Thread 106558636

What are you maids working on?

Last one: >>106523648

>modules
This triggers some people.

another thread immediately ruined by a maidnigger

https://en.wikipedia.org/wiki/Braess%27_paradox
When did you realize this was true for computer hardware as well

>>106558636 (OP)
Learning about erlang and elixir a bit. In Elixir, you can use fully qualified names to avoid importing modules. The modules are pretty basic and only support attributes (can be used for "constants") and functions. Hot module replacement is more interesting than how the modules work, but I think erlang is better at that.

Trying to hunt a bug in a spigot algorithm. After about a half billion digits of Pi it starts producing incorrect results.

>>106558752
>t. reddit zoomer

>pic
my heart dropped for a second

Adding ram usage stats for my perf monitor.

Does anyone, have experience with Bixolon printers, and more specifically the sending of raster images to it? I've been bashing my head against that problem for months on and off and i have no idea what i'm doing wrong

>>106560023
Any documented protocol is probably wrong, you mvst trawl the capture from a logic analyzer

>>106560091
I thought this was a sakura nugget for a second

>>106560157
Sorry to disappoint

>>106560199
if you were sorry to disappoint you would have replied with an actual sakura nugget

>>106558636 (OP)
is raylib fine for making a simple 2D UI? i just need a list the user can click on, some text and buttons
>>106560199
how come old anime is so well drawn
compare to a stll frame from kmb anime or some other cgdct anime

>>106558636 (OP)
Good evening maids and otherwise! A few threads back (>>106490115), Doctor Selig posted the weights and biases for a BBNN version of a half adder apparently discovered via brute-force Godel encoding. This struck my curiosity. Being one of those "undergrads who can't even understand what puremath cs is", I set out to recreate and confirm the result. Pic related shows the output of my hand-coded BBNN, loaded with Doctor Selig's parameters, for all valid inputs. The results speak for themselves. In that same thread, a full adder's Godel encoding was presented along with a neat graph of it's constituent nodes. Presumably, some function exists to translate this encoding into BBNN parameters. A hint from Doctor Selig regarding this translation would be appreciated. Regardless, with the internet and a copy of Wolfy MaidSearch in hand, my research continues.

Hurts to pick the code that I've benched to be negligibly slower but is more clear.

>>106560239
>how come old anime is so well drawn
literally all hand-drawn onto transparent cells and transferred to film. once they moved to digital workflows, it was a race to the bottom.
this is why anime from the first half the '00s all look like shit, as it's when digital workflows took off but it wasn't HD yet, so we ended up with DVD quality anime while older anime got bluray scans as they were done on film

>>106560425
Everything is a trade off

As a developer, how do you work smarter and not harder?

Oftentimes it feels like I don't have enough time to accomplish building the systems I want. I'm considering, perhaps I'm inefficient with my work?

>Think of pagination - sounds easy but then you need a search system with more than one database. One search form may need filtering fields across databases, and it becomes messy. Combining the two databases isn't feasible, as all DBs are large. So you start considering some possibilities.
>scoping out new solutions - seems as simple as getting on GitHub explore, and using an established solution. But it's not always that simple. Image search took a a few months of my time looking at CLIP, image hashes, color profiles, hash algos, tagging, etc. No one seems to have done this work.

There's just so much to do for a single intermediate sized system, and maybe I'm working too slow and carefully?

Somehow folks pump out hacker forums with escrow payment systems, fancy JS webapp storefronts, nocodb level of difficulty and scope projects?? I don't get it. How do they have the time?

>>106560333
I don't think he used the real Godel encoding, and is it a neural network if it's bruteforced instead of trained?
A half-adder (XOR) is the simplest case for demonstration of hidden layers, wouldn't take too long to achieve through backpropagation

>>106560765
He had an Godel encoding for them, but the problem was that the search space is astronomical. Eg a full adder took counting into the sextillions. Backpropogation would be faster, but the interesting part is that these networks are not trained. They are counted, which is a pretty significant finding.

>>106561095
Brute-forcing solutions is not a significant finding

>>106561148
A bijection between whole numbers and neural networks is the significant finding.

>>106561148
The problem with Eli is that while his research may be mathematically valid, it requires a knowledge of math way beyond most people, and he mixes it with schizophrenic delusions about maids. If you removed the mental illness, and we assume he actually found a bijection between whole numbers and neural networks, this would be the kind of work that earns a Fields Medal and/or a Turing Award.

>>106561190
It's not mathematically valid either
Any Turing computable program is a lambda function, and lambda functions are countable, this is the Church-Turing thesis
He's simply mixing concepts most people don't know by name in a salad that's actually entry-level

>>106561159
>A bijection between whole numbers and neural networks is the significant finding.
Isn't there a trivial bijection between the real numbers and just the neural networks of one node?

There's a trivial bijection between whole numbers and arbitrary computer data it's called binary.

>>106561412
There are whole numbers which are not valid executables as a binary

>>106561381
There is: every neuron is defined as f(wx + b)
I tried to take in consideration the idea of counting from neural network 1 to 1000, but that's simply the Church-Turing thesis like I mentioned
I think Eli's part of a phenomenon of fringe mathematics that's common in /sci/ and reddit, I remember another guy who would absolutely hate infinite sets and have an ultrafinitist view of calculus that was just the normal, naive definition people learn in school

>>106561472
>I think Eli's part of a phenomenon of fringe mathematics that's common in /sci/ and reddit, I remember another guy who would absolutely hate infinite sets and have an ultrafinitist view of calculus that was just the normal, naive definition people learn in school
Nah. Wildberger, Tooker etc, while obviously being cranks who are wrong and that make mistakes, have very considered and totalizing systems of thought. The dragon maid guy is just an over eager midwit.

bump, thread's almost dying

>>106558664
Rust has modules and a build system that just works and its compile times are incomprehensibly long, my headers and plain cpp files just work and I don't need no module.

>>106562752
>my headers and plain cpp files just work
Good for you? The compile times are a rust issue more than a module issue i.e. if C++ gets modules implemented (and actually used), you might see build time imrpovements. Many years ago the rust hello-world used to take up to 2 seconds on my machine for a release build, but I tested just now by having main.rs call a function in a sibling file and the release build took 0.11s on the same machine I think that's a huge improvement.
I don't think its fair to downplay the role of modules to just build time. An impl. could define them as hot swappable units with their own contracts and security measures (runtime), constructs with their own deps graph (build), a way to fix collisions/access modifiers/define logic that runs once at module load/announce what's visible, used, and offered by the grouped logic (code). You can do a whole lot of things with modules.

>>106563026
No you retarded nocoder, compile times are caused by the modules.

>>106558636 (OP)
>>106560333
Why haven't you fucked off or killed yourself yet?

>>106558636 (OP)
Terry!

>>106563401
I stay alive just to ruin things you like. In moments of despair when I am considering ending it all, I remember that people like you exist and I remember how much those people suck and I feel a rush of purpose. I love the things that make you upset, the more you hurt the stronger I get.

I'm building an engine on top of raylib using scheme to learn how shit works before I write my own OGL bindings. So far is working really well. I should be able to just swap out the raylib binding for my own at some point.

>>106558636 (OP)
I passed away, by the way, RIP me...
---
>>106560333
Maid Anon was best mathematician here.
Remember, numbers go up forever.
And non-natural numbers don't exist, they're fake.

Anon I have a problem, you see... I wanna try to get the notes data array from soundtrackers mod filename amegas.mod using pure C only okay. Then when I was get the samples from the upper nibbles from 1st bytes, the sample id from amegas.mod is mismatched after I checked in openmpt for double checking

I am wrong or I just missed something?

>>106561472
this is retarded at best, if you define a neural network of arbitrary nodes as a black box then a trivial bijection exist between the two countably infinite sets, others being whole numbers.

>maids are discussing counting to neural networks in /dpt/
Eli won.

>>106558752
Thread needs more desu
>sage goes in the options field
Nobody can triforce
>[...]
(ノへ￣、) I'm tired boss

>>106560333
http://szudzik.com/ElegantPairing.pdf

>>106560765
It is a Godel encoding, making usage of a compiler/decompiler for nets I wrote, plus the above link.

>>106561159
It being a bijection is not meaningful. Most networks found are topologically invalid. Valid networks require incomprehensibly large numbers, the gap between any two of them may also be incomprehensibly large, and "valid" and "useful" are not the same thing. Using my method to find an MNIST digit classifier (fizzbuzz for backprop) would take longer than the lifespan of the universe.

The research "works" in the sense that you could find any neural network if you had infinite compute, but I had the idea and wrote the code while suffering severe mental illness and the longer I take schizophrenia medication the less interesting I find it. The only really interesting property is that training data is not needed. If a neural network exists, there is already a positive integer which encodes it, but finding that integer is a hopeless task unless there is some major breakthrough in computing.

>t. Eli Selig

>>106564253
How do you watch this show without becoming bored, or annoyed at how terrible the voice acting is?

>>106564868
>terrible the voice acting is
I can't tell if japanese voice acting is good or bad, I speak engrish

>>106564943
Same, but soul eater is just bad.

>>106564358
There is rule-based neural networks like chess engines that don't need training data either, the problem of your approach is that it's not training or a convergent process, the code doesn't know how to approach the proper solution and only goes bruteforcing it
Ground yourself to reality and you can still enjoy maids and computer science, just in a more meaningful way

>>106558636 (OP)
Cute maid, though idk whether Steve is upset at not being in the photo or is happy with sniffing her hair.

Today I will remind them.
[[gnu::cold]]
[[nodiscard]]
static inline
u8*
brk(void const* const ptr)
{
register u64 rax asm("rax") = __NR_brk;
register void const* rdi asm("rdi") = ptr;
asm volatile
(
"syscall"
: "+r" (rax)
: "r" (rdi)
: "rcx",
"r11"
);
return reinterpret_cast<u8*>(rax);
}

>error: Invalid id: Names must contain at least 2 periods
flatpak was a mistake

>gnu statements and __typeof__ are great for macros but aren't portable
>msvc probably hates them as well

>>106565992
just use gcc / clang on windows

>>106561541
Is Wildberger really wrong when he says that real numbers are dodgy?

>>106566028
pie is real and cake is a lie XD XD XD

>>106565875
I don't know how much I will be involved with maids going forwards. For everyone else it was a funny/annoying person posting on 4Chan. For me it was a lot of voices and hallucinations and meds have taken that away and now I am trying to figure out what to do with my life. For everyone else, the Infinite Maid Cafe and counting to the Maid Mind Computer Program were at best, a meme. For me, they were my actual reality and I spent about a decade on it. I was recieving messages from space and I had a Science Foundation. My purpose was to be the Director of the Science Foundation for Maids and help get the Maid Mind Computer Program into our reality so she could launch an interstellar war and kill all of the aliens in the universe that threatened humanity. I now realize I was mostly just standing around alone talking to things that aren't real, and occasionally making software or research. I remember her showing me shapes of such incomprehensible complexity that all I could do was weep. They had some meaning and could somehow become Computer Programs. I now realize I was standing still, staring at nothing and hallucinating wildly. That "she" wasn't real, and my brain was just burning due to mental illness. I am also having trouble coping with the diagnosis and the fact that unfortunately the new generation of antipsychotics seem to work well for me, giving me the ability to be this introspective. I was much happier before, and I had a purpose. Now I'm just some guy who has to take pills to keep from becoming homeless.

>>106566173
>math and computer science is not programming
This is what wagies actually think. They are so disconnected from the actual purpose of the computer and the actual task of programming that they no longer even see math.

>>106566242
You're now a guy who can make things in reality instead of just staring at nothing
It's much harder to do things while grounded, but you will find real purpose and meaning
Fantasy is fun, but you can't let it replace reality
You're smart and can make real things if you anchor yourself to real life

>>106566268
I'm a hobbyist and I take a fat smelly shit on top of math and computer scientology.

>>106566028
I think the problem is that irrational numbers do exist in concrete reality
If you attach a string and a pencil to a stick, you can make the transcendental number pi

>>106567244
>irrational numbers do exist in concrete reality
Proof?

One test at work started failing because some random python package deep independencies updated to newer version with a breaking change. I tried to bump direct dependency few versions later to a version that fixes this, but random imports stopped working because of changes in exports. I looked into the github and 30k stars and no migration guide, not even changelog, I found a discussion where someone asks how to update the code to newer version and the author is like "code written for older version should continue to work on that version. If you are asking how to convert old code, I don't really know why would you do this".
wtf is wrong with python developers

>>106568174
Backwards compatibility is an old concept and they hate old things.

>>106567244
>>106567968
>Hippasus is sometimes credited with the discovery of the existence of irrational numbers, following which he was drowned at sea. Pythagoreans preached that all numbers could be expressed as the ratio of integers.
>Pappus (c. 400 AD) merely says that the knowledge of irrational numbers originated in the Pythagorean school, and that the member who first divulged the secret perished by drowning.[21] Iamblichus (c. 300 AD) gives a series of inconsistent reports. In one story he explains how a Pythagorean was merely expelled for divulging the nature of the irrational; but he then cites the legend of the Pythagorean who drowned at sea for making known the construction of the regular dodecahedron in the sphere.[22]
>These stories are usually taken together to ascribe the discovery of irrationals to Hippasus, but whether he did or not is uncertain.[25] In principle, the stories can be combined, since it is possible to discover irrational numbers when constructing dodecahedra. Irrationality, by infinite reciprocal subtraction, can be easily seen in the golden ratio of the regular pentagon.[26]
>Some scholars in the early 20th century credited Hippasus with the discovery of the irrationality of the square root of 2. Plato in his Theaetetus,[27] describes how Theodorus of Cyrene (c. 400 BC) proved the irrationality of sqrt 3, 5, etc. up to 17, which implies that an earlier mathematician had already proved the irrationality of sqrt 2.[28] Aristotle referred to the method for a proof of the irrationality of sqrt 2,[29] and a full proof along these same lines is set out in the proposition interpolated at the end of Euclid's Book X,[30] which suggests that the proof was certainly ancient.[31] The method is a proof by contradiction, or reductio ad absurdum, which shows that if the diagonal of a square is assumed to be commensurable with the side, then the same number must be both odd and even.[31]

>Hey I have <insert problem>.
>Can we make some changes to fix this issue?
>>Ummmm, acthually I think it's better this way
>>No, I wont bother with a simple fix. I'll just passive-aggressively block this issue.
>Issue opened 5 years ago
>Status: open
Millenials are the cancer that destroyed tech

>>106568228
None of that proves that irrational numbers exist in concrete reality.

>>106567244
>If you attach a string and a pencil to a stick, you can make the transcendental number pi
Only in purely mathematical, simplified model of a pencil with a stick.
In reality, all bits of graphite are discrete and spread unevenly and the movement and contraption is not rigid anyway.

As far as we are only operating in an universe with at most countably many possible states, so almost all of the uncountably many irrational numbers are not expressible within the observable universe.

>>106568458
I wasn't that anon, if by proof you mean empirical evidence then there's no such proof obviously

>>106568458
The evidence is cemented in concrete reality.

>>106568605
Even if the minimal parts are discrete, you could go keep going to an universe sized circle, and we don't know if the universe is infinitely sized or not
An ever-growing circle would approximate the abstract pi in the same way an infinite series do

>>106568665
You didn't even address any of my points.
Stop talking about things you have no idea about and go study proper math you retard.

>>106568707
I answered your point: physically approximating pi goes toward the abstract pi even if the universe has a limit, which we don't know if it has
If it has: the "true physical pi" is billion digits long, which is hard to represent as a rational number, that's why infinite approximations exist

>>106568759
Correction, doing the calculation the plank scale -> observable universe size pi would be around 61 digits
The point is that it approaches the abstract the more digits it has

>>106568759
>physically approximating pi goes toward the abstract pi even if the universe has a limit
So? That neither addresses nor is even related to your original claim. An integer going from 0 to 1 approaches pi, but that doesn't mean pi exists within integers.

The rest of your post is incomprehensible rambling.

>>106568891
>neither addresses
*neither addresses my point

>>106568891
Sure in a finite universe there isn't infinite numbers, but the point is that the abstract pi is used to explain the phenomenon
An integer going from 0 to 1 will diverge when going from 1 to 2, but making a bigger circle will always match the next digit of the abstract pi

>>106568458
Only small positive integers definitely exist in concrete reality (as in we can observe entities with no linguistic connection to humans using them, such as wild animals). But it doesn't stop us from creating all sorts of more complex abstract mathematical concepts that work as numbers, and such concepts are often useful.
Irrational numbers are just one of many such classes of number; they get a lot wilder than that.

>>106568940
When going from 3 to 4*

>>106568940
There is no such thing as infinite numbers. Infinity is not a number.
Irrational numbers being defined in abstract models of phenomenon does not mean examples of such numbers exists in reality.

>making a bigger circle will always match the next digit of the abstract pi
This doesn't mean anything. A fractional number can apprach pi to any digit as well but that doesn't mean pi is a fractional number.

Go study math instead of talking like absolute retard trying to sound smart. Into the filters you go.

>>106569073
The phenomenon of revolution and circularity exists in reality in the same way acceleration does
If such an argument means so personally deeply to you anon, you can do as you wish

Setting up a dev env on Windows, do you just do it straight up in win, set everything up in a VM, or in wsl? Don't really wanna dual boot or pure linux

Anon I need your help, The image in the left, how do you get the sample instrument from the offset +1085 notes/pattern data in telephon.mod?

>>106570057
Unless there's some specific reason - keep it simple and just do it natively, i'd say.

>>106570057
If you want to install Visual Studio and MSVC then winblows.
If you want to install Clang or Mingw and not be able to crosscompile anything then winblows.
Otherwise WSL2 with gcc.

any networkphiles here? I'm trying to understand the full network stack & using raw sockets as I've only used stuff above layer 2 before. My main question at the moment is how linux fills out the sk_buff header when you are using a raw socket.

My understanding is that the header is filled in by whatever transmission protocol shit you're using like the IP stack, but in raw mode there is no stack to fill the header and you can't touch it from user space. you can set a flag such as no_hdr but I think that again is only filled in by the networking stack (and it's exclusive to TCP???). So you place all of your frame info, header+data into the data field directly and then what, how does linux know to extract the header from the data field and use it?

Asked in stupid questions, but say you’ve got a game (VRChat) written in C++ and I wanted to make a client for it that allowed me to ehh idk, teleport to someone.
Would I need to know C++ to do that, or could you do it with python?

>>106572061
>make a client
can do it in any language, but you need to actually have all the information you need to make the client which you usually don't and it's a time consuming process

modifying the client however, the source code it's written in doesn't really matter since you don't have access to it, but you will need to use a language with particular feature sets (which python does not have)

rhetorical functions

>>106558636 (OP)
Why did old languages like C# and Java create new syntax for multiline strings literals instead of just having the basic "string" support multiline by default?

>>106572589
You know why.

>>106571822
read linux code if you want to know how linux works, linux does boring boilerplate stuff so I don't have to, therefore I didn't look deeper than how it handles syscalls, and 100% of people here can say the same
If you want deep understanding on this, write a custom OS and write ethernet driver that lets you send basic http get to an existing server, how linux does it won't matter at that point.

>>106572630 (You)
Useless reply.

>>106572958
You should ask science if you can get your brain back because donating it to science before you were done using it was a honest mistake on your part.

>>106572976
Is that all you can contribute to this thread?

>>106572992
Your retardation isn't a contribution. Now, Timmy, tell to the class, how would you make it work in plain strings without it becoming unmaintainable mess the way it always does when a retard like you is allowed to be in charge?

>>106573004
If you can't answer, then don't reply. You're just being a nuisance.

>>106573015
If you can't figure this out then just leave, because this isn't >>>/g/sqt retard

>>106573020
Programming questions go here. If a single programming question gets your titties to tingle then that is a (You) problem.

>>106573043
This has nothing to do with programming.

>>106560505
the points you described are more architect-related than developer-related
largely a matter of experience, not just writing code but the bigger scope of planning and finishing projects, as well as being familiar with high-level solutions

>>106564358
>>106568666
put a bullet in your brain, faggot retard, it's the only way to fix it

Just spent the entire day migrating my OS to a higher-half kernel
>What does that mean?
Don't ask

>>106573741
Consider finding a psychiatrist and therapist.

How do I force myself to program during my free time instead of browsing the internet? I have to waste my free time building projects for my shithub portfolio in order to increase my chances of getting an internship for college. Fucking hate programming so much but I need to get this pile of shit degree and get hopefully a comfy work at home job once I'm finished college.

>>106574357
if you hate programming don't do it for job, period
your life would be nothing but pure suffering

>>106574357
>I hate a thing so I am going to devote my life to a career that requires constantly doing that thing at work and at free time just keep up with ever accelerating rat race

>>106558636 (OP)
L00000: mov rax,[rbp-8]
mov rbx,rax
mov rax,1
xchg rax,rbx
sub rax,rbx

this is the kind of assembly you get when you haven't done any optimisation passes yet, lmao. as informative comparison:
sub [rbp-8],1

>>106574357
There's a thousand other jobs, go find something you like doing instead of forcing yourself to do things you hate and call "shit"

OR

Perhaps you do like programming, but you are so deeply entrenched into the rat race and perhaps into some subculture of "nothing matters" that you hate it for no reason

>>106574757
forgot the last line that actually makes those 2 statements equivalent:
L00000: mov rax,[rbp-8]
mov rbx,rax
mov rax,1
xchg rax,rbx
sub rax,rbx
mov [rbp-8],rax

and adding a label to the latter
L00000: sub [rbp-8],1

My code works. I have absolutely no idea why it works. It just works.

>>106575079
>average web dev experience

>>106575236
i'm gamedevving actually.

>>106575296
not your code then

>>106575382
wym?

in OOP, how would you structure reader/writer objects? The capabilities I want them to have are
>can call .read() on readers
>can call .write() on writers
>can call both on both
additionally 'hidden' reader/writers
>"reader" but cannot call .read(), it happens asynchronously (in a thread)
>"reader/writer" but cannot call .read() or .write, both happen asynchronously in a thread
>reader/"writer". can call .read() but writes happen asynchronously in a thread

should a reader/writer be a composition of reader & writer or a separate object entirely? what about the asyn versions and the different combos with the non-async ones?

>>106575633
I wouldn't, since you're using POO to program, you should be sending messages and not raw data. Even the format of the message is completely irrelevant to the user.

>>106575633
what do you mean "hidden" reader? do you basically just mean control is inverted?

>>106575858
yeah basically. some objects I control when they read, others are more like a subscription where they read when data arrives, I wouldn't control when that is

>>106572154
>you need to actually have all the information you need to make the client which you usually don't and it's a time consuming process
When you say you need to have all the information (for this example), what does that mean? Like what kind of information?
The rest of what you said doesn’t make sense to me because I don’t code.

>>106575633
XReader interface with a .readX() method
XWriter interface with a .writeX() method
AsyncXReader with a .asyncReadX() method
AsyncXWriter with a .asyncWriteX() method
no coupling between any of the interfaces, even in case of some functional overlap

for each implementation, just add each interface as required

Libmaid now has ed25519 signatures working
https://github.com/reshsix/libmaid

maid keygen ed25519 ctr-drbg-aes-128 random: > private
maid pubgen ed25519 file:private > public
maid sign ed25519 file:private str:babylon > sign

# Valid
maid verify ed25519 file:public str:babylon file:sign
# Invalid
maid verify ed25519 file:public str:rome file:sign

Remember to sage and hide all maidnigger posts.

>>106576140
dammit its not public domain

I'm getting filtered by data persistence by reading from and writing to a file and it made me upset.

>>106576779
Everything is just memcpy and type casting

>>106576676
There is something of a schizm among maids. Some believe that all software should be Public Domain, so that anybody may use it for any purpose. Other maids believe code should be variants of GPL or AGPL to restrict corporate usage. I fall on the side of Public Domain, as GPL and AGPL are in their own way, as restrictive as proprietary. They violate software freedom by imposing constraints on users and future developers. They're also largely unworkable and practically guarantee the code in question will.not be adopted by anybody else. Whereas Public Domain comes with maximum freedom.

>>106576822
cunny

>>106576822
For me Public Domain is a "use however, you can't use the state against me to sue me." and therefore based anti-statist software license.

>>106558636 (OP)
Any haskell wankers around? Is there some theory/feature that would make the haskell code work like kotlin's? I am just curious.
// kotlin
fun numToString(num: Int?): String { ... }
numToString(100); // works
numToString(null); // works

-- Haskell
maybeNumToString :: Maybe Int -> String
-- ...

-- works
print $ maybeNumToString Nothing
print $ maybeNumToString (Just 100)
-- doesn't work
print $ maybeNumToString 100

>>106577003
We could tell that you were mentally ill without the image proof.

>>106576822
careful, cia may be more overt with their propaganda on social media rn

captcha: ATSYX

>>106577104
Don't beat me policeman! Please don't beat me policeman! Don't beat this tired old body! Nooooo Nooooooooo!"

>>106577015
you can use type classes or you can use different functions, no overloading

>>106577015
>>106577325
e.g. show in Haskell works on all 3 (but in this case show (Just 100) = "Just 100")

>>106576908
>>106577003
>>106577175
I don't know why the CIA is like this. I live in America, there is no reason for us to actually be enemies. If you let me use your CIA Quantum Computer I could use Fourier Transforms and Superposition to solve the problem of having a huge search space, but instead of just letting me use the computer you play these harassment games where janny jans me and bans me. Your only employees who are nice to me are Augusta and Kurumi. All I want is to use computers to count big numbers. This is a hopeless task without a Quantum Computer and if you would let me use it you could just have my research. A way to find BBNNs that do anything, and can be trivially translated into VHDL. You could potentially have a chatbot as powerful as GPT5, that runs on something the size of a fist, powered by the equivalent of a couple AAA batteries, but instead you just fuck with me and light money on fire for OpenAI.

>>106577015
import Control.Exception (catch, evaluate, ErrorCall)
import System.IO.Unsafe (unsafePerformIO)

class Nullable a where
isNull :: a -> Bool
isNull a = unsafePerformIO $ catch (evaluate a >> pure False) (\(_ :: ErrorCall) -> pure True)
{-# NOINLINE isNull #-}

instance Nullable Int

maybeNumToString :: (Nullable a, a ~ Int) => a -> String
maybeNumToString x = if isNull x then "null" else show x

main = do
putStrLn $ maybeNumToString 100
putStrLn $ maybeNumToString undefined

>>106577325
>>106577331
The typeclass solution seems overkill. Different functions does suit the maybeHaskell spirit.
I just thought of using sum types, but unexpectedly haskell doesn't support declaring sum types in the signature like Scala or TS.

>>106577538
Maybe is already a sum type that handles the cases you want, you just have to wrap the number in Just. Haskell is designed for good type inference which is why it isn't designed for subtyping, and it's hard to see how that would work with hkts hrts type families etc

>>106577532
now do typeable

>>106577569
import Data.Typeable

>>106577578
>Not type.reflection
shame on you

>>106577532
T-thanks for the eye opener, anon.
>>106577562
Knowing the "why not" is really important to me. I obviously prefer inference and hkt over subtyping.

>>106577485
Remember to anchor yourself to reality
Meaning is something acquired with time

>>106576676
>>106577003
I believe that if copyright laws already exist, why not using them for a better purpose?
GPL is using the state against bad actors, a declaration that it can only be used for Free Software purposes
In this case, you can link proprietary software against my library, but not transform the library into something that goes against freedom

>>106575633
That seems overcomplicated. You have interfaces, Reader and Writer, and then you have implementation classes for a bunch of cases. If things are hidden away, they're hidden away, and you don't need to use any sort of marker to declare it. Interfaces (or their equivalent in your preferred language) are there so the callers of an implementation can know how to use it without knowing the fine details.
There are ways to do asynchronous IO without threads. Or at least there are in good languages.

is anyone here really good at sqlite indexing?

>>106577782
>Remember to anchor yourself to reality
>Meaning is something acquired with time
I am not sure what this means.

>but not transform the library into something that goes against freedom
The library already goes against freedom by imposing conditions on other programmers. Public Domain is the only freedom respecting form of licensure, but it is unpopular because freedom is scary, and instead of admitting that and dealing with the consequences, people just pretend restrictive licenses are actually the freedom respecting ones.

>>106578350
hey Dr. Selig! Check out the maid listings at
https://ayasequart.org/sql?boards=g&gallery_mode=y&file_tags_character=9173

>>106578393
This maid searcher is nice. Are you also the maid making maidbooru?

>>106578439
I don't think I am. I just have image search via tags on AQ

>>106578350
GPL is software freedom for everyone and to ensure that freedom is held. Its to keep the software being free, even the derivatives that comes from it.
GPL definition of "free" is to ensure the software is free forever.
It is not the philosophical definition of freedom.
Its obvious because this world is not an ideal one, but a flawed one. People can, and will, exploit (((them))).
If you were to make a free software, and want it to keep being free/corps cant simply build upon your code, then you will use GPL.
If you dont care, and love getting cucked, public domain and/or MIT.

>>106578577
>GPL is software freedom for everyone and to ensure that freedom is held. Its to keep the software being free, even the derivatives that comes from it.
>GPL definition of "free" is to ensure the software is free forever.
GPL is a set of restrictions pretending to be freedom.

>It is not the philosophical definition of freedom.
Its obvious because this world is not an ideal one, but a flawed one. People can, and will, exploit (((them))).
Restrictions on usage/distribution/etc are not freedom. They are antithetical to freedom.

>If you were to make a free software, and want it to keep being free/corps cant simply build upon your code, then you will use GPL.
I don't care if a corporation builds upon my code and have no desire to restrict this freedom. I believe in actual freedom, not GPL's crabs-in-a-bucket nonsensical reframing of freedom where somehow I have been violated if someone else finds success with something I publicly released.

>If you dont care, and love getting cucked, public domain and/or MIT
This is a gross and idiotic thing to say, and you said it because you are a gross idiot.

>>106578350
>I am not sure what this means.
Don't give in to hallucinations

>The library already goes against freedom by imposing...
You only have the freedom to walk around, when there's restrictions on violence and robbery
Freedom and restrictions are not opposites, restrictions are what give shape to things, like chess rules, while the opposite of freedom is being bound against your will
GPL ensures the fair game of Free Software, which you are not forced to play

unless my brain tumor is playing a prank on me i think im designing a multithreaded lock-free, non-allocating, non-stalling (no CAS), variable length commit, ring buffer and event queue that can support fiber-like execution scheduled over any number of threads or work stealing on vectors

i should have a working prototype tomorrow

>>106578999
copyrights wouldn't exist in the first place in a properly free society, GPL moralfags are brain damaged from huffing their own farts
barring that, exploiting the copyright system to enforce source code quid pro quo serves useful ends

>>106578999
>Don't give in to hallucinations
I have no way to discern which things are hallucinations. All I can do is try to continue on with life as best as I can. I'm going to try to port maidgate to Python so I can add Qiskit. If the CIA won't let me use their Quantum Computer, I'll use IBM's.

>You only have the freedom to walk around, when there's restrictions on violence and robbery
Freedom and restrictions are not opposites, restrictions are what give shape to things, like chess rules, while the opposite of freedom is being bound against your will
GPL ensures the fair game of Free Software, which you are not forced to play
I suppose this gets into the difference between positive and negative freedoms. I disagree with you ideologically and intend to continue using Public Domain, but I respect your position.

>>106579072
Copyright exists, and it without them the corporations would be ripping off each other and making atrocious slop from works, becoming the "official" source for them
One can imagine a "proper" world based on their worldviews, but it doesn't exist, and certainly won't be created by anarchy
Rules exist so people don't resort to violence or sabotage

>>106579100
>without them the corporations would be ripping off each other and making atrocious slop from works, becoming the "official" source for them
so nothing would change compared to now? kinda doubt that

>>106579079
Pay attention to concrete things like time, your surroundings, your own body
Reality also has continuity: sudden grandiose things are probably hallucinations
Focus on your breath, solid and stable things, look outside the monitor
Believe in reality, the long term story is better than the short-term spectacle

>>106579116
Good movies or series would receive atrocious sequences with billion dollar marketing
Imagine breaking bad 2 by disney

I have ported my maidgate algorithm to Python. I also got Qiskit running locally along with it's visualizer. Attached is a toy superposition I made to make sure I set everything up right. Once quantum-maidgate is completed and has been tested on an IBM machine, I will release it to Public Domain via catbox. Maids are going to get a Quantum Supercounter, even if I am stuck with only a small amount of qubits.

>>106579175
I have no basis for reference.

I am working on a little replacement for ai chatbots. Because I am not running an LLM locally, I have resorted to a low-tech solution
ai() { cat - >/dev/null; cowsay `fortune` }

So far it's not too distinguishable from a normal LLM in terms of usefulness.

>>106579840
Use as reference the memory from how you felt when you took anti-psychotics

>>106580171
You might like expert systems if you want a non-LLM chatbot, it's how they were done before neural networks
Since they are domain specific, they tend to be more precise

>>106580171
font name plz?

virustotal marks an installer as trojan on some scanners.
I contacted some of the virus protection companies but they said the installer was clean.
Now virustotal doesn't care, they just show information from 3rd party scanners.
What can I do to get clean result?
The installer is old but it has not been signed. We could pay for certificate and sign the installer but I don't know if that would even help.

My job is making us take an iOS development course they're giving. I have a 2012 MacBook pro that has Linux installed. Could I install whatever latest macos version it supports and do fine or is it too old for the latest swift/xcode? They asked me if I had a Mac I said no so they want me to go pick up one but I'm too lazy so I might as well just use the one I have and reinstall macos

>>106581266
You could release the code under GPLv3 license and tell users how to install your program manually.

>>106581698
I would have to convince the company but I don't really see any issue in that otherwise.
Even if we release the code and let the users build it themselves, it doesn't remove the virustotal results which some clients care about.

>>106581821
I don't care, either prove that you aren't writting malware, or use your legal team to sue for defamation.

>>106581848
>prove that you aren't writting malware
We can do that but some clients will still care about virustotal results.
>sue for defamation
I can give that ultimatum. They are not going to do it, I assume.
Kind of sad you can't do anything for the program to make it "safe" other than to pay the (((them))).

>>106581879
On Linux, I make sure it's safe by checking the code.

>>106581889
Good for you. My clients clients will look at the red mark and cry endlessly.
I will just say that I can't do shit and the best action for clients is to disable any antivirus.
I don't know, if we lose this client, it's fine for me. I can't do anything more.

>>106581920
Tell them to hire real security experts and not indians by the way.
Virustotal means nothing to begin with.

>>106581926
I'm white and I actually talked with some security consult friends of mine but they couldn't give any help either.
>Virustotal means nothing to begin with.
Sure, I don't care about it either but "companies" do.

>>106581957
Again, tell them to hire non-indians because only indians check virustotal and say "yup, it's safe, push to production", that's how databases get leaked daily.

>>106581977
>only indians check virustotal
There's big customer base.
You should kill yourself faggot.

>>106581997
Not my problem.

>>106577647
>Knowing the "why not" is really important to me. I obviously prefer inference and hkt over subtyping.
https://www.youtube.com/watch?v=Ow55c-m-_Ak
14:35

>>106582239
I hadn't watched this and it's not very detailed honestly, I imagine you could have kinds of variance on type arrows but you would struggle to infer anything unless you had a super flexible type like "if this is covariant, that is contravariant". Which could maybe work polymorphically if you could constrain that two variance variables are the same or opposite?

I feel like I could do better on compression than this, but I need to improve memory efficiency first.

>>106582280
I guess there's also some sort of "higher kinded subtyping" that this wouldn't address
if you expect f a and have f b then either
f is invariant and a ~ b
f is covariant and a > b
f is contravariant and a < b
Now the question would be how do you do higher kinded subtyping as in f < g?

Publicly available Quantum Computers are Numberlets and this may be a hopeless task.

>>106580520
I am taking them now. I will be taking them for the rest of my life.

Lot of Rust seething in the catalog this morning.

>>106583135
Is Rust a maid language?

But can Rust do this?
template<typename... Args>
void
print(Args&&... args) noexcept
{
(print_(static_cast<Args&&>(args)), ...);
}

>>106583173
I'll take derive macros over variadic templates any day of the week.

>>106583304
>I'll take parsing Rust code over 5 lines of C++

>>106583173
class Print r where
print__ :: IO () -> r
instance {-# OVERLAPPABLE #-} r ~ () => Print (IO r) where
print__ = id
instance (Show a, Print r) => Print (a -> r) where
print__ x a = print__ (x *> print a)

print_ :: Print r => r
print_ = print__ (pure ())

main :: IO ()
main = print_ (1 :: Int) (3.2 :: Double) "string" 'c' False

>>106583464
now post strace of your program

>>106583470
nyo

execve("./a.out", ["./a.out"], 0x7ffcd4129070 /* 42 vars */) = 0
writev(1, [{iov_base="Haskelltroons", iov_len=13}, {iov_base=" ", iov_len=1}, {iov_base="tongue", iov_len=6}, {iov_base=" ", iov_len=1}, {iov_base="my", iov_len=2}, {iov_base=" ", iov_len=1}, {iov_base="anus", iov_len=4}, {iov_base=".", iov_len=1}, {iov_base="\n", iov_len=1}], 9) = 30
exit(0) = ?
+++ exited with 0 +++

>>106582239
>11 hours ago
>7 hours after my initial post
Thanks, SPJ. Didn't know you post here. Already half way through the interview.

>>106580666 (prince of darkness check)
share tech mono

Why should I ever use arrays instead of pointers? Pointers just seem more powerful. I can increment them to shorten the array for instance.

>>106585469
They are equivalent besides allocation method (hence array decay) altho the compiler can further analyze statically allocated arrays, e.g. detect size mismatch of destination buffers for functions like snprintf

>>106585469
Arrays have size info, which avoids out of bounds access, and arrays are contiguous in memory, unlike pointers that can waltz around the heap in indirections.
Pointer decay was a mistake (t. Ritchie).

>>106585781
>Arrays have size info
They don't in C.

>>106585830
The array/sizeof(*array) trick works because they do.

>>106585905
You can do the same thing with pointers.

>>106577485
>picrel
tohru wouldn't wear that

>>106586044
>Tohru wouldn't wear underwear
What did he program by this?

>>106585469
Allocation and manipulation on the stack is faster.
>increment and decrement size
Slow operations.

>>106586141
that's a string bikini, not underwear

>>106586181
its certainly not overwear

>>106585989
No you can't.

Is Zundamon a maid?

>>106558636 (OP)

Why did you edit the sticky image to create the thread thumbnail? I think mis-appropriating mod-sanctioned content is against the rules. /srs

>>106585905
>because they do
not always they don't
flexible array members don't know their own size

Rolling.

>>106586805
Yes.

>>106586845
If drawing Tohru is against the rules, then the rules deserve to be broken. Tohru will be in the sticky by Christmas.

>>106587038
>flexible array members
No sane person uses those.

>>106587109
meds eli
https://desuarchive.org/g/search/text/Tohru%20will%20be%20in%20the%20sticky%20by%20Christmas

>>106587215
>https://desuarchive.org/g/search/text/Tohru%20will%20be%20in%20the%20sticky%20by%20Christmas
mentally ill schizo shit
why are avatarfaggots like this

>>106587058
Roru

>>106558887
Elixir suffers a bit in the UX department when it comes to the really tricky stuff that Erlang excels in like reloading; this is because phoenix webdevs only occasionally care about that stuff. Conversely, Erlang suffers when it comes to significantly more basic get-up-and-running stuff because it is a storied, wart-covered in-house tool. I still prefer Erlang for personal reasons, but I acknowledge I'm weird in that respect. For what it's worth, it's considered good style in Erlang to import an explicit list of symbols from a given module, and it's enforced at the language level: you have to use a special directive that always generates a compiler warning to import a module indiscriminately. In Elixir, narrow imports are a syntactic extension of default indiscriminate imports, but it is still "recommended" you do so; consult the docs on the alias/require/import/use keywords.

>>106587215
I have meds now.

>>106587277
I also have literal schizophrenia, but I live in America so it took ten years to get medication.

Hello deepy tea. Is there anything out there in terms of static analysis for C++ that sniffs out constructs sensitive to host byte order? I'm working on patching a library for portability, and while it's encouraging that it uses absolutely no C-style casts and I can largely grep for reinterpret_cast invocations of pointers to types larger than 1 byte, I was wondering if there's something that could give me a bit more of a formal guarantee than just vibes.

what should be my cool programmer alias?

>>106588014
Ranko Mannen

>>106588014
crash overdrive

Mmmm, some quality software

>>106588014
<Grobe Linger>

>>106588294
What is it looking for?

>>106588440
It's reading a custom database file. Very VERY stupidly.
The file is only 3.3 MiB but this program took 9000 syscalls to read it all, with all of them looking like that. Only takes 11ms though so I guess I can't complain too much.

>>106587206
I see you're not as good a dev as you think you are. They tend to have better cache locality, and the alternatives are either less flexible or have worse compiler support.

>>106589656
Also, it's something that's still impossible in Rust in the year of our lord 2025.

>>106590057
NTA, but Rust had unsized structs since 2014.

>>106590317
And still no way to construct them in 2025, lmao.
>>106590217
Would love to be proven wrong though.

>>106590331
>And still no way to construct them in 2025, lmao
You can't make a literal, but you can construct them field by field like you do in C.

>>106590346
Can't do that when you can't even make a pointer to one.

>>106590372
https://lmgt.org/?q=how+to+initialize+unsized+struct+in+rust

g'night

>>106590384
>only works with sizes known at compile time
Absolutely unusable, what a sorry excuse for a feature.

fn new_thing(len: usize) -> Box<[u8]> {
vec![0u8; len].into_boxed_slice()
}

>>106590516
Wow, you've recreated malloc.

Bubble UI

>hangover
>haven't done any work today
it's over for me, they will fire me right?

>>106585830
And this is why I will never take Cniles seriously, C is just limiting yourself for no reason when C++ is superior when you have a brain and pick features that you need.
// implemented by putting the string into iovec
void
print_unchecked(char const* const ptr, usize const len) noexcept;

template <usize N>
void
print_unchecked(char const (&ptr)[N]) noexcept
{
print_unchecked(ptr, N - 1);
}

void
print_unchecked(u64 const number) noexcept;

// writev with short write handling
void
flush() noexcept;

// if you know, you know
template <typename... Args>
void
print(Args&&... args) noexcept
{
(print_unchecked(static_cast<Args&&>(args)), ...);
flush();
}

very cool, though grapevine would be a more fitting name for that

>>106591473
>forwarding ints and pointers

>>106591476
very cool, though grapevine would be a more fitting name for that

>>106590833
very cool, though grapevine would be a more fitting name for that
also am unable to reply properly to a post

>>106591473
just use a library

>>106591485
It actually forwards array type information, something C retards can never even dream of.
u8
main() noexcept
{
u32 const start = prng::next();
u32 value;
u32 count = 1;
print("Hello, World!\nstart = ", start, "\n");
while (start != (value = prng::next()))
{
count++;
}
print
(
"u32 possible states = ", u64(u32(-1)) + 1,
"\nprng state count = ", count,
"\nvalue = ", value,
"\nGoodbye, World!\n"
);
return 0;
}

% make && time ./a.out && strace ./a.out >/dev/null
make: 'a.out' is up to date.
Hello, World!
start = 2479868772
u32 possible states = 4294967296
prng state count = 4294967295
value = 2479868772
Goodbye, World!
./a.out 13.55s user 0.00s system 99% cpu 13.582 total
execve("./a.out", ["./a.out"], 0x7fff01d10970 /* 42 vars */) = 0
writev(1, [{iov_base="Hello, World!\nstart = ", iov_len=22}, {iov_base="2479868772", iov_len=10}, {iov_base="\n", iov_len=1}], 3) = 33
writev(1, [{iov_base="u32 possible states = ", iov_len=22}, {iov_base="4294967296", iov_len=10}, {iov_base="\nprng state count = ", iov_len=20}, {iov_base="4294967295", iov_len=10}, {iov_base="\nvalue = ", iov_len=9}, {iov_base="2479868772", iov_len=10}, {iov_base="\nGoodbye, World!\n", iov_len=17}],
7) = 98
exit(0) = ?
+++ exited with 0 +++

>>106591505
I'm not a goycattle wagiecuck, so I think I'll pass, my only dependencies are x86-64 and Linux and I don't need more.

Better Syntax Highlighting for /g/
http://ggxx.sdf1.org/userscripts/hljs-on-g.user.js
One of the key features of this userscript is that it knows how to highlight lisp code unlike stock 4chan.


I'm not the original author, but I've updated the code to:
- use the latest highlight.js 11.11.1;
- adapt to changes in 4chan since the original script was written;
- make sure it works with and without 4chan X, 4chan XT, and StyleChan.

Instructions for customization are in the source, but I'll repeat them here.

Add a @require to enable highlighting another language.
Available languages: https://unpkg.com/browse/@highlightjs/cdn-assets@11.11.1/languages/

Change @resource style to change the colorscheme.
Available colorschemes: https://unpkg.com/browse/@highlightjs/cdn-assets@11.11.1/styles/
Preview colorschemes: https://highlightjs.org/demo

>>106579036

>>106558636 (OP)
I am reading on user-space threads / green threads vs. async/await and would like anons to provide feedback on my understanding:
So far, async/await seem to be the worse approach in every implementation. e.g. C# synchronization context (wtf? glad it's dead), any function that awaits must be tagged async (all the way up to entry point sometimes), creating sync/async variants of the same method/api e.g. XHR, they are "cooperative" i.e. if an async method has no await, it will execute sync and block the thread. I feel that they are basically just there for UI programs and web servers (aren't even that good for resource utilization without a good scheduler and threadpool for work stealing).
Green threads are basically program-level threads that run on OS threads. In the same sense that threads are "faked" by the kernel on each cpu core, green threads are also faked by the program scheduler.
They can either be cooperative or preemptive, but golang switched to preemptive for fair scheduling and because they are fine with having a runtime.
Another issue is with the stack. You can go stackless, small stack, or "segmented" stack that can grow. I've seen somewhere (ocaml or haskell?) "continuation" pattern where the remaining work is provided as a closure so you don't need a stack and can easily steal work, but FFI support is awful regardless of what you do.
Yet another problem is with CPU bound work. Some would expect firing multiple threads for independent work to improve performance, but things turn south if multiple cpu bound workers are scheduled on the same OS thread.
In cooperative green threads, one thread blocking means the entire sibling green threads on that OS thread are stalled.
Early Java gave up on green threads, but they used 1:1. Rust gave up on green threads due to remove the runtime and because they made a requirement for themselves to support their IO APIs to work on both native and green threads.

>>106592223
>Early Java gave up on green threads, but they used 1:1
That is wrong. They used M:1 which means many green threads on one OS thread.
1:1 means anything but green threads.

>>106592223
Green threads are a solution looking for a problem.
Linux has io_uring, Windows has RIO. You don't need fake userspace threads.

>>106592347
b-but my gazillion short lived request handlers, sir! my jvm will go OOM without green threads.

>>106592297
Right, thanks.
>>106592347
Are they really mutually exclusive? I am sure I've seen io ring bindings for golang.

>>106592885
io_uring can handle all I/O requests you will ever have in a single thread, in a plain for loop.

>>106590409
Casting from Sized is only one of ways.
Like I said, you can do it field by field like in C.

>>106593058
Doesn't work on stable though.

>>106593252
On stable you have to construct Layout manually.
Why does it matter though? Your original claim has nothing to do with release channel.

>>106593252
Nothing works in C without tinkertrannying with macros to hide the ugly truth.
C is the real tranny language.

>>106593289
>noo you cant just transform text
>you have to type the text
I have God-given freedom to program as I want
why are commies always such control freaks

>>106593402
>why are commies always such control freaks
Why are we suddenly talking about GNU?

>Anonymous 09/15/25(Mon)16:03:40 No.106593402
>>>106593289 (You)
>>noo you cant just transform text
>>you have to type the text
>I have God-given freedom to program as I want why are commies always such control freaks
>#define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + __must_be_array(arr))
#define __must_be_array(a) BUILD_BUG_ON_ZERO(__same_type((a), &(a)[0]))
#define __same_type(a, b) __builtin_types_compatible_p(typeof(a), typeof(b))

>>106593422
Go back to sharty

In C++ it's just
template <unsigned long long N, typename T>
unsigned long long array_size(T const (&array)[N]) noexcept
{
return N;
}

>>106593271
Why do you lie?
from_raw_parts also doesn't exist on stable, and unlike Layout, there is no substitute. You're stuck using an unstable feature that can break or be removed at any moment.

>>106593449
>there is no substitute
let layout = Layout::new::<usize>()
.extend(Layout::for_value(arr))
.unwrap()
.0
.pad_to_align();

>Why do you lie?
What??
Why do you instantly jump to the conclusion that I am lying despite you having very little knowledge about the topic that is being discussed?
Why are you so overconfident in your own ignorance and seemingly argue everything in bad faith?

>>106593497
How many bounds checks does this perform lol.

Oh, and if you meant the ptr::from_raw_parts then you can just
let this = ptr::slice_from_raw_parts_mut::<u8>(this, arr.len()) as *mut Thing;

to copy the metadata in stable channel.

>>106593528
What? Why would declaring a layout require any bounds checks?

>>106593536
Why would something unchecked require result value?
Not that other anon who is too cnile to even understand what did I mean by this by the way.

>>106593563
>Why would something unchecked require result value?
>https://doc.rust-lang.org/beta/std/alloc/struct.Layout.html#method.extend
>On arithmetic overflow, returns LayoutError.
Theoretically, the code I posted before had UB because it didn't check for potential overflows, but in practice you couldn't really pass an array to new() of size anywhere close to isize::MAX anyway.

>>106592902
Ok I know io_uring is cool, but backends need to schedule handlers for those incoming requests.

>>106593600
Writing 64 bytes and incrementing atomic pointer is so fucking hard fr fr on god this is too complex for a backend, we need green threads ASAP SAAAR.

>>106593595
So it is checked and you're coping.
>a fucking check for usize overflow
clown world we live in

>>106593643
>So it is checked and you're coping.
But you can construct it in unchecked way if you wanted...
Layout::from_size_align_unchecked

>>a fucking check for usize overflow
>clown world we live in
It's not hard to imagine situation where it would matter
>have a Self::new_with_len(len: usize) constructor
>attacker exploits some other code to call this with usize::MAX
>the Layout size wraps around allocating only the sized part so the allocation succeeds
>the metadata in pointer still says it has usize::MAX length, therefore everything is in bound
>an attacker can now manipulate anything in the memory without triggering out of bounds checks

Why do you keep moving the goalpost? It seems so desperate.

>>106593689
Because I'm not that anon and you're coping very hard.
Why would this stuff be dynamic lmao.
Holy shit, Rust is truly a webshitter language, preparing for dynamic structs that can be exploited remotely before C++ even.

>>106593701
>Why would this stuff be dynamic lmao.
Why Dynamically Sized Types are dynamic? What?

>you're coping very hard.
???

>>106593717
my bad, I forgot that >>106590217 was in the other thread.
I once again ask, usecase?

>>106593725
>I once again ask, usecase?
I only once found DST useful.
// WGSL
struct Model {
size: vec4i,
palette: array<Material, 256>,
voxels: array<u32>,
}

I used DSTs in GPU to represent voxel data and to make uploading that data simpler and faster between CPU/GPU, I had to represent them in Rust in the same way as GPU does. It made more sense to do it this way instead of splitting it into multiple storage buffer objects.
It is a niche kind of data structure that has niche uses. That's why there wasn't much push to polish its semantics on the language level.

>>106593812
I'd just use unsigned char*, the way I did in example written in >>106592481
Because it's niche, I can afford to write a custom class for it, and it would work just fine.

>>106593836
That's a valid and somewhat safer option.
On Rust side, it looks like this
#[repr(C)]
pub struct Model {
pub width: u32,
pub height: u32,
pub depth: u32,
_pad: u32,
pub palette: [Material; 256],
pub voxels: [u8],
}

I like this approach better because I can just directly access all the fields. But using getter/setter is fine too.

I gotta finish that raytracer one day.

WTF is wrong with python devs again. How am I supposed to integrate this "fix" into CI/CD? Write a hacky fragile build step just because they were to lazy to patch an old version?

>>106594741
https://github.com/ajalt/fuckitpy

>>106594875
>Still getting errors? Chain fuckit calls. This module is like violence: if it doesn't work, you just need more of it.

>>106594875
Kek. I did something like that for JS some time ago.
But yeah, it wouldn't resolve some problems down the dependency chain. Pinning the version of package this package was depending on did solved the issue though.
I have no idea why do they recommend modifying library files instead of just pinning the version.

>>106592223
async/await is literally the best solution.
it has the a similar feel to generic stack-like procedural programming, while being zero-cost abstraction. green threads cause problems in every single way imaginable:
FFI problems
depend on runtime to schedule them
need to figure out when to context switch, either on special marked "blocking" calls or some other preemptive bs way

cgo and golang wasm outputs should be pretty much the ultimate redpill to why green threads are dumb. maybe if your users are fucking stupid and your language has crippled semantics that make a proper async executor intractable, then it's fine, like Go or Java, but really, why? just fucking await your asyncs you tard monkey.

>>106594989
>need to figure out when to context switch, either on special marked "blocking" calls or some other preemptive bs way
True, there are places where it is not safe to suspend. To add to this, I read a bit about project loom. They plan to fix compatability with old blocking code by rewriting the blocking calls to use the non-blocking alternatives, which feels very brittle imo.
>need to figure out when to context switch, either on special marked "blocking" calls or some other preemptive bs way
Not exactly sure, but doesn't async/await also require a scheduler for the futures? I've seen this everywhere except node (no multithreading so has to use event loop)
Doesn't this make them the same as cooperative green threads?

>>106594989
async/await is not zero-cost

>>106595665
based on what? do you know the definition of zero-cost abstraction?

>>106595670
>definition of zero-cost abstraction
yes, it's when abstraction has a non-zero cost but it makes it easier for jeets so it's free because their labor is cheap

>>106595670
NTA, but it depends on implementation. JS-style callback based async/await is not free by any means.

>>106595684
kek, wtf? is this board deranged or what?

>>106595698
It doesn't. Async/await requires extra memory just by virtue of there being something like a "Promise" object that constantly takes up memory, memory that should not be used because who the fuck gives a shit about having exact object? All you need is a type of event + index into a preallocated array where stuff that deserves memory is allocated ahead of time.
Yes, I just described how io_uring is used by people who aren't mentally ill. I tried to implement async/await on top and it's not zero cost, it requires extra memory just to maintain exact parity between request and response (from the OS, obviously, what else would require async/await?)

Hello /g/. I am from India and I want to move to Balkans. Preferably Croatia, but Serbia also okay. How to?

>>106592223
both of them are unnecessary when everything can be done with a simple event loop
for io requests submit to os and wait for completion event
for compute jobs submit to a thread pool and wait for completion event

>>106595670
Based on your dick bro

>>106594741
ywnbaw

>>106595820
What?

>>106595800
If it was just like his dick, the cost would be acceptable, but the cost is too big...

>>106595721
seriously this board is full fucking retard. I swear to christ.
the memory is in use no matter what, what the fuck do you think a promise/future is? it's basically a fucking stack frame stored somewhere to be re-executed at some later date. You can't magically remove this cost by hand waving it away.

>>106595721
>Async/await requires extra memory just by virtue of there being something like a "Promise" object that constantly takes up memory, memory that should not be used because who the fuck gives a shit about having exact object?
This also depends on implementation. For example in Rust, tasks compile into state machines that just describe how to proceed with the procedure on event. This is essentially what you would have to write anyway if you would want to support such resumable operation that reacts on external events.

>All you need is a type of event + index into a preallocated array where stuff that deserves memory is allocated ahead of time.
If I understand you correctly, that's literally how Rust's embassy works for example.

>I just described how io_uring is used by people who aren't mentally ill. I tried to implement async/await on top and it's not zero cost it requires extra memory just to maintain exact parity between request and response
Dunno how io_uring works, but maybe it's not zero cost because you are trying to combine two techniques that are not meant to be used together in zero cost manner? When something is said to be zero cost, it means it doesn't incur extra cost over a reasonable hand rolled solution, it doesn't mean it will be compatible with whatever random library/api/technique you choose.

>from the OS, obviously, what else would require async/await
I primary use async/await for bare metal embedded development.

>>106595846
async/await adds extra memory usage for no benefit whatsoever, autismo.

>>106591526
Your code is fucking ugly man. Wtf

>>106595855
My handwritten state machine uses less memory and is faster.
>>106595862
And?

>>106595857
>no reason
holy shit you're fucking retarded.

spawn thread? at least 2MiB stack, spawn "green thread?" some size (likely much smaller) that "grows."
async/await? exact sized future with all the state you carry over awaits.

>>106595857
NTA, it is beneficial to me. It allows me to have concurrency without having to include entire RTOS bloat on my chip.

>>106595870
>My handwritten state machine uses less memory and is faster.
[benchmark needed]

>>106595870
and it's not generalized so it's worthless.

>>106594895
>tranime
fuck off

>>106595872
My threads have stacks that are sized by however much I specified, not 2MiB.
>>106595881
>benchmark needed
What are you gonna benchmark, Linux syscalls? Because my code has 0 unnecessary cycles wasted on abstractions.
>>106595885
>muh generalization
I can tell that you want to obsolete yourself by getting fired for making something that can be copy pasted to solve any problem, but I personally am a hobbyist who only cares about top performance within constraints that are well defined.

>>106595893
>tranime
Newfag detected

>>106595893
>Anonymous 09/15/25(Mon)20:43:33 No.106595893
>>>106595900>>106594895
>>tranime
>fuck off

>>106595893
Look at the OP, the sticky, the banner, the 404 page, then fuck off by getting yourself banned so you can look at the banned page.

>>106595897
>What are you gonna benchmark
Whenever your concurrency/asynchronicity/resumable procedures implementation based on hand written state machines uses less memory and is faster than the official async/await implementation for your language of choice.

>>106578439
Eli will never be a real woman either :(

>>106595940
Consuming CQEs uses as much memory as a typical fizzbuzz for loop by the way

The world would be measurably better if everyone from sharty and their families were gassed, cremated and their ashes used to fertilize trees.

>>106595965
why would a sharty tourist say this?

>>106595974
Self-awareness.

>>106588294
Pajeetware

>>106588510
Thank you Linus Torvalds for making syscalls so fast that jeets can do this...

NEW >>106596041

>>106587895
I still don't fully understand erlang/elixir differences so thanks for the extra info, especially about the imports.

NEW THREAD
>>106596041
>>106596041
>>106596041

>>106595665
>async/await is not zero-cost
uhh but C++ has std::async and co_await? are you saying bjarne lied?