>>108861159
shitty slop OP

File: heart hands.jpg (267 KB, 850x850)

267 KB JPG

>>108861167
You're welcome to make better language maids if you want. There is always room for more maids in these threads <3

>arguing about maids
>when we could be arguing about better languages
Y'all are literal low IQ retards whom no one would miss.

>>108861190
>Y'all
You are as bad as them

>>108861159
Nocoder here, which maid should i start with?

>>108861216
You mean, you're a cuck? I don't think it's up to you, it is probable that claude will do all of them and you will just sit and watch.

>>108861216
you're mom

>>108861206
You haven't got a fucking clue.

>>108861216
whichever one most immediately supports your interests, which none of us know about because you didn't share them. Programming is first of all a practical goal-oriented activity, like carpentry. Even as an aimless intellectual pursuit, you'll want a connection to other interests or you won't have anything to do and won't end up doing anything but slowly forgetting the knowledge that you never seriously applied, like what's destined to happen to most of your schooling.
anyway the answer is OCaml.

>maidnigger immediately races to hijack yet another thread and force his obnoxiousness on everyone

>>108861253
weird way of spelling Haskell

>>108861216
really depends on what you want to do

but most devs end up knowing at least the basics of SQL, Bash and JS simply because they're everywhere

>>108861159
>She trusts you completely and expects competence
WTF I love C now

>>108861253
>>108861253
Nothing precise comes to mind but scripting and making websites.

>>108861270
a lazy language? How about a gluttonous language? How about a wrathful language? How about a prideful language?
Do you sin so infrequently in your daily life that you need to compensate with a sinful programming language?
>>108861284
she also murders you if you make a mistake.
>>108861306
start with JS and https://eloquentjavascript.net/ . Avoid frameworks, and avoid server-side JS and node. Put inline script tags in HTML5 and start making immediately interesting things. Then add MithrilJS for vDOM and some narrowly useful JS libraries. Then reconsider what you want to do because JS sucks.

>>108861178
real maids are better than slop maids

>>108861363
Why would you trust clueless autists though.

>>108861159
>AI slop in the OP
God damn.

>>108861306
learn some basic html5 and css, then learn a good backend language like Go or Clojure, and use Data-Star so you can make good reactive websites with little to no JS (which is cancer).
Learn SQL or Datalog later, when you need to deal with large amounts of data.

>>108861396
I thought AI was only hated when it replaces wagies? No wagies were harmed in the making of the language maids.

>>108861532
You need to get out of your pajeet techbro bubble.

File: maid-dance.mp4 (3.49 MB, 1620x1080)

3.49 MB MP4

>>108861377
I will use real maids next time. Would you rather have 2D or 3D?

>/dpt/
>30 posts in
>not a single one about programming
????

>>108861532
>only cares about wagies
>not about users
I really really really really REALLY wish we would get rid of all autists.

File: 1779067017426586.png (324 KB, 1920x2182)

324 KB PNG

This is complete. Now to apply the same streaming logic to my password manager.

>>108861532
The same could then be said about vibe coding anything we are doing in these threads.
Just stop.

>>108861555
No users were harmed by the language maids either though?

>>108861563
>still using one-call-per-character nonsense
>still NGMI

>>108861330
>avoid server-side JS and node
Why? What else are you gonna do server-side programming with? PHP?

>>108861574
Anyone with a modicum of assembler knowledge not only knows you're lying, but also that humanity hasn't devised an appropriate way of torturing you
>and that include scaphism

>>108861576
getchar is buffer and since I setvbuf() to fully buffered putchar is also buffered.
You ignore that I am on parody speed wise with tr right there in the screen shot.
I'm making it.

>>108861595
>getchar is buffer
Doesn't.
Fucking.
Matter.

Look at the fucking machine code.

>>108861588
Seek help for your obvious mental health issues.

>>108861599
Seek help, you have a mental illness.

File: squint.jpg (38 KB, 429x410)

38 KB JPG

>>108861595
>I am on parody

>>108861595
>on parody
On parity.

>>108861552
This is a maid appreciation and discussion thread. Programming comes up because it is a primary hobby for maids.

>>108861653
I guess I'm out then.

>>108861600
>>108861604
>say the autists whose opinions literally don't fucking matter

>>108861696
You have the bad annoying kind of Asperger's.

>>108861563
>modchar function pointer array
i have not seen that before. that looks really useful.

File: 53808d5e1db7b5b339cd010d6(...).jpg (57 KB, 736x1030)

57 KB JPG

>>108861645
>>108861652
How do I delete my 4chan account?

>>108861839
echo 127.42.0.69 4chan.org >> /etc/hosts

>>108861577
>What else are you gonna do server-side programming with?
anything else.. Anything capable of standard I/O can do server-side, so you are free to pick any language, so you are free to pick an actually decent language.
If you want something mainstream, go/http is fine.
>PHP?
anything else except for PHP.

File: 1779230350919401.png (19 KB, 764x95)

19 KB PNG

AI turns a 100,000 cycle operation into a 3.75 billion cycle operation.
AI needs to be stopped.

>>108862052
Wrong. Learn what tool calling is.

>>108861159
sexo with c#-tan and typescript-tan

>>108862052
i thought most AI responses were cached

File: 1779231362378267.png (59 KB, 760x314)

59 KB PNG

>>108862055
>>108862091
You thought web apps were bad wait until we have AI apps.

>>108862055
Anon ur so dumb. To call that tool for "1+1" shit it would waste more or less the same amount of cycle operations as when it does (((reasoning))) and such.

File: cat maid.jpg (390 KB, 850x1202)

390 KB JPG

>>108861159
I made a simulator where the CPU plays the CPU at my card game, so I can test the game for balance without having to play it 500 times. Working on a program that auto-balances decks by random walking values, within constraints, and then running the simulator to see if the game got closer to balanced.

>>108862127
It's twice as much. One prompt to reason about making a tool call. A second prompt to read the tool call result and tell you what it says: 2

>>108861159
For me it's Go, JS, and Bash. I mean the maids, not the languages.

File: ocaml-getchar.png (100 KB, 864x960)

100 KB PNG

>>108861563
>arg-handling but not usage
>no validation of args, ignores excess args, ignores invalid args
>array lookup of function per byte of input despite the decision only being needed once
>avoiding magic numbers for universal constants like STDIN_FILENO, then using magic numbers in the arg handling
>treats I/O errors as EOF
>tolength handling prints negative lengths for a 2147483652-length line
>per-byte handling is so abysmally slow that printing that negative length takes THIRTY SECONDS
>tolength has an implicit return in the c=='\n' case
>-m prints \0 for every byte in addition to the line lengths
>per-byte handling prevents reddit-casing of e.g. cyrillic text
>function call per byte even to do nothing -> more than a 1000 times slower than cat on long inputs
>libc already does tty-conditional buffering, but does it better
the ^clock() is alright. It's still toy init, but it's useful.
attached: OCaml that's about 2x as slow as C on the per-byte runs, 20x faster than C on the cat run, and 4x faster than C when printing lengths, and which has dumb manual arg handling that's still sane

arg handling is actually a big deal. Sloppiness here will have you writing admin tools that turn a flubbed ctrl-R rewrite into a deleted database instead of an error message

>>108862233
bash is such a comfy language though. half the time it's complete gibberish but still works.

>>108861159
Why the slop OP?

>>108861159
JavaScript should be chained to the floor in the Google headquarters.
C++ being the strongest Waifu is accurate.
Rust should be a tranny holding a chastity cage telling the user that 'this is for you'.

I asked AI to see if I could not only remove a file from a public repo, but basically make it look like it was never added in the first place. What ended up happening was a "Merged branch" commit, and every commit before it was sort of duplicated

I ended up nuking the whole repo. It was just a barebones website to play with the "Sakila World Database".

Fuck me though

File: 1779235130359558.png (199 KB, 1392x1056)

199 KB PNG

>>108862268
ok yes my argument handling and length calculating is a meme BUTT...
I went through all the effort of writing this unreadable slop only for it to run at the exact same speed, these micro optimizations are not worth it.

>but but it makes a difference if you run terabytes of data through it
When I need to uppercase a petabyte of data I'll just use tr I promise.

>>108861330
>she also murders you if you make a mistake.
Hot

File: alexandrina.jpg (352 KB, 850x1283)

352 KB JPG

>>108861159
I need a feet enjoyer to tell me what makes good feet pics.

>>108862525
>fun lock file

>>108862318
I don't really like bash as a language desu. The syntax is needlessly cryptic. I just use it, but when I need something more complex in it, I use AI.

>>108862525
nvm I was looking at the wrong time.
It's 5x faster :|

I'm using C++
I have an array meant to represent a 3x3 grid. What's the math behind navigating each cell in this grid when the array itself is not a grid but a horizontal arrangement?

>>108862627
index = y * 3 + x
where x is the inner loop variable from 0 to 2, and y is the outer loop variable from 0 to 2

>>108862627
using linear arrays are often easier

int array[9]

>>108861576
getchar and putchar are both actually inline macros in every libc I'm aware of. They pull a byte off/push a byte into the FILE*'s buffer and increment the buffer pointer directly. They only actually perform a call if the buffer is empty/full or if you've asked for thread safe I/O.

>>108862569
I might not be competent enough, perhaps further training on this subject is required. But from what I understand, there is nothing special about this fetish besides the fact that it is a particular part of body instead of the whole thing. So mostly it's just detalisation. You need segments of the fingers to be clearly defined, thin and slim in the middle with big soft tips, also women often prefer their nails to be not very short and they might make them pointy as well, but that's optional and attracts attention a lot.
Also there clearly is a big difference between male and female feet. Academia artists spend absurd amount of time to learn it and get the anatomy right, it's not easy to just describe like that.
And perhaps worth mentioning that it is likely a female fetish originally. Due to their idea of asserting dominance.

>>108862569
perceived popularity it's a forced meme

>>108862569
Alright I shared stolen secrets of feet worshippers.
You tell me why you need them.

>>108862687
aaaah ok that makes a lot of sense and really simple. thanks anon

Why you delete my maid pictures? I was writing a story and I was coming back to continue making it.

>>108862862
Janny is non-deterministic.

>>108862788
>>108862810
I specifically need info about women's feet, to create art for an anime booba game.

>>108861563
>>108862525

You guys have bum fucked and confused me.
I am suppose to be writing this in Pure ISO Standard C not POSIX.

What is the best way to learn Kotlin?

>>108863003
Use it

>>108861330
Laziness is a virtue.
t. Larry Wall

File: crying maid.jpg (211 KB, 850x963)

211 KB JPG

>>108862129
My balancer made balance worse =(

>>108863070
if he'd put more work into designing his language it might still get use outside of legacy and bash-tier admin tasks

>>108863073
Interesting. How did it do that?

>>108863073
One time I wrote a btree optimizer that, due to a flipped test, actually turned the btrees into basically just sorted singly linked lists.

I have an encoder providing position values over gpio and I wrote a program that reads that value and then writes it to a serial device. I'm just doing a blocking read on the gpio device file and then immediately writing whatever I get to the serial device. this is fine 99% of the time, however the encoders data rate is variable based on the speed it's moving which means if it goes too fast I begin writing messages to the serial device faster than it can handle (internal buffers or something I guess?) and it starts behaving badly.

I can 'resolve' the queue in my loop as the device I'm writing to only cares about the difference, so I can process the entire queue and then write a single value, but I'm not sure if that's the best approach here since I think it introduces some stale data issues? Basically as is the serial device gets the data as soon as its generated, but if I want to resolve this rate limit I need to add a timer to my loop capped at the rate limit and then process every encoder value obtained up to it and then write. the last value I read is unlikely to have occurred right as I unsleep my loop which means I'm introducing random delay from the sampling to the device getting the value. idk if there is a way to fix that but the the behavior seems objectively worse to the immediate read/write I had before

>>108863337
Depends on your hardware of course, but in general my instinct would be to keep two variables, one storing the current position of the encoder (that updates whenever the encoder sends new data) and the other storing the position of the encoder at the last point when the serial device was informed. Then, just set up an interrupt timer (triggering at whatever the max rate your serial device can handle) that periodically subtracts current from previous, writes current to previous, then sends the result of the subtraction if its not zero.

>>108863421
I've never set up an interrupt timer before, how does that work? I think the equivalent of what you're suggesting is just 2 threads, one reading at a variable rate and the other writing at a fixed rate, right? Isn't that the same as what I suggested as a solution however? main difference would be my suggestion does the processing of the reads when the loop fixed rate write loop unlocks and yours would be doing it beforehand

I would still like to try and get things to how they are in the immediate mode I have where you write data as soon as you get it (if possible) so that as the speed changes the latency is minimized. The problem scenarios I envision are things like no write occurring on the previous cycle because the incoming data was too slow, but suddenly it picks up and you're unable to write because you're waiting for the next timer event despite no write occurring last cycle because of no diff.

>>108863484
actually thinking about it, VRR monitors probably have to solve this same problem huh? when the frame rate is low they need to sync up with it and when it's higher than what they can handle they need batch frames. wonder how they do it

>>108863484
Threads are almost certainly easier to use if they're available to you, just my brain hears encoder and starts thinking in embedded. :P

As for latency, it depends on your specific application but generally if you're measuring the position of an encoder it's changing in like... human-scale time, not computer-scale time, if that makes sense. Even if your serial device is really slow I wouldn't think it's likely an occasional skipped update would matter.

>>108863337
Why can't you use an interrupt timer to measure it? Just write a function to read it every x clock cycles

>>108863484
To actually answer how timers work, on POSIX you use timer_create() and timer_settime() to set it up then it sends your program a signal periodically which you can catch and do whatever you want with. It's pretty complicated but the man pages explain it in depth. On bare metal you'd use whatever hardware interrupt facility your platform of choice has (the PID on x86, most embedded platforms have similar functions available too), which causes your program to jump into the interrupt handler every X microseconds with some kind of return pointer on the stack to take you back to where you were when you're done.

>>108863559
>PID
*PIT

>>108861159
Where is Lisp, slopmaid?

File: 1779252165253603.png (571 KB, 1920x3738)

571 KB PNG

I have sped this up dramatically by chunking the input.
Violated DRY but what ever saves a function call while streaming.
Replaced rand() and modulo with a simpler operation.
handled argv properly.
Separated streaming vs interactive modes.
Length is still fucked even if I replace printf with a custom integer printer I perhaps need to look at chunking the output.

>>108863762
>Length is still fucked

#include <inttypes.h>

printf("%" PRIu64 "\n", length);

>>108861744
>>108840546

>>108862787
Fascinating heacanon: https://godbolt.org/z/MdY8x75WT

"foo":
        mov     rsi, QWORD PTR "stdout"[rip]
        movsx   edi, dil
        jmp     "putc"

File: memcpy2.png (46 KB, 1630x720)

46 KB PNG

>>108863762
>I have sped this up dramatically by chunking the input.
Yeah, it's almost as if I was right the last two threads.
Although, if you want to go hog you may want to CPUID your L2 cache so that you can determine the optimal chunk size at runtime.

File: 1779254086043158.png (101 KB, 1920x575)

101 KB PNG

>>108863796
Length is being line buffered which is slowing it down.

File: 6CACC0BA-648F-4369-968A-7(...).png (2 MB, 1085x1450)

2 MB PNG

>>108863626
here's what she came out as

>>108863884
she appears to have two left feet

>>108863896
It is AI slop after all.

>>108863814
>if you've asked for thread safe I/O
use putchar_unlocked().

>>108863920
Nice, now all I need is to actually HAVE it on my system (Windows).

>>108863960
https://learn.microsoft.com/en-us/cpp/c-runtime-library/reference/putchar-nolock-putwchar-nolock?view=msvc-170
I make no guarantees for how Microsoft actually implements it but they do seem to have an equivalent.

>>108863973

call    __acrt_iob_func
...
call    _putc_nolock

No surprises here though. Inlining essentially removes the ability to change the ABI later on. Honestly surprising that glibc is willing to solidify its interfaces in such a way.

>>108863995
BSD libc inlines it too. I'm surprised Microsoft doesn't since it's one of the places the C standard explicitly permits (and arguably encourages) it. musl doesn't (under the argument they optimize for space not speed) and I had thought that that was the only one.

>>108864065
MS "appreciates" the ability to change things under the hood constantly, even if it makes no fucking sense whatsoever. Anyone still remember when their memory management retards hallucinated that hugepage requests on Windows weren't already explicit enough, so they made the kernel just fail if there wasn't enough memory? https://devblogs.microsoft.com/oldnewthing/20110128-00/?p=11643

Kinda strange how both Windows and Linux manage to fuck things up in completely different ways.

File: 1779260189089557.png (184 KB, 1920x1055)

184 KB PNG

Astronomical speedup on length.

File: beating_printf.png (40 KB, 1020x700)

40 KB PNG

>>108864171
And snprintf isn't that fast to begin with either.
Imagine what you could do with SIMD operations.

>>108864171
yes|head -2050|./your-program -m

>>108864221
oh I missed the numbers check
why wouldn't you test against the length of the buffer?

File: 1779261402802823.png (8 KB, 692x50)

8 KB PNG

>>108864227
I'm already testing with 123985 lines.
I did have a check output[4063] != 0 but it was actually slower.

File: 1700643147824272.png (96 KB, 1465x1007)

96 KB PNG

>>108864237
>archlinux
Immediate auraloss.

>>108864267
That image you posted is older than you are.

File: 1779261869792177.png (13 KB, 291x214)

13 KB PNG

idk why this would be slower, it should be less writes.

>>108864267
I used archinstall.
Debian won't let me use admin as my user name and takes long to boot.

File: the_autist_doesnt_underst(...).png (6 KB, 680x100)

6 KB PNG

>>108864273
I severely doubt that.

>>108864281
I'm pretty sure that image is from the early 2010s, and saying "auraloss", surely you're like 12 or something.

>>108864288
Yeah, because the internet *totally* doesn't allow millennials to pick up on alphaspeak, you fucking retarded idiot.

>>108864297
no skibbidi cap 6 7

>Try to find a job
>Oh this job description sounds nice
>Try to apply
>[ ] I agree that the documents I have uploaded will be analyzed by generative AI software.
fuck you stupid fucks.
This fucking fuck triggered me so hard that I sent them an email asking if they are serious about that and if they are searching for a good employee or someone that just gets the highscore on one equation.
I will seriously apply at aldi very very soon

>>108864311
PS: Not possible to upload without checking the box. It's mandatory

>>108864275
I would need to see the actual instructions to perform a diagnosis. Yes, I am too drunk to care to replicate it for myself.
>again
Listen here you little shit.

>>108864309
Everything has its time and place.

>>108864315
>he doesn't know how to circumvent the check
ngmi

>>108864311
try a little prompt injection in 1pt white font

File: programming萌え擬人化メイド.jpg (193 KB, 1672x941)

193 KB JPG

>>108861159
only the Assembly maid turned out well imo

>>108864349
i was considering to add something like that in general to my applications, but i am too lazy to come up with something that is not too edgy

>>108864377
guess why though

>>108864171
Nearly all that time will actually be the tty, not your code. Try redirecting the output to a file, you'll see.

>>108864317
>>108864511
I actually think puts grinds to a halt when you pass a really large string to it.

>>108864528
When you're outputting to a tty you're not sending data to a neatly paged file or pipe, but to an emulated (or possibly real, who knows) 38400 baud serial link. There's all kinds of nasty legacy garbage going on there that will take up time and skew your benchmarks. Not to mention whatever the terminal on the other end is actually doing.

File: 177926755482350.png (98 KB, 769x782)

98 KB PNG

>>108864585
I have started outputting to a file, I am currently being bottle necked by sprintf() for -m. I will replace it and then it should be on PARITY.

using zig for the first time

really funny how people will say advertise their language as "simple" and then come out with monstrosities like zig, which looks to me more bloated than C++

C is a simple language. lisp is a simple language. you can learn all the syntax in an hour and understand most code fairly quickly

ziggers call their language simple and decide to bloat it with all that ceremony and decorations

at least rust doesn't call itself simple.

>>108864743
Rust was an attempt to solve several very specific and well defined problems, intended mostly for system level dev. What is zig for? It's a "better C". Also it's dev like red color more than orange. It's his favorite type of crayons.

File: 1779273144931338.png (176 KB, 1920x1055)

176 KB PNG

I removed the strcat().
I'm so fucking close but I can't beat this sprintf. It's the modulo required to convert an int to a string.

>>108861216
Started Python last week because I was in your shoes and realized I wanted to learn by copying code for old classical games like Pong or Frog I'd source in printed books or magazine and Python has a goldmine of that.

It's an high level language (closer to talking in English than talking to a machine) so really easy to write and read, plus there's like a dozen of books written for kids so you won't be filtered. I browsed them all and I'd say go for the female authors first: most of them are teachers and understand you're a nocoder that will need real world analogies to get your head around it. Males tends to think coding is an excuse for doing more maths and some will deep dive into that starting from chapter 3, while you haven't seen much more than installing Python and typing print('Hello World').

Plus you won't waste your time because Python is still relevant today when it's time to automate stuff and sort data. It's also behind Raspberry Pi so you can start some engineering projects once you're settled.

>>108865032
you only care about /usr/share/dict/words which has a very small domain of ranges
put that entire domain in an array in your code

the real answer to anyone asking what language they should learn is

1 - start with C if you want to learn fundamentals or Python if you want to build something quickly. anything else is just 4chan contrarianism
2 - your third language MUST haskell or ocaml or ML. there's a reason most elite universities force you to learn functional programming early on, even if you never use them again
3 - SQL no excuses
4 - only AFTER the above, if you absolutely must, try Java, but know you are likely going to develop the OOP brain diseases if you stick with it for too long, and stay away from design patterns. if someone is talking about the "factory pattern" kill them. (alternatively you could try Kotlin instead for less brain damage.)
5 - next choose whatever language fits your domain, there are no more requirements. typescript/javascript for frontend, C++/Rust for low level programming, go for backends, or whatever will get you a job. it doesn't matter.

t. 38 year old who wishes he had done the above instead of wasting 20 years on C# and Java shite before learning programming properly.

>>108864377
>Fish
Isn't fish a bash replacement scrpit? I get that Bash in there there but you don't see a lot of info of what you can replace Bash with.

>>108865474
1 - it doesn't fucking matter which way you go. Just educate yourself and gather needed knowledge for your goals
FTFY

File: Screenshot_20260519_105719.png (17 KB, 636x600)

17 KB PNG

working on a gameboy emulator

>>108865525
>>108865593
these lukewarm answers get you frustrated looks rather than grateful looks, because the original question is only ever asked from a position of such ignorance that the questioner doesn't even understand what's possible and what isn't. It's a question like "where should I go?" from a guy that hasn't heard that we have practical FTL space travel already, or that doesn't know that other planets even exist, or who is communicating somehow from a damp cave that he thinks is the whole world.

Pick a language based on if you can make a funny logo for it.

>>108865638
Yeah it's like when you're trying to buy a beginner piano and redditors say 'just try them out and see what you like! ;)" as though someone is supposed to even understand what they like before being able to play and having built preferences.

>>108865579
Better than mine.

>>108865593
no, the real answer is C and has always been C
even if you don't stick with it, you start with C

>>108865680
>What I want to make in the first place?
asked: >>108861253
answered: >>108861306
goal-oriented response: >>108861330
do you see the difference between that exchange and "just pick the tool that's most appropriate despite you not knowing what the tools even do"?

>>108865748
>advising node ever
now you're dispensing bad advice because of technical feasibility
hey, you can hammer nails in with a screwdriver, by holding it by the opposite end and smashing the handle against the nail
>>108865785
nobody hammers in nails anyway. You're supposed to hire Mexicans to do that. Start with high-level carpentry, it pays better.

>>108865638
The point is: It doesn't have to be C.
C is good. If you leave out UB, then C is easy to learn.
But it's not needed.
You need to understand (and apply) the fundamentals. Almost any language can teach you those.
I dont really understand how maths works if i can solve a quadratic formula that is already normalized

>>108865824
you are applying a constraint of "learn only one language", which should be obviously discarded because JS is necessary for the one task while being cancerously unsuitable for the other task. This is as bad an answer as "Nim, because it can also compile to JS"

>>108865804
>>advising node ever
better recommend bun, because it's so stable, eh?

>>108865876
the immediate topic is useful answers to questions like that. The useful answer to "what jelly-like substance should I use to flavor my donuts and also prevent my bike tires from going flat?" is not "this bottle of green slime I guess", but actually "use this slime for your tires only. Put something normal on your donuts, holy shit, something like strawberry jam, what psycho said that other thing?"
if asking you a question is no better than asking it of a constraint solver, or asking it of an AI exegesis of reddit, then what are you even bothering to answer for? Wait a bit and the guy will ask AI anyway.

>>108865933
no, JS sucks at one of those tasks, and so obviously so that you should whine less about your advice getting called 'lukewarm', and worry more about professional retaliation from people that take your advice and come to hate you for the results.

>>108866013
those are called light switches. they go on the wall.

>>108866041
no, they're purely mechanical devices that rely on electrical contacts. zero computers involved.

File: file.png (2.57 MB, 1661x943)

2.57 MB PNG

>>108861159
ftfy

>>108866081
>ruining perfectly good art by surrounding it with AI slop

File: 1776379585700354.webm (3.78 MB, 1080x1080)

3.78 MB WEBM

>>108865032
>I'm so fucking close but I can't beat this sprintf. It's the modulo required to convert an int to a string.

for (int n = length, i = 0; n > 0; n /= 10, ++i) {
    outputIndex[i] = n % 10 + '0';
}

You'll need to reverse outputIndex or length, whichever is faster

>>108861159
how do you guys approach optimization? i'm writing a compiler in C and replying on simple pointer-based data structures which are not cache friendly. do i start changing my data structures and running benchmarks now even if the compiler has barely any functionality? do you treat optimization like a final pass after most of the functionality has been written?

>>108866121
Treat optimization as a continuous engineering constraint, not a final “cleanup phase” but also not as the primary goal while functionality is immature.

i can guarantee you, like i would bet everything i have on this, that every single person talking about "performance", say "x is slow, use y", is a posturing little shit who has never actually had to deal with real bottlenecks and thinks faster means better

i can also guarantee that for 90% of people even java is fast enough for you, you little shits

>>108866135
>>108866203
i will focus on using algorithms/structures that have the correct time complexity, not leaking memory and producing the correct output for now. i will kick the can on further optimzation. thanks for the input.

>>108866224
>90% of people even java is fast enough for you
of course, almost all noticeable issues with software speed has to do with bad algorithms or other bad practices, not the language/platform doing it

>>108866121
Rewrite the same code 20 times

File: lispy.png (2.47 MB, 1122x1402)

2.47 MB PNG

>>108863896
>she appears to have two left feet
here I fixed her. she still has two left feet. but now she's an alien so it makes sense.

File: akane.jpg (475 KB, 850x1518)

475 KB JPG

>>108863240
Overfitting. It went from matchups like (W-L-D) 15-1-3 to 2-1-17. Nobody is going to like a game where there are ties 85% of the time, but nobody is going to like a game where you have a 95% loss rate against a specific opponent. I need a way to make it balanced and decisive as much as possible.

I am thinking I might just make the CPU worse. Or make difficulty adjustable. There is no reason to make it hard for a casual player who wants to see and unlock a bunch of fanservice art.

File: 1778708637696964.jpg (441 KB, 1279x1950)

441 KB JPG

>>108866224
t. went out in Q1

>>108866081
>saar rust bad it is le tranny lang saaaaaar
go back ranjeet

>>108866486
seethe tranny kek... your lang is only used in crypto scams and every big project where your tranny lang was used got into problems.

>>108861216
Get a taste of coding with python for a few months, whip up some projects that have to do with your other interests. If you like it get with C++ or something else. Starting with C will overwhelm you with theory without utility, JS is more syntax heavy than python, SQL is for databases and if you are nocoder you will have no use for databases.

File: cutemaid.jpg (436 KB, 850x906)

436 KB JPG

>>108861216
>Nocoder here, which maid should i start with?
Whichever one you think is the cutest! <3

>>108865474
>second
>not mentioning assembly
OK, 38 year old autist.

>>108866679
not that non, but assembly is pretty self explaining. There is not THAT much need to learn it in depth. It can be fun, but it's not needed
While learning C you can just disassemble your shit and understand what it means.

File: format.png (148 KB, 1894x966)

148 KB PNG

>>108866115
>no PBCD conversion into SIMD registers to avoid memory accesses
NGMI.

>>108866224
I can't wait for your registry dumper that beats my 7 seconds.

File: compilers_are_garbage_sta(...).png (300 KB, 1920x8440)

300 KB PNG

>>108866701
>There is not THAT much need to learn it in depth
Riiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiight.

File: sakuya shrug.jpg (228 KB, 850x1247)

228 KB JPG

>>108866679
>obsessed with assembly
>lives on 4Chan
>calls other people "autist"
You're in denial about your own neurodivergent status, and seem to be projecting a great deal of self-hatred onto others.

>>108866733
Nah, I'm pretty open about being ADHD.
It's just not autism.

>>108866742
None of the described behaviors are attributable to ADHD. Nobody with poor focus is playing with assembly.

>>108866742
Oh, also I'm not really obsessed with it. I don't *like* it, and I wish it wasn't required - but unfortunately compilers haven't been able to produce proper code for decades. If you gave me a language that wasn't married to fucking ABIs and knew how to generate SIMD instructions properly I'd leave assembly and C immediately.

Or, in short: my focus is ALWAYS on the user.

>>108866772
>>108866774

File: lispy2.png (1.43 MB, 816x1024)

1.43 MB PNG

>>108866336

>>108866709
>I can't wait for your registry dumper that beats my 7 seconds.
would it matter if it took 10 seconds? you proved his point

>>108866847
Actually yes. The reason I needed a fast one was that there was a very small time window after boot in which the OS did things that I had to track, so not only did every second matter - it's also proof that no one, not even Microsoft's own Regedit, knows how to work the registry properly.
>except me

Nandgame to learn how the CPU works.
Vulkan triangle to learn how the GPU works.
Simple as.

>>108866871
>Vulkan triangle
>still no vkCreateImage[s] call

>>108866713
okay maybe my wording was stupid.
95% of the people don't need to know the specifics of that, because 5% do all the optimizations that are critical.
But everyone should be aware of what, in general, happens under the hood and be able to make sense of most assembly instructions.

>>108866884
>because 5% do all the optimizations that are critical
Bro, I still remember when SSDs provided massive performance improvements.
>inb4 OK and
NVMes didn't.
>inb4 OK and
I/O batching has been around since 1989 for servers and 2004 for consumers.

Kinda wish we could dig up the 5% and torture them posthumously.

>>108866921
ok you can stay delusional, if you want to, but you don't have to.
We would live in a better world if people cared more about performance and less about ultra slow javascript animations, but that's just not reality. We gotta find a middleground that is acceptable

>went through the Godot tutorial as a beginner programmer
>started making a basic game while asking ChatGPT for assistance rather than going through documentation or Google
>decide to download Codex to see if it would help
>managing to do everything 100x faster than it would take me otherwise

Is there any catch at all to this? As long as I ask for little steps at a time and test things manually, I'm not seeing any downside, apart from a lack of learning. But writing code manually is maybe finished as a real skill.

File: hard_data.png (128 KB, 1911x989)

128 KB PNG

>>108866952
>autist calls me delusional
>when I have actually seen with my two own eyes where most bottlenecks are
And that is why normal, well-adjusted people hate autists and want to see them being killed on national TV.

Not murdered.
Murder is reserved for human beings, and autists are not.

SNMP is the worst protocol ever

Let's add some special snowflake encoding rule just for the first two numbers of an OID so we save one byte (those numbers could just be dropped because they're completely useless anyway)
Meanwhile they waste 6 bytes in every single SNMP packet just to store a value that can only range from 0-3 with their retarded TLV shit.

>>108866999
>And that is why normal, well-adjusted people hate autists and want to see them being killed on national TV.
I don't think normal, well-adjusted people want to see anyone being killed on TV? Having violent revenge fantasies against image board posters you dislike not well-adjusted behavior.

>>108866956
yeah everyone else has this feeling after about 5s with AI. But all programming mistakes are felt over time, and especially after maintaining software over time. Like the guy who thinks JS is good for scripting because he controlled some smart lights with node once, you can get some immediate use and be impressed by it. But one day (5 days for a nocoder; 5 weeks to 5 months for someone who can actually program and get use out of an agent) you'll realize you have a pile of shit and that very small changes require a strangely large amount of work and that the results are strangely very bad.
You're not going to believe me anyway. Just enjoy yourself for as long as you can and see what you can do. Your attitude now is not actually that different from millennial RAD programmers who "wrote one to throw away"

>>108866336
>making the LISP horse dog sized
Can you make it horse sized and have her riding the horse?

>>108867065
Little bit old, but still: https://www.nbcnews.com/id/wbna4353934

>>108867086
>implying most Americans are well-adjusted
They literally have a culture designed to make them stupid and aggressive.

>>108867106
>implying most Americans are well-adjusted
For the American society, yes. That's how that works.

>>108866999
I don't use windows, but good numbers tho. checked

>>108861955
>Anything capable of standard I/O can do server-side
Yeah, but Node's event loop is probably the best general purpose, out of the box async model out there. With V8, obviously, it compiles down pretty efficiently to machine code. Yes, you don't get the absolute greatest efficiency of resources, and in some applications, that's really important. But given how much of the contemporary web is distributed, TypeScript is very much a viable path towards working in basically any environment.

>>108867121
strace.

>>108861955
>>108867130
Also, PHP isn't actually terrible these days. It's got a lot of modern language constructs, even if it's still unwieldy and inefficient as is any genuinely interpreted language. A lot of the web still runs on it, and the combination of TS/PHP makes you theoretically employable to huge amounts of employers.

behold: a generic data structure operation in C

#define VECTOR_PUSH(v,item,type) \
    type *array = (type *)((v)->array); \
    array[(v)->size] = (item); \
    ++((v)->size); \
    if((v)->size == (v)->capacity){ \
        (v)->capacity *= 2; \
        array = realloc(array,sizeof(type)*(v)->capacity); \
    }       

>>108867206
>realloc
Y-i-k-e-s.

>>108867130
you're dying of cholera while praising the quality of your plumbing, because you're familiar with rivers of shit (callback hell) and bone-dry deserts (GILs). But literally any civilized water infrastructure is better than what you have. The fucking Romans and their deadlock-prone manual thread management can look down on you in pity and disgust.

>>108867212
something's clearly wrong because it leaks a big chunk of memory. back to the drawing board.

>>108867276
That's not the only problem though: >>108855970

File: VID_20260520_175949922-[0(...).webm (1.05 MB, 480x854)

1.05 MB WEBM

>>108861159
my work should let me ship this

>>108867206
mmap a lot of memory and divide that between all the arena allocators you'll need (temp, global, thread).

>>108867349
>mmap a lot of memory
You mean "Virtual Address Space".
The memory commitment comes later.

never mind, my vector macro thing with realloc works, i just forgot to free the vector.

>>108867349
i'll get into custom allocators someday. i'll wrap all of my mallocs in case i decide to to use one.

>>108867206
There's a memory leak in there, retard

>>108867410
>i'll wrap all of my mallocs
Too late at that point: >>108854215

>>108867421
nuh-uh, valgrind says zero errors

>>108867431
What if realloc fails? It returns a NULL pointer without freeing the existent allocation.

>>108867294
this could change the world

>>108867451
i thought of that, i have a wrapper called "xrealloc" that prints an error message and terminates, i just didn't use it in my code example for some reason.

void *xrealloc(void *ptr, size_t sz){
    void *new_ptr = realloc(ptr, sz);
    if(ptr == NULL & sz > 0){
        const char *err_msg = "Error: out of memory\n";
        write(STDERR,err_msg,sizeof(err_msg));
        exit(1);
    }
    return new_ptr;
}

>>108867466
OK, but you're *still* using realloc.

Which, if the bucket cannot be resized in place, WILL cause a new block to be allocated at the same time the old block still holds data.

Meaning that the chances of you running out of memory are much higher than if you made sure that nothing can write into addresses reserved for the growth of your structure - i.e. VAS reservation through mmap/VirtualAlloc.

>>108867431
>nuh-uh, valgrind says zero errors
Wow, retard, calm the fuck down.
Valgrind doesn't simulate OOM. It's a dynamic analysis tool, it can't detect bugs in code paths that aren't executed. It only checks execution paths that actually run.
Look:

array = realloc(array, ...);

If

realloc

fails, it returns NULL. Say goodbye to "muh pointer". Memory leak.

realloc

just works unless your PC is shit.
Try requesting

SIZE_MAX

bytes / using

LD_PRELOAD

and see if Valgrind cries foul.
Also, if

realloc

actually moves the memory block, you never update

(v)->array

, so it's left pointing to freed memory.
Nice showoff, you dumbass retard.

>>108867498
i noticed the issue where i didn't update (v)->array and fixed it. also see what i posted above. i will trying the SIZE_MAX and LD_PRELOAD.

>>108867497
i'll look into mmap/VirtualAlloc

>>108867466
>

if(ptr == NULL ...)

You still let it compare with the old pointer instead of the new one. If

realloc

fails, ptr is not NULL, if evaluates to false.
So

*xrealloc

just returns

new_ptr

.
>

ptr == NULL & sz > 0

Use the logical AND, not the bitwise AND, dumb fuck
>

sizeof(err_msg)

The fuck are you even doing?

err_msg

returns the size of the pointer, not the string length.
>

write(STDERR, err_msg, sizeof(err_msg));

Standard POSIX uses STDERR_FILENO, not STDERR
>>108867535
>also see what i posted above
I saw it and it's shit

>>108867561
i tried it with SIZE_MAX and it exits but prints only "Error: o" for some reason instead of the full error message. Valgrind doesn't appear to care either way.

>>108867561
you should save the anger for vibecoders, this guy is just doing his own thing and learning

>>108867583
I warned you so, kiddo

>>108867359
Why are you mad about an image of a green guy, and what does Barney have to do with the green guy?

>>108866803
damn NOW THOSE are some heavily nested parens

File: 1748623741526323.png (250 KB, 1788x1372)

250 KB PNG

anyone done the 1 billion row challenge?

https://github.com/gunnarmorling/1brc

apparently java sisters are winning

my current naive implementation is taking 6 seconds lmao

>>108867597
Well even if that cocksucker's learning, if there's one thing I can't stand, it's that he acts like a smug retard

>>108867651
as a challenge it is not that interesting but I did still enjoy the results. I really liked this guy's C# attempt/explanation
https://hotforknowledge.com/2024/01/13/1brc-in-dotnet-among-fastest-on-linux-my-optimization-journey/
it was a bit surprising how fast C# can be if you just ignore the standard stuff he

>>108867712
this is another one that didn't follow the spec, same as the rust guy who said he beat the java implementation

would be nice to see how everyone compared following the same rules

>>108867233
I prefer an environment and language built around callbacks, personally -- especially in webdev where you can assume that everything will end up returning a Response. The most satisfying (to me) projects are those engineered in sufficient detail such that every option becomes processing responses -- incoming and outgoing -- along their lifecycles, creating side effects, maybe changing some state, and then functionally composing your response.

It's not the same kind of satisfaction as a tight, brutal imperative algorithm, but it's got its own kind of beauty.

>>108867932
Literal autism.

>>108867932
>along their lifecycles, creating side effects, maybe changing some state,
OO callback hell

>>108867953
Nonlocality can be difficult to reason about, but OOP is almost always avoidable.

File: 1704893622857209.jpg (135 KB, 1080x1304)

135 KB JPG

My python script is coming along well, AI has been assisting me in computing the 2^1 millionth fibonacci nunbmbers

got trolled hard by vscode for like an hour. it decided to "shadow" one of my source files with a stale version for no apparent reason. tried deleting my cmake cache and all manner of other shit to not avail.

>>108868005

a = b = 1
for _ in range(2**1_000_000 - 2):
  a, b = b, a + b
print(b)

>>108868093
DUDE you have no idea how much I hate VScode for that garbage. I use it at work because its the only fucking thing they allow and its just garbage when it comes to that shit. If AI makes a change or you revert something with git or you have two instances of vscode EVERYTHING goes to shit. It's such a pile of trash.

am I right in thinking that C + sanitizers + fuzzing covers all of Rust's garauntees except data races and exhaustive match statements. Also are custom allocators a meme, can I just use malloc and free like a caveman

>>108868136
>>108854215

File: clara.jpg (297 KB, 850x1251)

297 KB JPG

>>108866358
The game is now balanced. Some matches are lopsided on "Insane" difficulty, but "Insane" is supposed to be hard, so this is good enough to call it balanced. No match is unwinnable.

>>108867932
this is the worst possible thing

https://journal.stuffwithstuff.com/2015/02/01/what-color-is-your-function/

especially in web development, where a client needs a response immediately and async in mostly pointless

>>108868136
i am really thirsty rn

>>108868123
uh you can just run neovim from normal user account. no need for root

File: 86a2769fc8613666fc2674d90(...).jpg (45 KB, 860x630)

45 KB JPG

>>108868136

>>108868785
>uh you can just run neovim from normal user account. no need for root
I'm on a windows machine. They locked down installing everything. Even Choco and NPM are locked. I hope I never have to take a windows job ever again.

File: 18108.jpg (184 KB, 1069x1312)

184 KB JPG

>>108861159
R011ing. 1-13 left to right and up to down. Skipping 11 (dubs). So last two are --12 and --13

Dubs = any two
Trips = any three

>>108868834
windows is always a red flag, i've been there
>manager that is tech illiterate
>laptop comes bloated with antivirus, one note and other microsoft crap
>remote access, screen reading is always accessible by admin
>expected to use shit like ftp to access files

just stay away. current company they gave me a laptop and the only requirement is using 2fa to access services

File: 1663895259162040.jpg (143 KB, 1024x1020)

143 KB JPG

>>108866774
>no response

>>108866115
Thanks, this is a really clean implementation.

>>108866224
It's a reaction to the enshitification of software.
Ranger, VS Code and kitty are slow pieces of shit.
lf, yazi, vim and foot run fast as fuck.

Just use them and you can literally feel the difference.

>>108866999
What am I looking at here?

>>108869206
He's our resident schizophrenic, he screams about direct CPU instructions daily. I don't think he's spoken to a human IRL in 20 years.

>>108869206
Partial log of the startup process of Factorio.

>>108869225
And it is opening several files?

>>108869248
Hundreds, if not thousands (especially if you use mods). A lot of it could be done in parallel - SATA HHDs have been out since 2004 - but kernel interfaces are too shit for that, both on Windows and Linux, by the way.

https://en.wikipedia.org/wiki/Tagged_Command_Queuing
https://en.wikipedia.org/wiki/Native_Command_Queuing

>and that's ignoring the NtReadFile calls that *copy* data in 4096 byte chunks from said files, rather than opening file mappings (allows using the page cache directly without having to copy shit)

>>108861159
>Some guy trashes my moderately popular open source project online
>Announces his own alternative
>Refuses to work together
>Thinks my approach is unsalvageable
>I have tried his approach and I know the exact optimization problems he will run into
>He doesn't listen
>Yoink his planned features before he can implement them
>Fix all his complaints that were legitimate
>A few months go by
>He says optimization is harder than expected and he should be done when he has a chance
>A couple years go by
>I wonder what happened to him
>No updates since then
>Now posts about his life falling apart on his main account
>His health, his love life, job, all trashed beyond saving
Jesus Christ, I don't know how to feel about this. I wasn't happy he was spreading wrong ideas about my software but he didn't deserve all that. The competition was good.

>>108869309
i didnt read your text, but your code is only limited by your imagination. An LLM is limited by compute power and your dollars

>>108869270
Is there anything to indicate that the requests are being done serially?
Unless the logging will combine them into single entries, even parallel tasks might be logged serially.

>>108869312
Not sure what that has to do with my post. Maybe you should actually read it next time so you know if your reply is applicable or not.

>>108869312
>i didn't read your text
>responds with something completely irrelevant

>>108869319
>Is there anything to indicate that the requests are being done serially?
The fact that NtCreateFile doesn't allow batching and always blocks? The fact that they're all happening on the same thread (0x1bd4/7124)? What else do you need?

>>108869312
LLMs are limited by the r*ddit data they're trained on.

>>108861577
>Why?
NTA but I used to use TS on the backend in the past and
1. The node/npm ecosystem is a security nightmare because you end up importing hundreds of micro-dependencies that you don't control, are way too verbose to read, and get hit by malware attacks all the time.
2. JS is a shitty language to write real software in due to mediocre performance, very small standard library and lots of weird quirks. The only reason to use it on the backend is if you're a frontend code monkey and are too lazy to learn a better language.
3. The multi-threading/async/concurrency story is actually a joke. Go routines mog it, Clojure atoms+futures+pmap and core.async mog it.
4. JS is utterly useless for finance or anything else where you need to do precise arithmetic because its runtimes have no native support for any data type akin to the JVM's BigDecimal, and libaries like bignumber.js are very slow. Python, Ruby and Lisp/Scheme all offer this by default, in C++, Haskell or OCaml you can at least get high-performance 3rd party libraries.
5. Compared to vanilla JS I haven't seen TS do much for codebases other than make auto-complete better, at the cost of making the code much more verbose. A static type system like TS' helps prevent some errors that are annoying but ultimately shallow. Making a codebase pass a basic typecheck doesn't automatically unfuck its logic. If you're going to invest into improving code quality, using a functional language like OCaml, F#, Elixir, Clojure or Haskell does much more for you because it practically forces you to write code that is easy to unit test and does not become spaghetti code full of mutable state.

>>108869342
Is that a kernel thread or one of the game threads?

>>108869542
One: game, or rather userspace thread.
Two: why would there be a kernelspace thread opening files for userspace? That would be a huge security risk.
Three: it barely matters anyway. If it was a kernelspace thread you'd get rid of the constant mode switching, but not of the constant blocking for single files.

>>108869325
it had nothing to do with your post. I just wanted to say that here

>>108869576
So it's just Factorio being shitass.

>>108869727
No? >>108869270
>kernel interfaces are too shit for that, both on Windows and Linux

>>108869757
NtCreateFile can have multiple concurrent threads referencing the same file, and Factorio could have multiple threads opening different files.
If they decide to serialize everything it's not the kernels fault.

>>108869831
It only makes it slower because windows was designed in ancient times.

>>108869831
>Factorio could have multiple threads opening different files
Congratulations, now you have introduced additional context switches and additional complexity in the kernel, which now has to keep track of requests and merge them by itself, rather than letting userspace do it much more efficiently and on a single thread.

ahhh so the anon screeching CPU instructions at me is the anon freaking out cause he can't run Factorio on his Apple II.

>detect running on apple OS
>uninstall and delete program and exit

Could you make a game like Factorio using the UNIX rule where everything is a text stream?

>his Apple II
And then autists wonder why we call them retarded.

File: 1779326182521417.png (197 KB, 1920x1055)

197 KB PNG

POSIX it is then.

>>108869916
>rather than letting userspace do it much more efficiently and on a single thread.
I mean, it could if they wanted to, but part of using the kernel is managing the concurrency of threads and users of the system as a whole and not having individuals going off doing their own things that could interfere and result in unexpected outcomes.

>>108870328
Dude, *the hardware capability has been there since 1989*. Not only does it not make any sense to only submit one request at a time, but MS already fell on its face with this nonsense. Twice, in fact. Once with its retarded I/O completion ports that ended up being so stupid that they put parts of IIS into their kernel for static content delivery, then with their scatter/gather API that is still limited to either all read OR all write operations *on one single file*.

You're defending stupidity and not even realizing it.

File: 1761101859053732.png (25 KB, 1531x233)

25 KB PNG

Who needs ai when you can make slop yourself

File: study.png (91 KB, 420x527)

91 KB PNG

chanto benkyou shite !

>>108870402
That looks very memory safe.

File: base.png (3.84 MB, 1920x1080)

3.84 MB PNG

>>108870240
>no response
Yeah, thought so. Autists belong in the gas chambers, all of them.

>>108861216
JavaScript, read Eloquent JavaScript.

>>108870345
>*the hardware capability has been there since 1989*
You're conflating calls to NtCreateFile with actual commands sent to the drives.
Do you have anything showing that the drive commands queues aren't being well utilized?

>>108870889
I crown you the Factorio schizo.

>>108870889
>plays factorio
>calls others autists
Next you'll say you code in assembly.

If anyone can make an ncurses factorio that uses text streams like unix philosophy I will suck your girl cock.

File: heapalloc.png (17 KB, 930x910)

17 KB PNG

>>108870921
>You're conflating calls to NtCreateFile
No. No, I'm not.

>Do you have anything showing that the drive commands queues aren't being well utilized?
Do *you* understand what "blocking" means?

>>108870951
No, next I'll say
>fix your fucking memory management, you fucking autists

>>108871045
You have 2000 characters to fully elucidate the issue.

>>108871091
No. If you can't even be bothered to look up what "blocking" means, and how the kernel cannot *possibly* know what files are going to be requested in the future, then you're not worth the effort, and we, as a society, should prevent you and your ilk from ever programming anything again in the same way we prevent sex offenders from accessing the internet.

File: 1779339373050337.png (36 KB, 750x203)

36 KB PNG

>>108871099
nta but...