see

[calm ruin]

I will soon steal all the vencord utility functions

[royal nymph]

function extractPrivateModules(wreq) {
    let modules = null;
    let cache = null;
    const sym = Symbol("wpgrabber.extract");

    let isSecondPass = false;
    Object.defineProperty(Object.prototype, sym, {
        get() {
            if (!isSecondPass) { // this is cache[sym] get
                cache = this;
                isSecondPass = true;
                return undefined; // return undefined so it skips cache
            }

            // this is modules[sym] get
            modules = this;
            return () => {} // modules are functions that it calls to get the exports. just return a noop function
        },
        set() { },
        configurable: true,
    })

    wreq(sym);
    delete Object.prototype[sym];
    if (cache) delete cache[sym];

    return modules;
}

this should work

[ornate quiver]

that's actually pretty neat

javascript is so cursed

[royal nymph]

essentially
modules is an object that holds each module (file)
cache is an object that caches the module exports. because if you import a module twice, it reuses the same variables

Source Code:

export let counter = 0;
export function increment() {
  counter++;
}

Output:

let modules = {
  1234567: (module, exports, require) => {
    exports.counter = 0;
    exports.increment = function() {
        exports.counter++;
    }
  }
}

then the first time you require 1234567, it will call the function and then put the exports object that it assigned counter and increment on to the cache object. in the future it will obtain it from the cache so it keeps reusing the same counter variable

each member of the modules object corresponds to one source file and all have the same function signature and assign their exports to the exports

it is called exactly one the first time you require it, and then never again

so what you want to do is find the member of the modules object that is the module you want to patch (based on source code, you need to stringify the function)

then you want to override that member with your own function. your own function should then call the original function and can alter the exports accordingly

[royal nymph]

you can see modules right at the top of ur file

81984 is the module id aka filename of the first module

re(7914) means require module 7914

then K.default = ... means exports.default = ...

[ornate quiver]

yeah

thanks

[royal nymph]

so this module requires module 7914 and all those other modules and defines a default export that has properties activityLogLevel and so on

[ornate quiver]

although

[ornate quiver]

isn't this slower than just making the cache a proxy and checking props
i only need to check by props, not by code

[royal nymph]

const theModuleYouWantToPatch = Object.entries(webpackModules).find(([id, module]) => module.toString().includes(".DEFAULT_COMMIT_MESSAGE_GENERATION_USER_PROMPT"));

if (!theModuleYouWantToPatch) throw "done exploder";

const [id, originalModule] = theModuleYouWantToPatch;
webpackModules[id] = function (module, exports, require) {
  originalModule(module, exports, require);
  // exports / module.exports are now initialised to whatever that module exports. you can access or modify it
  module.exports.whatever = 42;
}

[royal nymph]

except you cant do that

because the require function accesses it via local variable which you cant change

[ornate quiver]

oh right

oops

[royal nymph]

the best you could do is find the module id from the modules object and then define a getter/setter on the cache with that

but at that point it makes more sense to just monkey patch the module

[royal nymph]

exactly like this

i made a slight edit because i made a istake

[ornate quiver]

oki

[royal nymph]

sadly the part where you stringify the modules to find the right one seems to consume some memory permanently

because js caches the string

so if there is 10mb of code and you stringify everything

this is why vencord increases memory usage somewhat

[ornate quiver]

hm

[royal nymph]

anyway doesntr matter much

[ornate quiver]

what if i go through each module in __webpack_modules__ and monkeypatch it's call method

that way it doesnt use a ton of memory and i can also check by props immediately

[royal nymph]

you would just replace the function instead of patching call

anyway you can do that but its probably not worth it lol

its not like it will use hundreds of mb of ram

[ornate quiver]

the thing is i dont know which module it is until it's run

[royal nymph]

its not that much

[royal nymph]

yes i know

im saying

you wouldnt monkeypatch .call

that makes no sense

you'd just do

for (const [id, originalModule] in Object.entries(modules)) {
  modules[id] = function (module) {
    originalModule.apply(this, arguments);
    if (module.exports.whatever) // do thing;
  }
}

but its probably not worth it lol

you create so many new functions and closures

and its more cpu intensive probably

[calm ruin]

you know its bad when you see "310 remaining items load more" in a issue

[ornate quiver]

insane

[placid cape]

just don't use WSL

[calm ruin]

I will

I was trying to figure out why it was using 40gb

I couldnt figure out

Probably half of it is cuda

[serene elk]

except you can place a Proxy in the prototype but the getter/setter approach is better

[royal nymph]

horror

the performance of that is probably atrocious

[serene elk]

don't think so

it's just another proxy

[royal nymph]

yeah but inside the prototype

it's kinda different

the second u start messing with prototypes and do weird shit performance dips

all the prototype set methods have performance warnings xd

[serene elk]

lol

[royal nymph]

well it might be fine who knows

[serene elk]

either way it needs some tricks to get it to work fully

so not worth it

depending on how you do it, some traps are required to avoid a stack overflow

[royal nymph]

speaking of proxies it's kinda silly that proxies even have apply trap

I guess it's for completeness sake

but there's really no reason to ever need that

you can just wrap the function with ur own function instead to achieve the same without a proxy xd

[serene elk]

might be more peeformant somehow

[median root]

yo how do i get jemalloc to use page size 16? im tryna build the nuclear music player for my asahi linux installation but jemalloc always stops it with unsupported page size.

the only solution ive been able to find online was to use:

JEMALLOC_SYS_WITH_LG_PAGE

but that doesent seem to work for me

[jade stone]

is it just me, or is this a macro

[ornate quiver]

yeah

[past osprey]

should i leave this as a bug to fix later or call it a feature and never fix it?

one of the side effects of rewriting the entire collision code

[jade stone]

windows 7
look inside
kde

[viscid grove]

looks fine to me!
I love features!

[dense sand]

has anyone ever here worked with antlr4

[ornate quiver]

worst nightmare

[pearl stag]

disable-copying.js: Lines 9-25

function disableSelect() {
  return false;
}

function reEnable() {
  return true;
}

// If IE4+
document.onselectstart = new Function ("return false");
document.oncontextmenu = new Function ("return false");

// If NS6
if (window.sidebar) {
  document.onmousedown = disableSelect();
  document.onclick = reEnable();
}

[void leaf]

[placid cape]

new Function("return false")

💀

[void leaf]

Me when code is 13 years old

[placid cape]

well heh

[placid cape]

automatically promotes to double bcs C does

[still jolt]

the fuck is blom

[placid cape]

my lang

[still jolt]

oh okay fair enough

[hoary sluice]

@valid jetty @placid cape yall need to do https://adventofcode.com/2017/day/23

its awesome

i (and a lot of ppl in the solutions megathread) were really close to the answer but ||off by 1 because the loop does g = b - c, then b+=17 and then checks if g == c but thats before adding 17 so its an inclusive range and eric made the last number always prime 😭 ||

[placid cape]

52min delta time

[hoary sluice]

1 year left

[hoary sluice]

yep and about 30 mins on the spoiler

[placid cape]

ill do it after finishing 2015

[hoary sluice]

another issue i had was that my || is prime function was inverted|| but i fount that quickly

[placid cape]

hmmm i should fix it

to show it correctly

nwm its broken only on 2015

[hoary sluice]

if u finish it then its not broken

[placid cape]

true

[spark tiger]

no way ffmpeg is hosting c & asm courses now
https://x.com/ffmpeg/status/1874474647988933010

FFmpeg (@FFmpeg) on X

Make your New Year's Resolution to learn C and Assembly: https://t.co/Q63dLuCgFb

We are making low level languages great again in 2025!

Twitter · Jan 1, 2025, 3:15 PM

[dense sand]

[fleet cedar]

Ah, so they're running low on contributors

[dense sand]

I doubt they would want some newbies to mess up their code

[hoary sluice]

why is 2017 so easy

[spark tiger]

i mean code reviews exist

[dense sand]

Does anyone know about any tool which can parse out javadocs?

Like into a json

[fleet cedar]

Javadoc

[winged mantle]

regex

[dense sand]

No like into an actual json structure

[winged mantle]

why into a json structure

[dense sand]

Out of an existing source

[dense sand]

Or any format basically

[winged mantle]

isn't it a flat structure

https://github.com/chhorz/javadoc-parser

maybe this

[dense sand]

Something i can read in javascript

[winged mantle]

what are your needs

what are you trying to do

[dense sand]

Better javadocs

[winged mantle]

if accuracy doesn't matter that much you could just try to write your own parser

because it's unlikely there are any for js

[fleet cedar]

That's a good idea, javadoc frontend sucks

[winged mantle]

ah well ig if you're trying to replace that you'll need accuracy

[dense sand]

Yea i have frontend rn which can fetch any of the maven central artifacts

[dense sand]

Java-parser is good, but i have backend written in JS and idk how to use teavm to port java to js

[winged mantle]

uh

just write your backend in java

[dense sand]

Ig if nothing else remains

[winged mantle]

or look at java parser's source code and work out what they do

but it's probably not the simplest thing to parse

[dense sand]

Its a really fucking abstracted clusterfuck

[winged mantle]

average java project

[dense sand]

Yea

[winged mantle]

[dense sand]

The JavadocNode is implemented by like 12 interfaces

[winged mantle]

they're probably all the different types of javadoc structures

[hoary sluice]

@valid jetty how to speed up general longest path

[fleet cedar]

Make graph smaller

[hoary sluice]

2017 d24 p2 takes 5s

[hoary sluice]

cant

whats ur runtime for 2017 d24 @fleet cedar

[fleet cedar]

No idea, my repo only contains stuff from 19 onward

[hoary sluice]

i brought p1 from 5s to 300ms with a cache but p2 is uncacheable

[hoary sluice]

ok the rust guy did it in 275 microseconds so its clearly possible

[valid jetty]

one day i will upsolve

one day

[hoary sluice]

im too incompetent to implement the other optimizitions

[hoary sluice]

u should do this one today

[placid cape]

make your day 10 y2015 O(n)

[hoary sluice]

i solved the halting problem!!!

[fleet cedar]

This seems to be my scores from long ago

[hoary sluice]

i cant keep procrastinating intcode for much longer 😭

[fleet cedar]

Yeah I hated intcode

[placid cape]

function overloading 🔥

[hoary sluice]

2017 d25 is literally just parsing

i wouldve had ~#80 on lb

[placid cape]

oh xD

okay im gonna do aoc now

[hoary sluice]

this is the worst problem for me rn cause im doing these on my new keyboard

and on a new layout

so i dont have symbol muscle memory

49% consistency

[hoary sluice]

alright its time to not do any aoc for the next 9 months

(i dont wanna do intcode)

[valid jetty]

is this arity based only or also type based

[valid jetty]

if i wasn’t super busy with school work i didn’t bother doing and taking down my christmas decors i would lol

[placid cape]

Also type based

[fleet cedar]

Hype based

[placid cape]

but if you have more than one overload then for example this:

fun a(a: i64) -> i32 {...}
fun a(a: i64, b: i64) -> {...}

fun @public main() {
    return a(5);
}

Will not compile because 5 is i32 by default

but if you have only one overload then it'll implicitly cast it

so this works:

fun a(a: i64) -> i32 {...}

fun @public main() {
    return a(5);
}

I'll try to improve this later but for now it's enough

[hoary sluice]

why the @ in @public?

[placid cape]

Because it's an annotation, I'll probably make it as a keyword

[hoary sluice]

u can put annotations after the fun?

that seems wrong

[fleet cedar]

Kinda makes more sense than putting it before

Making it so function declarations always start with the fun keyword is nice

[spark tiger]

why does it show AoC++ on EACH year

[hoary sluice]

cause i bought it on EACH year

[fleet cedar]

Simp

[spark tiger]

ah. i thought it’s like one-time purchase

[hoary sluice]

it is

[placid cape]

I think you can also do this in java but not sure

[valid jetty]

so monomorphization but for arity and type

[placid cape]

yea

[valid jetty]

fun

[hoary sluice]

u do @annotation void whatever, before the type

in kotlin its before the fun keyword

[placid cape]

you can do it also after the type if I'm not mistaken

[valid jetty]

why not above the type like rust

#[macro_export]

[placid cape]

that's probably even better yeah

this was easier to parse

[valid jetty]

idk in elle i have it after the param list but before the return type

fn foo(i32 a) @volatile @fmt -> string

[placid cape]

eventually i want to make it above like in rust/java (java allows it almost everywhere)

[hoary sluice]

nope

[placid cape]

no?

Okay

[hoary sluice]

after the visibility mod but not after the type / fun

[valid jetty]

i made modifiers in any order because it’s easier to parse

pub external fn and external pub fn

[hoary sluice]

do u have annotations

[valid jetty]

yea

@alias(“foo”), @volatile, @fmt, @manual

[placid cape]

you're just setting booleans, don't you?

[valid jetty]

i call them “attributes” tho

[valid jetty]

yea

[hoary sluice]

can u do pub @volatile external @manual fn

[valid jetty]

no

[valid jetty]

.

[hoary sluice]

oh

[valid jetty]

attributes come before the return type but after param list

[hoary sluice]

meh i think it makes more sense to put them first

[valid jetty]

idk

this is what they do

[hoary sluice]

thats not what volatile means is it

[valid jetty]

it’s not

lol

volatile in C says to the compiler it shouldn’t optimize, reorder, or remove the variable/function through dead code elimination

but it works fine here i guess

[hoary sluice]

i just know that if a variable is modified in an interrupt service routine it should be made volatile

[valid jetty]

yeah it mostly says to the compiler “don’t make any assumptions about how to treat this variable”

which means the compiler can’t do instruction reordering and optimization as a result

[placid cape]

yea https://en.m.wikipedia.org/wiki/Volatile_(computer_programming)

[hoary sluice]

i thought it meant dont cache this anywhere

don't reorder volatile reads and writes

this is prob the important part

[hoary sluice]

@placid cape how do you parse >=

[placid cape]

it's in lexer same like @valid jetty has

I know it shouldn't be in lexer

[hoary sluice]

is that GreaterEqual or [Greater, Equal]

[placid cape]

GreaterEqual

[hoary sluice]

idk maybe thats easier

[dense sand]

does anyone here know how to use teavm

[dense sand]

@ionic lake you have worked with shadcn i suppose

[ionic lake]

sup

[ionic lake]

shoot your question(s)

[dense sand]

well theres this thing called sidebar

[ionic lake]

oh yeah that's new

[dense sand]

the thing is that i dont really know how to make it show under the top menu bar

it overlaps it

[ionic lake]

yeah that's ok

oh nevermind

how are you doing this, show your app

[dense sand]

currently its like this, where Navbar is my component

this is kinda what i want if you get me

[hoary sluice]

hop on hypixel disasters

[dense sand]

you mean hypixel skyblock?

i really didnt want to make my own sidebar but i guess ill have to

[hoary sluice]

nah hypixel disasters

[dense sand]

This is literally copy of a roblox game

[hoary sluice]

sad think is ik which one ur talking abt

[ionic lake]

ok what if you put your sidebar above nav

[dense sand]

like the whole sidebarprovider or what

cause siderbar has to be inside siderbarprovider

[ionic lake]

yeah move that entire block above

[valid jetty]

i fucking HATE c#

[dense sand]

nope

well ill try to tinker with the internal code

[valid jetty]

instead of just letting me do this

[Serializable]
class GameState {
    public Grid playerFleet = new();
    public Grid computerFleet = new();
    public Grid targetTracker = new();

    public void Clear() {
        playerFleet.Clear();
        computerFleet.Clear();
        targetTracker.Clear();
    }
}

class SaveManager {
    private const string SaveFileName = "gamestate.dat";

    public void Save(GameState state) {
        using (FileStream fs = new FileStream(SaveFileName, FileMode.Create)) {
            BinaryFormatter formatter = new();
            formatter.Serialize(fs, state);
        }
    }

    public GameState Load() {
        using (FileStream fs = new FileStream(SaveFileName, FileMode.Open)) {
            BinaryFormatter formatter = new();
            return (GameState)formatter.Deserialize(fs);
        }
    }
}
``` to save and load data

they deprecated BinaryFormatter

because its "unsafe"

so i literally cant use it without installing an external package

my other options are all convoluted bloated things like converting to json or using a BinaryWriter/Reader with reflection

when it would just be this simple

[ionic lake]

what if i show you some example code

[dense sand]

that would be cool

[ionic lake]

https://github.com/ewgenius/ui/blob/main/components/preview/dashboard.tsx

or, even if i hate to say, you can try shoving it to v0 and let it do it

look at the <aside> chunk in that code

[hoary sluice]

its intcode time

[spark tiger]

var stream = new MemoryStream();
var writer = new BinaryWriter(stream);

writer.Write(Something1);
writer.Write(Something2);
writer.Write(Something3);
writer.Write(Something4);
writer.Write(Something5);
writer.Write(Something6);

also i recently was wondering if there's an easy way to serialize an object in rust into a binary file

[fleet cedar]

Serde+bincode?

[spark tiger]

oh yeah sounds like it

i kinda wish it was simpler to parse binary file in rust though

like in c# it's just

var reader = new BinaryReader(stream);
Size = reader.ReadInt32();
FormatSize = reader.ReadInt32();

while in rust it's something insane

[fleet cedar]

It can be exactly that in rust too

Depends on how you define the apis

[spark tiger]

theo jumpscare

[fleet cedar]

But ReadInt32 is quite ambiguous - is it le or be?

[ionic lake]

that's not theo

[spark tiger]

i swear to god i've browser hundreds of forum posts, crates, and i haven't found anythnig as easy as c#'s impl

[spark tiger]

oh mb

i forgot his name

[ionic lake]

theprimeagen

[spark tiger]

yeah this

[fleet cedar]

It doesn't have to be harder than this https://github.com/Aureole-Suite/Factoria/blob/main/src/dirdat.rs#L109-L129

[hoary sluice]

val reader = DataInputStream(FileInputStream("file.bin"))
val size = reader.readInt()
val formatSize = reader.readInt()

[spark tiger]

are u64, u32 the built-in funcs?

[fleet cedar]

What do you mean built in

They're completely normal functions, defined on a completely normal struct

[spark tiger]

oh

[hoary sluice]

confusing theo and prime is crazy

[fleet cedar]

(Or rather, u32 is defined in the Le trait, and just calls u32le)

[spark tiger]

I'M SORRY

[hoary sluice]

you are proving his point

[spark tiger]

what's the difference though besides different method names

[hoary sluice]

thats kotlin

[spark tiger]

oh

ffs

[fleet cedar]

What point, that reading and writing bytes is easy in any language?

[hoary sluice]

you can tell because function namer arent PascasCase (worst decision ever)

[valid jetty]

i wrote this beauty

class SaveManager {
    public const string Name = "state.bin";

    public void Save(GameState state) {
        using (BinaryWriter writer = new(File.Open(Name, FileMode.Create))) {
            WriteObject(writer, state);
        }
    }

    public GameState Load() {
        using (BinaryReader reader = new(File.Open(Name, FileMode.Open))) {
            return (GameState)ReadObject(reader, typeof(GameState));
        }
    }

    private void WriteObject(BinaryWriter writer, object obj) {
        Type type = obj.GetType();

        foreach (var field in type.GetFields(BindingFlags.Public | BindingFlags.Instance)) {
            object value = field.GetValue(obj)!;
            if (value is Grid grid) WriteGrid(writer, grid);
        }
    }

    private object ReadObject(BinaryReader reader, Type type) {
        object obj = Activator.CreateInstance(type)!;

        foreach (var field in type.GetFields(BindingFlags.Public | BindingFlags.Instance)) {
            field.SetValue(obj, ReadGrid(reader));
        }

        return obj;
    }

    private void WriteGrid(BinaryWriter writer, Grid grid) {
        writer.Write((Byte)grid.kind);

        for (int row = 0; row < Config.GRID_SIZE; ++row)
            for (int col = 0; col < Config.GRID_SIZE; ++col)
                writer.Write(grid[row, col]);
    }

    private Grid ReadGrid(BinaryReader reader) {
        Grid grid = new((GridKind)reader.ReadByte());

        for (int row = 0; row < Config.GRID_SIZE; ++row)
            for (int col = 0; col < Config.GRID_SIZE; ++col)
                grid[row, col] = reader.ReadChar();

        return grid;
    }
}

its almost the binaryformatter

just a little extra bloat

[hoary sluice]

ignore what i said

[hoary sluice]

since when do u c#

[valid jetty]

since i can write in any language??

[fleet cedar]

PascalCase function names is the worst decision ever yeah

[valid jetty]

and i need to for school

https://github.com/acquitelol/battle-boats

if i get in trouble for publishing the source code its whatever

[nimble bone]

microsoft java

[spark tiger]

i just looked and this just uses a 3rd-party crate. my point was that without crates reading binary is too hard in rust, which it seems is true

[fleet cedar]

First party tho

And yes, if adding one line of toml counts as hard, you're gonna have trouble

[frosty obsidian]

they didn't say including the library is hard

they said that it's hard without the library

[fleet cedar]

Yeah and driving a car is difficult without tires

[valid jetty]

they said "without crates reading binary is too hard", adding that one line of toml breaks that invariant of "without crates"

[royal nymph]

virus

[valid jetty]

yeah

sorry

[frosty obsidian]

i think reading binary files should be a simple thing in any language

you shouldn't need a library to do it

[fleet cedar]

If you want to add nonsensical constraints to your language ratings you should state so upfront so I can disregard them

[frosty obsidian]

i don't think trying to avoid third party libraries is that nonsensical of a constraint

[fleet cedar]

It is

[spark tiger]

i don't like the idea of having to use a crate for such a simple thing as just parsing simple types from a binary file

i expected messing with binary being easier in low-level languages but it seems like it's not

[fleet cedar]

Imo that is exactly the kind of thing that has no place in a stdlib since there's a billion different ways to do it

Choosing one of those ways to be Objectively Superior is exactly what a stdlib should avoid

[spark tiger]

Fair, but reader already exists in rust's stdlib, it shouldn't be that hard to implement a few helper methods to make parsing easier

not "hard" but uhhh

[fleet cedar]

Easy to add, hard to design

And zero (or imo negative) value in adding, since using a crate is more adaptive

Rust's stdlib is designed around the assumption that you will use external crates for more specialized things

And that is the correct design

[spark tiger]

you have a point i guess, but that's still quite confusing to me 😔
i guess i'll probably stick with doing my own small implementation of stuff like read_u32, read_u64, etc., but i'm still not quite sure about all these types of uh readers. like each solution i've found on forums suggest different ones: Cursor, BufReader, Vec, [u8], ...
arghhh why is rust so hard

[fleet cedar]

And that is exactly why such a thing has no place in a stdlib

[placid cape]

Did you make it?

I want to do 2019 intcode so much tbh

[fleet cedar]

Who's stopping you

[valid jetty]

me

@placid cape dont do it im warning you

[placid cape]

I feel like I have to solve all previous years before 2019

[fleet cedar]

Ah, superstition

[placid cape]

¯_(ツ)_/¯

And rosie is warning me

[fleet cedar]

But you will do it anyway

You cannot resist fate

[hoary sluice]

I redid day 7 with todd ginsbergs async approach

[hoary sluice]

It feels more like implementing some sort of enterprise sotware instead of aoc

if you do it properly

[hoary sluice]

dont listen to rosie

[fleet cedar]

Implement aoc in enterprise style https://github.com/EnterpriseQualityCoding/FizzBuzzEnterpriseEdition

[hoary sluice]

Ok its not that bad

Its a sealed class with 1 abstract method thats inherited by 10 other classes who implement it

and coroutines invoking the correct class' method according to the opcode

Channel for io

[valid jetty]

i fucking love this project

[placid cape]

sealed class 😍

[ornate quiver]

https://github.com/acquitelol/battle-boats/blob/44fb6aa60383e02c95a3ea2cf1631adea5780607/src/Utils.cs#L7

[pearl stag]

Utils.cs: Line 7

public static char[,] Fill(this char[,] array, char filler) {

[ornate quiver]

scary syntax

i didnt know char[,] existed

[fleet cedar]

Is that a 2d array

[ornate quiver]

yeah

you can add more commas for more dimensions

[valid jetty]

why are you reading through it horror

[ornate quiver]

fun

[placid cape]

C#

[valid jetty]

yeah

i would do it in elle if i had the choice

[ornate quiver]

"hey i wrote my project in a language i made myself is that okay"

[valid jetty]

i mean im doing that for the bigger version of this project

im gonna write a compiler for a dummy language in elle

they say it can be any language so it should be fine i hope

[placid cape]

elle in elle 🔥

[ornate quiver]

lol

[placid cape]

tmrw ill probably start working on field access

[valid jetty]

good luck that took me a while to get right

[valid jetty]

elle calculator

use std/prelude;
use std/time;

fn calculate(string input) {
    let exec_path = "{}".format(time(nil));
    let src_path = "{}.le".format(exec_path);
    let src = "
        use std/io;

        fn main() {{
            io::println({});
        }
    ".format(input);

    if !io::write_to_file(src_path, src) {
        return false;
    }

    if !libc::system("ellec {} --hush".format(src_path)) {
        libc::system("./{}".format(exec_path));

        io::remove(src_path);
        io::remove(exec_path);
    }
}

fn main() {
    while true {
        calculate(io::input("-> "));
    }
}

or elle repl technically

depending on how you wanna look at it

it doesnt do line backhandling like the py one tho

but i can make it do that

[nimble bone]

can it remove my system’s french language pack

[valid jetty]

yeah probably

[valid jetty]

backlog repl rewritten in elle

use std/prelude;
use std/time;

const string SHORT_EXTENSION = ".le";

fn main(string[] args) {
    let program = args.remove(0);
    string[] lines = [];

    while true {
        let line = io::input(".ೃ࿔* -> "));

        if line.ends_with(";") {
            line.pop();
        }

        let backlog = "{}{}".format(
            lines.join(";\n{}".format(" ".repeat(16))),
            lines.len() > 0 ? ";" : ""
        );

        let code = "
            use std/prelude;

            fn main() {{
                {}
                io::println({});
            }".format(backlog, line)
         .split("\n")
         .map<string>(fn(string x) x.replace(" ".repeat(12), ""))
         .join("\n");

        if line == "<!dbg>" {
            io::println(code);
            continue;
        }

        let src_path = ".repl-{}{}".format(time(nil), SHORT_EXTENSION);
        let exec_path = "./{}".format(src_path.replace(SHORT_EXTENSION, ""));

        if !io::write_to_file(src_path, code) {
            io::eprintf("Failed to write to file at path: {}", src_path);
            continue;
        }

        let src_code = libc::system(
            "ellec --hush {} {}".format(
                args.join(""),
                src_path
            )
        );

        io::remove(src_path);

        if !src_code {
            let exec_code = libc::system(exec_path);
            io::remove(exec_path);

            if !exec_code {
                lines.push(line);
            }
        }
    }
}

ive GOTTA make a file abstraction soon

[dawn ledge]

calculate when my input is a string 🔥

[dense sand]

Is Prisma bad?

Like i hear all these opinions to use drizzle or typeorm instead

But im pretty comfortable with prisma

[placid cape]

I always used prisma

It's just probably a bit slower

[dense sand]

well yea same

prisma is really comfy to use

maybe i should stop watching theo

[placid cape]

totally agree lol

[ornate quiver]

yadayada inefficient orm

i like writing my own sql

[hoary sluice]

the only real database is a plaintext file and you use the kotlin stdlib to write queries

@valid jetty when is elle gonna compile to intcode

[placid cape]

real lol

intcode IR

[hoary sluice]

@placid cape i am the biggest 2019 day 7 procrastinator

i have had 2019 day 8 solved and gitignored for almost 2 years cause i wanted to push in order and didnt have a proper day 7 solution lol

i think 2019 d7 is my longest delta

[placid cape]

xdd

[placid cape]

lol

i'll have big delta for 2019 d2

idk why i solved p1 before

[hoary sluice]

omg day 9 is so free now

[dense sand]

https://fxtwitter.com/jiratickets/status/1874902644906717600

@placid cape XDDDDDD

look at the file

[hoary sluice]

@placid cape i submitted nv as my answer again

so annoying

couldve had a new delta pb

17s

finally done with intcode

didnt even take that long

[placid cape]

XDDDDDDDDDD

[placid cape]

nicee

I just solved d16 y2015

2015 is so easy tbh

[viscid grove]

am I correct in assuming they should've censored the secrets

[dense sand]

yes, but its a troll post i suppose lol

[viscid grove]

oh yeah looks like it

@jiratickets
made a bait post that blew up and now just feel bad about all of the nice people replying with kindness and encouragement. how does yacine live like this

[hoary sluice]

my fucking gcd function is wrong

wtf???

[placid cape]

Kotlin doesn't have gcd in std?

nvm it doesn't

[hoary sluice]

idk

nvm its not wrong

i just needed to take the abs

for 2019 day 10

[hoary sluice]

@valid jetty you are REQUIRED to do https://adventofcode.com/2019/day/10

i took an hour for p1 and 19 mins for p2, lb took 13min/42min

this is so cursed

[fleet cedar]

Show code

Looks like mine does funny numpy trickery in 28 lines

[hoary sluice]

https://github.com/eagely/adventofcode/blob/main/src/main/kotlin/solutions/y2019/Day10.kt

part 1 is a hack

the number of things a node sees is also the number of unique minified (divided by gcd(dx, dy)) distance vectors

[fleet cedar]

I just used atan2 for that

[hoary sluice]

if 2 nodes are collinear they share the minified vector

[hoary sluice]

yea im not that goot at math so i didnt immediately jump to angles

[fleet cedar]

I can then ||sort by that angle|| for p2

[hoary sluice]

i used it in part 2 cause there was a similar problem in aoc that required atan2

[hoary sluice]

thats what im doing

i think my p1 probably has a faster runtime

but p1+p2 is slower cause i compute the angle anyways

[fleet cedar]

Mine allocates a n^2 array which is kinda unnecessary

[hoary sluice]

i wrote a proper blocking detection system, but it wasnt giving me the correct answer for p1 so i rewrote it with the minified vector approach and then had to remake the blocking detection in p2 cause i deleted it

[fleet cedar]

Actually, looking at the numbers, much of my p2 is unnecessary

[fleet cedar]

I think something like this would be more up to modern standards||```py
import cmath;P=cmath.phase;E=enumerate
p={i-j*1j for i,l in E(open("10.in"))for j,c in E(l)if"#"==c}
print(max((len(x:=sorted({P(s-a)%9 for a in p})),(m:=s-min((s-a for a in p if P(s-a)%9==x[199]),key=abs)).real-m.imag100)for s in p))

[hoary sluice]

disgusting

[dense sand]

this is becoming such a scuffed project

[placid cape]

2019 aoc is really interesting

can't wait for it

[valid jetty]

right now check your mailbox for the install key

https://tenor.com/view/neuro-neuro-sama-vedal987-plush-pipebomb-gif-10189757888080839591

[valid jetty]

lol what is this

looks like an atan2 thing

i think my vector impl in elle would come in useful for that

[dense sand]

https://youtube.com/shorts/5cRztbMf0To

There is no way this is real

[hoary sluice]

i only realised that in p2

doing p2 without atan2 would be horror i think

[dense sand]

yall are good optimizers, is there any way to optimize this to run faster? For guava it took like 3 seconds which seems horrible(2k entries)

public Set<String> getAvailableClasses() {
    URI sourceFileUri = UriComponentsBuilder.fromUriString(getUri())
            .path("/")
            .path(getArtifactId() + "-" + getVersion() + ".jar")
            .build().toUri();

    try (var is = new ZipInputStream(sourceFileUri.toURL().openStream())) {
        Set<String> classes = new HashSet<>();
        ZipEntry entry;
        while ((entry = is.getNextEntry()) != null) {
            if (entry.getName().endsWith(".class")) {
                classes.add(entry.getName().replace(".class", "").replace("/", ".")
                        .replace("$", "."));
            }
        }
        return classes;
    } catch (IOException e) {
        throw new RuntimeException(e);
    }
}

[fleet cedar]

Run a profiler to find what takes time

[valid jetty]

@placid cape

[placid cape]

[dense sand]

apparently you can declare a class inside of a method?

[placid cape]

Funny

I never tried it lol

[valid jetty]

you can also have anonymous classes iirc

and they act as just objects

[dense sand]

you mean those that you pass in lambdas right?

like these, they do the same thing

[valid jetty]

like you can do stuff like this

public class Main {
    public static void main(String[] args) {
        Object obj = new Object() {
            @Override
            public String toString() {
                return "foo";
            }
        };
        
        System.out.println(obj);
    }
}

yah

that

[dawn ledge]

they be copying rust

smh

[valid jetty]

@hoary sluice [solution.le:30:13] Array<Tuple<i64, Triple<i32, Tuple<i64, i64>*, Array<Tuple<i64, i64>*>*>*>*>* queue = <[(12, (0, (0, 0), [(0, 0)]))] at 0x1400094f0> LMAO

the let is saving my life

let queue = [$(dist(start, t), $$(0, start, [start]))];
io::dbg(queue);

[placid cape]

[fleet cedar]

Why so many pointers

Why >*>*>*>*>* and not just >>>>>

[valid jetty]

because arrays need to be pointers to be able to be pushed to

and idk why i made tuples and triples pointers

[fleet cedar]

An array needs to contain a pointer, sure

But why would that be part of the type? It's just the struct's internals isn't it

[valid jetty]

you could technically write Array<Tuple<i64, Triple<i32, Tuple<i64, i64>*, Array<Tuple<i64, i64>*>*>*>*>* as (i64, Triple<i32, (i64, i64), (i64, i64)[]> *)[] lol

[valid jetty]

thats not what i meant

it has that

struct Array<T> @nofmt {
    i32 size;
    i32 capacity;
    T *elements;
};

the pointer to Array<T> is so the structure can be modified in place

[fleet cedar]

So things not in pointers are immutable...?

[valid jetty]

no but theyre copied

[fleet cedar]

So you don't have anything like rust's &self

[valid jetty]

if you want to have Array::push and it takes Array<T> not Array<T> * you wont edit the array youll just edit a clone of the array

[fleet cedar]

Kinda like c++ but you chose the different choice?

[valid jetty]

i do, but sometimes you may not be able to get the address of a stack variable to pass that into push

[fleet cedar]

Why not

[valid jetty]

so its easier if the array is just always a pointer

if the array is a param of a function

like

fn foo(Array<T> x) {
    x.push(1);
}

you dont explicitly allocate stack memory in the IR for this so you cant get its address

[fleet cedar]

So you're letting implementation details guide your design

[valid jetty]

sort of, i do have a workaround

if there is some value T and there is no address, it goes and allocates stack memory for it of size T and stores the value there, then returns that address

but again, that wont modify x

[fleet cedar]

Implicit copying is the devil

One of my (very many) least features of c++

[valid jetty]

which is exactly why it’s always a pointer

you never deal with Array<T> itself you just have a T[] which is an alias for Array<T> *

and so it doesn’t copy all over the place and actually mutates the array when you expect it to

if you were to do

fn foo(i32[] x) {
    x.push(1);
}

fn main() {
    let x = [100, 39];
    foo(x);
}
``` you would expect `x` to be mutated surely

if x was a Array<T> and not Array<T> * it wouldnt be

and if you dont want it to mutate there is a .clone() method on Array

[fleet cedar]

And when is the array's memory freed?

[valid jetty]

currently at the end of the program's runtime, along with every other allocation

i still havent come up with a good solution for this

[fleet cedar]

So, leaked

[valid jetty]

no

all allocations are pooled into an arena allocator who's pointer is stored in static memory and the arenas are all freed at the end of the program

there are no leaks

[fleet cedar]

"Not freed until the program ends" is the exact definition of a memory leak

[valid jetty]

technically yes but the allocations are still tracked by the compiler so i want to eventually expand this into a gc lol

[valid jetty]

mostly this

[dense sand]

it takes 2 seconds to fetch this data 😭 im so cooked with my code

[valid jetty]

day 18 done in elle

day 19 in elle

use util;

fn count_ways(HashMap<string, i64> *memo, string d, string[] ps) {
    if memo.contains_key(d) {
        return memo[d];
    }

    if d.is_empty() {
        return 1;
    }

    i64 ways = 0;
    for p in ps {
        if d.starts_with(p) {
            let remaining = d.slice(p.len(), d.len());
            ways += count_ways(memo, remaining, ps);
        }
    }

    memo[d] = ways;
    return ways;
}

fn main(string[] args) {
    let parts = io::read_to_string(args[1]).split("\n\n");
    let ps = parts[0].split(", ");
    let ds = parts[1].split("\n");
    let memo = HashMap::new<string, i64>();

    i64 c1 = i64 c2 = 0;

    for d in ds {
        let res = count_ways(memo, d, ps);
        c1 += res >= 1;
        c2 += res;
    }

    io::dbg(c1, c2);
}

[valid jetty]

why tf does python sort sets

maybe its just when printing

{(0, 0), (0, 1), (0, 2), (0, 3), (0, 4), (0, 5), (0, 6), (0, 7), (0, 8), (0, 9), (0, 10), (0, 11), (0, 12), (0, 13), (0, 14), (1, 0), (1, 4), (1, 14), (2, 0), (2, 2), (2, 4), (2, 6), (2, 8), (2, 10), (2, 11), (2, 12), (2, 14), (3, 0), (3, 2), (3, 6), (3, 8), (3, 10), (3, 14), (4, 0), (4, 1), (4, 2), (4, 3), (4, 4), (4, 5), (4, 6), (4, 8), (4, 10), (4, 12), (4, 13), (4, 14), (5, 0), (5, 1), (5, 2), (5, 3), (5, 4), (5, 5), (5, 6), (5, 8), (5, 10), (5, 14), (6, 0), (6, 1), (6, 2), (6, 3), (6, 4), (6, 5), (6, 6), (6, 8), (6, 10), (6, 11), (6, 12), (6, 14), (7, 0), (7, 1), (7, 2), (7, 6), (7, 10), (7, 14), (8, 0), (8, 1), (8, 2), (8, 4), (8, 5), (8, 6), (8, 7), (8, 8), (8, 9), (8, 10), (8, 12), (8, 13), (8, 14), (9, 0), (9, 4), (9, 5), (9, 6), (9, 10), (9, 14), (10, 0), (10, 2), (10, 3), (10, 4), (10, 5), (10, 6), (10, 8), (10, 10), (10, 11), (10, 12), (10, 14), (11, 0), (11, 2), (11, 6), (11, 8), (11, 10), (11, 14), (12, 0), (12, 2), (12, 4), (12, 6), (12, 8), (12, 10), (12, 12), (12, 13), (12, 14), (13, 0), (13, 4), (13, 8), (13, 12), (13, 13), (13, 14), (14, 0), (14, 1), (14, 2), (14, 3), (14, 4), (14, 5), (14, 6), (14, 7), (14, 8), (14, 9), (14, 10), (14, 11), (14, 12), (14, 13), (14, 14)}

[hoary sluice]

when 2019 day 10 in elle

[placid cape]

no

sets are unordered

[valid jetty]

thats what i thought but for some reason when i print a set its always in order

[valid jetty]

when i upsolve all the days in elle nyaboom~7

will take a while because im procrastinating so much i cant even finish this year in elle

[placid cape]

what

no way

[valid jetty]

whaaaaa

[valid jetty]

why is this in order then idk

[fleet cedar]

Sometimes, especially for small integers, the hashes just turn out that way

[valid jetty]

oh

true

[fleet cedar]

Mine doesn't print them in order ```py

{(0, 0), (0, 1), (0, 2), (0, 3), (0, 4), (0, 5), (0, 6), (0, 7), (0, 8), (0, 9), (0, 10), (0, 11), (0, 12), (0, 13), ...
{(12, 4), (4, 0), (5, 1), (8, 0), (14, 13), (5, 10), (8, 9), (10, 6), (0, 5), (2, 2), (11, 14), (0, 14), (2, 11), (13, 8), ...

[valid jetty]

did i just get super lucky doing this then

c = set(list(c)[:b])
``` where `c` is a set

and the order is preserved

[fleet cedar]

Which python is it?

[placid cape]

it still prints in same order

[valid jetty]

Python 3.10.14 (39dc8d3c85a7, Nov 09 2024, 22:52:48)
[PyPy 7.3.17 with GCC Apple LLVM 15.0.0 (clang-1500.1.0.2.5)]

[fleet cedar]

Yeah seems pypy prints them in order

Cpython does not

[valid jetty]

ah

i wonder if my solution wouldve been wrong if i use cpy instead of pypy

[fleet cedar]

Weird to have pypy aliased as python though

[placid cape]

send your sol

ill run it

but probably yeah?

[valid jetty]

yes it is 😭

[fleet cedar]

If you rely on set ordering then you're wrong

[valid jetty]

its wrong with python3

[valid jetty]

i dont anymore

[placid cape]

i had to do this after bron kerbosch

[valid jetty]

but that was a bug in my original solution

[placid cape]

because sets are unordered

[valid jetty]

def solve(c, s=71, b=1024):
    c = set(list(c)[:b])
    return find(c, s)

with open(sys.argv[1]) as f:
    c = set()

    for line in f:
        x, y = map(int, line.strip().split(','))
        c.add((x, y))

    print(solve(c))
    print(solve2(c))
``` i was doing this and it just shouldnt work but pypy orders sets in order of adding items so it does work lol

should be this instead

def solve(c, s=71, b=1024):
    c = set(c[:b])
    return find(c, s)

with open(sys.argv[1]) as f:
    c = []

    for line in f:
        x, y = map(int, line.strip().split(','))
        c.append((x, y))

    print(solve(c))
    print(solve2(c))

[placid cape]

i wonder why pypy has ordered sets lol

[valid jetty]

idk if them being ordered is an invariant it may just be a random side effect

wtf did i do

why is it doing that

im doing like the most basic formatting thing ever

and the weirdest thing is it breaks for a second then returns back to workig fine

?????????????????????????/ what in the

maybe im incorrectly assuming the arena has enough space and then it doesnt

[hoary sluice]

@valid jetty https://ioinformatics.org/files/ioi2019problem2.pdf

pretty sure this is just jgrapht docs

find 2 subgraphs of size a, b such that there is at least one edge between them

last question of day 2 is literally just bfs?? https://ioinformatics.org/files/ioi2019problem6.pdf

[valid jetty]

she has decided she wants to visit a b and c attractions so she has decided she is gonna spill math words

[valid jetty]

this looks slightly more annoying than just a bfs

[hoary sluice]

parse it into a graph with each building that a skywalk touches being an edge and then its just textbook dijkstra

[valid jetty]

ah yes

the bfs

my favourite textbook dijkstra

[hoary sluice]

you can also make a graph with a bunch of degree 2 vertices instead of making it weighted

its the same algo

different sorting

[valid jetty]

how long do you have to solve this problem

[hoary sluice]

idk

5 hours per problem apparently

gn

In 5h i could make a 3d visualisation of this

[valid jetty]

lmao true

[valid jetty]

@hoary sluice @placid cape https://github.com/messcheg/advent-of-code/blob/main/AdventOfCode2015/Day07/Day07.rock

[valid jetty]

@royal nymph opinions on this

fn $scoped(fn *func) {
    let current = #env.allocator;
    #env.allocator = ArenaAllocator::new();
    func();
    #env.allocator.free_self();
    #env.allocator = current;
}

fn main() {
    let a = 5;
    $scoped(fn() {
        #env.allocator.alloc(1024 * 1024 * 1024);
    });
    // All freed after this function is called
}

insta husk holy shit

[royal nymph]

make it a macro instead so you dont have to use an inline func

and dont use $ its weird

[valid jetty]

"#" is reserved for compiler builtins
"@" is reserved for function and struct attributes

calling it just "scoped" is weird

and there are no macros yet

[royal nymph]

oh it's elle

why tf u using rust language block

i thought u meant in rust

anyway just add using/with blocks

[fleet cedar]

Probably because discord doesn't have elle highlighting yet

Someone should make a plugin for that

[royal nymph]

using(mem = #env.allocator.alloc(1024 * 1024 * 1024)) {
}

[valid jetty]

no because thats not the purpose of it

thats just an example of what you can do

the purposes is like

because the allocator doesnt clean up after itself

if you have to allocate memory during a loop like a game render loop

itll just keep piling

if you $scoped it, everything will be allocated and freed on every iteration

like a temp allocator

the purpose isnt to scope only a single allocation and free it afterwards its to allocate a bunch into an autofreed chunk

sort of like the way stack allocations work but for the heap

specifically like, here

.format allocates heap memory because the string has to grow in size so it cant be stack allocated

and if this is ran in a loop itll just keep allocating

with $scoped itll free itself after its done

[wooden walrus]

helfo heres a pic of some python code i did

[valid jetty]

oh right wait elle doesnt let you capture things

so actually itll be more like this

fn $scoped(fn *func, void *arg) {
    let current = #env.allocator;
    #env.allocator = ArenaAllocator::new();
    func(arg);
    #env.allocator.free_self();
    #env.allocator = current;
}

fn main() {
    let a = 5;

    $scoped(fn(i32 *x) {
        #env.allocator.alloc(x);
    }, &a);
    // All freed after this function is called
}

ARE YOU SERIOUS

[fleet cedar]

Mangle your idents

[valid jetty]

so it has to be

fn mem::scoped(fn *func, void *arg) {
    let current = #env.allocator;
    #env.allocator = ArenaAllocator::new();
    func(arg);
    #env.allocator.free_self();
    #env.allocator = current;
}

external fn mem::scoped(fn *func, void *arg) @alias("$scoped");

[valid jetty]

what

[pearl stag]

index.html: Lines 107-113

<script src="./Javascript/72b5c5874930d6222007ea7b99269673dad47bcb7355815024742ddf575ca55f.js"></script>

<script src="./Javascript/clipboardInterceptor.js"></script>

<script disable-devtool-auto src='./Javascript/bee50f64f68a2feb114d80f4a2f1337fc164bc5574f6689ce9464be7f5ccf1ed.js'></script>
<!--<script src="./Javascript/d178e32824befa1fe8305a075094c8b669b88e27b4dba91c55abca73b3f081f0.js"></script>-->

[placid cape]

lol and this is a working legit solution lmao

[placid cape]

Why introducing $scoped? Why not just s regular block:

{...}

maybe because you don't allow inner variables in lambda 🤔

[valid jetty]

because scoped is performing a special task of providing a clean allocator which automatically frees everything it allocates after the function call

a regular block doesnt imply the allocator will free everything inside afterwards because the arena allocator isnt recreated

you also cant free a part of an arena only the whole arena

so you either free all or nothing

[ornate quiver]

holy christ

https://github.com/messcheg/advent-of-code/blob/main/AdventOfCode2015/Day07/Day07.rock#L52-L67

[pearl stag]

Day07.rock: Lines 52-67

Robert wants control
Cut control into pieces
The code is attractive
Put pieces at the code into the box
A ray is very gleaming
Cast a ray
The room is large enough
Build the room up
Cast the room
while the box ain't misterious
if the box is smaller than a ray or the box is bigger than the room give back lies
build the code up
put pieces at the code into the box
baby
the truth is right
give it back

[valid jetty]

thats fizzbuzz

ok well

this isnt the prettiest thing ever

$scoped(fn(i32 bounces) {
    string bouncesString = "Bounces: {}".format(bounces);
    raylib::draw_text(bouncesString, (screenWidth - raylib::measure_text(bouncesString, fontSize)) / 2, 100, fontSize, YELLOW);
}, bounces);

but it doesnt segfault anymore

and i discovered a huge bug in my formatter code lol

let result = (string)#env.allocator.alloc(fmt.len());

the result used to be preallocated to at least the length of the formatter

but never memset

which is just UB

the length of that string is just whichever null byte it finds (there isnt any set)

its much better to make it like just an empty string

let result = "";

because the string.push and string.extend/concat are designed to make it grow

[dense sand]

what is a reasonable amount of props to have on a component

[placid cape]

hm?

[dense sand]

cuz im using nextjs and i might be passing a lot of stuff from the server component into client subcomponent

[placid cape]

how many?

[dense sand]

its 5 now

[placid cape]

i think that's okay

[hoary sluice]

omg i was looking for that language

ty

[dense sand]

why is there 0 parsers support es6 modules

what is this

[placid cape]

babel?

babel supports everything

[dense sand]

Mb i meant xmlparsers

Most of them still use commonjs

But i found some fast-xml-parser whic works ig

[placid cape]

ohh

it should be possible to use commonjs modules with esm

[dense sand]

Do yall know of any optimized version of ZipInputStream

[royal nymph]

browsers have XML parsers built in

https://developer.mozilla.org/en-US/docs/Web/API/DOMParser/parseFromString

[hoary sluice]

@valid jetty you are going to do 2019 day 12

crt - ish problem

[dense sand]

Eh its on server

[hoary sluice]

i had to look up a hint in the subreddit cause i was completely lost after an hour

the solution is so simple 😭

[royal nymph]

can u parse it on client instead

[dense sand]

well i would get cors errors then

comedy 😭

i should probably extract it into layouts instead

[viscid grove]

Guava

[dense sand]

yea im making javadocs 2

[hasty vessel]

sure do love pygame

[celest lodge]

@torn seal

you may start crying

[torn seal]

lmao

[celest lodge]

it is not good

[torn seal]

are you trying to make a coreutils replacement or what

[celest lodge]

i got bored one day

i wanted to see how hard it was to make a posix-y ls

[valid jetty]

@fleet cedar i finally fixed the struct double alloc thing

no more double alloc to dynalloc a struct

the issue was when i was doing autoderef on Foo*

it makes no sense to hold a Foo

so doing a load instruction on that ptr would give you a nullptr deref

[dense sand]

Is there any way I can pass something from layout to page in nextjs?

[valid jetty]

i just removed that part and it works for both Foo and Foo* field access

now i need to go and fix all of the places i did double alloc 🥲

[placid cape]

🔥

i need to work on field access and oeprator overloading

[hoary sluice]

@deep mulch HII

[dense sand]

we are finishing this one 🗣️

[placid cape]

nice design

looks reallly nice

[deep mulch]

hii

[dense sand]

yea im just hoping that converting this hell lot of classes into package based tree wont slow this site down by a lot

[valid jetty]

vcotd

[elder yarrow]

The cozy of the day is mint green! (#98ff98)

[placid cape]

you can theoretically generate every html page at build time

e. g. https://mappings.xhyrom.dev/ this site has like 10GB if im not mistaken lol

[dense sand]

well i can scrape any maven repository for javadocs right now

so it cant really be done on build time

[valid jetty]

its kinda weird how this works

to deref a structptr into a struct i kinda just do .. nothing

i just return the same value with a different type (Foo * and Foo are equivalent actually)

its weird that it works like that

in this

use std/prelude;

struct Foo {
    i32 a;
};

fn main() {
    Foo *foo = #env.allocator.alloc(#size(Foo) * 4);
    foo[0].a = 42;
    foo.a = 42;
}
``` those are the same thing

[dense sand]

okay i added class lists :D

[valid jetty]

guaba

[dense sand]

yea i mean its kinda the best thing to see it on

[placid cape]

foo[0].a thats weird

[valid jetty]

how

you would do like
foo[0].a = 42; foo[1].a = 39; etc

the allocation is for 4 Foos

[placid cape]

ohh didnt see that

[valid jetty]

but yeah because of auto deref rules foo[0] and foo and *foo are actually the same thing

its really weird but the c compiler does the same thing for llvm so

[placid cape]

i have to rewrite my semicolon parsing

it's horrible and has bugs

[winged mantle]

got a mechanical keyboard and idk if it's better for programming tbh because it's really hard not to make typos

[dense sand]

you will get used to it

i cant type on membrane now lol

[winged mantle]

how do you type without knocking the arrow keys

[dense sand]

huh lol

what size do you have?

[winged mantle]

no numpad

[dense sand]

tkl?

i have tkl too

i just... dont touch the arrow keys at all?

[winged mantle]

my palm presses them

[dense sand]

wth idk

[winged mantle]

and it can only be avoided by lifting my hands an uncomfortable amount

[dense sand]

wtf lol

[winged mantle]

the tiniest bit of pressure registers on the connected device

it's also very easy to press ctrl by accident meaning if you try to type an R it reloads

also this is horrible lol

[placid cape]

lol

https://monkeytype.com/

[hoary sluice]

@placid cape 2019 is really fun

[placid cape]

make 15s english

[hoary sluice]

and probably also the most difficult

day 13 requires you to write a breakout ai

[placid cape]

interesting

[hoary sluice]

but there is no anticheat

cause its intcode

so there are a few silly solutions

[placid cape]

ill try to finish 2015 until end of next weekend

[hoary sluice]

by not having arrow keys

[placid cape]

rust????!

[hoary sluice]

yes

the keyboard is called ferris sweep, its a fork of the ferris keyboard and ferris is the name of the rust crab

thats why hes holding a broom

cause sweep

get it haha

[hoary sluice]

are you touch typing properly

cause i think ur turning ur hands to not have to angle your wrists in a silly way and hence pressing arrow keys