#programming
1 messages · Page 532 of 1
except for some reason with the new 7tv extension it defaults to the ffz chat so 7tv becomes completely useless
Well that clears it up!
Describe
No, i have exams
So I'm working on an ai architecture that instead of storing data and weights it creates tags that it uses the reconstruct the concept
Elaborate
That’s deep do you use uuid with sqlite3
So essentially every bit of detail is given a unique tag that the neurons create after shredding them. To give an idea of the tags We have different Qualia groups
Photonic-Q
Acoustic-Q
Haptic-Q
Olfactory-Q
Kinesthetic-Q
Emotional-Q
Spatial-Q
Cognitive-Q
That combine to form a concept like an apple
Unfortunately not because the neurons will speak in (-,+,×)
Will I’m doing something in my self hosted llm, for it to remember stuff about the conversation and about the user, consider it a multi memory layer
Interesting my system is similar in the fact that it's also multilayered
Tags > Concept > Scene > Event
I feel like your exam season is incredibly longer than mine
Ohh, yeah that is longer than mine
We have only one week for exam
Everything that can't be crammed into those one week are to be scheduled another time by the lecturer. Which to be fair, doesn't really matter to us but it is a PITA for the administration
Mine has a limited capabilities since I didn’t want to over complicate it, it’s already using gguf and mlx llama cpp and fathcing them, let alone apis and tools that require each model trained tools, so adding memory is a hassle even with the right model
I see
I really which I could run my own fine tuned model, i already have a good one pre trained, but t unfortunately i lack enough hardware
Dame how much ram
I DID IT
:D
i finally made a functioning ML archetecture that proved itself to learn
and it uses as much cpu as those solar panel calculators
while using 10gb of ram for 10B neurons :D
now comed the fun part
having to scale this ;/
it has the most potential
V1 failed bcus i went too complex and ram exploded
V2 failed bcus there was no learning signal or gradient
V3 is yet to fail :D
wont know until we see “soul” in an integral

I vote neither
they have a history of ~500 messages that are rather neurotypical
the real play would be it’s an elaborate prank
maybe they explored https://discord.com/channels/574720535888396288/1497044635900444782 and got some ideas
good morning
Best thread on the server btw
its very rough
but it sort of works
needs more improvements
uses neurons to steer a line to a output neuron
but only actually wakes the neurons that are in range to influence
so it keeps active processing low
running Gemma-3-270m adaptation running via a diy implementation of this paper that I’ve been waiting forever to try: https://huggingface.co/papers/2409.11901

good idea, waste company money and time as much as you can get away with it

what did I miss this time
not much
was it another ai bro/psychosis
nah, well maybe slightly
Idk I count it
20mhz single thread CPU and 10gb RAM is not a viable AI combo if one takes the explanation literally.
they're probably doing the math wrong, but it might be a working llm
rough but has more potential
exactly why im trying to make my own
I'm calling hard bullshit on the math here
I've experimented on and off in the space, and that combination doesn't make sense
Even for someone who is minmaxing performance as hard as I am
It doesent calculate everything
Not a transformer
Sparse routing
I just refuse to believe you did what you said you did in the compute quota you specified
End of that story
just look at the code
Ok?
smells ai generated
I can simply just use my own NN work to disprove based on raw logic alone
As that is the densest thing that current hardware can do, and that isn't even functional yet
So I need more compute than that
If I assume yours is 10x worse than my own. Then that'd be visiting about 200 params per second per CPU core. Quite the sparsity, however routing costs are going to eat you alive. Let alone attempting any GPU compute which surely it can and will rely on.
the code has
self.neurons = np.random.randint(0, 64, NUM_NEURONS, dtype=np.uint8)
so each neuron is a byte, with 5 bits for strength and a bit for the sign.
so 10B would indeed by 10GB
this is completely different form how normal NN work tho
you need weights between the neurons
TBH I haven't even gotten into the code yet
normal NN work via matmul, this is path-finding
I'm just using what I learned from NN to disprove the compute claim of running on a sunpowered calculator
Effectively so was claimed

I think konii likes cute animals
grrrrr
each token seems to be assigned a random start and end pos on the 2D grid, and when a token is processed it takes 128 steps across the grid???
as it moves it looks at a 7x7 grid around itself and each cell exerts a force on the particle that decides in which direction it moves???
Random walk 
its based on the random numbers that were assigned to the grid at the start
The neurons are the weights tho
The connections are implicit simply via matmul
this doesnt use matmul
We love cache miss 

run the code and see if it do thing
not on my pc lol
Surely it has a benchmark 
the biggest flaw that i can find so far is it has 0 memory
pred, _ = self.brain.run_token(current)
its based on the last word
Next token prediction 
i.e. markov chain
Not a flaw, just a con of early development
I'm not even insane enough to attempt next token prediction yet
this model would force the entire english language onto a 256x256 grid.
Extra dimensional has a pass
and there are slightly more than 65536 words
But this isn't thst
train a SNN instead
there's aslo ne semantical understanding
the word bank for "river bank" is the same as bank where i get my money
Because next token, it doesn't contain the resources to even carry that semantic difference in the first place
It is limited ye
its a fancy markov chain
I never doubted that it could actually train, I just question the effectiveness at scale
Once again, cache miss on 16kx16k array my beloved
lets just hope you don't end up killing the younglings
oh
they left the server
Who?
AGI the xth
this guy
(real)
agi posting lmao
+1 to the AGI counter
Lol why did he leave? Was too scared to show his AGI results ? 

He did show code, proceeded to be a poorly make markov chain
So eh token prediction? Aren't LLMs these days already using something like that? Lol
no, its a lot more basic than LLMs
The problem area lies in scaling a single 2d grid to encapsulate 10B params. 
Idk i was just reading what a Markov chain is
In probability theory and statistics, a Markov chain or Markov process is a stochastic process describing a sequence of possible events in which the probability of each event depends only on the state attained in the previous event.
Good luck filling the grid when you can only move in a 7x7 square
basically, llm looks at a long list of previous word to decide the next word, markov chain looks only at the last word
and this specific implementation did it in a manner that idk if its erven efficient but i doubt it
Oh i see so kinda like LIFO lol
Bro just use 1000000-grams markov chain
1 million context markov chain


1 million hop limit per token*
Train on the library of babel or whatever that one was called
What
Don't panic lol
The babel fish out of hitchhikers guide to the galaxy
The artificial intelligence company announced the data center will be positioned in the heart of the small, race car-themed bedroom where 8-year-old Billy Treaker fights a rare kidney disease on a daily basis.
Become an Onion member while it's still optional: https://membership.theonion.com/?campaign=701a500001geULHAA2
Finally, real news
agi datacenter vs coughing baby
this game is so ass
Why no nuked audio 
It's normal audio 
269 commits now
https://github.com/weenachuangkud/FastCast2/pull/39
Summary by CodeRabbit
New Features
Added serial and parallel caster modes, Motor6D transform movement with pooling, and client-side serial/parallel FPS benchmarks.
Bug Fixes
Fixed high-fidel...
@olive sable I FEEL THE SLEEP IT BECKONS ME
but I gotta actually do stuff today
cuz im listening to that
Look at the bright side, it is Friday
I'm gonna vibe code some shi
then sleep

sleep is a made up construct
Sam did you see me terrible soldering job from last night
then dont sleep, i odnt care
i fell asleep before i could see probably
cheese boigers
but anyway nini hope u all get good food and eep
goodnight
I still feel a bit stupid because I spent ages trying to solder to a ground
it probably did help you get used to soldering, making the other ones turn out better
I mean
I can already do soldering alright
like if you look at the wires in the picture I sent they're perfectly fine
or well specifically 3 grounds right next to each other
solder thrice, measure once 


Ok i just found Something cool that i Always wondered if it exists and its as beautiful as imagined
Diffusion based llms-
Whats the opinion on em? (The example i found was Mercury 2 by inceptionlabs, are there any other?
Up your solder temp
So you can heat it quicker than the heat dissipate to gnd
it kinda just wasn't worth it
It may be needed if the signal integrity is shit tho
I already had a ground from somewhere else
Holy endgame mic 
I have a TLM-102
I like it, it has a brighter sound to it that makes the harmonic frequencies in voices sound really good
yeah those are also good
i think i wont ever upgrade from this akg c414 xls
theres literally nothing else to upgrade to
maybe a neumann u87 but that's studio schizo shit
meh, u87 is overrated
I don't like the sound of it, tbh
I mean, it's a... FET mic? I think?
the Rode NT1 is a great "budget" option, though
wtf this mic sounds fucking god tier
I still don't see much of a point beyond just getting a dedicated setup. I say that as I opted for an SM58 and M2 for my setup. But that is pretty much all I need ever.
wayyyyy better than my sm58
sm58 is dynamic, yeah...
and only 100 bucks
so good quality but not the best
you need $200-400 range
it's multi pattern so yeah
TLM-102 is only cardoid
you know what mic I like
uh, I gotta find it...
I can at least tell better mics apart, I just don't have the energy to care beyond not static. I only got my setup just for consistency rather than audio quality.
Guys I need fuckin help like actually
ah, LCT 1040
Zero refuses to connect to a discord call, no matter what I do with her code. Its usually a connection error. Pycord 2.7
Should I rewrite the discord bot in C#?? Would that do any good
I don't know if the langugage even matters vs. reliability of extensions/imports
It just seems like a pycord issue
I once used discord.py and had no issues
tts in discord voice call worked fine
Similar with cameras, don't really care about lenses as long as I have a lens of the type. I don't hate phone cameras, I just choose not to expect much from them.
Joins for a while then disconnects with a connection failed.
But never says anything
Every hobby has a heavy curve to spending a lot, I just choose to value the few funds I have and accept mid-level everywhere instead of trying to max out on only one thing.
I think that can happen if the Version of pycord/PyNaCl/certifi is oudated/incompatible
Discord VC is like the one thing that genuinely is making me bash my head against a fucking wall because somehow I made a near agentic minecraft script but cant get her to join a vc
I have mulled over that already
I'm running 2.7
$3499
just the mic or everything
Have you enabled the required gateway intents for the bot ?
Don't think that's enough
Enable the gateway intents on the discord developer page for your bot and also enable them in your discord bot code
In dev portal right> Where is that at?
Yea meant those intents
Those are enabled bot side on dev portal, not seeing matching creds on discord side
Maybe in my script
i have no context for the actual issue i just came in here after 12 hours of vr and saw "needs gateway intents" and "where are those" so i dutifully looked them up and posted them rq
@floral hawk
Im seein
I get that I have to rewrite this entire thing in JS
Which
Not happy about but se la vie

why would you have to rewrite in js
also
another soul claimed by discord voice issues
I should really finish the thing I was writing
I'm seeing code online for JS gateway intents, but I found it for python
But yeah oh my fucking God
this isn't related to gateway intents if you're only getting the error when trying to join a vc
pycord kinda sucks for docs so I'm not surprised you're struggling
I'm not super surprised but this is fucking rediculous
yes it is
I've had to stub this 3 times
I've never had to stub anything else more than once
stub in which sense?

She is literally playing minecraft by herself and I can't get her to join a VC
tbf voice is one of the least well documented aspects of discord
anyways quizzing time, what do you need discord for?
i.e. is it just your bot talking, or also listening to people
For collabs with some streamer friends of mine
For both, I suppose. Talking and listening,
this example shows using their pre built sinks
that's such a stinky api
"time": "2026-05-22 04:24:53",
"type": "error",
"event": "DISCORD VOICE JOIN ERROR: RuntimeError("Task <Task pending name='pycord: on_message' coro=<Client._run_event() running at C:\\ZeroTwoAIProject\\.venv\\Lib\\site-packages\\discord\\client.py:465> cb=[set.discard()]> got Future <Task pending name='voice-connector:0x1b293b3cb60' coro=<VoiceConnectionState._wrap_connect() running at C:\\ZeroTwoAIProject\\.venv\\Lib\\site-packages\\discord\\voice\\state.py:480>> attached to a different loop")"
Logger throwing ass,
once again reminder why I don't like pycord 
I can't figure out what loop tho, it was a fresh launch
Voice loop isnt running
And that's her main runtime rn
Holup
One sec
I'm still fiddling with it
Imma send in a sec
Im hoping this fix works but lowk not optimistic
for my agent i tried vibecoding the discord integration and pycord just straight up didn’t work so we switched to discord js
what is voice_handler
because if it's something you diy'd I can probably guess what you did wrong
Tis for VC modulation int
It isnt the cause of the issue I don't think
Stuff like TTS, "Emotional" Modulation, and voice command injections
is it not directly responsible for joining the VC too?
because your command otherwise has nothing to join a vc
it would be the logical place to look imo
if I had to guess, you're manually creating an asyncio loop in there for some reason
join_voice isn't long 
Oh hey neurobot
also this - just stream pcm in whatever sr discord gives you, preferably pulling and pushing audio chunks in the library-consuming code directly - not over network
the same way you would listen to a mic
idk if this makes sense
I'm in a little bit over my head, tis not my forte
for getting the bot to speak, would it be useful to have "play this audio in its entirety" and "play this chunk, expecting more soon"
All signs are pointing to "rewrite in JS"
so you have a good reason to not send your voice manager file or are you just gonna lay down and keep complaining
because I am almost certain you're doing something dumb in there
I am too,
so send it

should probably just always have a queue/buffer and then have a flag whether to interrupt the current output and clear the queue on new chunks
probably both with the former for simplicity sake and the latter because streaming tts
Yeah she does use streaming with pocket tts
It does queue on its own, in theory
Not sure how the discord side interacts with that but in main runtime it works
🐇

ok, disregarding the sad reality that this situation reflects, all power to the maintainer
since my new ram hasnt died yet so far im blaming my old mobo
oooo you want to feed the new rmad ram to your old mobo sooo baaaad
I know, I'm trying to follow your advice
I'm just overwhelmed
Genuinely nothing about developing zero has been this taxing somehow
time to use xmpp
oh my god why does whisper suck so much
What
ye
To fix it u mean?
Wym better one
Wait what cpu do u have even
5950x
the mobo that broke was a msi x570 tomahawk wifi
my new mobo is a gigabyte x570 aorus ultra
i think x4 ye
the specs and shit are here https://www.gigabyte.com/Motherboard/X570-AORUS-ULTRA-rev-10
it has nice vrms
milba toast
im only using the primary pcie slot anyways. altho ive been thinking baout a usb thingy
it does to x2 on the final slot
so it'd be about half speed on a gen4 drive that maxes that interface
mine does the same thing unfortunately
cant run fullspeed gen3 either cuz it'll just negotiate x2 3.0
since it's electrically 2 lanes
if you have 2nd pcie slot open tho
expansion card
it might only support one in that slot when running x8 mode which it is if slot 2 is populate, idk if it supports x4x4 or x2x2x2x2 for an x8 slot
save the usb card for the chipset bound pcie lslots
well
my mobo has like 6 extra usb headers
i dont really need pcie, i just need to use the physical location underneath my gpu

Pink
urple
Hmm new headcannon: Pink stands for "Purple Ink"
The ram doesnt want to sync on my new mobo, probably some bios setting
Time to update my Openmediavault which hosts a nfs share that gets mounted by my host which then bindmounts to my vms, what could go wrong?
Dont you need special software for the rgb to be set?
i have said software
And it don't work? Huh
Also god damn that cooler fits on the multimeter exactly huh?
did you plug it into rgb header
my ram into the rgb header?
Oo, hows openrgb? Never used it.. tho odd that the mainboard would have a effect on thaz
its fine
-ish
The Gigabyte X570 Aorus Ultra is unique because Gigabyte implements strict SMBus/I2C blockades on this specific chipset generation to prevent users from corrupting their RAM profiles.


lemme try something else first
this wasnt necessary on my old mobo, but i should probably enable i2c to begin with in the configuration.nix
nah didnt work

ye shouldn't be necessary if OpenRGB is configured via the NixOS module
it already does that
yeah gigabyte does this
i cant read my modules properly either from OS
sudo i2cdetect -l
i2c-3 i2c AMDGPU DM i2c hw bus 1 I2C adapter
i2c-1 i2c AMDGPU SMU 1 I2C adapter
i2c-8 i2c AMDGPU DM aux hw bus 2 I2C adapter
i2c-6 i2c AMDGPU DM aux hw bus 0 I2C adapter
i2c-4 i2c AMDGPU DM i2c hw bus 2 I2C adapter
i2c-2 i2c AMDGPU DM i2c hw bus 0 I2C adapter
i2c-0 i2c AMDGPU SMU 0 I2C adapter
i2c-9 i2c Synopsys DesignWare I2C adapter I2C adapter
i2c-7 i2c AMDGPU DM aux hw bus 1 I2C adapter
i2c-5 i2c AMDGPU DM i2c hw bus 3 I2C adapter
none of this is ram i think
nasw the slopper is right
wtf is a synopsis?
that's the i2c controller
oooh
maybe the one for the onboard mobo rgb
you probably just need to use gigabyte's application
i know linux
but
like i can set these things from wandows and itll persist to linux
nah, none of that. im figuring out how to do it without dum gigabyte app
based tbh
needing to use the gigabyte app for performance modes on my laptop is the worst part of my laptop
and thats saying a lot since that laptop bluescreens every week
the one for my mobo (it's shit)
that said there is this option
so there must be another path

it takes 2 minutes to launch on my laptop, and its not even a slow laptop. it has a 12500h. shoudl be fast enough to open up a dum stoopid GUI

but i havent had that issue in my case
also
services.hardware.openrgb.motherboard must be set to "amd"
otherwise i2c-piix4 module not loaded
it might be cuz of the performance profiel or something. but even when its set to "low power usage" it shouldnt take that long.
i kinda need that gui to make it not use "low power usage"
i dont see how that would affect it that adversely
id probably reinstall it or windows entirely lol
i tried both already
well, windows reinstall was more of an attempt at stopping the bluescreens
updated every driver too
to whatever was the newest last august
maybe your motherboard on that thing is cracked as well
maybe
idk
the laptop works fine most of the time
only issues is bluescreen form some type of invalid ram acces, and GCC being shit
convincing pitch norton
but i will still neglect to opt in to your malware installer
ok im going into the bios 
o7
Refind
Nixos
New windows on the 2nd m.2 for school adobe stuff
Old windows on the external ssd
Random windows from a used m.2 drive i bought

Chayleaf is famously greek
The english is whatever. A lot of countries ive been in have random english everywhere so i dont even notice it being out-of-place
wdym interleave thr sockets? There is only 1 socket 
cpus can have multiple nodes
but yeah that's worded for a 2p system idk why it says it like that
also i dont think they really have multiple on one cpu unless it's like
manycore
epyc
etc
threadrippers, i know win10 will split froge 3990x into multiple nodes, win11 wont
setting leftover from bios being reussed
yeah but those wont usually be more than 1
yea
it's based on core count and stuff as well
it's just
a group of cores
that all can access some resources
apparently some motherboards do have a setting to expose the CCXs as NUMA nodes
quicklly

you can
because
of x3d
it's advantageous sometimes on x3d
to split it into multiple
I see
so you can force things onto the x3d ccd
I doubt id want that tho
Unless one ccx is god bin
even then i doubt
you still have L3 cache locally to each CCD even without 3D v-cache
my 9900x3d i had "X3D TURBO" turned on in bios and it completely fucking disabled the non x3d ccd so i had half the core/threads
just X3D CPUs have a lot more on one of them
right there' juist no reason to really do anything to that via numa
usually
unless you have a fuckload of cores
like 64+
ye I agree 
not make sense with only 2 CCDs probably
Where the hell do these guys put their "SPD Write" block thingy in this bios???? Its a fucking maze
lopsided cache setup on x3d makes it kinda worth if the os scheduling is dogshit
did you already try the kernel parameter and stuff
use the search
there's a search in my last couple uefi mobo
including a gigabyte
There are THREE places you can enable PBO. But i cant find what i need
There is no search
SPD write is not available on AMD X570 platforms because the SMBUS controller does not support blocking certain address ranges, meaning it will be missing from the BIOS.
that's what i'm finding

like i said i ran into the same thing
My old mobo could do it tho
was it x570 and also gigabyte
that'd probably be it then
it's "Generally" not available
probably isnt a hard block
might just be funky
luv me gigabyte but they make strange choices in bios at times
always have
Only sdp thing i could find
yup
it sucks cuz you cant even read the actual DIMM serial numbers on these boards with that blocked
you have to take the fucking modules OUT to read the serial on the module
that one t says there isnt the actual one
xd
that sounds like my boi

@sage crag quick what is the equation for parabola with known height and width?

Is sine considered parabolic? 
you expect too much from my smooth brain
i odnt think so???
Nono, as in, the first half of sine
maybe if you limit the domain
local to one curve sure
Hmmm
equation i cant fucking remember it's like 4a(x)= something
y=x^2?
so i can remember this shit and not look like a moron

the normal one is, yeah
but there are general equations
with multiple parameters

bro quack was going through fucking lorentz equations the other day and my brain actually deep fried it felt like
This is what I need 
lorentz is magnetic force right?
he wa slike so this goes to infinity pointing at velocity and asking what that meant
and i was like
le finger

its so funny watching everyone in the exam hall getting their left and right hands out to do these stupid rules
naur special relativity lorentz factor & transofrmations
No, lorentz is a person 
no
In all seriousness, lorentz is THE force name
It is caused by electric and magnetic field
was
Hmmm
AI Overview Indonesia Yes, thousands of people named Lorentz (both as a first name and a surname) are still alive today.
You're wrong 
not the indonesian ai overview
"Indonesian" 
i know one indonesian for sure and theyre very smort
not for long
say goodbye to everyone named lorentz in indonesia

holy shit he's gonna activate goblin mode
you were gpt 5.5 all along

im onto you, son of altman
https://www.youtube.com/watch?v=r9zRKOSvjK4 @nocturne olive
But his dimwitted dumbot does something dangerous. Damn.
robotnik@pingas
robotnik looks so much more evil here damn
im used to him being round head
Get Surfshark VPN at https://surfshark.deals/APOLOGIST - enter promo code APOLOGIST for 83% off and 3 extra months for free!
The greatest villain of Sonic the Hedgehog has come in many shapes and sizes...and names. Today in "What's in a Name" we look at almost all the varieties of the good doctor and talk about what may be the greatest compromi...


SamVanAltman 
snooPINGAS usual i see
derive it
i just finished feeling sorry for myself due to puking
i wont help

i wonder how weird re:zero would be to watch if you skip the times he died
like form the perspective of the people there
the canon timeline i guess
quadratic in the form ax² - (b + c)x + bc = 0
width = c - b
height is solveable by subtracting h from the left hand side
you can parameterise b and c in terms of a, w, and h
go solve it
Yeah, but the roots are known
The width forms the root no?
not /x
i was going to make it so everything was divided by a
but i just multiplied by a instead

I was wrong, the width does not always forms the root 
ye
it generalises to any degree polynomial, for some roots b, c, d, e etc
axⁿ + (b + c + ...)x^(n-1) + (bc + cd + bd + ...)x^(n-2) + ... + bcd... = 0
the coefficients of powers of x less than n are referred to as the sums of the first, second, third etc differences
o also i forgot a - sign
it alternates, -, +, -, +
for each coefficient

orm
a isnt a
root
nth degree polynomial
b, c, d, e, ... are the roots
a is the scale factor for the whole equation
it doesnt need to exist but its good to add it
im surprised you didnt know about this actually

i know that i just dont know what the equation is 
nth degree polynomial
why online learning materials teaching it this way
🐇

the
thingies im talking about
are referred to as vieta's results
sometimes vieta's formulas 
sounds less cool


french
ye this is the first part of the pure maths curriculum for further maths A level
very simple conceptually but easy to make mistakes with
they are both named Viète apparently 
actually
same person

oh yeah i remember this
actually complex numbers should come first

https://alevelmathsrevision.com/pdf-viewer/?file=/downloads/questions-by-category/further-maths/pure-core/y1-pure-roots-of-polynomials-exam-questions.pdf&qms=true
some medium difficulty sample questions
Improve Your A-Level Maths With A 30-Day Money Back Guarantee
homework
GitHub is silly 3.5Flash cost less than 3.1Pro by Google themselves
https://fxtwitter.com/FitzGPT/status/2057418463824482632
Soygun.
Quoting GitHub (@github)
︀
📣 @GoogleAI’s Gemini 3.5 Flash is now generally available and rolling out in GitHub Copilot.
︀︀
︀︀Early testing shows
︀︀➡️ It has strong tool use, fast response times, and high cache efficiency
︀︀➡️ It is it well-suited for fast, iterative agentic coding workflows
︀︀
︀︀Try it out in @code. ⬇️
︀︀github.blog/changelog/2026-05-19-gemini-3-5-flash-is-generally-available-for-github-copilot/
Tri-OS code is a bit very annoying
Claude 5 is going to be 100x, can't wait 
it is friday 
Now do some cursed stuff 
Lmao even vim does AI now
AI haters will have to go to emacs
https://www.phoronix.com/news/GTK4-For-Vim
That's gvim tho
no
i have exams
Gvim is vim
On the weekend? Surely not right? 

Add GTK4 GUI backend using separate source files instead of adding conditional branches to the existing GTK2/GTK3 code. GTK4's API changes are too fundamental for #if branching to be mainta...
have you not studied a day in your life?
Nope 
I am making
SPA thing
with jQuery
and HTML templates
me neither, but im hoping to change that
Gonna get forked hard
It's not as cursed tho. jQuery is still useful even if we have ES6 
No THIS is cursed 
and get abandoned within 2 weeks 
Depends
People are stubborn
Remember my tips to cheat during exam 
If it ends up like bun then it would 100% happen
it currently has a server that renders templates based on request path, but me not want that
there are only 3 actual theory exams, the other ones are handing in a project
so need client-side routing
people are also lazy and don't want to maintain things themselves
Lucky 
with exceptions of course
Hash based routing? 
Please don't 
Seek help instead 
no
i wish i had less projects
because
im dying
Hi dying 
hi starting to think konii is sending password in partial format 
My tips for uni projects are to not go above and beyond 

my tips for uni projects is to work on them, cuz if you wait till the last week you're severly fucked

I always did them on the last day
Correct. But also make the mvp and you can then safely submit those 
yep, hence why i have at least 4 retake exams now
No need to add features that are not asked or to be fancy 
2 from last semester, and 2 things im already 100% sure are fucked beyond repair
Never had that issue
I was still on the deans list
the problem is my projects are stuff you get 2 months for cuz its impossible to do it in less than a week
I like it when my cat kneads me. But man why is cat claws as sharp as a razor 
Slendercat
Has no face
🧀
Crème de Gruyère
yes
oh boy
oh boy?
with a sunny disposition 
we killed the schizoer again
Why are you wearing socks indoor?
we should keep track of em or smth
maybe we can make a graveyard in the schizo thread
Maybe because I live in the tropic so it's weird for me to see people wearing socks indoor
????
why would i be barefoot?
thats disgusting

im barefoot rn
thats disgusting
I think it’s just dependent on how much of a neat freak you are with your floors
You guys are not like the american that wears shoes indoor right?
most of the world does that im pretty sure
I don't think so 
apparently not
🍋
🍋

I don’t think Americans do that
Source: me
shoes inside = explode
Drink lemon, not eat 
eat
Lemon drink, raw not good 
why
Lemon cooked, good 
oben
idk where these maps get their sources from, i see at least 5 countries ive been to that are labeled wrong
micorvewa plasma
wrong map bad map
No, grill 
shoes off inside
Or, caramel lemon 
Lemon candy 
not a single one of the maps i find label uk as shoes off. and when i was there i didnt have to do shoes off myself
so idk
Lemon cooked, good. No oven 
i refuse to believe not every asian country has shoes off
emotional damage
bro my mom would nuke me if i wore shoes indoor
i see
like until the shoe thingie its fine
thats because as a guest you dont have to remove your shoes
where u put ur shoes
but people will dislike you
Shoe rack 

but why?
ye forgor name
They are british Sam. What are you expecting. Look at Vedal 
wdym look at vedal?
cant relate
they're weak
cull the weak
eat the weak
You just culled lorentz population in Indonesia 
good
'cause there might be rabbits nearby

arent both options saying you’re afraid of rabbits
A spider is also food
How did bun rewrite in rust went
Plenty people don't like spiders
mmm never ate one before
Bun is dead and the company went banckrupt
there's a dutch song about a boy and his pet rabbit, and the rabbit goes missing and then its served for dinner
it did not go yet
wow
not yet released
English moment
The problem is there is no notice or any info if the branch is just silly experiment or serious one
If the maintainer spent some times reviewing and vibe code properly, it would not went that bad
oh it's already merged
just not released
It was probably made by superbox
I say that but we actually have horse meat here
@prime spade
here too

horse meat
Someone sent that here on Christmas I think
Like rewrite bun in rust, does it have unit-test? Are there verifications that the code actually works?
isn't horse meat super tough, why would you want to eat that
it passes the entire test suite that they already had
Idk man, people are weird here
they like tough ones
















