🪅-progaming

[fleet cedar]

I don't like crate::parser::parser

[hoary sluice]

oh yeah idk why i dont have pub use for it

[winged mantle]

wait i am probably wrong

OH yeah

i am right

[hoary sluice]

but like is it preferred to do pub use parser::Parser in mod.rs or put the contents of parser.rs in mod.rs

[winged mantle]

you need to use pub mod to export the sibling modules

[deep mulch]

@hoary sluice 🦅ly

[hoary sluice]

yes but i dont know if mod.rs is meant to only house exports or if it can house the entire module

or a big part of the module

[fleet cedar]

It houses the contents of the module

[winged mantle]

sorry i was just getting confused

[fleet cedar]

What the module contains is up to you

[winged mantle]

do you mean conventionally

[hoary sluice]

yes

[winged mantle]

you can have structs there i am pretty sure

and fns

[hoary sluice]

that doesnt help me, should i put my parser into mod.rs or keep it in parser.rs

[fleet cedar]

Search your feelings

[hoary sluice]

my feelings are keep it in parser.rs but clippy whines about that so i was wondering if its convention to put it in mod.rs

[fleet cedar]

What does it say?

[hoary sluice]

https://rust-lang.github.io/rust-clippy/master/index.html#module_inception

tho i dont have mod foo { .. } twice

[fleet cedar]

You have a mod parser; inside a mod parser;

Which is indeed questionable

But if you want it that way, simply #[expect()] it

[blazing haven]

https://30fps.net/pages/palette-lighting-tricks-n64/

this is cool

[dense sand]

do yall know about any good website which would teach math for game dev? like i know what is matrix but how does relate to some position or shit in games, no clue

[spring schooner]

I don't really know of any (neither I have a ton of experience with matrixes), but you can view them as a generic tool for problem solving. Both CPUs and even more GPUs can do matrix calculations very fast, and in some cases this gives you a great advantage. An example is simple image filters, you can define a matrix that maps ARGB in to ARGB out for each pixel, and then any of those simple color filters can be applied using the same code instead of reimplementing the logic for each of them. At first it sounds like waste of computing power, but even for something as simple as a grayscale filter it's much faster

[fleet cedar]

In graphics they're mostly used for converting between coordinate spaces

[valid jetty]

@hoary sluice look at how easy it is to add new functions with C interop

[valid jetty]

lmao i just went down a massive rabbit hole

this is real debugging done by real software developers

[hoary sluice]

i have rust interop

(one way)

[hoary sluice]

whats that

[valid jetty]

@hoary sluice LOOK AT THIS https://gist.github.com/acquitelol/787ce7af581c87523e355c1ab714c2b5

[hoary sluice]

why are you screaming at me 😢

[valid jetty]

3:

qbe is being annoying

[hoary sluice]

shouldve compiled to haskell

fac n = if n == 1 then 1 else fac (n-1) * n
fac 50

icypeas to haskell compiler for factorial functions in one line of python

for line in open(0): print(line, end='')

get it because its also valid haskell

[valid jetty]

😭

are your things lazy

[hoary sluice]

idk

i didnt explicitly implement it like that

it might be lazy by accident

i didnt implement multi arg call parsing and i have working multi arg parsing

so maybe the lazy evaluation also just happens

[valid jetty]

if i do

loopForever = loopForever
f x y = x
f 42 (loopForever)
``` do i get 42 or a hang

[hoary sluice]

you get 42

well you get expected expression but thats because functions require an argument

[valid jetty]

oh

ok well what if i do

loopForever a = loopForever a
f x y = x
f 42 (loopForever 0)

[hoary sluice]

stack overflow

however if you dont pass the 0 it for some reason compiles

well

interprets

idk why

oh yea cause loopForever by itself is a function so you just pass the function to f and it gets ignored

ok well ig tomorrow i make lazy eval

its not hard

[valid jetty]

lmao

YES it works on linux now

[hoary sluice]

does it work on windows

i hope not

[valid jetty]

no ofc not

all of the stdlib is built on libc

[hoary sluice]

good

@valid jetty pov: you spent 10 minutes thinking of a commit message and came up with this https://github.com/eagely/icypeas/commit/6eb5af70413f25d1094624d85f86c569bc6c740d

had to look up if scopes was the correct word for this but then i remembered that i dont actually care

[valid jetty]

lmaooo

[hoary sluice]

@valid jetty when do you bump version

why here https://github.com/acquitelol/elle/commit/58871d678c6200b54a4a06e01b2ac4f47c6e3aae

[valid jetty]

whenever theres a potential breaking change with past versions of the compiler

or, if i forget, whenever theres a refactoring commit

as im not in 1.0 yet i can bump minor for breaking changes

if the change doesnt break past versions of the compiler but the change is significant enough i will bump patch

@hoary sluice look at how many import duplications im saving

each Module '...' is already imported. is an early return

[hoary sluice]

im saving a lot more duplicate imports by not having imports

[deep mulch]

EVN THE FILE NAME 😭😭😭😭😭😭

WHY

[fleet cedar]

Mincho font in terminal

[valid jetty]

https://www.youtube.com/watch?v=z7cfF7O7pGg WHO HAS TIME FOR THIS

[frosty obsidian]

balala

[valid jetty]

yeah

[deep mulch]

balala

[valid jetty]

@hoary sluice do i allow constants to be defined on enums

on one hand

on the other hand

this isnt ambiguous

you can define functions on enums but the syntax is a little nicer for lambdas

for example

// なんちゃって型エイリアス
enum rune @repr(i32) @nofmt {}
const rune::__fmt__ = fn(rune self, i32 nesting) self.encode();

instead of

// なんちゃって型エイリアス
enum rune @repr(i32) @nofmt {}
fn rune::__fmt__(rune self, i32 nesting) { return self.encode(); }

its as simple as 1 comment

i can technically lie and do this which is honestly funnier

// なんちゃって型エイリアス
enum rune @repr(i32) @nofmt {}
external fn rune::encode(rune self) @alias(rune::__fmt__) -> string;

😭
$dbg("猫".utf8_decode());

[hoary sluice]

@valid jetty this is such an insightful an interesting video especially towards the end https://youtu.be/Zv-XNlE1s8E

nah i just got to 17:30 and it gets crazy

[royal nymph]

rosie moment

[native spruce]

@woven mesa idk how im gonna do this but I wanna add a swipe gesture to my app that reveals a sidebar

sidebar would contain settings

[woven mesa]

meret also gonna do that too

[native spruce]

yk how the grok app does it

well just for chats

that but uikit

idk

[native spruce]

I kinda want

like

the whole app to shift right

would be neat

okay I should make the controller but with certain constraints so it matches a sidebar

overengineering ui my beloved

[valid jetty]

[deep mulch]

@valid jetty @valid jetty @valid jetty @valid jetty

[valid jetty]

@deep mulch have you made any progress

[median root]

yo W rice bro just add perkeo to dupe windows and its perfect

rofi pack theme too

[deep mulch]

Im at work rn

can't until later

[woven mesa]

@native spruce marvis does it so the gesture is continuous (bad demonstration bc i was eating)

you drag to open and then move finger up or down whilst continually holding

let go to open selection

[native spruce]

yummy

[valid jetty]

"add bowls and domes, and subtract saddles"

V - E + F = 2 MENTION???

why does she sound like furina

what the hell is this 😭

LMAO

WHAT THE FUCK AM I WATCHING

HOW DOES IT JUST GET WORSE AND WORSE

[deep mulch]

how to turn a sphere inside out

[valid jetty]

OH MY GOD

[deep mulch]

you should do it

@valid jetty @valid jetty @valid jetty @valid jetty @valid jetty

[valid jetty]

watch the full video

i dare you

[deep mulch]

I think I have a long time ago

@valid jetty you're so smart you probably turn spheres inside out in your mind

you can imagine every single step with visualization

@valid jetty probably solved some unsolved thousand year math problem

[lavish cloud]

@runic sundial do you also happen to know how to assimp? If I load all the meshes directly from scene.mMeshes they don't get the right materials, but for some reason building a scene tree from scene.rootNode down doesn't render anything at all for some of them

[valid jetty]

@hoary sluice @deep mulch do this https://www.youtube.com/shorts/ehv-Ls9Idw8

[runic sundial]

what format?

[lavish cloud]

wavefront atm

[runic sundial]

how are the materials annotated in the obj?

[valid jetty]

i did it in probably the simplest way possible in like 3 mins

use std/prelude;

struct Node {
    Node *left,
    Node *right,
    i32 value
}

fn Node::print_on_sep_lines(Node *self, i32[][] levels, i32 nesting) {
    while levels.len() < nesting + 1 {
        levels.push([i32;]);
    }

    levels[nesting].push(self.value);

    if left := self.left {
        left.print_on_sep_lines(levels, nesting + 1);
    }

    if right := self.right {
        right.print_on_sep_lines(levels, nesting + 1);
    }

    return levels.map<string>(fn(i32[] level) level.join(" ")).join("\n");
}

fn main() {
    root := Node {
        left = &Node {
            left = &Node {
                left = nil,
                right = nil,
                value = 4
            },
            right = nil,
            value = 2
        },
        right = &Node {
            left = &Node {
                left = nil,
                right = nil,
                value = 5
            },
            right = &Node {
                left = nil,
                right = nil,
                value = 6
            },
            value = 3
        },
        value = 1
    };

    $println(root.print_on_sep_lines([i32[];], 0));
}
``` surely its not *that* hard

[lavish cloud]

I know the obj is correct cuz online viewers handle it just fine

[lavish cloud]

mtllib bmw.mtl

[runic sundial]

How are you loading the file into assimp?

[lavish cloud]

    fun loadScene(path: String, target: GameObject) {
        val flags = Assimp.aiProcess_Triangulate or
                Assimp.aiProcess_GenSmoothNormals or
                Assimp.aiProcess_FlipUVs or
                Assimp.aiProcess_CalcTangentSpace or
                Assimp.aiProcess_JoinIdenticalVertices or
                Assimp.aiProcess_FindInvalidData

        val io = makeIO()
        val scene = Assimp.aiImportFileEx(path, flags, io)  ?: throw RuntimeException("Unable to load scene: ${Assimp.aiGetErrorString()}")
        println("Imported scene: $scene")
        io.OpenProc().close()
        io.CloseProc().close()

        // attempt at using the tree representation
        loadObject(scene.mRootNode()!!, scene, target)

        // Everything renders, but materials don't load 
//        if (scene.mNumMeshes() == 1) {
//            val meshRef = AIMesh.create(scene.mMeshes()!![0])
//
//            loadMesh(meshRef, scene, target)
//        } else {
//            for (meshIdx in 0 until scene.mNumMeshes()) {
//                val meshRef = AIMesh.create(scene.mMeshes()!![meshIdx])
//                val meshName = meshRef.mName().dataString()
//                val child = GameObject(meshName)
//                target.addChild(child)
//
//                loadMesh(meshRef, scene, child)
//            }
//        }

        scene.close()
    }

but the scene tree is also weird because a lot of objects are in there multiple times

sometimes standalone, sometimes as child of another part

[runic sundial]

I'm unsure if scene.rootNode would even work right with an obj

[lavish cloud]

private fun loadObject(node: AINode, scene: AIScene, target: GameObject) {
    println("${node.mName().dataString()} => [${node.mNumChildren()}, ${node.mNumMeshes()}]")

    if (node.mNumMeshes() == 1) {
        val meshRef = AIMesh.create(scene.mMeshes()!![0])
        loadMesh(meshRef, scene, target, null)
    } else {
        for (meshIdx in 0 until node.mNumMeshes()) {
            val meshRef = AIMesh.create(scene.mMeshes()!![meshIdx])
            val meshName = meshRef.mName().dataString()
            val child = GameObject(meshName)
            target.addChild(child)

            println("${node.mName().dataString()} => mesh $meshName")
            loadMesh(meshRef, scene, child, null)
        }
    }

    println("${node.mName().dataString()} => End meshes")

    for (childIdx in 0 until node.mNumChildren()) {
        val childRef = AINode.create(node.mChildren()!![childIdx])

        if (childRef.mNumMeshes() == 0 && childRef.mNumChildren() == 0) continue

        val childName = childRef.mName().dataString()
        val child = GameObject(childName)
        target.addChild(child)

        println("${node.mName().dataString()} => child $childName")
        loadObject(childRef, scene, child)
    }

    println("${node.mName().dataString()} => End children")
}

[runic sundial]

I believe it's more so for gltf/usd

[lavish cloud]

any suggestions on what to do then?

[runic sundial]

not too experienced with assimp, but I would validate that it even finds the mtl

assimp has an error message out iirc, might try malforming the mtl and seeing if it logs an error or not

Just to validate that it can find it

[lavish cloud]

I mean it can find it if I go through the scene root

just not if I loop over the meshes

[runic sundial]

Might be just how assimp is structuring the data internally, check their docs

or the rust docs of the assimp library lmaoo

Personally? I wouldn't bother with *.obj past simple demo meshes.

You should just make the jump straight into something like gltf or another more advanced format that assimp supports.

Obj is only really worthwhile if you need to "load a mesh quick and ez", past that it really sucks.

mostly because dealing with sub-meshes, materials, and additional vertex data with objs is a huge pain

[fleet cedar]

Obj is maybe easy, but it's definitely not quick

[runic sundial]

?

I mean quick to implement.

Not quick as in performant

[fleet cedar]

You'll want a binary format for quick

[runic sundial]

Define quick

You can make like a 100LoC Obj parser in nearly any lang from lua to c++

That's what I define as "quick and easy"

But yeah, binary *.glb/gltf files are way faster to load huge meshes with.

[fleet cedar]

And you can parse most formats in one line of cargo add

[runic sundial]

Didn't know cargo worked with Kotlin, will need to look into that

Most other formats also have "scenes" and some have pre-packed buffers you may need to decode or hope the library has utilities to do that with.

While objs are just verts and faces in a plain text file, easier for novices to deal with.

[deep mulch]

h

[runic sundial]

Obj as bootstrap, gltf as real deal

[runic sundial]

https://media.discordapp.net/attachments/850410310559989805/1368934204192329739/brandbird_2.gif

[formal belfry]

@woven mesa does @Observable have an equivalent to ObservableObject().objectWillChange

[woven mesa]

no

[formal belfry]

https://tenor.com/view/yandere-tagged-knife-anime-gif-16524623

[woven mesa]

it works differently

no combine

(combine isnt on linux

@formal belfry use withObservationTracking

[formal belfry]

i couldnt figure out how that works

im just gonna get this working with combine first

[woven mesa]

you have to create a new observation trackign thingy after the property changes and the callback fires

[woven mesa]

if you end up stickign with combine ig use opencombine

idk what u working on tho

[formal belfry]

xrpc 2 electric boogaloo

[woven mesa]

o

does that mean i have to review a massive pull request

i will be prepared

[formal belfry]

i dont know if i even should pr this because its an entire rewrite

https://tenor.com/view/pipkin-pippa-pippa-pipkin-vtuber-cringe-gif-13876806171127454285

[valid jetty]

finally

[woven mesa]

@formal belfry https://x.com/kuroijart/status/1923797017530372307?s=46

KuroiJ (@KuroiJArt) on X

🍈Shadow🍈

#ZenlessZoneZero #绝区零 #kuroij #zzz #zzzero

X · May 17, 2025, 5:45 PM

[formal belfry]

I am being trolled so hard right now

this is so close to working but rpc just doesnt show up

@woven mesa fix

this code is so clean too bad it doesnt work

[valid jetty]

YESSS @hoary sluice its so real

theyre printed in the right order

[formal belfry]

I made a swordrpc wrapper months ago and it still holds up its so insanely clean

pohm y godasvbdhsda

[formal belfry]

huge okay

now time to want to die by simply using AXSwift

@woven mesa@woven mesa@woven mesa@woven mesa@woven mesa@woven mesa

[shrewd canopy]

I think i might need to make RPC client soon

[deep mulch]

what did you do

[valid jetty]

qbe types have to be written in a specific order to maintain C abi compatibility

[valid jetty]

^^^

Box<Option<Node>> holds an Option<Node> *, so as its a pointer its just l (a 64 bit int) and doesnt depend on anything

but Node depends on Box<Option<Node>> and Option<Node> depends on Node

so i had to write this thing to print them in the right order

(they wont be in the right order by default, the compiler doesnt guarantee that)

the result is this https://github.com/acquitelol/elle/blob/rewrite/tests/manual/tree.le

generic recursive structures

with pretty-printed types and stuff

[fleet cedar]

Why Box<Option<Node>> rather than Option<Box<Node>>

[valid jetty]

true actually

yeah that actually simplifies it quite a lot

because it can infer T for Option::None here now

works

absolute explosions

[deep mulch]

terrifying

@valid jetty now sleep after a long hard day of Elle work

Rosie would rather die than sleep

[pseudo sierra]

tbh mood

[valid jetty]

it was not a hard day

i just messed around and things worked out

who needs sleep anyway

[deep mulch]

you are not human I think

[woven mesa]

so true

[digital plinth]

Yo, any good way of fetching all users from servers to compare? Aka find overlap.

[fleet cedar]

Discord intentionally makes it hard to get all members, to hinder spambots

[digital plinth]

rip

[shrewd canopy]

if you have ManageRoles/BanMembers/KickMembers permissions then you can send OP 8 (requestGuildMembers) with query:"" and limit:0 and you'll start getting GUILD_MEMBERS_CHUNK events, containing all guild members

otherwise no

[digital plinth]

yep. It is ripperoni

[median root]

If I have a user inputting an api key somewhere, what is the best way to store it so that you cant just access it anywhere and you need some kinda passwd? Like even if someone has my computer they cant just go onto the app and grab the key without a password or run around my file system to get it.

[lavish frigate]

store it in a file called IMPORTANT_API_KEY_DO_NOT_OPEN.txt on the users desktop

[fleet cedar]

Store it on the system keyring

[median root]

ah yes the good old honor system

[median root]

does linux have one? and does the keyring do all the actual encryption stuff itself?

so it asks for passwords whenever it is read from too?

ive only seen it used a few times on macos

[fleet cedar]

Linux itself does not, but both gnome and kde do

[median root]

ah lovely so what do i do about the linux people

[fleet cedar]

Check what other programs do

I've had to set an envvar to get python poetry to not crash from not having a keyring, so that probably supports it

[median root]

oh yippie more rust dependencies

just what I wanted in life

me when I get a new blazingly fast 🚀 memory safe 🦺 library for my birthday

I might just do the obvious option and encrypt the api key making use of the password hash from the user

and just store that encrypted thing in a file somewhere in sharedprefs

This looks pretty promising

AES-GCM

now to find a library that supports it

[valid jetty]

@hoary sluice i added tuple destructuring !!!!!!!!!!!!!!

a bfs hasnt looked this ergonomic in years

fn Node::print_on_lines(Node *self) {
    i32[][] levels = [];
    queue := [$(self, 0)];

    while !queue.is_empty() {
        node, level := queue.remove(0).unwrap();

        if levels.len() == level {
            levels.push([i32;]);
        }

        levels[level].push(node.value);

        if _, left := node.left {
            queue.push($(left.to_ptr(), level + 1));
        }

        if _, right := node.right {
            queue.push($(right.to_ptr(), level + 1));
        }
    }

    return levels
        .map<string>(fn(i32[] level) level.join(" "))
        .join("\n");
}

but yeah basically you can do this

the variables are assigned to x, y, z in the RHS, and its also an expression which evaluates to the LHS

for Option<T> and Result<T, E> the LHS is is_ok or is_some

which means you can write this

while _, current := lexer.next_token() {
    tokens.push(current);
}

[lavish cloud]

https://light-theme-hurts.my-ey.es/2pnb9xR.mp4
I figured out the issue
apparently despite opengl remembering everything all the time
it doesn't remember shader uniforms
so if I glUseProgram(a) and set uniforms, then go glUseProgram(b) and set uniforms, going back to glUseProgram(a) keeps the uniforms set on b

Now I just need to figure out why it's stuttering and why that weird artifacting happens on the side of the car

[supple whale]

why do i recoginise those menus

and why cant i remember what its from

[valid jetty]

imgui

lol

[supple whale]

sounds about right

its jsuch a genericly used lib everywhere

i think i first saw it for unity mods, then probs rtx remix the most often

its a workhorse and a half lol

[chilly viper]

ok uh

stupid question but

how did the person who made this put 2000 images of cheese into a slideshow

no way you do that by hand

[fleet cedar]

There's billions of ways to do that

• script the mouse/keyboard interactions
• generate a powerpoint file with a script
• generate the video with a script without going through a slideshow

But however they did it they're a hero

[chilly viper]

can you show me some example or something

[fleet cedar]

No

[chilly viper]

ok

[fallen nebula]

Me when ffmpeg + bash

[placid horizon]

is there a way to prevent Discord from automatically unmuting my mic whenever i join a VC?

[winged mantle]

i think it's very possible to hand select 2000 images of cheese as a long term project

if you picked 10 cheese a day it would take less than a year

I suppose ordering the cheese would be hard

wait

there's no way this video has 2000 cheese

you would need to show 10 cheeses a second for it to be 200 seconds

[pseudo sierra]

no that's 200 minutes

[winged mantle]

oh

i might be stupid

[spark tiger]

huh i don’t think it does that for me

[hoary sluice]

be honest yes it has

however now u can actually do aoc purely in elle

@valid jetty is elle lazy evaluated

[valid jetty]

nop

[valid jetty]

i know i was joking

[hoary sluice]

smh

[valid jetty]

it doesnt need to be ..???

its procedural

short circuiting things like && and || are lazy

or ?:

[hoary sluice]

but does it hurt

[valid jetty]

i mean not really other than to implement

[hoary sluice]

like is there ever a scenario where youd rather evaluate strictly

[valid jetty]

i mean no but the implementation of it is pointless

[fleet cedar]

It makes implementation harder, costs more memory, costs a branch on every variable access, and makes it harder to reason about performance

[lavish cloud]

@hoary sluice you love
(the model also gets a convex hull assigned)

All in the kotlinx

[fleet cedar]

Oh and it gets weird with side effects

[valid jetty]

this is true actually

i didnt think about that

[hoary sluice]

im surprised it runs faster than 1 fps

[winged mantle]

https://nodejs.org/en/learn/test-runner/using-test-runner why does it feel like this explains nothing

[valid jetty]

why are recursive generic structures so fucking painful

like a ```rs
struct Node<T> {
Option<Box<Node<T>>> left,
Option<Box<Node<T>>> right,
T value
}

monomorphization for this is a nightmare

[fleet cedar]

How so?

[valid jetty]

because i have to define what Node<T> is before i know what Node<T> is

[dense sand]

make it Node<JS> please

[fleet cedar]

In rust that is solved by Box<T> being the size regardless of T (at least of T: Sized)

[lavish cloud]

Wdym it's just a gl progm

[valid jetty]

the problem isnt the size, that works fine

i need to generate this struct in 2 passes

first pass, just collect the fields, so that I know what fields exist
second pass, partially monomorphize what I can

[lavish cloud]

I had to put 90% of my code in a Box<T> and even then it didn't allow for some stuff that needed generics

[valid jetty]

but the problem is that, when trying to monomorphize Node<T>, it reaches the definition for Node<T> which is exactly the same as the current thing being monomorphized

so it just infinitely monomorphizes Node<T> causing a stack overflow

[fleet cedar]

How is that case different from non-generic recursive types?

[lavish cloud]

iirc it was something like

DataColumn<I, V> {
    withIndex(index: Index<A>): DataColumn<A, V>
}```

[valid jetty]

because non-generic types dont need to be partially monomorphized

a ```rs
struct Node {
Option<Box<Node>> left,
Option<Box<Node>> right,
i32 value
}

[fleet cedar]

I don't see why this would be a problem

[valid jetty]

while a ```rs
struct Node<T> {
Option<Box<Node<T>>> left,
Option<Box<Node<T>>> right,
T value
}

[valid jetty]

its technically not if i just limit the monomorphization depth to 1

[fleet cedar]

What does partial monomorphization mean

[valid jetty]

if you have a ```rs
struct Bar<T> {
T x
}

struct Foo<T> {
Bar<T> x
}
``` when generating Foo<T>, you need to monomorphize Bar<T> with T, but you dont actually know what T is yet, so you have to partially monomorphize it into a non-generic structure that includes the generic T, that is then fixed up later when monomorphizing Foo<T> with some T

[fleet cedar]

I don't understand the "you need to monomorphize Bar<T>" part

[valid jetty]

its kinda hard to explain but its for cases like this

struct HashMap_Bucket<T, U> {
    T key,
    U value,
    bool full
}

struct HashMap<T, U> {
    HashMap_Bucket<T, U> *[] table,
    u64 size,
    u64 capacity
}
``` you know that the `Array<T>` struct holds a `HashMap_Bucket<T, U> *`, but you dont know what `T` and `U` are so you need to generate a partially generic struct `Array<HashMap_Bucket<T, U> *>`, which has a defined `T` being `HashMap_Bucket<T, U> *`, but that now depends on `T` and `U` being defined later

i probably could explain this better if i knew the theory lmao

i have no real way to explain it so, sorry if im making no sense

[fleet cedar]

Why are you trying to monomorphize things at all before you have the types

[valid jetty]

because Array<T> lives in a different context to HashMap_Bucket<T, U>

you know what the T for Array<T> is at all points, its HashMap_Bucket<T, U>

[fleet cedar]

??

[valid jetty]

hold on

[fleet cedar]

Which is not a type yet

[valid jetty]

.

[valid jetty]

it technically is as far as the type system is concerned

^^^

[fleet cedar]

It is not

It's simply an invalid expression

Just like x is not a valid lambda calculus term, whereas \x. x is

[valid jetty]

i mean i get what you mean but you have to describe the relationship between (in this case) what Array<T>'s T is and what T is when monomorphizing HashMap<T, U>

because T means something different depending on the pass of monomorphization that youre on

[fleet cedar]

They are different variables in different scopes

There is no relationship

[lavish cloud]

Wait so this doesn't work? But that's a super easy to do thing in any lang with generics or even in C++ templates

[valid jetty]

no this does work

[valid jetty]

this is what doesnt

[fleet cedar]

Anyway, why are you monomorphizing types before the arguments are given

[valid jetty]

because im messing around and seeing what sticks lol

[deep mulch]

@valid jetty hiii

[valid jetty]

i didnt research into how youre actually supposed to monomorphize things

[deep mulch]

im eepy @valid jetty

rosie does not experience the feeling of eepy

[valid jetty]

sleep !!

[deep mulch]

nopp

too late for more eep

[valid jetty]

smh

[winged mantle]

i find it very funny running grep -r stackoverflow\.com node_modules

you can see all of the lib devs rely on stackoverflow as much as you

[valid jetty]

ok anyway whatever

@hoary sluice @deep mulch you can do all of this now

you can also use let syntax or specify a type and itll cast the variables to that type

ie

[deep mulch]

i dont get it

youre redeclaring variables already in scope

[valid jetty]

thats just shadowing

lol

but i was talking about the destructuring

instead of ```rs
res := queue.remove(0).unwrap();
node := res.x;
level := res.y;

you just do
```rs
node, level := queue.remove(0).unwrap();

[deep mulch]

oh

no multi variable initializing like

x, y =: 0

?

[valid jetty]

you can do x := y := 0

you can define destructuring on any type by defining the __tuple__ and __triple__ methods

struct Foo {
    i32 a,
    string b,
    f32 c
}

fn Foo::__tuple__(Foo self) {
    return $(self.a, self.b);
}

fn Foo::__triple__(Foo self) {
    return $$(self.a, self.b, self.c);
}

a, b := Foo { a = 1, b = "a" };
a, b, c := Foo { a = 1, b = "a", c = 1.3 };

[deep mulch]

why not something like kotler and do a componentN N being the place

would allow for more than just 2

[royal nymph]

bad

[deep mulch]

oh my god $$

rosie fix

[valid jetty]

because ill do that when tuples exist with arbitrary length

[fleet cedar]

Why not just have __tuple__ return an arbitrary length tuple

[deep mulch]

good

[royal nymph]

@valid jetty

struct Baller {
  i32 foo,
  string bar
}

.foo, .bar = Baller { foo = 42, bar = "explode" }

add

[valid jetty]

because that means i need to have compiler sugar to generate arbitrary length generic structs which i havent done yet

[deep mulch]

struct Baller {
  i32 foo,
  string bar
}

val foo2: .foo, bar2: .bar = Baller { foo = 42, bar = "explode" }

[valid jetty]

ill do that along with getting rid of $ to initialize it

i do want to generalize the syntax

[deep mulch]

rename Elle to Rust and change logo to a Rust

[fleet cedar]

Tuple and triple is a very silly terminology at any rate, since a triple is a tuple

[valid jetty]

because the tuple type is (T, U), the triple type is Triple<T, U, V> the init function for a tuple is $(x, y), the init function for a triple is $$(x, y, z), and the destructuring syntax is x, y := z or x, y, z := w

its all over the place and i kinda hate it

[royal nymph]

AWFULLLLLLLLLLL WHAT IS THIS

[deep mulch]

you love

[valid jetty]

atp just use js 😭

[deep mulch]

@valid jetty@valid jetty

[fleet cedar]

%s/tuple/pair/g

[deep mulch]

rosie doesnt know how to find and replace

[valid jetty]

interface Baller {
    foo: number,
    bar: string
};

const { foo: foo2, bar: bar2 } = { foo: 42, bar: "explode" } satisfies Baller;
``` @deep mulch

[deep mulch]

satisfies Baller; is ulgy

[valid jetty]

sure i guess

[royal nymph]

you don't need that

rosie is insane

[deep mulch]

rosie is insane

[valid jetty]

you do if you want it to have interfaces like in a strongly typed language

or make a class

but cl*sses

[royal nymph]

satisfies is like a cast but not a strong cast

[valid jetty]

yeah

[fleet cedar]

But interfaces suck though

I can list half a dozen ways traits are better

[valid jetty]

lmao

but this is typescript

everything kinda sucks

[fleet cedar]

True

[deep mulch]

ellescript

[royal nymph]

interface Thing {
  numberOfTimesItExploded: number
}

const someThing = { numberOfTimesItExploded: 42 };

const asThing = someThing as Thing;
const satisfiesThing = someThing satisfies Thing;

[valid jetty]

honestly speaking though, there hasnt ever been a time when i needed a tuple with more than 3 elements inside

[royal nymph]

satisfies is cool

[valid jetty]

if you need a tuple with more than 3 elements we have a tool for that: its called a struct

[deep mulch]

@valid jetty what about like

[fleet cedar]

If you see tuples bigger than that then either you want a named struct, or you're doing something very cursed like bevy or something

[deep mulch]

a use case of more than 3 would be colors

rgba

[valid jetty]

yeah this is my primary usecase for it

[valid jetty]

exactly

[deep mulch]

r, g, b, a := col;

[valid jetty]

thats why i havent had the incentive to make arbitrary named tuples

[valid jetty]

i did actually have this problem in my head

[royal nymph]

rosie will add iota keyword

[deep mulch]

i knew it

[valid jetty]

which does what 😭

[fleet cedar]

Oops I accidentally ```
a, r, g, b = col;

[deep mulch]

you solve problems in your mind

[valid jetty]

i already have enums

[deep mulch]

fix

[royal nymph]

https://go.dev/ref/spec#Iota

sooo good

[deep mulch]

a: .a, r: .r, g: .g, b: .b

[royal nymph]

idk why it's called iota

[deep mulch]

[fleet cedar]

Yes but type checker didn't catch it so it broke in prod

[valid jetty]

the only real usecase for iota is for enums when go doesnt have real enums tho

[deep mulch]

its some greek word

[deep mulch]

developers fault i think

[valid jetty]

its a greek letter

[royal nymph]

😭

[deep mulch]

@royal nymph

[fleet cedar]

Developer's fault for using a tuple instead of a named struct yeah

[deep mulch]

V

[valid jetty]

in computer science its like "a small increment" because it goes up by 1 each time you reference it

thats why its called iota afaik

[royal nymph]

i mean

[valid jetty]

that too

[deep mulch]

idk

r, g, b, a := col
``` is way neater than
col.r, col.g, col.b multiple times in a function

[frosty obsidian]

android moment

[deep mulch]

if the dev cant read that the order is rgba thats their fault

[fleet cedar]

let Rgba { r, g, b, a } = col;

[frosty obsidian]

sometimes it is argb and that is annoying

[valid jetty]

^^

enum Foo {
    startAt100 = 100,
    a, // 101
    b, // 102
    c, // 103
    d  // 104
}
``` there is no other purpose for iota surely

[deep mulch]

you love

[fleet cedar]

Argb is the correct one

[deep mulch]

CMYK

soon

[frosty obsidian]

i agree actually

consistently sized number

easier to do bitwise stuff

[deep mulch]

@frosty obsidian soon RGB will be replaced by CMYK

[frosty obsidian]

cmyk is terrible for anything other than printers

[fleet cedar]

ACMYK or go home

[deep mulch]

@frosty obsidian

[valid jetty]

contrast, meep, yellow, kontrast

oh cyan magenta yellow, key

i got it

why do they call it key just call it lightness

[frosty obsidian]

@deep mulch you forgot to rewrite pipette to use oklab

[frosty obsidian]

bc its a different concept

cmyk is subtractive

[deep mulch]

i cant

[valid jetty]

oh

[deep mulch]

would be weird

nothing uses OkLab pickers

except professional software

[frosty obsidian]

you'll be the first

[deep mulch]

never

[frosty obsidian]

you could be rich

[deep mulch]

how

[fleet cedar]

Should be just called black

[deep mulch]

rgb

pantone

[valid jetty]

Black bird, black moon

Black sky, black light

Black, everything black

Black heart, black keys, black diamonds

[fleet cedar]

Is that poetry

[valid jetty]

Blackout, black, everything black

no

https://open.spotify.com/intl-ja/track/3UEnF6y5tyHVtMzldS3svp

[deep mulch]

@valid jetty@valid jetty@valid jetty@valid jetty@valid jetty

[valid jetty]

@deep mulch minklang

[deep mulch]

yes

[valid jetty]

i think ill just push tuple destructuring like this, whatever

im not doing arbitrary-length tuples today

that will take forever

[deep mulch]

do

[fleet cedar]

Btw are you making elle and いちご just for fun or do they have any specific purpose

[valid jetty]

いちご is supposed to be a school project but its very overkill, so both are for fun

im learning a lot

hmm i just realized what having tuple destructuring means

thats cool

[jagged cedar]

i took exam about this yesterday bruh

why when i open the server this is the first thing i see

[hoary sluice]

zooter im a big fan of kotlin but not everything has to be exactly like kotlin

[deep mulch]

it was just an example

variable length destructuring isn't a Kotlin specific thing

[hoary sluice]

componentN is kind of a meh solution

[deep mulch]

it's the only one I could think of

[royal nymph]

yes

kotlin destructuring is ass

[hoary sluice]

you can make variable length destructuring just not with arbitrary functions

[royal nymph]

js has peak destructuring

[hoary sluice]

nop

[royal nymph]

yop

idk how anyone likes parameter order mattering for destructuring

it's so ass

[hoary sluice]

js is unequivocally terrible no matter where

[lavish cloud]

[𝐄𝐗𝐓𝐑𝐄𝐌𝐄𝐋𝐘 𝐋𝐎𝐔𝐃 𝐈𝐍𝐂𝐎𝐑𝐑𝐄𝐂𝐓 𝐁𝐔𝐙𝐙𝐄𝐑]

[deep mulch]

js only good for websites

[hoary sluice]

be honest componentN is not the best solution

especially when the stdlib doesnt do more than component5

component(n) would be awesome i think

[lavish cloud]

How would you do it then
Especially when you need to preserve compat with JVM

component(n) would be slower

And also impossible to type correctly

[hoary sluice]

how about componentN and component(n) but componentN is preferred and the stdlib has up to component5 and then arbitrary component(n) (for collections)

[deep mulch]

what's the difference between componentN and component(n)

[hoary sluice]

or dont put it into a function and make an actual language construct for it

[hoary sluice]

componentN doesnt have branching, component(n) does

[deep mulch]

I don't get it

[hoary sluice]

u can directly call componentN

u need a switch case in component(n)

[deep mulch]

wouldn't that be worse on performance

unless it can be inlined

[lavish cloud]

And also component(n) would need to be typed as Any?

And you'd have to manually cast it to the correct type

[lavish cloud]

But for collections wouldn't you just do collection.get(n)

[formal belfry]

@woven mesa im making a proper swift rpc package

20 projects are already named SwiftRPC but im naming mine that

it won't use sockets so you can use it in a sandboxed app

[shrewd canopy]

doesn't discord recommend using slayer sdk instead of rpc for rich presence now?

[ornate quiver]

yop
we need name based destructuring

val [name, id] = data

has anyone noticed that different languages have a different "feel"
kinda like driving different cars
idk how to explain it idk if im just crazy

[random maple]

does vencord have that plugin where if u put over 2000 characters it splits it into multiple messages

[ornate quiver]

vns

[elder yarrow]

👉 #🏥-vencord-support-🏥

[woven mesa]

so true

[formal belfry]

i dont really know

its not public though

[shrewd canopy]

Whats not public

[formal belfry]

social sdk, you cant download it

need to sign up or something

[winged mantle]

// node does not currently support unit tests using typescript >:(

import { register } from "node:module";
import path from "node:path";
import { pathToFileURL } from "node:url";

register("@swc-node/register/esm", pathToFileURL(path.join(import.meta.dirname, "node_modules")));

[shrewd canopy]

yes you can, its public

[formal belfry]

https://tenor.com/view/castorice-hsr-honkai-star-rail-frosting-cupcake-gif-16525147167601276816

[shrewd canopy]

just fill form in and they will automatically give access

[formal belfry]

o

[winged mantle]

TIL node has swc built in

they use it for --experimental-strip-types (which no longer needs a flag but ye) and --experimental-transform-types

but for some reason --test does not support it

and using https://github.com/nodejs/amaro didn't give me much luck either

(which is just a swc wrapper tbf)

@royal nymph why do you bundle using esbuild in venbot instead of using https://github.com/egoist/esbuild-register which adds a hook to require

[hoary sluice]

dude

my cat got hit by a car

[valid jetty]

what

youre joking right

[hoary sluice]

no

his paw got hit and his face was full of blood

we found him in a ditch at the side of the road

[valid jetty]

wtf :(

[hoary sluice]

wondering why he didnt come home today

[valid jetty]

i hope hes ok :c

[hoary sluice]

no hes dead

getting hit by a car at 9 years old was not how i was expecting him to die bro

[valid jetty]

nooooo :c

poor cat

[hoary sluice]

it was like 30m away from our house i dont understand why he got hit

its not like he was lost

weve had cats die before but this just hits different

ive known him for more than half my life

[supple whale]

honestly? its still fucking dogshit, for me by far simply tsx is by far the best

not perfect when it comes to workers, but as good as it will get tbf

but when i want typescript with nodejs i usually jusst go for deno

because ts in nodejs is still cancer and has issues no matter what u try

[winged mantle]

wdym

[supple whale]

node's ts impl

[winged mantle]

it works fairly well

[supple whale]

its ass

doesnt work for dependencies which need to be in js not ts, its file extension handling is fucky and ignores tsconfig

[winged mantle]

it's unfinished and you need to pass a flag to actually transform but i think in the future it will be better

but for my case it works just fine

[supple whale]

for simple things its fine

problem is, TS is useless for small simple libs

TS is useful when you're doing complex shit, which is when node's ts impl starts to become problematic

https://cdn.discordapp.com/emojis/776909496764071977.webp?size=48&name=woe~1

[winged mantle]

it works just fine for me in my pretty complicated project - i don't expect that much lol

i don't expect it to typecheck

i dunno what the dependency issue is

most typescript options only have an effect on typechecking

i don't use jsx

[supple whale]

nah i mean raw ts

fuck tsx

react is a cancer XD

like for example

import lib from 'lib/src/main.ts'

wont work

so like, cool you can use TS, your deps cant

https://cdn.discordapp.com/emojis/776909496764071977.webp?size=48&name=woe~1

legit just deno index.ts

works better

[winged mantle]

well - that's just something it doesn't do - not a problem with what it does
every dependency that supports node.js will work just fine

I do see the annoyance if you have a monorepo and are depending on other internal packages

[supple whale]

did they finally fix import ./src/file.ts?

or was it ./src/file

[winged mantle]

wait whgar

[supple whale]

i dont remember one of them didint work

[winged mantle]

that's an esm thing you need extension

and you need explicit index.ts

unless the type stripping also requires that

but i'm using esm anyway

[supple whale]

yeah it ignores tsconfig

[winged mantle]

i see

well

[supple whale]

yeah its defo not usable imho, i just wish deno was more widely supported in shit like android or electron or vscode

[winged mantle]

i just appreciate node somewhat catching up with other runtimes so i can just use what's familiar and stable and have nice stuff like that even if it's not quite on the same level

[supple whale]

so i could fucking ditch node

[winged mantle]

i don't really care about these problems

i just use esbuild or something else if i have to

[supple whale]

yeah, when i have no errors in my editor during dev, i want to have no errors at runtime

tsx is bae in that regard

like it just works

its insane to me how many alternatives to tsx there are there which just dont work

[winged mantle]

isn't it really sloow by default because it just uses tsc

[supple whale]

you run tsx and it magically works first try

[winged mantle]

ts𝕏

[supple whale]

nah it uses esbuild

[winged mantle]

i swear i used another library where it used tsc

[supple whale]

yeah there's one that uses import hooks

i mistakenly used it in prod one time

the memory usage and cpu usage on that shit was NUTS

[winged mantle]

wait

it must have been older version of tsx

[supple whale]

SA comes up to me like "yo your k8s instance is using like 2 gigs of ram for some simple script, wtf u doing"

[winged mantle]

???

i am confused

it does seem unlikely they would switch from tsc to esbuild

[supple whale]

node-ts/ts-node?

[winged mantle]

oh probably

[supple whale]

yeah that shit implodes RAM and CPU usage

[winged mantle]

yeaa

weird

[supple whale]

idk how it has 30 mil dls per week

[winged mantle]

i swear i copied the lib i was using from another project

[supple whale]

when its so bad

[winged mantle]

and that other project uses tsx

(until it was rewritten in rust just casually)

[supple whale]

ts-node uses tsc afaik

[winged mantle]

yeah it was totally that one

https://github.com/TheKodeToad/Foxy/blob/master/package.json#L5-L19

[pearl stag]

package.json: Lines 5-19

"scripts": {
    "start": "ts-node src/main.ts"
},
"keywords": [],
"author": "",
"license": "ISC",
"dependencies": {
    "@discordjs/rest": "^2.2.0",
    "@types/express": "^4.17.21",
    "discord-api-types": "^0.37.75",
    "dotenv": "^16.4.5",
    "express": "^4.18.3",
    "ts-node": "^10.9.2",
    "tweetnacl": "^1.0.3"
},

[supple whale]

yeah i made that mistake once

it seems like a "good choice" because it has nice clean hooks into node

but nope, bad idea

[frosty aurora]

Hi

[winged mantle]

i feel i have come a long way since then even thought it was only a year ago

[supple whale]

yeah i've known ts for years, but only seriously started using it like a year ago

i still fucking hate typescript because of its insane per-project setup and god awful default settings

which is why i've put off using it for years

but its gotten better now

[winged mantle]

i was wanting to make a bot

i started with java

[supple whale]

XD

GG

[winged mantle]

then i used C++

then I used python

then i used rust

then i used typescript xD

(do NOT ask about C++)

[supple whale]

you procedurally used worse and worse languages for the job XD

[winged mantle]

i don't know why i considered c++

[supple whale]

rust had awful support for discord bots for ages, mainly because json and web and yeah

[supple whale]

yeah bro i couldnt tell you

i will never touch C/C++ for a personal project

its fucking hell

[winged mantle]

i think just because it had coroutine support

[supple whale]

i barely know any rust at all, and its so much better to compile and import dependencies its insane

like even with my utter incompetence in rust i still get the job done faster than in C/C++ because i can actually compile things and import dependencies and it just fucking works

FUCK MAKEFILE

[winged mantle]

rust is decent for discord bots however I to this day have concerns regarding using serenity for a large scale project as it doesn't seem to handle ratelimits correctly?

[supple whale]

rate limits.... are a complex thing

like hardest things in programming

1. naming shit
2. caching
3. rate limits

no cap.

buckets, bursts, refreshes, retries

its... hard as shit

[winged mantle]

oh there was also

the fact you couldn't set allowed mentions globally

[supple whale]

https://cdn.discordapp.com/emojis/1207393766458007562.webp?size=48&name=kekwolf

[winged mantle]

that's probably fixed by now idk

can't remember

[supple whale]

sounds about right

[winged mantle]

poise is a pretty nice framework though and it uses serenity

[supple whale]

yeah libraries for web tech in non-web languages is usually 🐶💧

[winged mantle]

however i prefer to roll my own command handling stuff

things irk me i love how every command handler seems to say type .help command_name for wrong usage instead of having a button you can press
it's not that hard to make a button... this is not an unsolved computer science problem...

[supple whale]

also fuck tsc

[winged mantle]

idk that might not be entirely accurate

go rewrite 🚀

[supple whale]

it fucking scans node_modules deps which are in ts and errors build if they have ts-config incompatibilties

https://cdn.discordapp.com/emojis/776909496764071977.webp?size=48&name=woe~1

[winged mantle]

skipLibCheck

[supple whale]

nope

[winged mantle]

but supposedly that does bad things

[supple whale]

skipLibCheck doesnt help that

my node_modules deps themselves are in ts, and have their own tsconfigs

but tsc ignores them

and it uses the root tsconfig from my project

ignoring the scoped tsconfig in the dep

skipLibCheck doesnt work around that, because tsc NEEDS to compile that dependency, as its TS only, there's no JS code for it to import

[winged mantle]

unless you're not using node, libraries generally compile to a dist folder and only publish that

[supple whale]

yeah i'm not using node libs

[winged mantle]

my answer to everything just use node

😎

[supple whale]

cant run node in browser

:^)

[winged mantle]

oh fair

why do these libraries not contain the compiled ts though

[supple whale]

esbuild handles it correctly

but tsc which i use in my action hooks errors

https://cdn.discordapp.com/emojis/776909496764071977.webp?size=48&name=woe~1

it's gotten to the point where i use eslint for type checking because it fucking works better

https://cdn.discordapp.com/emojis/1233149945465536522.webp?size=48&name=cocainewolf

[winged mantle]

shrimply get it to cd into every package and run tsc and rewrite the package.json to import from the output folder

[supple whale]

unironically that's the official solution

[winged mantle]

use jq in the github action

WHAT

[supple whale]

a script which adds // @ts-ignore at the start of every .ts file in node_modules

[winged mantle]

wait is jq read-only

dhfgj

unsane

[supple whale]

thats the current solution on tsc's github regarding the problem

[winged mantle]

i feel like some developers of big libraries are trolls they want to annoy people who open issues as much as possible

[supple whale]

https://github.com/microsoft/TypeScript/issues/40426#issuecomment-2522221597

https://cdn.discordapp.com/emojis/1233149945465536522.webp?size=48&name=cocainewolf

its a ts problem not a tsc problem

https://cdn.discordapp.com/emojis/1293934183307022387.webp?size=48&name=troll_unhinged

[winged mantle]

isn't tsc part of ts

[supple whale]

right

i keep forgetting that

[winged mantle]

It is good that TypeScript errors here, even if it is an overall terrible user experience.

expert troll who managed to be employeed by microsoft /s

[supple whale]

yeah

this is legit why deno

it just works with ts dependencies no problem-o

[winged mantle]

i think if somebody opens an issue saying something is broken i should gaslight them into thinking they have problems

[supple whale]

i need to look at this https://gist.github.com/felskov/84fa477e87fb18f3ff7bee4ee2ce1c57

[winged mantle]

and they should get a therapist

i will just close every issue from now on

"get a therapist"

[supple whale]

supposedly its tsc but without dep checking

[winged mantle]

in all seriousness honestly with issues like these i think there's usually some technical reason that's hard to understand

[supple whale]

it really is not

but eh

[tawny umbra]

I wonder if Lina is in a state now where I can submit to testflight hmmm

[shrewd canopy]

Whats Lina

[tawny umbra]

I accidentally posted it here but I guess it fits

App I'm making

[reef whale]

out of curiosity how different is react js to react native and what's the learning curve like trying to pickup react native if you already know react js

[deep mulch]

im biased but I highly recommend against it. it's buggy and performance isn't gonna be like a proper native framework
it's just not a good solution for multiplatform applications

[shrewd canopy]

Its mostly identical, just use <Button>foo</Button> instead of <button> for example, and RN has own styles thing instead of css
but anything that is JS/TS shouldnt be used outside of web browsers

[lavish cloud]

^

If you want something proper with a similar-ish API, give Compose a try.

[deep mulch]

give

im jealous

[supple whale]

i want one for svelte ngl

[valid jetty]

only took half a year

[deep mulch]

@rosie.pie hiiiii

@valid jetty

[spark tiger]

resell for $500

[supple whale]

would it even sell?

its kotlin...

[deep mulch]

silence

id buy

not for $500

[royal nymph]

buy for 100$ @deep mulch

[deep mulch]

okay

[spark tiger]

probably not cuz coders are dirty nerds who are not interested in spending $$$ on clothes 😭

[supple whale]

looks at furry community... fair....

[valid jetty]

i don’t even pay for my software why would i pay for merch

i would much rather spend the money on 地雷系

https://tenor.com/view/地雷系が大好きな人-地雷系-gif-6958619523295176256

[supple whale]

mfw man is talking to himself

[spark tiger]

i would spend all my income on an extremely rare and pretty piece of clothing for $2000 ngl

yall would not..?

[supple whale]

my limit is $100 or $150 for a single piece of merch

[fleet cedar]

I would not spend $2000 for a tshirt

[valid jetty]

exactly

[spark tiger]

still can’t believe ts is real 😭😭😭

[deep mulch]

id wear

[spark tiger]

i mean the fact that vetements has a fucking java t shirt not the price (it’s a fair one imo)

[fleet cedar]

Who the would pay $878 for that

[deep mulch]

me

[fleet cedar]

Looking at a random clothes store's website, I could buy 125 tshirts for that money

[royal sphinx]

How to increase the size of server icons?

[elder yarrow]

👉 #🎨-theme-development

(Auto-response invoked by @jade stone)

[spark tiger]

me if it’s a design i like, a brand i like, rare and interesting piece

[spark tiger]

what store

[runic sundial]

Vro sounds like a rich kid

[spark tiger]

no i’m actually not that’s why i don’t have any expensive clothes 😭😭😭😭

[runic sundial]

Then what the hell are you on about

Who pays more than like

20 dollars for a shirt ever

Better be a fucking golden shirt

[spark tiger]

i’m saying if i was earning shit i would consider paying that much for a tee if i fw sm

[runic sundial]

But have u considered how much ice cream you could buy instead

Or frozen pizzer

[spark tiger]

bro fuck ice cream u can’t rock dat

[runic sundial]

I can eat it

I want ice cream now

Accept my apolocheese for the outburst

I will now investigate the viability of ice-cream

[spark tiger]

u gon b sick from that much ice cream

[runic sundial]

I only want like a scoop

But the idea is that you can always get a scoop

And if you can control yourself

[spark tiger]

or a t shirt

[runic sundial]

It is good

[spark tiger]

need 😭😭😭

[fleet cedar]

You need to charge your phone

[spark tiger]

I KNEW SOMEONE WOUKD SAY THAT

it’s at 13% it’s gonna be fine

[valid jetty]

@hoary sluice @deep mulch @placid cape I DID IT

I ACHIEVED PEAK

fn Array::map_with<T, U, V>(T[] self, fn(T, V, u64, T[]) -> U mapping, V arg) -> U[] {
    let new_self = Array::with_capacity<U>(self.len());

    for u64 i = 0; i < self.len(); i += 1 {
        new_self.push(mapping(self[i], arg, i, self));
    }

    return new_self;
}

fn Array::map<T, U>(T[] self, fn(T, u64, T[]) -> U mapping) -> U[] {
    let new_self = Array::with_capacity<U>(self.len());

    for u64 i = 0; i < self.len(); i += 1 {
        new_self.push(mapping(self[i], i, self));
    }

    return new_self;
}

STRONGLY TYPED FUNCTION SIGNATURES IN THE TYPE SYSTEM

!!!!!!!!!!!!!!!!!!!!!!!

[winged mantle]

i tried shortening it but it made it longer

[valid jetty]

wtf am i looking at

[deep mulch]

this is hard to read fix

[valid jetty]

how

[winged mantle]

wdym...

[deep mulch]

too much self

why can't self just be implicitly an argument

developer shouldn't have to always specify it as the first arg

I think it would be easier

@valid jetty @valid jetty

[valid jetty]

magic

[fleet cedar]

Decades of experience with c++ and java have learned that explicit self is better

[valid jetty]

^^^

[deep mulch]

does this support non local functions too@valid jetty

[valid jetty]

wdym

[deep mulch]

map(someFn)

[valid jetty]

yes you can pass any function to it

fn Array::truncate(i64[] self) {
    return self.map(i64::truncate);
}

fn Array::promote(i32[] self) {
    return self.map(i32::promote);
}

as long as it has a signature and isnt just a void* or something

[deep mulch]

fn Array::truncate() {
    return self.map(i64::truncate);
}

fn Array::promote() {
    return self.map(i32::promote);
}
``` fixed

[winged mantle]

how do ints have functions

scary..

[valid jetty]

any type can have functions defined on it

[fleet cedar]

Int is a type

[valid jetty]

it sounds cursed but it is actually really really powerful

[deep mulch]

@valid jetty can function be defined on null

[valid jetty]

because you can do stuff like this

global pub, external;

fn isalnum(char arg) @alias(char::is_alphanumeric) -> bool;
fn isalpha(char arg) @alias(char::is_alphabetic) -> bool;
fn iscntrl(char arg) @alias(char::is_control) -> bool;
fn isdigit(char arg) @alias(char::is_digit) -> bool;
fn isgraph(char arg) @alias(char::is_graphic) -> bool;
fn islower(char arg) @alias(char::is_lowercase) -> bool;
fn isupper(char arg) @alias(char::is_uppercase)-> bool;
fn isprint(char arg) @alias(char::is_printable) -> bool;
fn ispunct(char arg) @alias(char::is_punctuation) -> bool;
fn isspace(char arg) @alias(char::is_whitespace) -> bool;
fn isxdigit(char arg) @alias(char::is_hexdigit) -> bool;
fn tolower(char arg) @alias(char::to_lowercase) -> bool;
fn toupper(char arg) @alias(char::to_uppercase) -> bool;

'a'.is_digit()
``` and this is a 0 cost abstraction which, in the IR, literally just does `isdigit('a')`

[valid jetty]

null isnt a type

nil is a void*

[deep mulch]

I null

[winged mantle]

not 0 zero cost it should skip function call

[valid jetty]

well

[deep mulch]

can you declare function on void*

[valid jetty]

it calls into crt

[winged mantle]

in rust can you call functions on ints idk

[valid jetty]

yes ofc

[winged mantle]

probably

[valid jetty]

there is a special compiler intrinsic specifically for pointers and void pointers

that being __void_ptr__ and __ptr__

fn __void_ptr__::__fmt__(void *self, i32 nesting) {
    return "<unknown at {}>".__internal_formatter_do_not_use_unless_you_know_what_youre_doing__(self);
}

!external fn __ptr__::__fmt__<T>(T *self, i32 nesting) {
    let res = "invalid";

    if #cast(bool, self) && #cast(void *, self) % #size(void *) == 0 {
        res = (*self).__fmt__(nesting);
    }

    return "<{} at {}>".__internal_formatter_do_not_use_unless_you_know_what_youre_doing__(res, self);
}

[winged mantle]

stop putting functions on primitives
ints were never meant to be .toStringed

[valid jetty]

ugly but it doesnt need to be pretty

[deep mulch]

Rosie will add proper visibility modifiers and remove __

[valid jetty]

its an internal function

[deep mulch]

@valid jetty tomorrow Elle will be released

[fleet cedar]

Where else would you put the function for tostringing an int

[spark tiger]

sounds like a common thing though ain’t it

[winged mantle]

sprintf

[valid jetty]

rust has stuff like saturating_sub which essentially does (x - 1) < 0 ? 0 : (x - 1)

its useful to have methods on ints

and any type in general

[winged mantle]

more like sub_without_overflow?

[fleet cedar]

The less special primitives are, the better

[valid jetty]

✨ saturating_sub ✨

yeah ^^

[deep mulch]

private fn voidptr::fmt(i32 nesting) {
    return "<unknown at {}>".format(self)
} 

fixed

wait

[winged mantle]

i think there's a good argument to not allow operator overloading and only make primitives have them

[deep mulch]

private fn voidptr::fmt(i32 nesting) {
    return "<unknown at ${self}>"
} 

fixed

[winged mantle]

but i understand why people like operator overloading

[fleet cedar]

Sure, if you want to encourage primitive abuse

[deep mulch]

good for working with special units

5.km + 2.mi

or something

[winged mantle]

compromise: make everything use awkward methods

5.times(10)

[valid jetty]

thank you rust

[deep mulch]

@valid jetty do you love my ideas

[fleet cedar]

Based

[deep mulch]

one day I will just rewrite Elle in Kotlin

[valid jetty]

yeah you can do this in elle

[winged mantle]

😭

allow doing not like in c++

/run

#include <iostream>
if (not false)
    std::cout << "hi\n";

[rugged berry]

Here is your c++(10.2.0) output @winged mantle

hi

[valid jetty]

oh actually

you did times like *

[deep mulch]

I'll make Elle feature complete in Kotlin before you finish Elle @valid jetty

[valid jetty]

i was thinking of the other times

[winged mantle]

what like 5555555555

5.repeat(10)

[deep mulch]

@valid jetty make compose in elle

[valid jetty]

i was thinking of this other example

[winged mantle]

make ellex

[valid jetty]

which is honestly funnier

[winged mantle]

@valid jetty like jsx but for elle

[valid jetty]

evil

[winged mantle]

is it weird that i named a file wish_dot_com

[deep mulch]

@valid jetty Elle will support this (syntax
(5.mi + 3.km) / 10.sec

[winged mantle]

imagine if you add units to primatives

print((5 miles).unit) // miles

print(typeof(5 miles)) // int

where it's still the same type

[deep mulch]

infix getters would be funny

[winged mantle]

infix is kinda clean

[deep mulch]

yes

[winged mantle]

/run

println 'hi'

[rugged berry]

Here is your groovy(3.0.7) output @winged mantle

hi

[deep mulch]

I like as long as it's not abused too much

[winged mantle]

why don't languages have a print operator if printing is used so much like
🖨️"hello world"

why husk

get special programming keyboard

all necessary symbols

why don't programming languages add support for exporting with tarrifs

[deep mulch]

5 mi / 3 hr // mph
9.8 m / 1 sec ^ 2 // m/s²

[valid jetty]

f# has this

its so cursed

[winged mantle]

i don't understand why you need units lioke that

[deep mulch]

@valid jetty Elle repl

[winged mantle]

just do 5 * MILES