programming

[olive sable]

i think c++ and python are both fine for this purpose

c++ will use less recources tho

[sharp stream]

How long do you think I’d have to code for a 24/7 monitoring system with like 350 inputs and outputs for a website?

[olive sable]

uhhh

[fast pagoda]

dataprocessing is kinda python's bread and butter

[olive sable]

its easier inpython

[sharp stream]

So python for data processing and C++ for functions and operations?

[fast pagoda]

generally for ease of implementation reasons yeah

[olive sable]

i wouldnt say its better tho

[tender river]

old website want to update

but

too much effort

[sage crag]

all the libraries are written in c

[tender river]

enjoy my 2022 writing

[sage crag]

no meaningful performance effect for website made by random person

average of 0.3 visitor a day

[olive sable]

true

[fast pagoda]

you say that like it's old 2022 was like la- oh

[sharp stream]

How do you get 3/10 of a person visiting your site?

[fast pagoda]

children

[olive sable]

a day, so 3 per 10 days

[fast pagoda]

they only count as 1/3rd of a person if theyre only 1/3rd the height

[sharp stream]

Me when

[rigid snow]

#programming message you did say rust > C as first lang, and C is objectively good as a first lang, which makes rust good as a first lang

i pulled up the logs

[fast pagoda]

r ecieipts

[sharp stream]

So then…What is Tutel?

[olive sable]

height isnt a good scale, you're better of with volume

[sharp stream]

1/90th?

[fast pagoda]

british

so not legally classifiable

[sharp stream]

Based

[olive sable]

[fast pagoda]

he exists outside the spectrum of things that can be defined

[olive sable]

the british catching strays again

[sharp stream]

Outside the spectrum

[olive sable]

what did they do to deserve this?

[sharp stream]

They drank too much tea

👍

[fast pagoda]

there's thousands of years of reasons that one could cite so pick anything from any history book

worst of all they indirectly caused my existence by founding burgerland

[sharp stream]

Exhibit A: Stolen Egyptian artifacts still on display in museums

[olive sable]

ok but so does every other country

[fast pagoda]

yes

which is why it's ok to randomly make fun of any country for a laugh as long as you're not constantly shitting on them

[tender river]

pick
want to scream at compiler -> rust
want to scream at yourself -> python
want to scream at program -> c++

[olive sable]

well thats the problem tho, people are

[fast pagoda]

define people and cite sources indicating it's an outsized phenomenon compared to any other country of similar status

[olive sable]

i mean, i dont see anyone hating on belgium

[fast pagoda]

of similar status

and influence

[olive sable]

bruh

belgium is the capital of europe

[opaque sigil]

Nice one

[olive sable]

i dont hink we're that far behind the uk

[rigid snow]

he's right tho

[opaque sigil]

[opaque sigil]

Brussels may as well be its own country

[rigid snow]

that can be said about lots of places

[opaque sigil]

[olive sable]

i agree, but that doesnt mean it is its own country

[fast pagoda]

i have nothing against belgium but from a worldwide influence & general mindshare perspective i apologize but it's not even a contest

[rigid snow]

this is true too

[olive sable]

i mean, if belgium kicked the bucket they'd jsut move everything to france or germany, but for now it is decently important

[fast pagoda]

again i'm more speaking towards influence outside of europe && cultural export

[rigid snow]

nuance

[opaque sigil]

We don't do nuance here

[fast pagoda]

in general i'd even say belgium's perception is overall more positive

[olive sable]

ok forget belgium

what about germany

[fast pagoda]

similar but doesnt have the world reach

[olive sable]

id argue it does

[opaque sigil]

[fast pagoda]

how many countries have german colonial history

[olive sable]

compared to the hate the french get germany should also get a lot of hate, but it doesn't

[opaque sigil]

We're not french that's why

Duh

[olive sable]

see

[tender river]

in terms of colonies nobody beats uk and france

[sage crag]

actually

i have

[real sierra]

[sage crag]

many

you may not know them they go to a different planet

[tender river]

tv colonies

[olive sable]

so the amount of hate is proportional to the amount of colonies then?

[real sierra]

guys?

[fast pagoda]

significant effect on overall cultural consciousness

[olive sable]

this is programming related surely

[sage crag]

rule 3 not real

[rigid snow]

we are a very civilized bunch here at #programming and no rule 3 violations have or will occur

[sage crag]

cant hurt you

https://cdn.discordapp.com/attachments/1362755858353885195/1367975291019989083/ElvynNyaFix.gif

[real sierra]

this seems like its one tangent away from me wiping the last 30 minutes

[olive sable]

im trying to make an ai vtuber the queen of england

[sage crag]

say something useful about chess bots so shiro cant remove it

[rigid snow]

[fast pagoda]

anyways all i was saying was popular country = popular to joke about

idk how that became a debatable thing

neuro is already the AI vtuber queen of england since vedal is the king

well, she will be when he croaks and she ascends to the throne

[sage crag]

https://tenor.com/view/cat-computer-typing-fast-gif-5368357

[olive sable]

vedal is king charly?

[fast pagoda]

yes

[olive sable]

charles?

wtf was his nmae again?

[fast pagoda]

i like king charly

[tender river]

shiro ban everyone ye

[rigid snow]

ban me

[fast pagoda]

ive been yelling it from the rooftops for like 6 months, vedal is definitely king charles himself

that's why neuro gives his address as buckingham palace

[olive sable]

i feel like i just wasted another 2 hours by arguing on discord over stupid things i dont actually have strong opinions about

[olive sable]

"royal family channel"???

why does vedal have a 2nd channel?

irl streams i guess

[fast pagoda]

"you know this would be a good fake persona to put on online"

i am a lipreader i saw him say it

you can see him saying it in this image

motion not required

[olive sable]

lipreading a static image

[rigid snow]

how does this affect the arg's legacy

[sage crag]

https://cdn.discordapp.com/attachments/1072697081443131476/1377728467025530950/NewLeafAbyss.gif

[fast pagoda]

it just reminds us it's bigger than you or me

it's a legacy of a country

[olive sable]

study sama is charles' daughter i guess

[fast pagoda]

"so yeah im gonna basically turn this into a drawn character and that'll be my online persona"

there's a reason the duchess' name is camilla and vedal is friends with camilla

coincidence?

i think not

[rigid snow]

she's dead
was charles' daughter

[olive sable]

[olive sable]

that woman is queenpb

trust me

[fast pagoda]

https://tenor.com/view/camila-imp-vtuber-clown-cute-gif-12054276830109292411

nono that's

[olive sable]

nah this is the moment he asked queenpb to make him a tutel mmodel

[rigid snow]

what is going on

[olive sable]

im not sure

[fast pagoda]

me continuing a stupid meme i randomly decided to declare as real at the start of this year

[olive sable]

i guess this counts as "pretend vedal is the king roleplay"?

[rigid snow]

is that in the bingo or something?? why is it in quotes

[fast pagoda]

if it's not, it should be

[olive sable]

[olive sable]

idk

i just formatted it like that

sue me

[sage crag]

phrrr phrr

[rigid snow]

filing a case rn

[sage crag]

sue sam

rr

[fast pagoda]

court room stream 3

[olive sable]

that does not help

[rigid snow]

good use of ai compute

[fast pagoda]

i have doomed the human race with my wasteful usage of resources wrt ai

[olive sable]

ok but i still don tknow what it means

it just gave me options

[fast pagoda]

phrr

for real

for real

[olive sable]

did your colour change mlntcandy?

[rigid snow]

it is a noise definitely. cat or horse

[rigid snow]

i removed osu role

[fast pagoda]

damn

[olive sable]

ye sure, but how am i supposed to interperet it?

[rigid snow]

you're not

[olive sable]

what

[fast pagoda]

it's probably nonsense

[rigid snow]

you're not supposed to interpret it

[olive sable]

ok but is it good nonesense? bad nonesense?

[fast pagoda]

ive seen people say "phrr" to say "for real" before

[rigid snow]

[fast pagoda]

in fact ive done it

so that's my only guess

[rigid snow]

good nonsense

bad nonsense

phrrr

[olive sable]

i dont understand humans man

[fast pagoda]

we are all nonsense on this blessed day

[olive sable]

nuh-uh

speak for yourself

[fast pagoda]

i am ALL nonsense on this blessed day :)

[tender river]

phrrr

[fast pagoda]

if you weren't referencing this then incredibly you accidentally did

[olive sable]

[olive sable]

i odnt remember charles being this young

oh its from 2013

[fast pagoda]

that's because you were like 2

[olive sable]

nuh-uh

i was 6

[fast pagoda]

lol

[rigid snow]

6 old enough to follow world politics

[olive sable]

nah

even at 18 i dont follow it

[olive sable]

i wasnt btw

[fast pagoda]

life imitates art

i was surprised that you would reference a ken M post from 2013

[olive sable]

i spiritualy channel amanda

[fast pagoda]

proof that humans are no more than stochastic parrots

[olive sable]

i need to google that ward

where is grok when you need him?

[fast pagoda]

one day vedal will hook into the discord in the same way

surely

[olive sable]

@ neurobot is this true?

running that would cost a decent amount of money, so we need neurocord Basic, costing $3/month, Premium at $8/month, and Premium+ at $40/month

that way we also get a gold checkmark tag on our discord profiles

[tender river]

[tender river]

just think of phrrr as

[olive sable]

shiro said is positive

[fast pagoda]

shiro has never lied in his life

[real sierra]

i may have guessed a few times tho

[fast pagoda]

plausible and deniable

[olive sable]

how could you?

do you have no shame?

i trusted you

[fast pagoda]

shiro's shame is accepting the role of moderator here

[real sierra]

but when an LLM makes stuff up it doesnt get any hate

double standards

[fast pagoda]

it's just a little baby it doesnt know

[real sierra]

maybe im just a little baby

big assumptions on your end

[olive sable]

[fast pagoda]

smart ass baby in college n shit

[real sierra]

why thank you

[fast pagoda]

lol

[olive sable]

i debated on saying somethingbut i decided not to

[rigid snow]

maybe the "college" is actually daycare

[olive sable]

dont want shiro to ban me

[fast pagoda]

shiro is stewie griffin

[olive sable]

everything broke

https://tenor.com/view/head-bang-dr-cox-ugh-angry-gif-16823810

[fast pagoda]

dont run update for one(1) day
171 packages with updates

ok

[uneven pulsar]

today's fun story

[tender river]

arch

[olive sable]

go on

[uneven pulsar]

keycloak takes their perfectly good method of boot time exporting entire realms with credentials and all (because using their web ui always filters out credentials)

[fast pagoda]

Packages: 2163 (pacman)[stable] total packages so like 10% of my system needs an update lol

[uneven pulsar]

and breaks it, so you can't do it in a running pod anymore, because that is very helpful for everyone

so i decide to run it locally in docker
at first i just quickly add it to my magical docker-compose of local development for ease and comfy, and it works

then i realize i'll need to bake an image from the container and run it again with a different entrypoint to be able to do the boot time export thingy, so i use this site to turn docker compose into just regular docker run (lots of env vars, who has time to type all that)

https://www.decomposerize.com/

doesn't work. maybe the site converts wrong? use https://www.composerize.com/ to see if it converts back

it does, it's not that

[fast pagoda]

docker

found your issue

[uneven pulsar]

imagine having employment

so why does it work in docker compose but not in regular docker

spend about 2 hours scouring the stars for an answer

[sage crag]

i approve this message

[uneven pulsar]

once again it was my arch nemesis, MSYS_NO_PATHCONV=0

[sage crag]

now for you to find out what means

[uneven pulsar]

[fast pagoda]

git bash

[uneven pulsar]

gibash

gymbash

[rigid snow]

how are we supposed to know what the codes you make up mean

[sage crag]

for me

not for you

[rigid snow]

i'll come up with some nonsense too

[uneven pulsar]

that's the point

if you use it well it becomes charming

[rigid snow]

jk i can't

[uneven pulsar]

https://tenor.com/view/elf-nonsense-elf-frieren-frieren-beyond-journey's-end-sousou-no-frieren-gif-2308403050531304831

[rigid snow]

[olive sable]

yep

[fast pagoda]

usually for anything involving docker first thought would be because docker was designed to be as obtuse and annoying as possible while abstracting away your ability to tell wtf is breaking and why and oh it's because this requires a very delicate balance of specific dependencies that are no longer current but docker makes it easy to distribute things with broken garbage inside of them behind a nice incomprehensible screen

[rigid snow]

evil

[olive sable]

== "phrr"
== evil
so
"phrr" == evil

[rigid snow]

now uhhhhh figure out what means

[fast pagoda]

if i had a time machine with one use i'd use it to go back in time and make whoever created docker in the first place touch grass before they ruined everything

[real sierra]

[uneven pulsar]

lmao

[real sierra]

overleaf added some LLM spellcheck garbage

and it cant decide on "nonempty" vs "non-empty"

[uneven pulsar]

honestly it's faster than VMs and i like that i cna just click the red button and make the oopsie go away

[fast pagoda]

neither can i tbf

[real sierra]

every time i write more, it highlights all past instances and insists i switch to the other

[fast pagoda]

lmfao

[real sierra]

[fast pagoda]

ignore that specific pair of words

[uneven pulsar]

change it to non-emptious

[real sierra]

"populated"

[uneven pulsar]

poppin

[fast pagoda]

docker just makes it easier for people to hide their broken garbage in a box that kinda works

[sage crag]

use nix instead

[fast pagoda]

99% of the things distributed as a docker container shouldn't be but are because it's slightly easier than juggling and providing correct dependencies the traditional way

[sage crag]

flaaatpaaaak

snaaaap

[fast pagoda]

but ya know gotta pull in 52 packages to run that calculator app

[sage crag]

aaaaapiiiimaaaaageeeeee

[fast pagoda]

i hate flatpak and snap as well tbf

appimage is slightly less garbage

[uneven pulsar]

honestly i think the main selling point is that you can run several of a thing without the app getting a mental breakdown

[sage crag]

nix os

nix os

[fast pagoda]

instead they offload the mental breakdown to the user

[sage crag]

[uneven pulsar]

like most things go

the moment you want to run 2 of it

[fast pagoda]

there are definitely real use cases for it

but like i dont need to run a single application that will never be run concurrently and doesnt need sandboxing
from a container

and yet "ohohoh we provide our package in a container"

bruh

[uneven pulsar]

me running 3 oracle and 2 postgres servers on my windows laptop be like

click green button watch all of them burn to ashes and then pure nonexistence and get replaced by new ones while i fetch coffee

goodbye ozone layer you should have been this automated too

[fast pagoda]

that's also not good, applications shouldn't expect and rely on the ability to just explode and restart to fix it

but they do because docker lets them

[uneven pulsar]

i'm not an application

[fast pagoda]

that's what you think

ive seen no proof

[uneven pulsar]

i'm just applied physics aren't i

[fast pagoda]

we are all applied goofiness

[uneven pulsar]

also

[real sierra]

oh thank god

[uneven pulsar]

imagine running your product in a docker container

[fast pagoda]

idk i cant really see a reason for docker to exist because all of the use cases besides just having an isolated development environment on locally can just be kubectl

[uneven pulsar]

so when it gets deployed to a customer

[real sierra]

i hit my token quota limit with this ai spellcheck garbage and it turned off

[uneven pulsar]

who still uses idk RHEL 6

you don't have to rewrite all your python scripts in python 2

so it works for them too

[fast pagoda]

the way that should work is that they go ahead and stop using out of support system software

but no they can use it with the bandaid that is docker

[uneven pulsar]

ok well that's not my job

bandaid

money

[fast pagoda]

then the powers that be that decided RHEL 6 is still fine because "all our stuff still works!! {life support by docker allowing it to still function for purpose, barely}" are like how could this have happened

when they get pwned and all customer data released and have to pay out for 6 months of ID protection

[uneven pulsar]

like all in all i'm much more glad dealing with magical blue boxes than trying to scream at sales and business people that no i'm not gonna do what it takes for our product to run on the customer's machine

[fast pagoda]

docker is probably somehow indirectly responsible for my SSN being out on the internet

[uneven pulsar]

who is, of course, always right

[uneven pulsar]

yeah well you shouldn't have used the grid

[fast pagoda]

i shouldn't have checks notes had dental insurance

[uneven pulsar]

just be a wild undocumented cyberpunk duck!

my other favorite thing is <Product Onwer> who can be scared into submission by "security alerts"

so i'm trying to iron things out for my beloved little team before i leave company, teach them the high magics they neglected in their years,now that they will have to maintain themselves

showing all the fucky wucky nightmare ropes of helm templating and the azure wizardry we sit upon

but some bureaucrat guy at IT finds some post on his favorite CVE site and tells everyone in the company that idk if some wizard does the right 59 step ritual to http they can make it crash out so every http server ever has to be upgraded to at least these versions from tomorrow ago:

so instead of teaching team how to be able to operate later on when i'm gone, i'm bumping up version 16.0.358 to 16.0.389 in no more than 72 hours, so when we eventually release in late october, and clients install the new version in 2027, they'll be safe from a thing that will never fucking happen anyway, because if someone even reaches the product they're already inside and can just blow the place up

it's called being agile

[fast pagoda]

lmfao

there might be a very prepared hackerman continuously monitoring your application for the precise moment they swap to that compromised version to strike

CVE sites do like to present these things in a way that to someone who may or may not know what they're talking about is going to see it and think the world will literally explode tomorrow if the thing isn't patched NOW

especially when they have to then report to someone who DEFINITELY doesnt know what the hell they're looking at

especially since CVE ratings are skewed high

7.5/10, seems dangerous!!!!!!!!!!!!!! FIX NOW!!

[uneven pulsar]

recently there was one about some

freeware windows ftp server

for versions released between like

2008 and 2014

and my PO is like "hey can you check if this software vulnerability affects our product"

how do you even put a freeware ftp server thingy in a java product running on red hat, oracle, and the blood of ancient vampires
with the initial commit being in 2018

like i was moderately offended just for being asked that

[fast pagoda]

hey it's hard to realize that running on red hat, oracle, and the blood of ancient vampires isn't typical if that's what you do all the time, easy mistake to make

[uneven pulsar]

he's the product owner tho

(in his defense he's kinda new)

[fast pagoda]

a lot of silly requests start out in a place of knowing you'll have to report it to the powers that be anyways so might as well rip the bandaid off and have an answer prepared for "WELL DID WE MAKE SURE IT WONT AFFECT US?!! DO IT NOW!!"

[silent cloak]

me raging at my docker setup earlier

[uneven pulsar]

it's just funny cause

ok another story

man i'm full of these lately

so there is a project that no team accepted but was already promised by some higher up buffoon to some client

there are two teams that have the domain smarts to make it: my team, other team
other team has the architect grand expert who came up with the idea

they asked both teams how long it would take to make
their estimate: 3 people, 3 months
our estimate: 3-4 people, 9 months

maximum that the client is willing to pay for: 3 people, 3 months

well then, our team is not doing it have fun

<it gets promised at this point>

other team goes "yeah we don't really have the capacity to do it rn"

so management comes to my team

"hey so remember that thing you didn't want and said would take 9 months? now you have 3 months to do it, it's your #1 priority over literally anything else you guys are doing"

PO does emergency meeting on reprioritizing things

priority #1:
security alerts being handled

[uneven pulsar]

like -that- will be something hard to explain to management

hey not only did we not finish the thing that was #1 priority and is gambling a customer, we barely even started because the reduced size team was busy checking all the security alerts about idk freezilla knockoffs from 2008

i'm so glad i'm leaving

[real sierra]

homework done

coding time

starting the chess vm

[sharp stream]

[olive sable]

shiro is cordinator of events

he decided to make a chessbot tournament cuz he hates himself

poor shiro

[real sierra]

castling with bitboards

[olive sable]

jsut check the conditions and move the bits

[real sierra]

the calculator uses the opposite endianness that i do

[olive sable]

isnt casteling a move to jsut a single other position? if you have a bitboard for the king jsut change it out

[real sierra]

i could but it would be slower

but this is almost certainly not gonna work

maybe you have a point

[olive sable]

jsut set a bool to true at the start of the game, and if either the king or rook move set it to false

[real sierra]

thats not the problem

im just trying to microoptimize

you're right tho i should just do this the braindead way

no

NO

i remember why im doing it this way

its because moving the rook is hard

[olive sable]

and that is?

[real sierra]

you cant slide the bitboard for the rook because it has the other rook on it

so you need to xor it with a mask to flip the relevant rook's old position off and its new position on

[olive sable]

still, why is it difficult to do it binary wise?

[real sierra]

because i have to figure out what these masks are supposed to be

and im confusing myself over and over trying to go from board squares to numeric constants

[olive sable]

bitshift the king 2 to the right
for the rooks do xnor with the eight space and then or with the 6th space

[real sierra]

but whats the number for that

notice also that the mask depends on which color is castling

[olive sable]

jsut do 64-8

ez

idk

[real sierra]

yeah thats all im stuck on

lol

the rest is straightforward

[olive sable]

i think the 8th position in binary is 128 no?

2^7

[amber fractal]

That sounds correct

[real sierra]

if i write it that way

will the compiler simplify it for me

[olive sable]

idk

[real sierra]

oh well

this doesnt really need to be quick

how many times does castling happen in a game

[amber fractal]

Literally once

I guess twice actually, one on each side

[olive sable]

the other rook is the 57th position, so 1 with 56 zeros

so 72057594037927936 in decimal i think

this would be a lot easier if you'd use the 1st and 64th position yo uknow

on avarage even less than once cuz i dont think castleing is done too common

around 90% of games at the grandmaster level

hmm

ok the grandmasters like their castles

[real sierra]

you get to move two pieces for the cost of one move

its pretty good

[olive sable]

ye but they move horizontally, my goal has always been to get to the other side

[real sierra]

i got this too

[olive sable]

the king is safer tho

so i guess its good

[remote atlas]

depends if the other player wants their opponents king buried in the corner

[real sierra]

wait

are we doing this right

[olive sable]

idk

[real sierra]

i think its actually 5,764,607,523,034,234,880

[olive sable]

why?

[amber fractal]

I don't have a c linter atm to test if this stupid idea works. Is this even valid somewhere? I just thought about it as Shiro was struggling to function with hex

0b\
11111111\
11111111\
00000000\
00000000\
00000000\
00000000\
11111111\
11111111

[real sierra]

that would definitely be easier to write

[olive sable]

5764607523034234880
would be 63 digits, with 1010000000000000000

[real sierra]

anyone reading the code would have a laugh

( )

[silent cloak]

https://tenor.com/view/aneurysm-hhhhomeboyyyy-kings-dark-tidings-saga-by-kel-kade-reaperleviathan-head-explode-gif-21578105

magic numbers my beloved

[real sierra]

now en passant

the one move where the piece taken is not on the tile moved to

requiring a whole separate piece of code

[silent cloak]

what are u actually making

[real sierra]

a chess bot api

[silent cloak]

whats the magic numbers for

test cases?

[real sierra]

the board is stored as a series of bitfields, with each bit in the field corresponding to a square on the board, and each bitfield being associated with a particular piece and colour

[olive sable]

@real sierra
you can just write int value = 0b11110001001000000;

[silent cloak]

thats probably a better way

[real sierra]

this code is performing castling, so it needs to clear the bits from where the king and rook were and set the bits at where they are now

so im doing that with an xor

[real sierra]

i am not typing out a 64-bit string in binary

[amber fractal]

copy paste

[silent cloak]

so its a bunch of hardcoded values?

[real sierra]

coding in a vm

didnt set up clipboard

[olive sable]

it would be less time consuming than your magic number lol

[real sierra]

you're always castling to and from fixed positions

so i know what bits need to be flipped

[real sierra]

not now that it's done

[silent cloak]

i usually use bits for flags honestly these days

rarely anymore than that

[amber fractal]

Your telling me your vm doesn't have a clipboard itself

[real sierra]

discord isnt in the vm

separate clipboards

[silent cloak]

why a VM

[real sierra]

api has to be as fast as possible so people dont blame me for their slow bots

[olive sable]

cuz its arbitrary code

on our end

[amber fractal]

Well just type it out once and keep it copied

[real sierra]

the end goal is that people will write bots and give them to me to run

im not running random internet user code on my actual pc

[silent cloak]

yeah i get that with some systems in game dev

[olive sable]

it we will stream it in vc right?

[real sierra]

depends how many submissions there are

[silent cloak]

the whole reason i memorized float16 conversion was an optimization method i used for minecraft

[real sierra]

at the very least i can surely stream the finales

[jagged turtle]

what is shiro trying to do?

[amber fractal]

chess battle (not) advanced

[olive sable]

its most likely gonna be about 8 i think

[silent cloak]

you could probably make a tool to just generate the values for you

[real sierra]

the time searching for or making such a tool would take longer

[silent cloak]

i suppose so

my programming brain would do it anyways

https://tenor.com/view/xddd-gif-11484761101775539239

[olive sable]

or jsut type the values directl in int value = 0b11110001001000000;

[silent cloak]

i love making tools for specific use cases!!!

[olive sable]

no need for conversion

[real sierra]

i bet a gpu would love this

[silent cloak]

bruhhh

[olive sable]

yes

feed me

[amber fractal]

@olive sable can you split these over newlines or is it a syntax error?

[olive sable]

i have no clue, lemme test

[silent cloak]

@real sierra u should write it in ASM

for 100% optimization

with how much bit management you're doing you might as well

[real sierra]

no thanks

this needs to run on at least a few different architectures

[silent cloak]

i know a bit of ASM but im planning on picking up more

mostly because i was building my own language

[olive sable]

doesnt seem to work no

[real sierra]

also i just found a bug in this, im glad i posted it

[amber fractal]

[olive sable]

you can use ' as a seperator, but still needs to be on the same line

[silent cloak]

any SIMD?

[amber fractal]

One board instance so probably not

[silent cloak]

ahh

[amber fractal]

unless shiro has multiple test boards

which will be a pain to API

[silent cloak]

had to build a bunch of API services recently for Hanakyos mc server

so i know the pain

[real sierra]

no simd planned

[silent cloak]

ive gotten them wired up with docker today though

[real sierra]

gonna be raw cpu time to finish this tournament

[amber fractal]

I have a stupid idea but it is probably cpu depentant

let me sanity check

[silent cloak]

projects like these are a good way to pick up micro optimizations

[amber fractal]

This idea is combining ASM and SIMD

[real sierra]

do && and || work as i'd expect in C?

[olive sable]

they should, and or?

[real sierra]

as in like

[silent cloak]

can you can write ASM in some C setups

[real sierra]

AND(A,B) = A ? B : 0

[silent cloak]

https://en.cppreference.com/w/c/language/asm.html

[olive sable]

&& is logical and, & is bitwise and

[real sierra]

OR(A,B) = !A ? B : 1

[real sierra]

i know that

[olive sable]

well then why ask?

[amber fractal]

Yeah, that definitely exists

[real sierra]

because that wasnt my question

[olive sable]

e

[hollow spruce]

Hmmm

[real sierra]

oh well

ill just use * for what i want

[olive sable]

?

[real sierra]

trying to write A ? B : 0 branchlessly

can do this as A * B

but if C implements its && in a particular way, then A && B would work too

and be faster

[olive sable]

int mask = -!!A;
int result = mask & B;
?

[silent cloak]

(!!A) * B

[real sierra]

A is already either 1 or 0

[olive sable]

thats fine no?

[silent cloak]

(A != 0) * B could also work

[real sierra]

means you can drop the !!

i think A * B is as good as it gets without special &&

[hollow spruce]

Hmmm ure logic is more focused of white
So u should try B algo

[silent cloak]

branchless programming is always fun

though sometimes the compiler will be a dickhead

so make sure to check ur shit

[olive sable]

multiplication is a lot cleaner yes

[amber fractal]

Using AVX512, you have 8 64bit masks that you can do operations on. You have 6 pieces, so I guess one can be empty and the other can be opp pieces. Meaning you'd be masking the opps side on attack.
2 AVX-512 registers total for board state storage.
Just the first thing that came in mind.

[silent cloak]

bit hacks my beloathed

[amber fractal]

This would actually work for en passant due to empties and deleted squares being separated in this case

Maybe I've been working on my NN for too long, but that is where my first instinct was

[real sierra]

ok

move making implemented

now for move undo

[sage crag]

bitwise would be better than multiplication due to platform inconsistency

also slightly faster if you can do it

but uh

not bother

simple best

[real sierra]

i know but idk if && works the way i want it to

most intelligent implementations return the second arg if the first is false

[silent cloak]

just test it

[sage crag]

dont use && when you are doing bitwise ops btw

[real sierra]

im not

[sage crag]

&& is short circuiting

[real sierra]

im using & for bitwise and

[sage crag]

yeye

[sage crag]

which mean branch on each bitwise op

[real sierra]

i know im an idiot but i do kind of know what im doing sometimes

[amber fractal]

Surely there is a variable where you can match for the piece that was deleted last

[real sierra]

the elegant solution is to store the move sequence and reverse them, and then store lost data such as captured pieces and castling rights in a stack

the way im doing it is storing a stack containing all past boards

[olive sable]

i have aquired crackers

[silent cloak]

is the elegance in the room with us?

[olive sable]

shiro

what the hell

[sage crag]

what

banned

make my own chess tournament

[olive sable]

me when i make my bot move left and right and left and right to fill the heap and crash the program

[real sierra]

its not so bad

boards arent really that big

they're like

what

[amber fractal]

96 bytes

[real sierra]

12*8 + 16*8 + 5/8 + 2*4 + 12 + ? bytes

[amber fractal]

surely I guessed it

[real sierra]

(the ? is because they store pointers to the previous board as explained, not sure if that counts)

[olive sable]

300 bytes roughly with the ?

[real sierra]

but yeah its a relatively small memory footprint

[olive sable]

256 bytes might be mnore fair tbh

[silent cloak]

shiro is going in the programming war crimes compilation

[scarlet arch]

I've been made aware there's war crimes happening here

Ignore that I'm German. Totally unrelated

[real sierra]

you should be able to store like

14 million boards

by my math

in 4GB*

[amber fractal]

Shiro is warcrimes

[olive sable]

The average chess game is approximately 30 to 40 moves long
so 40 x 256bytes is 10KB

[real sierra]

and then you undo when you unwind that search stack

[olive sable]

i still dont like it tho

[real sierra]

which frees the memory as you pop boards off the history

[silent cloak]

there are 4.8*10^44 possible legal chess positions

[real sierra]

if this was true you would be a millionaire

[olive sable]

are you sure about that

[silent cloak]

[real sierra]

if you remove the "legal" from that you might be right

unsure

might have to remove "possible" too

[olive sable]

AI Overview
There is no single answer to "how many legal chess moves" exist, as it varies significantly depending on the stage of the game. From the starting position, White has 20 possible first moves. During the middlegame, the number of available legal moves usually ranges from 30 to 40, while in the endgame, this number typically declines as pieces are exchanged

thanks google

[scarlet arch]

I haven't read back all of this, but I remember seeing a video where a dude stored the whole chess board as bitflags in a u64. Like per character. Does that make things smaller?

[silent cloak]

https://en.wikipedia.org/wiki/Shannon_number#Number_of_sensible_chess_games

[olive sable]

the answwer is 40 i guess

[amber fractal]

[real sierra]

if chess is analyzed for the number of "sensible" games that can be played (not counting ridiculous or obvious game-losing moves [...] )

[silent cloak]

me picking all the losing moves on ur API

[real sierra]

"here is how many moves are in chess (i didnt count most of them)"

[olive sable]

when my bot makes a -75 trillion IQ move but the enemy llm wasnt trained to handle it and crashes

[real sierra]

[silent cloak]

neuro

[scarlet arch]

Why that face

[olive sable]

these are 64 bit ints

but you cant use a 0 or 1 to show the position of 13 diffrent kinds of pieces, so you have to use multiple

[silent cloak]

https://www.chess.com/article/view/what-is-the-worst-chess-move

time to study this for shiro's API

[real sierra]

go for it

[amber fractal]

Shiro is only using bit masks

[real sierra]

finding the worst move is equally as difficult as finding the best move

[silent cloak]

shiro is dressing for halloween with all these masks

[real sierra]

you have to check just as many boards, and verify the same conditions

if you make the worst chess bot ever, i'll be impressed as heck

[amber fractal]

Though this also exists @scarlet arch

[silent cloak]

is this a challenge

yes or no

[amber fractal]

(yes)

[silent cloak]

fuck it im in

added to my list

[real sierra]

+1

[silent cloak]

ill make the worst possible chess bot

[olive sable]

the worst chess move physically possible is almost definetly what i do. researchers shoudl see me play

[silent cloak]

this weekend ill probably start on it

[olive sable]

ill wait till the api is done

[silent cloak]

i want it to be so bad that a brain dead person could battle it and win

[amber fractal]

Time to beat worstfish

[scarlet arch]

Oh neat

[silent cloak]

welp time to plan an algorithm to see how to lose in the fewest moves

[olive sable]

i move my rook from A8 to A9

[scarlet arch]

I have no clue what this means. It's 03:37

[scarlet arch]

Reminds me of that poker bot story where it just always goes all in and the other bots fold

[silent cloak]

a big thing ill need to focus on is avoiding stalemates where possible

[olive sable]

the smolest file size would probably be 208 bits

[silent cloak]

i should call it the SlopEngine

[olive sable]

wait lemme check if im doing my math correctly

for a single space there are 13 possebilites, so if we use 4 bits to represent that we're only using 13/4 = 3.25 bits per space. so 3.25 * 64 = 208 bits total

the poblem is that requires a lot of decoding and encoding so not really good for speed

[silent cloak]

i think i should start experimenting first with an inverted minimax

[olive sable]

nah my math is still worng

log2​(13^64) = 64 * log2​(13) = 64 * 3.70044 = 236.83 bits

so 237 bits

[amber fractal]

Yeah, bit mask for each piece type makes more sense. but that is 832 bits

[olive sable]

shiro is combing similair ones

or wait, thats still 832

[amber fractal]

This follows 8x8 and 13 types

[olive sable]

what is shiro doing then?

oh wait he's doing bytes

not bits

[amber fractal]

/8 time

[olive sable]

256 bytes

he could be so much more space efficient smh

[amber fractal]

104 bytes

[remote atlas]

Dude this isnt a story

Its a bot

The Poker Night bots

its so funny

playing against them

[amber fractal]

Surely he is talking about a video he watched on it

[scarlet arch]

Idk where I've seen it. Was some sort of Reddit post

[remote atlas]

Oh my bad

[olive sable]

imagine wasting a whole 0.00001% of your storage space

those 100 bytes

[remote atlas]

Idk its probably the case for a lot of poker game bots

[olive sable]

i have no clue what i did wrong but sure

[amber fractal]

I'd still be doing 128 bytes per board state as described in the funni ASM SIMD idea. So not exactly the most space efficient but it is worth it for easy impl.

[olive sable]

space efficient doesnt matter when you can have pure speed

[amber fractal]

The only reason the SIMD method is fast at all is because you can cast a 64bit register to a SIMD one like numpy does.

Otherwise you'd be iterating over all of the opps pieces to remove them on taking

I mean having all of the board on 2 AVX-512 registers is also a benefit I guess.

[real sierra]

move undo is complete i think

now i need to finish the UCI handler

how much of this spec do i really need

surely just implementing like 10% is sufficient

[faint sandal]

why is vmware so slow on my computer

borderline unusable pos

[olive sable]

i may have done some race conditions

single slice

[real sierra]

i actually think maybe i can get away with this little of UCI

i was joking before but

mutex

[sage crag]

shimo

[real sierra]

hiii

[sage crag]

your api sounds like an affront to god

[real sierra]

whaaaaat

[sage crag]

i look forward to using it

[real sierra]

i thought my concern for thread safety was quite a green flag

[olive sable]

i fixed what i thought the issue was, but its still not fixed so the broken code was responsible for a bug i wasnt even aware of i guess

[sage crag]

you not need thread safety its single threaded

[real sierra]

its not

the bot runs on a single thread but the UCI spec says it must always be listening to the server

so the UCI listener runs in a separate thread

[sage crag]

[olive sable]

shiro not using async

[real sierra]

async

C

just about bricked it

[sage crag]

ye state machine with mutexes

[sage crag]

pthread

not portable to msvc

[real sierra]

but posix...

[olive sable]

i didnt know async worked diffrently in C vs C++ but i guess thats where those plusses came form

[sage crag]

its ok they will just have to use windows gnu target

[real sierra]

surely if you build on non-posix system it just makes an alias

[sage crag]

right mhm

surely

it will probably work on windows gnu target

almost certainly wont on windows msvc target but who cares about that garbage

[olive sable]

OM i was interpreting vec3's as vec2' and vec2's as vec3's cuz i switched the numbers by mistake

its still broken

bwaa

[sage crag]

[olive sable]

i have this rn

so indexbuffer and vertexbuffer work. the uv buffer doesnt tho

and normalbuffer i cant tell with the textures like this

[real sierra]

when the bot indicates it has submitted its final move

i would like that submit method to block that thread until the UCI listener hears that its time for the next move

idk a good way of doing that tho

i guess another mutex

[olive sable]

i have no clue

[real sierra]

surely this wont make a race condition

(i already know what the race condition is and am pretending it doesnt exist)

[tender river]

semaphore

[real sierra]

explain

[tender river]

no

[real sierra]

ah i see

that'll work

[olive sable]

semaphores are for between gpu and cpu no?

or does it have a diffrent meaning when not vulkan?

[real sierra]

i think it probably functions on the same principle as a synchronization primitive

but here it's working between threads instead of between cpu and gpu

[olive sable]

om

not only did i switch vec2 and vec3 in 1 part of the code

i did it again in the shaders themsleves

maybe it used to be the opposite way and i accidentaly changed everything when restructuring the code?

[real sierra]

wow

i think its done

v1 of the api

how do i test this

how do-

how do i build a c program again

[unkempt citrus]

this is why I like go

among othe rreasons

[real sierra]

its ok

i got this

[olive sable]

fixed

the pictures will get more interesting once the individual objects code gets finished

[real sierra]

it builds

now i guess i have to do something scary and actually try to start the api...

moment of truth

well thats a bad sign

oh i didnt recompile

thats why

the warnings gcc just spit out exceed my terminal history

need to pipe into less

none of these errors make any sense

what conflicting types

[olive sable]

huh

[real sierra]

gcc smoking something

[olive sable]

did you maybe not define the function before using it?

it needs to go above

[real sierra]

...

[olive sable]

bad

ye this is one of the worst parts of c and c++

[real sierra]

but recursive functions work fine

[olive sable]

yep

idk why

i guess cuz the function knows it is the function?

imo the compiler should just see what stuff exists in the file beforehand but i guess we're not doing that for some reason

[sturdy wing]

after working on my PC when using it my ears began to hurt. I do hear some high pitched noises, but nothing too unusual from normal PC sounds, however, my ears hurt after longer term exposure
I began tests to attempt to determine the source of the inaudible painful sound

after yesterdays PC noise tests here's what i know:

• the noise began after i worked on my pc -
  I swapped both case fans and my CPU fan,
  I attempted to unscrew my motherboard from my case multiple times before realizing it is stuck due to a broken screw,
  I attempted to remove the screw by force with no success
  I didn't in anyway interact with the PSU, SSD, RAM
• I noticed that my GPU's right fan would make an almost engine rev up noise when increasing or decreasing in speeds
• after realizing the high pitched noise (not the rev up) here's what I attempted:
  1. I disabled both case fans. noise seemingly persisted, but i felt that it got slightly better
  2. I disabled CPU fan (re-enabled case fans), I enabled power saving in windows settings. noise seemed to get notably better, but still persisted.
  3. I swapped GPU, both case and CPU fans were enabled. noise seemingly persisted

conclusions:

• As I did not interact with my PSU my assumption is that in my messing with the motherboard something got slightly damaged causing the noise
• Alternatively my exposure to this noise made me more sensitive to the natural noises of my computer making all tests inconclusive

[olive sable]

hmmm

idk

[real sierra]

👍

well it compiled finally

what if i -Wall

[sturdy wing]

also based C enjoyer

[real sierra]

why thank you

[sturdy wing]

my second favourite language after Rust
if C had the ability to overload operators like C++ and some syntax improvements it would be the best language imo

sadly no one has made C, but with minimal improvements™

[tender river]

typedef

psh

https://fuckingfunctionpointers.com/

[real sierra]

endless warnings

i made the mistake of assuming & took precedence over >

the compiler has warned me such but the fixing is annoying as hell

ok

all the warnings i care about are fixed

lets run it

[olive sable]

why not all of them?

[real sierra]

wtf

it hasnt crashed

its also not behaving as i expected

if the semaphore has a value of 0

then sem_wait should make it hold until a sem_post right

[hoary lion]

this is this and that is that

https://tenor.com/view/thats-that-and-this-is-this-roland-library-of-ruina-roland-library-of-ruina-ruina-gif-15323178589486145178

[olive sable]

https://tenor.com/view/python-path-meme-didnt-add-gif-22594006

[real sierra]

looking good

[olive sable]

fix your errors

my -Wall is completely empty

[nocturne olive]

Turns out, using Python on web is a horrible idea

Who could have ever expected this

[olive sable]

me

i expected this

i did this before

[nocturne olive]

Yeah unfortunately you didn't tell the person that wanted to make a Neuro attorney visual novel thingy that

[olive sable]

still, 40fps at 720p?

thats a skill issue

i did vsync

at 1440p

[nocturne olive]

Well it's not native WebGL, it's using a library so there's like 2 extra layers

[olive sable]

oh

must be a shitty library

[nocturne olive]

https://discord.com/channels/574720535888396288/1338282572689440860
You can read up here, in the conversation between them and Xoda

Who knows, maybe you could help them translate the whole thing to something more performant

[real sierra]

stupid question

does scanf consume input

something is not making sense rn

[tender river]

shiro fish

[real sierra]

chayleaf

i dont know why this isnt working

it no make sense

[tender river]

good luck

[real sierra]

its reading the first line of input

over and over

how is that even possible

[tender river]

\n format string maybe

[real sierra]

but

i thought input was consumed

by scanf

it shouldnt be there to be read a second time

[tender river]

[real sierra]

[tender river]

write small test

[real sierra]

i am running my small test

thats why i dont understnad

oh

OHHH

i get it

char line[1024];
while (running) {
  scanf("%[^\r\n]", line);
  // ...
}

this seemed pretty foolproof to read in lines

but it doesnt ever read the newline chars so it doesnt clear them from the buffer

then it just infinitely reads 0 characters over and over

which leaves line unchanged

i dont know how i feel that it doesnt read the null terminator but

whatever

[tender river]

shiro buffer overflow

[real sierra]

i know

surely the chess program wont send me more than 1k

[amber fractal]

Sending you inputs from yes bwaa

[real sierra]

well it was a good start

don't like how the board looks? just segfault

segfault fixed

the UCI interface seems to work on a very basic level

lets see if it knows enough words to talk to arena

do all writes happen from the main thread

arena starts the bot, asks if its there, and for some reason the bot doesnt reply for like 30 seconds

so arena is like "a new game is starting, ready?"

and it still doesnt reply

so arena gives up and just sends it the board info and says "ok you can start thinking"

and only then do the replies for all the commands before get sent, all at once

and i think its no coincidence that this is where the semaphore is released

[real sierra]

segfault but no stack trace?

[tender river]

shiro c

[real sierra]

explain

where is my stack trace

no make sense

[tender river]

main ()

[real sierra]

literally the only thing it could be doing during this segfault

is running that function

[tender river]

[real sierra]

[tender river]

dont use vtables

[real sierra]

already am

[tender river]

useless without 2 impls coexisting

[real sierra]

convenient because i can refuse to give the user my API until it's ready to be used

no calling my functions until i say so

[tender river]

__attribute__((constructor))

[real sierra]

python lookin line

[tender river]

shiro did you know

c function pointers cant be used in some languages

[real sierra]

bad languages

[tender river]

nop

bad shiro

[real sierra]

they'll have to wrap the python bindings

[tender river]