wheres marie?

I wanna throw whatever work I have done on Zetha, my ANSI C compiler, so far, and start anew.

I made a mistake by focusing on arenas and string pools. D has memory management. Use it.

I want to begin with a new name. Give me a good name. I don't like Zetha.

I've never been the one for 'wasted work fallacy', or whatever that fallacy is called. You can tell by me having close to 300 repos on my Github, most of them unfinished projects. If I feel it's not working, I throw it away.

And I just did that. I threw it away.

But I need a good name to start again. gib gewd name.

>>107756119
My cute brown poor French larper wife

>>107756169
have a French name: Frank
command: frankly hello.c -o hello
do straightforward compilation with minimal optimization. Speak directly and without subtlety to the machine.

>>107756169
>I need a good name for me umpteenth AI slop project.
No you don't.

>>107756169
Zurethra

>>107756322
The compiler is modeled after lcc, which does minimal optimizations, anyways. Just CSE with DAGs.

>>107756391
Nice. I'll choose this one. Sounds kinda like Zoroaster.

File: Screenshot_20250822_030245.png (216 KB, 346x337)

216 KB PNG

Now, Zurethra (name chosen thanks to anon) is just my warm-up compiler. I wanna go through it with:

1. Minimal use of LLMs
2. Without getting distracted away by bullshit like arenas and string pools

I wanna just make an lcc-like compiler in D.

The main compiler I wanna make, Cephyr, is a different story. I have *plans* for this one.

But I feel like, unless I don't make a smaller compiler, I'd be lost.

File: 1767446658113391.jpg (172 KB, 1320x1865)

172 KB JPG

Threadly reminder that any attack on programming yourself is jeets trying to shill AI slop.

Do not redeem the AI.

My boss told me he'll pay for the remainder of my college tuition if I enroll back in.

If I go back to college now, I can probably get my PhD at 38.

I don't understand why you folks are so anti-college. The reason you got no job, and jeet has a job, is that jeet has a degree.

I can view this situation at a good distance. Persona A opens a business. Person A considers hiring Person B, and Person C. Person B has no degree, and needs insurance, and loads of other expenses. Person C has a degree, and needs no expenses because he lives in Nepal or some place like that.

Would Person A be insane for choosing Person C over person B?

At least, try to get your degree, since a degree, even from a shitty American college, is better than a degree from the best Nepalese college.

>>107756119
Je t'aime mon petit oignon doux.

>>107756119
TABARNAK

>>107756958
the way i see it a degree is only worth something if you use your time there to network with fellow students, professors, recruiters etc. If you're just gonna spend it neeting in your room doing college work and exams might as well do those things on your own. Your chances of landing a job are about the same since you arent getting referred

>>107756455
>The main compiler I wanna make, Cephyr, is a different story. I have *plans* for this one.
Are you going to make your own language, instead of just another C compiler?

File: 1754287236546975.png (50 KB, 1200x858)

50 KB PNG

>>107756119
how do you even pronounce this?
q-tea or cute?

File: letmewikipediathatforyou.jpg (4 KB, 318x39)

4 KB JPG

>>107758236
Although personally I say ku-te

>>107758236
'TRA-sh, 'trăsh, /tɹæʃ/

>>107758273
damn it, I was explaining about it to my friend and I was saying q-tea all the time

This protocol defines a cyclical state machine with a specific focus on computational density and precision scaling. The architecture suggests a system that ingests data at standard precision, expands it for complex internal processing, and then compresses it before looping back for refinement.

https://colab.research.google.com/drive/10HngoUgOMg6_lWXGbsQJZkijhWgDfVdg

>>107758407
I'm sure that's fine.

File: screenshot-04-01-2026-20:10:51.png (76 KB, 803x748)

76 KB PNG

>>107758564
Causes /g/ to have a mental break down.

>>107758749
>not using the old vpmovmskb + tzcnt trick
The only breakdown here is one of never-ending laughter.

>>107758564
If you want *actual* people to use your program, you have to rely on already-established libraries and toolchain.

Dem's the breaks anon. Sorry.

Overreliance on externals, now, is also a mental illness. There are libraries on NPM that check if a number is odd or not. But I shove that to beginners being too stupid.

Here's a library that you'd be stupid to reimplemnt: ncurses.

Also, never roll your own crypto.

My compiler, Zurethra, has zero dependencies. But that's only because it's super-minimal.

My other compiler, Cephyr, has loads of dependencies. Chief amongst them, Goblint-CIL.

how do I filter someone based on their posting style?

>>107758829
Loaded gun to your head, safety off, pull the trigger.

>>107758850
get new material

>>107758801
How many hours have you wasted making your ultra optimized fizzbuzz as portable as possible only for literally no one to use it anyway.

I usually use ublock origin to block posts I don't like.

Like that retard who keeps annoying me with 'le ai le bad' shit.

Sorry, hun, if nobody is hiring you no more because you can write two lines of JS.

>>107758871
>optimized
>portable
I love how the nocoders are constantly telling on themselves.

>>107758877
Just answer the fucking question.

>>107758882
And if not?

>one retarded schizo leaves
>even more retarded pops up
where do these guys spawn from?

>>107758871
Plenty of times. You seen my Github have you not.

I think for an open source program, in this day and age, to become popular, it's mostly about the provenance of your attention-whore-iness.

Like, you gotta have a blog with posts that begin with "X is a hot topic of discussion these days" followed by lazily throwing wrongly-cited papers you haven't read (or have had AI read it for you), followed by a "Hot Take(TM)".

The more you advertise *yourself*, the more your library or program takes off.

Ditto of the program or library is in/for Rust. 70% of which is LLM-generated.

No shade at using LLMs to generate a 'rough' version of your program, I do it myself, but I always end up writing the main matter myself.

But, yeah, if you want your software to take off, make a blog, advertise yourself, and also, create a Discord server for it. That always pays off.

Just be careful, if you make a Discord server, some stupid fucking teenage twats will be attracted to it, 'community seekers', basically. And you get branded as a pedophile.

Speaking of blogs, I think it's time I made myself one. I won't share "Hot Takes(TM)" on it. Just share educational stuff. And my revisionist ideas on Claude Lanzmann's ""The Shoah.

>>107758887
>And if not?
<tumbleweeds>

>>107758829
Sounds like an actual legitimate use case for AI. You'd need to train a model on the style you want to filter.

>>107758871
>wasting hours making your ultra optimized fizzbuzz as portable as possible
Sounds like a good learning experience that teach you knowledge you can use for real projects.

>>107758962
>prefers simplicity
Sounds like a (You) problem.

>>107758988
What server? What code?

>>107759019
I think I'm going to be juuust fine without your imaginary service.

>>107759029
Yeah, like posting about code that doesn't exist, like you have been doing for the last 30 minutes.

>>107759040
The same goes for Linux and Mac, too.

>>107759076
Yet you don't name them. Probably because you don't want to endure another never-ending laughter attack.

>>107759094
Pffffff.
Right, nocoder.

My university's servers ran on neither Linux, nor Windows nor macOS, but some unemployable retard has an opinion on what matters.

File: 1767435286725222.png (185 KB, 437x403)

185 KB PNG

>>107756125
>>107757283
Oui oui je suis un faguette

>>107756169
>>107756455
>>107758523
Report this AI trash for automated spam.

>>107759351
Can you even send posts automatically on 4chidori with this new captcha?

File: (0) ∕g∕ - ∕dpt∕ - Daily P(...).jpg (78 KB, 959x502)

78 KB JPG

>>107759455
I dunno.

As someone who has already done the one page basic programs on notepad++ like, creating a shop, ATM, showing a star on the console panel, creating a registry and other stuff, what should I now do where I have to use multiple pages and have to use OOP using visual studios?

>>107759632
whatever you want
it doesn't have to be OOP

>>107759632
>I don't know what to program! What should I program?
Why does this question keep popping up lately?
Why did you get into programming in the first place?

I've finally taken the snake_case_pill for function names. Still PascalCasing types though.

>>107758901

function on_schizo_exit() {
    spawn_schizo(rnum);
}

>>107759632
>creating a registry
Show code.

>>107758928
yeah, for the discord stuff, I just disabled chat for everyone and only wait for them to express themselves via adding reactions on my posts, also vc and stuff disabled

I'm writing an snprintf that doesn't suck.
%s formats take 2 arguments: a char * and size_t length (not an int like %.*s does)
It's actually an iterator, so it can stop in the middle of a format because the output buffer is full, then when it will be called again it will continue where it left off and finish outputting the current format, and again if necessary.

>>107760560
All of these functions always suck.

>>107760560
It will always suck because of parameter code generation - i.e. having to copy raw values from complex structures onto the stack.

File: Screenshot from 2026-01-0(...).png (88 KB, 816x419)

88 KB PNG

>>107760560
i was gonna make a remark about fprintf having a return value
<but then this retarded shit in the man page

how is this even allowed?
we have a typeof() in c23 but this shit isnt fixed?

>>107760587
You mean non-libc format libraries in general?
It does exactly what I wish [sp][n]printf would do so this works for me.
>>107760598
>parameter code generation
>i.e. having to copy raw values from complex structures onto the stack.
I don't understand waht you mean.
The fact calling the format function is a moutful to call because of the struct parameter and initialization? It's intended to be used as an internal library, I'll make wrappers that will handle all this and that you will need to call only once.

>>107760660
I know all this but so what? I'm going repeat myself:
What if it didn't finish, what do you do now? There isn't a good mechanism to handle the case where it didn't finish outputting a format.
It returns an int instead of size_t.
%s takes nul terminated so if you have a length terminated string, like you should have in modern C code, you'd have to append a 0 (if you can) to the string you want to print and you're fucked if you want to printf a section of a string, or it gets messy.

>>107760743
>What if it didn't finish, what do you do now?
yeah, thats what i was saying
its a failure in interfacing. thats exactly the type of things the commitee should be dealing with instead of bloating c with sepples like features

i never dealt with this because i just dont use the printf family of functions outside of debug outputs
its a fucking tank, if i need to output anything i write an adequate function.
also because i deal with strings using a modernized type- compatible with null terminated c-strings, but i also have a size and i pad my strings depending on the type of operations im gonna be using (up to 32 null bytes if i were to go full auto retard with vector extensions)
in other words: modern fucking strings which allow modern hardware to enable all its features
another thing the commitee should be dealing with

>>107760743
>It's intended to be used as an internal library, I'll make wrappers that will handle all this and that you will need to call only once.
That fixes the code size problem, but doesn't get rid of the fact that you have data being copied from struct memory to CPU register back to stack memory, only for it to be copied back from stack memory to CPU register during ASCII conversion (or whatever the fuck you do). Better than nothing, but still fucked.

>>107756119
is zig worth learning? I see some projects that are starting to use it exclusively like tigerbeetle, bun and ghostty

File: tincan.jpg (149 KB, 922x1472)

149 KB JPG

>>107756119
>added one time secrets
>made the browser UI more modern for normie appeal
>voice chat is somewhat more stable now between browser and desktop
>added reporting feature and some basic moderation tools in case someone decides to be an asshole and dump bad stuff that could get me in trouble, so I'll start working on user uploaded avatars soon

Been in the works for over 2 years now. Switched from cross-platform TUI, to WPF, now settled on GTK. This side project has become one of my main quests.

>>107760916
it has plenty of good ideas (going all-in on custom allocators, build.zig, explicit imports of files=modules=structs, unconfigurable but easily hinted fmt) and special utility (bundled clang and cross-compilation toolchains) and is only a little retarded (explicit to the point of being a complete pain in the ass, long names to the point of becoming a denial-of-service attack on the reader, compile-time duck-typing). zigguy's objective was that people, even not liking Zig, would still use it, and he's doing a good job of pursuing that goal. Zig's something that you can generally write instead of C, and even where Zig sucks, you'll probably prefer it a lot to C. Other contenders fail to do this in a lot of ways that Zig doesn't.
In particular if you'd rather write mainly in in some more ergonomic language, like Lua or Erlang, writing the fast bits in Zig just makes more sense than writing them in anything else, even Rust.

>>107760813
About the copying on the stack, it's necessary to copy some state since it's an iterator, so it can continue where it left off but it doesn't copy the vararg argument itself. The vararg arguments are passed with a pointer to an array and a length, but the wrappers will use <stdarg.h> so they will need to copy the arguments indeed.

It's not meant to be efficient, but it shouldn't be too slow. There is only one simple function that does everything. It doesn't call other functions for now (except for %f and %m), it doesn't allocate anything, it doesn't use too many variables, it doesn't need to use a mutex either. https://aras-p.info/blog/2022/02/25/Curious-lack-of-sprintf-scaling/

I didn't use mempcy though, I usually do but here it's intended to be used on short strings for mostly logging stuff and since I need to iterate over the format string one character at a time anyway and %s parameters should be small...

>>107760811
>up to 32 null bytes if i were to go full auto retard with vector extensions
impressive level of autism

But yes, string handling beyond trivial shit is such a pain in the ass in general and in particular in non scripting languages, both for parsing and for generating. All the string handling tools suck balls.
If you think about it, a good deal of why the web cuck users so much is because everything is so much of a pain to even parse.

File: 1755376910116192.jpg (176 KB, 1080x1337)

176 KB JPG

>using library
>the trannies change the build system from make to cmake

>>107760916
Absolutely if you're writing something close to the metal, it's perfect for that. The further away you get from the metal, the more burdensome it becomes.

>use python3.x
>your python version is too new, we require python3.y
why is python allowed to break compatibility on every update?

>>107762401
Because despite all your rage you're still just a rat in a cage.

File: 5001858.jpg (33 KB, 430x316)

33 KB JPG

>>107762401
They are refining Python at each release into the perfect programming language. That's what gives them that right.

>protocol specifies CRC of non-divisible-by-8 number of bytes

File: sunzu.jpg (25 KB, 640x428)

25 KB JPG

>>107762565
>clients software uses 12bit crc with custom polynomials
>they just expects you to "work with it" without original source code

>>107762634
>disables bit integrity check

>>107762686
>clients software fails all your messages without any indication or return

>>107762401
>>107762529
I am writing an asyncio-based server program. Every basic feature depends on some barely maintained third-party library with hundreds of open github issues and little to no activity in the past year. At this rate I expect it to constantly rot and require major code-rewriting maintenance every 6 months. There are no alternatives in the standard library. No, I don't want to write a hundred thousand lines of code just to support a few complicated protocols with asyncio.
How do pythoneers cope with this horrendous state of the world? I wonder if it's a good idea to invert the entire codebase and write everything in the pre-asyncio generator coroutine style, yielding "effects" and accepting results from some kind of executor loop. That way all the third party crap is sort of isolated from the "pure" part of the program.

>>107762718
>How do pythoneers cope with this horrendous state of the world?
Most of the them are researchers or students so they do their one shot and forget it forever.
I think people that use python for work are stuck on their old system that they run in some docker container / virtual machine.

>>107762183
>>up to 32 null bytes if i were to go full auto retard with vector extensions
>impressive level of autism
id argue this is just competent design
padding the string with the width of the type youre working with allows you to drop all the code thats gonna deal with avoiding overflowing when you apply your code to the data

it can really slow things down
instead of doing one algo cycle utilizing the whole width of a vector type you will be parsing one byte at a time, REGARDLESS whether you look for null termination or you use a size with the string
if you can run your vector-sized algo on everything its not only faster, but it makes also less code to write

also, people instinctually want to pad to the nearest multiplier of your data width
dont do that on strings.
this way you can start working on your string with the full width of your types, starting at any point, even if you launch your function on the null terminator.
also also once you work with vector types this means youre on x86-64 and you have 32 padding bytes to spare for your hot loop strings
youre most likely working with an overallocated buffer already, anyways

Please retry operation after closing inhibitors and logging out other users.
'systemd-inhibit' can be used to list active inhibitors.
Alternatively, ignore inhibitors and users with 'systemctl poweroff -i'

I hate nu-linux so fucking much

>>107763108
Why do you have inhibitors on your system? I don't know what they are but it looks like they are doing their job, or your system is misconfigured for your needs.
Also is this really the right thread for this?

Is there anything I can do with these?

func u64 set_neighbors(u64 x)    { return x | ((x << 1) | (x >> 1)); }
func u64 keep_gaps(u64 x)        { return ~x & ((x << 1) & (x >> 1)); }
func u64 fill_gaps(u64 x)        { return x | ((x << 1) & (x >> 1)); }
func u64 remove_loners(u64 x)    { return x & ((x << 1) | (x >> 1)); }
func u64 keep_loners(u64 x)      { return x & (~(x << 1) & ~(x >> 1)); }
func u64 remove_edges(u64 x)     { return x & ((x << 1) & (x >> 1)); }
func u64 keep_edges(u64 x)       { return x & (~(x << 1) | ~(x >> 1)); }
func u64 keep_left_edges(u64 x)  { return x & ~(x >> 1); }
func u64 keep_right_edges(u64 x) { return x & ~(x << 1); }
func u64 keep_edges_remove_loners(u64 x) { return x & ((x << 1) ^ (x >> 1)); }
func u64 remove_first_run(u64 x)     { return x & (x + (x & -x)); }
func u64 remove_run_at(u64 x, u32 n) { return x & (x + (1 << n)); }
func u64 set_run_transitions(u64 x)  { return (x << 1) ^ (x >> 1); }
func u64 set_1_transitions(u64 x)    { return x & ~(x << 1); }
func u64 set_0_transitions(u64 x)    { return ~x & (x << 1); }
func u64 majority3(u64 x)            { return (x & (x << 1)) | (x & (x >> 1)) | ((x << 1) & (x >> 1)); }
func u64 equal_neighbors(u64 x)      { return ~((x << 1) ^ (x >> 1)); }
func bool is_zeroes_then_ones(u64 x) { return -x == (-x & x); }
func bool is_compact(u64 x)          { return (x & x + (x & -x)) == 0; }

>>107762700
>block clients software
>continue doomscrolling

>>107763395
>get fired from your job for not doing your work
>can't afford house or internet

>>107763388
what do you mean

>>107763388
You can fill gaps and remove loners.

>>107763388
also what does -x do when x is u64 (presumably unsigned)

>>107763452
>ruin entire network beyond repair
>fly to non-extradition country
>permanent vacation

>>107763479
~x + 1

>>107763363
>try to close system from command line
>try to close system from the gnome power off button
>redhat / gnome / microsoft wants to keep your system on for eternity
I hate linux so fucking much

>>107763506
does it give you back a signed value? i guess a + b is the same either way but what if it was something like a * -b where b is unsigned, would it be signed or unsigned mult.

>>107763520
u64 a = 3
u64 b = 5
u64 c = a * -b = -15 = 0xfffffffffffffff1

>>107763520
You can use speedcrunch calculator to visualize how the bits change.

>>107763561
that doesn't answer my question, I'm asking if in that language -x is u64 or i64 if x is u64 and if you have a mix of i64 and u64 which ops does it use

>>107763542
so presumably it gives you i64 and does signed multiplication?

>>107763671
C++, u64 stays u64

>>107763716
does it just happen to work out to the same thing? does signed vs unsigned mul only matter if theyre both negative [if interpreting it as signed] or something?

>>107763488
every country is an extradition country if you piss someone off enough
but probably not for failing to put up with BS from an undocumented CRC

>>107763795
>every country is an extradition country if you piss someone off enough
Tell that to Snowden.

Signed and unsigned numbers are same thing under the hood, and use the same arithmetic operations. The is no signed and unsigned multiplication, they are identical. It's just a matter of how you want to interpret the bit patterns of the arguments and the result in the end.

File: cbdc.jpg (113 KB, 1087x610)

113 KB JPG

>>107763827
>not a word about kikes
hes a crisis actor and his whole escape to russia is grand cinema
also... pigrel
in things that matter the east and the west are on the same page
picrel is map of digital currency, a project under the atlantic council's oversight
>rasha, china : pilots
yeah, theyre in on the plan and the apparent inimicity is all bread and circus

>>107763921
so why do MUL, IMUL and PMADDUBSW exist

>>107763940
like is it only the highest bit that is affected? or flags

File: Screenshot from 2026-01-0(...).png (2 KB, 102x48)

2 KB PNG

>>107763940
you will never unsee this
youre welcome

>>107763985
W

>>107763994
i thought youll appreciate
it always puts me in a good mood, too
finding positives in the little things, you know

>>107763511
you need it to receive updates goy

>>107763936
No one cares about your schizo narratives.

>>107763936
>yet another reason to immigrate to mongolia

>>107763660
You got told already. Unary negation applied to u64 produces u64.
There's multiple sane models for numeric types, but that's one.

>>107762195
There's nothing wrong with cmake that couldn't be fixed by nuking it from orbit just to be sure.

>>107764088
I didn't until here >>107763716

File: Screencast_From_2026-01-0(...).mp4 (3.04 MB, 1354x1008)

3.04 MB MP4

>>107756119
Basic Context Menu for TempleOS

File: Mount Stupid copy.jpg (159 KB, 1120x630)

159 KB JPG

>>107763940
I don't know. My understanding of how thins work may have been overly simplified and I may have spoken out of turn.

>>107764166
I can make it even easier for you: MOVSX vs MOVZX.

>>107763985
kek

>>107756119
dashi dedans

File: c43.jpg (30 KB, 450x450)

30 KB JPG

>>107764046

>>107764031
>schizo narratives
>straight from the atlantic council's mouth
thats some very strong cope you got there
>>107764046
i thought about this
mainly bc the country is empty as fuck and its all either steppes or deserts
but then i learned mongolia is china's bitch

i have no idea where tor un to, anon
not even africa is safe. maybe the fukken moon or something
and kessler synrome the earth for gatekeeping purposes

>>107764278
Take your meds.

>>107764286
nah, youre just pants on head retarded

File: 1728574471373651.gif (177 KB, 276x320)

177 KB GIF

>>107764315

>>107764331
>t. ostrich
its a retarded survival tactic
insofar it doesnt work
whose team youre on?
brics? or nafo?

>>107764358
The one that locks you up for literally forever if you don't take meds. As in, "life without parole or even pardon" forever.

File: same-same.jpg (22 KB, 321x300)

22 KB JPG

>>107764384
>mfw i finished reading your post

>>107764435
Good. Then I cannot steer wrong.

>>107764196
I don't get it. I think I understand those, but what does that have to do with multiplication?

>>107764461
Not "multiplication". Signed vs. unsigned integer promotion.

>>107764444
weird cope
the only way to win at this game is not to play

>>107764482
For you, yes.
For me, I just want you behind bars until they force-feed you meds.

>>107764501
>t angry. mad, even
tough tiddies. im gonna keep fucking up your psyops instead, faggot
not out of a sense of duty
but just because im bored and i find it funneh to dunk on you and your likes

>>107764520
You can also just spend the rest of your life locked inside, completely forgotten forever. That works too.

File: ukaki.png (15 KB, 400x400)

15 KB PNG

>>107764579
then why hasnt it yet hapened then?
oh, thats right

youre actually impotent
and im still doing exactly what i want, whenever i want

>>107764637
Same reason you survived school: systems fail.

>>107764654
>school
i went to a freemason school
they tried to break me

big mistake lamao
im an european
the harder you try to bring me down, the harder i fight
you should have covered me in gold and i would have been your most loyal ally

>>107763921
https://www.felixcloutier.com/x86/mul
https://www.felixcloutier.com/x86/imul
the upper word result has a different value wether you use mul or imul

File: 139224994_p0.jpg (586 KB, 768x1280)

586 KB JPG

programming is too tedious nowadays.
too much work to implement my imaginations.
too much researching, too much troubleshooting, too much learning datastructures and algorithms, too much dealing with system and language caveats.
it's all too tiresome.

>>107766708
then ack lamao
nothing of value will be lost

>>107766708
language design is the cure

>>107766721
nono
for him the only cure is a rope