Thread 107133087

What are you working on, /g/?

Previous: >>107113418

cniles be like
>i need to use a boolean!
>#include <sexually transmitted disease bool dot h>

>>107133087 (OP)
handy chart. have a you

>>107133099
>cniles
>bool
int or setting bits
bool is gay

>>107133388
bools are for the compiler, not storage

I'm trying yet another way of doing gitignores:

# Ignore all directories at root
/*/
# Allow source dir
!/src/


Most programming langs don't produce _files_ straight into the repo root (some do), so this works out quite well. Some langs will still need a bit of tuning, like __pycache__, but it's not bad.

I've tried e.g. `*` and `/*`, and then allow listing, but that's tedious.
I think this will also have the side effect of not dumping random files at repo root, I'll rather put them in a directory if I don't want them in git.

>>107133087 (OP)
>cm
>km
huh?

>>107133442
mm->cm->dm->m->dam->hm->km
mili->centi->deci->unit->deca->hecto->kilo

on another note i think i got my version of seeking bits
i have yet to fix my mask, and test it, but i think its correct
idk how its gonna stack up with other versions though
i have only 3 conditionals so it should be decently fast
but because i have only 3 conditionals, theres some operations that i do but i dont actually need
it should perorm decently well on small lengths because instead of looping through 1 byte at a time i parse them by bunches of 0s and 1s

i still have to test it though
knowing myself theres 100% something ive forgotten about
its gonna happen later though its been something like 3 hours im hacking at this problem, i need a break

>>107133388
my leetcode problem requires me to use bool

>>107133435
You can just modify the gitignore file when you need to.

>try to solve a leetcode problem for an hour using C
>fail
>switch to python
>solve in 5 minutes
fucking crazy
and yes it's 100% a skill issue but god damn

>>107133987
its a different class of language.
completely different usecase

>>107133987
>>107134003
cont
its like comparing a scalpel (C) to an axe (python)
you can cut a tree with both
same with performing surgery
but for either case what makes sense is using one or the other

>>107134003
im just not used to thinking at the autism low level
like once I got the implementation in a higher level language it's relatively easy to translate it to C since it basically just adds more steps
but starting from scratch in C is rough since it doesn't give you many of the comfy abstractions

>>107134084
>comfy abstractions
such as?
also, which leetcode problem filtered you?

>>107134061
C is closer to a sharpened rock than a scalpel, given how both are unfixably primitive as fuck and require a lot of hacking to get even simple things done

>>107134084
heh. my brain is thoroughly warped low-level-wise because ive been using C more or less exclusively, during the past 6 years
but if i dont need to go low, i dont.
amdahls law:
you dont want to waste time optimizing code that accounts for 0.1% of your runtime
you want to optimize stuff that actually matters

this also goes for languages.
if i have to do some scraping, you can be sure ill be using paiton + beautifulsoup
unless the data collection takes longer than the desired data resolution

>>107134153
nono
a well sharpened scalpel
you have compiler hints, in syntax, or as pragmas/attributes, various flags, intrinsics
its the most granular language short of asm
>primitive
thats its selling point.
both as a learning tool, and in situations where individual reads matter or a float multiplication accounts for half your runtime
attempting to satisfy the usecase for C with higher level, more feature rich languages is like writing a letter while wearing boxing gloves

>>107134199
>feature rich
even that is not accurate
you got features of the wazoo with the c ecosystem
gorilions of compiler flags
attributes and pragmas for everything you might want
mature analysis tools

the language itself looks barebones.
but even that is misleading

a * b * c * d is not the same as (a * b) * (c * d)
c is actually subtle as fuck. extended c that is
the standard is kinda shit and the comittee people are autistic (mentally retarded)

>>107134263
>comittee people are autistic (mentally retarded)
minus eskil steenberg.
i like that guy

>>107133556
What does masking_mask do?

>>107134199
That's why non-larpers use C++.

>>107134263
integer multiplication is associative. fp multiplication isn't because it's ieee, c had no hand in that.

>>107134570
fp addition can't be associative in any useful way can it
dunno about multiplication

>>107134130
>such as?
nigga you serious? literally just go and compare python vs C solutions to anything lmao

>>107134676
you can't name one?

>>107134696
it's very unhealthy for your brain to try to ragebait people on the internet
ps. I'm exiting the thread now.

The year of C++ modules.

I'm building a static site generator, or what I think Hugo does. Using this as Rust practice.
I'm getting bored though. Another project that I probably won't finish.

>>107133556
Tried running it, it failed in this case that doesn't use the mask table.

Mismatch: len = 2

New 0 = 0 Old 0 = 0
New 1 = 139 Old 1 = 66

{0x0, 0x55555555555555d, 0x7fff80007fff800,

>>107134199
it's a rock
a fucking fossil that has no place anywhere near modern applications

>>107134570
its not what this is aboot
parenthesis hint to the compiler the dependency of operations
because of that when you use parenthesis the compiler stores the intermediary result in discrete registers.
and thanks to that the contents of the parenthesis can be computed separately, allowing the compiler to leverage instruction level parallelism

well, in this exact case thats what happens if you compile with -O0, any optimization turns that into a regular multiplication because in this exact case its not worth it i suppose

>>107134438
umm, masking? kek.
it sets the bits that have already been processed to zero so that tzcount gives an accurate offset

>>107134542
sepples is c but for big corpos
and no, this isnt a compliment if you give it a fink

>>107134905
kek, didnt even have time to run it myself, yet
i just came back from a grocery run
{0x0, 0x55555555555555d, 0x7fff80007fff800}
this is the array i presume?

>>107134928
you have no idea what youre talking about
to illustrate it:
take a guess what language this code is from?
float tmpSum = 0;

if (ROW < N && COL < N) {
// each thread computes one element of the block sub-matrix
for (int i = 0; i < N; i++) {
tmpSum += A[ROW * N + i] * B[i * N + COL];
}
}
C[ROW * N + COL] = tmpSum;

>>107135003
whatever the language, it's written by a caveman that doesn't use iterators or proper function/method splitting

so again, proving the point that c is primitive and has no place anywhere near modern applications

>>107135080
its cuda
now feel free to take your redditspacing back where you came from <3

I'm currently working with C. How useful do you think a function like the following is:
bool equal(const char *s1, const char *s2) {
return strcmp(s1, s2) == 0;
}

I'm a bit torn apart, on one hand I think it makes the code more readable, on the other hand I don't want to clutter my program with helper functions like these that solve only a very little problem, as strcmp(s1, s2) == 0 is probably almost as readable to a c programmer. What do you think?

>>107113513
>>107113592
>>107134930
This is the test generator I'm using: https://pastebin.com/wxvV6Q1y

>>107135112
thanks
ill certainly give it a whirl.

>>107135110
iunno.
if i were you id reimplement string ops instead.
for proper string types: wrapepd in a struct, with a size, and extra padding so you can make maximum use of vectorization while keeping your strings compatible with c interfaces

>>107135110
>>107135154
abt vectorization: you have to align your data to 32 bytes when you use __mm256 types and to 16 bytes when you use __mm128
for maximum comfiness i use/allocate unions instead to have a comfy and legal way of accessing the underlying data reagrdless of the type

>>107135112
>>107135154
Crap the pattern count at the main loop only goes up to 9...

>>107135191
? whats the implication?

>>107135209
It's a mistake, set it to 21

>>107135110
should be called str_iseq or something. But yeah !strcmp can be confusing even for someone experienced.

>>107135223
ah, in the utility
ok
ill keep that in mind
i was compiling the code for the first time on my end
aaaaand... (rolling drums)
41 lines of errors. kek
gimme a moment

>>107135185
No you don't have to.

>>107135154
>if i were you id reimplement string ops instead.
that's overkill for the program I'm writing, I really just need to occasionally compare small strings, this is really not the bottleneck in my program. but sounds like a fun project
>>107135239
alright, I will keep it then. I just thought maybe I'm retarded for writing this function

>>107133442
>Distance which light travels while the operation is performed
did you miss that?

>>107135296
if you dont you cant pass an union member or the union itself as argument without using an unaligned load
which has ~4cycles latency iirc
i noticed even aligned loads and casts have a cost it seems.
but maybe thats bc of data dependency which forces a fence which prevents the cpu from ovelapping latencies in the following operations
didnt go to the bottom of things on that one

>>107135309
>that's overkill for the program I'm writing
100% fair but its a necessity whenever stringops are a meaningful portion of performance critical code
the libc is gimped by the fact it assumes non-padded null terminated strings
you can be an entire order of magnitude slower using vanilla libc compared to semi competent homebrew, if not worse

>>107135110
>I don't want to clutter my program with helper functions like these
your program is literally just one function: main. everything else is a helper, so don't be shy about writing small utilities. depending on the size of your project, you probably want to do this regardless, in case you want to change the string type in the future (to use slices, for example, as the other poster suggested)

>>107135369
>muh unions muh latency
that's enough lil bro, asking each character in a string to be aligned to 16 bytes is peak mental illness.

>>107135416
were talking about vector types here, try to keep up

>>107135427
Then mention of unions makes even less sense you stupid nigger.

>>107135504
dats bc u dum
and flamey

arent you a shartoid perhaps?
on another note, if you ask nicely i can explain to you things and the stuff theyre made of

>>107135080
what about a for loop scares you?

>>107135518
I already know that your skull is filled to the brim with diarrhea because the BBC stuck in your ass still wasn't removed.

>>107135583
>bibisi
i knew youre a shartoid
youre easy to spot
ur just fucking retarded

>>107135645
nah its a sharty thing
>cobson ridin that bibisi
have a look at their site sometime
their whole culture fits on one post it
once you understand what theyre about, theyre easy to spot
report and move on
theyre here to troll
thats why they dont make sense whatsoever

>>107135635
>>107135645
>>107135670
what does big black coalstone have to do with any of this

>>107135683
idk, you brought it into the discussion
i am perfectly happy discussing vectors

>>107135369
thanks for the insights, senpai
I didn't understand all of it, but a custom string data type sounds like a cool thing to do to get into vectorization, which I have no experience in currently
>>107135378
>don't be shy about writing small utilities
will do

so how would you handle helper utility functions in c, in the same file, include another .c file, or write a .h and .c with these functions and include the .h?

>>107135729
for something like that, static inline in a header file that gets included everywhere.

>>107135817
ah yes, I thought that a header without the implementation would be weird, but including a c file is also weird, and keeping it in the same file is also weird... this solution seems to be the best

pretty comfy general you have here

>>107135729
>senpai
i barely learned vectorization myself
but its indeed a powerful tool
i wrote an ascii to float that takes 1.5-3 cycles per character read, with better results with bigger strings
ngl im pretty happy with myself, light testing yields that im ~2.5x faster than strtof on a typical string
maybe more bc theres the launch overhead, and the control additions, which souldnt amount to much on 22.6 and 50.9 seconds respectively
heres the code if you wanna have a look
https://files.catbox.moe/1zq8zw.c
you need -mavx2 and -mbmi flags to compile it
and yeah, i cheated a little bit. i take in max 32bytes for integers, 32max for decimals but its not hard or expensive to generalize it into arbitrarily long sequences of numbers

>>107135877
>takes 1.5-3 cycles per character
how would you know that? so far, I have only used tools like perf to see how much time which function takes, but how would you estimate the number of cycles used?

>>107135942
its an estimation
but i take the time a run takes, in seconds
like 1B iterations
then plug in the clock speed
so
(time * clock speed)/iterations = cycles
and it more or less works. it gives figures in the ranges presented in ops picrel when you compare changes

>>107135983
>it gives figures in the ranges presented in ops picrel when you compare changes
(or these you see in the documentation)
im conservative in speeds bc i heard ryzens slightly throttle when using avx2. actually all consumer grade processors do

>>107135983
i aim for a test time more than a specific numberof iterations
like when im bust working i aim for a couple second long test
when i do a full test, then i aim for something like 45 secs
its to drown out possible variations and overhead, including launching the program in the whole test time
it doesnt change bw iteration numbers
so i aim for a total test time instead of a specific number of iterations
also bc stress test of the machine
~45 secs is long enough to see if its gonna throttle i think
this is easy to spot with dual times bc you will see that your big test takes longer per iteration than your fast test. when it should be the opposite

setting up a proper test main takes more effort
you would have to run the test, and the iterations separately, then compute the exact time your functions take
run the test multiple times to average everything out
and then on top of that adjust your results based on processor speeds

its a hassle
i like my tests. quick and dirty, give you all the info you need. its not exactly precise but youre after the relative changes not the exact figures
its like:
you quesstimate that youre gonna see a change of 5 cycles
but you see only 1 cycle difference
its not very important that its actually 1.1 and 4.5 cycle respectively

>>107135942
but also yeah. profiling tools
i used callgrind + a gui interface, i dont remember the name...
its very useful even if in this case it just told me that i got pretty much as far as i could and all i could do now is either fuck around with various elements, try to remove memory fences (thats a very interesting concept btw, if you dont know it you should look it up)
or totally reforge the algo and try completely different ways
i saw that i had no performance hog, btw. thats what i saw. everything pretty much equally balanced throughout the program
i certainly could further refine the algo
but i wanted something fast because i wanted to see if i can hide my parsing completely within the io delay
and so with my ssd, on cold cache it takes ~270k cycles to load a file
my atof takes around 60
its gonna be used ~400 times

thats fukken 24k cycles out of 270k available
80k cycles with warm cache, even if thats not the usecase
i think its already fast enough, kek

>>107136197
>try to remove memory fences (thats a very interesting concept btw, if you dont know it you should look it up)
I know a bit about memory ordering, but I'm not sure it's about that because we're not talking about atomics, but it's also not an easy topic and I haven't done much with it (actually just used sequentially consistent ordering, meaning the worst performing one, but at least I know that there is potential for improvement)

sourcegraph.com is so good

>>107136248
aaah. honest mistake, i should have been more specific-
memory fences in the context of instruction level parallelism and out of order operation
the way you present data dependency is mirrored in the output of the compiler, and it allows, in turn, for your processor to utilize instruction level parallelism
x86/64 processors have multiple instruction slots, and they do run as much instructions as they reasonably can
this depends on the structure of data dependency
which is totally abstract and doesnt exist as a specific syntactic form
it is encoded in the way you design your algorythm
it needs to align loads/stores/various operations in such a way that it makes it easy for the processor to do its out of order magic, so that your instruction slots are as saturated as possible
yes. your cores have cores.
iirc its 4 for regular operations and chud gpt told me its 2 for avx-tier ones. didnt verify it yet. didnt really need to

and the autism doesnt end there because individual instructions have individual latency lists so various combinations can interact in various ways but i didnt need to go that far yet
it is a thing doe
it has an enormous impact on avx2 instructions performance, but i hope this kinda averages out
and apparently it does

>>107136328
>x86/64 processors have multiple instruction slots, and they do run as much instructions as they reasonably can
*simultaneously

Reading OOP code gives me severe fatigue. I think it's all the jumping between little functions across multiple files.

>>107136328
>instruction level parallelism
damn I wasn't aware of that. anyway, good read, bookmarked, will look into this stuff further when I have the mental capacity to do so

>>107136328
>iirc its 4 for regular operations and chud gpt told me its 2 for avx-tier ones.
it depends on the CPU and the operation

>>107136248
>>107136328
ill give you an example of how these phenomena come together
with the atof-
the difference bw parsing 8 chars and 16 is the difference bw 19.5 and ~20.5 seconds for 1B iterations
thats a difference bw ~57 and 60, lets say around 4 cycles.
despite the fact i have half the memops and half the compute operations
but the difference is ~1/15 which is ~6-7%.
why?
because i bump my head against the latencies. the time it takes for the cpu to prepare to run the instructions given. and when i effectively run, i leverage ilp so doing the twice amount of work doesnt change runtime at all
in a way while the cpu is waiting to be able to run certain operations it runs other ones
and so on one side it still has headroom to calculate stuff, an on the other it can also run twice the amount of operations.
the caveat - rule of thumb is that they have to be more or less identical

>>107136484
cheers

>>107136493
noted

I miss registrydumber and maidschizo, this words words words schizo is tiresome.

>>107136529
im high on some fat kush
dw, this wont happen often

>>107136517
you can also get negative cost instructions like xoring a register with itself to tell the CPU to break dependency chains

>>107136549
noice
but for me thats still a couple chapters down the line
i barely got comfy with intrinsics
i think ill take some time to get used to them
like, just finishing a couple projects im currently doing
then ill look into asm, seriously this time
couldnt do that concurrently im working on stuff thats complicated in itself- numbrecrunching and gpu shenanigans of various sorts
the atoi was kindof a workshop to get comfy with intrinsics, i will certainly make heavy use of them down the line

// This
// ```
// #[default_args]
// fn foo(x: i32 = 123, b: bool = true) {
// // ...
// }
// ```
// should expand to all of the stuff below

struct FooArgs {
x: i32,
b: bool,
}

impl Default for FooArgs {
fn default() -> Self { Self { x: 123, b: true } }
}

impl From<(i32, bool)> for FooArgs {
fn from(value: (i32, bool)) -> Self { Self { x: value.0, b: value.1 } }
}

impl From<i32> for FooArgs {
fn from(value: i32) -> Self { Self { x: value, ..Default::default() } }
}

impl From<bool> for FooArgs {
fn from(value: bool) -> Self { Self { b: value, ..Default::default() } }
}

impl From<()> for FooArgs {
fn from(_value: ()) -> Self { Default::default() }
}

fn foo<T: Into<FooArgs>>(args: T) {
let FooArgs { x, b } = args.into();
// ...
}

fn test() {
foo((12, false));
foo(12);
foo(false);
foo(());
}


Would this be based or cringe?

Excellent auto-complete no-more menial typing.

>>107136612
>rust
cringe by default

>>107133087 (OP)
>>what are you working on?

I'm working on a simple command line tic tac toe game in C, as im learning it as my first language

>>107136470
with non-shit OOP code you need to read only very little of it at a time
if you have to read a lot and jump a lot, chances are it's poorly written code, which would be even worse to follow if it the same person that wrote it did not use OOP

>>107133099
Bro 1999 was like 100 years ago.

>>107133987
import minecraft
minecraft.run()

"Look mum I made minecraft"
"Oh you're so clever little timmy"

>>107133987
which problem?

I don't really understand that mindset.
>b-but python has the library
Yeah if it's not there, you can write it. Share it to others. Be a good sport.
>b-but you can't use libraries in C
I want to kill every sepplestard and rust tranny.

>>107133442
Those are called metric units anon.
The europeans use them, I hear.

>>107133442
>cm
30cm ~ 1 foot
>km
30km ~ 20 miles

>>107137171
$ cat test.c
int main()
{
bool gender = 3;
return gender;
}
$ cc -std=c23 -Wall -Wpedantic test.c
$
What did C mean by this?

>>107137649
anything > 0 is male, 0 is female, < 0 is a failure

>>107137485
People already use metric time for arithmetic with hours (1.5 + 2.7 = 4.2h, rather than 1:30 + 2:42 = 4:12) but cope about the république

based rust
https://www.youtube.com/watch?v=Klq-sNxuP2g

>>107137777
>destructively moves penis

>>107137777
Quads

Referenced in the video. What an unserious language, great defect handling
https://cplusplus.github.io/LWG/issue2116
>Opened: 2011-12-09
>[2017-01-27 Telecon]
>Gave the issue a better title
>Ville would like "an evolution group" to take a look at this issue.
>[2020-08; LWG reflector]
>A poll was taken to close the issue as NAD, but only gained three votes in favour (and one vote against, which was subsequently withdrawn).
>[2022-03; LWG reflector]
>A poll was taken to close the issue as NAD, with six votes in favour. (and one vote against, subsequently withdrawn).
>"Write a paper if you want something else. These traits have well established meaning now.

>>107137777
qrd?

>>107137990
Non-destructive moves are the worst of all worlds. It's a thorough video that can't be quickly summarized without seeing the several examples provided.

You can read the description yourself

>>107138064
I mean what should I do.
const auto& everything and forbid std::move or what.

>>107137649
C does not mean anything by it, the compiler is evaluating that as gender = 1.

Python interprets the exact same way, this is not a smoking gun.

>>107133087 (OP)
How is it so expensive to catch an exception ? Are they measuring an "average" exception catch that needs 5 context switches to be caught ?

>>107138076
It's much more complicated than that

>>107137777
Sad that this is going to go way over the heads of the typical Cnile in these threads.

>>107138124
>video ranting about c++ bullshit
>HURR DURR CNILE HAHA COPE
You say this like we haven't been complaining about C++ for decades.

>>107136883
https://godbolt.org/z/45PbsTd1x
int putchar(int);
void put(const char* s) { while (*s) putchar(*s++); }
void print(int board) {
for (char a[] = " XO.", i = 0; i < 9;) {
const int c = (board >> i & 1) + (board >> (i + 8) & 2);
*a = '1' + i;
putchar(" ["[!c]), putchar(a[c]), putchar(" ]"[!c]);
putchar(" \n"[i++ % 3 == 2]);
}
}
int victory(int board) {
const int base = board >> 18;
const int lines[] = {0111, 0222, 0444, 0700, 070, 07, 0124, 0421, 0};
for (const int* line = lines; *line; ++line)
if ((board >> base & *line) == *line) {
print(~(*line << (base ^ 9)));
put("Player "), putchar("OX"[!base]), put(" Wins!\n");
return 1;
}
return 0;
}
int tied(int board) {
print(board);
if (((board | board >> 9) & 0x1FF) != 0x1FF) return 0;
return put("Tied game!\n"), 1;
}
int getInput(int* board) {
int getchar();
const int base = *board >> 18;
for (;;) {
put("Player "), putchar("OX"[!base]), put(" [1..9]:\n");
for (;;) {
const int c = getchar();
if (c < 0) return put("Bye!\n"), 0;
const unsigned d = c - '1';
if (d >= 9) continue;
if ((*board | *board >> 9) >> d & 1) {
put("Position "), putchar(c), put(" is not empty!\n");
break;
}
return *board |= 1 << (base + d);
}
}
}
int main() {
for (int board = (print(0), 0);
getInput(&board) && !victory(board) && !tied(board);
board ^= 9 << 18);
}

>AVX512_BMM
>2 parallel 16x16 non-transposed fused BMM-accumulate (BMAC) with OR/XOR
reduction
bit wizards we fucking won
https://sourceware.org/pipermail/binutils/2025-November/145449.html
Fuck off encryptards your GF(2) is ours now

>>107138076
There is nothing you can do. You either add an small overhead and extra logic to every form of RAII or you write things in C way. There is just no way to archive both zero cost and safety in C++ in this case.

If my company doesn't pay me for two months claiming to pay me later is this true or are they going to string me along claiming they'll pay but never do?

>>107138076
#define def auto const

>>107138441
There's no extension called that

>>107139011
As far as I can tell, that's the first public reference for it. That's coming from AMD so unless they change their mind, Zen 6 will have it.

>>107133087 (OP)
>Actually informational image
>Furry

>>107139070
Many such cases!
(Welcome to tech).

>>107138973
Aliases time?
using namespace std;
#define null nullptr
#define ct const
#define cx constexpr
#define let const auto
#define letx constexpr auto
#define stx static constexpr
#define inl __attribute((always_inline)) inline
#define noinl __attribute((noinline))
#define expect(x) __builtin_expect(bool(x), 1)
#define uncommon(x) __builtin_expect(bool(x), 0)
#define countof(x) (sizeof(x) / sizeof((x)[0]))

I will now use C++

>>107136612
Just write a simple macro
macro_rules! with_defaults {
(fn $name:ident( $($aname:ident : $atype:ty $(= $aval:expr )? ),* ) $(-> $ret:ty)? $body:block) => {
fn $name(args: impl Into<${concat(__,$name,Args)}>) $(-> $ret)? {
let ${concat(__,$name,Args)}{ $($aname,)* } = args.into();
$(let $aname = $aname $(.unwrap_or_else(|| $aval))? ; )*
$body
}

#[allow(non_camel_case_types)]
struct ${concat(__,$name,Args)} {
$($aname: with_defaults!(@if($($aval)?) { Option<$atype> } else { $atype }), )*
}

with_defaults!{@gen $name () () ($( $aname : $atype $(= $aval )?, )*)}
};

(@if($($arg:tt)+) { $($then:tt)* } else { $($else:tt)* } ) => { $($then)* };
(@if() { $($then:tt)* } else { $($else:tt)* } ) => { $($else)* };

(@gen $name:ident ($($sname:ident : $stype:ty,)*) ($($uname:ident : $utype:ty,)*) ()) => {
impl From<($($stype,)*)> for ${concat(__,$name,Args)} {
fn from(($($sname,)*): ($($stype,)*)) -> Self {
${concat(__,$name,Args)} { $($sname: $sname.into(),)* $($uname: None,)* }
}
}
};
(@gen $name:ident ($($set:tt)*) ($($unset:tt)*) ($nname:ident : $ntype:ty, $($rest:tt)*)) => {
with_defaults!{@gen $name ($($set)* $nname : $ntype,) ($($unset)*) ($($rest)*) }
};
(@gen $name:ident ($($set:tt)*) ($($unset:tt)*) ($nname:ident : $ntype:ty = $nval:expr, $($rest:tt)*)) => {
with_defaults!{@gen $name ($($set)*) ($($unset)* $nname : $ntype,) ($($rest)*) }
with_defaults!{@gen $name ($($set)* $nname : $ntype,) ($($unset)*) ($($rest)*) }
};
}


with_defaults!{
fn kek(a: i32, b: bool, c: f32 = 0.5, d: String = "foo".to_string()) {
let g = a;
println!("{a} {b} {c} {d}");
}
}

#[test]
fn test_kek() {
kek((1, true, 3.14, "hello world".to_string()));
// 1 true 3.14 hello world
kek((1, true))
// 1 true 0.5 foo
}

>>107139593
How could you do it for methods though

anyone here used DDS before? Are you supposed to use different topics for different instances of the same data stream, or are you supposed to add a @key field and send everything on the same topic?

>>107139620
macro_rules! with_defaults {
(impl $sname:ident { $(fn $name:ident($self:ident, $($aname:ident : $atype:ty $(= $aval:expr )? ),* ) $(-> $ret:ty)? $body:block )* }) => {
impl $sname {$(
fn $name($self, args: impl Into<${concat(__,$name,Args)}>) $(-> $ret)? {
let ${concat(__,$name,Args)}{ $($aname,)* } = args.into();
$(let $aname = $aname $(.unwrap_or_else(|| $aval))? ; )*
$body
}
)*}

$(
#[allow(non_camel_case_types)]
struct ${concat(__,$name,Args)} {
$($aname: with_defaults!(@if($($aval)?) { Option<$atype> } else { $atype }), )*
}

with_defaults!{@gen $name () () ($( $aname : $atype $(= $aval )?, )*)}
)*
};

(@if($($arg:tt)+) { $($then:tt)* } else { $($else:tt)* } ) => { $($then)* };
(@if() { $($then:tt)* } else { $($else:tt)* } ) => { $($else)* };

(@gen $name:ident ($($sname:ident : $stype:ty,)*) ($($uname:ident : $utype:ty,)*) ()) => {
impl From<($($stype,)*)> for ${concat(__,$name,Args)} {
fn from(($($sname,)*): ($($stype,)*)) -> Self {
${concat(__,$name,Args)} { $($sname: $sname.into(),)* $($uname: None,)* }
}
}
};
(@gen $name:ident ($($set:tt)*) ($($unset:tt)*) ($nname:ident : $ntype:ty, $($rest:tt)*)) => {
with_defaults!{@gen $name ($($set)* $nname : $ntype,) ($($unset)*) ($($rest)*) }
};
(@gen $name:ident ($($set:tt)*) ($($unset:tt)*) ($nname:ident : $ntype:ty = $nval:expr, $($rest:tt)*)) => {
with_defaults!{@gen $name ($($set)*) ($($unset)* $nname : $ntype,) ($($rest)*) }
with_defaults!{@gen $name ($($set)* $nname : $ntype,) ($($unset)*) ($($rest)*) }
};
}

struct Foo;

with_defaults!{
impl Foo {
fn kek(self, a: i32, b: bool, c: f32 = 0.5, d: String = "foo".to_string()) {
let g = a;
println!("{a} {b} {c} {d}");
}
}
}

>>107113592
>>107133556
Searches with popcnt check, around 32 bits, in random data, are getting wrecked by normal distribution. They keep entering the false positive branch.
Any ideas?

if (num_bits < 64) {
u32 bk = 0;
for (; bk < num_buckets - 1; ++bk) {
const auto left = BIT ? buckets[bk] : ~buckets[bk];
if (popcnt(left) >= num_bits) {
auto runs = left;
auto num_b = num_bits;
while (num_b > 1) {
const auto step = num_b >> 1;
runs &= runs >> step;
num_b -= step;
}
if (runs) return idx(bk, tzcnt(runs));
}
const auto right = BIT ? ~buckets[bk + 1] : buckets[bk + 1];
const auto ones_left = lzcnt(~left);
const auto ones_right = tzcnt(right);
if (ones_left + ones_right >= num_bits) return idx(bk, 64 - ones_left);
}
const auto last = BIT ? buckets[bk] : ~buckets[bk];
if (popcnt(last) >= num_bits) {
auto runs = last;
auto num_b = num_bits;
while (num_b > 1) {
const auto step = num_b >> 1;
runs &= runs >> step;
num_b -= step;
}
if (runs) return idx(bk, tzcnt(runs));
}
}

>>107139698
make it uglier

>>107139739
I can't. It won't fit on single post.

>>107139726
Found a way to make the worst cases ~30% faster.
#define bytes_less_cnt(x, n) (((~0UL / 255 * (127 + (n)) - ((x) & ~0UL / 255 * 127)) & ~(x) & ~0UL / 255 * 128) / 128 % 255)
#define full_bytes_cnt(x) bytes_less_cnt(~x, 1)

const i32 min_bytes = num_bits / 8 - 2;

if (popcnt(left) >= num_bits && i32(full_bytes_cnt(left)) >= min_bytes) {

>>107133087 (OP)
adding twitch shitposting to mpv

>>107137756
A mile is called that because it is defined as 1,000 paces (left foot hitting the ground 1,000 times). Unless you're in the "space is fake" camp, It's a measurement that connects humanity to the cosmos:
https://youtube.com/watch?v=Eahn_ttWG28

>>107139726
>>107140167
--disclaimer: i havent drunk my morning kofi yet--
id say: use different methods for different lengths
its one conditional branching out at the beginning of your code then its all benefit.
im too asleep to think straight yet, this the the 4th thing i do today

>>107139726
>>107141679
cont
and since its 32 bits were talking about
maybe the following might be fast:
you reinterpret your array as 32bit wide numbers
and then you can merely check for leading ones.
a 32-bit+ wide space is guaranteed to either straddle two 32bit wide numbers or fill one completely so youre guaranteed to find a valid solution by looking at leading ones and ignoring the rest

is there a good scripting language that you can use to replace Posix-shit? i guess Perl, but not sure if there are better alternatives in the current year, the big ones, ruby, lua, python, are quite clunky and shitty at executing shell commands

>>107139011
>npc brain moment

>>107141773
>poosix
idk i use bash, but theres a variety of shells around

>>107141798
bash has some extensions on top of posix, but it doesn't make it much better

>>107141810
yeah, it doesnt look/feel much different from og shell
but theres a ton of alternatives like zsh.
im not shilling em, just saying they exist
personally i dealt with zsh at school for all of 15 seconds until i decided its not worth it and fell back onto shell for my scripting needs
otherwise i use bash on my machine. or paiton.

>>107141773
Go ahead with perl, it's excellent for scripting.

I finally installed Eitaa, the Iranian Telegram clone. I'm free of Russian tyranny.

These are the hottest hashtags on the app. Have ChatGPT translate them for you. You'll find out how much us Iranians hate certain countries. Enough to make it our national personality to hate them.

(I'll save you the work, it's Death to Israel and Death to America, but see, if two down from the 1980s, when we used to shout Death to Britain and Death to USSR as well).

And, we literally mean "Death". It's not a figure of speech, we really wish 'Death', with capital D, upon these nations. Some people translate it as "Down with..." which is literal historical revisionism, as the history is happening. Disgusting.

>>107142161
Here's how the Linux version looks like. It's literally a copy of Telegram. It has a "Saved Messages" feature as well, so I can back up my files.

It's not like Discord where you can register without a phone. You need to register with a phone first. Thankfully, my boss gave me some cash to buy a phone with. I bought some cheap Xiaomi. I think the total cost of the phone ended up 81 USD. Phones are cheap these days.

I don't know why anyone would buy a more expensive phone. My brother installed F-Droid on it. I installed a terminal emulator (Termux) and Vim on it. Vim on a phone! Amazing.

I realize why Americans buy iPhones instead of Android phones. They are slaves to late-stage capitalism. I don't know why Iranians would even bother with this "Death to America" malarchy, when America itself it digging its own grave. I mean, 1000 for a phone? Besides the camera, what other features down it have? Your messages not showing as green in text threads? Why do you allow this shit to happen?

>>107137582
>30cm ~ 1 foot
gnometric system

>>107141773
I use fish

>>107142472
Up top. Yay. Been using it for 3 years now. POSIX shells (Bash, Korn Shell, ZSh) are for stuck up corporate losers. Fish is better for personal use. I, not only use it as an interactive-friendly shell, but also, as a tool to manage my system. e.g. backing up.

This girl, Bread on Penguins, she uses ZSh. I fucking hate her for that. Don't get me wrong, I'd forcibly make her my wife if I could, but she should not make Zsh.

https://www.youtube.com/watch?v=oTpA5jt1g60

> solution to Windows is more telemetry

Goddamn, davepl is one retarded scammer. Why do I even click on his videos. This is the same retard who thinks LLMs are going to fully replace programmers "except those who do embedded development". This is the same person who was told by another scammer that he's 100% autistic, at the age of 50 no less, and proceeded to write a book about being a successful autismo.

Also, he's from Regina. lol.

>>107142726
AI is going to replace youtubers streamers and shills

>>107142775
It already did. I just watched the latest video by "YouTux Channel". This guy unashamedly uses a TTS for his videos, but I still prefer it to whatever davepl craps out. YouTux Channel manages to put out one video a day, just because he does not care if it's a TTS. He's probably an EFL speaker like me who knows nobody likes foreign accents. I especially hate it when speakers of other Germanic languages speak English. It sounds 'wrong' somehow. Like when a Kurdish person speaks Persian (which, given Kurdish and Persian are both Iranic language, is a similar situation). When, say, a Swedish person speaks English, I wanna reap my soul out and sell it to the devil. It just sounds 'wrong'.

I cannot wait until we get to generate our own entertainment. I wanna see people like Asmangold suck dick for money. I don't know why people compare Charlie to Zack. They physically resemble each other, however, Charlie is a university-educated, pleasant, funny person. If he was not a Youtuber, he's be training Hollywood starts (I think his major in college is related to physical education?). Asmangold is one of those guys who, in absence of their edgy fans, would be working in a gas station right now. No shade on gas station workers.

I think people are tired of people like Zack. Most people watching his streams are clippers, or rage-watchers. At some point, these edge lords are going to grow up and smell the cinder. The only streams I watch are those who I am sure will 100% respond to one of my messages, that is, people with like 3000 concurrent viewers.

>>107142858
what the fuck are you talking about

>>107142880
Zack -> Asmangold
Charlie -> Moistcritical

Do you live under a bridge?

>>107142890
Why do you know so much about streamers?

>>107142910
The Youtube Recommender system was not kind to me. I did not sacrifice enough goats at its altar.

>>107138204
Why are you complaining about C++ if you're the so called C enjoyer? People whose languages aren't garbage don't have to complain about anything.
>>107138845
There's no overhead in my RAII code, just because you're a nocoder, or majority of subhumans who call themselves "programmers" are total nocoders, doesn't mean C++ is broken.

>>107142890
>Asmangold
I've seen him in my recommended list, watched maybe 2min worth of his content
>Moistcritical
never heard of him, you spend way too much time on youtube

>>107142910
He's unemployed.

>>107143096
> you spend way too much time on youtube

You think? This video just popped into my feed:

https://www.youtube.com/watch?v=ZRJTgmGWY5c

Can you believe this a real man? As in, he did not used to be a lady. I don't believe it myself. But I remember in one of his videos, he said he's 'le cis'.

The fact that I even know this, makes me sad. I don't know what to do. I hate being on jewtube. I used to nuke my history every month. I've set the quality to 144p so maybe I'll get tired of it. But it's gotten worse than before. I used to pause the videos when I started coding, now a video must be yapping in my ears.

I do have an excuse, though. They help me learn English. Not as much as interactions on 4chan do, but still, I've gotten a lot better at parsing sentences ever since I fell into the jewtube trap 10 years ago. I can watch any movie without subtitles. In fact, I hate subtitles now. I don't watch non-English movies because I don't like subtitles (I watch Persian movies too, but only in the theater --- that's the only place you can get nachos here).

>>107142658
>Bread on Penguins
she looks kinda weird, is she Jewish?
Wouldn't mind pumping her full of my seed regardless.

ZSH isn't too bad, it's just that you need like 10k lines of config to get it to the same level of comfiness as fish's defaults.

>>107143183
She looks "approachable". Not too beautiful, not too ugly.

She needs some orthodontic work, though.

what git host do you use? github? gitlba? gittea? gitgud? bitbucket?

>>107143422
Github

Early MMC5 support in my nes emulator
https://files.catbox.moe/83rudq.webm

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

This is why I love Youtube. Absolutely worthless info. I wonder if they were gonna teach us any Beowulf had I stuck to my English literature major (which is wishful thinking, English lit in non-Anglo countries is basically American high school --- they had us reading fucking The Great Gatsby when I dropped out. I hate that book so much).

>>107143422
git-daemon

>>107143422
git-daemon

I fucking love chugging down a glass of milk. My superior lactose-tolerant Iranian genes allow me to.

>program seemingly runs fine on Arch
>move to Debian
>same program segfaults
>debug and find the cause (using a va_list twice instead of a copy), and it really should be segfaulting on Arch, but isn't
Is this the fabled "undefined behavior" in action?

>>107143640
Arch is a rolling distro. Debian is not. That's your answer. Check the changelog for libc, they probably changed something in the latest version that causes this.

>>107143614
If you're thinking about this at all or can't eat a whole meal of cheese until you're not hungry, you're a shitskin.

>>107143736
Damn, you're totally right.
Debian's on 2.41, Arch on 2.42
The following bug got fixed in 2.42 (after being reported 10 years ago lol: https://sourceware.org/bugzilla/show_bug.cgi?id=18982

>>107143865
Butthurt chink detected. Look at that map I posted, zero percent of Iranians are lactose intolerant. I just drank two liters of milk (honest to G-d), chased it down with a glass of olive oil, and a bottle of cold water. You must see our alcohol spirits. Araq, it's called. It's mostly alcohol. Only pissant homosexuals drink fermented beverages like wine. I'm a superior milk/olive oil/araq drinker. These are drings a chink like you cannot handle. Don't worry, at least your kind is better at math (like NERDS lol).

>>107143872
Yeah never test on rolling distros. The idiots who made Ghostty (I just got AIDS thinking of this piece of shit, troon-tier terminal, I use Alacritty, because I am an Aryan) had their Fish init function made on a rolling distro, so it did not work on Pop!_Os. I switched to Arch (through EndeavorOS) just because I was tired of the idiots who weren't mindful of distros that use older versions.

>>107144017
>map
>iran
>no data
lmao, yet another instance of schizo not being able to read. Go take your meds.

>>107144040
> no data

It does not say that, troll.

Anyways, I've never seen a single person in my life who's been lactose-intolerant. We drink all kinds of milk. Sheep, goat, cow. Go fuck yourself, chink.

I will no longer have this disucsion, because it's silly.

>>107144235
Typical schizo behaviour

For benchmarking with rdtsc, you want variable non-constant tsc right?
So if the CPU goes into turbo/sleep, it won't make your code unfairly faster/slower.
The tsc should slow down if the code slows down.
But that could still be biased if RAM and CPU reclock out of sync (does that even happen? I'm keeping my memory usage low anyways [but that doesn't guarantee it will be in the CPU cache]).

>>107144262

I found the official dataset for the yearly production of cow milk in the past 6 years:

https://amar.org.ir/Portals/0/Statistics/2702z142911404-14040812172721.xlsx

Have ChatGPT translate it for you. 2721 thousand metric tonnes, JUST THIS LAST SUMMER.

Get fucked chink. Your ilk cannot digest milk. So you are misrepresenting that map. If there's no data on Iran, then there's no data on other countries of that color (which is a lot of the map). Are you telling me there's data on African countries, but not Iran?


btw, this website has tons of datasets if you are looking for them: https://old.sci.org.ir/english

Just search for 'dairy' in this website. But you won't because you're a tofu-eating chink.

>>107144270
What kind of retarded benchmark are you running where CPU isn't pinned at 101%?
Either way, rdtsc is meaningless and mostly useless, its value means absolutely NOTHING, you can only compare if one benchmark produces consistently lower number than another, that's it.

>>107144340
Uhhh, a game?

>>107133087 (OP)
How THE FUCK is main ram read SLOWER than C++ virtual function call?

>>107144371
Why would dereferencing a pointer that points somewhere in general icache be slower than accessing main RAM?
>>107144369
You don't benchmark "a game".

>>107144371
the assumption is that the function is cached

>>107144336
that website is either down or geoblocked

>>107144235
he's right, Iran is marked grey, i.e. "no data" on that map

>>107144415
You could have just said you don't know how to do it.

>>107144415
>You don't benchmark "a game".
modern AAA dev spotted

>>107144433
You can only profile a game, midwit, if you want to benchmark a function, go ahead, don't call it a game.
>>107144432
I'm not the one expecting good advice on 4cuck. But sure, go ahead, tell me more about things I don't know, nocoder.

I'm writing an article based on this paper for my SCA job:

https://www.ndss-symposium.org/wp-content/uploads/2025-559-paper.pdf

So I implemented it using Sonnet 4.5 through a 36-chapter book (and several appendices):

https://chubak.neocities.org/xcrunity-dossier

(if you want a thorough introduction to UAFX, read Appendix A).

This is admittedly grade-a slop, but it will help you get familiarized with pointer analysis, kernel idioms, UAF, and most importantly OCaml, because it uses OCaml and its ecosystem to implement XCrutiny (my implementation of UAFX).

It uses the CIL library as a C frontent, and SMTML + Z3 to solve the constraints.

If you like this book, help me create a better Pandoc template for them. I've made 77 "dossiers" so far (I call my slop 'dossiers'). For example, this is my "Ultimate Neovim Dossier":

https://chubak.neocities.org/nvim-ultimate-dossier

It uses a prior template, which is much worse than the current one. Still, if anyone can help me make a better template, lemme know.

I explained my process of creating these slop a few days ago. If you're interested, I'll explain again. I have made several Fish functions that help me create these slop in a streamlined manner.

https://hackage.haskell.org/packages/tag/aws
>1 million amazonka-X packages
>1 million stratosphere-Y packages
Why is this allowed

>>107144584
I'm not in-the-know of Haskell ecosystem (despite wanting to be). Assplain why this is bad anon.

I'm so tired I can't keep my head up.

>>107144705
because theres a million packages specific to one fucking thing by the same people? how are you supposed to see whats new or find shit when theres a million fucking ecosystem-within-an-ecosystem packages

>>107144811
This is why Haskell will never be good, it's not the syntax or monoidism or whatever, it's same cancer that's killing Python, JavaScript and Rust.

Please help me get a into Jai beta. What do I need to write or do in order for JBlow to look at me

>>107144849
Don't bother. I tried the version that leaked on here a few months ago and it sucked.

>>107144720
Why not go to sleep?
Is the 20hrs a day locking in meme real?

>>107144869
Afaik that version was very very old

>>107144958
I've taken 5 Ritalins and 2 Vyvanses since I woke up 12 hours ago and I think I'll be woke for the next day or so.

Get on my level, nerds.

One negative side of my usage of Methyl Fenidite and Lisdexamphetamine is, I am kinda afraid to consume caffeine now, since I might actually get a heart attack. So I drink tea mostly. My mom just bought me packets of Stevia (which is hard to find here in Mashhad) and I drink like 30-40 liters of tea every fucking 'session of wokeness' (I realize 'wokeness' is not a word, but people use it for other semantics, and I'm going to repurpose it).

>>107145064
If I got to your level I'd hit a new PR in lowest point in my life because I was everything, but drugcattle? Never.

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

what a loser man. collecting coins? lol. I had a friend in college who collected coins, he ended up getting his master's in archeology, but he dealt coins and made money from it. If you're into numismatics, or any 'collection-based hobby' and you don't make money from it, you're a fucking loser.

No shade, I once collected language implementations and OS implementations. I formatted my SSD so it's all gone. I wanna get back at it because honestly, it's a literally free hobby, and besides, it's related to my job (SCA, as established).

So, do you guys know where I can restart my hobby/profession-necessary act of collection OS's and language implementations?

I did not just collect 'mainsteam' OS's and languages. I had some obscure ones.


>>107145207
Drugs are awesome. Shut up. My dad did drugs and he was a successful dentist (he smoked opium, and it gave him the energy to work til odd hours of the night, on a thankless, boring job which he hated --- he wanted to be an electronics engineer. He quit opium, and he felt drained, so he had to prescribe himself Ritalin and shoot it up, finally, his body gave in. Take drugs, but never, ever quit them).

>>107145243
>Drugs are awesome.
>I am fatherless btw.
Thank you for your input.

>>107145267
My father died at 41 but he would be 59 now, so he's be dead either way. He was bipolar, and people with bipolarity usually kill themselves, even if they are happy. I'm bipolar too, and I plan to kill myself when (and if) I reach 54. I think it's a nice age to off yourself. Everything after that would be pain. Your brain stops working, your body stops functioning, and people will call you an old man. Better just flip the switch and kill yourself.

>>107145294
Could you hurry up and do it right about now?

>>107145316
I wanna fuck a girl at least one before I die, so no.

>>107139593
I cooked a proc macro https://pastebin.com/bZCXBR6B

default_args! {
fn foo(x: i32, y: i32, b: bool = true) {
println!("{x} {y} {b}");
}
}

fn main() {
foo(1, 2, ());
foo(1, 2, (false,));
}

>>107145431
Just use the default macro constructor. What you made does not make hygienic macros anon.

>>107145431
Use proc_macro_attribute so you can just do
#[default_args]
fn foo(x: i32, y: i32, b: bool = true) {
println!("{x} {y} {b}");
}

I am getting filtered by react-native-google-signin on android emulator

literally cannot get it to work with any method. sha-1 matches, google enabled as sign in method on firebase console.
I had this working on ios for months. you'd think it'd be straight forward with google.

>>107145492
The input has to parse correctly for those I guess so it wasn't letting me do the defaults.

>>107145431
>>107145453
>>107145492
In C++ this is just:
#include <iostream>

auto foo(int const x, int const y, bool const b = true) -> void {
std::cout << x << " " << y << " " << (b ? "true" : "false") << '\n';
}

int main() {
foo(1, 2);
foo(1, 2, false);
}

>>107145590
Yeah but it's not worth dealing with all the other shit for that.

>>107145590
>not array[b]
>not "false\0true" + b * 6

>>107145691
I don't have to deal with anything, C++ is perfect and satisfies all usecases.

>>107144584
>>107144811
This is hardly a Haskell unique problem. All ecosystems have ENTERPRISE (tm) libraries where they want to modularize everything to save space and improve maintainer experience (shit is easier to test and distribute) and that's what you get.
In the Maven repository you see the same thing.
AWS services has a HUGE MCGIGANTIC surface and amazona reflects that.
The (now on permanent hiatus) Haskell ecosystem surveys have developers complaining about library documentation (low on startup examples) and library discovery (what the fuck do I use) and lack of enterprise libraries (DB2 drivers for old mainframes instead of raw dogging ODBC).

>>107145571
Ah, ok.

>>107145590
Yea, but that's not fun and doesn't play well with function pointers.
For functions that take variable number of arguments, it's more common to use a structs like:
Foo::new(FooConfig {
arg: val,
arg2: val2,
.. Default::default()
})

With #[non_exhaustive] you can leave room for future additions to these arguments. And you can also replace Default::default() with some builder method FooConfig::some_preset() to have various presets of default values that can be further configured via the syntax above.

>>107145784
You could have described Haskell as on permanent hiatus, it feels dead even though there are some GHC improvements.

>>107145798
>With #[non_exhaustive] you can leave room for future additions
how the fuck is that implemented at runtime

SO WHY DOES EVERY WEBSITE TREAT ME AS IF MY IP ADDRESS IS A SHITTY VPN?

For example, Pastebin just loops its captcha. 4chan keeps asking me to wait 120s.

This does not happen to my LTE subscription's IP. But that connection is too slow (40mbps only, I rarely use it).

Funnily enough, this does not happen when I start a private session (Firefox). Is there an add-on that is causing this?

>>107145827
Oh shit I forgot to cover up one of them.

oh well.

>>107145841
>this IP was reported 80 times

>>107145870
8 times you dyslexic moron.

>>107145821
It doesn't do anything at runtime. It's a marker that prevents users of your crate from assuming that they know all the fields/variants of that type. It just bans things like Foo { a: 1, b: 2 } and match foo { Foo::A => "a"; Foo::B => "b"; } without fallbacks like default branch for match or ..default } for struct construction. This makes sure that if you add another field or variant it won't break downstream code because of non-exhaustive match or missing struct field.

And don't bother DDoSing it, or reporting it to that website. It's a dynamic IP.

I'm going to buy an static one. Except, they make you wait in a queue before they assign you one. And it's very expensive as well.

You know what, I'm going to switch to my slow connection. At least it's got a good IP address.

launching low orbit ion canon rn
IMMA CHARGIN MAH LAZER

>>107145827
I'll report you to Mossad and tell them you're one of Iran's most important programmers

>>107145827
>Iran
There's your problem.

Thinking about it more probably want to try to do something like
#[default_args(b = true)]
fn foo(x: i32, b: bool) { /* ... */ }

and then with impl blocks
#[default_args]
impl Foo {
#[with_defaults(b = true)]
fn bar(&self, x: i32, b: bool) { /* ... */ }
}

>>107145875
didn't get the joke award

>>107145948
I hope you won't use it for anything serious.

>>107145908
but you'd still need to recompile?

>>107145989
Yes, this is not some polymorphism or unsizness. It doesn't change ABI of a function or binary representation of that struct. It's only a hint to compiler to act as if this type had some hidden private fields or variants and a hint to developer that this type might get new fields or variants in future version of the library.

>>107145983
Everything I do is srs bsns

>>107145989
>>107146024
Also it's means of futureproofing. Adding new field to a struct that could previously be fully constructed by client code would be a breaking change. This way, you are sure that client code has fallback for the new behavior or state you want to introduce. This of course is a double-edged sword because even if the code still compiles, it doesn't mean it will work correctly and it might be more difficult for the developer to figure out where exactly something new was added that is causing problems.

>>107145942
You could do that, but after we embarrassed the respectable Jewish folk this summer (www.bbc.com/news/articles/c2dj217z2w6o & www.youtube.com/watch?v=UFTONTxiZGE), the government 'fixed' all the leaks, so there's literally zero "Fifth Column" people remaining to assassinate me.

I know you're joking, but my SCA job concerns national security -- not cyber security, application security. I'm still a junior, I'm still translating papers and learning the theoretical foundations of SCA. But one day, I'll validate the programs that is hopefully going to kill respectable Jewish persons of the totally legit land of Israel. I'm not joking.

My boss told me that, they had hired 5 programmers for some nook shit, and these were top-notch programmers, but they made an error that caused 2.5 billions of damage (not sure if USD or IRR, I think it's USD because 2.5 IRR is price of a Chinese EV, or, it could be back when IRR was worth something). My ultimate mission as an SCA specialist is to aggregate all the research on program security, and discover ways of making code more secure.

Building compilers and operating systems comes later. Some sort of "National Compiler" or "National OS". I call them "NaCC" or and "NaOS" (like Nabisco).

>>107145943
Could be. My LTE connection's IP address (the one I'm posting with rn) does not even show up in that website's database. I think the government uses that IP address in cyber attacks and stuff like that.

I still run into the 'endless captcha loop' problem with this one, too, but I'm sure the issue is with one of Firefox extensions, because they work in private mode.

My bet is on uBlock Origin. Nobody likes this addon.

>>107145805
It's the same status quo as before the years where FP was the fad of the moment (2010s). A test ground for FP ideas. It's now THE FP language despite Jane Street pumping OCaml with money and development, and Scala waning.
In truth everything is declining towards a plateau of enshittastability. People leaving Java for Python and Go for Java and JavaScript for TypeScript and TypeScript for Go and Python for TypeScript and just making more and more unmaintainable slop.

>>107145798
I can do this in C++ too.

>>107146198
FPs are fun to make. I've started several FP projects. The one I most progressed with was VEBLEN (Veblen is Church's professor). Check it out: https://github.com/Chubek/VEBLEN

It has several useful shit in it. For example, the AST for Lambda calc. The one in SPJ's old book.

>>107146198
It's not really a test ground anymore, they've been pushing "professional" bullshit, committeeism and d*mocracy, code of conduct faggotry and "beginner friendliness"
they're crippling all the new features and libraries

erlang is King

>>107146211
Yeah. But that's also the general convention in Rust.

>>107146272
>they've been pushing "professional" bullshit, committeeism and d*mocracy, code of conduct faggotry and "beginner friendliness"
All of these people left for Rust actually. Only a bunch of fags remain.
>they're crippling all the new features and libraries
Crippling how? The community adopts new things as they increase the language edition. There are no pointless breaking changes like outright removing partial functions because all breaking change integrations are reserved for like Linear Haskell or Dependant Haskell which are lone wolf efforts because GHC is huge and the spergs work for Rust now.

>>107146283
If you love Erlang so much, read Joe Armstrong's thesis (if you haven't yet):

https://github.com/Chubek/chubek/blob/master/joe-armstrong-concurrency-thesis.pdf

Correct me if I'm wrong here, but Erlang's model of concurrency is based on Hoare "CSP" model. It's quite unlike those reasoned with by Lamport and others.

Golang's model of concurrency is similar. Based 'Green Threads' --- or, lightweight, 'virtual' threads of concurrency, that rely on cooperative multitasking (as opposed to preemptive). In practical implementations, we use "N:M" model of Green Threads, wherein, we map N 'hard' threads, where M > N, to the pool of cooperative threads. In order not to confuse them, we call cooperative threads of execution 'fibers'.

I plan on implementing ECMAScript with focus on systems programming. This implementation with expose two namespaces, "Fiber" and "Channel". They are basically goroutines and go's channels.

const foo = Fiber.spawn(x => x + 1);
const fourchan = Channel.createBidi();


I have generated a 52-chapter book with Sonnet 4.5 to help me implement it.

https://chubak.neocities.org/quishejs-dossier

In reality, they are interoperable C libraries. Because my implementation will have a top-notch C API, that enables all the native libraries to be implemented via the said interop layer.

(I know, I know, Node.js exists, but it relies on V8, it does not implement its own engine).

I wanna make another version of the dossier, which implements it in Rust instead of C. But I am not as comfortable in Rust, as I am in C. I think I need to implement something simple, like Awk, first in Rust. I already have several Rust programs, but they are all smol, like a pager.

>>107146462
They didn't, these commitees they have now are fucking useless, in fact it's probably on purpose so that nobody proposes anything and they have no obligations.

>>107146482
The committees are a relic maintained by inertia because Haskell was always a design by committee language since the beginning.
People still propose things, the issue is that proposals should have a viable implementation to figure out if it should go or not, otherways you end up with OverloadedRecordUpdate, a half baked accepted effort with no end of sight because it's not trivial to solve without giving up half of the proposal objectives or shitting all over compilation times, compiler complexity and overall performance.

>>107146574
All that really matters is who is proposing it, it's longhouse bullshit
We're still missing tons of shit in base meanwhile they expect every proposal to build all of fucking stackage instead of accepting there will be breakage, it's literally insane

>>107146598
Also overloaded record update is worse than generic lens. It's astonishing there's still so much denial about the better ways of doing things that have been developed in Haskell

>>107146574
>>107146598
Chris Okasagi says "Type system in functional languages is like the dancing bear, wonder is not that it dances good or bad, it's that it dances at all".

I read in a book from 1987 (I think a book on category theory) that "Purely functional languages are only possible with the super-parallel computers of tomorrow".

Haskell should not exist, and GHC is not be this fast to compile. It's a wonder that it exists even.

>>107146634
Honestly we need a new Haskell, preferably based on something like mu mu tilde / sequent calculus rather than lambda calculus / natural deduction and possibly with linear types from the start. It's also interesting how you can get all 4 evaluation strategies into one language that way. Not sure how side effects would work. Haskell has become too stale

>>107146598
I agree but I also emphatize with the maintainers that are spreaded thin and are responsible for huge swaths of abandoned hackage things. You can convince them of accepting breakage if you bundle a ton of base changes.

>>107146627
Yes and no. The easy implementation of record update is indeed just lens. But there's always somebody that wants to do multi field multi polymorphic type changes that require destructuring and reconstructing the whole thing in one go (one of the things that block implementing overloaded record update).

>>107146634
Yes.

>>107146661
Gonna need to convince college prof nerds to rally around.

>>107146725
>But there's always somebody that wants to do multi field multi polymorphic type changes that require destructuring and reconstructing the whole thing in one go (one of the things that block implementing overloaded record update).
You can't do polymorphic update with the class they added. You can do that with lens just by chaining the changes.

>>107146746
A lens can go from "Thing a" to "Thing b" focusing on one field, and a traversal can go from "Thing a" to "Thing b" where you touch all related fields of the type. But having to touch multiple fields with interconnected types makes impossible to chain the changes. An edge case but the bare bones nature of records give them this power.

>>107146806
You can chain the changes, I'm not sure why you think you can't. You can't do it with the class they've got for overloaded record update.

>>107146817
I mean you can't do polymorphic update with it
https://github.com/ghc-proposals/ghc-proposals/blob/master/proposals/0583-hasfield-redesign.rst

I can't come up with a fun side project to do some programming in my free time. Feels like I've already automated most of the annoying tech related areas in my life. Do you guys know any open source projects in Go or C# that are cool to contribute to?

>>107146926
sorry mate, all the cool projects are in c, c++ or python

If you had to choose one single programming language to write the internet from scratch, which one would it be(assuming it exists)?

>>107147047
what projects, for example, do you consider cool?

>>107136612
gay
Rust is like the kid which tries to be better than everyone; it isn't

i will stick to my c, but do whatever you want

>>107137303
For me the question is about who controls the code. My code is my problem and I can't depend someone else making it theirs if I use their library. I enjoy it more too, may have a slower development speed but I know that my code works as it is supposed to.

>>107138204
fuck that shit, it is just hard to read
also CMAKE

How many Japanese computer scientists do you guys know, off the top of your head.

I know Sato, the author of the two-volume "Theory of S-Expressions".

That gook incarnation of Larry Wall counts, too, if we stretch the definition.

>>107147260
Holy inferiority complex.

>>107143460
>>107143422

Honesty Gitlab is so much better than Github. Sure it needs some getting used to but then you have basically everything there.
Also Github is owned by the same company who develops Windows so I wouldn't put to much faith into them in the long run.

Personally I just have my git repos on my own server. https://git-scm.com/book/en/v2/Git-on-the-Server-Setting-Up-the-Server

>>107147400
I was actually thinking of hosting a gitea instance but my only server is at home

>>107146817
You can chain the changes if you can build a lawful lens on each field.
data Thing a b = Thing
{ field1 :: a
, field2 :: b -> a
}

If you have this you can't do a lens per field or a traversal but you can update everything at once with a record update.

>>107147458
Gitea is great. Have tried it before and really enjoyed it, especially the runners.

If you pull/push a lot you should probably connect it to a domain or add the ip to your host file so that you don't have to constantly update the origins.

>>107147502
o I see what you mean
you could have a lens onto the tuple
also still doesn't benefit from that hasfield class

>>107147400
I honestly do not care. I often contribute to other projects, make forks, etc and having everything in one place is convenient.
I just have to set up private git server that will mirror my github repositories in case they get upset with one of my shitposts.

>>107143422
github is the best if you want to release binaries since people know github releases (but it's a bit hidden, but you can point URL's to the release).
if you want to share full debug C/C++ binaries and open core dumps, it is possible to do everything manually for free, but it's painful.
The easy way is to just copy your source code and rebuild everything into a folder every time you release, so that when you open the core dump, it will match to the version. Then to make the core dump, you can either do it in-process or or use procdump manually/AeDebug or task manager.
This is fine for infrequent full debug builds, but if you had updates every day, this is unmaintainable and this can't be automated into your CI/DI system since the files need to be on your PC you open the core dump on.
So, the solution is symbol servers with symbol indexing, so the core dump can use the source code SVN to find the correct version, but doing this yourself is excruciatingly painful and not documented anywhere.
Microsoft Azure has it's own built in symbol indexing system, and this works with your microsoft credentials (VS studio).
Azure sounds like shit, but you need to remember that automation is not free, github will ask for CC details for closed source code (gitlab seems to be free, but you need to host the server).
And the convenient crash report libraries (using google crashpad like sentry and bugsplat) are not free either (bugsplat is free under a quota, but they require CC details).
BUT you should pay for the services for release builds, because full debug binaries can easily be 500mb uncompressed for a 1mb executable (depending on your code) AND it will combine crashes that are identical.
Note that sentry supports coredumps, which avoids the need to manually send a core dump (but the file must be under 20mb compressed 100mb uncompressed, which might not fit full dumps).
And technically core dumps don't require debug info after it's on a server, but no stacktraces.

>>107147637
If you need to make non lawful lenses then it's over, you make a tuple to do polymorphic wide updates and if you want to do a non polymorphic single update you need to provide either more lenses or dummy values for the tuple.
You're right that the current HasField is just lenses. What I'm saying is that the vanilla update "case Record {a. b. c} -> Record {a = f a, b = g b, c = h c}" is more powerful and has no issues typechecking. Therefore when somebody says "ok lets compromise and do setField just like lenses" somebody says "uuuuh in MY usecase I do this fucky thing with this fucky record, OverloadedRecordUpdate is supposed to just work like the powerful update chud" and progress stalls again.

>>107147804
edit
>full debug binaries can easily be 500mb uncompressed for a 1mb executable
*full core dumps, can easily be 500mb uncompressed for a 1mb executable
>Note that sentry supports coredumps
Storing coredumps as a file, of course it's redundant because you already have a stacktrace.
>And technically core dumps don't require debug info after it's on a server, but no stacktraces.
I mean, you don't need to distribute the pdb file. And I am talking about printing a stacktrace inside your own code.

>>107147858
>If you need to make non lawful lenses then it's over, you make a tuple to do polymorphic wide updates and if you want to do a non polymorphic single update you need to provide either more lenses or dummy values for the tuple.
I don't know what a lens is but it sounds a bit like the Expression Problem, how you can't have it both ways.

>>107148124
A lens is an abstraction of a pair of functions, a getter and a setter. You focus on a part of a thing, and you can get the part or put another part in there. Getting and then putting is the same as doing nothing. Putting and then getting gives you what you put in. Putting the same thing twice is the same as putting it once.
In Haskell, you have lenses as functions and lenses as datatypes depending on the library.
The issue is that lenses aren't as powerful as basic record updates, but that loss of power gives you better modularity and more succint expressions.

https://www.youtube.com/watch?v=c8Phszco8TY

This guy has been paid to made this video.

I don't give a shit about China, I'm not here to defend its honor. But this video sounds to me like the Western propaganda against Iran in the early 2010s.

Chances are, one or two bathrooms have implemented this system, and this guy is making it look like it's end of the world somehow.

In the early 2010s, Youtube wasn't very popular, so the propaganda against Iran came in text format. I can name a dozen instances of outright lies being spread by people who were probably paid to do so.

Again, I don't care about China, I'm just posting this video as an example of how paid shills, knowingly or unknowingly, use propaganda tactics.

I don't understand why Westerners don't want to accept the fact tha their media, although much more freer than media in non-Western countries, are not above spreading propaganda.

Now, this system honestly sounds great. Toilet paper is made from non-renewable resources, and people should not waste it. This is a good system to implement in order to avoid waste.

Also, one thing I notice about this video is, that most public Chinese bathrooms are squat toilets. We have both a squat and a seat toilet at home, and I use the seat toilet because I have a lame leg. Otherwise I would use the squat one because it's easier to defecate with one. Let's not get carried away, what I mean to say is, this system cannot be implemented in more than a few places, because they don't use the kind of toilet shown in the video in many places.

Just be vigilant of shills playing with your perception of reality, anons.

I liked this guy, I used to watch all his videos, but I'm going to subscribe if he's going to spread propaganda so shamelessly like this.

>>107147858
>progress stalls
overloaded update was never going to be useful

>>107148266
And oh, what this guy does not get vis-a-vis weak plumbing is, that if your building has weak plumbing, scrunched up toiler paper is going to clog it, too. I avoid using toilet paper (use water instead) because I have to dispose it in a plastic bag, out building is very old and it uses pipes made out of concrete and, generally, it's a mess. Situation in China must be the same.

This video is so fucking off, and so fucking unrelated to cybersec (which this guy's videos are mostly about) that it must be 100% a shill propaganda video.

If Tim Pool is getting paid by Russia, this guy is getting paid by some entity. I don't knwo and I don't care. I just feel insulted by this video, because I liked this guy, and this video is fucking stupid.

Also, he made another video last week that was in the same vein as this one, sounded so unnaturally stupid.

I've learned not to trust anyone who mentions the so-called "Chinese Social Credit System" because it's been proven time after time that such thing does not exist. It's propaganda cooked up by whoever benefits from making China appear as a draconian Black Mirror-esque society.

I repeat, for the third fucking time, I don't care about China, and why the fuck would I. But I hate it when people spread propaganda, I got PTSD from the early 2010s Western propaganda against my own country, so now I have blood lust for any fool who partakes in West's new boogeyman.

>>107148266
>>107148366
Programming?
Nice matching doubles though

>>107148355
If you write datatypes like relational tables and normalize your datatypes (i.e have dog_name inside Person instead of a nested Dog element with a name) then it's useless because you have 1 depth level. But the "industrial base" has OOP tendencies and writes nested records inside nested records so you need to do sweeping updates.
r { timezone = r.timezone + 3
, residence.country = "Japan"
, things.car.inventory = addTo r.things.car.inventory "Suitcase"
, economics.transactions.travel.tourism = newPlan
}

Lenses work, mostly, because they don't need polymorphic bullshit for data, and so they use that. But there's always some degenerate case that asks for the full record power.

>>107148404
Much obliged.

It's related to programming because that channel is an 'orbiter' of programing world.

>>107148228
Nice, I get it. It's equivalent to a function taking and returning a tuple of the same type, but thanks to the lens indirection you can swap out the tuple independently of the data transformation.
I don't understand why type polymorphism would be an issue though. It should be possible to pattern match and construct a tuple with variable type fields.

>>107148460
what we're talking about is this
if you have (a, b) you can use a lens to the second component to turn the b into c, getting (a, c), you could then swap the a out for d to get (d, c)
but for some types, the types of the two fields are linked, e.g.
data Closure a b = Closure { fun :: a -> b, arg :: a }

with a lens onto fun, you could modify the function to change the return type to c, giving you Closure a c from Closure a b. But you can't modify a by modifying 1 field. You have to modify both fields simultaneously because they both reference a.
Another case is existentials

>>107148266
>I don't understand why Westerners don't want to accept the fact tha their media, although much more freer than media in non-Western countries, are not above spreading propaganda.
I stopped watching TV around 2003 because the Iraq war propaganda was unbearable, and a few years later I found out that the reason for the propaganda is that the media is controlled by kikes.

Now fuck off to /pol/

>>107148460
>It should be possible to pattern match and construct a tuple with variable type fields.
This is called row polymorphism. There are libraries for this: vinyl, HList, large-anon.
The current record implementation is not quite like that because it's a simple bolted on implementation.
>why not have row polymorphism then
https://github.com/ghc-proposals/ghc-proposals/pull/180
It's yet another stalled GHC proposal because it makes signatures look like ass and it also runs like ass because they aren't data constructors with data like normal records so the proposal stalled.

>>107143614
>france is less white than Spain
Baguettebros...

>>107148556
Well it's great to know Americans have gone to the Semitic shop, and swapped out their Jewish Semites with Arab Semites. At least, the people of New York did. When my brother said someone named "Zohran" has won the NYC mayoral election, I said "oh, so another Jew is running NYC now?" and he said "Nooo baawww he's an Arab shut up" (my brother is a "Twitter leftie", too late to save him, he's also flunking dental school because he hates the profession that will make him a millionaire by 40 for some reason?).

Wake up you morons. Jews and Arabs are both semiteshit. Bibi Netanyaho could be the doppelganger of this guy from Mashhad's "Kheen-e Arabha" (basically Chinatown, but for Arabs) whom I used to buy bootlegged araq from.

And these are the same braindead idiots who think Iranians are Arabs. lol. I think they haven't seen enough Iranian faces, or the Arabs they've seen are from Iraq which is genetically more Iranic than Semitic. Iranians and Arabs look nothing alike. Iranians look like Ashkenazi Jews, because we are people of Indo-European origin mixed with Semiteshit, and Ashkenazi Jews are the inverse, Semitic people mixed with Indo-European peoples. Still, there's a lot of variety in the Iranic phenotype. Persian Iranians in general look darker than Kurdish Iranians, I've never seen a Kurdish girl who was not pale af and most Kurdish girls have dirty blonde hair.

So when I see murikkka electing a Semitic person, because they think these kind of Semitic scum is better than the other kind of Semitic scum who has been ruling them for centuries, I just feel cathartic, full of joy, knowing that, Westoids have been ruled by Semitic people for much longer than Iranic people have been ruled by Semitic people.

10,000 people identify as being "Mashhadi Jews", and sadly, some are still living in the city. My dad one bought loads of processed meat products from a deli near his dental practice, and when he found out the deli owner is a Jew, he threw it all out!

>>107144017
>G-d
kike detected

>>107149037
By "G-d" I'm referring to the Indo-European Skyfather. Not Jehovah the storm deity, divorce God of Semitic people: https://www.youtube.com/watch?v=m60p7BEnN5Y

In Persian, we use the word "khoda" to refer to G-d, and this word is cognate with 'god', both referring to the Indo-European Skyfather.

It's sad to see people worshiping a Semitic deity instead of a chad Aryan god.

>>107149088
Jesus fucking christ, take your meds.

>>107149115
I did. But not because you told me.

---

So, I was wondering, 'speech' is a facility yielded by all humans. Even deaf people know speech. In history, 'dumb' people were always deaf, otherwise, all humans can speak... except non-verbal autistic people.

So why do we consider non-verbal autistic people 'humans', again? They are non-human. If you cannot speak, you are not a human. End of story.

>>107133099
>cniles be like
>>i need to use a boolean!
no they don't, you fucking liar. you've never heard a cnile say that in your life. you're just making up fake scenarios and getting bent out of shape over it like a delusional tranny blaming children for taking their rights and shooting up a school. stupid fucking tranny liar. you're lucky i'm too drunk to drive or i would come over and beat you with my stupid fists.

Is the only quality tutorial for OpenGL 2.1 is a fucking 2005 book?

Because I can't find good shit on google and I don't want to use chat gpt to read made up shit.

>>107149403
why 2.1? anything made in the last 20 years should handle 3.3 at least. everything after 3.3 is pretty inconsequential in comparison.

>>107149403
Just read the spec.

>>107149456
Is there any practical advantage of 3.3 over 2.5?

>>107149403
>>107149463
If you're looking to do "le hecking old version because le new = bad", you should be using OpenGL ES 2. It has more practical advantages for using on really old shit and was the basis for WebGL 1.

>>107149483
I mean, OpenGL 2.1 was the shit I was trying to learn in 2014.

I guess I'm now old enough to be stuck with PS2 era shit.

next.js personal website coming along nicely saars

>>107149491
OpenGL and OpenGL ES are different things. ES is the "embedded" version that throws out a bunch of unnecessary shit.

>>107149463
3.3 introduces significant improvements to glsl for shaders. it also backports parts of gl4 which you can ignore if you want but if you want to use some of those features later on the option is there. i don't recommend going past 3.3 honestly. 4 doesn't provide anything you actually need and it limits the hardware it can run on which is not worth it imho. as the other anon suggested gles2 is a stripped down opengl that is much more simple and direct. it's worth using if you don't like dealing with all the bloat that comes with opengl, and there is quite a lot of bloat.

>>107149677
>>107149506
I actually wanna learn retro homebrew shit.
I want to move away from le unity-unreal-godot modern stuff.

This Crystal language seems promising. Ruby, but compiled, using LLVM.

I don't know if the language is isomorphic with Ruby or homomorphic. But a basic eyetest says, the syntax is nigh-ident. Semantics, not sure. Dynamic dispatch differs in a language that runs on a VM, and a language that is compiled. Crystal has static typechecking, and that's endearing. I like Ruby's style of "duck typing", but I fucking hate such 'Silicon valley bro' terminology like 'duck typing' (another SVB term I dislike is 'transpiler'). Be scientific, dammit!

>>107149689
IMO don't be too picky if you're just starting out with the graphics stuff. All the gay ass APIs are just obfuscation over some really simple ideas. Get data onto the GPU => draw it with shaders. 99% of everything else is just attempts to optimize that process in various ways.

This playlist has some good stuff if you're into watching lectures:
https://www.youtube.com/playlist?list=PLplnkTzzqsZS3R5DjmCQsqupu43oS9CFN

>>107149689
i agree that general purpose engines suck but you don't have to use older versions of gl to do retro style graphics. everything after version 2.0 uses the same object binding paradigm. basically you put your data into an array, bind the array, and call the draw function.
>https://learnopengl.com/
this site is a good intro to what they call modern opengl which is everything 2 or later. each version of opengl is just a backward compatible extension to version 2 so you can write whatever basic retro style you want and run it with any later version.

Wow MAKE is a real piece of shit. Never using it again.

what can i make as a solo person to have a realistic shot of making money?

>>107150376
ransomware

is putting something like this in a c headerfile retarded
#ifdef USING_NAMESPACE_VEC
#define I32V vec_I32V
#define I32V_init vec_I32V_init
#define I32V_free vec_I32V_free
#define I32V_resize vec_I32V_resize
#define I32V_push vec_I32V_push
#endif

If Sam Altman had asked me, yes ME, about justifying DNNs as a technology. I would have told him, "my brother is Moses, and respectable Jewish person, never, ever say 'We don't know how it works' or 'It's a black box'".

Because, believe it or not, I fell into this trap about 2 years before the explosion of GenAI. I had made a facial recognition server, with liveness detection (it's the 'PuertoRECO' project in my Github) and the guy, some Turk or Arab in Germany, asked me 'so how it works? explain to me how it works!' --- and Godgang it, Arabic is enough of a 'violent-sounding language', add German, another such language, to the mix, and when the guy spoke English, I literally felt threatened.

So I told him "It's a black box, sir" and we got into an argument. I was still paid, but not fully, and I had already finished the project, so I put it up with GPL on my Github. That repo was what started my journey in 'computer science excellence'.

So, when these idiots say 'we don't know how it works', they mostly mean 'we don't know how it OPERATES'. Thank Lord English is such a lexically diverse language!

These idiot Jewtubers have taken this comment buy (I think Sam Altman?) and are running with it, but it's like running with scissors, even an idiot like me who did ML engineering for 2 years, and is only up-to-date with the academic work of 3 years ago (which is difference between Jurassic period, and the dictatorship of Nero, given how quickly the academia is stacking up!) could tell you, that's NOT what they mean, pretty lady making anti-AI slop!

>>107148266
other than when I look for some specific tutorial or whatever, there are only 2 youtube/odysee channels I watch regularly:
https://youtube.com/@cjbbooks8622/videos

https://odysee.com/@KnowMoreNews:1

>>107150633
Using C is by default retarded so you don't have to worry about it anymore.

>>107150136
Yeah it's yet another piece of hot garbage tooling that Cniles have Stockholmed themselves into thinking is usable.

why is recursion so important? why do they focus so much on it in CS

>>107153042
1) because it is the foundation of lambda calculus and therefore entirety of functional programming
2) because it is often easier to reason about in mathematical sense
3) because there are functions that can't be computed without recursion (non-primitive recursive functions)

>>107150136
>>107152765
Why? Make just works. When I use more opinionated build systems I often miss the simplicity of make.

>>107153042
Try and implement a tree traversal with recursion and then without recursion

>>107153042
who is they?

>>107153136
NTA, but doing it without recursion and with a manual stack is much better since you won't overflow the call stack on a very deep tree.

>>107153250
At the expense of causing one or more allocations.

>>107153261
If that's a concern, you can start with a stack buffer and move to the heap if you overflow.

>>107153133
Why use Make at all if you need "simplicity"?
Your fizzbuzz is very simple to compile, just gcc main.c, fizzbuzz is fast enough so if you forget to pass 9 gorillion compiler flags into it, that's fine, it will still run faster than in Python.

>>107153269
>Why use Make at all if you need "simplicity"?
Because make is simple.

>Your fizzbuzz is very simple to compile, just gcc main.c , fizzbuzz is fast enough so if you forget to pass 9 gorillion compiler flags into it, that's fine, it will still run faster than in Python.
I have project that is currently composed of 8 Rust crates that target different architectures, couple of Rust and python scripts and bunch of different assets that need transforming before compilation. Although Cargo workspace support is pretty handy in general, it handles poorly multi-target projects and do not work well as general command runner and asset processor. I am currently using cargo-make to somehow get this all working together, but I hate how opinionated and verbose it is. I would use make but I need it to reliably work on windows as well.
What I would absolutely love is something like Ninja, but with a few changes to make it a little bit more friendly for hand writing and to make it work as command runner, not just build system. I really want a simple, unopinionated but powerful build system, I might try to develop one at some point.

>>107153250
yeah, but one's easier than the other. And for the other anon, stacks are also very focused on because of its usefulness

>>107153366
I accept your concession.

>>107153042
Some shit is easier to program as the result of a function that calls itself.

>But a for loop
Some problems like A* is not clear if you need to allocate 10 or 50 or 500 or 5k iterations.

>>107153042
Some really important data structures like Trees and Lists are defined recursively and recursive algorithms are the cleanest way to operate on recursive data structures.

>>107153632
Trees and Lists are so important that we do anything to avoid using them.

>>107133087 (OP)
What's a good cross-platform TUI library for c++?
I just want to make a small terminal game to practice but all the good terminal libraries are for unix systems only.

>>107154002
pretty sure javascript, python and java shit are cross platform.

>>107154038
But I want to practice c++
That was part of the request.

>>107154048
if you want to practice, then It doesn't matter if it's linux only stuff.

As for something to make a binary to share with others, maybe use Qt or wxwidgets.

>>107154067
Can you please stop changing the request?
Either mention one or admit you don't know any.
I want to share what I make with my windows only friends and they can't be bothered to use a VM or to setup WSL.

>>107154103
you're trying to reinvent a problem that is already solved.

ask chat gpt then.

>"Hey does anyone know any good c++ library that..."
>"Uhh actually sweetie, you want to program in python."
Why are retards like this?

>>107154115
You are an unhelpful retard, did you know that? You are the reason this board is so shit. Nobody is reinventing anything, a person has a problem and you don't know how to solve it, so instead of saying "idk" you say "well change your problem", whioch is the most lazy and stupid way of solving shit.
Unironically, you should kill yourself.

>>107154144
>retarded newfag uses C++ for le epic bragging rights
You should just use python or java or C# or javascript, kid.

>>107154161
Typical, you are just seething at someone who dares to learn something. Bragging about what?
Do you consider c++ hard? That's ridiculous. Maybe for your brain it is. No surprise you are unable to give any good advise.

>>107154161
Shut the fuck up, >>107154144 is right.
>>107154002
Windows Terminal supports ANSI escapes now, just like Linux terminals. I'd recommend looking up a table of ANSI escapes like this and doing it yourself: https://en.wikipedia.org/wiki/ANSI_escape_code
There may be a library out there somewhere, but if your game is simple enough, the raw escapes should be fine.

>>107154180
C++ is garbage.

Use C is you want low level shit, or a more modern language if you want to code.

C++ is just a language that only has a reputation of being le hardcore shit for true developers.

When linux literally said that C++ is for retarded high level programmers, unsuited for kernel code.

>>107154196
You know what?
Yeah. I might just put all the characters manually and then clean the screen with "\033[2J\033[H"

>>107154002
0 wintoddlers will run your toy and those who will likely are intelligent enough to at least use WSL, cross-platform tui has 0 usecases.

>>107154197
And Rust is a low level language perfectly suited for Linux, I'm also trans btw.

>>107147126
emulators to play my retro games

>>107154674
uhh my friends say they will play it, and they are more than 0

I didn't knew matches! can take an if-like syntax inside. Pretty neat.
let start = logs[0..idx].iter()
.rposition(|log| !matches!(log, Log::SCR(n, _) if n == name))
.map_or(0, |idx| idx + 1);

I started programming last year, and I currently consider myself to be in the range of a 10x to 25x engineer. Full stack. My friend who started around the same time is probably a 5x to 8x and he says I am easily 25x+, but I am considerably more skilled than him so I'm not sure if he can adequately assess my level. I would eventually like to be a 100x but let's say the next milestone is 50x, how do I best go about getting there? I am using Claude Max but I would consider other models if it helps me grow as an engineer. Any advice is appreciated, but especially that of 100x or higher engineers.

>>107155186
sorry anon, this isn't the GPT prompt, you need to go on the GPT website to get responses.

>>107155220
Oh believe me, I will be asking an AI the same question, but I like getting input from multiple perspectives. It's what sets me apart and has let me become at the very least 10x in such a short time.

>>107133087 (OP)
>C++ exception thrown+caught
>5000-1000
C++ issue, in C longjmp should take < 100 cycles

I'm using my last hours of Claude to refactor/improve most of my codebase

I'm going back to writing code by myself

Despite finally getting the hang of meta-prompting and context engineering, I feel dumber, even when I have actual knowledge of the field

I hope that my project without using AI and the advent of code will help me getting back my edge

Goodbye, world!