re: CheckSoftSNData
I love the use of rust as the decomp target. If you want something do-while shaped but still valid Rust, you can do:
while {
i = &i[1];
if *(i as *mut u8) as u32 != 0 {
r1 = 0;
}
i != ((&var_204).byte_offset(3) as *mut u32)
} {}
I love the use of rust as the decomp target. If you want something do-while shaped but still valid Rust, you can do:
while {
i = &i[1];
if *(i as *mut u8) as u32 != 0 {
r1 = 0;
}
i != ((&var_204).byte_offset(3) as *mut u32)
} {}
January 24, 2026 at 12:20 AM
re: CheckSoftSNData
I love the use of rust as the decomp target. If you want something do-while shaped but still valid Rust, you can do:
while {
i = &i[1];
if *(i as *mut u8) as u32 != 0 {
r1 = 0;
}
i != ((&var_204).byte_offset(3) as *mut u32)
} {}
I love the use of rust as the decomp target. If you want something do-while shaped but still valid Rust, you can do:
while {
i = &i[1];
if *(i as *mut u8) as u32 != 0 {
r1 = 0;
}
i != ((&var_204).byte_offset(3) as *mut u32)
} {}
OsRng.try_fill_bytes to get 1M u32's would probably get you a lot closer to that if you wanted, mildly funny to turn the u8s into u32s but Welcome To The Fast Zone
January 23, 2026 at 12:30 AM
OsRng.try_fill_bytes to get 1M u32's would probably get you a lot closer to that if you wanted, mildly funny to turn the u8s into u32s but Welcome To The Fast Zone
there's a hardcoded 32-byte AES key the firmware uses to encrypt passwords... except they fucked up (made an array of u32 and not u8) and 3/4 of the key is zeroes
not that it matters a lot given it's hardcoded
not that it matters a lot given it's hardcoded
January 23, 2026 at 8:38 PM
there's a hardcoded 32-byte AES key the firmware uses to encrypt passwords... except they fucked up (made an array of u32 and not u8) and 3/4 of the key is zeroes
not that it matters a lot given it's hardcoded
not that it matters a lot given it's hardcoded
That's why I usually avoid functional notation for min/max and explicitly write u32::min(x, y). In that context the name makes much more sense.
January 16, 2026 at 12:37 AM
That's why I usually avoid functional notation for min/max and explicitly write u32::min(x, y). In that context the name makes much more sense.
ABC439おわりっ！45分42秒でABCDE問題をノーペナ5完！
A問題はそのまま計算
2^Nの部分はビットシフトで記述しても良かったな…（2_u32.pow(n)で記述しちゃった）
B問題全然わからんからゴリ押し
置き換えの履歴をHashSetで保存しておいて、操作後にHashSet内の値になったらループに入ってることが確定するのでNo
B問題にしては難しすぎないか……？
C問題はよく見るとほぼ全探索できるやつ
1≦x＜y＜√Nの範囲で探索すればいいから、計算量はO(N)
A問題はそのまま計算
2^Nの部分はビットシフトで記述しても良かったな…（2_u32.pow(n)で記述しちゃった）
B問題全然わからんからゴリ押し
置き換えの履歴をHashSetで保存しておいて、操作後にHashSet内の値になったらループに入ってることが確定するのでNo
B問題にしては難しすぎないか……？
C問題はよく見るとほぼ全探索できるやつ
1≦x＜y＜√Nの範囲で探索すればいいから、計算量はO(N)
January 3, 2026 at 1:41 PM
ABC439おわりっ！45分42秒でABCDE問題をノーペナ5完！
A問題はそのまま計算
2^Nの部分はビットシフトで記述しても良かったな…（2_u32.pow(n)で記述しちゃった）
B問題全然わからんからゴリ押し
置き換えの履歴をHashSetで保存しておいて、操作後にHashSet内の値になったらループに入ってることが確定するのでNo
B問題にしては難しすぎないか……？
C問題はよく見るとほぼ全探索できるやつ
1≦x＜y＜√Nの範囲で探索すればいいから、計算量はO(N)
A問題はそのまま計算
2^Nの部分はビットシフトで記述しても良かったな…（2_u32.pow(n)で記述しちゃった）
B問題全然わからんからゴリ押し
置き換えの履歴をHashSetで保存しておいて、操作後にHashSet内の値になったらループに入ってることが確定するのでNo
B問題にしては難しすぎないか……？
C問題はよく見るとほぼ全探索できるやつ
1≦x＜y＜√Nの範囲で探索すればいいから、計算量はO(N)
god finding the right formula for this took YEARS off my life
it's really annoying how simple it is… completely belying how i spent like 3 hours confusing myself trying to get it to work
it's really annoying how simple it is… completely belying how i spent like 3 hours confusing myself trying to get it to work
![/// Convert from bitwise pixel data (`0b00110110`), as used in the
/// XBM file format and this program's Bitmap structs, to nybble-wise
/// (`0x00 0x11 0x01 0x10`), as used in the BMP file format.
fn binary_to_nybbles(x: u8, null: &mut [u8; 4]) {
for i in 0..8 {
let cellindex = i / 2;
let j = i as u32 + 1;
let assign = x.rotate_left(5 * j) & 1u8.rotate_left(4 * j);
null[cellindex] |= assign;
}
}](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:yatiw4nkhbnyzvk32h7q7r6i/bafkreibigzvnx4r3wtdj5vopchd3zsaisxj6r4yil2xnz27fc7kekxvcri@jpeg)
December 20, 2025 at 7:16 PM
god finding the right formula for this took YEARS off my life
it's really annoying how simple it is… completely belying how i spent like 3 hours confusing myself trying to get it to work
it's really annoying how simple it is… completely belying how i spent like 3 hours confusing myself trying to get it to work
Does it make any difference what the underlying type is? For example, does the complexity of the wrapped type play a role—am I less likely to see an impact from wrapping u32 than a large struct from std?
December 15, 2025 at 9:23 PM
Does it make any difference what the underlying type is? For example, does the complexity of the wrapped type play a role—am I less likely to see an impact from wrapping u32 than a large struct from std?
📹 María Corina Machado llega a Oslo después de más de un año en la clandestinidad social.elpais.com/u32…
December 11, 2025 at 6:17 AM
📹 María Corina Machado llega a Oslo después de más de un año en la clandestinidad social.elpais.com/u32…
half of part 2 was stubbornness to the bruteforce half of it was me forgetting that u32 sucks and i can just fucking use u64
December 7, 2025 at 7:06 AM
half of part 2 was stubbornness to the bruteforce half of it was me forgetting that u32 sucks and i can just fucking use u64
Fwupd 1.9.19 Linux Firmware Updater Supports Acer U32 and Luxshare 7-in-1 Docks
9to5linux.com/fwupd-1-9-19...
#Infosec #Security #Cybersecurity #CeptBiro #Fwupd1.9.19 #Linux #FirmwareUpdater Supports #AcerU32 #Luxshare7in1Docks
9to5linux.com/fwupd-1-9-19...
#Infosec #Security #Cybersecurity #CeptBiro #Fwupd1.9.19 #Linux #FirmwareUpdater Supports #AcerU32 #Luxshare7in1Docks
Fwupd 1.9.19 Linux Firmware Updater Supports Acer U32 and Luxshare 7-in-1 Docks - 9to5Linux
fwupd 1.9.19 Linux firmware updater is now available for download with support for the Acer U32 and Luxshare 7-in-1 docks.
9to5linux.com
May 3, 2024 at 1:13 PM
Fwupd 1.9.19 Linux Firmware Updater Supports Acer U32 and Luxshare 7-in-1 Docks
9to5linux.com/fwupd-1-9-19...
#Infosec #Security #Cybersecurity #CeptBiro #Fwupd1.9.19 #Linux #FirmwareUpdater Supports #AcerU32 #Luxshare7in1Docks
9to5linux.com/fwupd-1-9-19...
#Infosec #Security #Cybersecurity #CeptBiro #Fwupd1.9.19 #Linux #FirmwareUpdater Supports #AcerU32 #Luxshare7in1Docks
i mean yeah when T: u32, you can’t have conflicting AsRef<u32> impls. AsRef isn’t for field access; write field accessors instead. it’s for subtyping in a way that the lang team doesn’t get sad about
March 20, 2025 at 1:49 PM
i mean yeah when T: u32, you can’t have conflicting AsRef<u32> impls. AsRef isn’t for field access; write field accessors instead. it’s for subtyping in a way that the lang team doesn’t get sad about
behold: my best attempt at printing the first 10,000 primes in rust, lua, fennel and haskell
![local function next_prime(primes)
local n = primes[#primes]
local n_is_prime
repeat
n = n + 1
n_is_prime = true
for _, p in ipairs(primes) do
if p * p > n+1 then
break
end
if n % p == 0 then
n_is_prime = false
break
end
end
until n_is_prime
return n
end
local function gen_primes(count)
local primes = {2}
for _ = 2, count do
local next = next_prime(primes)
table.insert(primes, next)
end
return primes
end
local primes = table.concat(gen_primes(10000), ", ")
print(primes)](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:zunc3ki5hdrgqadatljl7znn/bafkreigas75xgmekvzubaozeqsnv4mitt3cg3nvttqmtkcsa6duu6zn4ku@jpeg)
![(fn prime? [n i primes]
(let [p (. primes i)]
(if (= (% n p) 0)
false
(> (* p p) n)
true
(prime? n (+ i 1) primes))))
(fn next-prime [primes ?n]
(let [n (or ?n (+ 1 (. primes (length primes))))]
(if (prime? n 1 primes)
n
(next-prime primes (+ n 1)))))
(fn gen-primes [count]
(var primes [2])
(for [_ 2 count]
(table.insert primes (next-prime primes)))
primes)
(local primes (table.concat (gen-primes 10000) ", "))
(print primes)](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:zunc3ki5hdrgqadatljl7znn/bafkreibgtskxh6q2moa4vmesjepl4qba3d433w5ekzmzjttovhkbdt3f5i@jpeg)
![import Data.List
import Data.Maybe
isPrime n p's = all (\p -> n `mod` p /= 0)
(takeWhile (\p -> p^2 <= n) (reverse ps))
nextPrime :: [Int] -> Int
nextPrime [] = 2
nextPrime p's = fromJust (find (\n -> isPrime n p's) [(head p's + 1)..])
primes = unfoldr (\list ->
let next = nextPrime list in Just (next, next : list)) []
primeString x = intercalate ", " (map show (take x primes))
main = putStrLn (primeString 10000)](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:zunc3ki5hdrgqadatljl7znn/bafkreiesxpanazwbjlrc5daqejdjscf47db243fuit2yxflxlrdfqjcqja@jpeg)
January 7, 2025 at 12:52 AM
behold: my best attempt at printing the first 10,000 primes in rust, lua, fennel and haskell
also it was really neat to write a couple arm instruction decoders for this to figure out relative addresses, `ADRP` and `ADD` my beloved. quite a big difference from x86's usual read<u32>(insn + 3) :3
February 26, 2025 at 5:30 PM
also it was really neat to write a couple arm instruction decoders for this to figure out relative addresses, `ADRP` and `ADD` my beloved. quite a big difference from x86's usual read<u32>(insn + 3) :3
desse post o problema é a struct consumir mais memória do que deveria
se os u32 tivessem em sequência não precisaria de padding entre os campos
se os u32 tivessem em sequência não precisaria de padding entre os campos
January 4, 2025 at 9:49 AM
desse post o problema é a struct consumir mais memória do que deveria
se os u32 tivessem em sequência não precisaria de padding entre os campos
se os u32 tivessem em sequência não precisaria de padding entre os campos
I still think it's pretty cool I managed to get this working at all
```rust
// Allows adding two `i32`, `u32`, or `f32` numbers
let reflect_add = add::<i32>
.into_function()
.with_overload(add::<u32>)
.with_overload(add::<f32>);
```
```rust
// Allows adding two `i32`, `u32`, or `f32` numbers
let reflect_add = add::<i32>
.into_function()
.with_overload(add::<u32>)
.with_overload(add::<f32>);
```
March 30, 2025 at 3:36 AM
I still think it's pretty cool I managed to get this working at all
```rust
// Allows adding two `i32`, `u32`, or `f32` numbers
let reflect_add = add::<i32>
.into_function()
.with_overload(add::<u32>)
.with_overload(add::<f32>);
```
```rust
// Allows adding two `i32`, `u32`, or `f32` numbers
let reflect_add = add::<i32>
.into_function()
.with_overload(add::<u32>)
.with_overload(add::<f32>);
```
So far I'm not able to reproduce this with random data and u16 or u32 hashes.
How many kmers are in the read sets that you test on?
Otherwise I should adjust to testing on real data as well, with more interesting collections of kmers.
How many kmers are in the read sets that you test on?
Otherwise I should adjust to testing on real data as well, with more interesting collections of kmers.
August 4, 2025 at 7:06 PM
So far I'm not able to reproduce this with random data and u16 or u32 hashes.
How many kmers are in the read sets that you test on?
Otherwise I should adjust to testing on real data as well, with more interesting collections of kmers.
How many kmers are in the read sets that you test on?
Otherwise I should adjust to testing on real data as well, with more interesting collections of kmers.
Finally, the code for TileRgbaSemiFogAcc:
![/// Accumulator for tile rendering, using Rgba colors and semi-fog effect for alpha over blending.
///
/// This is used to compose colors from triangles into a final color.
pub struct TileRgbaSemiFogAcc<const TILE_SIZE: usize, const ACC: usize> {
pub buf: [[[(f32, Rgba); ACC]; TILE_SIZE]; TILE_SIZE],
pub len: [[u8; TILE_SIZE]; TILE_SIZE],
}
impl<const TILE_SIZE: usize, const ACC: usize> Acc for TileRgbaSemiFogAcc<TILE_SIZE, ACC> {
type Data = ();
type In = Rgba;
type Out = Rgba;
fn new(_: ()) -> Self {
Self {
buf: [[[(0.0, [0.0; 4]); ACC]; TILE_SIZE]; TILE_SIZE],
len: [[0; TILE_SIZE]; TILE_SIZE],
}
}
fn clear(&mut self) {
self.buf = [[[(0.0, [0.0; 4]); ACC]; TILE_SIZE]; TILE_SIZE];
self.len = [[0; TILE_SIZE]; TILE_SIZE];
}
fn upd(&mut self, i: u32, j: u32, depth: f32, color: Rgba) {
let len = {
let len = &mut self.len[j as usize][i as usize];
if *len as usize >= ACC {return};
*len += 1;
*len
};
self.ins(len, i, j, depth, color);
}
fn acc(&self, i: u32, j: u32) -> Rgba {
let len = self.len[j as usize][i as usize];
let buf = &self.buf[j as usize][i as usize];
let mut color = [0.0; 4];
let mut fog = FogState::None;
for k in 0..len {
let (d, c) = buf[k as usize];
fog = fog.update(d, c);
fog.acc_alpha_blend_over(d, c, &mut color);
}
fog.acc_end_alpha_blend_over(&mut color);
color
}
}
impl<const TILE_SIZE: usize, const ACC: usize> TileRgbaSemiFogAcc<TILE_SIZE, ACC> {
pub fn ins(&mut self, len: u8, i: u32, j: u32, mut depth: f32, mut color: Rgba) {
let buf = &mut self.buf[j as usize][i as usize];
for k in 0..len {
let (d, c) = buf[k as usize];
if d == 0.0 {
buf[k as usize] = (depth, color);
return;
} else if …](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:olippfg4rukwzrkxsdwpd5bc/bafkreic2ll7vc5qklo4thsdnj7hsynbnshokpdqwdughdf2eyuyfrzs37m@jpeg)
November 23, 2025 at 2:32 PM
Finally, the code for TileRgbaSemiFogAcc:
(@mtgamesgnm.bsky.social)
VEE, G!
...
Not The Best.
But This Will Do. (I Hope...)
Because I Had Left This On My File For A Long Time Due To Me Playing WWE 2K24.
So... Yeah.
#StickNodes #StickNodesPro
#StickFigure #StickFigures
#Human #HumanOC #HumanAU
#Vee #VeeG
#U32 #Universe32
#OC #AU
VEE, G!
...
Not The Best.
But This Will Do. (I Hope...)
Because I Had Left This On My File For A Long Time Due To Me Playing WWE 2K24.
So... Yeah.
#StickNodes #StickNodesPro
#StickFigure #StickFigures
#Human #HumanOC #HumanAU
#Vee #VeeG
#U32 #Universe32
#OC #AU
September 11, 2024 at 1:10 PM
(@mtgamesgnm.bsky.social)
VEE, G!
...
Not The Best.
But This Will Do. (I Hope...)
Because I Had Left This On My File For A Long Time Due To Me Playing WWE 2K24.
So... Yeah.
#StickNodes #StickNodesPro
#StickFigure #StickFigures
#Human #HumanOC #HumanAU
#Vee #VeeG
#U32 #Universe32
#OC #AU
VEE, G!
...
Not The Best.
But This Will Do. (I Hope...)
Because I Had Left This On My File For A Long Time Due To Me Playing WWE 2K24.
So... Yeah.
#StickNodes #StickNodesPro
#StickFigure #StickFigures
#Human #HumanOC #HumanAU
#Vee #VeeG
#U32 #Universe32
#OC #AU
Ever bothered by binary search being slow?
I spent some time optimizing static B+-trees as explained by algorithmica, and got 35x increase in throughput over plain binary search, from 1200ns/query to 35ns/query, on a 4GB array of u32.
Another blog to write :)
en.algorithmica.org/hpc/data-str...
I spent some time optimizing static B+-trees as explained by algorithmica, and got 35x increase in throughput over plain binary search, from 1200ns/query to 35ns/query, on a 4GB array of u32.
Another blog to write :)
en.algorithmica.org/hpc/data-str...
Static B-Trees - Algorithmica
en.algorithmica.org
December 11, 2024 at 10:41 PM
Ever bothered by binary search being slow?
I spent some time optimizing static B+-trees as explained by algorithmica, and got 35x increase in throughput over plain binary search, from 1200ns/query to 35ns/query, on a 4GB array of u32.
Another blog to write :)
en.algorithmica.org/hpc/data-str...
I spent some time optimizing static B+-trees as explained by algorithmica, and got 35x increase in throughput over plain binary search, from 1200ns/query to 35ns/query, on a 4GB array of u32.
Another blog to write :)
en.algorithmica.org/hpc/data-str...
今日の東北新幹線E5系U32やまびこ🚄✨
何気に、今年初めてU32編成に会ったかも💦💦
何気に、今年初めてU32編成に会ったかも💦💦
July 22, 2025 at 10:45 AM
今日の東北新幹線E5系U32やまびこ🚄✨
何気に、今年初めてU32編成に会ったかも💦💦
何気に、今年初めてU32編成に会ったかも💦💦
When I draw e.g. a button, I just use a function that I call manually, that uses generic code on top of Piston2d-Graphics:
![pub fn draw_button<T: UiTypes + UiButton, G: Graphics>(
ui: &T,
panel_pos: [u32; 2],
button: T::Button,
texture_data: &TextureData<G::Texture>,
c: &Context,
g: &mut G
) {
use graphics::*;
let button_visible = ui.button_visible(button);
if !button_visible {return}
let button_enabled = ui.button_enabled(button);
let button_pos = ui.button_pos(button);
let button_size = ui.button_size(button);
let pos = [
(panel_pos[0] + button_pos[0]) as f64,
(panel_pos[1] + button_pos[1]) as f64,
];
let size = [
button_size[0] as f64,
button_size[1] as f64,
];
let bg_color = ui.button_bg_color(button);
rectangle(bg_color, [
pos[0], pos[1], size[0], size[1],
], c.transform, g);
if let Some(ind) = ui.button_icon(button) {
image(&texture_data.icons[*ind], c.transform.trans_pos(pos), g);
}
let border = 2.0;
let alpha = 0.5;
// Shadow.
rectangle([0.0, 0.0, 0.0, alpha], [
pos[0] + size[0] - border, pos[1], border, size[1],
], c.transform, g);
rectangle([0.0, 0.0, 0.0, alpha], [
pos[0], pos[1] + size[1] - border, size[0], border
], c.transform, g);
// Brightness.
rectangle([1.0, 1.0, 1.0, alpha], [
pos[0], pos[1], size[0], border,
], c.transform, g);
rectangle([1.0, 1.0, 1.0, alpha], [
pos[0], pos[1], border, size[1],
], c.transform, g);
if !button_enabled {
rectangle([bg_color[0], bg_color[1], bg_color[2], 0.5], [
pos[0], pos[1], size[0], size[1],
], c.transform, g);
}
}](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:olippfg4rukwzrkxsdwpd5bc/bafkreic4bhx3b6vgbjlhb5oeyugoyuu375jlosgpphwhjp6byszwdjxjzq@jpeg)
August 2, 2025 at 11:20 PM
When I draw e.g. a button, I just use a function that I call manually, that uses generic code on top of Piston2d-Graphics:
if you just want the current executable's path, you can use `std::env::current_exe`, which will call `GetModuleFileNameW` behind the scenes. make sure to use a structure to ser/de from, instead of relying on TOML values directly. you can also use `context` from `anyhow` to annotate errors
![#[derive(Serialize, Deserialize)]
struct Config {
test: u32,
}
const DEFAULTS: Config = Config { test: 5 };
static FULL_CONFIG: LazyLock<Config> = LazyLock::new(|| {
std::env::current_dir()
.context("Failed to get current directory")
.and_then(|path| {
Ok(path
.parent()
.context("The DLL doesn't have a parent directory!")?
.join("config.toml"))
})
.and_then(|config_path| Ok(fs::read_to_string(config_path)?))
.and_then(|config| Ok(toml::from_str::<Config>(config.as_str())?))
.unwrap_or_else(|error| {
println!("Reverting to defaults due to error. ERROR: {error}");
DEFAULTS
})
});](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:wamidydbgu3u6fk3yckaglnz/bafkreihvkdl2xouwfuytnppk3bx35yp7737iqgukhq7w2id6h7bwoyl4mq@jpeg)
September 28, 2025 at 2:56 PM
if you just want the current executable's path, you can use `std::env::current_exe`, which will call `GetModuleFileNameW` behind the scenes. make sure to use a structure to ser/de from, instead of relying on TOML values directly. you can also use `context` from `anyhow` to annotate errors
got the most basic excuse for shape compositing working then added a second non system* library. now I can use the command list from microui as renderer TODOs.
*sdl might as well be a system lib on linux, but it's only a fallback till all the native platforms are done
*sdl might as well be a system lib on linux, but it's only a fallback till all the native platforms are done
November 29, 2025 at 5:20 PM
got the most basic excuse for shape compositing working then added a second non system* library. now I can use the command list from microui as renderer TODOs.
*sdl might as well be a system lib on linux, but it's only a fallback till all the native platforms are done
*sdl might as well be a system lib on linux, but it's only a fallback till all the native platforms are done
Non, débordement arithmétique. En debug, ce code panique:
let mut x: u32 = 4294967295; // = 2³² - 1
println!("x: {x}"); // OK
x += 1; // Panique ici
println!("x: {x}"); // Jamais exécuté
let mut x: u32 = 4294967295; // = 2³² - 1
println!("x: {x}"); // OK
x += 1; // Panique ici
println!("x: {x}"); // Jamais exécuté
December 23, 2024 at 6:26 PM
Non, débordement arithmétique. En debug, ce code panique:
let mut x: u32 = 4294967295; // = 2³² - 1
println!("x: {x}"); // OK
x += 1; // Panique ici
println!("x: {x}"); // Jamais exécuté
let mut x: u32 = 4294967295; // = 2³² - 1
println!("x: {x}"); // OK
x += 1; // Panique ici
println!("x: {x}"); // Jamais exécuté
Hmm. Single precision float (f32) can represent ±2^24. Double precision (f64) up to ±2^53. So i16/u16 -> f32 is lossless, and i32/u32 -> f64. Or larger, if appropriately bounded. Or maybe as programming languages gradually develop support for arbitrary-width integers one could just reach for an i54.
January 24, 2025 at 10:27 PM
Hmm. Single precision float (f32) can represent ±2^24. Double precision (f64) up to ±2^53. So i16/u16 -> f32 is lossless, and i32/u32 -> f64. Or larger, if appropriately bounded. Or maybe as programming languages gradually develop support for arbitrary-width integers one could just reach for an i54.