I would have made a better thread.

>>108740300
>ugly unfunny soi slop
The maid threads were better.

>>108740300
>ai coming for soislop shitposters' job too
Kek
Maybe let the bubble stay unpopped for a little more while

Non-slop thread:
>>108740331
>>108740331
>>108740331

File: 0c5617a0dd479953.webm (3.15 MB, 716x576)

3.15 MB WEBM

Made a somewhat sloppy program that previews fountain screenplays with an 8x8 pixelfont
(Tested it on the screenplay for Wall-E

>>108740300
>pic rel
reminds me of a yt video about the history and production of game and its management, World of Warcraft I think

I'll take this over the maid thread.
Picked up a couple more PS2 memory cards to use for testing.
I've got to start testing individual commands so I can see what happens when communication goes bad and how to recover from it, if possible.

>>108740300
Fucking lol this 'jak is so relevant to my company it's almost not funny

>>108740489
Lol lmao are ur tits jacked, bro/he/him? Let's talk about it at the barcade tonight!

>>108740300
My current boss is smart as fuck but he also constantly uses corpospeak like
>let's sync once we have the bandwidth, so we can circle back and identify the learnings from this workstream and align on the next steps
And I've started picking it up too, to my great dismay

Like mate we're a startup with 8 people, we don't need to synergise the eschaton

>>108740573
It does sound like mental illness.
Yeah, it is hard not to pick up language being used around you.

File: 78A7BF30-FCB7-446B-8840-5(...).jpg (1.29 MB, 3464x3464)

1.29 MB JPG

>>108740300
You must all learn to vibecode. Stop coding manually like a trans (snailcats). VibeGODDING is the future

>>108740650
I haven't written a single line of code manually for month but the snailcat just looks better, sorry

>>108740657
Transgender behavior (liking snailcats)

God, this is depressing.
https://desuarchive.org/g/search/image/XShUWhMhwvEXOCXYpZTG-w/

>>108740650
reported for begging

2026 and still no "psyop" or "demoralization" reporting option
yep, this is accurate >>108740409
>corpo-govt psyops and programming

>>108740754
link doesn't inline, is this because of the fuckery with the URLs the glowjannies are doing

File: IMG_2253.png (1.88 MB, 1807x1750)

1.88 MB PNG

>god, this is depressing

File: ayy pepe.jpg (11 KB, 225x225)

11 KB JPG

>>108740300
Beautiful C code to solve Advent of Code 2024 day 3:

#include <stdio.h>
typedef long long Z;
static int get(Z* pn) {
    Z c, k = 0; Z n = 0;
    while ((unsigned)(c = getchar()) - '0' <= 9) n = n * 10 + c - '0', ++k;
    return c >= 0 && ungetc(c, stdin), pn && (*pn = n), k;
}
#define $(s) (c == s)
#define N(n) && get(&n)
#define _(c) && (getchar() == c)
int main() {
    for (Z ok = 1, p1 = 0, p2 = 0, a, b, c;
        (c = getchar()) >= 0 || (0 & printf("%lli\n%lli\n", p1, p2));
        ($('m')_('u')_('l')_('(')N(a)_(',')N(b)_(')') &&
            (p1 += a * b, ok && (p2 += a * b))) ||
        ($('d')_('o') && (c = getchar()) && (
            ($('(')_(')') && (ok = 1)) ||    
            ($('n')_('\'')_('t')_('(')_(')') && (ok = 0))))
        );
}

https://godbolt.org/z/rs3ahMs6j

File: 1775742778840664.png (71 KB, 214x235)

71 KB PNG

>>108740816
This looks like something that could make mustard gas strong enough to escape from Chromium so not clicking that link

I haven't used an AI Coding Assistant in 6 months (and haven't coded in 6 months) and I started Codex CLI just now with my ChatGPT Plus account and I asked it just one request and was already down to 93% of my weekly limit afterward. I guess it's true what they say AI Coding is being taken away from normies quickly.

>>108740780
how long has it been since urls are like that? I've been away from this shithole

>>108740868
>I guess it's true what they say AI Coding is being taken away from normies quickly.
Get them onboard at a loss and then raise the prices to gain profitability.

>>108740902
I only noticed it today, there was maintenance recently
I think its a glowfag tracking thing

https://github.com/microsoft/lib0xc
Microsoft I...

File: 1744821455319q.jpg (200 KB, 1024x1024)

200 KB JPG

class Solution {
public:
    int romanToInt(string s) {
        std::map<char, std::tuple<int, int>> romanMap = {
            {'I', {1, 1}},
            {'V', {5, 10}},
            {'X', {10, 10}},
            {'L', {50, 100}},
            {'C', {100, 100}},
            {'D', {500, 1000}},
            {'M', {1000, 1000}}
        };

        return std::accumulate(s.cbegin(), s.cend(), 0,
            [&romanMap](int result, char c)
            {
                auto [value, place] = romanMap[c];
                return result + (value - (result % place) * 2);
            }
        );
    }
};

I like this leetcode solution but I am pretty sure I could just check the 5 scenarios and then dubs it instead.

>>108740313
>they automated the soijakking
>millions of soiteens now unemployed
weird futures

New hire code janny here
What're some good resources or practices for improving C++ code in place. Inherited a big old C++ codebase complete with its own Bison/yacc implemented DSL

>>108741213
>Bison/yacc implemented DSL
slop. I would rewrite the parser using a recursive descent parser. Probably rewrite the rest of the compilation/interpretation too.

File: ymawky 503.png (696 KB, 2740x1666)

696 KB PNG

just added the ability to serve custom error pages in my arm64 assembly webserver. apple (ew) has an undocumented syscall, proc_info(), that gives you info about a given PID, including the number of children that process has. since it just uses fork() on each request, i can use that enforce a connection limit and deny any requests if it's at max capacity. if this was gnu/linux i'd have to parse /proc/<pid>/task/<pid>/children, which would be a major pain in assembly.

>>108741337
>if this was gnu/linux i'd have to parse /proc/<pid>/task/<pid>/children, which would be a major pain in assembly.
Or you could just remember how many times you've called fork and reaped children? You have to do that anyway, so why waste time on unnecessary syscalls?

>>108741359
i used to do it like that, but the reaping loop wasn't working right. at *least* one zombie would remain, only sometimes getting reaped on the next connection while accept() was blocking. so now instead of calling wait4() in a loop before accept(), i use sigaction() to SIG_IGN SIGCHLD with SA_NOCLDWAIT, so the kernel automatically reaps them. but if the kernel's reaping them, i can't keep track of how many have been reaped. and if a child crashes and can't clean up, i'd be unable to decrement the counter.

File: akiba.jpg (281 KB, 1920x1080)

281 KB JPG

>https://leetcode.com/contest/weekly-contest-500/
contest time girls

>>108740300
why does it say "cold pizza" on the box? is it already cold when you order it?

>>108741615
>is it already cold when you order it?
That sound's like something that would be in silicon valley

>>108741615
>>108741758
>“Cooling starchy foods—from pizza to rice—creates ‘resistant starch,’ a carb that behaves like fiber…”
>https://www.scientificamerican.com/video/why-leftover-pizza-is-actually-healthier-the-science-of-resistant-starch/

cold pizza is more healthy! why buy it fresh!

File: sst1_fastfill_impl.png (467 KB, 1916x1048)

467 KB PNG

Implemented the Voodoo 1's FASTFILL command. This was missing and so I'm getting really whacky output in 86Box. The Y co-ord starting at 1 looks wrong, so I probably need to debug it, but that should be an easy fix.

This should (according to Sonnet) just bolt onto the current pixel pipeline with a few changes to the busy signalling

>>108742110
>The Y co-ord starting at 1 looks wrong, so I probably need to debug it, but that should be an easy fix.
I'm retarded, that's actually within the clip rect, so this is expected

I've avoided doing any graphical programming (not counting basic HTML rarely) for like 20 years.
One day I'm going to get into making non-console applications and I hope the libraries are good by then.

>>108742123
I find myself still going back to Java to make anything GUI/Tooling related. It's just way easier to smash something out quickly that works as compared to using something like Qt or whatever fucked up C/C++ libraries exist

>>108742128
All the graphical Java applications I've used are pretty good so I believe it.
When I was learning C from a course, the GUI toolkit somehow interfaced with a Java graphics library / window toolkit, and only the game logic was in C.
What I imagine is that the cross platform support for the JVM(s) is the biggest appeal.
What I remember from looking into C++ stuff like wxWidgets etc. was not great for native look and feel, but maybe that's changed.
Qt huge lol

I suspect TCL/tk or something using Lua could be good too.

File: Screenshot 2026-04-19 105736.png (417 KB, 1402x700)

417 KB PNG

>>108742128

I do ML and my data pipeline is in Java for all the obvious reasons, but unfortunately the ML layer itself is in python, and its kind of retarded how you have to have these huge monolithic 8K line single .py file, yikes, very antiquated language

>>108742123
I attempted to make my own gui library on top of webgl 2 for a C application I am building just for the shits.
it was super fun but I would never do it again. it's hell to debug when something looks wrong and there is a whole class of bugs I will never be able to fix in it

So it seems that when a memory card goes incommunicable, or you change memory card, you have to go through an auth process before you can communicate with it again, which involves reading some auth data from the memory card, sending it to the mechacon (cdvd controller I think) getting some data back that goes back to the memory card as a challenge/auth and that response back to the mechacon.
Which I was not expecting to have to be writing today.

>>108742224
did you look if pcsx2 had to do a similar thing?
I would have thought it would be memory mapped like a gameboy cartridge or maybe you would just need to send read/write requests to an mmaped I/O register

>>108742427
>did you look if pcsx2 had to do a similar thing?
I'm looking at the ps2sdk to see sorta what is going on.
I think the controller in the memory card essentially runs in two modes, pre-auth and authed. When in pre-auth all the regular commands will return a "no device present" result, but the auth commands will return "device present" if the card is there.
So if I change a card (or send a bad command so the memory card crashes or something) I end up in that pre-auth mode and I have to go through this process to get the card authed so I can communicate with it again (without resetting the console)
https://github.com/ps2dev/ps2sdk/blob/master/iop/security/secrman/src/secrman.c#L565

Resetting the console works because when you first boot up into the BIOS it will load mcman and detect/auth the memory cards, and it doesn't deauth when you then load an elf.
Getting this working isn't really a priority, I suppose, but it would be nice to have a more complete memory card handler.

>>108742427
>I would have thought it would be memory mapped like a gameboy cartridge or maybe you would just need to send read/write requests to an mmaped I/O register
On this point, the memory card is connected to a serial interface, the same one that the controllers are connected to essentially.
So you load up commands into a buffer and then trigger to send them, and then read out the results when it comes back.

File: screenshot-03-05-2026-17:38:29.png (49 KB, 590x408)

49 KB PNG

Has anyone ever made software like an interactive program in bash?

>>108740650
>CRITICAL THINKING IS TRANS (BAD)
kys

>>108740300
that picrel hits way too close to home

;; send up to two 8byte values to mechacon and read up to two back
;; $s0 carries the command value to send to the mechacon
;; $s1 is how many bytes to send (8 or 16 probably)
.align 128
mecha_input:
    dc64 0, 0
mecha_output:
    dc64 0, 0
.align 32
Mechacon_auth:
    addi $sp, $sp, -48
    sd $ra, 0($sp)
    sd $s0, 8($sp)
    sd $s1, 16($sp)
    sd $s2, 24($sp)
    sd $s3, 32($sp)
    sd $s4, 40($sp)
    jal CDVD_wait
    nop
    li $s0, CDVD_STATP
    li $s2, mecha_input
Mechacon_send_loop:
    lbu $s3, ($s2)
    addi $s1, $s1, -1
    sb $s3, ($s0)
    bne $s1, $zero, Mechacon_send_loop
    addi $s2, $s2, 1
    li $s0, CDVD_SCMD
    lbu $s1, 8($sp)            ; read the command out of stack
    jal CDVD_wait
    sb $s1, ($s0)            ; send command and wait
    li $s0, CDVD_RESULT
    li $s2, mecha_output
    ori $s4, $zero, 0x10
Mechacon_get_loop:    
    lbu $s1, ($s0)
    addi $s4, $s4, -1
    sb $s1, ($s2)
    bne $s4, $zero, Mechacon_get_loop
    addi $s2, $s2, 1
    ld $ra, 0($sp)
    ld $s0, 8($sp)
    ld $s1, 16($sp)
    ld $s2, 24($sp)
    ld $s3, 32($sp)
    ld $s4, 40($sp)
    jr $ra
    addi $sp, $sp, 48

Because the "Mechacon" is in the CDVD area, it is communicated with using the same SCMD process as reading the RTC, so I can easily copy and adapt that routine and add a couple of little buffers to hold the input and output.

>>108742224
>>108742471
>>108742838
You working on a PS2 emulator anon?

>>108742882
No, I'm writing a gameboy emulator for the PS2.
In assembly. All assembly. Even the controller and memory card communications.

How much would it cost to hire someone to make a new C-like programming language? I'm too dumb and lazy to make it myself.

>>108740300
The head honcho at a company I used to work for unironically purposed we do a "hackathon". His idea is that his employees worked for free and come up with new ways for the company to make money in a 24 hour sprint. What the fuck? lmao

>>108742894
You're the autist the world needs.

>>108742588
>Has anyone ever made software like an interactive program in bash?
never in raw bash, but ncurses was fun though.
>https://hackaday.com/2025/06/17/a-gentle-introduction-to-ncurses-for-the-terminally-impatient/

>>108742894
>No, I'm writing a gameboy emulator for the PS2.
>In assembly. All assembly. Even the controller and memory card communications.
based. I kneel. I started one in c, but the sound channel apu stuff crushed me. cpu and gpu was fun though.

>>108743646
Their idea is to catch all the internal hacks and then patch them.

>>108743537
It would cost 4 trillion dollars and it will take 20-30 years.

>>108743867
APU is what I should be working on instead of getting distracted on making saves.

>>108743537
C is perfect

>>108743537
you're not satisfied with the current dozen C clones?

File: cute_dance.gif (1.56 MB, 568x640)

1.56 MB GIF

>>108744008
It will be close once defer gets into the standard.

>>108744760
>didn't animate the clap
bad copy

>>108744846
Good Copy Bad Copy (subtitled Good Copy Bad Copy: A documentary about the current state of copyright and culture) is a 2007 documentary film about copyright and culture in the context of Internet, peer-to-peer file sharing and other technological advances, directed by Andreas Johnsen, Ralf Christensen, and Henrik Moltke. It features interviews with many people with various perspectives on copyright, including copyright lawyers, producers, artists and filesharing service providers.[1]

File: laputan.png (93 KB, 1432x3132)

93 KB PNG

ive upgraded PHP's type system a bit so the de-macro-ification became possible. now extending runtime is much simplier and its proper C lang (without macrolang retardation)

here you can observe the module spec. anyone who tried extending runtimes may find it superior to Z-API or N-API. it doesnt require full runtime re-compilation but only this file, so those funcs become available fast.

>>108745007
Thank you AnonAI

>>108745099
uh, is this some old system or do you unironically use a font that gives you cancer?

File: Gengedance.gif (879 KB, 512x384)

879 KB GIF

>>108744846
NTA but here's one for u anon

>>108745300
jerky af animation but at least it has the clap

I use C because I find automatic memory management to be more confusing than just remembering to write free.

>>108745433
I use C because I'm a memory control freak, and also because I like pointers

>>108745433
I use C because that's the only compiler that ships with my embedded platforms

>embedded dev so I need to use C
>also write my PC applications in C mostly because I have lot of libraries written in C
TCL for GUI is usually enough and you can extend it with really easily.

>>108744558
I've tried every one of them and not a single one fixes any of the grievances I have with C, instead they focus on stupid retard things like forbidding me from using tabs to indent code.

thoughts on OCaml?

>>108745137
it gives me the revelation, yes, it gives you the cancer. otherwise, do you have anything to say on topic? or you too wrote a web server in hexadecimal codes? hehehe

>>108746189
i got this book called ml for the working programmer which was published in like the 90s and in the prefece it talked about c being dead in 10 years and i closed it and returned the book

 
$(TARGET): $(OBJS) $(SPV_SHADERS) | $(BUILD_DIR)
        $(CC) -o $@ $(OBJS) $(LDFLAGS) 

How do I make it so that objs and flags are separated by slashes and newline?

I want to learn to program c# what is the best way to start?

>>108746382
Use a real language for your build scripts.

#!/usr/local/bin/tcc -run
int main() {}

File: test.mp4 (3.24 MB, 1920x1080)

3.24 MB MP4

>>108740300
I am working on a video generator that walks over annotations in a pdf file and generates a video that scrolls through the pdf and puts that text in the spotlight, like they do on television

>>108747288
cool idea

Virtual memory is just Keynesianism for computers

>>108747288
Ooo that looks nice. Good job.

I think I'm about half way through the auth process with the PS2 memory card.
A few bugs at the beginning I had to fix, but now it gets about half way and then errors out.
Annoyingly, I can't really use PCSX2 to test it because the virtual memory cards just don't auth at all. It doesn't seem to be emulated at all. The mechacon can send it's data but it doesn't fucking matter. The cards don't really seem to ever exist in an unauthed state.

File: 1768937932954001.jpg (15 KB, 320x213)

15 KB JPG

Today I finished a small server that I wrote by hand in x86-32 assembly. I think doing it in 32 bits ended up making it harder since it seems like there are less resources online for 32 bit.
Next up will be the client, which I expect will be easier.
It's nice that I've been doing this stuff in C already, since it seems like C has only the thinnest possible wrapper around the POSIX syscalls.

it feels like this thread is on repeat mode

>>108746159
>the grievances I have with C
what's your list?

File: fedora-linux-44-1-816x345.jpg (49 KB, 816x345)

49 KB JPG

>>108746213
I don't have anything specific to say, but I will upgrade to fedora 44 now

>>108749180
>I think doing it in 32 bits ended up making it harder since it seems like there are less resources online for 32 bit.
but why? the only thing you need to know are the common instructions and the calling conventions

>the auth bytes coming from and being sent to the memory card need have their orders reversed
Mistakes were made.

>>108742142
>native look and feel
Doing correct native look and feel takes a ton of work, whatever tool you use. Different platforms want their GUIs to work differently, and it's a whole load of little changes in how everything interacts so the toolkit can't do the whole job.
Fortunately, lots of apps use electron now and their look and feel are wrong every platform, and most users are used to that and you can do whatever the fuck you want. lol

>>108749455
I got tripped up by some of the syscalls and the weird calling convention related to them. Several syscalls like bind and listen are actually subcalls within socket and take an array of arguments with a pointer to it stored in a register, rather than all arguments being loaded in registers. This doesn't happen in 64 bit where they're all independent syscalls. If you don't already know that 32 bit has different syscall numbers for these things, like I didn't, you get weird and confusing failures.
Anyway, it's not like it was exponentially harder, it's just that you have to know to look for these things.

File: mcprobe.png (1.05 MB, 845x8930)

1.05 MB PNG

Ok, memory card auth is working.
Just ignore the Memorycard_set_cmd and Memorycard_comm routines, or use the mcdump/mcnuke source as reference.

>>108749702
based.
Are the NOPs after the jumps required or do you have a return bug on your emulator? :^)

>>108750053
I just have nothing to put in those branch delay slots (on the EE when you branch/jump the instruction following the the branch is also executed, unless it is a branch if likely and the branch is not taken).
I should actually remove those vsync_wait jumps as well, they're not needed.

>>108749371
>what's your list?
Verbosity.

>>108740300
>picrel
Jesus that's literally us atm. Backlog filled to the brim, customers churning left and right.

"We need to use more AI guys" is all I'm hearing from both my direct manager and from upper management.

I'm tired, chief.

>>108750186
Bro just have an monster ultra white and watch the LLM do the work

>vibe team is responsible for generating tests and checking that the code passes the tests
>they push broken code
>there is no tests and nobody has done anything
>work off the clock to get project working
programming has always been and will always be a mistake

Is it better to generate a file piecemeal with fprintfs and fputcs all over or should I be allocating a big buffer, generating the file into it and then dumping the entire thing with one fwrite call?

>>108750459
if we are talking performance less calls the better

>>108750476
i mean, libc is certainly going to have its own internal buffering to minimize syscalls

>>108750482
>i mean, libc is certainly going to have its own internal buffering to minimize syscalls
imp lying

>>108750493
i will minimize your cock with a knife. In minecraft

File: Annotation 2026-04-18.png (19 KB, 514x349)

19 KB PNG

https://pixel<delete it>drain.com/u/QAmX4mm5

>>108740573
Interesting how much of what is now corpospeak was adopted from jargon that began among the programmers themselves.
>bandwitch
>cycles
>blocking
>take it offline
But it seems "layer 8 problem" never made it into the mainstream.

>>108750459
Use writev.

File: snailMAID.png (3.42 MB, 1807x1750)

3.42 MB PNG

>>108740672
Snailcat is a cute maid now, and everybody likes her.

>>108750482
>libc is certainly going to have its own internal buffering to minimize syscalls
Yes.

>>108750459
Alternatively, if you know the file size ahead of time, fopen -> ftruncate -> mmap.
Although sometimes that is not actually faster than just doing a copy.

>>108751586
Correction: open -> ftruncate -> mmap -> msync

>>108751544
wow is this a real animal?

>>108751613
Yes
t. snailcat maid rancher

>>108751648
Thank you for training the snailcat maids! <3

>>108751613
its a french delicacy

How do you code when chapGPT is down? A friend said main pages, but i dont know what that is

>>108752284
Use deepseek

>108752284
>This is the grim reality we live in

File: YOUR-A-NIGGER-FUCK-YOU.png (530 B, 80x80)

530 B PNG

>108752789
>Anonymous 05/04/26(Mon)18:40:01 No.108752789▶
> >108752284
> >This is the grim reality we live in

>the anti-maid janny is back
Hopefully the Iran war resumes and he has to spend a couple more weeks in his bunker in Tel Aviv.

>>108752902
maidsnailcat is unaffected

>>108740313
No they weren't, you tranny

>>108752951
I don't think janny actually reads threads or looks very closely at images. I think janny just deletes obvious maidposts.

>>108753049
Maids are in every way superior to soislop.

>>108753409
>programming
>fun

>slop'ing
>fun

>>108753606
>>108753617
these are both correct

Is there a Japanese programming language?

>>108753660
ruby

>>108753666
I meant like, with Japanese as the syntax

>>108753677
https://pastebin.com/DCtsRu3v

https://pastebin.com/HVWdj7h5

#include "nihongo_c_kana.h"

せいすう しゅ(む)
{
    せいすう ごうけい = 0;

    くりかえし (せいすう i = 1; i いか 10; i++) {
        ごうけい += i;
    }

    もし (ごうけい おなじ 55) {
        ひょうじ("せいこうです。ごうけい = %d\n", ごうけい);
    } でなければ {
        ひょうじ("なにかがおかしいです。\n");
        もどる しっぱい;
    }

    もどる せいこう;
}

is dependency injection just a fancy way of saying call a method with a parameter? What the fuck is wrong with software devs why did they need a word for that.

>>108747288
smooth scrolling effect I like it. Idk why but okular and the others suck ass these days and barely render a single page without lagging.

>>108753677
ruby supports unicode with all the emojis and shit

>>108753775
yes
it's insane

>>108753775
academics (and others) love to come up with new jargon for already known often simple things in order to sound like they have discovered something profound, it is just how it is

>>108754019
>academics
This shit is by software engineers. Academics create cool terms like sheafification.

>>108753791
For some reason, GNOME 3's Evince is by far the fastest one, for me at least. Selecting text is instant too, everywhere else the text selection kind of lags behind the cursor.

File: Untitled.png (291 KB, 1600x900)

291 KB PNG

>>108753409

File: 1777692788600455.png (630 KB, 638x960)

630 KB PNG

Is it possible to get a job?

Is it possible to make a true no-fap app? the goal is to block all apps except the ones you white-list. but it cant be uninstalled (while the timer is running) or cant be closed by task manager otherwise it opens opportunity to relapse. not sure how to achieve that

>>108754256
in welding, sure

>>108754317
Cock cage

>>108754317
>it cant be uninstalled (while the timer is running) or cant be closed by task manager
That just sounds like a virus with extra steps.

>>108754317
As long as you have root access to your PC you can't prevent yourself from doing anything. Also most people have multiple devices so what prevents you from grabbing your phone if your PC is locked down, or vice-versa?
It may be possible to do something like, set up secure boot, whitelist only your current OS (to prevent booting into a live distro to bypass it), then setting something up which changes the root/admin password and revokes your access temporarily, and then runs in the background as root/admin and blocks your shit.
On phones I'm not sure, since by default you don't have root, it means you can't touch system apps but it also means you can't install system apps. You'd likely need a rooted phone and then have the same mechanism to temporarily lock yourself out of root.

Even that may not be 100% foolproof, I'm not sure yet. You might be able to boot into safe mode with that process disabled or some shit like that.

The truth is that the best option is often to add enough friction and extra steps that even though you CAN disable it, doing so gives you an opportunity to think "why am I going to such lengths just to relapse, what am I doing", catch yourself and back out.

File: er.png (119 KB, 1563x843)

119 KB PNG

>>108740300
I'm building a document management system and am kinda lost in the DB system.
I have some data that needs to be kept safe from modification for ten years after a document is minted because of EU financial bureaucracy, but some data needs to be modified for future minting for example when a client changes their name or a record needs to be deleted due to EU data-protection bureaucracy without nuking the "historic" data.

I've been beefing with a few different LLMs for three days over this, the probably best solution yet was to just say "fuck disk-space" and keep shadow tables of everything with a revision-timestamp and -counter.

Pic-Rel is my current ER-Diagram, what I used to do in an earlier version was to build a generalization / inheritance ("is-a" Triangle) and copy the needed data into a new record in another Entity, for example: Position became a carbon copy of Ware, changing Ware after creating an Order containing said Ware would not change the corresponding Position entries unless you "re-minted" the Order after the fact.

TL;DR: How do I keep historic records in a database without going to jail for breaking laws that pursue opposing goals?

>>108754418
that's what I meant

>>108754710
assuming your documents are plain text based: just use git, dummy

>>108754093
I genuinely don't understand why more people don't just reserve a terabyte worth of memory and commit as needed with a glorified bump allocator tho

we have virtual memory for a reason, just let the kernel do most of the work

>>108754256
No. I got the very last one. Sorry.

File: dc6b603e4f0bb5f3aa2e09bfb(...).png (2.7 MB, 2246x2900)

2.7 MB PNG

fixed occasional playlist loading issue for my youtube client
youtube is ab testing again

>>108742128
i made my own gui toolkit to avoid dependencies

>>108740300
>"thanks for all the hard work. i got you pizza! i'm sorry i can't stay late, i have prior plans"

>>108755063
Kys troon

>>108740650
spelled Autists wrong

>>108754710
just have a separate database/data warehouse for the historical data, don't complicate your live database over this

File: screenshot-05-05-2026-11:11:06.png (31 KB, 734x386)

31 KB PNG

NULL terminated arrays.

>>108755660
Devilish
>>108755274
Evil

>>108741111
Is that rust?
Elegant solution!

status update:

i blocked 4chan and a bunch of other sites on my main PC a few weeks ago. this lead to a downward spiral of blocking everything i could conceivably waste time on but my average productivity failed to rise.

i then installed server linux on my laptop so that i would have a distraction-free computer. this lead to me creating my own "ai overview" command line tool in python because lynx doesn't support javascript. it's substantially slower than gemini in a modern browser and if i use my shitty local llm it remarkably follows the system prompt but hallucinates terribly.

anyway, i would up circumventing all of my blockers on my main PC by just installing WSL and then typing sudo apt install firefox.

i am currently going down a rabbit hole of learning about unix v6 and pdp 11 emulation because i have lost control of my life.

File: 1769816729358.jpg (172 KB, 960x1280)

172 KB JPG

Let's say I want to be as autistic about memory as possible and make desktop applications. What programming languages let me go as low as possible on RAM/Memory?

Is there anything in between Assembly and C in terms of RAM footprint? Like maybe a language that targets embedded systems but can also compile to PC targets?

>>108756226
You don't need to block stuff you just need to learn self control.

>>108756229
forth, maybe? no clue if it actually performs better than C but it's definitely lower level and requires you to manage the stack manually.

>>108747288
Nice. I’d like to send people this in response to their stupid questions that are answered in our manual

>>108756234
i am a lot more productive than i used to be but i freak out sometimes and start taking drastic measures that don't pan out or backfire in some way. i have to settle for being productive most of the time and not all of the time. it's just not going to happen.

>>108756260
They back fire because you think you need to take drastic action.
You just need to develop more healthy habits over a long period of time.

>>108756241
Ooh, thank you anon. I will look into that one.

>>108755660
there was some GNU shit I used that did this and it felt pretty fucking retarded to write

File: 1757869870458681.png (8 KB, 365x89)

8 KB PNG

>>108756297
here it is, argp options array. an entire zero'd struct as your null terminator. at least with strings / primitive arrays the null character is the size (or smaller) of size_t where you would have stored the length anyways, but using it for large objects is a crime

>>108755371
You are gay

>>108755660
>>108756319
People forget that argv is like that, i.e. argv[argc] == NULL is guaranteed.
It shows up in execv too.

>>108756229
LLVM intermediate language.

File: check'd67.png (581 KB, 1210x680)

581 KB PNG

>>108743867
SIX SEVEN!!! CHECKED!!!

File: 1769241046437259.jpg (10 KB, 563x256)

10 KB JPG

Does anyone else use 4chan-XT? It's been archived by the "original" fork author but I have too many threads in it to switch.
If anyone else is using 4chan-XT and wants the video filename not showing up fix, you can find it here:
https://github.com/nick-s-b/4chan-xt

>>108755660
Why are you defining a pointer instead of an index in the for loop? I know it works but it looks confusing

for (int i = 0; items[i] != 0; i++)

Is better for readability and the compiler probably optimizes it the same as the pointer version anyway

>>108757220
Because K&R told him to write like that.

>>108757220
>the compiler probably optimizes
>probably
okay either you fucking know and care or you don't.
muh probably it's good. Either you want it good and look it up or you don't care. it's reallly that simple, faggot

>>108757302
Ok, I went ahead and tested both cases at -O3 with gcc
For the original code, it compiles to:

.LC0:
        .string "banana"
.LC1:
        .string "orange"
.LC2:
        .string "apple"
.LC3:
        .string "pear"
main:
        push    rbx
        mov     edi, OFFSET FLAT:.LC0
        sub     rsp, 48
        movq    xmm0, QWORD PTR .LC4[rip]
        mov     QWORD PTR [rsp+32], 0
        mov     rbx, rsp
        movhps  xmm0, QWORD PTR .LC5[rip]
        movaps  XMMWORD PTR [rsp], xmm0
        movq    xmm0, QWORD PTR .LC6[rip]
        movhps  xmm0, QWORD PTR .LC7[rip]
        movaps  XMMWORD PTR [rsp+16], xmm0
.L2:
        call    puts
        mov     rdi, QWORD PTR [rbx+8]
        add     rbx, 8
        test    rdi, rdi
        jne     .L2
        add     rsp, 48
        xor     eax, eax
        pop     rbx
        ret
.LC4:
        .quad   .LC0
.LC5:
        .quad   .LC1
.LC6:
        .quad   .LC2
.LC7:
        .quad   .LC3

Meanwhile the for loop with the index one compiles to:

.LC0:
        .string "banana"
.LC1:
        .string "orange"
.LC2:
        .string "apple"
.LC3:
        .string "pear"
main:
        sub     rsp, 8
        mov     edi, OFFSET FLAT:.LC0
        call    puts
        mov     edi, OFFSET FLAT:.LC1
        call    puts
        mov     edi, OFFSET FLAT:.LC2
        call    puts
        mov     edi, OFFSET FLAT:.LC3
        call    puts
        xor     eax, eax
        add     rsp, 8
        ret

It seems in the first case the compiler decides to do some vectorization fuckery, while the latter case the compiler decided to simply unroll the loop, so maybe the first version is faster because of SIMD? but I can't really tell what's going on actually, is the compiler concatenating strings on the fly before passing them to puts?

>>108757525
Well, I guess unrolling is always better, if you're not size constrained.
But my actual point was: It's not important. If it is important, then you'll notice it. Apart from that just do it the way that works for you or your project

File: screenshot-05-05-2026-21:28:11.png (52 KB, 551x614)

52 KB PNG

>>108756701
HOLY SHIT! I never knew this.

I use goo mamo btw ;-)

File: jackie thumbs up.gif (1.94 MB, 389x185)

1.94 MB GIF

>>108756835
Thanks a lot, anon.

I just discovered that apparently 32 page per block 8MB PS2 memory cards exist, which my current stuff won't work for.
But that's ok, get the basic cards (ones I have) working and restrict to them first, then make it more compatible across different card geometries.

>>108757790
there's also envp, supported on both unix and windows. iterated the same way. which also explains why you need to restart applications to see environmental variable changes.

>>108757302
If you know, the compiler knows.

>The compiler will optimise this
Sirs... the compiler is MSVC

>>108756835
>4chan-xtx
heh

>>108760498
is 4chan-xt better than 4chan-x? maybe I should look into it. I have a bunch of 4chan-x annoyances that have been bothering me for years.

>>108760562
Look at the readme, I'd say it's better

>>108760568
>sounds off
>he doesn't even listen
https://www.youtube.com/watch?v=4352rEUgM60

>>108760568
>>108760650
Ok I finally made the effort to find the video. It's this one:

2017-02-06T00_01_00+00_00 - Terry A Davis Live Stream (Nh0xnEG3oGo).mp4

https://i.4cdn.org/wsg/1778011035296940.mp4

Fuck you

File: 1760778915110.gif (545 KB, 320x320)

545 KB GIF

C# has a standard library CBOR reader/writer now and it’s really comfy
I love a good serialization format

I use KrashDE, AI keeps returning with QT as my best option for making desktop gui programs. Anyone got any other suggestions?
Getting started with C then C++ wasn't too bad for basic quick-starting, but the tutorials for QT itself seem all over the place. The official "academy" feels like something you woefully get forced to do by some obsolete sr position.
Going to keep bashing my head into this brick wall until it makes sense.

>>108755660
Null terminated array are nice, but that style is too verbose.
C23:

#include <stdio.h>
int main() {
    char* a[] = {"banana", "orange", "apple", "pear", 0};
    for (auto i = a; *i; puts(*i++));
}

https://godbolt.org/z/j9od5dv94

File: 1765509052274808.jpg (31 KB, 402x577)

31 KB JPG

>>108760588
>Look at the readme, I'd say it's better
Thanks! Just installed it. Having a preview of images is worth it alone. Thanks guys.

>>108756835
call it
>4chan-ext

>>108760807
4chan special/super ext (4chan-sext)

>>108756835
Does it also include whatever other manual fix we had to add to the script a couple months ago? Forgot what it was.

Should I use an MCP to make a decomp of an old PC game for shits and giggles?

>>108761107
not him but got some more info? I'd like that implemented too.

>>108761163
Not 100% sure but I think it's this
https://github.com/TuxedoTako/4chan-xt/issues/207

>>108760226
why doesn't bash need to be restarted?

>>108756835
>>108760807
call it
>4chan xp sp2
Tho, I won't install it anyways. I have never understood 4chan extensions.

>>108761176
I'm pretty sure that fix is in the original XT since I can click on "Get Captcha" and I get "Verification...." and can Submit just fine.

>>108761203
Ah, you're right, it's literally the last update of the old fork.

>>108761202
>I have never understood 4chan extensions
XT is one of the reasons why I still use this site. It makes it actually usable. I can't even imagine using 4chan without one of these extensions... since it would be such a massive PITA that I would just give up.

https://github.com/GitFrog1111/OpenWhip

No one is programming.

>>108762817
That's what happens when you get a soislop thread instead of a maid thread.

>>108762847
Still better than the maid threads.

File: 1774516658107448.png (101 KB, 788x1131)

101 KB PNG

you WILL rotate the box

>>108762892
Objectively wrong

File: gbemu-badmc.png (3 KB, 300x200)

3 KB PNG

These images take up more space than pokemon yellow now, uncompressed.
It's fine.

>>108762817
I go through periods where I don't

File: animesher.com_nichijou-ca(...).gif (890 KB, 1000x458)

890 KB GIF

>>108762897
lol I know we just did it days ago but I already forget. actually the python version isn't so bad its just

[row for row in zip(*grid[::-1])]

but imagine explaining that to some normie.
>"just list comprehension the reverse the splat then zip the pairs"

Why I Love Ada
I love Ada because it allows me to precisely describe domain concepts using types. Thanks to the language’s properties, the compiler can detect many errors in advance, and readers can more easily understand the essence of the code. Ada is focused on reliability, safety, and large, long-term projects.

>so embarrased by my lack of programming ability that I don't want to do leetcode problems to actually improve my programming ability.
why am I like this

File: screenshot-06-05-2026-16:42:39.png (151 KB, 895x1037)

151 KB PNG

I made this to sanitize images after finding out the hard way you can get doxed by your metadata.

Yes this should be a bash or python script but it isn't and wont be.

>>108764128
just create a new online personality, with new email and nickname you've never used before, to dump all of your embarrassing stuff

File: mpv-shot0003.jpg (261 KB, 1280x720)

261 KB JPG

ok lets get back to koding

File: 1777766684209883.png (802 KB, 1230x1348)

802 KB PNG

>>108764864
I love kuuuding!!1

>>108762897

array.transpose

>>108764202
Why C?

File: Programming Notes.png (148 KB, 749x816)

148 KB PNG

Sigh... I was getting ready to /l2code/ then AI happened...

I have decided that my memory card messages are too large, so I'm going to make them smaller.
So now I have to redo them all again.

File: Brainlet-meme-8-3246057797.jpg (121 KB, 1024x576)

121 KB JPG

>>108765548
>c#
you were saved. c# might be the worst written language of all time. I'm currently working with literal retards.

>>108765573
Are you neet?
Are you actually autistic?
How come you don't have a mental breakdown yet?
Impressive work and I am a bit jelly about your perseverance

>>108765623
Because I actually spend very little time each day on it because I'm lazy.

>>108765741
based
ok then i dont have to feel bad either

>>108754710
I'd just treat it as two separate problems:
1. Storing a revision history to be able to view each previous version of a document
2. Amending the historical data according to data protection laws

Start with 1, i.e. just store every revision of a document. This could just mean copying the document on every change and saving the old version as a new historical record, or it could mean delta records or someshit, or if your documents are large files then chunk-based copy-on-write might be worth it (so changing a client name in the text of a PDF document with 50MB of images does not duplicate all 50MB).
The CoW question is going to depend on your tradeoffs. If you can just tell the government to pay for more hard drives, or if there's only ever going to be a TB or two of data, it might be whatever. If you can mandate the use of btrfs (or zfs), you can get CoW for free, at the cost of being tied to a specific operational constraint (xfs supports it too but opt-in I think). You can also implement CoW yourself for your custom DB but the cost here is the complexity of implementing and maintaining it which may or may not be worth it.

Then for 2, you just need a record management system that can make targeted edits to historical versions. The exact requirements depend on your laws, but if it's something like redacting a specific part of PII, the operation would be "edit all the historical records (to amend a specific name, or delete it)", which is easy as long as the diff is well-defined across different document versions. (E.g. a search-and-replace is easy. A well defined diff patch is also fine as long as it's unambigous across all previous versions.) Nuking all previous versions is also easy. This should probably cover 99% of data protection compliance usecases.

RE doesn't seem relevant unless there are things that must remain pointing to old revisions of other things; if a customer changes their name, shouldn't orders also be updated to the new name?

>>108740313
>t. Wasnt invited to hackathon by the boss

File: 2026-05-06_15-59-22.png (38 KB, 292x471)

38 KB PNG

>>108761180
It does.

>>108763501
Ada is just a worse version of Java. Or, I guess since Ada predates Java, you could call Java a better version of Ada.

>>108763501
We are no longer living in the 90s. Most of the mainstream languages now have very expressive type systems, even more so than Ada.

>>108756835
see >>108766605

>>108766918
Yet they still all suck, because normies don't want correct software. You'd quickly run out of business otherwise

>>108767023
Hi, I'm the one who forked it.
Thanks for pointing out this to me. I'm reluctant to change the name right now since many people won't be able to find it when they're looking for a fix. How about we revisit this issue in a few weeks?

I do plan on fixing bugs and maybe adding more features to it. I will also accept PRs. I love XT too much and use it all the time to let it just die.

>>108767371
Things like modern C++, Typescript and Rust exist precisely because people want correct software.

What LLM you guys use?

>>108767400
thank you
I also have this issue I posted in the other thread if you wanna look into it

I've also noticed that deleted posts that are revived from the archive, show up at the bottom of the page instead of their 'normal' position.
basically always after the quote, which doesn't make sense because the post was made before being quoted

>>108767400
I dunno anon, I don't use XT but keeping multiple forks with the same name is usually how confusion starts and and trust erodes. Then you end up with several different "4chan XT" all with different features, updated at different times, from different maintaners. People will complain that XT can or cannot do this, or has this or that bug, someone will make a malware fork, etc.
Even if nobody else forks it, if the original guy ever comes back and suddenly updates his repo again then immediately there will be two different 4chanXTs with different features and bugs and developers.

>since many people won't be able to find it when they're looking for a fix
I sincerely doubt that a lot of people will find some random guy's repo called "4chan-XT" and use it, but if it were named even slightly differently would fail to find it.
If anything, the fork on github having the same name as the upstream project is the default, so at first glance it just looks like someone's personal fork (like the 18 other forks https://github.com/TuxedoTako/4chan-xt apparently has), not an actual serious project in its own right.

A lot of my solutions to programming problems tend to be inelegant and bordering on brute force and that makes me feel retarded. Like I feel like I'm supposed to be doing something differently but I cannot see it.

>>108767467
None

>>108767467
>>108767702

File: rotate_fall.png (346 KB, 582x476)

346 KB PNG

>>108762897
this one feels like a part 1 of a aoc kinda.

>>108767794
based maidposter

>>108767794
cringe maidposter

>>108767794
seek Medical Assistance In Dying MAIDposter

>>108767843
>>108767897
>t. butt-blasted sharty tourists who don't know how to program and are intimidated by the raw talent of maidposters who solve problems you don't even understand for fun while posting cute maids

>>108767917
>solving actual problems
>working on actual projects
>calling other people tourists
you've exposed yourself

>>108767467
the Chinese one

>>108767600
>>I've also noticed that deleted posts that are revived from the archive, show up at the bottom of the page instead of their 'normal' position.
>basically always after the quote, which doesn't make sense because the post was made before being quoted
I've noticed this too! Feel free to open an issue when you get a chance.
>>108767660
>If anything, the fork on github having the same name as the upstream project is the default, so at first glance it just looks like someone's personal fork
Fine. I see your point.
https://github.com/nick-s-b/4chan-xt/issues/1
I've opened the issue. We can pick a new name.
If anyone has any suggestions, just leave it there.
4chan-X -> XT -> ?

"neXT"? "sXT"? kek

File: deus_ex_jc_cyberpunk_jc_d(...).jpg (511 KB, 1920x1080)

511 KB JPG

I manage a repo at work. We don't have any formal guidelines around AI usage so I wrote these in the readme for the project. What do you all think?

>1) You may use AI coding agents to generate code which is committed to the repo
>2) You must personally read and review every line of code generated by AI. Do not expect a code reviewer to do it for you.
>3) You must test all code that is committed to the repo, AI generated or not.
>4) AI generated code must still comply with stylistic and linting rules (run bun run lint in order to check the entire monorepo). Some rules or heuristics are not covered by eslint (for example, "prefer stateless functions to classes"), so they may come up in code review. This is fine, it's a part of the process.
>5) Disclaim the use of AI coding agents where appropriate. If a large amount of your code was generated by AI, simply note it in the merge request. You do not need to disclaim if you consulted an AI chat bot for advice, only if the AI generated code that appeared in a commit.
>6) Do not commit any Agent-specific file to the repo (such as claude.md). Add whatever agent-related files you create to the .gitignore file before committing your code.
>7) Do not commit any Agent-specific VS Code configurations to the .vscode/settings.json file.
>8) You may use AI to generate unit tests, but you must also read them and confirm that they handle all necessary cases

>>108767991
The worst problem is code bloat which isn't covered by the guidelines. AI is good at generating code that LOOKS useful.

>>108767991
You should ask them to include AI generated anime pornography to prove it's not a censorslop model

File: Xubuntu-Lunar-Lobster.jpg (226 KB, 1920x1080)

226 KB JPG

>>108768014
I think that's just gotta come up in code review. How do you objectively define code bloat? You can't legislate bad taste away

>>108768014
100% this.
Not even only looks useful, it might even is useful, but it's just fucking bloated and solves a non-existing problem.

>>108767948
I'll make the logo

>>108767991
Is this a public project or an internal one? How massive is the corpo, and how retarded are your colleagues?
Most of the rules boil down to "you are responsible for your own AI generated code". If this is going to be a public open source repo, then maybe it's reasonable. If it's a public repo but you don't intend for there to be many public contributors, you don't need to make it this verbose. If it's an internal repo them IMO this is way too autistic: colleagues should be expected to commit good code, AI or not, and if someone is submitting massive AI slop PRs then that's something to take up with them personally or with management, not to fight in guidelines in a README.

Other than "write good code, and AI-generated code is still your code", I'd say:
>Disclaim the use of AI coding agents where appropriate.
Only worthwhile if AI use is still rare in your company, or if you expect tons of unrelated contributors you can't vet. Otherwise you can just enforce coding standards and if someone is submitting lots of clear AI slop you take it up with him for submitting lots of slop, and attribute the slop to him personally, i.e. not "you are submitting too much LLM code!" but "the code YOU personally are writing is extremely over-designed".
>If a large amount of your code was generated by AI
Discourage large PRs in general. A large amount of code is hard to review even when human-written; AIs just make it easier to make large PRs, it's not a new problem. You can make it a general guideline, unless it's absolutely vital to merge an entirel massive feature in one go.
> Do not commit any Agent-specific file to the repo (such as claude.md)
I think claude.md edits are a change to review like any other, and if it were me I'd review them on their merits. Objectively, lots of projects don't need one so you can also just reject PRs to add one. If someone commits one as part of a larger PR, tell him not to bundle unrelated changes in one PR, and then reject it later.

File: 1764560186805131.jpg (52 KB, 570x540)

52 KB JPG

>>108768100
ty

File: enjoy your fungus water.jpg (71 KB, 809x717)

71 KB JPG

>>108767991
>having a job

>>108754752
The "minted" documents are PDF Documents with a few images but mostly text and tables, I mean it would be theoretically possible with git lfs but training the users to properly use git would be impossible.

>>108755458
>>108765994
>just have a separate database/data warehouse for the historical data, don't complicate your live database over this
>I'd just treat it as two separate problems
I thought about that as well, that's basically what I meant with shadow tables.

>This could just mean copying the document on every change and saving the old version as a new historical record, or it could mean delta records or someshit, or if your documents are large files then chunk-based copy-on-write might be worth it
They're PDF Documents, but I don't store the minted Documents, I only store the data. The minted PDF are stored by the users on a separate network drive.
>If you can just tell the government to pay for more hard drives
>CoW
Can't do that and I don't have any say about the hardware / OS, but I'll ask my colleagues
>or if there's only ever going to be a TB or two of data
A year of Data is wort about 1GB, maybe 10GB if we scale this to everyone.
>You can also implement CoW yourself for your custom DB
Way too complex for what I'm doing lmao.
>Then for 2, you just need a record management system that can make targeted edits to historical versions.
I thought about normalizing the db to the highest degree and just working with joins that cannot be changed for historic data. Would take some effort, but it would be a one and done thing I think.
>if a customer changes their name, shouldn't orders also be updated to the new name?
Documents are not allowed to change at all once they've been brought into circulation unless you officially cancel and re-issue with a paper trail, otherwise you're committing a felony here.

>>108768466
Store only the data and generate the PDFs on the front-end?

>fullscreen an OpenGL application in Windows
>black screen for several seconds, HDR breaks permanently in the entire OS until reboot
>use a borderless window the exact size of the screen in Windows
>same thing
>use a borderless window 1 pixel larger than than the screen
>everything is fine
Fuck this OS

>>108768775
erm ... that's ... that's literally what I'm doing? The PDF Documents are not saved on my end, they're generated in situ from data and downloaded straight to the client (webapp running in a browser).
The problem is that you sometimes you need to re-mint a PDF from historic data because someone somehow lost the file or deleted it by accident, which has happened more than once before.
If this scenario happens:
>operator creates data
>operator mints pdf
>some of the data changes
>pdf is misplaced
>operator re-mints PDF
the PDF will now not display the accurate historic situation of the past, but it is mandated by German federal tax code that it does; be it wrong tax-percentages, names, prices, etc.
Everyone in the chain of events, including the poor sap who programmed it (me lmao), will be on the hook for federal prison.

>>108768875
Event sourcing

>>108768903
I'll read up on it, thanks.

>>108768875
if the policy only requires storing documents, just archive the minted PDFs and don't bother with the data used to create them if that is not also required - it'll only be an extra burden to prove that your process of re-creating the documents from archival data is consistent

>>108768944
So basically just say fuck this and create another hard copy on my end that cannot be fucked with?
I'll have a talk with my supervisor about that tomorrow.

>>108767794
for a keyword that is to be avoided in Python at all cost you sure have many for in there!

Haskell is basically Python but fast

>>108769002
Python is awful. Haskell is a toy.

>>108769024
>Haskell is fun
True!

>>108768971
>create another hard copy on my end that cannot be fucked with
exactly
that way you don't have to maintain the process of transforming data that's consistent with how you create "live" data, which you would need to maintain the current and all future iterations of, to be able to recreate the correct document for the period
if you need some additional data-specific features, for example searching and indexing, build a database specifically for such features but tie the results to an already baked document. also consider looking into already existing solutions for such features, eg. a dedicated PDF indexing service/application

le epic fast inverse square root in quake xD

>>108769090
le le ironic xD post xD

>>108765398
It's the best multipurpose language and the easiest to work with.

Am I stupid for considering sticking to Odin instead of both Go and/or Rust?
I mean, I can't get a job either way so I might as well just use an interesting language, no?

>>108769037
I agree that it is fun, I'm just saying it doesn't really have applications beyond being fun. It is an academic toy.

>>108769442
>Go
>Rust
These are downgraded versions of C.

File: iamfreee.png (1.8 MB, 1297x1212)

1.8 MB PNG

>>108767823
>based maidposter
>>108767843
>>108767897
>cringe maidposter
>>108767917
>solve problems for fun while posting cute maids
all true!

>decide to learn a bit about sepples
>classes in header files must contain private variables and functions
What the everloving fuck am i reading. Why are they like this. Ive read its supposed to help to know the size of the glorified struct they call class, but still. Forcing to recompile something just because you added a private function must suck.

>>108769442
This is a tough one. If you have no hope it doesn't matter. But if you have a little hope its worth switching. Normies (eg hr, recruiters, managers, old devs) are just going to be like "whats that?" or not care at all... which is kind of worse when you're talking about stuff.

>>108769442
just do whatever makes you happier

>>108769442
Msg
Just use both Odin and Go. They are very similar languages.

>>108769570
Yeah I've been progressing a lot on Go, actually, so I don't feel completely bad. Right now what I feel is more like an itch of switching to Odin and not full "guilt" for choosing Odin over Go. But now I also kinda feel like Rust is a middle ground? I don't know.

>>108769556
Thanks. That could be Odin, at least for now, maybe.

>>108769529
I have a little hope, yeah. And, I mean, maybe I should have some hope still. So yeah, considering Rust as some form of middle ground between niche and employability. Not sure.

>>108769521
.h only with inlined functions, PCHs to reduce compilation time.