🪅-progaming

[supple whale]

aka go down the list if there's no break

[royal nymph]

go switch makes fallthrough explicit

[supple whale]

a LOT of fancy algos do that

[fleet cedar]

Switch statements are slight syntax sugar for goto

[royal nymph]

i.e. instead of writing break you write fallthrough when you do want to fallthrough

[fleet cedar]

Completely worthless in modern coding

[royal nymph]

switch foo {
  case 1:
    print("1")
  case 2:
    print("2")
    fallthrough
  case 3:
    print("2 or 3")
}

this is go switch

[dusty moth]

my language has switch as a slightly different alias for match

[supple whale]

so

[fleet cedar]

How does that work

[supple whale]

what ya'll meant to say

is switch statements suck

but insert unorthodox switch implementation here is good

aka that assigment sucks for normal switch

[dusty moth]

my language has switch_expr as a copy of match_expr with match changed to switch (and case is optionally accepted)

[supple whale]

these kidn of statements make me want to spend the next 4 days tracking you down just so i can cut off ur balls

becaue people like you is why UE5 exists

[dusty moth]

unrelatedly, got an error

[supple whale]

"yeah we dont need it, just use something slower thats simpler, throw more compute at it, it will fix itself"

https://cdn.discordapp.com/emojis/1216760206705557605.gif?size=48&quality=lossless&name=axorager

[fleet cedar]

You're forgetting that match is infinitely more powerful than switch can ever dream to be

[supple whale]

brotha

they are different things

[fleet cedar]

Since it's pattern matching, not just constants

[supple whale]

they act similarly

but they are different things

[dusty moth]

unless switch is just match in a trenchcoat (like java)

[supple whale]

with different purpoe

as you said, switch is just goto

match isnt just goto

there's extra shit there

[fleet cedar]

There are zero uses for switch that are not better handled by structured control flow

Other than legacy languages

[supple whale]

that just means you havent worked with low level code which needs to be executed billions times per seconds where thoes uses matter

i'm not saying they are common

but switch has its place

and while I agree that match is often a better option for what programmers need

saying "match is better than switch" is like saying "just use salt its better than pepper"

no, they are different things

and i also fucking hate switch and always use some form of match

[fleet cedar]

You can write the exact same control flow graph in better ways

[supple whale]

switch is amazing for math, such as exponent functons etc

[fleet cedar]

??

[supple whale]

ex:

let expo = 0
swtich (order)
  case 9:
    expo += 1
  case 8:
    expo += 2
  case 7:
    expo += 7
  case 6:
    expo += 42
  
return expo

fallover has its uses

they are rare but they exist

i've also seen it a lot in networking hardware, because its simply fast for weridge non-standard shit

[supple whale]

dis kinda a bad example cuz its performing the same math operation, but it could be doing other shit that'd need to be summed in order

its minor, but this is the shit that matters when you're transferring terabytes of data per second

[fleet cedar]

I would have to compare the generated asm before I could say anything about that

[supple whale]

fair

for match you'd need to put all the prev expressions in the next call, which means bigger jumps, which means slower

[fleet cedar]

I'd try with something like ```
if order >= 9 { expo += 1 }
if order >= 8 { expo += 2 }

Any decent compiler should optimize that to the same

[supple whale]

in a simple single operation case yeah

which is whyi

[supple whale]

.

[fleet cedar]

The body of the cases is irrelevant here

But I do admit that I typically work in fields where code quality is more important than squeezing cpu cycles

[dusty moth]

speaking of matches, how should I codegen them

should I convert rs match foo { Some(Some(a)) => 2 Some(None) => 1 None => 0 }tors if Some(_1) = foo { if Some(a) = _1 { 2 } else if None = _ { 1 } } else if None = _ { 0 } or would something else make more sense

[fleet cedar]

If the patterns are disjoint, probably a jump table

[dusty moth]

ah, that makes sense ig

[long sparrow]

oh

[dusty moth]

and unflatten the matches?

[fleet cedar]

What do you mean

[long sparrow]

i was about to submit a cool vencord plugin idea, are general plugin requests not allowed?

[dusty moth]

like rs match _1 { Some(Some(_2)) => _3 Some(None) => _4 None => _5 }tors match _1 { Some(_a) => match _a { Some(_2) => _3 None => _4 } None => _5 }until I add niches

[fleet cedar]

Yeah sounds like a decent approach

Or patterns would complicate slightly maybe, but they're probably just a goto

[long sparrow]

maybe the wrong place to ask... sorry

[dusty moth]

or patterns would just be two gotos to the same place probably

let r = match foo {
  1 | 2 => bar
  _ => baz
}``````rs
jump_table!(foo) {
  1 => 'a
  2 => 'a
  _ => 'b
}
'a:
goto 'c(bar)
'b:
goto 'c(baz)
'c(r):
/* ... */```

[dusty moth]

oh no what did I fuck up this time

[fleet cedar]

Tree sitter?

[dusty moth]

using tree-sitter to procrastinate on actually writing the parser myself

[fleet cedar]

Yeah I'm procrastinating mine too

I need to add parsing and serializing so I can compile my decompiled scripts back, but I'm stuck on figuring out exactly which subexpression each line number instruction belongs to :/

[dusty moth]

insane

[dusty moth]

did i never actually test multi-digit numbers properly

[fleet cedar]

Eh, who uses those anyway

[dusty moth]

nvm they're getting bound as identifiers somehow

[dusty moth]

dark blue means it's an identifier getting bound somehow

what the fuck

[fleet cedar]

\w does match digits, so maybe that's somehow

[dusty moth]

they're being matched as emojis

[fleet cedar]

Why do you have emojis as syntactic constructs in your language

[dusty moth]

@analog sail

[analog sail]

Any Unicode works

[dusty moth]

eh I'll allow it

except for operators

[analog sail]

make quotes allow operators

[dusty moth]

my identifier regex is \p{XID_Start}\p{XID_Continue}* for textual identifiers and \p{Emoji} for emojis

[dusty moth]

wdym

[analog sail]

like "[operator]" is valid

and it stores without the quotes

[dusty moth]

that's already valid?

that's a regular string

[analog sail]

allow overriding operators

[dusty moth]

you can

just ```rs
fn (a: Foo) + (b: Bar) {

}```

[analog sail]

I want + = -

things like that

for = +

[dusty moth]

syntactic nightmare

[analog sail]

DO IT

= = ==
== = =

[dusty moth]

fn self + other {

}```compiles to a metatable

you can do something similar with metaprogramming

[analog sail]

temp = ==
== = =
= = ==

[analog sail]

allow that to happen

[dusty moth]

use swap_eq;``````rs
schema swap_eq {
  fn preprocess(ast: SourceFile) -> SourceFile || Error {
    ast.map_exprs(fn match {
      .Assign(l, r) => Expression.Equal(l.cast(Expression), r)
      .Equal(a, b) => Expression.Assign(a.try_cast(Pattern), b)
    })
  }
}```

[valid jetty]

how in the heck are you gonna parse this

"the = has no = behind it so it must be a declaration"

[analog sail]

that's not up to me to figure out

[dusty moth]

i could pull a perl and not allow deterministic parsing

[dusty moth]

@nimble bone @nimble bone what sso do you use

[nimble bone]

authentik

[dusty moth]

why is the ↓ button just not
https://jsfiddle.net/PoolloverNathan/msnozpL7/12/

[dusty moth]

i could just make operators identifiers and make tree-sitter figure it out

[dense sand]

hi, im here once again with react problem lol, i have a component, but this useeffect in the component gets run twice, i suppose this is again the same problem with the strict mode, but im not entirely sure when the problem could be in dev

[autumn sigil]

have you considered not using react

[dense sand]

the project is due tommorow >:3

[autumn sigil]

just enough time to rewrite everything

[dusty moth]

show rest of component

including where setSupabase is called

[dense sand]

supabase is not a usestate

[dusty moth]

what is it

[dense sand]

call to createBrowserClient(...)

[dusty moth]

in what context?

[dense sand]

export function EndMenu({ score, allChick }: EndMenuProps) {
  const router = useRouter();
  const supabase = createClient();

  useEffect(() => {
    async function updateScore() {
      console.log('[CDS] Updating score...');
      
      const user = await supabase.auth.getUser();

      if (user.data.user) {
        console.log('[CDS] User found:', user.data.user.id);
        
        const resp = await supabase.from('leadeboard').insert({
          "score": score,
          "user_id": user.data.user?.id,
        });

        if (resp.error) {
          console.error(resp.error);
        }
      }
    }
    updateScore();
  }, [supabase]);

  return (
    <div>...</div>
  );
}

heres the whole thing(stripped of stuff you dont need i suppose)

createClient(); calls createBrowserClient with few extra params

[dusty moth]

ah, you're creating the client in component context

hint: put a log inside createClient

[dense sand]

mm i suppose wrapping it in usestate and remove the supabase dependency would fix the issue, but its not entirely correct, is it

[spark tiger]

Is anyone familiar with how chrome devtools' overriding content works? trying to make a change to some site's js file but pressing override content doesn't seem to do anything on the site

[dense sand]

nvm it doesnt

the thing is, even if i leave it as [] and put createClient into useeffect block, it gets ran twice

[dusty moth]

usually it's initialized to null then properly initialized in a useEffect with []

[fleet cedar]

useEffect is run twice in debug mode, yes

To ensure that there are no side effects

[void leaf]

why did i just get a notification from here

?

[dusty moth]

wait, effects are run twice?

[void leaf]

the fuck you doing discord?

[dusty moth]

i thought effects were explicitly not run twice

[fleet cedar]

https://react.dev/reference/react/useEffect#my-effect-runs-twice-when-the-component-mounts

[dense sand]

so would there be any way to prevent it running twice in dev?

[lost pawn]

Does anyone know why discord's useSpring makes some modals open in slowmo?

[spark tiger]

also how do I create a chrome extension that would patch the needed JS script when it's being loaded on a webpage? like what event should I look for and how do I determine when all the needed scripts have loaded?

just trying to create a simple patch that should simply turn false in one function to true

[lapis dove]

not really programming just some fun xd

[jade stone]

i love when im trying to poke at a website and it does this to __REACT_DEVTOOLS_GLOBAL_HOOK__

[nimble bone]

husk

why

[jade stone]

they dont want people using react devtools

i already made an extension to bypass it

(cant use userscript)

[past moon]

hi everybody, i got a vencord plugin idea but idk if it's even possible so please tell me. Since the last attachment size limit change i'm constantly bothered by a popup asking me to get nitro if the file is >10MB, but would it be possible to make a plugin that tries to compress files over 10MB down to <10MB ? Since that's what I end up doing with third party tools I feel like there could be something to do here but i don't know enough about vencord and how it works to tell if it's even a good idea

[autumn sigil]

zig server so based

[shy mauve]

I have tried ai code for small projects before, it's good if you rlly specify what you need

[frosty obsidian]

i had this policy in vendetta

i deleted any ai generated content

[balmy lintel]

you should enforce in this server

(even though its not really a problem here)

[frosty obsidian]

vee said no

[ornate quiver]

whY

[frosty obsidian]

vee doesn't really want to restrict what people post

unless it's hate speech

[ornate quiver]

fair

[winged mantle]

i hate vee

[nimble bone]

wing hates ai

[lost pawn]

I could see where they are coming from

like

"there are ppl here dedicating hours to this lang, and then there is you, the guy who thinks mr. ai will land you a job"

or I could be completely off

prob the latter one

cuz in the end, both humans and ai can write shit code, it's just that ai is more likely ig

[valid jetty]

int main() {
    for (int i=10; i-->0;)
        cout<<2*i+-+-+-+-1<<endl;
    return 0;
}

quality

(i wrote this a while ago and found it lol)

i think you could go as bad as ```cpp
main() {
for (int i=10; i-->0;)
cout<<2*i+-+-+-+-1<<endl;
}

at this point this isn't even cpp anymore

[dusty moth]

that's valid cpp i think

with -Wno-all

[dense sand]

Can i somehow get discord cdn shit in my app? It seems to have cors

Id like to just have gif in my app which is on discord cdn

[valid jetty]

no

discord cdn links have expiry arguments

they expire after 24 hours if no client fetches them

[dense sand]

so i cant fetch gif into my app

[valid jetty]

not from discord

if you want a gif that’s always fetchable upload it to imgur or something

[jade stone]

is there a guide / docs on discords gateway format

its just all binary

[royal nymph]

cause its erlpack

enable gateway log in discord developer settings (enable experiment plugin)

and it will log every message

[frosty obsidian]

its zstd

compressed

do what vee said

[ornate quiver]

https://discord.sex

[dim cave]

why cant i resize an image in css

  .socialmediaicon{
   width: 32px;
  }
``` doesnt work

[jade stone]

dont know much about css, but there might be a max/min width set

[dim cave]

there issntt

but

i added display: inline-block; to it

and um

wtf

[jade stone]

use devtools

so good for css

[dim cave]

devtools?

[jade stone]

yes 😭

ctrl + shift + i

it will tell you things like this

(not saying thats your issue here)

[dim cave]

oh shit wait

im adding the class to the LINK not the image

im fucking stupid

[umbral mirage]

https://cdn.discordapp.com/attachments/911223311709638678/1304756714700144730/append_top.png?ex=67308ce1&is=672f3b61&hm=200e9059ea1286a3ce6f668d8ffa5e2762989ee2edba236d57d36dd0ca4421a6&

[fallen nebula]

Lmao

[cerulean plover]

mkdir --rate 😠

[median root]

ok incredibly random question but are there any other packages that improve commonly used terminal commands

like zoxide for cd?

[fleet cedar]

Eza, rg, bat

[pseudo sierra]

atuin

dust for du

[valid jetty]

:3?

its slightly convoluted

but its cute

i think

it basically implements a builtin method called __fmt__ on all structs created in elle which produces this nice output

and then functions can choose to have structs diminish into a string via the @structfmt attribute

and if you want no fmt function on your struct for whatever reason you put the nofmt attribute on your struct

HOLY SHIT LMAO

IT WORKS WITH THINGS NOT CREATED IN ELLE

[royal nymph]

valibot is so nice

[autumn sigil]

valibot very good

[dusty moth]

how does that work

also you should make it a regex environ

so you can set it to /v(e+\s*😭?)?\s*/i

[ionic lake]

i just let neon-env handle it

[royal nymph]

it's very similar to what I do

i just use valibot to parse process.env with with schema

https://codeberg.org/vee/bot/src/branch/main/src/env.ts#L10-L55

[pearl stag]

env.ts: Lines 10-55

const configSchema = object({
    PREFIXES: pipe(
        string(),
        transform(s => s.split(/ +/).filter(Boolean)),
        array(string()),
        minLength(1)
    ),
    DISCORD_TOKEN: string(),
    DATABASE_URL: string(),

    NODE_ENV: optional(picklist(["development", "production"])),

    GUILD_ID: string(),

    COMMUNITY_CATEGORY_CHANNEL_ID: string(),
    COMMUNITY_POST_PASS_ROLE_ID: string(),

    DEV_CHANNEL_ID: string(),
    SUPPORT_CHANNEL_ID: string(),
    BOT_CHANNEL_ID: string(),

    MOD_PERMS_ROLE_ID: string(),
    MOD_ROLE_ID: string(),
    MOD_LOG_CHANNEL_ID: string(),

    MOD_MAIL_CHANNEL_ID: string(),
    MOD_MAIL_LOG_CHANNEL_ID: string(),
    MOD_MAIL_BAN_ROLE_ID: string(),

    HTTP_SERVER_LISTEN_PORT: pipe(
        string(),
        transform(Number),
        number()
    ),
    HTTP_DOMAIN: string(),

    GITHUB_PAT: string(),
    GITHUB_CLIENT_ID: string(),
    GITHUB_CLIENT_SECRET: string(),

    CONTRIBUTOR_ROLE_ID: string(),

    NINA_CHAT_TOKEN: optional(string()),
});

const parsed = mustParse("Invalid environment variables", configSchema, process.env);

[jade stone]

i am going insane

[autumn sigil]

no youre just dumb

[jade stone]

guhhh what am i doing wrong

[autumn sigil]

pull prop acces into variable so ts knows its number

[jade stone]

can tsserver not infer if its not in a var

[autumn sigil]

no, it doesnt know e is the same

i mean it does i guess, idk how to explain it

but no it just doesnt work like that

[jade stone]

i think i will just as any

[autumn sigil]

oh is this an enum

[jade stone]

uhh

one moment

yea

it is

[autumn sigil]

why not push(...Object.values(enum).filter(isnum))

[jade stone]

idk

wrote this code a while ago

just coming back to it

[autumn sigil]

oh does filter even do result inference anymore i forgot

[jade stone]

prob if the function asserts that the input is a number

[autumn sigil]

too much zig lately

[jade stone]

let intents: number[] = Object.values(GatewayIntentBits).filter(e => !isNaN(Number(e))) as number[];

[autumn sigil]

no not really

i sleep bye

[dusty moth]

it does now

if it constrains the input's type

T[].filter(fn: (x: T) => x is U): U[] (simplified)

[ornate quiver]

does anyone remember when a guild becomes "large" (requiring an op 13 event to be sent to receive future events)?
was it 50 or 300 members or something

[frosty obsidian]

i thought it was 100

but the api will tell you if a guild is considered large

you shouldn't need to calculate that

[ornate quiver]

oh right

[valid jetty]

who wouldve guessed that this code causes SO MANY ISSUES

global pub;

struct Node<T> {
    void *next;
    T data;
};

struct LinkedList<T> {
    Node<T> *head;
};

fn LinkedList::new<T>() -> LinkedList<T> {
    return LinkedList { head = nil };
}

fn LinkedList::add<T>(LinkedList<T> *self, T data) {
    Node<T> *new = malloc(#size(Node<T>));
    *new = malloc(#size(Node<T>));

    if !new {
        return;
    }

    new.data = data;
    new.next = self.head;
    self.head = new;
}

basically because Node is also generic it needs to also generate monomorphic definitions for that when making them for LinkedList

which it was doing on the parser level for structs that define it directly (some struct that defines a field like Node<i32> instead of Node<T>)

but in the compiler it was only doing the root struct, which is weird because???????? it was working before

i have been experiencing the worst form of torture a human can experience

[deep mulch]

my professor uses var for constants interchangeably and uses single quotes in js 😭

@royal nymph

[valid jetty]

imo single quotes arnet bad as long as youre consistent with it

[deep mulch]

@valid jetty

[valid jetty]

the other stuff should be banned though

doesnt it literally tell you "var is a deprecated keyword and breaks scoping rules. use let and const instead"

unless your profsssor uses a super old editor

[deep mulch]

vscode

[valid jetty]

it should tell you iirc

[deep mulch]

idk it doesn't tell me either

[valid jetty]

horror

[deep mulch]

maybe I need some extension

[valid jetty]

it should just be part of eslint

[royal nymph]

LOVE

singlequotes in js is sadly very common

people are insane

[valid jetty]

no i like them

they look more demure than double quotes

[deep mulch]

I only use single quotes if it's the language convention

[fleet cedar]

I like double for string and single for stringly-typed enum

[deep mulch]

do I need to install eslint globally to use with vscode @valid jetty

[crimson sparrow]

Oops I shouldn’t be doing that?

Literally did a replace all on a multi thousand line project recently from double to single…

[royal nymph]

i mean it's purely preference xd

but I prefer double quotes because in other languages, single quotes stand for chars

double quotes = strings
single quotes = chars

[deep mulch]

yop

[royal nymph]

and imo double quotes look better

[deep mulch]

kotler does that

yop

[royal nymph]

nah but i would do it anyway

makes it easier to run from the terminal

[deep mulch]

I can't get vscode to use it

help

[valid jetty]

v3 is written entirely with single quotes

i am not joking

[deep mulch]

you should delete the entire project

when no one's looking

[valid jetty]

:3

[deep mulch]

vscode LSP when

[spark tiger]

https://vxtwitter.com/astrodotbuild/status/1856011142390513977 that’s pretty cool i guess

though i’m honestly still not sure what so good about astro is

[royal nymph]

it's amazing

super nice DX

[valid jetty]

you can define your own fmt method now

[dusty moth]

my fucking rust code segfaulted

[spark tiger]

I actually never heard about it before browsing the source of vencord.dev, but it seems to be getting more popular
and I generally just can’t tell what the difference is between all the big js frameworks, like they all… look the same to me. how would i even choose one if i were to do some web dev

[valid jetty]

the compiler can do that

[dusty moth]

no the build output segfaulted

[valid jetty]

..?

unsafe block

??

[dusty moth]

none

[valid jetty]

bad wrapper somewhere then

[dusty moth]

lemme grab code

can only paste once every 10 minutes

https://nathanlaptopv.axolotl-snake.ts.net:10000/l1zxnbwisyywy6f9djpzfsynv28ccn1v-source/src/

[deep mulch]

oh no

• before the name

is this like C pointer or something else

[valid jetty]

yes

• before the name because Array::new returns a pointer to Array<T>

technically pointer formats should be like

fn __ptr__::__fmt__<T>(T *self, i32 nesting) {
  return string::format("<{} at {}>", *self.__fmt__(nesting), self);
}

but theyre more like

fn __ptr__::__fmt__(i64 self, i32 nesting) {
    return string::as_string_wrapped(self);
}

atm

need to fix soon

[frosty obsidian]

it really depends on your needs

astro is really good for more static sites, although it does allow for more dynamic stuff

its also really good for documentation

starlight has been the best experience I've had making documentation

[valid jetty]

HOLY SHIT YES

finally

[deep mulch]

personally I hate that

I doubt you'll change it but if I could it would be before the type

*Array<i32>

[valid jetty]

idk

probably too late to change it now

[deep mulch]

wyd @valid jetty

[valid jetty]

being able to visualize arrays again after all this time

is so nice

[deep mulch]

that's good

[valid jetty]

yay

[autumn sigil]

[dusty moth]

is this C semantics

where you only actually give the innermost type

[valid jetty]

not really

the main reasons you get a pointer instead of a struct directly are:

• to move it around youre only passing 8 bytes not 16 (size = 4, cap = 4, ptr = 8)
• methods can edit the array in place (otherwise youd have to do Array<T> arr = arr.push(1) constantly)
  TECHNICALLY elle automatically takes a pointer to the array when calling a method that requires self by reference instead of by value, however in cases such as you passing the array to another function by value, the argument will not have an address that you as the developer have access to, so you cannot call any of those methods that take a reference unless you create a new stack variable and set the array to it and then take the address of that instead

this totally isnt gonna be confusing for anyone righttttt

[deep mulch]

Rosie so smart

[valid jetty]

React.__SECRET_INTERNALS_DO_NOT_USE_OR_YOU_WILL_BE_FIRED part 2

[royal nymph]

ELLE_INTERNALS_DO_NOT_USE_OR_YOU_WILL_BE_FIRED

[dusty moth]

what's an ElleMeta

[valid jetty]

its a this

runtime information about what the function is being called with and from where

[dusty moth]

does writing to types reinterpret?

[valid jetty]

no, types is just a static array of strings, the string is the type formatted as a string by the compiler

[dusty moth]

oh

[valid jetty]

thats how i get like these types here

[dusty moth]

wait what does variadic type the variable as

[valid jetty]

variadic isnt a type lol

well

oh

[dusty moth]

ik; what type does the variable end up being

like what's args's type

[valid jetty]

i64

technically it should be a va_list but its not

i64 is just like, a raw pointer

[dusty moth]

oh it's just a raw pointer?

[valid jetty]

yes

[dusty moth]

interesting

[valid jetty]

because this is all the compiler is doing

just allocates stack memory worth the size you specify and then calls vastart on that ptr

[dusty moth]

oh

[valid jetty]

its then a compiler gimmick but you can do args.yield(some type)

[long knot]

I made a script that adds the title of folders below each of them

function insertAfter(referenceNode, newNode) {
    referenceNode.parentNode.insertBefore(newNode, referenceNode.nextSibling);
}
let folders = Array.from(document.getElementsByClassName("folder_bc7085"))
document.querySelector('[aria-label="Servers"]').style = "text-align: center;";
folders.forEach(function(e){
    const label = document.createElement("p");
    label.innerHTML = e.parentElement.parentElement.parentElement.parentElement.getAttribute("data-dnd-name");
    label.style = "color: white; font-size: 12px;"
    insertAfter(e.parentElement.parentElement.parentElement.parentElement.parentElement, label)
    // console.log(label, e.parentElement.parentElement.parentElement.parentElement.getAttribute("data-dnd-name"), e.parentElement.parentElement.parentElement.parentElement)
})```

[hollow locust]

can I make proxyLazy handle undefined properly?
this is the current result when factory returns undefined:
Uncaught TypeError: Reflect.get called on non-object

[serene elk]

what would be handle properly

[hollow locust]

that

[royal nymph]

no

and I don't understand this use case

[hollow locust]

okay I'm insane

I need to find another way

thanks

[serene elk]

you can actually

but there isn't really a point

[hollow locust]

I mean, I'm using this for a lazy workaround for a complex situation

but I guess I need to resolve this the hard way

[royal nymph]

explain ur use case 😭

why does the error matter

ur code should never run into error

[hollow locust]

like I said it's kinda complex,
well I built the plugin loading system in you know what project so that first the plugin code is ran and then when preparation is done a plugin details object is returned but the thing is that some plugins want to use their own plugin details during initial execution and it doesn't exist yet at the time lol

so I thought that maybe adding a proxy lazy would make a workaround but this introduces more issues

[serene elk]

I dont think using a proxy lazy to throw errors when its ran to early is a good solution

proxy lazy is more suited for when you need to return an object which is only going to be available later

like webpack finds

but not for errors

you can probably get it to work but there might be a better solution

[hollow locust]

yes a better solution is to first push the plugin details and then fill them as needed

but I didn't think of that a year ago

[serene elk]

how exactly are details accessed

[spark tiger]

let's say for a "about me" site but with some fancy visuals. i'd like to create something similar to one of those sites from https://brutalistwebsites.com/

[serene elk]

code example would help

https://github.com/Vendicated/Vencord/blob/immediate-finds/src/utils/lazy.ts#L63-L76

[pearl stag]

lazy.ts: Lines 63-76

/**
 * Wraps the result of factory in a Proxy you can consume as if it wasn't lazy.
 * On first property access, the factory is evaluated.
 *
 * IMPORTANT:
 * Destructuring at top level is not supported for proxyLazy.
 *
 * @param factory Factory returning the result
 * @param attempts How many times to try to evaluate the factory before giving up
 * @param err The error message to throw when the factory fails
 * @param primitiveErr The error message to throw when factory result is a primitive
 * @returns Result of factory function
 */
export function proxyLazy<T = any>(factory: () => T, attempts = 5, err: string | (() => string) = `proxyLazy factory failed:\n${factory}`, primitiveErr = "proxyLazy called on a primitive value.", isChild = false): T {

[serene elk]

however this implementation might help

but first lets try to get a better solution

[hollow locust]

PluginsHolder = {
// ...
  get: function (name) {
      return this.getAll().filter(x => x.name == name)[0] ?? this.getAll().filter(x => x.originalName == name)[0];
  },
  getAll: () => {
      return window.GeneratedPlugins as AssembledPlugin[]; // idk why is this on window, don't remember
  },
// ...

this is the Plugins.get and getAll
GeneratedPlugins is populated after completing of running the actual plugin code

[valid jetty]

you can just wrap it in a proxy which calls Reflect.get on {} if the receiver/target is undefined

you don’t need proxyLazy js has it as a feature itself

[frosty obsidian]

it would be perfect

[serene elk]

then you could just throw in one of those

[spark tiger]

neat thanks!

[frosty obsidian]

how the site looks is still just css

astro has support for preprocessors like scss though

[hollow locust]

some plugins check for the return value of get to check if the plugin is valid and proxied returns a proxy, so I guess forget the initial question

[frosty obsidian]

and also has support for mdx

[serene elk]

damn

do plugins store the return value of get in a top level variable or something

and never call get again

[spark tiger]

not sure why to use scss either though. from what i've heard it essentially just simplifies some uhhh tasks you do with css (actually that sounds like a description suitable for any framework or something lol)

[frosty obsidian]

scss is very good

[spark tiger]

i'm no webdev guy

[frosty obsidian]

i prefer it for more complex styling

theres also tailwind but that takes some getting used to

[spark tiger]

oh yeah i've heard of that one a lot recently

[lament bay]

scss was useful when css didn't have all the bells and whistles it has today but now scss is just an unnecessary abstraction in most cases and can easily vomit out some horrible css

[frosty obsidian]

it still massively improves organization

imo

[shell crag]

i like scss mixins

[hollow locust]

yeah some do that

const plugin = Plugins.get(config.name) // top level

but other one does this:

getInstalledPlugin (plugin) {
    if (!plugin || typeof plugin.authorname != "string") return;
    const iPlugin = SomethingSomething.Plugins.get(plugin.name);
    if (iPlugin)
        doSomethingWith(iPlugin)

[frosty obsidian]

and I didn't really notice anything wrong with the output

[serene elk]

okay yeah

unless you change when details are set you are gonna need to use a proxy

[hollow locust]

as you can see if (iPlugin) will success even if the real result is undefined

[serene elk]

do you want this type of error message

or

[hollow locust]

and that introduces more issues

[serene elk]

but that's what is going to happen once you use a proxy

[hollow locust]

yeah that's why trying to find some other solution right now

[serene elk]

I dont think there is another sadly

besides fixing the real problem

[hollow locust]

yeah i guess

thanks anyway

[serene elk]

do you still want help with implementing the proxy lazy solution

[hollow locust]

nah it wouldn't really help, it was my first thought before I noticed other problems

[serene elk]

yikes

goodluck then

[dusty moth]

this line (let mut db = ) segfaults

found the culprit I think

one of the blocks in this mf...

specifically checking if it's threadsafe lmao

and... creating a connection reaches that codepath unconditionally

[severe zealot]

does anyone know how webpack patches work? im trying to create a client mod for slack and im struggling.

[spark tiger]

this and the next few messages might help you understand it ig #🪅-progaming message

[dense sand]

could someone tell me the cause of this? _cpu function symbol is defined in vmdef.h which has include guards, theres no reason to have this defined twice

[fleet cedar]

What do you mean has include guards

Include guards don't prevent it from being included in multiple different compilation units

[dense sand]

so how would i solve this then

main.c includes vmdecode.h which includes vmdef.h

[fleet cedar]

Don't define function bodies in headers

Unless they're static or inline

[dense sand]

ah i see

[valid jetty]

you can actually visualize the garbage collector now lmao

that’s fun

[umbral mirage]

it's beautiful

[dense sand]

java

[fleet cedar]

I love immutability that throws at runtime

Especially when mislabeled

[royal nymph]

getMutableSystemProperties
look inside
immutable

[dense sand]

Peak gradle coding this is

[spark tiger]

RUST

[dense sand]

i cannot believe this shit

[viscid grove]

are you watching in browser bc youtube mobile app is bugged?

[clever musk]

Hey @winged mantle A friend of mine wanted a plugin that also removes profile effects. I was wondering if you wanted to merge it as one plugin for themes and effects or I could just make a seperate plugin for it.

[winged mantle]

you can remove effects with css

i made the plugin because it was hard to do right with css

[clever musk]

alright then

I just thought it was more ideal to implement it into one (noprofilethemes) then make others search for css?

[spark tiger]

wdym that’s mobile app

[viscid grove]

oh
why did it say Safari

oh wait it's < Safari
so ig you opened youtube from safari?

[dusty moth]

yep

[spark tiger]

yeah

[jade stone]

this feels wrong

https://gist.github.com/sadan4/7cea927110e3c81fe9f44f2461e7d78f

[formal belfry]

@livid light what do you wish to do?

[livid light]

msg me its lil private

[small osprey]

Hemmliii

[civic flare]

this is not dating channel

[livid light]

That is not what's going on gang

[dense sand]

Which C library to draw simple text and pixels would yall recommend? raylib works, but is kinda uhhh

[shy mauve]

29 husks

O dyamn

That bro is hella cooked

[woven mesa]

gm

hiiii oomf

[winged mantle]

SDL exists

SDL is very popular

[jade stone]

GUHHHH

WHY DO DISCRIMINATED UNIONS LOSE INFRENCE AFTER DESTRUCTRING

[viscid grove]

i read that as a weird spelling of onions at first

[winged mantle]

you gotta stop discriminating

[valid jetty]

i love how you can do this

structs are genius

[fleet cedar]

Are you not quoting your strings when you debug-print them?

[valid jetty]

its just display not debug print

however no

technically they should be tho

[fleet cedar]

It looks like a debug print

[royal nymph]

why the hell are you storing it in an i32 pointer tho lmao

kinda misleading

[valid jetty]

specifically strings in formatting functions arent quoted

if you printed just a "string" it would be quoted

[valid jetty]

do you know what structs are

sorry that came out wrong

lmao

but basically when you allocate the memory for a struct you get back a pointer to the start of that segment, which is the first field

[royal nymph]

yes and it makes no sense to store a struct in an i32 pointer lol

[valid jetty]

in this case the first field is an i32

[royal nymph]

the Foo* in question

[valid jetty]

you do not get Foo* you get just Foo

[royal nymph]

why are u even adding pointer arithmetics 😭

unsane

[fleet cedar]

Does that mean type Foo is pointer sized?

[royal nymph]

why lol

[valid jetty]

because that was meant to be a joke 😭

obviously you woudlnt actually do that in a real codebase

yopu can just foo.a or foo.b

[fleet cedar]

Does type Foo mean a pointer to struct Foo?

[valid jetty]

but when you do this

use std/libc/io;

struct Foo {
    i32 a;
    string b;
};

fn main() {
    Foo (2)foo = (1) Foo { a = 1, b = "hi" };
    (6)io::cprintf("%d, %s\n", (4) foo.a, (5)foo.b);
}

the compiler does this essentially

export function w $main() {
@start
        ; (1) declare the foo struct
    %struct.Foo.1720 =l alloc8 12
    %offset.1721 =l add %struct.Foo.1720, 0
    storew 1, %offset.1721
    %offset.1723 =l add %struct.Foo.1720, 4
    storel $.1722, %offset.1723

        ; (2) store it as a stack variable (when you do Foo foo = Foo {} it puts that struct into `foo` thats what this is)
    %foo.addr.1724 =l alloc8 12
    storel %struct.Foo.1720, %foo.addr.1724

        ; (3) load the struct back from `foo` (intermediate step)
    %foo.1719 =l loadl %foo.addr.1724

        ; (4) load member a
    %offset.1726 =l add %foo.1719, 0
    %field.1727 =w loadw %offset.1726

        ; (5) load member b
    %offset.1728 =l add %foo.1719, 4
    %field.1729 =l loadl %offset.1728

        ; (6) print them
    %tmp.1730 =w call $printf(l $.1725, ..., w %field.1727, l %field.1729)
    ret 0
}

[valid jetty]

no it just means struct Foo

[fleet cedar]

In that case casting a i32* to a Foo is nonsensical

[valid jetty]

not really

the way Foo is made is that it allocates memory of its size right, and what you pass around is a pointer to the start of that allocation

well the first field is an i32 so that pointer would be an i32*

[fleet cedar]

Either a value of type Foo is a pointer, or it is not

[valid jetty]

it is a pointer internally, yes

but not a Foo *

specifically here

%struct.Foo.1720 =l alloc8 12
%offset.1721 =l add %struct.Foo.1720, 0
storew 1, %offset.1721
%offset.1723 =l add %struct.Foo.1720, 4
storel $.1722, %offset.1723

[fleet cedar]

So then #size(Foo) is the size of a pointer?

[valid jetty]

uhh it is the size of the struct

when you put a struct into another struct it is memcpy'ed into it

[royal nymph]

that makes no sense

[fleet cedar]

So it is not the size of a value of type Foo

[royal nymph]

you're saying Foo is a pointer, not a value, yet its size is not that of a pointer

[valid jetty]

yes

that is how structs work in C too

it is simply semantic that they are a value

[fleet cedar]

So a local variable of type Foo is a pointer but a struct field of type Foo is the raw data?

[valid jetty]

you would never directly work with that pointer though

the compiler would automatically calculate offsets for you

[valid jetty]

yep

because getting a field from a local variable Foo is deref(fooptr + offset)

[royal nymph]

no i dont think so lol

in C, structs are values

[valid jetty]

getting a field frm a field of foo is deref(otherstructptr + fooptroffset + offsetintofoo)

[valid jetty]

semantically yes

[fleet cedar]

Structs are passed by pointer in many ABIs

[valid jetty]

but if you look at the llvm ir they are not

i can actually show you this right now

[fleet cedar]

But the ABI is not exposed in language semantics

[valid jetty]

1 sec

^^^

[royal nymph]

struct whatever {
  int x;
};

struct whatever w1 = {0};
struct whatever w2 = w1;

w2.x = 42;

w1.x == w2.x; // false

it's a value not a pointer

[fleet cedar]

How values are passed in the ABI does not leak into the language

[valid jetty]

in c when you make a struct and pass it around it passes a pointer to the start of the struct, allocates new memory in the other function, and copies the original struct's memory into it

so that when you use it it feels like its a value

so in the example you showed when you set w2 to w1 it actually allocates new memory and copies w1 into it

[fleet cedar]

In most abis yes, because that's faster than copying the whole thing

[valid jetty]

^

[fleet cedar]

But that is not part of the language

[royal nymph]

it doesnt matter what it does under the hood lol

it only matters what it is in usage

and that's a value

[fleet cedar]

It does not allocate

It's on the stack

What happens under the hood is irrelevant

The language is the hood

[valid jetty]

stack allocation is still allocation

[fleet cedar]

What matters is language semantics; compilers can then write whatever asm they want as long as all observable behavior matches those semantics

[fleet cedar]

No

When you say allocate or allocation without qualifying, that means heap

[valid jetty]

and for the record you can do this thing in C too

just pointer arithmetic is done by element and structs are padded by default and stuff like that

[fleet cedar]

Yes but then you cast the pointer to a pointer

[valid jetty]

so its a little bit more annoying

[fleet cedar]

Which makes sense

[valid jetty]

true

i suppose that is something that i did wrong in the language on a semantic level

[valid jetty]

is this easy to read 🙏

im still kinda torn on whether <> is a good idea for concatenation

[fleet cedar]

<> for string concat
<> for monoid operator

[valid jetty]

TRUE

i was gonna overload + and make it just use concat if its a string and add if its a number

however,,

this is a low level language

[fleet cedar]

Js ptsd

[valid jetty]

sometimes if you wanna do pointer arithmetic manually for example

like if you wanna do str + 1 to get the 2nd char

[fleet cedar]

Kinda nasty, but legit

[valid jetty]

you would expect the b not abcd1

[fleet cedar]

How does memory management work with those strings?

[valid jetty]

well

thats essentially non existent lol

theyre null terminated

but any operation you do on them it creates a new string and leaks memory

[fleet cedar]

That's not excellent

[valid jetty]

its not

a todo i have is to make it sized

technically you can work with it as sized by turning it into an Array<char>

there are apis in place to go from 1 to the other

im also torn on whether this is a good idea

[fleet cedar]

But array<char> takes four times the memory

[valid jetty]

Array<T> is just size, cap, ptr, so 2 bytes

therefore 2 times the memory

not 4

[fleet cedar]

I mean for the actual content

"a" is one byte, ['a'] is one char so four bytes

[valid jetty]

how 4

wouldnt that be 3

2 bytes for the struct itself then 1 byte to store that char

[fleet cedar]

A char is 21 bits, rounded up to 32

[valid jetty]

even so, that fits in 1 byte

or o

no

wait

[fleet cedar]

A char does not fit in one byte unless you're racist

[valid jetty]

in the case of elle chars are 1 byte

[fleet cedar]

💩

[valid jetty]

also cstrings are not necessarily 1 byte thats just if you assume theyre utf8

(they are typically not)

[fleet cedar]

Okay sure, if you want the null terminator it'd be two bytes

[valid jetty]

either way you can do this idk

[valid jetty]

and anyway this doesnt hold as the string size grows

its just a constant term

[royal nymph]

broke: i live in the hood
woke: i live in the javascript

[royal nymph]

you can also do this in C

int main() {
  int x = 42;
  *((char**) &x) = "lol";
}

doesnt mean it makes any sense

[valid jetty]

that would segfault if you try to use it tho

in the c code i provided it actually results in a working Foo struct

[royal nymph]

no it actually wont

[valid jetty]

well

it would print the pointer to the start of that string truncated to 32 bits

still not anything useful

[fleet cedar]

What does from_string do with non-ascii?

[valid jetty]

what you would expect

[fleet cedar]

I expect complete bullshit, if char is a u8

[valid jetty]

it interprets each byte as a character and prints gibberish

yes

[fleet cedar]

Meaning the char type is complete bullshit

[valid jetty]

i will make a c_char type u8 and then make a normal char type which is a u32

so that unicode still works

[fleet cedar]

That sounds better

[valid jetty]

the reason its not always u32 is because sometimes you interact with padded structs from C libraries like raylib

if char was u32 it would calculate a completely wrong offset here

[fleet cedar]

Using char for numeric values is already nonsense

[valid jetty]

no, actually

[fleet cedar]

Yes please set my red value to {

[valid jetty]

in c people use char all the time to refer to 1 byte

because theres no explicit byte type

[fleet cedar]

uint8_t

[valid jetty]

true

but in the case of raylib it is just unsigned char iirc

[royal nymph]

the point is that you can do anything with pointer casts so "you can do this thing here actually" is a meaningless statement

[valid jetty]

well what i did with pointer casts is just what the c compiler does under the hood to populate a struct so its not always meaningless

[fleet cedar]

Except you're making it part of the hood

[valid jetty]

true,, but that was also kind of a joke originally,,,,

like youre not actually meant to do that to make a struct its just fun that you have the ability to do so if you wanted

[valid jetty]

i added 1 + 1,,,,

[fleet cedar]

A bit clunky, but looks legit

[autumn sigil]

zig is fun

parsing to u1234

[valid jetty]

what the fuck

[dense sand]

Anyone who knows spring security here?

[spark tiger]

how do u fuckin read c code. i wanted to look at this method implementation which call i found in the code and i found NOTHING. and this func should definitely be in this repo judging from ff (fastfetch) prefix

[fleet cedar]

Could be that the function name is spliced together by the preprocessor somewhere

Which is cursed

Or maybe that function actually doesn't exist and the code that calls it, which you're looking at here, is ifdeffed out

[spark tiger]

I’ve only found ffWinrtDetectMedia but uhhhh it’s missing another ff

also how do I make image rendering work in windows terminal

need it for fastfetch

[dense sand]

for loop through pixels and https://learn.microsoft.com/en-us/windows/win32/api/wingdi/nf-wingdi-setpixel

[spark tiger]

nah i mean from user perspective

[dense sand]

i dont quite get what you mean

[spark tiger]

in some terminal emulators (kitty, iterm2, wezterm, etc.) there is support for cli apps to render images in terminal. yazi docs page (https://yazi-rs.github.io/docs/image-preview) says windows terminal (>= v1.22.2702.0) has support for it but it doesn’t work for me for some reason

[spark tiger]

ok i cloned the repo and opened it in clion and ctrl-clicking the func call redirects me to line 9

fucking macros 😩

i dont even wanna know why they dont wrap ff inside double quotes

this language is cursed icl

[fallen nebula]

Macros w

[spark tiger]

😩

this might have worked ig??

[dense sand]

please is there anyone whos knowledged enough in multiboot2

[valid jetty]

try the next planet along

2.3 ly

[fallen nebula]

why not just use git bash ?

[fleet cedar]

Why not just use linux

[formal belfry]

who actually uses windows in 2024 tho

[still jolt]

a lot of people unfortunately ​​​​​​​​​​​​​​​​​​​​​​​​​​

[fallen nebula]

because it's actually usable
sure you can use linux, but it has more annoying problems than windows

[fleet cedar]

But most of the annoying problems can be solved

Unlike on windows

[fallen nebula]

eeeeeeeh

while i like using linux and i wish i could daily drive it, it's just too much of a pain because issues are too common
especially because stuff like nvidia

and also linux people are the champions to make 10 apps that do the same exact jobs except they all have flaws when they could unite and make 1 good software

[formal belfry]

in my experience linux is super reliable up until the point where you want to do something new

but i also did debian minimal and wiped windows cuz i'm a dumbass

not to mention is always heavily depends on distro

[fallen nebula]

i mean i daily drived linux during 1.5y and i'm still alive
and some of the stuff i do on windows is linux heavy
like i mostly only use git bash because bash > cmd/powershell
most of the software i use is FOSS
like i use Gimp, LibreOffice, Firefox, etc

[formal belfry]

how do i submit quickcss code stuff

[formal belfry]

@programmers

grr

ignore my stupidity lmao

i'm bored and tryna make vesktop look more like the DC mobile app

i got the channel list looking identical

[viscid grove]

me unfortunately :(
I'm lazy, busy, switching to linux would make me never get anything done because I would be endlessly customizing everything, and some things I use like sharex don't work on linux

[viscid grove]

in #🎨-css-snippets ?
open a css ticket in #📩-modmail

[formal belfry]

thanks

[willow knot]

clock :)

[formal belfry]

wait a sec that's more than 7 segments

[willow knot]

3x5 pixel font defined by a u15 (zig)

[viscid grove]

that's so hard to read
why are they numbered

[willow knot]

its a typographic clock - looks better when smaller / at a distance

[autumn sigil]

now make it a quine

[willow knot]

holy shit thats actually a great idea

[celest lodge]

in my experience it just werks 😶‍🌫️

[serene narwhal]

How do I get accsss to #🧩-plugin-development

[cerulean plover]

ask for permission through #📩-modmail

[serene narwhal]

their is no option for accesss too plugin channel

[cerulean plover]

pick the "i need to talk to a moderator" option

[serene narwhal]

okay

[autumn sigil]

skill issue

[valid jetty]

i added object files to elle

[formal belfry]

what should I learn for typescript with making vencord plugins

:P

[autumn sigil]

what is @volatile, the returntype?

[valid jetty]

no

@volatile is an attribute that basically says “don’t perform dead code elimination on this function and don’t optimize it, please just leave it alone” to the compiler

without that the function won’t be generated in the object because it’s not used anywhere

i did make that the default for generating object files specifically since that screenshot

when you pass -c:

• all public functions are exported
• it no longer expects an entry point
• the implicit string module methods are removed right before the IR is generated
• it generates .o instead of an executable by default
• creating generic functions or structs throws an error

[valid jetty]

static libraries are real now too i guess (i fixed linker flags)

this may be cleaner idk

i got rid of the -Dtime and -Clink-flag stuff because that was ugly

[dawn dragon]

@valid jetty

[valid jetty]

hi

[celest lodge]

waow,,

[autumn sigil]

ret is inferred then?

[valid jetty]

yes

if you return from multiple places it will throw an error if they dont all match

you can also specify an explicit return type

for recursive functions you kinda have to

otherwise the compiler wont be able to infer it

[autumn sigil]

blobcatcozy

[winged mantle]

does this support tailrec

[valid jetty]

technically yeah sure

it wont optimize it for you but you can do it

use std/io;

fn sum(i32 x, i32 total) -> i32 {
    if x == 0 {
        return total;
    } else {
        return sum(x - 1, total + x);
    }
}

fn main() {
    io::dbg(sum(12, 0), sum(8, 0));
}

the traditional way

use std/io;

fn sum(i32 x) -> i32 {
    if x == 0 {
        return 0;
    } else {
        return x + sum(x - 1);
    }
}

fn main() {
    io::dbg(sum(12, 0), sum(8, 0));
}

note that elle has no default parameters so you must pass 0 as the initial total

i can actually even

use std/io;

fn tail_call_sum(i32 x, i32 total) -> i32 {
    if x == 0 {
        return total;
    } else {
        return tail_call_sum(x - 1, total + x);
    }
}

fn traditional_sum(i32 x) -> i32 {
    if x == 0 {
        return 0;
    } else {
        return x + traditional_sum(x - 1);
    }
}

fn main() {
    io::assert(tail_call_sum(12, 0) == traditional_sum(12), nil);
    io::assert(tail_call_sum(5, 0) == traditional_sum(5), nil);

    io::println("The test has passed!".color("green").reset());
}

[ruby shuttle]

ew rust

[valid jetty]

its not rust

[sick swift]

.

[winged mantle]

I mean converting to a loop

otherwise it will stackoverflow

[valid jetty]

no but that is technically easy to do

i just need to convert function calls into a jmp to the start label and replace the parameters

[errant mesa]

Hello everyone!

I'm a happy coder

[viscid grove]

wow

[fleet cedar]

Make your lies believable next time

[valid jetty]

maple mono

[royal nymph]

it wont optimize it for
so it doesnt have it xd

https://cdn.discordapp.com/attachments/1038550076534042726/1305055557463314442/guh-cat.gif

silly banana

[winged mantle]

is it insane to edit nautlius source code for personal use to change it to use iec

[fleet cedar]

That's pretty much the whole point of open source

[winged mantle]

is there a way to automate updates for a personal patch

it'd be cool if i could have automatic updates in general

[pseudo sierra]

there are a few ways most depend on what package manager you use

for arch you can use .patch files with PKGBUILDs

gentoo and nix also have a way to patch stuff but it's a bit more complicated

apt rpm etc idek

[winged mantle]

dnf

whatever imagine a cve in a file manager

would never happen

wait

i could use hooks

modifying os in the same way as video game haks

[autumn sigil]

idk how to stream decompress discord ws zstd

[ornate quiver]

gzip still works fine too

[autumn sigil]

oh yeah i was using zlib but i wanna switch ^^

[winged mantle]

i wonder why this doesn't work....

[autumn sigil]

it doesnt look like theyre doing anything special to the buffers but ofc i dont have access to their node module src

wonder if they append some stuff

? it wants to read 3 bytes but there arent any more

[ornate quiver]

decompile the android app
the decompressor is in java and isn't obfuscated

[autumn sigil]

im not decompiling the mobile app

thank you :3

[ornate quiver]

here's the patch of how i implemented in cadence

its kotlin but whatever

gu

[autumn sigil]

@frosty obsidian when will you give me repo access

hm

[royal nymph]

you will write a zstd inflater from scratch

[autumn sigil]

might be fine if you dont impl dicts

[frosty obsidian]

never

[autumn sigil]

why :c

[frosty obsidian]

private

[autumn sigil]

youre mean

[autumn sigil]

it looks like the single block i get doesnt have the last block flag set

so it tries to read more block headers and ofc it fails

now what do i do?

does that mean discords serverside impl is faulty?

it looks like i can reasonably expect a frame to have more than a single block so i cant just set the flag for every block..

[autumn sigil]

setting the flag does spit out valid etf so

wait nvm ofc it doesnt have the next block, its a stream

[dense sand]

okay so i just finished the end of evangelion, what the actual fuck was that

wrobng chane

l

[autumn sigil]

are you sure this is the right channel?

[dense sand]

mb

[autumn sigil]

xd

[dense sand]

even wrong server huh

[lost pawn]

I love when I can't create a custom hook with svelte's runes

[hazy pine]

i beat minos prime after so long

i think i finally might be pro gaming

[dawn ledge]

hate building from source
no patches in my house

[pseudo sierra]

hop on unstable-small

[dawn ledge]

[winged mantle]

ended up using LD_PRELOAD

don't ask why i need so many minecraft launchers

[pseudo sierra]

what did you even do?

[fleet cedar]

LD_PRELOAD my beloved

[winged mantle]

replace the formatting function

[pseudo sierra]

I see

[fleet cedar]

What formatting? Size? Date? Filename?

[winged mantle]

size

GB -> GiB

[crisp harness]

Is there a better way to see the lifetime of my code without having to run npm build and pnpm inject?

[autumn sigil]

kp was du willst, sag mal auf deutsch

[crisp harness]

haha

ja ob ich immer discord neustarten muss

oder ob es nen dev build gibt. der live updatet oder sowas

kp

[median root]

is "was du wilst" just what you will?

and does that mean wilst is "will"?

[fleet cedar]

Just run pnpm watch

No need to reinject, it's already inside

[crisp harness]

Is it possible to create a fullscreen modal? So that you only see the servers on the left and I can do everything myself on the right. Like the SHOP tab, for example.

[fleet cedar]

Check how the shop tab does it

[solid gazelle]

is this approach good

let fecha = null;
let faltas = [];
for (let elem of doc.querySelector("tbody").querySelectorAll("tr")) {
    if (elem.children[0].tagName == "TH") {
        fecha = elem.children[0].innerText.trim();
        continue;
    }
    faltas.push({
        fecha,
        ...[
            'hora', 
            'materia', 
            'tipo', 
            'motivo', 
            'comentario'
        ].reduce((a,b,i)=>(a[b]=elem.children[i].innerText.trim(),a), {})
    })
};

[signal oak]

It's functional but can be improved. Directly accessing child elements by index (elem.children[i]) is fragile. Use more descriptive selectors or a mapping to improve readability and robustness. Error handling (e.g., if a tr lacks expected children) is missing.

Message URL

#🪅-progaming message

Question

explain

[crisp harness]

Where can you look that up?

[fleet cedar]

In the code

Same way as literally anything else

[median root]

Use the search function in the dom

[valid jetty]

holy shit

[royal nymph]

@lone helm @ornate quiver WTF whatsapp web started using metro

cursed

so it's react native web ?

@calm ruin

[calm ruin]

probably

[royal nymph]

lol yeah it is

you can tell if it's react native web by checking css classes

react native web makes individual classes for every single css property

[calm ruin]

webpack was fine 😢

why did they switch to metro 🤢

[royal nymph]

cause react native uses metro xd

[calm ruin]

worse

[royal nymph]

facebook is so insane

[calm ruin]

I think twitter also uses react native

yet they use webpack

[lone helm]

yeah

[lone helm]

yeah what the hell xD

looks like it

[royal nymph]

they've been using it for ages

maybe metro didn't exist back then

[autumn sigil]

always has been

strencher tried to make a mod many years ago

[frosty obsidian]

not on android

wait i didn't backread enough

web is just regular react iirc

[royal nymph]

nope

react native web

[frosty obsidian]

how can you tell

nvm i looked back literally 10 messages

[royal nymph]

yea xd

css classes

and because it uses react native components like View and Text

[frosty obsidian]

is the ios app react native or something

bc otherwise i don't know why they would use that over regular react

[ornate quiver]

insane

react native Web needs to die

[spark tiger]

what's so bad about it

[nimble bone]

next time stop at react

react needs to die

[ornate quiver]

it turns an already bad platform (react native) into somewhere it shouldn't run (web)

sooooo husk

[ornate quiver]

.

[royal nymph]

it makes sense tho

the entire point of react native is write code once use everywhere

but in reality it doesn't really achieve that because normal react and react native use very different components and styling approach so you still have to write two separate ui implements

so the logical conclusion is to make react native run on web so you truly only have one single codebase

[ornate quiver]

so bad

just because it's cost efficient doesn't man it's good

[still jolt]

electron apps

[frosty obsidian]

companies don't care about good

they care about profit

[ornate quiver]

i agree

[dawn ledge]

enshittification of the web was spearheaded by them

return to jquery

[dense sand]

i just tried astro and i absolutely hate it

please add usetstate to astro 🙏

[dawn ledge]

whar

astro is supposed to be static rendering

there's no client side interaction

you have to byof

[dense sand]

bruh i want to have 2 buttons which toggle rendering of 2 different components

imma just do it on a different page

[dawn ledge]

youre solving the wrong problem with the wrong tool

normally i would have agreed and trashed on astro (or any other js framework fwiw)
but this is a you issue

[dense sand]

its just a simple utility site, what other thing would you recommend? i didnt want to generate bloated nextjs project for such a simple site

its literally supposed to be 2 forms

lol

[dawn ledge]

perhaps write a simple js snippet then?

if it is a simple site and youre looking for usestate somethings wrong

[ionic lake]

if you want a simple store theres nanostores

[autumn sigil]

this guy has never heard of inline js..

[dense sand]

elaborate

[autumn sigil]

https://docs.astro.build/en/guides/client-side-scripts/#handle-onclick-and-other-events

you just put css display none on the forms and toggle it with event listener

[autumn sigil]

idk why i called it inline js

i just meant you need to get used to writing simple code if you want to use pure astro

[dawn ledge]

and fwiw, its perfectly valid to have custom props on html elements
you can opt for data-foo pattern (and access it in js with elem.dataset.foo) or just elem.foo = thing

[dense sand]

yep, already discovered that, but it will always be converted to string >:( so cant pass down functions

[dawn ledge]

then do the latter

(which supports functions, well any type really)

[royal nymph]

that's like saying your microwave isn't good because it can't cool food and they should add a fridge to it

if you want reactivity, use svelte

[dense sand]

it was a joke cmon 😭

[royal nymph]

https://docs.astro.build/en/guides/integrations-guide/svelte/

[dense sand]

altough it mightve sound pretty seriously given the shit i wrote afterwards

[royal nymph]

astro is build framework + ssr adapter + static site generator

it is not an ui framework

that's why for anything with lots of reactivity you combine astro with a ui framework like svelte

[frosty obsidian]

astro is really great if all you want is reusable components

[dense sand]

youre true, seems like i picked wrong tool for what i want, i guess ill keep using react