🪅-progaming

[royal nymph]

it's mainly meant as a type safe compiled ruby alternative

crystal and ruby have essentially the same syntax

[spark tiger]

is it like mature enough tho

like the ecosystem and stuff

tooling etc

[royal nymph]

depends

if you're a big company I wouldn't use it

[spark tiger]

nop i’m me

[royal nymph]

then yes

[deep mulch]

.getOrNone

[fleet cedar]

Doubt

[spark tiger]

🐈

[royal nymph]

crystal is really fast while being super easy and fast to write

it feels more like a scripting language than a compiled language

you only rarely have to write explicit types

[valid jetty]

// overloads
fn Array::__load__<T>(ElleMeta meta, T[] self, u64 i) {
    __internal_array_bounds_check(meta.file, meta.line, meta.column, i, self.len());
    return self.elements[i];
}

fn Array::__load_ref__<T>(ElleMeta meta, T[] self, u64 i) -> T * {
    __internal_array_bounds_check(meta.file, meta.line, meta.column, i, self.len());
    return &self.elements[i];
}

fn Array::get<T>(ElleMeta meta, T[] self, u64 i) -> Option<T> {
    if i + 1 > self.size {
        return None();
    }

    return Some(self.elements[i]);
}

fn Array::get_ref<T>(ElleMeta meta, T[] self, u64 i) -> Option<T *> {
    if i + 1 > self.size {
        return None();
    }

    return Some(&self.elements[i]);
}

[spark tiger]

idk why but i’m just really interested in exploring other new languages. like zig, gleam, crystal, nim

[deep mulch]

rosie add visibility operators

[lavish frigate]

zig is irrelevant to me

[valid jetty]

wdym

[deep mulch]

private fn Array::load

[spark tiger]

wh

[valid jetty]

array::__load__ should not be private lol

its what is called when you do array[x]

[royal nymph]

/run ```cr
def shout(thing)
puts thing.to_s.upcase
end

shout "hi"
shout 1234
shout ["Hello", "World"]

[rugged berry]

Here is your cr(0.36.1) output @royal nymph

HI
1234
["HELLO", "WORLD"]

[deep mulch]

__internal_array_bounds_check

[valid jetty]

and anyway that exists the operator is !pub

[deep mulch]

[valid jetty]

thats only because no mangling, as i have mentioned several times before

a function can be private but theres no mangling so if it doesnt have a unique name itll fail at linking time because duplicate symbols

[deep mulch]

fix

[valid jetty]

the function is private

[spark tiger]

my list of languages i want to learn is like elixir, zig, rust, crystal, c, swift, objective-c, nim, gleam, haskell, go, v, scala, julia, odin, ts. is it too many?

[valid jetty]

you cant call it yourself

[deep mulch]

!pub is such a weird choice

[royal nymph]

it's essentially duck typed but at compile time instead of runtime. the compiler infers types for you

but you can also explicitly type stuff if you prefer

[spark tiger]

oh it uses ends eh

[valid jetty]

its not, it allows the toggling to be very simple

[royal nymph]

you can also use brackets

[deep mulch]

okay but what if you add more visibility types

it wont work then

[spark tiger]

oh that’s neat

[valid jetty]

thats exactly the point, !x is a way to make it general across all modifiers

either it has modifier x or it doesnt

[deep mulch]

thats way more confusing than just explicit words

[valid jetty]

whats the opposite word for external

[deep mulch]

internal

[royal nymph]

/run ```cr
2.times do
puts "hi"
end

2.times {
puts "hi 2"
}

[rugged berry]

Here is your cr(0.36.1) output @royal nymph

hi
hi
hi 2
hi 2

[valid jetty]

internal doesnt mean what it should here

[spark tiger]

wtf the syntax is unsane

[valid jetty]

because !external means the function isnt externally defined and needs to have a body

that is not an internal function

[deep mulch]

[spark tiger]

2.times is crazy 😭

[valid jetty]

i love this lmaoo

[royal nymph]

/run ```cr
puts [1, 4, 2, 3].sort.map(&.to_s).join

[rugged berry]

Here is your cr(0.36.1) output @royal nymph

1234

[spark tiger]

im confused do u like need brackets for func arguments or not

[valid jetty]

i think puts is a builtin?

or something

idk

i guess its optional

[royal nymph]

they're optional

[spark tiger]

oh

[valid jetty]

if i had to guess, func(x) is the same as wrapping an expr in parens

so its as if you did func((x)) in a language where its mandatory

[royal nymph]

/run ```cr
puts [1, 4, 2, 3].sort.map(&.to_s).join '-'

puts([1, 4, 2, 3].sort().map(&.to_s).join('-'))

these are the same

[rugged berry]

Here is your cr(0.36.1) output @royal nymph

1-2-3-4
1-2-3-4

[spark tiger]

learning new languages is so fun

if it's not c# or python or java

[valid jetty]

how does it differentiate between fields and function calls 😭

[deep mulch]

sort.map

scary

[royal nymph]

by simply not allowing fields and methods to have the same name

most languages don't allow that

[spark tiger]

wtf are macros just regular funcs or what

[fleet cedar]

I know of Rust and Java that do, and C++, python, and js that don't

[lucid trail]

this is so clean

I really want to do something with crystal now

[royal nymph]

crystal is really cute

crystal also has yoda expression

[deep mulch]

love

[fleet cedar]

Cringe, yoda expressions are

[royal nymph]

/run ```cr
puts "balls" if 42.positive?

[rugged berry]

@royal nymph I received cr(0.36.1) compile errors

Showing last frame. Use --error-trace for full trace.

In file0.code:1:20

 1 | puts "balls" if 42.positive?
                        ^--------
Error: undefined method 'positive?' for Int32
Showing last frame. Use --error-trace for full trace.

In file0.code:1:20

 1 | puts "balls" if 42.positive?
                        ^--------
Error: undefined method 'positive?' for Int32

[royal nymph]

how

[lucid trail]

hmm this seems hard to follow control flow

[royal nymph]

maybe old crystal version that doesn't have positive? yet

[spark tiger]

macros in crystal sound powerful

[royal nymph]

if you overuse it sure

I find it easier to read during other times tho

especially with continue

[fleet cedar]

Are there any benefits to return x if y == 7 over if y == 7: return x

[lucid trail]

continue if makes sense actually

[royal nymph]

arr.each do |e|
    next if e == 42

    puts e
end

I think it reads nicer cause it has the more important bit first

makes it easier to scan

[fleet cedar]

The important part, such as being a conditional?

[valid jetty]

is that the only way to do it or can you do it the other way too

[royal nymph]

the important part is that we continue :P

[royal nymph]

other way too

[valid jetty]

then imo thats worse

because now you have to scan both ways

[fleet cedar]

No the important part is that we maybe continue

[royal nymph]

there's also unless keyword

[valid jetty]

like ruby

lol

[spark tiger]

it's one char shorter

[royal nymph]

arr.each do |e|
    next unless e == 42
end

I keep writing continue cause I'm used to it

[fleet cedar]

If I were to write something in crystal (an unlikely prospect though) I would set up a linter rule that prohibits such out-of-order statements

[valid jetty]

yeah

nice in theory but uhh

[deep mulch]

https://tenor.com/view/absorption-absorb-cat-absorb-le-epic-cat-big-ahh-eyes-gif-4195419603504489856

[valid jetty]

^

[spark tiger]

do yall think it’s possible to make a vc plugin that adds more languages to code blocks?

[fleet cedar]

That shouldn't be hard

hljs is designed to make that easy

[royal nymph]

you're only saying this cause you're used to non yoda 🐮

[fleet cedar]

So just gotta get hold of the hljs instance, which probably is already in vencord.common

[spark tiger]

oh alright then i think that’d be very useful as a plugin though

[fleet cedar]

Partly that, sure

But also partly because the feature objectively sucks

[royal nymph]

nuh uh

[valid jetty]

can you just pr tree sitter grammar to hljs

[fleet cedar]

No, hljs uses hljs grammars

[valid jetty]

husk

[spark tiger]

/run ```raku

grammar Parser {
rule TOP { I <love> <lang> }
token love { '♥' | love }
token lang { < Raku Perl Rust Go Python Ruby > }
}

say Parser.parse: 'I ♥ Raku';

[rugged berry]

Here is your raku(6.100.0) output @spark tiger

｢I ♥ Raku｣
 love => ｢♥｣
 lang => ｢Raku｣

[lucid trail]

seems like it's more regex for hljs grammars

so much regex

[fleet cedar]

Yup

A simple one I wrote some time ago ```js
hljs.registerLanguage("calmare", hljs => {
const TERM = [];
TERM.push({
className: "number",
begin: /0[xX][0-9A-Fa-f]+/,
});
TERM.push({
className: "number",
begin: /(-|\b)\d+(.\d*)?((mm|ms|deg|mdeg)(/s)?)?\b/,
});
TERM.push({
className: "tag",
begin: /\b(flag|system|var|global|name|bgm|magic|quest|shop|sound|town|battle|item|look_point|entrance|object|trigger|label|anim|chip|vis|fork|eff|eff_instance|menu|sepith|at_roll|placement|fn|file|char|field_party|party|custom)[/,
end: /]/,
contains: TERM
});
TERM.push({
className: "tag",
begin: /\b(self|null|random)\b/,
});
TERM.push({
className: "string",
begin: /"/,
end: /"/,
});
TERM.push({
className: "string",
begin: /@\w+/
});

return {
    unicodeRegex: true,
    aliases: ["clm"],
    keywords: [],
    contains: [
        ...TERM,
        {
            className: "function",
            begin: /\{/,
            end: /\}/,
            contains: [
                {
                    className: "keyword",
                    begin: /\{/,
                    end: /\}/,
                    contains: TERM,
                },
                {
                    className: "keyword",
                    begin: /#\d*[A-QS-Za-z]/,
                },
                {
                    className: "keyword",
                    begin: /#\d*R.*#/,
                }
            ],
        },
        hljs.COMMENT(/\/\//, /\n/),
        {
            className: "symbol",
            begin: /[-+*/%&^|<>!=~]+/,
        },
    ]
}

});

for(let e of document.getElementsByClassName("language-clm"))
hljs.highlightBlock(e);

[lucid trail]

write a TextMate grammar and use shiki instead

[spark tiger]

wow so simple!

[royal nymph]

crystal is definitely not perfect by any means but I still love writing it

it's such a fun language

I just wish the language server was better

and need more opportunities to use it

[fleet cedar]

Rewrite vencord in crystal

[winged mantle]

https://github.com/ivangreene/java-to-zod 🔥

[fleet cedar]

What's the difference between a pojo and a bean

[royal nymph]

YOP

[winged mantle]

i think pojo is plain object java object which is a more general term for a data-only class

[royal nymph]

I just took a shower and came up with a project idea during it

#showerthoughts

[winged mantle]

you take showers?

even though my project pure typescript i should write the schemas in java and convert them because i love java

[fleet cedar]

And bean is a plain java object with getters and setters for each field

Thereby inflating the line count by 5x and gaining absolutely nothing whatsoever

[winged mantle]

true

i love just using public fields in oop languages

remembering how you need to manually write constructors in java

[royal nymph]

I hate getter setter pattern

[valid jetty]

getter setter is the worst thing to ever fucking plague programming

like it completely voids all invariants you can have about fields in a class or struct

the fact that i could write a program

int x = foo.a;
int y = foo.a;
``` and x and y can be different is the worst fucking thing ever

[fleet cedar]

Yes but bean getter/setter pairs also have no invariants

[valid jetty]

the problem is that the syntax is the exact same as regular fields

[fleet cedar]

No, getFoo() is a very different syntax from foo

[royal nymph]

getter setter is more bearable if the language has sugar for it

[valid jetty]

oh well i mean yeah

[winged mantle]

true but i think we were talking about java getters

[fleet cedar]

The problem is that the semantics are the same even though the syntax is boilersome

[valid jetty]

those are fine because you know youre calling a functon

[winged mantle]

which are often used when they are just directly returning the field value

and not doing anything

just for no reason

[hoary sluice]

you can say this about any impure function

[royal nymph]

tbf just don't write bad code and if the language is not single threaded you have that problem either way

but yes I hate getter and setter too

rw field >>>> get and set

[valid jetty]

no, because you know youre calling a function at those points

at this point unless you read the impl you cant know whether this is a getter or not

[hoary sluice]

solution: everything is a function

[nimble bone]

just use NixOs

[hoary sluice]

variables are functions in icy

[deep mulch]

elle will be used as config language in nix

[valid jetty]

well yeah but thats a functional lang

that has no place in oop

[deep mulch]

elle will replace nixlang

[valid jetty]

yeah exactly

[winged mantle]

wdym

[hoary sluice]

what if you dont depend on side effects, only on internal state

[winged mantle]

give example

[royal nymph]

public class Person {
    public int age;
}

[hoary sluice]

val x: Int get() = y / 2

[deep mulch]

@hoary sluice

[hoary sluice]

@deep mulch

[valid jetty]

the point is that you cant know whether simply accessing a field has side effects

[winged mantle]

are you saying you dislike getters even with sugar

[deep mulch]

document the getter

[royal nymph]

yes

getters are stinky

[deep mulch]

vender

@royal nymph vending language

[valid jetty]

in any sane language, doing a field access is invariant, it cant possibly have side effects because all it does is a memory access

[hoary sluice]

just dont use dependencies

[valid jetty]

as soon as you bring getters and setters into the equation, unless you know what the field is represented as internally, you no longer have that guarantee

[hoary sluice]

in any sane language, calling a function is invariant, it cant possibly have side effects because all it does is a memory access

[valid jetty]

that is not true lol

[deep mulch]

in 1 year elle will look and behave exactly like rust and rosie will rename to rust

[hoary sluice]

depends on your definition of sane

[winged mantle]

if your code has no side effects it just does nothing

[valid jetty]

in what language is it sane to have side effects accessing values with syntax exactly the same as normal field accesses

[hoary sluice]

its not sane

[valid jetty]

lmao exactly my point

[deep mulch]

lazy variable initialization with caching maybe

[hoary sluice]

i didnt disagree with you

[hoary sluice]

altho ive never had problems with it in kotlin

[valid jetty]

the whole thing is that youre turning a pure operation into an impure one (a function call)

[deep mulch]

insane

[valid jetty]

there is no equivalent things when calling functions because in procedural languages every function is impure, thats not an invariant

[hoary sluice]

just say constant

javascript brain where constants arent constant

[winged mantle]

i should make a scripting language called lasagna

[hoary sluice]

and then make its syntax javascript

@valid jetty do u use vim-surround

[valid jetty]

yeah constant

[hoary sluice]

i think its built into zed

[deep mulch]

add JIT and AOT to elle

[hoary sluice]

what is the type of x

[deep mulch]

oh no explicit type

fun

[hoary sluice]

is it () => i32

[valid jetty]

internally yes

[deep mulch]

i suggest making rand:randomi32

[hoary sluice]

this is not sane

[winged mantle]

are you inlining constants

[hoary sluice]

didnt you just complain abuot this

[winged mantle]

[deep mulch]

random::next_i32

[hoary sluice]

:kotlin

[valid jetty]

well yeah because the behavior is more like this

thats kinda the idea behind it

[hoary sluice]

this is so awesome https://github.com/Wansmer/treesj

[winged mantle]

rename const to constexpr

[valid jetty]

its not a constexpr

its not computed at compile-time

[vocal ether]

😭

[valid jetty]

the difference here is that i dont make an invariant that accessing constants is pure

[hoary sluice]

i dont think you should be using const for that

[valid jetty]

in fact i specifically state its not

[winged mantle]

inline

[valid jetty]

idk what keyword to use

[hoary sluice]

const is usually not associated with macros

define

[valid jetty]

yeah probably

[valid jetty]

its not a macro either

[winged mantle]

tbh i just use constexpr for stuff i want inlined

not really clear on the difference between inline and constexpr in c++

[hoary sluice]

its a less powerful #define no?

[deep mulch]

wbat the fuck

[valid jetty]

the internal representation is like this

use std/prelude;

fn x() { return rand::random(0, 6); }

fn main() {
    $dbg(x());
    $dbg(x());
    $dbg(x());
    $dbg(x());
}

[deep mulch]

why is it a const that returns a different value

HUSK

[valid jetty]

much more powerful define

[deep mulch]

const is a constant, its value should not ever change

[valid jetty]

you can make a constexpr function that takes a constexpr computable value and does stuff like recursion in it at compile time

thats why c++ takes so long to compile

[winged mantle]

ohh i guess you use constexpr for it to compute it, not just inline it

[valid jetty]

you can make a fib constexpr function where you pass it 142 and it computes the value from your function at compile time

[valid jetty]

yeah, const is not the right keyword to use here

but it works in most cases

[deep mulch]

wouldnt it just be a function

[hoary sluice]

rosie is a javascript enjoyer

[valid jetty]

its a function but its called automatically

[deep mulch]

yop

[valid jetty]

^^^^^

[deep mulch]

hmm

[valid jetty]

im very clear about this in the documentation

It is labelled as a "constant", because although it can return a different value (it can call any function), it cannot be redeclared.

[deep mulch]

that kinda seems like a weird choice

[valid jetty]

yeah but idk what keyword to put

[deep mulch]

function

[valid jetty]

no because its not a function per-se

its called automatically when you reference it

[deep mulch]

,

[valid jetty]

well yeah

[deep mulch]

getter then

[valid jetty]

its internally a function not semantically

but the difference here is that im not doing a pure operation

when you access a top level "thing" you know it can return something else

[deep mulch]

In computer programming, a constant is a value that is not altered by the program during normal execution.

[valid jetty]

idk what keyword to use

[deep mulch]

getter or function

[valid jetty]

use std/prelude;

const x = rand::random(0, 6);

fn main() {
    $dbg(x);
    $dbg(x);
    $dbg(x);
    $dbg(x);
}

use std/prelude;

fn x() { return rand::random(0, 6); }

fn main() {
    $dbg(x());
    $dbg(x());
    $dbg(x());
    $dbg(x());
}

[valid jetty]

a getter keyword is so horrible

[deep mulch]

i dont really see the issue with fn

[valid jetty]

its not a function

[deep mulch]

its easier to understand that way

its a function

[valid jetty]

you can define lambdas in constants

[royal nymph]

you will do lambas like kotlin @valid jetty

kotlin compiles lambas to classes and gives them insane names

[royal nymph]

look at the stacktrace here

[fleet cedar]

Why does the stacktrace have a sunset background

[valid jetty]

yea elle does that too lol

[royal nymph]

beautiful

[deep mulch]

why is it mangled

[royal nymph]

cause its Aliucord

[deep mulch]

did you enable proguard

[fleet cedar]

Lambdas do get ugly names yeah

[valid jetty]

it just calls the function lambda.{tmp_counter++}

[royal nymph]

the user who posted that screenshot was using that theme

[royal nymph]

NO

kotler named it that

[deep mulch]

i think newer kotlin its simpler

[fleet cedar]

/run ```java
Runnable r = () -> {};
System.out.println(r);
System.out.println(r.getClass());

[rugged berry]

Here is your java(15.0.2) output @fleet cedar

temp$$Lambda$192/0x0000000800c77a48@27fe3806
class temp$$Lambda$192/0x0000000800c77a48

[deep mulch]

java 15

husk

[fleet cedar]

Very beatufl

[valid jetty]

yea anyway

maybe not as ugly as java

[deep mulch]

@valid jetty Elle v3

best idea

[royal nymph]

/run ```kt
val l = { -> Unit }

fun main() {
println(l)
}

I forgot kotlin syntax

[rugged berry]

Here is your kt(1.8.20) output @royal nymph

() -> kotlin.Unit

[royal nymph]

is that even right

[deep mulch]

no

1.8.20 so OLD

FileKt$$Lambda$13/0x00007fbeb0003a20@1b2c6ec2

val l = {}

fun main() {
    println(l)
}

[royal nymph]

why is it just {}

[deep mulch]

i think its classname/address/hashcode

[royal nymph]

horror

[royal nymph]

INSANE

[deep mulch]

oh

[royal nymph]

Unit.a @deep mulch

[deep mulch]

its a lambda inside a lambda i think

idk

@royal nymph alibabacord

[royal nymph]

kotler so unsane for not having void

[deep mulch]

how

[fleet cedar]

Just want to add that c++'s handling of the void type is hilarious

And very very painful if you're doing template crimes

[still jolt]

isn't Unit void

[fleet cedar]

Yes, unit is what it's called in languages developed after type theory was invented

[still jolt]

sounds fun

[shrewd canopy]

void*? More like Any*

[fleet cedar]

That's C

C++ has std::any, and also templates

[ionic lake]

does qbe generate debug symbols?

[valid jetty]

i think you can get it to

there’s something like dbgloc as an instruction iirc

i haven’t looked into it yet

[deep mulch]

@valid jetty Elle VHDL

@valid jetty make Elle for FPGAs

[valid jetty]

what are either of those words 😭

[deep mulch]

hardware

like programmable circuitry

https://en.m.wikipedia.org/wiki/Field-programmable_gate_array

[fleet cedar]

Hardware isn't real

[strange matrix]

yeah it's hard 👍

[lavish frigate]

@valid jetty how are you not going insane making elle

[hoary sluice]

you cant go insane when youre already insane

[lavish frigate]

thats a good point

[strange matrix]

(since this is a prog question)

@tribal sapphire
how do i make getline() get from stdout ?

OH MY GOD

its supposed to be STDIN

😭
mfw im r*

immana' cry man why am i so dumb

[valid jetty]

you cant go insane when youre already insane

[median root]

wait so if the ascii table goes up to 127 why is it signed to go to -128 and why can it go up to 255 when unsigned

whats the actual use for that

wouldnt it be much more efficent to just have it limited to the ascii table size?

[fleet cedar]

If you're talking about C's char type, that's not a character

That's the smallest addressable size

Which is called char because... reasons I guess

[strange matrix]

people only needed 127 characters back in my days 👴 /jk

[median root]

yeah the char type.

why does it allow me to use values

that arent in the ascii set?

[fleet cedar]

Because it has nothing to do with textual data

[median root]

huh

[strange matrix]

wait doesnt char support unicode?

no wait i'm way off

[fleet cedar]

How would an 8-bit integer support unicode

[strange matrix]

would it at least support UTF-8 ?

[median root]

i didnt know there was this much to C data types until i saw "ungisned char" in wiki

and decided to google this

now it feels like im going down a sinkhole

[fleet cedar]

UTF8 is defined in terms of bytes, yes

Just like every encoding of everything ever, since cpus and storage both deal in bytes

[median root]

ive always assumed char was limited to 127 since if I was a prgorammer back when computers needed to be the most efficent things immaginable I would also have char limited to a 0-127

[valid jetty]

yes, but adressing things in 7 bits is inefficient

we want a byte not almost a byte

so its 8 bits instead

[fleet cedar]

Powers of two are good

[valid jetty]

yeah

[median root]

I mean you can fit a boolean there

[valid jetty]

you can

[strange matrix]

so there's no reason why it's called char for it to store a character?

[valid jetty]

people have used that bit for that purpose before

even i have

[strange matrix]

can i store a character in a int32_t ?

[deep mulch]

@valid jetty make Elle have 1024 bit numbers

[deep mulch]

yea

it would just have lots of padding

[median root]

in that case why not just make up a character and shove that in as 128

[strange matrix]

mind absolute blown

[valid jetty]

you can, but its pointless because ascii chars fit in 1 byte

[strange matrix]

(rlly)

[median root]

then have it 0-128

[strange matrix]

wb emojis

why am i asking i could literally try

[valid jetty]

yes, you can fit 1 whole unicode grapheme in an int32_t

[fleet cedar]

No you can't

You can fit one codepoint

[valid jetty]

1 byte for the length and 3 bytes for the actual data

you can

is that not the same thing

[deep mulch]

Rosie so smart

@valid jetty one day you will teach me everything you know

[fleet cedar]

I don't know if "grapheme" is a defined unicode concept

But grapheme cluster is

[strange matrix]

u smort
me dumb
pls no mad at me okay ? :3

[valid jetty]

rune, grapheme, codepoint

whatever you wanna call it

[median root]

then why the negative numbers? and why let it run up to 255 when unsigned? I swear thats just not-memory-efficent

[strange matrix]

i'm storing my strings in size_t from now

heap-maxxing

[fleet cedar]

Wdym

8 bytes can hold 256 different values

[strange matrix]

wait no its not in the heap

[fleet cedar]

-128-127 when signed, 0-255 when unsigned

[strange matrix]

stack-maxxing

[valid jetty]

if you make it unsized you can fit an extra 128-sized enum in the empty space by doing x-127

[median root]

why not just keep it unsigned? do negative numbers not make it really weird?

[fleet cedar]

If you want unsigned numbers, use unsigned numbers

[valid jetty]

well actually it is unsigned on some platforms lol

[fleet cedar]

If you want signed numbers, use signed numbers

[valid jetty]

its signedness depends on the platform

[fleet cedar]

The cpu doesn't care, it's just 8 bits

[valid jetty]

yeah ^

[median root]

eh fair enough.

[fleet cedar]

Types exist in programming languages, not cpus

[strange matrix]

but you could make cpus in programming languages

[valid jetty]

signed vs unsigned is just a difference of how the byte(s) are interpreted by the language

[median root]

"when i grow up im making my own char type which goes from 0-127 and cannot be signed, i will also force the cpu to store bools after each char if avalible for less fragmentation"

[deep mulch]

Elle will only have signed types

[valid jetty]

that’s not how fragmentation works

[median root]

yeah my concepts of it are really hazy

[strange matrix]

mm yes my bool goes from -1 to 0

[valid jetty]

storing a bool right after a char means if you want to retrieve that bool you would get a non-aligned pointer

[strange matrix]

isn't bool just an enum tho ?

genuine question

[fleet cedar]

It would also make memcpy not work

[valid jetty]

i mean i guess

[fleet cedar]

Semantically yes

But in most languages it's a primitive type

[valid jetty]

but an enum is also a number in most languages

[strange matrix]

how do languages

uh

that bind enum to strings or other types

like

ENUM : String

[valid jetty]

it can fit into 1 bit but usually we store it into 1 byte for alignment

[strange matrix]

store that

uh

like how does that work

[valid jetty]

enums are entirely compile-time sugar

[fleet cedar]

C is not most languages

[valid jetty]

when you do Enum.Foo it just substitutes the literal for that inplace

[strange matrix]

C is the father of most languages

so in preproc

it all gets

[valid jetty]

other than rust, name 3 languages where enums are not named constants

[median root]

oh yeah wait how do pointers actually work anyways? like when I ask the os for memory it gives me a pointer to the memory but then how do I know where my range of memory stops? Like im not given 2 pointers right?

[strange matrix]

like replaced?

[fleet cedar]

They are named constants in most, yes

They are not integers

[valid jetty]

you dont, youre the one to tell that to the computer

[fleet cedar]

In c/c++, discipline

[strange matrix]

chatting with absolute choding people and not knowing anything is amazing

[median root]

but does that not mean i could just overwrite some other programs memory space cuz I dont know where to stop?

[fleet cedar]

No, the OS typically won't let you

[valid jetty]

on older computers yeah

[fleet cedar]

It knows which memory ranges are assigned to your process and gets whiny if you violate that

[valid jetty]

but nowadays memory space is lazy and every program thinks it’s the only program that exists in terms of memory

and if you access something you’re not allowed to, you get a segfault

[median root]

so if I need to write something bigger than the space I have then what happens?

does it write over? does the os kill the process?

[deep mulch]

@valid jetty hiii

[fleet cedar]

You get a segfault

[median root]

ah those lovely things

[valid jetty]

when are you writing more than 16gb of memory

[fleet cedar]

Which typically kills the process, unless you have a funny signal handler

[valid jetty]

yeah

[median root]

when I suck at programming (or im using python)

[strange matrix]

System | OOMD handler ?

or i guess they're both the same

[fleet cedar]

Consider asking questions that are pareseable

[median root]

wait so the OS always gives the same amount of memory?

[strange matrix]

yes sir.

[valid jetty]

no, the OS gives an amount of memory at least as large as the amount you tell it

if you ask malloc for an amount bigger than it can give you, the pointer will be NULL.

and you try to use that pointer you will segfault

but if you don’t have enough memory left you have bigger problems

[median root]

oh and thats hidden from me in 90% of the languages I use

joverly

[valid jetty]

well yeah because usually you’re not doing manual malloc calls

[median root]

and im guessing having something like

char d[4];

just uses the stack instead calling for heap?

[fleet cedar]

Unless you ask for heap, you're not getting heap

C the language does not have a concept of a heap, that's a library thing

[median root]

then how can I still write to my char even if i have about 15 return adresses pushed onto the stack?

[deep mulch]

@valid jetty is made of pie

[median root]

does it all get popped off or does C remember the stack offset for me

[fleet cedar]

What do you mean

You can write to it since you took a pointer to that region and passed it down those 15 calls

[median root]

oh yeah but then how does the program know where the bit of memory stops? like how does it know my char stops there and my boolean starts here

and doesent read the boolean as part of the char?

[fleet cedar]

Discipline

If you try to read d[4] then you do indeed get your bool

[median root]

what so it just doesent stop reading from the stack?

[fleet cedar]

Or something else that happens to be nearby, stack layout is not guaranteed

[median root]

it starts at the pointer adresss and just keeps reading

until the os complains

[fleet cedar]

It reads exactly as much as you tell it to read

If you tell your program to read 8 bytes starting at d, then it reads 8 bytes starting at d

[median root]

wait what does that look like in C?

[fleet cedar]

*(uint64_t*)d for example

Or d[7]

[median root]

hmm

im starting to see why the rust borrow model is so appealing

[lucid trail]

It is, but it hides pointers from you really well

[median root]

im sure im gunna have fun in uni

[lucid trail]

I really didn't understand them until I did a project in C

[median root]

what about pointers is actually so horrible anyway?

from what I know they are just a memory adress

[fleet cedar]

They're very very easy to use wrong

[median root]

"use wrong"...

use wrong how

now im scared

[fleet cedar]

d[7]

[lucid trail]

7[d]

[fleet cedar]

Use after free, double free, null pointer, out of bounds, aliasing violations

[median root]

never seen double free or aliasing violations before

I always get hit by the segfault errors

[fleet cedar]

Double free isn't anything special, it's just a subcategory of UAF

[median root]

oh

aliasing violations?

[fleet cedar]

Compiler can often make much faster code if it knows two pointers are not the same

If you tell it that these are not the same but they are, then you have an aliasing violation

[median root]

im guessing this one isnt really done directly?

[fleet cedar]

C has noalias annotations which tells it that "these pointers do not overlap"

[median root]

so compiler flags can actually altogether stop this error?

since its an error purely thrown because the compiler tried to cut corners?

[fleet cedar]

C does not assume pointers are nonoverlapping by default

But things can get a lot faster if you tell it to

[median root]

oh yeah is compiler entropy a thing in every C compiler or is it just windows?

[fleet cedar]

Uh, wtf is that supposed to be

[median root]

like the compiler decides to change some bits of code becase... it can

ig

[fleet cedar]

That is complete nonsense

[median root]

sick alr imma go spend the rest of my night reading wiki pages

[lucid trail]

why would it do that

[strange matrix]

void trim_whitespace_entire_string(char *str);
void trim_whitespace_begin_string(char *str);
void trim_whitespace_end_string(char *str);

am i supposed to use something else?

[royal nymph]

wtf are those names

[lucid trail]

doesn't trim imply whitespace

[strange matrix]

im sorry 😭

[lucid trail]

trim_left

[royal nymph]

trim
trim_left / trim_start
trim_right / trim_end

[strange matrix]

trim_start
trim_end
trim
??

ig that good ?

no but more like

doesn't C have trim funcs ?

or glibc ?

[lucid trail]

C doesn't really have strings

[strange matrix]

right

[fleet cedar]

C has nothing

[lucid trail]

yeah

[fleet cedar]

You shouldn't use it

[strange matrix]

C is a standard i guess

glibc is the meat

or musl

[lucid trail]

it made me appreciate rust a lot more

[strange matrix]

am i right ?

[fleet cedar]

These days, C is an abi, nothing more

[strange matrix]

i'm contemplating which one to learn
with the goal of generally learning low-level computing

[fleet cedar]

In other words, how different languages can talk to each other

[strange matrix]

what would you suggest for me tho @fleet cedar ?

[fleet cedar]

I'm not a teacher, I don't make curriculums

[strange matrix]

you have experience

and you know a large amount

[lucid trail]

probably rust, but do some things in C too

[strange matrix]

oki

[hoary sluice]

LEFTUNPAD!!!

[gray heron]

chat why the fuck does using "*" for app.get not work and give me this shit

C:\Users\zander\Documents\GitHub\Parlix\node_modules\path-to-regexp\dist\index.js:73
            throw new TypeError(`Missing parameter name at ${i}: ${DEBUG_URL}`);
            ^

TypeError: Missing parameter name at 1: https://git.new/pathToRegexpError
    at name (C:\Users\zander\Documents\GitHub\Parlix\node_modules\path-to-regexp\dist\index.js:73:19)
    at lexer (C:\Users\zander\Documents\GitHub\Parlix\node_modules\path-to-regexp\dist\index.js:91:27)
    at lexer.next (<anonymous>)
    at Iter.peek (C:\Users\zander\Documents\GitHub\Parlix\node_modules\path-to-regexp\dist\index.js:106:38)
    at Iter.tryConsume (C:\Users\zander\Documents\GitHub\Parlix\node_modules\path-to-regexp\dist\index.js:112:28)
    at Iter.text (C:\Users\zander\Documents\GitHub\Parlix\node_modules\path-to-regexp\dist\index.js:128:30)
    at consume (C:\Users\zander\Documents\GitHub\Parlix\node_modules\path-to-regexp\dist\index.js:152:29)
    at parse (C:\Users\zander\Documents\GitHub\Parlix\node_modules\path-to-regexp\dist\index.js:183:20)
    at C:\Users\zander\Documents\GitHub\Parlix\node_modules\path-to-regexp\dist\index.js:294:74
    at Array.map (<anonymous>)

Node.js v20.18.0

[hoary sluice]

i hate i hate i hate

[strange matrix]

im sorry im sorry im sorry 😭

[hoary sluice]

musl so annoying

probably a great concept but stuff depends on glibc

i thought oh cool glibc alternative

wasted a day

[gray heron]

@nimble bone is nina nina-ing

[hoary sluice]

windows

[nimble bone]

hii its been a while

send your entire function

[gray heron]

there's no function just some server code

UNLESS U MEAN

[nimble bone]

whatever send your code

[gray heron]

app.get("*", (res) => {
  res.sendFile(path.join(__dirname, "public", "index.html"));
});

that's what's erroring out

[nimble bone]

first argument has to be the request afaik, 2nd has to be the response

i think

so do like (req, res)

unless i'm blatantly wrong

[signal oak]

The res argument should be req, res. It's missing the req (request) object.

Message URL

#🪅-progaming message

Question

Issue with this code?

[gray heron]

same error

😭

[nimble bone]

insane

[gray heron]

lemme send my file

[nimble bone]

if you change * for / does it work

[gray heron]

yep

[nimble bone]

try /*

[gray heron]

same error

same with /*/

[hoary sluice]

if you dont have a lot of experience with programming, especially with languages that dont give you everything for free at the cost of performance, you will have a hard time understending why rust does stuff the way it does; there are 2 extremes to this, a university i know starts their students with pascal because its simple, and many others start with python, which is one of the worst choices imo

you should learn c first, everything everywhere is compared to c, if you know how it works in c you can understand the comparison and will have a much easier time learning other systems languages

but dont spend too much time on it - dont master it

[gray heron]

webdev was easy to learn for me and the best option i think

[hoary sluice]

if you start a 14 year old (technical high school) on javascript, they will probably have a hard time finding the motivation to learn anything but webdev

my high school started with C, and after 2 years started oop with C++ and Java, and a bit of js at the end; nobody in that class can or wants to use anything else

and they were taught for 5 years

and in 2025 its quite likely they will just become an ai bro

[gray heron]

css was the bestlanguage first for me because it's what you can SEE the fastest and i learn visually and by doing

[strange matrix]

well for the past year (now it's over, i'm doing my final exams, 3 left) our programming was C#
which is kind of basically microsoft java (in terms of level)

so im not really beginning with python

i think my problem is more like
i don't stick to things

but i want to

[gray heron]

i went css>html>js>ts

i'll learn c one day but that shit is complicated as fuck and COMPILING BRO i cant

[strange matrix]

i think my problem today mainly with C was just
how am i supposed to know a functions name
because man pages are really
making my head spin

[royal nymph]

vibecode it

[hoary sluice]

see this is what i meant, its easy for someone who started with C to do webdev because its easier than what they started with, but hard the other way around

[strange matrix]

like it's fair for C to be hard for me, because i'm not a person that worked with memory since my experience was in C#

but like

i wanna

i want to try

i'm just mad at manpages

and well i'm just sometimes dumb
like setting getline(*buf, *buf_size, stdout)

[lucid trail]

Yes same exact, path, except i went to rust like a year ago

[shrewd canopy]

Python is very easy

[hoary sluice]

Its 53 years old and has barely changed since, it has some weird semantics because of how bad hardware was back then but you just have to deal with that

[shrewd canopy]

I went js then java then cpp then python

[strange matrix]

but is there some advice you could give for documentation ?

like

rust api docs are so easy

can search and everything

examples too

[gray heron]

😭 idk if u saw but i said it didnt fix

[strange matrix]

but manpages

well

[gray heron]

i've tried like everything

[hoary sluice]

i did scala > java > c > cpp > kotlin > rust

[gray heron]

yeah you did the whole midlevel programming pipeline

[hoary sluice]

probably just an llm

[strange matrix]

i should get ollama fr

[gray heron]

much different than the skilltree i fell down

[hoary sluice]

c is big enough for llms to properly document it

[hoary sluice]

minecraft pipeline

[gray heron]

real

[lucid trail]

Java to write minecraft mods fr

[hoary sluice]

at ~11 years old my dad dropped me into a scala repl, ran help and told me to make smth

[lucid trail]

i was forced to learn java against my will

[fleet cedar]

I started with java too (for minecraft)

[shrewd canopy]

I learned Java to make Bukkit plugins :D

[hoary sluice]

i think like 50% of all devs who started young started with minecraft

in the 2010s

[lucid trail]

I've only been programming for 1.5 years

is it over for me

[gray heron]

i started my life off sticking to science until i got a computer at 13 so it was a hard switch

it was after i lost my extreme love for mc modding

[shrewd canopy]

I started programming since 2019
Back then I used only javascript

[fleet cedar]

I'm not really sure how long I've been programming, but I had a backpack from an international programming competition labeled 2015, so probably started several years before that

[hoary sluice]

i have a hard drive with an early version of https://github.com/eagely/meowmod from 2020 so i started minecraft mods in 2019 or 2020, not sure about stuff before that

i think i did some scratch

and the scala thing

[gray heron]

@solid gazelle

[solemn ravine]

@woven mesa @young flicker I made a simple swiftpm package which changes your keyboard color

but I havent figured out how to change the predictive bar yet

it uses swizzling to force the color change

[woven mesa]

omg

so cool

what did u swizzle

[solemn ravine]

uikeyboard

@woven mesa we need more of this

[woven mesa]

im not as smar as u

[native spruce]

@woven mesa I think you can customize more than just the colors, I did something funny with the keys

[woven mesa]

so true

funny letters

[valid jetty]

https://www.youtube.com/watch?v=nkpo7XCJMUY

3d game in a 2d display in a 3d game in a 2d display in a 3d world 😭

[valid jetty]

tsoding server is so funny 😭

[gray heron]

@nimble bone i ended up doing some fucked up regex Lol, my discord clone "PARLIX" is coming along swimingly so far, very slowly tho

those things are attachments i'm gonna add a header or smthn and put the attachment count with it too

even has deletion

just realized the icon isnt aligned this is bothering me

there 😭

[valid jetty]

@hoary sluice is this true

this is the message im replying to

read it out-of-order i guess

[frosty obsidian]

completely agree 100%

theft machine 9000 that only makes brains become mush

[valid jetty]

i feel like ai shills going "we need to make {field XYZ} ai first!" and then people with experience in that field going "no?? dont do that wtf it ruins any enjoyment working in the field" is the biggest thing

the disconnect is huge

[valid jetty]

so true

[frosty obsidian]

i feel like we ran out of stuff to invent so we just resort to scraping the bottom of the barrel

[valid jetty]

lmao literally

i saw a video titled "Minecraft 1.8 is finally dead. We killed it" and i think it embodies that perfectly

it died because everyone eventually got bored of nothing new and moved to newer versions while at 1.8 its the same thing with some spin on it

same thing with ai except people arent updating

or well

the ai shills arent updating

actual software developers are

[frosty obsidian]

as tom scott once said, "people like a mix of novelty and familiarity"

change nothing, people get bored and move on

change everything, people hate it and move on

[valid jetty]

so true

people have become so familiar with ai that their brains have turned too much into mush to realize it

one day ai websites will be down and theyll be told "write this algorithm" and theyll have no idea because they used to just have ai write it, even if they could write it by hand in the past, ai polluted their brains and made them stupider "oh yeah why should i need to remember this information, ai will regurgitate it for me when i need it"

[frosty obsidian]

im just waiting for this to crash like the dotcom bubble

[valid jetty]

lmaooo

imagine @deep mulch working on hyperion without his trusty ais

[frosty obsidian]

imagine @deep mulch working on hyperion without his trusty ais

he left

[valid jetty]

no way he actually left

insane

[lucid trail]

true

I've found it useful for learning some things though

And AI is sooo bad at debugging, if it doesn't one shot something its basically cooked

[lucid trail]

i ordered a corne with an aluminum case, i'm so excited

[formal belfry]

how can i use css to change server icons from the client side?

[frosty obsidian]

you can use inspect element to find the class for the server icon

you can then do content:url("yourimages.com/url")

[pearl vine]

Replace the number with the server id and the link with your replacement icon

[data-list-item-id="guildsnav___1015060230222131221"] 
[class*=icon] {
  content:url("https://files.catbox.moe/39yx8f.jpg");
}```

[formal belfry]

may try this in the morning, thanks

[strange matrix]

if i want to ignore an error is it good to do this ?

int main(int argc/* ,char **argv */)

[fleet cedar]

Uh, what is that supposed to mean

[strange matrix]

i mean inline commenting out the part that is causing the warn

[fleet cedar]

What warning

Unused variable, or what/

[strange matrix]

yeah unused variable
both argc and argv
so i just commented them out

like this

int main(/*int argc, char **argv*/)

[fleet cedar]

If you do not use argc and argv, just do int main(void)

[strange matrix]

oki

[hoary sluice]

Why not mention typing assistant

Refactoring a 300 line match statement into separated functions takes a long time even with vim

llms solve that

[fleet cedar]

Yeah and insert a few subtle bugs in the process

[hoary sluice]

well no i can see what it did

and the task is simple enough for claude 4 not to hallucinate anything, its very good at moving code around

[ionic lake]

my take is that ai makes producing code cheap so if you need some new functionality you should scrap the existing implementation out and let ai do it anew with your new requirements

[hoary sluice]

thats not fun to do and everything will collapse quickly

[fleet cedar]

My take is that throwing out a working implementation in favor of ai slop is a terrible idea

[hoary sluice]

a decent part of my voice assistant is made with llms and while it works now, it was painful to implement and i wish i wrote more of it myself

[hoary sluice]

not always

if you remember my unsafe impl Send, Sync

i threw that out in favor of an ai impl with a separated thread for recording and it works

[valid jetty]

yeah, but 300 lines is like nothing

it’s almost always better to do it by hand even if it’s slightly inconvenient lol

[hoary sluice]

its not fun to extract 30 arms into functions

i did most of it by hand and got extremely bored

[valid jetty]

no, but when it’s done you know it’s correct because you wrote it

[strange matrix]

i feel like you should use LLM for when you don't know some sort of docs (at least for my case it's been like that.)

i feel like knowing the return value and func parameters are the only thing i'd really need an llm to answer me

[fleet cedar]

Why would you use a llm for that rather than docs

Unless your language docs suck ||c++||

[hoary sluice]

when i told claude 4 what to do, have an example of how its done (manually done 1 or 2) and its just moving around code, and it passes all tests, then i have nothing to worry about

[strange matrix]

well my brain too dumb for docs

[hoary sluice]

nix

[strange matrix]

the only good piece of docs i've seen so far was in a pdf

[fleet cedar]

That's an oxymoron

[strange matrix]

and even now i'm having a hard time finding it

hold on

[fleet cedar]

"Good docs" and "pdf" are mutually exclusive

[hoary sluice]

i dont think we should "give in to the vibes" but completely rejecting all llm code as slop is also less productive

[hoary sluice]

the atmega 32u4 datasheet is pretty good id say

and its a 450 page pdf

[fleet cedar]

Imagine how good it would be if it was in a useful format, then

Pdf is good for printing, nothing else

[hoary sluice]

using llms as advanced vim macros is completely fine imo

[strange matrix]

Here’s an example of how to use waitpid to get the status from all child processes
that have terminated, without ever waiting. This function is designed to be a handler for
SIGCHLD, the signal that indicates that at least one child process has terminated.

void
sigchld_handler (int signum)
{
int pid, status, serrno;
serrno = errno;
while (1)
{
pid = waitpid (WAIT_ANY, &status, WNOHANG);
if (pid < 0)
{
perror ("waitpid");
break;
}
if (pid == 0)
break;
notice_termination (pid, status);
}
errno = serrno;

}

[fleet cedar]

And who the fuck prints docs

[strange matrix]

this

[hoary sluice]

datasheets are printed very often so it makes sense for it to be pdf

one use case today is if you have an electronics exam and the student has to do something with an arduino, you might print the section of the datasheet that they need

and back when these were created these scenarios were more prevalent

[strange matrix]

mupdf doesn't like me copying text

so i have to OCR it with tesseract

😭

[hoary sluice]

there were schools/unis that didnt have computers for every student

so datasheets were made to be printable

some of them are garbage, some of them are awesome, i dont think pdf limits docs

there are good pdf docs and there are non-pdf Nix docs

software docs will almost always be a website, because writing software means you have a computer to look at them; if you are working with electronics, you might not have a computer and the docs will therefore be in a pdf to be printable, but this has nothing to do with how good they are

[valid jetty]

hey you can print websites

[fleet cedar]

Yeah and it usually turns out like shit

[hoary sluice]

yea go give your students a printed readthedocs page, they will laugh at you

ok no they wont but it wont be the best experience for them

[strange matrix]

  char *path_name = "/usr/bin/";
  char *prog_name = va_arg(args, char *);
  char *comb_str = path_name + prog_name;

why does this not work?
the first two are string literals so i can't add them up in like
prog_name itself

but then i try making a new string which combines those two
and it doesn't let me?

[valid jetty]

over time i learned that readthedocs aren’t real docs

[hoary sluice]

why

[valid jetty]

that’s um

not how C works

[hoary sluice]

it completely depends on whose readthedocs it is

[strange matrix]

😭 help me ma'am i stoopid

[hoary sluice]

null terminator stuff

isnt there a function to combine strings

strtok or smth

[strange matrix]

i haven't found it

[hoary sluice]

strcat

[strange matrix]

i found strcat

but it does it backwards

?

[hoary sluice]

then do it backwards

[valid jetty]

path_name is an address in memory, comb_str is an address in memory. adding them will give you an invalid address in memory. what you want to do, is to use malloc with strlen(path_name) + strlen(comb_str) + 1, then iterate both strings and set chars in your new string, and finally insert a null terminator (‘\0’) at the end of the str

[strange matrix]

hm

[valid jetty]

no not even

[valid jetty]

oh i didn’t know that existed

[fleet cedar]

That looks like a buffer overflow

[valid jetty]

how tf does that work

it can’t resize the string what is it doing

[hoary sluice]

its not a buffer overflow if you dont care

[fleet cedar]

It just overwrites whatever's after the string in memory

[hoary sluice]

probably overwrites starting at the \0

[strange matrix]

mmkay thank you :3

[valid jetty]

C is very low level. you don’t get almost anything for free :3

any feature you want, you have to implement manually

[strange matrix]

i mean, i kinda like that

but then i realize im stoopid

[valid jetty]

even as simple as string concatenation

[strange matrix]

and i don't know how to

yeah rexim built a whole strbuilder

[valid jetty]

and then you learn !!

[strange matrix]

i might do that atp

[valid jetty]

oh yeah that’s just an abstraction for this lol

it’s just a dynamic array you can push to

[hoary sluice]

c is built for use in 1972, not 2025

[strange matrix]

so is this safe then ?

  char *comb_str;

  comb_str = strcat(path_name, prog_name);

[valid jetty]

i don’t think so

[strange matrix]

i still don't even know if i'm supposed to use free() or not

[valid jetty]

nop, that’s allocated on the stack

[fleet cedar]

That's fine if path_name has enough space to hold the concatenation

[valid jetty]

which it doesn’t because it’s a string literal

[strange matrix]

would it be okay if i just allocate 1024*4 for all of these ?

[valid jetty]

yeah probably

[hoary sluice]

youre overwriting strlen(prog_name) - strlen(path_name) bytes of memory

if its 0 then its fine

but theres probably a safer way

[strange matrix]

its probably not 0

[valid jetty]

this is probably the simplest way???

char *comb_str = (char *)malloc(strlen(path_str) + strlen(prog_name) + 1);
strcpy(comb_str, path_str);
strcat(comb_str, prog_name);
``` and get out a null terminated string

[hoary sluice]

buffer overflow

strncat?

[valid jetty]

well sure

[hoary sluice]

with error detection

[valid jetty]

there

safe

[strange matrix]

wait but this

char *comb_str = (char *)malloc(1024);

is it the same as this :

char *comb_str = malloc(1024 * sizeof char);

??

[valid jetty]

yes

[fleet cedar]

Nope, that's ub

[valid jetty]

on most platforms the size of a char is 1 byte

[fleet cedar]

You need comb_str[0] = 0; first

[valid jetty]

but you memcpy into it

[fleet cedar]

No, you strcpy

[valid jetty]

that inserts a null terminator for you

[valid jetty]

that’s what i meant

[fleet cedar]

Which means it skips past any non-null bytes and then pastes it there

Which can be at an arbitrary point since malloc doesn't zero

[hoary sluice]

why

[valid jetty]

wait actually???

so then can you just zero out the memory and use strcpy twice ???

[hoary sluice]

isnt strcpy the same as strcat then

[fleet cedar]

Wait mb misread strcpy as strcat

[valid jetty]

lmao

ok

[hoary sluice]

oh

[hoary sluice]

but now what if accessing prog_name inside the comb_str assignment caused a side effect

and made prog_name grow by 2MB

[valid jetty]

what

[strange matrix]

yeah i'm getting segfaults

now i know i'm doing stuff wrong

great indication

[hoary sluice]

do you not remember yesterdays conversation

[valid jetty]

did i put it the wrong way round

[valid jetty]

no i remember

but nothing should be growing

[hoary sluice]

i dont think field access can make side effects in c

[valid jetty]

it can’t

and if it’s a good c spec, strcat will accept a const char*

[strange matrix]

oh my god i so stoopid i make malloc for a string i already know the length of

[hoary sluice]

dont think that causes a segfault

[strange matrix]

no but

just live reporting my dummyness

:3

makes me remember later

if a char is 1 byte

then i don't need to

i could just do malloc(1024)

right ?

[strange matrix]

am i right ? ^^

[valid jetty]

malloc(1024) will break if strlen(path_name) + strlen(prog_name) > 1024

[valid jetty]

look at this

[hoary sluice]

why specifically 1024

[strange matrix]

no specific reason

[hoary sluice]

do you know that path_str + prog_name + 1 will never be longer than 1024?

if not, malloc according to their lengths

[strange matrix]

well i don't know how to keep track of it

[hoary sluice]

.

[strange matrix]

but its user input

[valid jetty]

you just take their length lol

[strange matrix]

it would change

you're telling me i can malloc twice ?

[valid jetty]

the way you keep track of it is you take its length

[valid jetty]

uhhh you can but it’s not necessary here

[strange matrix]

no look

user input comes in

ohhhh

wait

[hoary sluice]

yes you can

[strange matrix]

no im dumb

okay

[hoary sluice]

youre mallocing another variable

@valid jetty can i

char *foo = malloc(17);
foo = malloc(26);

theres nothing wrong with that right

itll just get 26 new bytes somewhere else

[strange matrix]

char *prog_name = (char*)malloc(strlen(va_arg(args, char *)));

is this right now ?

[fleet cedar]

What are you doing with that va_arg btw

[strange matrix]

um

so i

let me show you

[fleet cedar]

Well you leak memory but

[strange matrix]

void execute_prompt(int n, ...) {
  va_list args;
  va_start(args, n);

  char *prog_name = (char*)malloc(strlen(va_arg(args, char *)));

  int status = 0;
  char path_name[10] = "/usr/bin/";
  prog_name = ;
  char *comb_str;

  comb_str = strcat(path_name, prog_name);
  // strncat(prog_name, path_name, strlen(path_name));

  for (int i = 0; i < strlen(comb_str); i++) {
    printf("%c\n", comb_str[i]);
  }

  // pid_t pid = fork();

  // if (pid == 0) {
  //   /* This is the child process. Execute the shell command. */
  //   execl(prog_name, NULL, NULL);
  //   _exit(EXIT_FAILURE);
  // } else if (pid < 0)
  //   /* The fork failed. Report failure. */
  //   status = -1;
  // else
  //   /* This is the parent process. Wait for the child to complete. */
  //   if (waitpid(pid, &status, 0) != pid)
  //     status = -1;
  va_end(args);
}

[hoary sluice]

and a lot of other issues

like no header

but my question was if that segfaults or not

im guessing not

[fleet cedar]

No, assigning a variable doesn't segfault

[hoary sluice]

bold statement

i wouldnt be too sure

[hoary sluice]

@valid jetty maybe assigning to that variable creates a side effect that segfaults

[valid jetty]

use realloc

[hoary sluice]

that was not the point

[valid jetty]

it wouldn’t do anything

it just makes the original memory before you reassigned unusable

[hoary sluice]

yes i know

[valid jetty]

it doesn’t free the previous memory it just overwrites what the pointer in your variable holds

[hoary sluice]

its cause of this

i was saying

[valid jetty]

yea

[hoary sluice]

theres nothing wrong with mallocing twice right

like that codeblock i sent

[fleet cedar]

malloc is just a function

[hoary sluice]

obviously its semantical nonsense

[valid jetty]

yea no there’s nothing wrong with it it just makes no sense

[hoary sluice]

it might have side effects if the program was written by rosie

[strange matrix]

um, @fleet cedar ?

[fleet cedar]

So you're doing a varargs function that takes exactly one extra arg

Sure, fair enough

[strange matrix]

😭

well i'm supposed to take

multiple

for argv of the program that is supposed to be ran

probably two actually

two strings

good thing you told me

[strange matrix]

*** stack smashing detected ***: terminated

what does this mean now?

[fleet cedar]

It means your buffer overflow corrupted the return address or similar

[strange matrix]

let me gdb it

[spark tiger]

help

[fleet cedar]

You are beyond saving

[spark tiger]

im saved

deleted some games

[strange matrix]

demonic amount of free space

[spark tiger]

tru

[strange matrix]

i have to implement vectors in C

now

no nvm i don't need it