File: fully trained discerning (...).png (764 KB, 1600x720)

764 KB PNG

Check this out! I took a fully trained network for 0-9 OCR, but now this is each neuron visualized as an activation matrix of the original images, you can see how it 'sees' the images as edges and a combination of them.
The structure is 784 inputs (not visualized) by 256x128x64 with 10 outputs.

>>101280210
I am making a streamer for music although
things are going tough these days

the biggest struggle isnt learning how to write code, its figuring out where the fuck things should go

>>101280210
currently tidying up source sdk code from an old mod that i had and moving it to a proper build system, mostly for fun, and because it'd be cool crosscompiling something for windows from linux
what valve intended by still supporting that ancient-ass shit of software that is VPC is beyond me

setting up git server without bloat

ERLANG

Creating a Huffman encoder and decoder is actually quite complex. I was surprised because the algorithm is so simple.

>>101281469
>noooticing patterns is complex
take vitamin D supplement and go outside more

File: lnx_computer.jpg (194 KB, 651x720)

194 KB JPG

I need advice on software. Complete beginner here, largest garbage I wrote was 2000LOC of spaghetti that "worked". Doing this for fun, not work.
Usecase : I often used libreoffice calc to plot graphs. Values could be of different types and all over the place so I wanted to automate the horrible experience of making graphs, using C and a plotting library on linux.
I've been having fun with the gnuplot_i library but I quickly reached its limitations. Specifically, I wanted to change the type, size and colors of the curves, but the functions are either too simplistic or don't work the way I want to.
Furthermore, it is an absolute pain because there is 0 support for it (I know I asked for it, should have probably gone the python route). Not only that but most of the links for the gnuplot documentation are dead, and they have a horrible way of writing doc where every little detail is hidden in another webpage. The man for gnuplot also might as well not exist. Gnuplot is horrible to use, I'd like a real plotting library, not a way to send commands to gnuplot like I've been doing.
Do you know a good C plotting library? I'm basically looking for:
- export png graphs, change the resolution, etc
- change the size and type of dots and lines
PLplot sounds good, what do you think? I also looked at C++ libs and omg they look so much easier and better. Should I just switch to C++? I'm a bit hesitant because I know nothing of it.

File: 1622528198216.jpg (46 KB, 235x243)

46 KB JPG

>Using new printf formats that C23 adds for the uintN_t types, like "%w64u" instead of "%"PRIu64
>ASan complains

>>101281504
>using C
>using printf
>complaining about daily humiliation ritual

>>101281503
>Do you know a good C plotting library?
i don't, but if you can find something that does what you want in python, you can pull if apart and take what you need. it is generally "cpython". i think c++ is a lot of bloat and obfuscation. tendency to get wrapped around its axles and distract from the goal.

>>101281503
just use python
it is horribly easy, that's why it's so popular
if you have coded in any other language, coding in python is a joke

>>101281515
The HRT certainly has made you crabby.

>>101281964
You can just memcpy into your own buffer then use write syscall by yourself without using bloated functions that aren't even convenient to use.

>>101282072
So how do you convert an integer into a decimal string?
>inb4 itoa
Thought so.

>>101282493
custom solution optimized for the problem

>>101282540
So do you have actual code to show for, or do you regularly spout utter nonsense?

>>101282556
He's the resident "le ebin h4x0r, i'm just like terry davis please notice please, my life is so fucking sad" guy here.

>>101282673
>no code
Your attempts at deflection are humorous in their ineffectiveness.

>>101280221
I like your work, anon.
I'd be very interested in your backpropagation algorithm.
>to steal it
'course!

File: code.png (32 KB, 1065x1353)

32 KB PNG

>>101282700
This is the code I'm working on.
I'm just doing some testing with io_uring. It's not intended to be high quality code.

>>101282758
Cute, but it has nothing to do with itoa.

>>101282783
I know. Why the fuck would I waste my time reimplementing that? I don't care what some random Terry wannabe thinks.

>>101280210
working on nothing of your business, faggot

>>101282797
So you're an ape that just likes to throw shit at random stuff? Got it?

>>101282816
You aren't fitting in

>>101282556
few months ago I made a thread about this, I posted the code that's generic yet still faster than 100% of your toy standard libraries, if you weren't there, you can only blame yourself.

>>101283171
If you mean the C++ garbage that had nothing to do with anything: keep it to yourself. No one needs code that issues one-byte memory writes.

>>101283208
I mean the code that beats anything in benchmarks, since then I improved it, and no, this time you won't see it, I had a problem when I made this thread, nobody helped me solve it, now I won't share either.

>>101283222
>I mean the code that beats anything in benchmarks
Also when the number goes above 0x2386F26FC0FFFF? Because that's when 64-bit-based PBCD will fail you.

>>101283256
then it becomes even faster, but I digress, I'm not interesting in nocoder niggerbabble seethe

File: 1693554491117055.png (134 KB, 518x408)

134 KB PNG

>>101283285
>then it becomes even faster
Ah, got it, you're just making shit up as you go.

@101283323
yes, I made up things like
>I totally need numbers bigger than 64bit

File: 1695299960265025.png (566 KB, 700x571)

566 KB PNG

>he doesn't understand the difference between 64-bit numbers and 64-bit PBCD-encoded numbers
Worse. You make shit up about things you don't even pretend to understand.

>hurr you don't know about my irrelevant thing
no usecase, WONTFIX, email put into spam filter

File: 1686511064733964.jpg (209 KB, 1024x1024)

209 KB JPG

>your irrelevant thing I had to watch out for, since my code was supposed to be oh-so-much-faster than other implementations out there
And that's why I'm so sure you're just making shit up.

>code was posted
>he didn't see it
case closed, summerfaggotry is in full swing, PBCD wasn't relevant way before you were even born, retarded zoomer

File: 1694888355721296.jpg (85 KB, 448x635)

85 KB JPG

>he didn't see it
No need to. No amount of seething and hallucinating will change the way hardware works.

>>101283437
Stop responding to its posts, it's early and I should be enjoying my peace and quiet god damn it!

>>101283481
you can always use ctrl+W if you don't want to see what a retarded tranny nocodeshitter humiliation ritual looks like

File: 1709420053120675.png (142 KB, 396x364)

142 KB PNG

>>101283481
>if you just ignore their posts they will go away
So how does losing for over thirty consecutive years feel like?

>>101280592
just what i am going to complain about.
>>101280210
I fkkin hate figuring out where things should go in big projects, i have like 5 "kinds" of memory pool and one of them is "each of these 4 guys are big arrays that you are going to have to put somewhere..."
like FUCK

>>101283523
wow gee it's so hard to create allocator package/module/file and put them there

>>101283537
it is about the mental overhead and my tendency to procrastinate in response to it.

Glad to have the schizos back.

>>101283521
srsbsns

File: 1661180967317096.jpg (8 KB, 207x253)

8 KB JPG

>>101283587
Schizo(s)?

>>101280592
>C Interfaces and Implementations: Techniques for Creating Reusable Software - David Hanson
https://github.com/drh/cii
https://drh.github.io/cii/

>>101283691
this function alone https://github.com/drh/cii/blob/master/src/arena.c#L52-L84 shows that whoever made this is a nocoder

Are those nocoders in the room with us now, or do they only exist in your mirror?

oh no, the nocodetranny took it personally, the thread is finished

File: 1691396804696495.jpg (94 KB, 1089x1080)

94 KB JPG

A preferable outcome, really.

File: file.png (27 KB, 735x222)

27 KB PNG

can this function be improved?

>>101283959
yes, don't allocate a new string

>>101283959
Use AVX.

>>101284022
where? in the format?

>>101283959
yes, dont use silly libraries.

>>101283706
>36 channers watching the repo
holy shit

>>101284075
in the function, formatting a hex string like that won't allocate so it will be fast enough

>>101284130
There is never "fast enough", there's only "too dumb".

>>101284181
not interested in nocodeshitter cope

>>101284193
Yet you're here.

>>101284115
there are no libraries, everything is standard c++
>>101284130
well, i am allocating a 2 char string on every iteration

>>101284217
std::string uses SSO

>>101284238
guess it is as fast as it can be then
>>101284070
this needs to run on arm as well, I'll just let the compiler optimize it

>>101283959
yea, don't convert to a string and instead just pass around the byte(s) until the actual point of use, then convert to a string (with whatever your print / writer function uses for number formatting)

>>101283959

char dec_to_hex(char c) {
    if(c >= 0 && c <= 9) {
        return c + '0';
    } else {
        return c - 10 + 'A';
    }
}

auto bytes_to_hexstring(std::span<std::byte> bytes) -> std::string {
    std::vector<char> v(2 * bytes.size() + 1);
    size_t i = 0;
    for(const auto byte: bytes) {
        char c = std::to_integer<char>(byte);
        std::cout << c << '\n';
        v[i] = dec_to_hex(c >> 4);
        v[i + 1] = dec_to_hex(c & 0xf);
        i += 2;
    }
    v[i] = 0;
    auto s = std::string(v.data());
    return s;
}

Something like this, I guess, I mean if you really want to make this fast, you want to avx this thing, there are some hex libraries on the web, like https://github.com/zbjornson/fast-hex

What happens when a C compiler sees that a variable is set to a value it already was set to? Does it just ignore that line?

>>101285764
Whatever it wants that still fits the semantics specified by the C abstract machine
Why would there be any issue? Do you think checking and conditionally assigning would be faster and not slower?

I am making a basic webpage/website scraper for fun. How can I avoid the website banning my IP or other methods?

>>101285823
You can't, buy more IPs (or quit being a nuisance for fun)

>>101285795
>Do you think checking and conditionally assigning would be faster and not slower?
I mean, maybe this is my noobiness showing, but I would assume that any redundant instructions that can be removed at compile time would make for a faster execution, no? The check isn't happening at runtime, it's happening during compilation

>>101285842
read the binary and report back already.

>>101285841
how is that being a nuisance? i just want to collect the information on your website that is there to be collected by the public in the first place.
What is wrong with you?

>>101285842
only if it's known at compile time and it usually isn't

>>101280210
boring thread, post more

>>101285913
You aren't smart enough to understand this, but the internet doesn't run on magic and every part of its operation costs money

>>101286367
I am smart enough to understand that that is completely irrelevant, me ordering 20gigs of documents from your shitty server is not a significant cost in 2024.

File: 1.jpg (25 KB, 423x303)

25 KB JPG

Explain this sirs:
>learning about time complexity
>code a linked list in python
>insert should be O(1)
>read should be O(n)
>write a python script to compare a 10k vs 1mi list
>start 10 minutes ago
>still running

What the fuck???? A modern pc cant handle 1 million items?

>>101286530
You are probably doing it in O(n^2)
If you do
>head.append(item)
In a linked list 1M times you do 1M×1M/2 operations.

>>101286654

class list:
    insert(self, insNode):
         node = self.head
               while node.next:
                     node = node.next
                node.next = insNode


for i in range (1,1000000):
   list.insert(i)

File: 1720037122936760.gif (1.35 MB, 346x261)

1.35 MB GIF

>>101286715
Yup, O(n^2). Try updating the head instead of going to the end of the list.

>>101284426
i did a test and this version computes the hex string for shakespare in 4-6ms and the format one in 220-250ms

>>101286715

    def insert(self, value):
        new_node = Node(value)
        new_node.next = self.head
        self.head = new_node

Just do this

>>101287008
thanks, this works

File: doubt.jpg (18 KB, 600x341)

18 KB JPG

>>101286494
>I am smart enough

File: 1719077446373560.jpg (69 KB, 640x644)

69 KB JPG

Trying to implement operator new in a c++ class, where everything needs to be dynamic in a nostdlib environment. Whole point is to have constructors called. Can't call constructors explicitly after allocation, which is throwing off vtable/virtual functions.
It's ok. Little by little I'm getting closer. Just need to figure out a way around having to set function ptr variables (heap allocation functions) needed in the operator new implementation as static.

File: img_6_1702480605853.jpg (124 KB, 1439x1495)

124 KB JPG

>>101287475
A BRIDGE FUNCTION
FUCK
NEW CALLS A FUNCTION IN CLASS TO DO THE THIIING IT'S SO SIMPLE FUUUUCK

>>101287475
You can call constructors explicitly after allocation
new(p) Type(...)
new(p) Type{...}

>>101287475
why would you ever want to do that instead of just doing alloc()???
>a bridge function
yay let us go so much slower for no reason!...
anon, have a regex macro.

>>101286530
>10k vs 1mi
You should have increased the length by a factor of 2, not a factor of 100

File: file.png (72 KB, 862x581)

72 KB PNG

>>101286888
thanks anon, it is much better and generates less assembly than the format one
posting improved version if anyone wants it

>>101288184
might as well use std::string directly, improved v2

auto util::byte_to_hex(const std::byte byte) -> char
{
    constexpr int to_char = 48;
    constexpr int to_char2 = 87;
    constexpr int limit = 9;
    const auto chr = std::to_integer<char>(byte);
    if (chr >= 0 && chr <= limit) {
        return static_cast<char>(chr + to_char);
    }
    return static_cast<char>(chr + to_char2);
}

auto util::bytes_to_hexstring(const std::span<const std::byte> bytes) noexcept -> std::string
{
    std::string result(bytes.size() * 2, 0);

    constexpr std::byte mask{0xf};
    std::size_t idx = 0;
    for (const auto byte : bytes) {
        result.at(idx) = byte_to_hex(byte >> 4);
        result.at(idx + 1) = byte_to_hex(byte & mask);
        idx += 2;
    }
    return result;
}

File: partially trained neural (...).png (1 MB, 1600x720)

1 MB PNG

>>101282756
Sure, here's a repo of the network and code to create that image, the js file is a little cluttered but if you go to the 'Network' class, and see the 'calculateDeltasSigmoid' function within, it does the training along with 'adjustWeights'.
The general procedure looks like:

let inputs = [new Data(0),new Data(0)] //input example to structure from
let structure = [3, 2] //example structure, one hidden layer of 3, and an output size of 2
let network = new Network(inputs, structure)

//then to train
let trainingData = [new Data(.7),new Data(.3)]
network.compute(trainingData)
let goals = [new Data(.3), new Data(.7)
network.calculateDeltasSigmoid(goals)
network.adjustWeights()

I hope I got the code tag right

https://github.com/Packmanager9/ActivationAtlas_OCR

Does anyone use solidity anymore?

File: 17043648937187781.jpg (331 KB, 1536x2048)

331 KB JPG

My python script is running substantially faster now.

I am almost ready to start calculating fibonacci numbers bigger than 4^2,000,000

>mfw used 10 mb to implement atoi like

char* my_atoi(int n) {
    if (n == -2147483648) return "-2147483648";
    int s = n < 0;
    n = n > 0 ? n : -n;
    switch (n) {
        case 0: return "0";
        case 1: return s ? "-1" : "1";
        case 2: return s ? "-2" : "2";
        case 3: return s ? "-3" : "3";
        case 4: return s ? "-4" : "4";
        case 5: return s ? "-5" : "5";
        case 6: return s ? "-6" : "6";
        case 7: return s ? "-7" : "7";
        case 8: return s ? "-8" : "8";
        case 9: return s ? "-9" : "9";
        case 10: return s ? "-10" : "10";
        // ...
        case 32768: return s ? "-32768" : "32768";
        default: return "NaN";
    }
}

>>101289393
Shouldn't that be itoa?

>>101289498
That one, yes.

>>101289223
I have a solditity book I ordered from Amazon.

>>101289393
>goes slower than the following anyway because of cache locality

void itoa(uint64_t num, char * string){
        uint32_t off = 0;
        uint64_t factor = 1;
        while(factor*10 < num){ factor*=10; ++off;}
        for (1; off; --off){
                string[off] = num/factor;
                num %= factor;
                factor /= 10;
        }
}
 
uint64_t full_atoi(char * string){
        uint64_t ret = 0;    
        uint32_t u = 0;
        for (uint64_t i = 0; i < MAX64BIT/10*10; i*=10){
                ret += string[u] * i;
                ++u;
        }
        return ret;
}

>>101289498
Does that stand for integer to ASCII?

muscle memory made i 0

>>101289723
your formatting is atrocious.
full atoi doesn't work.

>>101282758
This looks advanced. What does it do?

File: 1720162864371581.png (935 KB, 1041x858)

935 KB PNG

>>101280210
Password manager in C++. Might post progress later if anyone's interested.

Anyone here code in Assembly?

>>101280210
what do the edges mean?

>>101283421
lost

>>101289781
sorry man, going from implementing actually complicated shit to highschool homework has me tripping

        void hard_faulthandler(void){
                void * faultee = __asm__ volatile("mov %%cr2,%%rax\n\t");       //DANGER syntax
                memstream_init(void);

                Kingmem * mm = get_kingmem_object();

                ustd_t offset = 0;
                ustd_t rotation = (mm->paging-1)*9+13;}
                for (ustd_t i = 1; i < mm->paging; ++i){
                        ustd_t rel_off = rotateRight(faultee,rotation) << 55 >> 55;
                        for (ustd_t i = rel_off +1; i; --i){
                                for (1; mm->vm_ram_table[offset] & 1; ++offset);
                                ++offset;
                        }
                        rotation -= 9;
                        if (mm->vm_ram_table[offset] & 64){ break;}     //ps bit in pagedirs
                }

                if !(mm->vm_ram_table[offset] & 512){
                        Thread * thread = get_thread_ptr();
                        Process * calling_process = thread->parent;
                        calling_process->sigset |= SIGSEGV;
                }

                ustd_t request_mode = mm->vm_ram_table[offset] << 12 >> 63;
                ustd_t pagetype = mm->vm_ram_table[offset] << 13 >> 62;
                ustd_t virt_disk = mm->vm_ram_table[offset] << 6 >> 58;
                if (request_mode == RESERVE){
                        void * phys;
////

>>101289856
>for (ustd_t i = rel_off +1; i; --i){
> for (1; mm->vm_ram_table[offset] & 1; ++offset);
> ++offset;
> }
good thing i post shit here sometimes because i have no idea what drugs i was on when i wrote this

File: art_graph.png (2.92 MB, 4112x3984)

2.92 MB PNG

>>101289822
They are linked based on visual similarity, and the force applied by the link is along a gradient from high -> green, to red -> low. It is a force graph, and naturally fell into that shape. It isn't a great representation, but the pokemon are grouped roughly by body hue and shape of the pixel data. I just wanted to start a new thread and I don't have any template images, but I've seen anime used before and graphs, so it seemed like fair game. Here's one where the link represents the shared hue average.

File: untitled-[00.04.167-00.18.916].webm (620 KB, 826x720)

620 KB WEBM

Either the last or second to last update I'll make on my doujin gui tool. I've now added a downloader as well as fixed a few ui things that bugged me. If I make another post it'll be to post the gitlab repo in case anyone wants to try it, but I need to go through all my files and clean up the absolute mess I made of the code, otherwise I'll die of shame if I post it publicly.

>>101280592
Where is my code editor where you can cross reference the same code in different files to keep a logical categorization of your code.

working on my own database system, its quite simple. The goal is log(n) retrieval times. I'll put it on my home server and keep my gf's recipes on it to be accessed with ssh and displayed on a simple flutter app.

File: file.png (1 MB, 1200x1200)

1 MB PNG

>>101290381
>tfw bf dumps the dinner table

File: 1695906123180405.png (424 KB, 969x966)

424 KB PNG

>>101289723
>single-byte writes because he can't into 64-bit PBCDs

>>101289810
>inb4 that's not REAL assembly

>>101289845
Reality must be a scary place for you.

File: code.png (16 KB, 520x1172)

16 KB PNG

Is there a more efficient way to do this?

File: 1699041712708363.png (1.4 MB, 4296x4340)

1.4 MB PNG

With python and powershell scripts I like to print out to the terminal a lot with the progress such as "downloading file in progress", "successfully downloaded file", "adding this and that" "removing this".
is it bad practice and just bloats up the script?

File: 541-5415825_pokemon-hypno(...).png (68 KB, 860x755)

68 KB PNG

>>101280210
?

>>101290647
hypno is next to jolteon in the yellow section

>>101290495
if you dont write it in avx you deserve to be hung from a lampstand while the crowd cheers and yells for more iron to trafict your limbs.

/s

>reflection
what a waste of time

>>101290732
This, but unironically.

>>101290537
>is having sovl bad practice?
it doesnt matter.

>>101289719
>I have a solditity book I ordered from Amazon.
Have you used solidity for a gig or a job or no?

>>101289223
I really don't get Blockchain and what it could be useful for

Are there any *good* tutorials for writing nginx modules? Most articles out there are utterly non-descriptive (what the fuck is the difference between the pool and the temp_pool? what the fuck does "merging" mean?)

>what are you working on
my job :DDDDDDDDD
just wrote some groovy lately, had a fucking blast, can say for certain that that scripting lang is very frenly, I guess it might be js's cousin, and it shows, if you wanna hit me up with some nice things to do in groovy please do, I'd love to learn more

>>101292760
>:DDDDDDDDD
Thanks for making it obvious from the get-go that you're not worth listening to.

>>101292824
fuck off, queer

>>101292824
https://ih1.redbubble.net/image.803510534.2789/fposter,small,wall_texture,product,750x1000.u1.jpg
plus here it is, lore, you fucking colossal faggot

Too late. :DDDDDDDDD

>>101283163
never was, never will

Why did github get rid of the ctrl+k palette?

File: carbon.png (129 KB, 1396x860)

129 KB PNG

>>101280210
going back to work on my C++ binding library for Duktape. today I will try to add support for [smart] pointers.

File: 1578860177062.png (83 KB, 860x791)

83 KB PNG

>crash if you call toString on a string

File: 1668082012599364.jpg (18 KB, 384x404)

18 KB JPG

>>101289787
io_uring is a fairly new asynchronous network/file API for linux.
You queue up a bunch of "submission queue entries" (SQE) and at some point the kernel will give you some "completion queue entries" (CQE).
What I did was queue up 4 SQEs:
- sock = socket(AF_UNIX, SOCK_STREAM, 0)
- connect(sock, ...)
- sendmsg(sock, ...)
- recv(sock, ...)
and I linked them together (IOSQE_IO_LINK) so they properly run in order, otherwise it might try to connect() a socket that doesn't exist yet.
In certain situations, using this can lead to performance improvements, but it's also giga unportable.

I'm toying around with writing a dbus client library, and while I think io_uring is massively overkill for this, I want it to be possible for a user to use io_uring "properly" if they want to, and basically eliminate ALL blocking.
I'm thinking how I want to write the API where you (optionally) farm the socket operations out, and not have it be a massive pain in the ass to use.

dot  {+/(+×)}

I'm learning.

>>101293662
Oh, fuck off moot.

I really want to make more gui app in C but It seems no widget library Is good enough for C

>>101293662
stay and explain the oldfag joke

>>101280210
>weekly programming thread

Do I add type checking AFTER I get evaluation working, or before, when writing a compiler? Types are part of the design already but I'm just executing the code regardless of the variable type. If the type is wrong, the program fails. I'm thinking I can add type checks at the end but maybe this will be a nightmare to later?

File: Screenshot 2024-07-06 101530.png (4 KB, 323x88)

4 KB PNG

>>101294009
Oldfag joke?

>>101294162
depends

>>101294752
Wanna elaborate? I've already done all the basics. Control flow, functions, scope, etc. but just evaluating the program without even looking at the types. Doing static type checking seems like something I can postpone until I've decided on the rest of the syntax/features, but maybe I'm just creating more work for later.

>>101294810
in haskell for example you have type classes where types effect evaluation by picking out a particular implementation for a matching list of type patterns

>compiler prohibits division by zero
how fucking dare you
that's a runtime error

>>101295080
should have just defined x/0 = +inf or x/0 = 0 and done away with this error nonsense

>>101295080
worse it's not even a runtime error it's well defined
https://en.wikipedia.org/wiki/Division_by_zero#Floating-point_arithmetic

>>101295080
good because its OS dependant, write better code.

>>101295162
you are retarded

>>101295462
go on

>>101295488
i am not going to help you stop being a cretin, everybody in the know knows already.

>>101295537
>>101283706
hello

>>101295162
>+inf
>>101295211
>floating-point
ngmi

>>101295453
you there's a standard for division by zero?

>>101295211
Assuming you meant to link to "Computer arithmetic" section, in theory it's well defined what happens for a given architecture, but the fact that the runtime behavior differs between systems is exactly what makes it undefined behavior in the sense of the C standard.

>>101283706
that "drh" guy literally made lcc, a reference c compiler, in the 90s and published an entire book about it
He's also a (former) microsoft programmer though so I can understand where you're getting at

>someone who looks for a Python/Qweb/Postgre programmer is calling me on Monday
>have only made a few scripts with Python and used Robot Framework once
My typical stack has been MERN and its variations for years. Can I convince him that I can do what he wants? I'm going through tutorials lol

>>101295582
Ieee754

>>101295582
if it's floating point yeah

File: GRzN05-XkAA8dun.jpg (166 KB, 1460x880)

166 KB JPG

Be honest, anons, is this enough in the year of our lord 2024? I've been working on projects to demo my skills because I'm a sperg and didn't netwooooork or make connections. Going through college as a background character.

Overall, what do you think? Appreciate any advice from anons who did this before me and got jobs in this economy.

>>101297125
On one hand jobs that you get via networking are usually dead-end, because who in their right mind would let you work on anything critical if your only qualification is that you once blew the right guy?
On the other hand I couldn't give a flying fuck *what* projects you did, but rather *how* you did them. Do you even know what TCP_CORK is? No? Well, you did a TCP and HTTP server, so one would expect you to know.

>>101297125
I would never hire anyone who wasted time on Brainfuck interpreter. this may impress /g/ autists, but it's cringe basically anywhere else. delete it, and never mention it again to anyone.

>>101297452
>cringe
Luckily that's a complete (you) problem. You feel the cringe, and I don't.

>>101297125
unfortunately that's a lot of nothing
reinventing the wheel only makes sense if you at least cover some rare use-case or requirement that the standard implementations don't. also outright stating they're toy versions isn't a good way to "sell" your stuff.

try writing something that is usable just on its own, like a smart diff tool for JSON or XML that compares nodes/elements rather than just text line-by-line

>>101297493
it's not about you. it's about the person you show it to. for the love of god, don't mention it in any interviews.

>>101296748
>> Python

How hard can it be anon. Just Lie. Get the bag and worry about the code later.

>>101297517
>it's not about you
Everything is about me, especially in times of skilled labor shortages.

>>101297530
>times of skilled labor shortages
lmao, ever been in /twg/ in the last 2 years?

I made a function 4 micro seconds faster. This is HUGE news.

>>101297547
I said "skilled" labor, 0%-reading-comprehension-kun.

>>101297125
Should be ok for a junior position especially since you mentioned you also have a degree. Anons are way too pessimistic.

>>101297573
>posts this list >>101297125
>skilled
delusional. you are checking off some baby's first steps tutorial projects. no one cares about any of this. this won't help you land a job.

File: outfuckingskilled.png (9 KB, 424x124)

9 KB PNG

>>101297614

>>101288682
Thanks, anon, looking into it.
Glad to see it work after finding so many bad/broken algorithms on the interwebs

any groovy bros here? just started using it and already lovingit, I just google things like "groovy find element in array" and it already exist, total bro of a language. Just an example:

def "givenListOfPerson_whenUsingCollectionMatching_thenShouldEvaluateList"() {
    expect:
    personList.any { it.age > 20 }
    !personList.every { it.age < 30 }
}

If you have any tips or cool groovy stuff for me pls share.

File: firefox_Pqy4NuJeNe.png (26 KB, 845x278)

26 KB PNG

>"I write in a subset of rust where I use unsafe code everywhere. If you don't like it, leave"
Is this fella based or cringe?
As a non-fan of rust I think it's kinda based if only to piss off rust cultists

>>101298161
It might seem based but it is ultimately cringe, because you are still using rust.

File: Untitled.png (44 KB, 848x608)

44 KB PNG

>>101283959
100x speed code for you

is 20 commits a day enough?

Any better c++ WS client lib than cpprestsdk?

File: c98.png (1.54 MB, 900x1200)

1.54 MB PNG

>>101298767
>more than 1 commit a day

>>101298767
You need to edit the readme at least once every couple of hours. Documentation is important.

File: lmao.png (114 KB, 2206x612)

114 KB PNG

>my model is so shitty the R2 is negative
kek

File: based.png (19 KB, 454x98)

19 KB PNG

>>101299798
Never mind, fixed it
Anyone know how I can interpret coefficients to figure out which features are the most important for the prediction?

>>101280210
making a manga reader in qt

but i guess it will really just be an image viewer with some settings for images to have different gradients and maybe some anime-themed icon/loading screen

>>101280210
>What are you working on /g/?
Nothing, unfortunately. I don't have any projects that I am enthusiastic about.

>>101291143
I was going to use it to do flashloans with crypto DEX arbitrage. I got bored of reading it though, like most programming books.

Are program functions suppose to mimic mathematical functions?

>>101300281
stop reading books and tinker instead.
The best programmer alive at any moment is the tinkertranny who decides to put down his ego and use the interfaces / libraries / whatever

>>101300293
yeah

Class math{
uint64_t integer_addition(uint64_t intone,uint64_t inttwo){
return intone + inttwo;
}
};
//in some other function
Math nigger;
nigger.value = nigger.integer_addition(something,something_else);

>>101298767
I make ~3 commits a month an they all happen within the span of 2 days

>>101280210
Is there any decent native alternative to arduino IDE? It works fine but I hate the fact that its electron soiware and literally takes one minute to start on my alder lake PC with 32gb of ram. I am mostly interested in atmega microcontrollers.

>>101280210
is that a merchant?

>>101300648
vscode with platformio, you get the arduino SDK with bit nicer experience.
Microchip own atmel now and their tool is the mplabx which sucks.
There is atmel studio which is decent.
If you use the older asf3 sdk, you can use pretty much any C IDE.

>>101300648
An Anon talked about the debian package being shit and the appimage being good, i have the appimage and shit is still slow as fuck, fails to launch most of the time even, though i am on wayland and the icon on the top left tells me it runs on an emulated X session so that is probablt the culprit.
Specs are jasper lake sse2, 8gig ram

>>101300751
>windows-trooning
no thank you.
But also
>having to compile all libraries by yourself
might as well install gentoo.

>>101299945
>Anyone know how I can interpret coefficients to figure out which features are the most important for the prediction?
ML models are notorious for being hard to persuade to spit out the "reason". The model doesn't "know" because it's just doing something like curve fitting (in a high-dimensional abstract space so yay), and that doesn't have "reasons".
You could claim it has "hunches" I guess, but it isn't really even that. These things truly do not think.

>>101300293
not always

is there a way to use C macros to turn

if (__builtin_expect_with_probability(x>0, true, 0.9)) {...}

to

if PROB(0.9) (x>0) {...}

>>101301100
no because of the order of parameters.
You could write a shorthand macro prob(probability, test)

File: shader triangle.gif (2.11 MB, 295x296)

2.11 MB GIF

>>101300664
wasn't supposed to be, but I think I see what you are seeing, now that you mention it.

>>101301166
would it be possible if i changed the order with an inline wrapper function?

File: mmm pre-standard C++.png (90 KB, 1718x928)

90 KB PNG

Anyone else programming with MSVC6?

>>101298565
>one-byte reads AND writes
You are all so bad at this holy shit.

A question to people with extensive Linux/Unix programming experience. I'm mostly a Windows guy. Why is the fork-exec model so beloved and hasn't been replaced by anything else?

The need to implement Copy-on-Write feels like a tacit admission that this model is wrong and bad for performance.

What's some good tips and tricks you can do in C++98 and earlier (aka pre-standard C++)?
inb4 >use C++11
I've restricted myself to using VC6 as a little challenge/exercise in masochism.

>>101301573
I dunno man, I'm a gamedev, I let SDL take the wheel for win32 and unix shit
t. Win10/Arch dualbooter

>>101301574
Are you using COM and ATL

>>101301610
>Are you using COM and ATL
Nope, I plan to use the Win32 API directly with DirectX7 for gamedev (or DX9, whichever's better).

>>101301573
What makes you think that's an exclusive *n*x thing?
https://learn.microsoft.com/en-us/windows/win32/api/winbase/nf-winbase-createfilemappinga#PAGE_WRITECOPY

>>101300799
Your problem then.

>>101301634
Lord be with you

I feel like you would have a better experience with Delphi or whatever the popular flavor of Pascal was at that time. And why DX7? It feels like the era between VESA/software rendering and DX9 was pretty short

>>101301646
I don't have anything against COW, it can be a useful feature. But CreateProcess on Windows has never even needed to consider it, because the model is fundamentally different. COW feels like unnecessary complexity.

File: firefox_wNWIuOVnji.png (66 KB, 1032x650)

66 KB PNG

>>101301694
>And why DX7
I literally slapped DX7 on a random picker wheel site among other old graphics api and platform/compiler names and that's what I rolled
Also VC6 and DX7 are from the same era

>>101301723
MSDOS has way more soul

>>101301753
Yeah but only old hardware renderer api for MSDOS is glide afaik
Could write a software renderer but
hmm

>>101301770
They're software but VGA and VESA are pretty comfy from what I've heard.

How are you supposed to use Glide on DOS? How do you load dynamic libraries on DOS?

File: wcslen.png (8 KB, 500x780)

8 KB PNG

>>101301694
>But CreateProcess on Windows has never even needed to consider it
There's a lot that Microsoft hasn't considered, from fixing up their kernel to fixing up their userspace runtime library. They are much less concerned with speed gains (otherwise their kernel-to-user copies wouldn't be so shit) and much more with vulnerabilities and backwards compatibility (which is why CreateProcessAsUser and CreateProcessWithLogonW came about). Don't assume they couldn't speed up their existing code base if they tried, because they absolutely can and don't.

>>101301799
https://dosbox-x.com/wiki/Guide%3ASetting-up-3dfx-Voodoo-in-DOSBox%E2%80%90X

>>101301802
post the assembly how bullshit are the comments

>>101301802
Oh to be a fly on a microsoft programmer's cubicle wall
Imagine the comments in modern windows source code just complaining about "why doesn't this shit work" or "why do we have to do this"
Though then again, iirc although that stuff is in the 2000/XP source leak, microsoft has a rule not to do that anymore

File: wcslen_assembly.png (35 KB, 1220x480)

35 KB PNG

>>101301815
I should point out that, even if they didn't want to use SSE/AVX (which memcpy does, by the way), they could still issue 8-byte loads here. With vector instructions ... well, the entire thing becomes a farce: https://lock.cmpxchg8b.com/zenbleed.html

>>101301802
There's lots of things I could shit on Microsoft for when it comes to Windows programming, and they have made many things unnecessarily complicated by trying to keep compatibility, even as far back as win9x. Don't even get me started on every -Ex function and -EX structure.

I just think that their fundamental approach, at least in the early days, on how programmers should interact with the OS was much better, and translates much better into modern principles like RAII, rather than trying to obey the 70s Unix dogma.

>>101301913
>I just think that their fundamental approach, at least in the early days, on how programmers should interact with the OS was much better
Their fundamental approach how programmers should interact with the OS was exclusively driven by the question how to make the entire bloody thing work on 16-bit registers and a 20-bit address bus: https://devblogs.microsoft.com/oldnewthing/20131008-00/?p=3003

Unless you meant 32-bit NT, which was such a clusterfuck that MS put parts of IIS in their kernel just to not lose against Apache on Linux.

>>101300835
What about just bruteforcing the answers by removing/readding input features successively until the R2 becomes dogshit?

>>101302011
That's more Intel's fault than anything. Segments were a stupid design.

I'm talking about managing system resources (memory, threads, synchronization, etc) in your program. For all the shit I give Microsoft, they did the right thing here - system resources are all HANDLEs, that you can wait for or cancel. There's no need for this "everything is a file" bullshit.

>>101302154
>That's more Intel's fault than anything
They knew what they were signing up for: https://devblogs.microsoft.com/oldnewthing/20041215-00/?p=37003

And it doesn't change the fact that they had the chance to learn from their mistakes when NT replaced the DOS underbelly, and they didn't.

>There's no need for this "everything is a file" bullshit.
And here I thought you *wanted* to wait for system resources?
https://man7.org/linux/man-pages/man2/select.2.html

>>101301277
No, but sometimes 8.0 (2005) because I've been provided APIs using that. It's amazing people still use it but I don't blame them, they have no reason to care. It's less sluggish than newer versions, but intellisense can't handle much complexity before giving up completely.

>>101286494
it really depends on how often you ask for 20 gigs and in what timeframe. And if you do this you can imagine that there are already 50 chinks robots doing the same
it IS a significant cost (time) for any server valuable enough to be targeted by scrappers
>not a significant cost in 2024.
you don't understand what you're talking about

>>101280210
non-qml qt has filtered me. Guess i will cut my dick off and scream at anyone complaining about clis.

>>101302154
segments were literally the only workable design. What the hell are you talking about?

Windows NT®

I've been told that if I manage to learn CUDA, I won't have to worry about job security anymore. Is it true? Is CUDA that hard?

>>101301277
I think the early touhou games are written in MSVC7 and the reverse engineering efforts for those are kind of interesting
https://github.com/happyhavoc/th06

>>101302890
>early touhou games
the Windows XP era of touhou isn't early 2hu. Early would be PC98.

>>101302954
That's a retarded distinction no one makes ("early" isn't a specific era) but those were written in Turbo C++ 4.0J
https://github.com/nmlgc/ReC98

>>101280210
idk what the fbi injected into my brain but maybe for the first time ever, things are looking so fucking bright bros.

>>101303193
Ride that Mania for what its worth.
Soon it's back to Dementia.

>>101280210
whats the graph for?

I'm forever a ngmi code monkey

File: 1704061928935305.png (1.39 MB, 1280x720)

1.39 MB PNG

>>101303539
I will never finish a project

>>101280210
Does anyone know a good, simple, language agnostic testsuite?
> black box testing based on executable inputs/outputs
> (optional) """wrapper""" unit testing, where it can generate dummy entry points based on a template and insert a few lines of code, compile and run them checking exit codes
i keep finding myself rewriting very similar functionalities in bash to this end...

>Arch updates Clang to 18
>Clang 18 ships with a compiler bug that had a fix merged back in early May but somehow didn't make it into this release

>What are you working on /g/?
Nothing because my project is now defunct.

>>101283959
WTF is this. If this is C++ then ill hang myself tomorrow.

>>101305296
I know, right?
C++ has gotten to the point where you can't make any guesses about how the code in the editor will relate to code in the binary. Not like it's easy with C, but C++ has given up any and all pretenses and just says "trust me bro".
>and then it constantly fucks it up
>and the hardware industry has to come in and fix its architectural fuckups with more speed increases
>to the point where not even Stroustrup is arguing for more speed (because he knows they've lost that race against C), only for more safety (because the jury is still out on Rust).

>>101306998
what does Add do?

>>101307030
nta but I guess it appends to whatever string is already there, instead of replacing it

>>101303814
pytest

>>101280210
6 months in to trying to find a damn job

>>101301574
>I've restricted myself to using VC6 as a little challenge/exercise in masochism.
Not only you will miss on new C++ features, but you will also get shittier binaries. Absolutely pointless. Use >C++11. Choose a different challenge. Code in a cock cage or programming socks, whatever.

>>101307096
>Code in a cock cage or programming socks
No thanks, I think I'll use VC6.

>>101301574
https://learn.microsoft.com/en-us/cpp/porting/visual-cpp-change-history-2003-2015?view=msvc-170#visual-c-net-2003-breaking-changes

>>101301574
write a compiler for a better language

File: Untitled10.webm (2.01 MB, 662x582)

2.01 MB WEBM

>>101307030
>>101306998

>>101307327
Why the need to activate/deactivate it? Why not just type in prefix if you need it?

>>101307338
because programmers should not design UI

>rust on windows needs c++ and installer recommends visual studio >5GB as the first option
I knew hackers hate windows but It's beyond my imagination.
anyway rust seems like a complicated system.
It's best not to get involved in these situations.

>>101307363
Well it would make sense if the prefix is stored for subsequent uses and you want to disable it sometimes without erasing it

>>101285764
It'd probably get rid of one of the stores unless it's a volatile pointer.

>>101280210
I don't know what your pic represents, but it reminds me of Vvardenfell, so it's cool!

>>101307978
>>101285764
See: https://godbolt.org/z/vj4b5E5G3

>>101307054
>pytest
>language agnostic
no, fuck you, you illiterate nigger

How do you cope with the fact that everything that you want to build has already been built by someone wayyy better than you? I was just hit by this and I'm devastated. I don't even want to program anymore.

>>101308458
fuck their gf

>>101308458
>How do you cope with the fact that everything that you want to build has already been built by someone wayyy better than you?
That's funny. I've just looked at the source code of nginx, and for a webserver that's supposed to be ultra hyper fast by some Russian superhacker the memory management is utter garbage, as it's just yet another layer around malloc/memalign. Igor's pathetic attempt to organize everything in pools may be laudable, but it cannot fix such inherent gashes.

>>101308458
Go read some open source code every now and then. Most of it is not as impressive as you think it is, and then also remember that proprietary code is easily 20x worse, because they're always hiding their shame.

>>101308515
>proprietary code is easily 20x worse
Unfortunately it's not so simple. For all the clusterfuck that MS Windows is it's yet somehow faster than the mess that Linux and its userspace is when it comes to gaymes. It's why ebussy doesn't receive nearly as many death threats as he deserves.

File: 1561299316554.jpg (64 KB, 548x548)

64 KB JPG

I've been reading through some parts of the DBus spec, plus the reference and systemd (sd-bus) implementation, and I see stuff about a direct peer-to-peer mode (skipping the DBus daemon) and launching processes with socketpair() to act that way.

While it seems like a fine idea on paper, has anyone EVER seen this used in practice?
I'm finding more and more things in the spec that'll just aren't worth bothering with.

>>101308626
a protocol that accomplishes nothing is a protocol not worth reading about
If anything read the posix socket standard

>>101308688
Why would you do that if shared memory and atomic intrinsics exist? There's no need to involve the kernel past setting up the connection.

>>101308458
I don't. I'm having a midlife crisis

I'm learning C. If I write code that requires C99 to compile, should I somehow use __STDC_VERSION__ to indicate that at least C99 is required? I can only find irrelevant results on search engines.

>>101308688
>Comparing an application protocol to a transport layer
That's like saying "stop reading about HTTP, read about TCP instead".

Speaking of TCP, that's another part of the DBus spec that I have my doubts about implementing.

>>101308706
I read an article many years ago, and I completely forget where so I wouldn't be able to dig it up, but I read that trying to do some kind of RPC that way is actually really fucking hard to do properly, and often doesn't even end up faster than just copying, via a socket in the normal case, especially if the message sizes are small.
Plus sockets are poll(2)able etc, which makes working with more than one of them fuckloads easier.

>>101308515
>Most of it is not as impressive as you think it is
it is impressive to me because I can't tell what they're doing with their code.

I hate that git integration fails half the time on vscode

>>101308706
atomics do fuckall so you are going to have to offload the work to the kernel anyway, you should have known that but you dont because you are larping, in fact talking about optimization when it was not the concern at all shows you are low iq and obsessed.

>>101308740
It's only something you'd maybe bother with when writing a header file for a library.
It's honestly not worth the effort. Even in C89 mode, so many people (accidentally) use GNU extensions which is basically C99 or even later standards without realising it. Maybe if you're using some more "cutting edge" C23 stuff that isn't widely supported/adopted yet.

Most libraries just say their headers are written for C99 and do nothing in the code to indicate that.

>>101308744
yeah that is exactly what i am saying because there is no point to having a protocol like http

>>101308773
He thinks he's really cool and a le ebin h4ck3rm4n like Terry Davis for implementing his own mutexes.

>>101308744
>I read that trying to do some kind of RPC that way is actually really fucking hard to do properly, and often doesn't even end up faster than just copying, via a socket in the normal case, especially if the message sizes are small.
Is that speed, or latency? Because I can believe that the mode switches and user=>kernel and kernel=>user copies can have a lower latency than whatever notification BS they were benchmarking against. But if you want to tell me that doing little work is faster than doing more work, then you've been had.

>>101308773
>atomics do fuckall so you are going to have to offload the work to the kernel anyway
>source: my bourbon-drenched asshole
And if not? Will you make it not work out of spite then?

>>101308808
Your jealousy is showing.

>>101308744
>shared memory in linux is not pollable
kek, do they even swap files contents? retards

>>101308848
>Doesn't even know what poll(2) is
My god, how embarrassing.

File: carrot_bait.jpg (408 KB, 849x1200)

408 KB JPG

>>101308839
>>atomics do fuckall so you are going to have to offload the work to the kernel anyway
>>source: my bourbon-drenched asshole
>And if not? Will you make it not work out of spite then?

>>101308853
>doesnt even comprehend what i am talking about
not the own you think it is
anyway >>101308888

>>101308876
I don't care what works on you, only what works for me. It works for me, and no amount of seething will change that.

>>101308890
I hate that this thread has become plagued with actual schizos.

>>101308946
Good thing I don't, since that's all that matters.

Anons
I made an app. It's for educational purposes and can be used by institutions. Should I make it free for now to grow users and add subscription later, or make it paid from the getgo?

>>101309066
make it paid. also make free demo version.

>>101309066
Be the worst misanthrope you can be. Humanity deserves suffering.

File: 360-tpose.webm (222 KB, 608x432)

222 KB WEBM

>>101307327
In unix this is just

f () { echo "$2$(shuf -zern $1 {A..z})" | xclip; }

$ f 32
rkVCc^sqzuhcuKlFeclH[NJDoxlK_MRr
$ f 32 prefix_
prefix_IYG^wQ]eUfLrfOCMgo_LBULwYxmbWYa

>>101309066
spread the word about it, pressure some friend of yours who works as a principal to use it, eventually contract the government for maintenance

>>101309092
>>101309115
I did make a payment kind of option but I have to set up a firebase database to keep track of which phone downloaded it else you can just delete the app and reinstall it to reset the tracker
but yeah maybe I should go this route
>>101309215
I have some previous teachers who would be very interested in using this app, and they even have some contacts in the education system. If I could get any sort of government contract I'd be set lmao

>>101308832
>gif
That's not how matamoran fusions work. If you don't have the same body shape, you can't fuse.

I wish I had more IQ bros. I am going through some fucked up C++ meta-programming mental gymnastics right now, and my brain can't keep the entire context needed to achieve what I am trying to do. I will get it eventually, but in times like this I can clearly see what an absolute brainlet I am.

>>101309334
Why do you assume it's an intelligence issue? Maybe it's your subconsciousness telling you that C++ is really really really fucking stupid. Writers block works like this, too.

>>101309469
>Why do you assume it's an intelligence issue?
I know the solution is staring me right in the face, but I just can't grasp it. it's a matter of time I get it right. I just wish that time was shorter and the entire process less frustrating.
>Writers block works like this, too.
that's not it. I get writer's block when I am looking for a perfect solution. now, I'll take any solution as long as it works, kek.

File: bg,f8f8f8-flat,750x,075,f(...).jpg (204 KB, 750x1000)

204 KB JPG

>senior colleague consistently writes absolute dogshit spaghetti code and gets pissy when I leave critical feedback on his pull requests (which he ignores anyway)
what do I do bros?

>>101310489
Are there other senior colleagues to pull into code reviews?

>>101310523
No

>>101310489
been there. do your review as usual. wait until he responds to all comments, and just approve. change jobs when code becomes unmaintainable.

>>101310489
jump ship, move to greener pastures

>>101310545
>>101310555
Good ideas desu

>>101307524
That is still possible with a single text input/output.
It would look cleaner but it would be harder to learn.

>>101309334
I have the same replicating ocaml/merlinsearch for haskell.
The pieces are ready but they're still too big.

I don't know how to express the idea that certain type
variable instantiations are possible but unfavorable.
For example there might be an
instance (Num a, Num b) => Num (a,b),
from which (Int32, ((Int8,Int16),Int64)) can be an instance of Num,
which might be useful if the query is looking for all those Ints,
but in general it's wasteful to look at what that (a,b) instance
allows.