🪅-progaming

[jade stone]

i think that's called powershell

[winged mantle]

[valid jetty]

😭

[split timber]

Java clases can have association, inheritance, dependency, composition, etc these are ways that created objects interact with other objects

[valid jetty]

yeah

[fleet cedar]

Sounds like something someone very bored made up

[winged mantle]

can we have a dynamic version of javascript called javascriptscript

[split timber]

It's literally java API

[valid jetty]

isnt that literally all of object oriented programming

[split timber]

^

[winged mantle]

java.applet.Applet api my beloved

[split timber]

I love oop so much

[valid jetty]

wait so then @royal nymph why have generics at all are they literally just like "type hints" in python

[fleet cedar]

I deal in programming, not bored people's fantasies

[winged mantle]

means you don't have to cast everywhere

[split timber]

It's programming 💔💔💔

[valid jetty]

oh i think i see

[winged mantle]

without generics you need to do this

List list = new ArrayList();
String thing = (String) list.get(0);

[valid jetty]

😭

[royal nymph]

they are still meaningful and better than using Object aka any type that you always have to cast

[winged mantle]

I believe it's essentially syntax sugar for that

[valid jetty]

to be fair it kinda makes sense that its fake because java is a dynamic language

just like how typescript types are fake

[winged mantle]

java does not allow duck typing

only for generics (kinda)

as soon as you try to get any data it fails though

you don't even have to assign it

i don't think type confusion is really possible in java without extreme hackery

[royal nymph]

yeah it isn't

you'd need to use reflection or unsafe

[winged mantle]

i don't think reflection will allow you to pass in the wrong type

i think to break java's memory safety you would need unsafe

[fleet cedar]

I've accidentally passed invalid types to interface-typed parameters in bytecode

It's not caught until you try to call a method

[winged mantle]

but it still wouldn't break the memory safety, you would get an exception

[fleet cedar]

Yep

[royal nymph]

yeah that it can't

[winged mantle]

is this with methodhandles

[royal nymph]

but reflection can cause a lot of other cursed stuff

you can write final fields with reflection

[winged mantle]

i was about to say

i remember doing that with some library

[fleet cedar]

But writing to static final fields of primitive/String type does nothing

Since they're inlined

[winged mantle]

using reflection on the reflection api to trick it that it's not final

but java 9 broke

[royal nymph]

also with usafe you can do shit like instantiate classes without calling their constructor

but once you're using unsafe you're down the deep end anyway

[royal nymph]

yes

[royal nymph]

I wrote this ages ago https://github.com/Aliucord/Aliucord/blob/main/Aliucord/src/main/java/com/aliucord/utils/ReflectUtils.java#L235-L253

[pearl stag]

ReflectUtils.java: Lines 235-253

public static void setFinalField(@NonNull Class<?> clazz, @Nullable Object instance, @NonNull String fieldName, @Nullable Object v) throws NoSuchFieldException, IllegalAccessException {
    if (accessFlagsFields == null) {
        try {
            accessFlagsFields = Field.class.getDeclaredField("accessFlags");
        } catch (ReflectiveOperationException ignored) {
            try {
                accessFlagsFields = Field.class.getDeclaredField("modifiers");
            } catch (ReflectiveOperationException ex) {
                throw new RuntimeException("Failed to retrieve accessFlags/modifiers field", ex);
            }
        }
        accessFlagsFields.setAccessible(true);
    }

    var field = clazz.getDeclaredField(fieldName);
    field.setAccessible(true);
    accessFlagsFields.set(field, field.getModifiers() & ~Modifier.FINAL);
    field.set(instance, v);
}

[winged mantle]

i thought java 9 broke this

[royal nymph]

this is for androd

[winged mantle]

wait is android s tuck on java 8

[royal nymph]

you mean jdk I assume

java has no relevance in this

[winged mantle]

jre and jdk

[royal nymph]

the runtime is what matters

Android doesn't use jdk

idk much about jdk

jdk is cringe and restricts reflective access to internals

[winged mantle]

jdk is not a single implementation anyway..?

[fleet cedar]

Openjdk is an implementation

[winged mantle]

isn't there hotspot and j9 and stuff

idk i have expert vague understanding of java

[valid jetty]

wtf is Class<?>

[fleet cedar]

Same as Class<Object> but less type safe

[winged mantle]

isn't ? short for ? extends Object

so it accepts anything other than Object

or is that wrong

[royal nymph]

Class is the parent type of all classes (the class itself, not its instances)

[winged mantle]

[tired vigil]

Why are java classes instances of a generic Class class

[royal nymph]

<> is generic

[fleet cedar]

Uh, what

[royal nymph]

<?> is like any generic

[winged mantle]

wait that has to be wrong

i swear i had issues where i could not put an <Object> value into a <?> value

😭

[winged mantle]

the Class class is like the metaclass for classes

[royal nymph]

actually not even that

it's hard to explain lol

[winged mantle]

I don't think there are any subclasses of the Class class

it's generic

[fleet cedar]

Yes, java.lang.Class is final

[winged mantle]

String.class is an instance of java.lang.Class

not a subclass

the type is Class<String> though

i am kinda rusty but i think that's right

[fleet cedar]

Class<String> is a subtype of Class<?>, but not a subclass

[winged mantle]

i wouldn't consider it a subtype so much as a type which fits into it (assignable type)

[fleet cedar]

Yes, that's what subtype means

[royal nymph]

Class<String> is like a reflective representation of the String class

Class<String> stringClass = String.class;

you can use it to query information about the class, look up methods, constructors, etc

it allows you to write dynamic code for example

[winged mantle]

i would call it a metaclass

class is a class for a class

[royal nymph]

as in call methods from strings

[winged mantle]

[royal nymph]

load classes from strings

[winged mantle]

ClassLoader.defineClass my beloved

[tired vigil]

wait so why is it generic when generics are fake

[winged mantle]

it allows things to be typed

[fleet cedar]

Generics pretend to be real

[winged mantle]

i think it means you can call the constructor and not need to cast

[tired vigil]

insane behavior

[royal nymph]

nah you have to

[winged mantle]

/run

public class Main {
    public static void main(String[] args) {
        "".__proto__.toInt = Integer::parse;
        System.out.println("4".toInt());
    }
}

java pro tip

[rugged berry]

@winged mantle I only received java(15.0.2) error output

file0.code.java:3: error: cannot find symbol
        "".__proto__.toInt = Integer::parse;
          ^
  symbol:   variable __proto__
  location: class String
file0.code.java:3: error: method reference not expected here
        "".__proto__.toInt = Integer::parse;
                             ^
file0.code.java:4: error: cannot find symbol
        System.out.println("4".toInt());
                              ^
  symbol:   method toInt()
  location: class String
3 errors
error: compilation failed

[royal nymph]

actually you don't have to cast yeah

[winged mantle]

v you should take an iq test

[royal nymph]

why

[winged mantle]

java iq test

[royal nymph]

Am I that stupid..

[winged mantle]

i remember somebody being wrong about c++ and i ended up making like 4 example programs to show each time they got something wrong

[valid jetty]

so ```java
void foo(Class<?> clazz) {}

is synonymous with
```java
void foo<T>(Class<T> clazz) {}
```???

[winged mantle]

[winged mantle]

<T> void foo(Class<T> clazz) {}

[valid jetty]

close enough idk java syntax lol

[winged mantle]

you don't write generics like that

[valid jetty]

but is the concept right

[winged mantle]

:german:

[royal nymph]

that's not how u declare generics in java 🤓

[valid jetty]

[royal nymph]

also yeah in usage they're the same but they are still different

[valid jetty]

so ```java
void foo(Class<?> clazz) {}

is synonymous with
```java
<T> void foo(Class<T> clazz) {}
```???

[winged mantle]

can you readd the german emote

[royal nymph]

did someone delete

or was it never here

[winged mantle]

i swear it was here

somebody sent when i said i was using openbox wm with no panel

[royal nymph]

they're not really synonymous no but in this case the method is called the same

the first method isn't generic at all

[valid jetty]

does java have covariance

[winged mantle]

<T> T make(Class<T> clazz)

this would make it useful

[valid jetty]

wait so ? is only a placeholder in arguments?

[fleet cedar]

Yes, List<? extends Foo> is covariant while List<? super Foo> is contravariant

[royal nymph]

? is like a generic wildcard

[winged mantle]

contravariant

[fleet cedar]

Java generics are cursed and half baked

Don't try to make sense of them

[valid jetty]

but isnt <T> with no constraints also a generic wildcard???

why does ? need to exist

[royal nymph]

no it isn't

[winged mantle]

so you don't need to add a generic param to the method?

[royal nymph]

T is a generic type

[fleet cedar]

How would you declare a List<T> in a non-generic function

[royal nymph]

you can use ? anywhere

[valid jetty]

hmm i see

[royal nymph]

List<?> list = idgafWhatsInThisList();

[valid jetty]

ok that makes sense

[winged mantle]

how does it differ from List<OBject> this has always confused me

[valid jetty]

im thinking of this from a static typing standpoint but i forgot java doesnt actually care about the inner type of generics

[royal nymph]

cause you have to cast to List<Object> don't u?

[placid cape]

i think you do

[winged mantle]

i swear this did not work last time itried

[royal nymph]

/run ```java
import java.util.*;

List<Object> list = new ArrayList<Integer>();

[rugged berry]

@royal nymph I only received java(15.0.2) error output

file0.code.java:4: error: incompatible types: ArrayList<Integer> cannot be converted to List<Object>
List<Object> list = new ArrayList<Integer>();
                    ^
1 error
error: compilation failed

[royal nymph]

guh it doesn't import

wait what does that work

[winged mantle]

because boxing?

[royal nymph]

yeah see

you have to cast to List<Object>

[winged mantle]

horror why

/run

System.out.println(System.getProperty("java.version"));

[rugged berry]

Here is your java(15.0.2) output @winged mantle

15.0.2

[winged mantle]

java 15

[valid jetty]

/run

import java.util.*;

List<?> list = new ArrayList<Integer>();

[royal nymph]

cause that's not safe

[rugged berry]

Your java(15.0.2) code ran without output @valid jetty

[winged mantle]

ohh

[valid jetty]

ok that makes sense

[winged mantle]

trueee

[valid jetty]

i get it now

[winged mantle]

but

how does ? make it safe

oh wait

no it is safe

[royal nymph]

Object is like any in typescript (but you have to cast to get to it)
? is like unknown

[valid jetty]

yeah

[winged mantle]

why would that not be safe wat

Object does not implicitly cast 😭

[royal nymph]

cause you can assign anything to object

[royal nymph]

List<Object> list = intList;
list.add("Hi");

the second line is valid

if the cast was allowed this code would pass but fail at runtime

[winged mantle]

wouldn't it be valid with ? too though

[royal nymph]

no

I think

/run ```java
import java.util.*;

List<?> list = new ArrayList<Integer>();
list.add("hi");

[winged mantle]

/run

List<Integer> intList = Arrays.asList(1, 2, 3, 4);
List<?> list = intList;
list.add("hi");
System.out.println("bazinga");

[rugged berry]

@royal nymph I only received java(15.0.2) error output

file0.code.java:5: error: incompatible types: String cannot be converted to CAP#1
list.add("hi");
         ^
  where CAP#1 is a fresh type-variable:
    CAP#1 extends Object from capture of ?
Note: Some messages have been simplified; recompile with -Xdiags:verbose to get full output
1 error
error: compilation failed

@winged mantle I only received java(15.0.2) error output

file0.code.java:2: error: cannot find symbol
List<Integer> intList = Arrays.asList(1, 2, 3, 4);
^
  symbol:   class List
  location: class temp
file0.code.java:2: error: cannot find symbol
List<Integer> intList = Arrays.asList(1, 2, 3, 4);
                        ^
  symbol:   variable Arrays
  location: class temp
file0.code.java:3: error: cannot find symbol
List<?> list = intList;
^
  symbol:   class List
  location: class temp
3 errors
error: compilation failed

[royal nymph]

CAP#1

my beloved

yeah it doesn't let u do it

[winged mantle]

i swear i never saw an error saying CAP

i think it was more verbose

idk

i useed eclipse ide for years

i remember now getting errors about captures

[royal nymph]

/run ```java
import java.util.*;

List<Object> list = (List) new ArrayList<Integer>();
list.add("hi");

[rugged berry]

@royal nymph I only received java(15.0.2) error output

Note: file0.code.java uses unchecked or unsafe operations.
Note: Recompile with -Xlint:unchecked for details.

[winged mantle]

fuire

[royal nymph]

yeah see that works

[winged mantle]

that's because you're duck typing

[royal nymph]

yes but this is why you have to explicitly cast to List<Object> xd

[winged mantle]

i understnad now

[royal nymph]

List<?> is safe, List<Object> isn't

[valid jetty]

@hoary sluice do you know about higher ranked trait bounds

https://doc.rust-lang.org/nomicon/hrtb.html

[winged mantle]

lol I am remembering this java project I made
(I was trying to create a fully fledged piece of software when i was like 15 )

it comes with solarized dark theme...

wait did i really create a http library too

[valid jetty]

i can believe that you manually typed out every single http code's name

[winged mantle]

probably

I USED SNAKE CASE BACK IN JS BACK IN 2020 TOO

chronic snake case user

[supple whale]

who said manually

this is where vim-like shortcuts rock

[winged mantle]

i was using eclipse ide

[supple whale]

https://cdn.discordapp.com/emojis/725734532111794246.webp?size=48&name=yikes~1

[winged mantle]

this was during lockdown

[valid jetty]

i used to do that too lol

[winged mantle]

i was probably like 13 or 14 back then

[supple whale]

i just use whatEverThisIs

[winged mantle]

camelCase

[valid jetty]

camelCase

[winged mantle]

i made so many webapps with xss

[valid jetty]

kebab-case

PascalCase

SHOUTING_SNAKE_CASE

snake_case

[winged mantle]

i basically just did fs.readFileSync("filename.html").replace("$CONTENT", content).replace("$USERNAME", usernamew)

[valid jetty]

objc_CursedCaseAkaMixedCase

[winged mantle]

and i parsed cookies manually

the golden age of thekodetoad

[royal nymph]

this is where ai is goated

[winged mantle]

these days my github followers fluctuate between 99 and 100 i don't get it

back then i had no followers

when somebody followed me i felt like stalking was a more appropriate word

[royal nymph]

or just copy paste a table from Wikipedia or smth and use ide shortcuts to turn it into code

@winged mantle I used ai to refactor venapp code to jsx

copiloter

[winged mantle]

HORROR

i would do it manually or use a jscript with regex

i remember one time using xdotool to export many images from gimp

[placid cape]

[valid jetty]

i love this bug

[fleet cedar]

Good ol' cve-rs

[valid jetty]

true

😭

[winged mantle]

would a good architecture for a jason parser be to have a function to read the next token and it returns a union of boolean, number, string, array start, array end, object start, object end, comma

or is that dumb

[supple whale]

a fucking what parser?

[winged mantle]

jason parser

[fleet cedar]

That's how most json parsers work, yeah

Don't know about Jason parsers

[winged mantle]

i mean json

[supple whale]

pov: jason parser

[winged mantle]

json ctron

[winged mantle]

ugh it feels weird that comma would be token that's all

[fleet cedar]

The specific list of tokens varies from parser to parser though

[valid jetty]

https://github.com/tsoding/jim

[winged mantle]

jim

immediate mode

[valid jetty]

tsoding 😭

[winged mantle]

what would a string token look like

would it contain the escape codes

[fleet cedar]

Depends on how zerocopy you want to be

[winged mantle]

or maybe you'd then tokenise that separately

[valid jetty]

trolley~1

[fleet cedar]

The obvious way would be "Foo\nBar" as ```
StrStart,
StrChunk("Foo"),
StrEsc('\n'),
StrChunk("Bar"),
StrEnd,

[winged mantle]

that means nested tokens though

which feels kinda 🥴

[fleet cedar]

No

That is a sequence of five non-nested tokens

[valid jetty]

you should be doing whitespace trimming in your lexer anyway lol

[winged mantle]

oh wait

i see ty

[valid jetty]

whitespace denotes separation but should be removed when actually parsing

[winged mantle]

what if json had significant whitespace

wait

[valid jetty]

so toml

[winged mantle]

what if you wrote a python object notation parser

pon

[valid jetty]

pon

[winged mantle]

jon

i already had the jon idea

parse syntax like

new Map() {{ put("key", "value"); }}

that would be so fire

wait can you do Map.of

[winged mantle]

chloe moment

[lavish frigate]

MAD cuz jealous

[formal belfry]

how do i access

[lavish frigate]

gave you perms

[winged mantle]

is it normal to have structs of function pointers in c

just like interface

[formal belfry]

WHY IS NEMMY THERE

insane

[fleet cedar]

Sure, manually implemented vtables are pretty common

[lavish frigate]

our resident idiot <3

[winged mantle]

struct InputSource {
char (*read)();
}

[lavish frigate]

we have to balance out the rust love with stupidity sometimes

dw about it

[valid jetty]

can i have access too 💔

pls

[formal belfry]

rosie is like 12

[winged mantle]

and then make a FILE based InputSource which implements buffering

and String which just moves cursor

[lavish frigate]

you arent in the server smh

[winged mantle]

that doesn't need a struct though

that could just be a single function

[valid jetty]

what server—

[winged mantle]

—

[valid jetty]

evil

[formal belfry]

—

[valid jetty]

—

[winged mantle]

絵希[猫ART]
— 15:48
—

[lavish frigate]

i forwarded the message from another server 😭

[formal belfry]

Roie 🌻

[valid jetty]

i know 😭

[winged mantle]

click here: [🔒 No Access]

[valid jetty]

絵希[猫ART]
— 15:48
—

[winged mantle]

maybe i should call this lib jay instead

people might think i'm weird for being obsessed with jason citron

[pearl stag]

Emoji not found. The emoji set chosen might not have this emote as an image.

[visual shell]

You need to specify an emoji to enlarge.

[winged mantle]

😭

hf!jumbo :jasonzoom:

[pearl stag]

Emoji not found. The emoji set chosen might not have this emote as an image.

[winged mantle]

horror

vprune by bots 3

https://images-ext-1.discordapp.net/external/r1JAAY6pUZEeGGo2XwLmM9osD1G1zWmS0Iw5HgsKrEE/%3F_%3D.gif/https/cdn.discordapp.com/emojis/1099771124172210306.gif?width=96&height=126

guh why is before: 2021 barely doing anything

i wanna filter ai slop

searching about tokenization has loads of ai results

[valid jetty]

watch the tsoding combinators video

https://www.youtube.com/watch?v=N9RUqGYuGfw

[winged mantle]

why not split number into tokens

why not have exponent as token

i have already written a json parser before i just did not conceptualise and separate the process of tokenisation

[fleet cedar]

What would be the benefit

[winged mantle]

whaty's the benefit of tokenisation for something as simple as json

the point is numbers have separate parts

the negative sign

the exponent

[fleet cedar]

What would be the benefit of that, other than making parsing harder?

[winged mantle]

well. it feels weird to do this for strings but not numbers

but fair

i guess at the tokenisation stage you need to recognise escapes in strings

with numbers you can just read until the next char which is not relevent

thank you for all the advice

is it bad to do barrel file in c

oh wait

[fleet cedar]

Do you need special woodworking tools for making barrels??

[winged mantle]

just separate .c files

i mean like json.h file which includes all headers

like barrel file in js

file which does nothing apart from reexport

c modules support when

cpp modules are fire

it just occured to me that in c i don't think you can bind functions

[fleet cedar]

Can't do much without a self yeah

[ornate quiver]

thank

[supple whale]

welp i made gemini shit itself

even AI cant handle service workers

https://cdn.discordapp.com/emojis/722917812649197711.gif?size=48&animated=true&name=PepeLaugh~3

telling it, try again or 12 orphans will die fixes the problem

https://cdn.discordapp.com/emojis/1327371909385555989.gif?size=48&animated=true&name=classic

[ornate quiver]

i had gemini keep failing last night with the same thing

no matter what i typed

[deep mulch]

sorry

[winged mantle]

c would be so much better if strings were like this

struct str {
    int len;
    char *d;
}

[fleet cedar]

Rust says hi

[winged mantle]

this would basically be like string slice

[valid jetty]

what’s stopping you from just making this

[fleet cedar]

Or std::string_view

[winged mantle]

at first i thought of

struct str {
    int start;
    int len;
    char *d;
}

but srting slices aren't generally aware of the whole string, are they?

[deep mulch]

roieeee

[valid jetty]

you can just offset the pointer

you don’t need to store start separately

[winged mantle]

and that also would be annoying

[fleet cedar]

If you did that you'd need to store the original length as well

[valid jetty]

not..really?

[winged mantle]

oh yeah

ofc

[fleet cedar]

That'd be messy and useless

[winged mantle]

you would need to do d + start

[valid jetty]

create a new str struct with the offset pointer

and a len-start len

[winged mantle]

i think just length and string is fine

wait how does string_view in cpp work

what if the source string gets deleted

[fleet cedar]

Well then you get a uaf

Just another day in c++

[winged mantle]

makes sense

[valid jetty]

string_view is like &str where string is String

[winged mantle]

i was thinking you could add to reference counter.. but c++ strings are like unique_ptrs

[valid jetty]

if you know rust

[fleet cedar]

The standard doesn't guarantee that

Some std::string implementations do indeed refcount I think

[winged mantle]

i think qstring does

i have mainly programmed with qt

why is there no | NULL type in C

(this is a joke)

[valid jetty]

there’s is, it’s called a raw pointer

🎀

[winged mantle]

🎀

[deep mulch]

🎀

[winged mantle]

but the problem with using your own strings in the stdlib does not support

[fleet cedar]

That's ok, c++ stdlib is crap anyway

[winged mantle]

C

[fleet cedar]

C barely has a stdlib

[winged mantle]

ig i could implement my own stdlib

that would be a cool project

[deep mulch]

@valid jetty what is starting nginx on my server

[valid jetty]

#define str_to_cstr(x) (str_append(x, '\0'), x.d)

[winged mantle]

mutates...

[ionic lake]

do you have systemd? it could be that

[winged mantle]

why is there no man entry for bufsiz

[fleet cedar]

Wtf is bufsiz

[winged mantle]

you mean i can't just use man as C google

[winged mantle]

it's your siz of your buf innit

[deep mulch]

yop it was a service

[winged mantle]

buffer size

[ionic lake]

nice

[winged mantle]

lol public exec endpoint
https://horizon3.ai/attack-research/disclosures/unsafe-at-any-speed-abusing-python-exec-for-unauth-rce-in-langflow-ai/

[supple whale]

nah its way more than that

that endpoint is giga sanitised, they just overlooked that python decorators can run code

because python

[frosty obsidian]

don't use ai garbage and that won't happen

[supple whale]

again if you read the code for more than a quarter of a second then you'd see its not ai garbage

[frosty obsidian]

no im calling langflow as a product garbage

[supple whale]

they had to work around dogshit problems in python with dogshit python solutions

just because python's ast doesnt process imports correctly

woeful

[winged mantle]

the fact is has comments is a red flag

a real experienced programmer human would write this confusing code with no comments

i have no idea if this is undefined behaviour

maybe i should just allocate buffer separately

[hoary sluice]

wtf is this

[winged mantle]

seems like fgetc buffers for you?

even java does not do that

[hoary sluice]

@valid jetty we need to hold a who knows more italian brainrot animals competition

[hoary sluice]

@valid jetty im scared, i vibe fixed my parser and it works now

on monday i will wake up and my parser will have exploded

woohoo you can remove the parens now

shiftLeft x n = x * 2 ** n
shiftRight x n = x / 2 ** n 
shiftRight (shiftLeft 2 8) 8

instead of

shiftLeft x n = x * 2 ** n
shiftRight x n = x / 2 ** n 
(shiftRight (shiftLeft 2 8)) 8

@fleet cedar @valid jetty i remember some language had a parser that had significant whitespace for unary -, so foo - x would be a binary op and foo -x would be passing -x to the function foo

is that or

[fleet cedar]

Haskell does that with an optional feature flag

[hoary sluice]

apparently like all of the popular functional languages do that

[hoary sluice]

isnt it the default?

i cant find a lot of info on it

[fleet cedar]

https://github.com/ghc-proposals/ghc-proposals/blob/master/proposals/0229-whitespace-bang-patterns.rst

I can't find anything about whether that's on by default though

Oh, here https://downloads.haskell.org/ghc/latest/docs/users_guide/exts/negative_literals.html#extension-NegativeLiterals

Or https://downloads.haskell.org/ghc/latest/docs/users_guide/exts/lexical_negation.html#extension-LexicalNegation

[hoary sluice]

yea should i do this

[winged mantle]

i love reading glibc source code to work out the best thing to do

https://github.com/bminor/glibc/blob/21e54f160f6245f959cca1f48bad9cca487c2570/libio/genops.c#L326-L349

[pearl stag]

genops.c: Lines 326-349

if (_IO_vtable_offset (fp) == 0 && _IO_fwide (fp, -1) != -1)
  return EOF;

if (fp->_mode == 0)
  _IO_fwide (fp, -1);
if (_IO_in_put_mode (fp))
  if (_IO_switch_to_get_mode (fp) == EOF)
    return EOF;
if (fp->_IO_read_ptr < fp->_IO_read_end)
  return *(unsigned char *) fp->_IO_read_ptr++;
if (_IO_in_backup (fp))
  {
    _IO_switch_to_main_get_area (fp);
    if (fp->_IO_read_ptr < fp->_IO_read_end)
  return *(unsigned char *) fp->_IO_read_ptr++;
  }
if (_IO_have_markers (fp))
  {
    if (save_for_backup (fp, fp->_IO_read_end))
  return EOF;
  }
else if (_IO_have_backup (fp))
  _IO_free_backup_area (fp);
return _IO_UFLOW (fp);

[fleet cedar]

Please light that on fire

[winged mantle]

i think i will just use fread

[hoary sluice]

im so glad im using rust

[winged mantle]

i barely have an understanding of threading maybe i should learn it more sometime

does anyone truly understand threads

or do they just pretend to

[fleet cedar]

Wanna see something cursed?

[valid jetty]

elle had that when i first started working on it

[winged mantle]

yes

[valid jetty]

because it was easier to parse

[fleet cedar]

/run ```java
\u0053\u0079\u0073\u0074e\u006d\u002e\u006fu\u0074\u002e\u0070\u0072\u0069\u006e\u0074\u006c\u006e\u0028\u0022\u0048e\u006c\u006c\u006f\u002c\u0020\u0057\u006f\u0072\u006cd\u0021\u0022\u0029\u003b

[rugged berry]

Here is your java(15.0.2) output @fleet cedar

Hello, World!

[hoary sluice]

its not about easier parsing it just removes the requirement for () for unary epr

expr

[valid jetty]

is that not making it easier to parse

[winged mantle]

does this bot run strings through eval(f"\"{str}\"")

[hoary sluice]

no cause sometimes its just ambiguous

[valid jetty]

ah

fair

foo x - 3

[hoary sluice]

@valid jetty would 100 silverback gorillas be able to beat 1 rosie

wait shit

[valid jetty]

foo x (-3)

[hoary sluice]

other way around

[fleet cedar]

No, bot isn't doing anything weird

That's just regular java

[valid jetty]

foo (x - 3)

[valid jetty]

no

i would win

[fleet cedar]

Bot does add public static void main though

[valid jetty]

can the silverback gorillas write compilers

didn’t think so

[hoary sluice]

well ok i didnt specify that you dont have an F35

[valid jetty]

it’s so cool

it’s basically

generic lifetimes but on steroids

[hoary sluice]

wtf is a generic lifetime

arent lifetimes always generic

[valid jetty]

to put it simply, you can define an Fn trait item on a trait impl, and that trait item may take an argument which has a lifetime, but you’ll only know that lifetime when calling the function, even though the lifetime has to be somehow specified trait wide for than Fn trait item

[hoary sluice]

i use literally 0 lifetimes anywhere besides impl debug and display

[valid jetty]

so the solution is that higher rank trait bounds

[hoary sluice]

this sounds super niche

[valid jetty]

you have a for<'a> Fn(&'a i32) and you can specify this lifetime trait-wide rather

for any trait 'a, that is the Fn trait item that type T is

[hoary sluice]

this is just not an english sentence

[valid jetty]

it’ll just infer the lifetime that you’re talking about at the call site

@fleet cedar can you explain this simpler lmao

it’s an advanced concept it’s kinda hard to explain simply

i’ll try to explain via code examples when i get home

i’m currently out at the front of my school at 9pm 🎀🎀

[fleet cedar]

I have never written a hrtb in my life

[valid jetty]

oh

i mean neither have I

[hoary sluice]

i think im mainly having trouble understanding this because ive never seen the syntax for<'a> for normal lifetimes either

[valid jetty]

but it’s still a cool concept to know

[hoary sluice]

this just sounds extremely useless

[deep mulch]

@valid jetty omg hiii

[hoary sluice]

oh its literally just Fn(&'a str) where 'a is predefined in the trait vs Fn(&'a str) for all 'a

ok

[valid jetty]

fn foo(x: &i32) { dbg!(x); }

fn main() {
    let cb: for<'a> Fn(&'a i32) = foo;
    {
        let tmp = 3;
        cb(&tmp); // lifetime &'a
    }
    {
        let tmp = 4;
        cb(&tmp); // lifetime &'b
    }
}
``` it’s a way to concretely type this

[hoary sluice]

what is cb

[valid jetty]

(yes it’s not just useful in traits)

[valid jetty]

callback, just a placeholder name

[hoary sluice]

yea i understand it

[valid jetty]

🙏🙏

[hoary sluice]

its just a generic lifetime

why is it called hormone replacement therapy

[valid jetty]

thats.. exactly what i said

[valid jetty]

.

[hoary sluice]

oh was steroids the joke here

[valid jetty]

😭😭😭

glad someone got it

[hoary sluice]

is it on steroids because hrt or is it on steroids because its somehow better than just generic

[valid jetty]

both

[hoary sluice]

how is it not just a normal generic

[valid jetty]

normal generic lifetimes are generic function-wide or trait-wide and this just lets you specify that some type is generic over ANY generic lifetimes

it’s like a level above generic lifetimes

[hoary sluice]

ohh

i thought this was what generic lifetimes were

[valid jetty]

because instead of saying you have an 'a for the duration of the function or trait, you’re saying you have a lifetime you don’t know yet but will when you call the function

it’s really hard to explain this lmao

[hoary sluice]

its not

instead of it being generic on the trait its generic on the function

[fleet cedar]

The trait bound itself is generic

[valid jetty]

yea this ^^

it can take any other lifetime

[hoary sluice]

its literally just ∀'a

[fleet cedar]

Yes

[valid jetty]

yeah that’s a good way to put it

[fleet cedar]

Hence why it uses for keyword

[valid jetty]

yep

[hoary sluice]

^

[valid jetty]

for all analogy

[hoary sluice]

for any not for all

?

[fleet cedar]

What would be the difference

[hoary sluice]

wait are they equivalent in this case

[valid jetty]

well it’s like “for all 'a that this Fn trait item is called with”

which is essentially “for any” as far as the type system is concerned

[fleet cedar]

It's just a for all

[hoary sluice]

whats the mathematical difference between for any and for all

or is there none

[valid jetty]

tbh i don’t know the formal difference but i’m sure mathematicians had nothing better to do than prove that they are different

[hoary sluice]

This seems to depend on the context: "For all x∈X P(x)
" is the same as "For any x∈X P(x)" On the other hand "If for any x∈X P(x), then Q" means that the existence of at least one x∈X with P(x) implies Q, so P(x) doesn't need to hold for all x∈X to imply Q.

[fleet cedar]

One of them is a mathematical term, the other is not

"For any" is an english phrase, not math

[hoary sluice]

who defines that

[fleet cedar]

Convention

[hoary sluice]

im gonna put for any in a proof and you cant do anything about it

[valid jetty]

anything can be a mathematical term if you use it in a paper and people cite it enough

[hoary sluice]

there seems to be many interpretations of for <insert word referencing every element> https://math.stackexchange.com/questions/430646/difference-between-for-any-and-for-all

[valid jetty]

i could call the theory of everything “ski bidi ri zz” (idk if those words are filtered) in my next paper and if people cite it in future papers it’s now a mathematical term

[fleet cedar]

I'm gonna put an apple pie in a proof

(Probably won't though, lazy)

[hoary sluice]

i wil make the elementary arithmetic operators the following: + - tralaleroTralala / ** %

[fleet cedar]

I remember in some uni lecture the professor redefined the symbol 4, just to show he could

[hoary sluice]

lol

[valid jetty]

lmao

i’m sure “for any” is definitely used throughout mathematical papers though

i’m willing to bet money that there are papers that use the exact phrase “for any”

[hoary sluice]

im willing to bet money that the sky is blue sometimes

[fleet cedar]

Since most papers are written in english, that seems exquisitely likely

[hoary sluice]

also you dont have any money

[fleet cedar]

Did you plunder her

[hoary sluice]

yop

[deep mulch]

@valid jetty

hi

[nimble bone]

@valid jetty huiii

[deep mulch]

@robust jackal

yes

[modern rune]

https://pastebin.com/n1z19Vb5 docker compose file for vencloud + valkey + behind traefik

[south moon]

on pastebin 😭

is @signal oak open source

[spark tiger]

holy 😭 https://chatgpt.com/codex

https://openai.com/index/introducing-codex/

[median root]

guys does this make me a good dev? im removing more code than im adding!1!1!!!

[lavish cloud]

Does anyone know how to OpenGL cuz I can't figure out deferred lighting

https://pastebin.com/KdrRLuTi here's my best attempt (only displays black, hiding anything rendered before)

[runic sundial]

Have you tried rendering out the separate layers into the window?

[lavish cloud]

how do

because I'm currently trying to render them all to the window

[runic sundial]

Essential debugging

[lavish cloud]

I know the G-buffer is fine at least

but trying to render only the ambient lights is still black (and still erases all drawn content)

[runic sundial]

No I mean like, can you render the gbuffers to the window to debug it?

Without the actual lighting passes

I'm on the train and the 4G is dying

[hoary sluice]

kotlin game engine??

[lavish cloud]

[hoary sluice]

gigabased

[lavish cloud]

I mean it'll only show the gPosition as color but that part works fine

[runic sundial]

So what part is busted?

[lavish cloud]

either the way I set up my gbuffer backing textures or the way I bind them to the light shader or the light shaders or the blending or the

[runic sundial]

Also, do you have a debug output wired up yet?

[lavish cloud]

I don't know anything opengl so half of this is just LLM stuff

[runic sundial]

Like, for debug error message

[runic sundial]

Oh, nvm

[lavish cloud]

I mean I can run it with renderdoc

[lavish cloud]

wdym?

[runic sundial]

Heretic.

[lavish cloud]

I didnt know opengl did error messages other than just printing to stdout

[runic sundial]

Go purchase the OpenGL Superbible 7th edition right now.

Opengl can give you asynchronous debug output for errors

Or general messages

[lavish cloud]

it doesn't crash or report to stdout rn tho, when there's a problem it usually does

[runic sundial]

How is it even printing to stdout

[lavish cloud]

¯_(ツ)_/¯

it does so when I fuck up my shader code fwiw

[runic sundial]

Well, just tell the AI it has to fix it

Or it goes to jail

[lavish cloud]

I tried on 3 separate AIs

all fail to fix the problem

[runic sundial]

Get the OpenGL Superbible and just read it

Find a pdf if you're unable to afford a physical copy

Just read it and follow the examples provided there

[lavish cloud]

but all opengl docs are written as if you understand opengl already

idk what a vertex or renderbuffer or framebuffer or other stuff is

[runic sundial]

Because you opened the unfunny docs

Anyway if you don't know what a vertex is I think you're cooked

[lavish cloud]

I know it's different from a fragment at least

cuz they both have different shaders

[runic sundial]

There's a lot of stuff but likr

If you make it using AI then it will uh

Explode

I can help explain terms but I will not fix code

[lavish cloud]

ok whats a g buffer, depth buffer, render buffer, draw buffer, frame buffer
and the like differences
and how does opengl know which one to use

[runic sundial]

A gbuffer is an abstract concept, it's a "geometry buffer" and what you usually draw the geometry onto in the middle stages of a deferred renderer

A framebuffer is a container which describes a buffer onto which you render. It has textures and renderbuffer attachments.

Renderbuffers are special "write-only" textures, you need to blit (manually copy) stuff from them if you ever want to read.

But, they are useful for depth testing.

A depth buffer is a greyscale image that is used to store depth. Depth here means how close or far away from the camera something is.

How does opengl know which to use? Framebuffers are used to describe what you're gonna draw onto etc, but the rest is more when you decide to use em

[lavish cloud]

[runic sundial]

So in a deferred renderer, you create a set of textures. These are your gbuffers. You take these textures and you attach em to a framebuffer. That's your rendertarget.

You can now render stuff with the shader to these textures if you bind that framebuffer beforehand.

Once you're done rendering stuff to it and want to light up the scene, you bind these textures as samplers and another framebuffer for the output for the light or whatever.

Once you're done with the lighting, you bind the window framebuffer (framebuffer zero is the window one usually), and render the final output there.

Generally the way a deferred renderer works

[lavish cloud]

ok I think that's what I do (except I draw light passes all to window framebuffer directly with gl blending on)

[runic sundial]

That also works

[lavish cloud]

but then why no work

[runic sundial]

Works if everything is correct

https://www.khronos.org/opengl/wiki/Debug_Output

Peek this

I'm assuming you're using LWJGL3 bindings with Kotlin, and that should have bindings for this included

This will let you route all errors to stdout, with stacktraces

[lavish cloud]

There's some stuff during model loading:```
33350 33361 131185 33387 Buffer detailed info: Buffer object 160 (bound to GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB (0), GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB (1), GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB (2), and GL_ARRAY_BUFFER_ARB, usage hint is GL_STATIC_DRAW) will use VIDEO memory as the source for buffer object operations.
33350 33361 131185 33387 Buffer detailed info: Buffer object 161 (bound to GL_ELEMENT_ARRAY_BUFFER_ARB, usage hint is GL_STATIC_DRAW) will use VIDEO memory as the source for buffer object operations.

But nothing during each draw step

[valid jetty]

the infamous -ljson

or whatever the lib is

[winged mantle]

i wish this would work :(

executable/.. does not seem to resolve to the directory the executable is in

[runic sundial]

nvidia drivers on windows detected lmao

[fleet cedar]

argv[0] is just a random string that's conventionally a path to the executable

There was some systemd tool or something that had a security vulnerability because it assumed argc >= 1

[winged mantle]

you mean it's not even guaranteed that it's present?

[hoary sluice]

it pretty much always is

[winged mantle]

i guess i will allow any path to be passed in

this is just an example executable

to test the lib

[hoary sluice]

enough to allow relying on it (unless youre systemd ig)

[winged mantle]

in this case testing the streaming functionality

[fleet cedar]

On linux the exe path can be found somewhere in auxv I think, or readlink(/proc/self/exe)

[winged mantle]

tbf it's not like i'

m bothering to make it work on anything other than linux

[fleet cedar]

On windows there's some api for it but I don't know it by heart

[hoary sluice]

realistically the only way for argv[0] to not be the path is if you purposefully make it not the path, and if the software only uses the path for usage then theres no problem with it

[winged mantle]

(in the makefile)

[hoary sluice]

based

[winged mantle]

idk how you would make it generate a dll tbh

[hoary sluice]

just dont

[winged mantle]

how would makefiles even work on windows

[hoary sluice]

just dont support windows

[winged mantle]

or what if the distro doesn't have a mkdir command for example

[hoary sluice]

if a windows user wants to use my software they can either switch to linux or pay me to port it to windows

[winged mantle]

but makefiles are sh based

do you use windows port of sh and coreutils

[hoary sluice]

why are you using mkdir for that

you mean in the makefile?

[hoary sluice]

you just dont use windows its as easy as that

[hoary sluice]

if someone uses a distro so obscure that it doesnt have mkdir preinstalled they can either install/alias mkdir or make it themselves, this isnt something you have to worry about

[balmy lintel]

developers who develop on linux and then say "sorry cant fix windows issues" are so annoying

[winged mantle]

still what about windoze

[hoary sluice]

what is $CC isnt set

same issue

[runic sundial]

Is this the shader ur using to render the initial geometry?

[winged mantle]

i would first think you'd build in wsl

[hoary sluice]

cry about it

[winged mantle]

but then you'd still be building a so

[balmy lintel]

You will be forced to use windows

sorry

[hoary sluice]

i was forced to use windows for an internship last year

genuine torture

[fleet cedar]

I'm more of a I don't look for bugs on windows but if someone reports one I'll take a look kind of guy

[winged mantle]

so i imagine there's some ports of make to windows which also have some port of sh and coreutils

[hoary sluice]

should be illegal

[winged mantle]

the prism launcher modrinth pack export was broken for some time because i never tested on windows

[balmy lintel]

companies who force operating systems are insane

[hoary sluice]

yop

its a war

they force me to use windows

i make linux only software

[balmy lintel]

like imagine getting some job and you never used windows or macos before and they just expect you to learn to use it within 2 weeks

[hoary sluice]

in the name of "corporate security"

[winged mantle]

use freebsd

[hoary sluice]

their threat model is "someone who got fired and is angry will publish the code from their linux computer, if we give him a windows machine we can lock it before firing him"

[lavish cloud]

no thats the fragment shader for the ambient light shader, the geometry shader is:

[winged mantle]

can't you lock down linux machine too...

[hoary sluice]

its one of those "digital transformation" companies that contract services for random companies, some of which require nda

[winged mantle]

digital renovation

[hoary sluice]

easier to bypass

if a company doesnt allow you to use your own laptop or your own os theyre generally evil and should not be worked for

[runic sundial]

epic

so dump the output this framebuffer onto the screen, or to an image or whatever

to validate their outputs

[hoary sluice]

why dont you import it from a file

[lavish cloud]

I will eventuallly™️

[lavish cloud]

I already did, and I know it's correct (color is the gPosition value iirc)

[runic sundial]

And the normals?

[lavish cloud]

how do I show those

[runic sundial]

layout (location = 0) out vec4 gPosition;
layout (location = 1) out vec4 gNormal;
layout (location = 2) out vec4 gAlbedoSpec;
layout (location = 3) out vec4 gTangent;
layout (location = 4) out vec4 gBitangent;

Whatever texture is in loc 1

[hoary sluice]

@valid jetty https://youtu.be/xP5-iIeKXE8

[runic sundial]

If lighting is broken, validate the TBN values. So Normals/Tangent/Bitangent

[lavish cloud]

so I just swap location=0 and location=1?

[runic sundial]

if you have render doc you should be able to just, find that texture and render it

one sec

https://docs.gl/gl4/glObjectLabel

[runic sundial]

And the T and B ones look different?

[lavish cloud]

but the model data shouldn't matter because the ambient shader should essentially just be a solid color

T and B?

[runic sundial]

Tangent/Bitangent

[runic sundial]

ah

anyway, send over the fragment shader for the actual render output then

[runic sundial]

yep, this looks correct.

[lavish cloud]

I mean the pastebin has all the fragment shaders

[runic sundial]

it's in this yucky blob

[lavish cloud]

@Language("GLSL")
private const val LIGHT_VTX_SHADER = """
    #version 330 core
    layout (location = 0) in vec3 aPos;
    layout (location = 1) in vec2 aTexCoords;
    
    out vec2 TexCoords;
    
    void main() {
        TexCoords = aTexCoords;
        gl_Position = vec4(aPos, 1.0);
    }
"""

private val ambientLightShader = Shader(LIGHT_VTX_SHADER, """
    #version 330 core
    uniform vec3 ambientColor;
    uniform sampler2D gAlbedoSpec;
    
    in vec2 TexCoords;
    out vec4 FragColor;

    void main() {
        vec3 albedo = texture(gAlbedoSpec, TexCoords).rgb;
        FragColor = vec4(ambientColor * albedo, 1.0);
    }
""".trimIndent())

val ambientLights = lights.filter { it.type == Light.LightType.AMBIENT }
val ambientColor = ambientLights.fold(Vector3.ZERO) { acc, light ->
    acc + light.color * light.intensity
}

ambientLightShader.use()
ambientLightShader.setUniform("ambientColor", ambientColor)
ambientLightShader.setUniform("gAlbedoSpec", 2)
renderFullScreenQuad()

[runic sundial]

So the vertex shader is just a passthrough meant to be used with a fullscreen pass

Have you tried on the ambientLightShader FragColor = vec4(1.0, 0.0, 0.0, 1.0) ?

Should make the whole screen red.

Just validates that you're doing the full-screen pass at all

[lavish cloud]

so even thats incorrect

[runic sundial]

Next I would:

FragColor = vec4(1.0, 0.0, 0.0, 1.0);
FragColor.rg = TexCoords;

[runic sundial]

Nope. that's what i expected to see

[lavish cloud]

but it should also be rendering the stuff I rendered before, no?

cuz I have blend enabled

[runic sundial]

You're doing a full-screen pass with red and alpha one, most blend functions will just treat it as full opaque

[lavish cloud]

glEnable(GL_BLEND);
glBlendEquation(GL_FUNC_ADD);
glBlendFunc(GL_ONE, GL_ONE);

// render to gBuffer

// bind textures

// ambient light pass

oh

how do I fix that

[runic sundial]

https://learnopengl.com/Advanced-OpenGL/Blending

This example uses glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);

Without setting the blend equation

[lavish cloud]

[runic sundial]

But yeah, validate that this shows you the expected UVs

[lavish cloud]

but that requires stuff to have an alpha

[runic sundial]

Yeah?

[runic sundial]

Ok, your UVs are perfect.

[lavish cloud]

but the magenta floor (regular render pass) has no alpha and the car model (deferred lighting render pass) has no alpha

[runic sundial]

My question is why do you intend there to be alpha to begin with?

[lavish cloud]

wdym?

[valid jetty]

i saw this, this is insane

[runic sundial]

Why do you need blending enabled?

[lavish cloud]

because wont every light shader overwrite all old data otherwise

[runic sundial]

Oh, you're trying to mix the color data from multiple shaders?

[lavish cloud]

well

yeah

thats how deferred lighting works isnt it

one pass from ambient, one from each directional and one from each point

• all the data that was rendered before the gbuffer/light passes (e.g. skybox, debug objects, etc)

[runic sundial]

Deferred gives you a lot of freedom, but mixing is generally done in post

Like with lights, the power of deferred is that you can do the lights all in one big go

[lavish cloud]

how does that work in one big go?

because best I could find with a single shader is a hardcoded limit on lights

[runic sundial]

Do you know how deferred is different from forward?

[lavish cloud]

but modern game engines support up to millions of lights

deferred is doing it on the 2d view based on the depth data and stuff iirc?
and forward is doing it as part of the vertex/fragment shader for the object I believe

that's what I understood from the wiki page at least

and because you're dealing with a simpler view deferred is faster (and can do any number of lights)

[runic sundial]

Forward is "what you see is what you get" lighting. Both can do "any number of lights", but the optimization strategies and pros/cons differ.

Like with forward, you light the geometry as it's being rendered.
With deferred, you split the info about the geometry into 'gbuffers' and 'do it later'

[lavish cloud]

[runic sundial]

The "one big go" has to do with you being able to treat the gbuffers as all the geometry as a set of textures.

So a single lighting pass can do all the lighting at once.

gbuffers -> (big fat lighting shader) -> window

Is the simplest

[lavish cloud]

ic

wait but then how does the big shader know how many lights there are and what types of lights, without hardcoding a limit

[runic sundial]

You, hardcode a limit?

[lavish cloud]

I thought the point of deferred lighting was to not have a limit

[runic sundial]

There is always a limit. Deferred is just one of the ways of having a higher limit.

[lavish cloud]

ic

also wait

if you have everything as a texture

how do you deal with reflective surfaces

cuz you'd need to know what's behind a wall in order for a mirror showing what's behind it to work

[runic sundial]

Screen space trickery

[lavish cloud]

or is that where you cheat and put a camera in the mirror, then render that camera output as texture on the mirror

[runic sundial]

What's behind that wall? You've discovered one of the limits of rasterized rendering

[lavish cloud]

so I guess I have to learn raytraced rendering instead then?

[runic sundial]

Raytraced, or one of the tricks to get around it to fake it.

Depends on performance targets

If you wanted to have "no limit" you can technically do the similar multi-pass you were suggesting, but then you need to have a seperate "luminance" gbuffer.

But you should still have a "hard-coded" limit per pass, so each pass can do more than one light at a time.

[lavish cloud]

surely modern gpus can do 4k@60 raytraced tho, we've had RTX GPUs for like 6 years

[runic sundial]

[runic sundial]

void main() {
    vec3 albedo = texture(gAlbedoSpec, TexCoords).rgb;
    FragColor = vec4(albedo, 1.0);
}

Anyway with this and disabling blending, you should at least see "something" on the screen.

[runic sundial]

Either uniform sampler2D gAlbedoSpec; is not bound correctly, or it is black.

[lavish cloud]

gAlbedoSpec comes from ```
vec4 diffuse = material.diffuse * VertexColor;
if (hasDiffuseMap) {
diffuse *= texture(diffuseMap, TexCoords);
}

// Specular color
vec4 specular = material.specular;
if (hasSpecularMap) {
specular *= texture(specularMap, TexCoords);
}

// Store diffuse and specular components
gAlbedoSpec.rgb = diffuse.rgb;
gAlbedoSpec.a = specular.r; // Store specular intensity in alpha

hmm yeah gAlbedoSpec is black

[runic sundial]

set gAlbedoSpec.rgb = diffuse.rgb; to gAlbedoSpec.rgb = vec3(1.0, 0.0, 0.0); for more red debudge

[runic sundial]

there u go

issue found

[lavish cloud]

so diffuse.rgb is wrong

[runic sundial]

yOp

You can go back and undo the changes made to the other shader, and mayhaps your funny blend modes are correct after all

[lavish cloud]

material.diffuse is set to Color(value=[153, 153, 153, 255])

so that should be fine

so does that mean VertexColor is wrong?

[runic sundial]

Up here you have 3 variables controlling the diffuse out:

VertexColor, material.diffuse and texture(diffuseMap, TexCoords)

Validate each

[lavish cloud]

out vec3 FragPos;
out vec4 Color;
out vec3 Normal;
out vec2 TexCoords;
out vec3 Tangent;
out vec3 Bitangent;
// ---
in vec3 FragPos;
in vec4 VertexColor;
in vec3 Normal;
in vec2 TexCoords;
in vec3 Tangent;
in vec3 Bitangent;
```order of in/out values seems fine

[lavish cloud]

disabled the diffuse map for that test

ok figured it out
now why is a light thats above it and pointing up still lighting it up from below

[runic sundial]

[winged mantle]

i tested this code on a 1gib file and it worked perfectly

idk if it should stop on null but that's invalid in json anyway 🤷‍♀️

[runic sundial]

https://tenor.com/view/green-alien-man-green-alien-green-alien-gif-16513909281920195232

[winged mantle]

i have identified a "problem" with splitting strings into multiple tokens

you would need to have context of what the previous token was

[hoary sluice]

@valid jetty https://youtu.be/7sl0e9yKwTk

[valid jetty]

see i knew this would happen

what hes doing is very smart but also very dangerous

[winged mantle]

the thing is, the first thing i want to implement is syntax highlighting

put in json string, outputs ansi highlighted string

and it's kinda useful for token parsing to be relaxed and not care about the previous token

so even invalid json gets highlighted

[fleet cedar]

What are your goals for your parser

[winged mantle]

mainly learning

[fleet cedar]

Performance? Ergonomics? Zerocooy? Resilience?

[winged mantle]

i can use it in my c discord bot!

simplicity

i guess

but some versatility because it will push me further

I think I will just make string one token like most parsers

[fleet cedar]

If you want simplicity, step 1 is to not write your own json parser

[winged mantle]

😭

it's mainly for learning

i just might use this library if i do other c projects

and say... not having a whole hashtable impl would make it easier and more lightweight

[nimble bone]

tode insane

[deep mulch]

tode insane

[valid jetty]

@hoary sluice you will not believe how hard this was but

self-referential structs are now possible

as long as you refer to a pointer

so you can do this

struct A {
    Foo *a
}

struct B {
    Foo *b
}

struct Foo {
    A *x,
    B *y
}
``` without forward declaring

which means that this is now cleaner than ever

enum AstNodeKind {
    Literal,
    BinOp,
    Funcall,
    Return,
    Declare
}

struct Literal {
    TokenKind kind,
    TokenValue value,
}

struct BinOp {
    Box<AstNode> left,
    Box<AstNode> right,
    string op,
}

struct FunctionCall {
    ObjectString name,
    AstNode[] args,
}

struct Return {
    Box<AstNode> value,
}

struct Declare {
    ObjectString type,
    ObjectString name,
    Box<AstNode> value,
}

struct AstNode {
    AstNodeKind tag,
    @unused Box<Literal> as_literal,
    @unused Box<BinOp> as_binop,
    @unused Box<FunctionCall> as_funcall,
    @unused Box<Return> as_return,
    @unused Box<Declare> as_declare,
}

the old(er) version is so much more insane lmao

struct AstNode {
    u32 tag;
    void *as_literal;
    void *as_binop;
    void *as_funcall;
    void *as_return;
    void *as_declare;
};

const ASTNODE_LITERAL = 0;
const ASTNODE_BINOP = 1;
const ASTNODE_FUNCALL = 2;
const ASTNODE_RETURN = 3;
const ASTNODE_DECLARE = 4;

struct Literal {
    string kind;
    ValueKind *value;
};

struct BinOp {
    AstNode *left;
    AstNode *right;
    string op;
};

struct FunctionCall {
    ObjectString *name;
    AstNode *[] args;
};

struct Return {
    AstNode *value;
};

struct Declare {
    ObjectString *type;
    ObjectString *name;
    AstNode *value;
};

struct AstNode {
    u32 tag;
    Literal *as_literal;
    BinOp *as_binop;
    FunctionCall *as_funcall;
    Return *as_return;
    Declare *as_declare;
};

im not sure how to approach this better than to do it the way i did

basically before i parse the real structs, in an earlier parser pass, i collect all structs enums and namespaces into the struct and enum pools

which, to allow for self-referential formatting, also means i need to forward declare the formatting function (if the user didnt specify it shouldnt be generated)

that means the number of functions for each struct effectively doubles

even though only 1 is generated

[hoary sluice]

@valid jetty woo finally

[valid jetty]

yay !!!

[hoary sluice]

also i notice that when i run fac 49 this is the error:

and with fac 50 its a stack overflow

ig cause the stack is big enough to compute 49!

[valid jetty]

tail call recursion time?

[hoary sluice]

this is like planned for 2026 levels of optimization

also is the mod.rs file meant to house only mod, pub use and type or can i put my entire parser in there

[valid jetty]

RUSTFLAGS="-C link-arg=-Wl,-z,stack-size=16777216" cargo build --release

or something

[valid jetty]

no idea

[hoary sluice]

cause clippy complains about module inception so im assuming it wants me to put the contents of parser.rs into mod.rs

@fleet cedar

[valid jetty]

-#
mini blobcat

ulimit -s 16384 then i guess

lmao

[hoary sluice]

stack is 16mb by default on nixos

probably

[valid jetty]

oh

mine is 8mb

[formal belfry]

anyone else hear that ominous bell tolling????🤣🤣🤣 no?? just me??????😭😭😭😭😭😭

[hoary sluice]

uh

what

[winged mantle]

i loove c

[valid jetty]

147 days until mizu5

[valid jetty]

ulimit -s

[hoary sluice]

i read that as omnibus bell trolling

oh its 8mb

nvm

yop

[valid jetty]

make ints internally i128 when

[hoary sluice]

they are

[valid jetty]

oh

i256 when

[hoary sluice]

when rust makes i256

[valid jetty]

https://crates.io/crates/i256

[winged mantle]

Quite frankly, even if the choice of Java were to do nothing but keep the Kotlin, Nix and Rust developers out, that in itself would be a huge reason to use Java

[valid jetty]

or just use bigint tbh

[hoary sluice]

i am not adding i256 as my only dep

[valid jetty]

bigint time

https://crates.io/crates/crypto-bigint

[hoary sluice]

toad when he finds out about java devs that use kotlin, nix and rust

[fleet cedar]

mod.rs, if you have one, is exactly the same as any other module

[winged mantle]

🤣🤣🤣 eagerly when he fidns oytu tthat z??? because the uses rust

[hoary sluice]

i dont understand this sentence

is the answer to my question yes or no

[deep mulch]

rosie explosion

[fleet cedar]

It contains the contents of the module

[winged mantle]

pretty sure you can put anything in mod.rs

even a recipe for a cake

[hoary sluice]

the contents of the module being the entire module or only the mod statements?

[winged mantle]

SMART IDEA

[fleet cedar]

Depends on if you want the module to contain things or only submodules

[winged mantle]

instead of storing 2fa backup codes in obvious places put them in comments in project source code

so if malware tries to find 2fa backup codes it will fail

[winged mantle]

pretty sure mod.rs is what is exported to outside the module - everything else is internal

but i might remember wrong

[winged mantle]

i last did rust in dec

[hoary sluice]

do i move parser.rs into mod.rs

theres no submodules in there