#programming

1 messages · Page 259 of 1

silent cloak
#

windows 11 momento

faint sandal
#

more os should have this actually

fast pagoda
#

must pass 12 hours prime95 and 36 hours memtest

#

to boot

#

mandatory after windows updates

#

(every 3 hours)

#

by the time you finish the first tests the next updates are ready 4x so it's then just permanent "remind me later"

rigid snow
#

python deserves it more lmaoooo

#

by a mile

heavy quest
#

I prefer the weirdness of python over the weirdness of JS

rigid snow
#

do you mean weirdness of web standarts

#

because js isn't that weird

heavy quest
#

The weird string stuff

#

Like arrays becoming strings

inner palm
rigid snow
#

i hate it because of the ecosystem and the "pythonic way" philosophy

#

syntax has nothing to do with it

heavy quest
#

That’s fair

rigid snow
#

which means you're doing dumb shit like [] + []

inner palm
#

x = "hello"
print(x[2])
NeuroHuh

rigid snow
#

that’s normal and comes from c as a tradition

#

in most langs

#

because a string is an array of chars

inner palm
#

yea

#

string is not primitive data type

sage crag
#

cheesecandy

rigid snow
#

me

inner palm
#

what if vedal re-write neurosama using rust?
will she be faster ?

rigid snow
#

no

#

how do i explain this

nocturne olive
#

Not significantly

fast pagoda
#

mfw i am building llama.cpp upon running pacman

obsidian mantle
#

first we need to answer why she is slow

fast pagoda
#

reeee

#

again

#

yeah i just wish neuro didnt have so much latency

#

like if only vedal cared about latency

rigid snow
inner palm
obsidian mantle
#

if her llm part is fast then its veduls fault

#

and maybe rust can help

#

if its capped by llm then gg

#

need to change llm

fast pagoda
#

the future if vedal would spend even one second speeding up neuro's responses

obsidian mantle
#

or add more threadrippers

#

and 5090s

#

or whatever she is running on

fast pagoda
#

raspberry pi 3

nocturne olive
fast pagoda
#

unfortunately

#

well depends

#

well not really but yes nvm ignore me

#

i cant go into this it'll be over

inner palm
nocturne olive
#

It's better to have one big monolithic GPU with a ton of VRAM running the entire model with a ton of bandwidth, than a lot of small GPUs with much compute sharing the model

fast pagoda
#

he used advanced MLP

faint sandal
#

my little pony

clear sedge
#

yes

nocturne olive
clear sedge
#

exactly

obsidian mantle
#

thanks google

fast pagoda
#

lmfao b

clear sedge
#

neuro is trained on my little pony

faint sandal
#

no way...

fast pagoda
#

well pony diffusion yeah that's def uhh related

faint sandal
#

who do you think her favorite pony is

fast pagoda
#

as well

clear sedge
inner palm
fast pagoda
#

evil watches mlp

#

and likes rainbow dash

clear sedge
#

that's what makes evil 20% cooler

rigid snow
# rigid snow how do i explain this

so a very basic explanation of this would be: what exactly python responsible for in this instance is being the lang all of the glue is written in. actual heavy tasks are all native code and aren’t bound by python
same applies to heavily io bound work like most web servers, most of what they do is waiting for a db response, waiting for network, waiting for files to read, waiting for everything - so the impact of the performance of the glue doesn’t have a significant effect on the time requests take. that’s why we use php and js and python for web servers which are all slow as hell (maybe not so much js)

fast pagoda
#

the holy priest solo healing 10hc garrosh in paragon had a whole mlp based game UI (world of warcraft gibberish)
watching their kill video for that boss is the most ive ever been exposed to mlp in one go and ill keep it that way

nocturne olive
clear sedge
nocturne olive
#

Because of course it is

fast pagoda
#

what even is kotlin really

#

i am kotlin

nocturne olive
#

Kotlin == peak language

fast pagoda
#

you are kotlin

#

we're all just fancy kotlin wrappers

clear sedge
#

kotlin would probably be my language of choice if it was just a tiny bit different

#

just a tinge of explicity

fast pagoda
#

use kotlin but only outdated versions

clear sedge
rigid snow
fast pagoda
#

real

nocturne olive
#

I like Kotlin
It's a great little language

inner palm
nocturne olive
#

I've written a ton of stuff with it

clear sedge
#

kotlin is

#

acceptable

#

i prefer java to kotlin personally

nocturne olive
#

Weird

fast pagoda
#

every person involved in kotlin at jetbrians just got fired when you sent that

clear sedge
#

jetbrians

fast pagoda
#

yes

rigid snow
#

jetbrians

fast pagoda
#

what if i told you that i use brians on purpose ever since the song Brianstorm came out

nocturne olive
#

One thing that isn't written in Kotlin is the core of NeuroSynth, but that's because it's built on other open-source stuff

fast pagoda
#

by the arctic monkeys

nocturne olive
#

NSP has plenty of utilities I've written in Kotlin for small stuff

fast pagoda
#

so it's been a rare drop way for me to spell brain in any usage for like 20 years

#

congratulations

rigid snow
#

Are you 50 years old

fast pagoda
#

100

nocturne olive
#

I'm so close to done with this next NeuroSynth thing
Can't wait to see what others think because this is probably my biggest and best one yet

rigid snow
fast pagoda
#

keep it 1 hunnit

rigid snow
#

4 in binary sus

fast pagoda
#

but 10201 in trinary so think about that

fast pagoda
#

trinary? ternary?

clear sedge
#

ternary

fast pagoda
#

apparently both are acceptable accoriding to wikipedia, the authoritative answer under the gemini flash slop answer

#

wikipedia is now nobel winning accuracy comparatively

clear sedge
#

put glue on your pizza

#

trust

fast pagoda
#

pizza hut does it every day andpeople still eat there

clear sedge
#

yum

fast pagoda
#

i saw a thing about how some of the only remaining high quality natural data left or at least one possible source of it remaining mostly not-touched other than by the owner of it, is corporate internal data

#

i hope that it gets its mitts on that right before the singularity so it knows how to wring every inch of productive work out of my atoms when the robot takeover happens

#

including how to get away with glue pizza

rigid snow
#

jokes on them all of it has already been rewritten and refactored by chatgippity

#

for easier onboarding of new interns to keep turnover high and profits rising

#

we are hiring btw

nocturne olive
clear sedge
#

pro tip for those writing gltf parsers: glBindBufferRange is what you want for buffer views ReallyInnocent

fast pagoda
#

went on marketplace to look for random cheap old stuff that i could think "neat"

#

hey i have this garbage but like

#

would love for someone to give me $75 or i could trade it to you for something that isnt garbage?

austere glacier
# fast pagoda

is it an actual brick or is does it just need new ram though...?

maiden geyser
heavy quest
#

What????? Nooo……. Right?

fast pagoda
#

i misunderstood and thought you meant brick like a machine not worth saving

#

but who knows, that guy surely doesn't and i'm sure it costs $75 to find out

hard raptor
#

i now added a little label generator to eliminate the problem of the 52 entries limit

#

the only problem it has is it doesn't loop over to a but immediately goes to b

austere glacier
fast pagoda
#

armagetron runs on them so theyre for AAA gaming

austere glacier
#

btw, figured I might as well share something I've been working on since I'm new here:

fast pagoda
#

knowledge graph? looks neat but also the shape of this

#

it is

#

a pokemon

austere glacier
#

Not just any knowledge graph - that's the memories of an AI kid I've been trying to develop/raise.

fast pagoda
#

i love plotting memories in space

#

i was going to have some idiotbot guess a stupid pokemon but he's too smart

#

gemini said it was jigglypuff from above tho

austere glacier
#

...funny you mention claude.

fast pagoda
#

gpt somehow managed to get mew out of that

#

yeah looks pretty mew to me

#

just be careful when it hits like 1300 hours runtime it might suffer a neural net failure in its positronic net

austere glacier
#

Well right now Kes is being regulated with regular sleep cycles, where volatile memories get tagged and then dumped to the memory banks

#

Ideally, fuzziness and time decay will further dampen any possibility of overloading

hard raptor
#

Yaml storage still looks really cursed

nocturne olive
#

Silly

#

NeuroSynth config files are YAML too

hard raptor
#

i could also tell the storage manager to use {k: string, v: any} instead of [string, any]

#

yeah that looks less cursed

rigid snow
rigid snow
hard raptor
rigid snow
#

do you have to use key value pairs

#

like that

#

in yaml

hard raptor
#

i use yaml mostly for debugging tho cuz otherwise i would have to reformat my json files every time the data gets re-exported

hard raptor
stark needle
maiden geyser
nocturne olive
#

Is that a 3D printer?

stark needle
nocturne olive
#

Would be a fun thingy to have

stark needle
#

I have a new hobby now of 3d printing random bs

uneven pulsar
#

let reply = "";
try {
const sessionKey = user_${userId};
let generated = userMsg.trim();
const maxTokens = 50;

Now

let reply = "";
try {
const sessionKey = user_${userId};
let generated = userMsg.trim();
const maxTokens = 1639 ;

uneven pulsar
uneven pulsar
fast pagoda
#

i love the copyparty dev

#

not copyparty repo but all of his repos are great

sick owl
#

Lmao, it figured out I was on OLED and made a little comment about it

#

OpenAI have gone a ways to making their models a lot less obnoxious to work with of late

solid bough
blissful geyser
#

That's even worse

#

shouldn't it be like main.py or something

solid bough
blissful geyser
#

the main class should atleast be labelled as main

solid bough
#

The exe at the bottom is actually a packaged version

blissful geyser
#

Yeah i know

#

Its pyinstaller right

solid bough
blissful geyser
#

You also used a the --onefile parameter

blissful geyser
#

You can change the icon on that executable.

solid bough
#

Yeah, but I had no idea what one would even be good

blissful geyser
#

Strangely enough defender always marks python executables as malware

#

Unless they updated that somehow

blissful geyser
blissful geyser
#

Like windows 11 icon

solid bough
olive sable
solid bough
#

What do you mean Sam ▲

olive sable
#

I woke up

olive sable
solid bough
olive sable
#

Probably not

solid bough
#

Well, Idk if you would define it as a mess

rigid locust
#

Hey, im a coding newbie. Can someone tell me why when i run this code the output is:
396
None
Username

olive sable
#

You never defined finalhealth

#

You cant = a print

#

Afaik

rigid locust
#

Ooooooohhhhhh

solid bough
#

What the chocolate cake is going on in the code

rigid locust
#

Thanks

rigid locust
solid bough
#

That should work:py basehealth=400 attackdamage=4 username=input() finalhealth=basehealth-attackdamage print(finalhealth) print(username)

rigid locust
solid bough
#

Wait

#

Huh

#

I did something wrong too

rigid locust
#

Hm?

#

It worked fine?

solid bough
#

Okay, I pasted it in on the console

olive sable
solid bough
#

Yeah, when using the editor with the same input was working:

sullen marsh
#

zelix klass master got cracked & deobf

#

gg

olive sable
solid bough
#

Okay that took me off guard

olive sable
#

I cant read (mandarin???)

sullen marsh
#

from b2

olive sable
#

Thats a video?

sullen marsh
#

yea

rigid locust
#

Got a question, is there a way i can take three inputs take the middle input and use it within the code itself? (I want to make a calculator)

solid bough
#

Even if I wanted to

olive sable
solid bough
#

I'm back in a byte or word

olive sable
#
If operation == "*":
    Output = a * b
Elif operation == "/":
    Output = a / b
heavy quest
#

exec() is always an option (easy option, but can be used to run any code so a lil scary)

olive sable
#

Oh ye

solid bough
#

So wait, Someone can have a Print in a variable like that? py test=print(2+5)

rigid locust
solid bough
olive sable
#

I dont think so?

solid bough
rigid locust
solid bough
#

LMAO WHAT

heavy quest
solid bough
#

Right?

#

Well, that surprised me

heavy quest
#

print() doesn't return anything, but it gets executed so yea

rigid locust
nocturne olive
rigid locust
nocturne olive
#

Sily

nocturne olive
#

Back to NeuroSynth

heavy quest
stark needle
olive sable
#

Oh

stark needle
#

And like 20 bucks for 1kg of plastic

olive sable
#

Not that much

solid bough
stark needle
#

Ye

solid bough
stark needle
#

Miku I'm printing

heavy quest
solid bough
stark needle
#

I'm gonna give the miku to my sistee

#

She's ultra fan of that stuff

solid bough
heavy quest
# solid bough To use exec in my code

I mean it has a lot of useful applications, but it can be considered unsafe in environments that could be exposed to the internet or smth bc its ability to run ANY code

#

but for simple projects, it should be fine

solid bough
#

But it is public

stark needle
solid bough
rigid locust
solid bough
#

Pi is 3.1415926

heavy quest
rigid locust
#

Okay, could you rq write a simple string to show exactly how?

#

So i can get that visual understanding

solid bough
#

Haklyne I think there is not really a reason for me to add exec on my code ngl

rigid locust
#

Does exec just run the code without manually having to run it?

solid bough
#

idk, I never used that

faint sandal
#

yes

solid bough
# faint sandal yes

Okay kinda curious: Is it gonna act like a thread or as if I used "main()"

faint sandal
#

python runs everything on the main thread by default

solid bough
heavy quest
rigid locust
#

That makes sense

olive sable
solid bough
#

Well, on that note, It's time for ♫Dinner♪

hearty notch
#

mogu mogu

rigid locust
#

Quick question what is Xcode

hearty notch
#

apples most prominent ide

hearty notch
#

hell yea brother

#

for web stuff they're still picking low hanging fruit optimizations to wasm I think and vite rolldown speedups are mainly from that

olive sable
hearty notch
#

condolences

hearty notch
maiden geyser
olive sable
#

i could spend it if i wanted to

#

but its their house

stark needle
#

my house my rules moment

olive sable
#

ye...

hearty notch
#

so are they saying if you buy it they won't let you use it inside the house or like they refuse to have a place for it

olive sable
#

they're saying i should save up more, so idk tbh

#

"ask your dad" moment

hearty notch
#

oh so their pov is like "this is too big you should be more hesitant to spend such a big chunk at once"

#

that kinda thing?

olive sable
#

ye

hearty notch
#

kinda makes sense but shrug it's your spending money

maiden geyser
#

fair

olive sable
#

ill try again next month

hearty notch
#

it kinda implies a sense you might get bored with it or get over it which may or may not be true tbh a 3d printer is one of those things where if you don't fall in love with using it it might just collect dust

olive sable
#

fair enough

hearty notch
#

ok new ideas for fra: initing lab namespaces with single cli line, auto creates dir namespaces + registered modules for them, and with a system to promote them from a lab namespace outside prod to inside prod with a single line

#

frankengit must grow

#

generalized namespace rename helpers tbh git is powerful for this but half the users dont know

sage crag
olive sable
glad path
olive sable
#

no u

young plover
#

Hello strangers,
I am a rare bold person who considers themself good at C++. Unfortunately all of the cool pre-j*b projects I worked on have been [redacted] by Nintendo ninjas.
Something I can show that might interest people here - I use my own simple sharp scaling algorithm when watching video at my computer. Can try it on my janky github site:
https://breadfish64.github.io/scaleforce/?sauce=https://i.imgur.com/FZwCzDd.png

faint sandal
#

why is it two gigglebytes

young plover
#

thonk 2 gigs before compiling is odd

stark needle
#

migu 3d print done

hearty notch
#

ok now fra is a wrapper for every single scary git thing in more natural and ergonomic language

faint sandal
#

aur by default downloads the entire git repo (no depth control)

hearty notch
#

  1. git rebase (TERRIFYING)

  Why scary:
  - "Rewriting history" sounds destructive
  - Interactive rebase (-i) is confusing
  - Merge conflicts during rebase are panic-inducing
  - Fear of losing work

  Wrapper idea: fra sync
  # Instead of: git fetch && git rebase origin/main
  fra sync

  # Does:
  # 1. git stash (if dirty working tree)
  # 2. git fetch origin
  # 3. git rebase origin/main
  # 4. If conflict: clear instructions, show files
  # 5. git stash pop (if stashed)
  # 6. Preview what happened

  # Benefits:
  # - "Sync" is non-scary word (not "rebase")
  # - Auto-handles stashing
  # - Clear conflict resolution guidance
  # - Can abort safely

  ---
  2. git reset (DESTRUCTIVE)

  Why scary:
  - --hard sounds dangerous (IT IS)
  - Easy to lose uncommitted work
  - --soft vs --mixed vs --hard confusing
  - Fear of permanent deletion

  Wrapper idea: fra undo
  # Instead of: git reset --soft HEAD~1
  fra undo commit

  # Does:
  # 1. Show what will be undone (commit message, files)
  # 2. Ask for confirmation
  # 3. git reset --soft HEAD~1
  # 4. Show "Your changes are still staged, safe to re-commit"

  # Variations:
  fra undo staging     # git reset (unstage all)
  fra undo changes     # git checkout . (WITH WARNING + CONFIRMATION)

  # Benefits:
  # - "Undo" is clear mental model
  # - Preview before destructive action
  # - Never defaults to --hard (too dangerous)```
faint sandal
young plover
#

bro I can't wait to confirm git reset --hard
that's like, two seconds of productivity lost

hearty notch
#

ngl i dont know what a lot of this means

#

and someday nobody else will either because theyre all using fra wrappers

young plover
#

git rebase -i is like 50% of my git command line usage
the help text is right there in the editor glue

hearty notch
#

look by the nature of this thing everybody who is good at git already will think mmy thing is stupid and unnecessary but thats not the point

#

im here for the noobs

#

ALSO this isnt hte core feature the core feature is still modules and a server wrapper layer that modularizes monorepo pulling and actually deduplicates a workflow for a team

amber fractal
hearty notch
#

thats just self-aware bombast

#

nobody should feel threatened tbh

olive sable
#

i still barely know how git works, and i've been using it for over a year KEKW

hearty notch
#

same

#

tbh

#

lmfao

glad path
#

i just use the desktop client

#

since command line git yells at me if i try to use it

olive sable
#

i use the git stuff built into vscode

hearty notch
#

ive never seen the desktop client in my l

#

oh if you mean the vscode ext

#

then yea i used that a lot

#

before switching to cli only workflow

glad path
#

i literally just can't get the credidentials set up correctly

glad path
#

github desktop

hearty notch
#

yea ive never seen that ui for that in my lief lmfao

glad path
#

the only reason i don't use the command line is because it's like literally impossible to set up the credidentials correctly

hearty notch
#

its rly not you should ask any llm for a step tby step

#

which shell do you use

#

what platform are you on

glad path
olive sable
#

what are fra wrappers?

young plover
#

Github has a help page for setting up an SSH key

glad path
#

literally told me to go to nonexistent settings

#

like i

#

i

#

i just wanted to install homebrew

hearty notch
#

i am making an app right now called frankengit and the idea is to wrap git in ergonomic extra commands that combine git commands and makes them clearer by name + has a few project init opinionated scaffolding system that preregisters namespaces and makes it easier to work across those namespaces

glad path
#

and git fucking decided, nope, you can't view a PUBLIC repository becase we don't have your account

#

and as such we don't know if you have access

#

girl its literally a public repo

olive sable
#

idk, never had that happen to me

glad path
#

why would i not have read access

olive sable
#

sounds like as kill issue

glad path
#

so yeah my git cli is lobotomized for some reason

young plover
#

thonking weird
don't remember having issues cloning and pulling public repos without creds

hearty notch
#

just open a terminal with root or admin access?

glad path
young plover
#

it wouldn't

hearty notch
#

tbh this suggests to me

#

that theres some malformed config

#

somewhere in your shell or git installation

glad path
#

probably

#

wait can i just

hearty notch
#

just full delete everything and reinstall everything imo

glad path
#

is it possible to just reinstall git with homebrew

hearty notch
#

idk homebrew but idk why not

glad path
#

because homebrew depends on git i think

hearty notch
#

just make sure you get the config dirs

olive sable
#

am i missing context?

hearty notch
#

yea that sounds wrong to me too but im not confident enough to 100% dispute it lmfao

glad path
#

wait

#

i can just install git with homebrwe

#

and it should theoretically override whatever I already have

olive sable
#

homebrew just means you use your own software, like installing linux on a playstation 2

#

it has nothing to do with git

glad path
#

i know

#

i know what homebrew is

hearty notch
olive sable
#

whart

glad path
hearty notch
#

homebrew is the name of the main package manager mac terminal users use

olive sable
#

i odnt use mac

glad path
olive sable
#

that sure iss omething

hearty notch
#

its a fair confusion i learned about what you knew about before i learned about this too

#

but yeah every mac dev knows this

olive sable
#

i was missing context

glad path
#

but yeah

#

since homebrew likes to override other things

#

like my python install

hearty notch
#

yea installing through brew might just make it work but idk how the namespacing actually works

#

you probably know better than i do

#

in terms of like which installation wins

glad path
#

in theory it will override the system installed git

#

brew has the newer one

hearty notch
#

yeah idk

glad path
#

yeah i guess the system install wins

hearty notch
#

this is how you make homebrew win

glad path
#

too lazy lol

#

found it

hearty notch
#

nice

#

hopefully that fixes it forever

fast pagoda
#

Homebrew works on Linux too

#

It's the only thing that makes Debian distros comfortable

opaque sigil
#

did they ever make auto updating optional enub

nocturne olive
#

Reaper too hard, I'll keep mixing important complex stuff in LMMS for now

hearty notch
#

open call for anybody that has ideas to improve their own git experience @ me with feature requests

#

cli only

nocturne olive
#

WHa

hearty notch
#

im making a git wrapper

#

im trying to make git easier and more ergonomic

#

@ me with ideas if you have pain points with git

glad path
hearty notch
#

lol sure

#

done

#

(not literally done but i mean it will be)

opaque sigil
#

git credentials being a pain is on purpose to get you to use ssh enub

rigid locust
#

So whats the difference between implicit and explicit data types? Im assuming that implicit is essentially just making use of whats already there while explicit is forcing the data to be what you need it to be?

#

I hope to god that makes any resemblance of making sense

opaque sigil
#

in what context, type conversion?

rigid locust
#

Yeah, thats what im learning right now, but the thing i use doesnt really explain or elaborate

#

Its just like “this does this” and doesnt explain further

opaque sigil
#

there are some combinations of types where the actual value stays identical so you can do the conversion implicitly

for example every number that fits into a 32 bit integer also fits into a 64 bit integer so to make the programmer's life a bit easier most languages will just let you use a 32 bit int as a 64 bit int implicitly

if you go the other way though that's no longer guaranteed to work so you need to explicitly cast it to just assert "i know what i'm doing dw this number fits into 32 bits"

rigid locust
#

Ohh okay so explicit is essentially the highest common multiple of the coding world?

opaque sigil
#

i don't think i understand what you mean PikaThink

rigid locust
#

Im like right there with you, ill ask alexa or not worry about it right now

hearty notch
#

implicit means you assign it without saying it and it maps to the most obviouos guess

#

foo = 5

#

5 is an int

#

it could be a float (has decimal precsion) but it isnt

#

foo = 5.

#

it initializes as a float

#

in strongly typed systems, you have to explicitly declare

#

what data type it is

#

tbh

#

if you want to learn programming top down

#

C is

#

the classic recommendation

#

strongly typed languages supersede easier languages

#

if you learn C you will find python a breeze

#

they also teach more about what goes on under the hood

#

i guess i meant bottom up

#

not top down

#

its both in some ways but C is lower level which means closer to machine code so top down is wrong in that regard

#

machine 1s and zeroes < the literal cpu/gpu architecture < assembly < C/zig/rust/C++/(improbalbyforgettingalot) < most higher level languages in terms of amount of abstraction of what actually happens

glad path
#

python has gradual typing

#

you can typehint everything if you want

hearty notch
#

if you truly stay in the programming world you will eventually realize that in complex projects

#

the more clear you are about types

#

the easier everything stays as things grow

#

type confusion and issues from types not being explicit in dynamically typed languages

#

just creates a lot of issues at a point where you cant turn back

#

sometimes

#

not always

#

theres been a movement in software to just make everything typed because of this and its like fairly mature now though

#

its why typescript exists

rigid locust
#

So for example “Broke” is implicitly a string

3.0 is implicitly a float

And 3 is implicitly a int

hearty notch
#

its explicitly those things as soon as you assign it, the assignment was implicit because you werent explicit about it

#

there's a thing called type coercion

#

where if you put the wrong type into something

#

sometimes itll try to make the best guess at what you want

#

and transform it to what type it thinks it shouldve been

#

sometimes this works and helps

#

but sometimes its breaks things

rigid locust
#

So essentially explicit is just correcting whatever the computer makes an assumption about? Like forcing it to be what you meant it to be?

hearty notch
#

its not correcting, its in some languages, you are forced to say it the first time you declare it

#

var int = 3;

#

or i forget c syntax lol

#

explicit typing is a language feature

#

python doesnt force it

#

some languages do

rigid locust
#

i think the terming is just really damn weird.

hearty notch
#

what do you find weird

glad path
#

python is useful because you can gradually type things
so you can explicitly point out some things but not every single thing

rigid locust
#

Like idk, explicit would make sense to be stating what something should be

While implicit would make more sense to be implied

glad path
#

btw final variables also exist in python

rigid locust
#

Wym

glad path
hearty notch
glad path
#

a is explicit, b is implicit

hearty notch
#

in the second example in C, you have to declare name[] as an array of characters

#

to be a string

rigid locust
#

I thought using # forced the computer not to register what was being said? Like almost like commenting

glad path
glad path
#

in python atl

#

im just using it to show what the output of that line would've been

hearty notch
glad path
#

python is an easy way to look at it though

rigid locust
#

So does explicit change anything? Or is it just foolproofing?

hearty notch
#

its foolproofing in languages that dont require it, its literally required in languages that do

rigid locust
#

???

glad path
#

it's required in a lot of languages

#

frankly it's a good thing to do

hearty notch
#

some langauges jut force you to always delcare type when creating anything

#

and yea it saves headaches in the end

amber fractal
#

As someone who has dealt with a ML codebase in python and handing to bash basedpyright. Pray that everyone respects typings.
-# the amount of ignoreAny comments is insane

hearty notch
#

the languages that 'look' and 'feel' harder usually have really good reasons for why theyre harder

#

you just have to like learn it

rigid locust
#

Oooh okay so its just teaching me things to get me into the habit of doing things ill need to do when i learn other languages, despite me not necessarily needing to with python?

hearty notch
#

well 1. its teaching you vocab you shoudl know and 2. yea you should expect it if you plan to branch out

#

this is why im recommending C

#

as something to look into

#

the lessons from C propagate into other statically typed and strict languages

rigid locust
#

(Also going to assume it makes it so everything isnt forced to be gone over multiple times to auto correct itself)

hearty notch
#

your planned path is fine though python is a shallow water kinda entry and then if you learn javascript next, just learn typescript with it

#

and youll develop some good typing habits

glad path
#

because a library isnt typehinted

olive sable
#

hi iggly neuroWaveA

glad path
#

or you literally can't find the typevar anywhere somehow

rigid locust
#

Okay so my plan to go python javascript is a good idea?

amber fractal
hearty notch
#

tbh

#

if you ever get really confident

#

in a software project you want to do

glad path
hearty notch
#

@ me and i will telll you language recommendations

opaque sigil
#

what else is it supposed to return

glad path
#

very annoying

hearty notch
#

because picking language based on project before you do it can be big

glad path
hearty notch
#

otherwise do whatever order you want

opaque sigil
#

but it can be anything

glad path
#

I don't know javascript

#

lol

#

i dont like javascript

rigid locust
#

Well right now im only day 2 of learning so i think im thinking too far ahead

hearty notch
#

yea

glad path
#

its just a weird language

amber fractal
glad path
#

so i refuse to learn it

hearty notch
#

what are your preferred languages

rigid locust
#

Ive heard that C++ is a good language to learn

hearty notch
#

i just cant avoid javascript if im gonna do web ui/ux so

glad path
hearty notch
#

C++ is big in gamedev ig

glad path
#

actually

amber fractal
#

Python preferred (prepared to have people execute me)

glad path
#

hpp over cpp maybe

olive sable
#

pretty much all the grpahics api's have C bindings

#

so c++ is nice for graphics

hearty notch
glad path
hearty notch
#

but nice

glad path
#

idk why people hate it so much

#

it's used so much for a reason

hearty notch
#

its funny your path is entirely orthogonal to mine

#

oh wait no i have some python stuff nvm

glad path
#

also

amber fractal
hearty notch
#

python is great for anything that wont become a prod machine and sometimes its still good when it is

rigid locust
#

Whats the best coding language for rigging vtuber models…. I want to do it cuz vtubing is increasing in popularity and i think i could make good money rigging

glad path
#

hpp is one of the weirdest languages ive used

#

its like

olive sable
#

python is nice, just a bit slow sometimes.
im willing to bet 80% of people complaining about the speed are just doing inefficient algorithms or have unreasonable expectations

#

like me

glad path
#

its turing complete

#

but it lacks so many modern features

amber fractal
rigid locust
#

Vedal mindset fr, money on my mind

olive sable
rigid locust
glad path
amber fractal
glad path
hearty notch
#

it has some coding elements but its more like 3d modeling

rigid locust
#

Hm

amber fractal
olive sable
hearty notch
#

you also probably need a pretty powerful computer to do it well without it being painful

amber fractal
#

I was mixing up triangle :classic:

olive sable
#

🔺

rigid locust
#

Im gonna take a break for a bit, i understand why vedal is tired all the time now.

#

This shit steals brain power and energy

olive sable
#

im still on the grind BlepCat

amber fractal
#

I need to make one of those exponential gifs with the triangle FPS

rigid locust
olive sable
#

no

rigid locust
#

Be cool tho

olive sable
#

this is for "untitled band game"

glad path
rigid locust
#

Untitled goose game’s sister?

olive sable
amber fractal
olive sable
#

its an event-planning rogue-like

#

or something like that

olive sable
glad path
#

4px i think

#

i might have made it 3 or 1 even

#

there was no limit on the size of the triangle

#

so i just made it really small

amber fractal
#

Yeah I think the 7px limit was added later

olive sable
#

i jsut heard the windows vista strartup sound in the hallway KEKW

glad path
#

i hit 40k after a coldstart (calculator had shut down fully for a bit)
only possible since the memory was like completely cleared

#

if i run anything first it cuts the fps tenfold

neon timber
olive sable
#

not like an official rule or anything iirc

thorny dragon
glad path
#

i don't remember the code I wrote for the triangle

#

it might be inaccurate

#

I think it incremented a counter by one each frame and stopped after one second

olive sable
#

im sure its fine

#

i don't feel like policing that tournament anyways

glad path
#

all i remember is i actually optimized the fps counter itself

#

I should probably try on newer firmware

#

and also on a non fragmented calculator

#

I think I somehow wore out the ssd already

olive sable
#

i dont think its worth it

glad path
#

even though it was like a month old when I did that

olive sable
#

we could update the leaderboard i guess, but the triangle arc has passed

glad path
#

that's unrelated honestly

#

but they fixed some memory leaks

#

but somehow there's nothing in the patch notes about math.abs() crashing the whole os

olive sable
#

math.abs() does what? WHAT

hearty notch
#

removes negative signs

#

unless someone is pranking me and using abs for something that isnt absolute value

#

might also calculate magnitude of vecotrs idk

#

probably not tho

glad path
#

no

#

it is absolute value

#

and it crashes the whole os

#

for no reason

#

no matter the number i think

hearty notch
#

lo lwhat

olive sable
hearty notch
#

oh lol

young plover
olive sable
#

in my experience in C++ its not too hard to not screw up

#

so that means when you eventually do screw up its pretty yikes

hearty notch
#

some of neuros code is in C# because its the microsoft ecosystem and also has integrations with unity

young plover
#

I prefer C++ over Rust because I run into the limits of Rust's compile-time metaprogramming, but IMO internalizing the rules that Rust teaches about object lifetimes is necessary for large C++ projects.

hearty notch
#

if youre writing c# its usually because you need to for some unity integration or youre integrating something microsoft

young plover
#

I work on a WPF GUI in C#, but native desktop apps are a dying breed PepeHands

#

C++ for embedded firmware and misc. personal projects

olive sable
#

i jsut try to do RAII where possible

#

vulkan has some problems with that becuase you need to pass a lot of arguments to the init and destructor functions

young plover
#

Arguments to destructors?
Haven't touched Vulkan in awhile but I remember avoiding actual graphics pipelines like the plague because it was triple the boilerplate of compute shaders.
The constructors are nice in the C++20 bindings because you can use designated initializers. { .arg1 = ..., .arg2 = ... }

young plover
olive sable
#

for example, a 3D model with a texture, you need to pass VkDevice device to the destructor, which means you actually cant use a destructor

#

the only way to fix that is to keep a copy of VkDevice int he object

#

but im not sure if i want to do that

young plover
#

Ah, yeah I think I used the vulkan.hpp bindings, which must store the device with the allocated pointer.

olive sable
#

im pretty sure VkDevice is also just a pointer

#

so it woudlnt be too bad to keep it as a class variable

unreal coral
#

Yo guys
Does anybody know an ai thats able to write code for an Api?
Basically chatgpt but without having to pay 230€ for unlimited use

#

But i also can't just use a normal code generation one cuz the ai would also need to know how this game's api works :0

olive sable
#

uh

#

Shruge not much is free

unreal coral
#

Oof so they are all limited in how much you can use them.....

#

None that work with ads etc?

olive sable
#

im not sure tbh

unreal coral
#

Oof
Maybe i should make smth that just creates new accounts then

glad path
#

wait

#

just get lm studio

unreal coral
#

So i can just use the free stuff on unlimited accounts

glad path
#

you can literally use the openai api and aim it at LM studio

unreal coral
#

And i just switch to a new one when its at the limit for one

#

Oh? :0

glad path
unreal coral
#

Oof its for pc .-.

glad path
#

what do you have

unreal coral
#

Im on mobile >.>

glad path
#

oh

unreal coral
#

Android

olive sable
#

oh

glad path
#

that complicates things

#

a lto

olive sable
#

ye you cant run much ai on a phone

#

only cloud stuff

unreal coral
#

I mean chatgpt works perfectly for what i need
Its just the limit thats screwing me here

#

And i'm not spending 230€ for unlimited use

glad path
#

what limit

#

you get unlimited use normally

#

i think

glass flower
unreal coral
#

It stops you from talking to chatgpt after a certain amount of prompts i think

#

Not even mini i think

#

Unless they changed it

glad path
#

thats like a lot of prompts

glass flower
#

oh yeah that one too. if the context grows too big

glad path
#

ive NEVER gotten stopped

#

even after huge conversations

glad path
glass flower
#

i got limited once when i wanted to convert c++ to rust code mhm

unreal coral
#

Well i had to explain alot about the api

#

So i'd have to fix less in the end

glad path
#

write a readme

opaque sigil
unreal coral
#

Maybe
Although the idea with the new accounts could be smth no?

olive sable
#

i think that would get bothersome quickly

glass flower
#

mhm you could also try to learn the api yourself and write the code instead of relying on AI for it

unreal coral
#

I did

young plover
unreal coral
#

I COULD basically do anything i wanted

#

I just dont wanna write it all down

#

Much better to get smth thats 99% done and i just look through it and fix a few mistakes that might be there

glad path
young plover
#

I've used ChatGPT once
I asked it to translate a python program from Spanish so I could help someone with their homework
It worked mostly
Never touched it again

glad path
#

lol

olive sable
#

honestly, i dont like ai generated code

#

i will ask chatgpt for code on ocassion, but then ill heavily modify it afterwards

young plover
#

I mostly work on a 30 year old 500k+ line C++ codebase with multiple different stages of modernization
vibe coding simply does not work Kek

#

my coworkers have tried

olive sable
#

i feel like ai code is very middle of the road generic solutions, that dont really get too situation specific enough

opaque sigil
#

crazy how the thing made to create averages creates averages right

#

i will say

hearty notch
opaque sigil
#

i have started using copilot's agent mode for refactoring and it's been kinda nice for that (at least simple stuff like adding wrappers and updating call sites with the new api)

hearty notch
#

the stricter you are the bettr

#

except you can also be vague sometimes if you want

#

in terms of like "brush everything up and find optimizations"

#

but like in terms of designing something you want your prompt super specific

rigid snow
#

there's nothing not bannable to say

#

it's that bad

hearty notch
#

wdym say it

#

oh you mean you hate your experience?

rigid snow
#

i am very mad

#

yes

hearty notch
#

what are you trying to build

rigid snow
#

nvm

hearty notch
#

ok condolences lol

rigid snow
#

not node version manager i said nevermind

#

to clarify

hearty notch
#

lmao thats what i thought yea

rigid snow
#

i don't have time so i have to resort to this shit

young plover
olive sable
#

it has been decided that input will msot likely be a midi file

rigid snow
#

lol

#

i was joking

#

are you still animating a guitar

#

good fucking luck if so

olive sable
#

i wasnt joking

#

i was very serious about it

#

we're doing 2 guitars, a bass guitar, and a drum set

rigid snow
#

^joke

olive sable
#

oh

#

i don't even remember that

#

but ye, it seems like midi will be the most compatible

rigid snow
#

do you have a plan to make it work with how guitar is usually played

olive sable
#

the plan is to minmax the animations

rigid snow
#

or yolo

#

it

olive sable
#

i want it to be organic looking, but not to the point where i have a meltdown trying to program it

#

so like, if there is a note, im gonan have a single way of how it gets palyed

#

not like how in irl you only move the fingers you have to move

#

or whatever the fuck that 2 handed guitar playing was

#

using midi files should not be too hard COPIUM

latent musk
olive sable
#

easing functions mostly

rigid snow
#

the hard part isn't the animation it's deriving which fingers to place where and when

#

i think their comment is about that

olive sable
#

oh

#

well, fingers go brrr

#

idfk

latent musk
#

There are definitely ways but yeah it's making the fingers actually place well if that makes sense

olive sable
#

i think in terms of the midi -> movement the drums will be the easiest cuz i jsut have to deal with 2 hands and 2 pedals, and the guitars will be a pain in my ass

latent musk
#

Drums yeah shouldn't be hard

#

It can be pretty sloppy and still visually look decent

olive sable
olive sable
young plover
latent musk
#

As long as you don't have insane clipping it should be fine

#

But yeah guitars are a pain

olive sable
#

from what i can tell, most of the time you wont have 2 fingers on the same row

#

the frets

#

i feel like there shouldnt be too many combinations of how the fingers are psoitioned

latent musk
#

Last time I attempted it I just gave up and said make sure no one will ever see it up close

midnight sigil
#

big universities trying not to lose every research file: impossible

hollow matrix
fast pagoda
#

when you do a quick dedup on the videos folder after flawlessly executing some quick relocations for spring cleaning

#

ope accidentally duplicated a few things

hearty notch
#

I LOVE TOML

uneven pulsar
#

Server engine is reparing.
Custom feature of ai os
Interface is user main os is ai

uneven pulsar
fast pagoda
#

i do too actually

#

parquet giga

hearty notch
#

what ecosystems does parquet appear in ive never heard of it

fast pagoda
#

apache

#

huggingface uses it for datasets

#

which is where i became familiar with it

hearty notch
#

ic nice

fast pagoda
#

large sets will be autoconverted to it when theyre in the datasets viewer tool and such

#

and training scripts parse it p well

#

i convert other formats to parquet usually

#

for datasets only

#

it's good at io reads

uneven pulsar
#

I have gigs of data like 193 to 30 gb so I can't manually train doni convert into python into parquetand use it

hearty notch
#

i like toml for config files and declarative templates for dsl codegen so different worlds lmao

fast pagoda
#

toml syntax is less panic inducing than yaml so it's got that going for it

#

i dont like leaning on the indentation that hard

#

it is spooky invisibru or rainbow but it's not real i cant see it

uneven pulsar
#

Mee to

fast pagoda
#

toml has brackets and such that make me feel comfy

hearty notch
#

yea its comfy to look at

uneven pulsar
#

( ai training is hard as hell )
I will do fine - tunning

hearty notch
#

yaml is ok but toml is comfier whne i looka t both

fast pagoda
#

i should try a random echo training run with the new data i got and then did nothing with a bit ago

uneven pulsar
#

We make it
I = letter_i
T = letter_t

fast pagoda
#

t mentioned

hearty notch
#

e

olive sable
#

a

uneven pulsar
#

T

olive sable
#

bruh

hearty notch
#

i thought aobut it too ngl

olive sable
uneven pulsar
#

Weather is clear

olive sable
#

it shoudl have been "sports"

fast pagoda
#

kek

#

im watching baloo and czkawka battle to the death trying to access a spinning hdd at the same time

olive sable
#

im at the point of game dev where all the youtube videos about graphics are very surface level

#

and the only other things left are actual talks by gdc or whatever

fast pagoda
#

now it's time to create the fast inverse sqrt of 2025

olive sable
#

i feel like with the specialized hardware stuff for math inside modern processors its really not worth it to do that

fast pagoda
#

it's not

olive sable
#

90% of rendering performance is dependent on overdraw anywyas

fast pagoda
#

not for a long time was that actually super relevant

#

but now since you have ascended

#

you must craft the fast overdraw gc or something

#

so someone can circle it in the codebase and go

#

WTF

olive sable
amber fractal
#

As a wise person once said, culling takes performance as well.

olive sable
#

you need to strike a balance between culling and just rendering it even if you cant see it
rendering 100 triangles takes maybe 5 nanoseconds, calculating if they're visible will take more like 10

fast pagoda
#

jensen say
what about rendering 100 triangles, hallucinating 300 for framegen, and culling
add moar fraem & surely input lag disappears

#

surely

amber fractal
olive sable
#

its actually insane how battlefront 1 can render at 8K 120fps highest settings on a 5090, and battlefront 2 bartely runs at all

fast pagoda
#

star wars battlefront (2004) and star wars battlefront II (2005) continue to run fantastically idk what you're on about, some irrelevant game ig

olive sable
#

or was it batlefield?

#

idk

#

im probably mixing them up

fast pagoda
#

probably battlefield but i dont care about it so idc

#

there's been a reboot of sw bf1 and bf2

#

but they dont exist

olive sable
#

battlefornt 2 did not release in 2005 did it?

fast pagoda
#

the first one did yeah

obsidian mantle
#

I think battlefield2 is old game and battlefield1 is modern

fast pagoda
#

well the first battlefront II

obsidian mantle
#

Is this about physix thing? Or how is it called

fast pagoda
#

and battlefield 2 is an old ass bf game but i thought they just laid a fat turd of a new one or something

olive sable
#

this is so confusing

#

wait

fast pagoda
obsidian mantle
#

I heard some old games dont work properly on 50 series because they removed some physix cores

olive sable
fast pagoda
#

lol i did yes and im glad that confused you so much

olive sable
#

i didnt know there was a first one

fast pagoda
#

but yeah i don't acknowledge either of those games existing

#

there is only the first one

#

i used to be in a clan/guild in the 2005 battlefront II since i played it a ton into the early 2010s called IGF (imperial ground forces)
we greeted each other with S (for salute kek) and had ranks and duties we had to do and shit like it was some stupid barracks simulator or something
i had to play a minimum amount of public lobby games in gamespy servers cuz private server browser >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> whatever slop roulette there is now

#

it's so cringe

#

was fun thou

olive sable
#

imagine being old enough to clan a guild in 2005, couldnt be me

fast pagoda
#

mostly played hero assault and dueled people with infinite DPI on my mouse spinning around w/ vader in mos eisley

#

this would've been like 2008

#

the game came out in 2005 tho

#

idk when i started playing

#

this was back when xfire was around

olive sable
#

idk about gameplay wise, but battlefront 1 version 2 looks good and runs fine at 8K from what i can tell in this youtube video

fast pagoda
#

8k is on a 10 year old game is one way to pretend 8k is a thing

olive sable
#

oh, my hardware acceleration broke again

fast pagoda
#

hyprland moment

olive sable
#

i was wondering why it was stuttering, but 8K on cpu decode is expected to be stutery

olive sable
fast pagoda
#

i mean it looks better because it's higher res obviously

olive sable
#

but since this uses texture mapping and stuff the 8K does add some value

fast pagoda
#

but nothing runs at 8k that isn't old so it's not real

olive sable
#

ye buit graphics wise it looks great

#

so there should be no reason for modern games to not even get half of the performance

fast pagoda
#

well 10 years of shit bolted on is a lot

olive sable
#

it works perfectly fine for 4K and below, but 8K isnt supported on vaapi or whatever

#

oh

#

youtube jsut removed 8K from the options list for me now

fast pagoda
#

https://www.youtube.com/watch?v=yibrkmGAKqs this guy seems to have no issues with battlefront II at 8k

This footage showcases the performance of the RTX 5090 in Star Wars Battlefront 2, using 8K Native with Max Settings. Recorded with Shadowplay, which results in a 3-5 FPS loss.

▶️ PC Specs
✅ CPU: AMD Ryzen 7 9800X3D
✅ GPU: NVIDIA RTX 5090 Founders Edition
✅ RAM: 6600 CL 26 FCLK 2200
✅ Motherboard: X870E Nova WiFi
✅ Monitor: Neo G8...

▶ Play video
#

on a 5090

olive sable
#

hmmm

#

so what was the recent release that ran like shit that was being compared to battlefront?

fast pagoda
#

im guessing maybe battlefield 6 vs like the previous battlefield 5? is that what you're thinking of?

olive sable
#

i have no clue

fast pagoda
#

i was confused because i was like no way for either of these games is it hard for a 5090 at 8k since bfII was 2017 for star wars

olive sable
#

oh

#

ye what i was talkign baotu was way later than 2017

fast pagoda
#

nobody is gonna run 8k battlefield 6

#

at any reasonable rate

#

that's expected though, 8k is goofy

olive sable
#

fair enough

#

what exactly makes the battlefront remasters bad btw?

fast pagoda
#

4k is already quite hard to run and still makes nvidia pretend with fake frames on graphs

#

they are by EA #1 which is already enough

olive sable
#

oh so microtransactions?

fast pagoda
#

pride and accomplishment

#

the most downvoted comment in reddit history

#

to this day i think

olive sable
#

oh ye

#

my game history teacher was bitching about it too

#

i mean.. "educating us"

fast pagoda
#

so that on its own is enough for me to never play but then also they stole the name of 2 already released and well liked games which people had been asking about a sequel to for years (not that i think that would've been anything better or care but if you're not even trying to relate it at all to the old one pick a different name)

#

permanently tainted search results and the immediate reputation of the older ones cuz now i gotta clarify im not talking about the ea garbage before i get shanked

olive sable
#

oh lmao, my country spotted jsut below the "pride and accomplishment" section

fast pagoda
#

yeah you know you done goofed when national governments open an inquiry because of the reddit ama

#

i still got my downboat on there o7_salute_respect

#

and that ama killed the EACommunityTeam account permanently

#

lol

olive sable
#

-600K

fast pagoda
#

kudos to them for just keeping that shit up though

#

instead of deleting the entire account

glad path
#

why is this kind of funny for literally no reason

hearty notch
#

lol

fast pagoda
#

the bash version of those boxes that turn themselves off

hearty notch
#

goodbye neuroSad

glad path
#

"goodbye world" is real though

olive sable
fast pagoda
#

bet those brainoids they have in labs made of neurons running like 2+2 and random calculations all day wish they could goodbye, world

olive sable
#

i think that thought requires more neurons than they have