programming

[silent cloak]

Its hard to get a full picture when ur stuck in the frame

Thats the issue with temporal theories

[frail mortar]

What Julian Barbour says

[silent cloak]

Thats also what I think could be behind virtual particles appearing and disappearing

Just particles moving into our frame

[kind nimbus]

my system is stateless since i use impermanence mode. My rootfs is on tmpfs / in RAM.

the nix-store is read-only anyway.

i do automatic snapshots via zfs every hour, day, month.

i also use home-manager so all my configs are 100% reproducible and read-only.

my entire config is on github and everything is tracked via Git.

i could re-setup my entire OS and setup within an hour with a livecd and boostrap everything directly from GitHub.

[tender river]

it not stateless since you have files in /home and /var

[idle dune]

I would like to think my Ubuntu installation is secure or safe
UFW, ClamAV, LUKS+Encrytion, and Mullvad

[kind nimbus]

i just use full disk encryption via ZFS

[frail mortar]

I'd imagine a grid. 4d geometry of gravity and mass.

[idle dune]

I didn't turn ZFS on becuase it had (experimental) next to it and I need a stable machine

[silent cloak]

Thats what ive had in mind yeah

[kind nimbus]

yeah those are the only non-stateless bits but everything else is 😛

[tender river]

"only" its literally only thing that matter for most people

[silent cloak]

I might be biased though because Ive always found the dimensional / phase approach the most fascinating

[kind nimbus]

? ZFS is super stable lol

i am running the unstable zfs version for years now so i can use the latest kernels and i never had a single issue

my servers and NAS also run ZFS

it's one of the most if not the most reliable filesystem

[silent cloak]

But its like the dream theory of the physics world because its hard to test

[idle dune]

Idk I prompted Claude with "I'm installing Ubuntu and I am new and want a secure system." andit gave me 24 commands to run. I've stopped search forums for the answer and just ask Claude how to fix my Linux and 99% of the time it works .

[frail mortar]

well

[idle dune]

Oh, well when I get my 1tb ssd to upgrade in the future I will do ZFS

[frail mortar]

if the math was there, it would be more provable.

Good example, rewrite Schrodinger with the theory of timeless universe.

[idle dune]

I've started jokinly referring to using Claude to fix my linux install as "Vibe Sys Admin" and it makes my IT freinds mad.

[kind nimbus]

good choice

[silent cloak]

With quantum computing down the line we may get more answers but for now its hard to peak outside our bubble of space and time

Its cool running extra dimensional simulations of 4D spaces though even if they are just approximations with 3D software

[frail mortar]

we dont need quantum computing to make a mathematical framework to tie GR and QM together.

[silent cloak]

Non-euclidean geometry is what got me interested in the first place

[frail mortar]

ah

whats your opinions on that?

[tender river]

nix allow you to overengineer config
easier solution -> not make complex config in first place just as reproducible

[frail mortar]

Remember you need sleep

[silent cloak]

I mean I find it fascinating and while bending space can be possible its not practical with how much energy it would take

[frail mortar]

gravity causing? Or something else?

Cause you have gravitational lensing.

[kind nimbus]

i feel like my config is already super complex but it works perfectly for me so i stick with it

[frail mortar]

and time bending

[idle dune]

IDK when I am gonna get that 1tb though, its looking like its further and furth in the future.

[tender river]

how many line of code

[silent cloak]

The Alcubierre drive is a good page on some of the practical downsides these practices could have

[kind nimbus]

of nix code ?

[frail mortar]

I must retrieve my wife. She's ready to get picked up

[kind nimbus]

need to count

[frail mortar]

Ttyl

May I dm you my work cipher?

[silent cloak]

Yeah

[idle dune]

All we need is Element 115 to be reproducable like Bob Lazar says and well have ftl tech.

https://tenor.com/view/schizo-cat-the-voices-gif-25415230

[kind nimbus]

18854 total (all nix files)

[silent cloak]

Until the ship is back in real space and destroys everything in front of it by accelerating all the particles caught in the field

[tender river]

me 21789 non-whitespace non-comment

[kind nimbus]

damn

[tender river]

22225 actually forgot one file

not good number

[kind nimbus]

ok without whitespace or comments i am looking at  15091 lines of code according to claude lol

[tender river]

vibe coder

[kind nimbus]

vibe coding is so good these days it would be a waste not to use it tbh

[tender river]

vibe coding make boring code me want make fun program not achieve goal with mass produced code

[idle dune]

https://tenor.com/view/vibe-coding-kai-lentit-gif-8970542015069914600

I joke but I would also vibe code

[jagged turtle]

ad break my ass

[kind nimbus]

the thing with vibe coding compared to manual coding only is it saves me so much time it's insane

[rigid snow]

save now pay later

[kind nimbus]

i can now do stuff within 1 month i could otherwise do within lets say 6-7 months

[idle dune]

Yeah but it can introduce alot of random ass bugs cant it? Like not immdedlty but further down the line.

[rigid snow]

that’s not the issue the issue is the code is shit

[jagged turtle]

buy now pay later moment

[tender river]

ai code is just bad me try with rust result horrible c/go/python-like mess

[kind nimbus]

it can also find bugs for you

but even if you code everything by hand you can also introduce bugs just as well

[silent cloak]

Most repos will deny any bug reports u make using AI lmao

[tender river]

good

[kind nimbus]

interesting. somehow the results i am getting from Codex with gpt-5 high are extremely good.

[rigid snow]

easier to dismiss immediately than to try to figure out if it’s even real

[silent cloak]

They usually arent

[rigid snow]

ye which is why i said what i said

[silent cloak]

They act like they found a CVE and its just a nothing burger

[tender river]

sound like task just equivalent of boilerplate

[silent cloak]

Literally

[idle dune]

I was going to vibe code a doom wad.

[tender river]

anyway you wont even know what good code looks like if you rely on vibe coding

[silent cloak]

AI often doesn't follow K.I.S.S as a practice and does the opposite

Makes it as over complicated as possible

[tender river]

me go feel vindicated

[umbral thorn]

shh wait till he get vulnerability issue

[kind nimbus]

lol

i know how to exploit vulnerabilities

[tender river]

ohh are you also a script kiddie

[kind nimbus]

no

[tender river]

me mean

[kind nimbus]

i can also write my own exploit POCs from scratch

i am not a programmer so the stuff i code is not being used in production anyway

[olive sable]

im not allowed to agree and continue

geuss i wont be trying out ff14

this is prob cuz it couldnt install edge webview 2 and is using something else now which it was complaining baout

[umbral thorn]

checks out

[quick condor]

I will never use gpt for anything other than review and planning after what it attempted to do in one project (I always at review the code before I apply the changes), it attempted to inject one of its own guard rails into a prompt for a local model that:

1. Was not asked for.
2. Would have completely changed the output of the model.
3. Was completely unnecessary for the request it was given.

[kind nimbus]

That's wild. Never seen anything like that happen yet so far.

[quick condor]

It was a literary analysis prompt that was being used to check for only grammatical and punctuation mistakes. It injected content moderation into the prompt. The worst part is that this was a project that was processing training data

[kind nimbus]

Damn

Well at least Claude shows you any changes and asks for confirmation or different instructions before applying anything to your code

But yeah ChatGPT absolutely loves to censor stuff or add guard rails lol

[sick owl]

https://fixupx.com/claudeai/status/2001748044434543082

Claude (@claudeai)

Claude in Chrome is now available to all paid plans.
︀︀
︀︀We’ve also shipped an integration with Claude Code.

**💬 171 🔁 415 ❤️ 5.3K 👁️ 751.6K **

▶ Play video

FixupX · Dec 18, 2025, 8:14 PM

[true hemlock]

vibe coding is shit

objectively shit

[sick owl]

Anthropic doing the sane thing and just making an extension instead of trying to fork chromium

[true hemlock]

last time i tried vibe coding i had to spent 8 fucking hours debugging and arguing with the clanker

restarted the whole codebase progress and i only need 1-2 hours to finish what i wanted to do

[quick condor]

Have claude max, got it after codex tried to gaslight me.

In the same project:
Codex keep trying to add a completely unnecessary "respond in strict JSON" segment to another prompt. The model which was not trained on code but English prose kept failing to respond correctly. It blamed the model performing the request not the completely unnecessary prompt. I say completely unnecessary because the api response was wrapped in json automatically.

In another situation it failed to correctly form the Api call. It kept insisting that the model that the call was going to was the problem and would "prove it" by adding a debug line, restarting the process (which killed the model) and immediately testing the code again (which returned that the api call failed to be responded to)

[kind nimbus]

Somehow my experience has been quite the opposite lol

[true hemlock]

i don't know, people that vibe coded and "made it work" never seem to work on relatively complex codebase

[quick condor]

The important thing is that claude doesn't have reasoning shortcuts and has safeguards to prevent editing without reading

[true hemlock]

that or good luck dealing with security and logic holes

[kind nimbus]

Which Codex model was that?

I usually only use gpt5-max

[sick owl]

Just avoid claude code, its borked atm due to context compression

Unless they've patched it already

[quick condor]

This was 5.0/5.1

[sick owl]

This is true, generally its low skill hobbyists creating low scale stuff for personal use that it works well for

I don't know of any businesses "vibe coding" entire commercial codebases successfully

[kind nimbus]

I already used Claude and codex on bigger codebases and the results were good

Gemini-3-pro on the other hand completely screwed up despite being hyped so much lately

[true hemlock]

what kind of codebase and how much you're making it work

[kind nimbus]

Yeah that one is annoying

[quick condor]

Works fine for me on a project with currently 45k lines of custom code+ a significant dependancy list and cross system calls

[sick owl]

It only kicks in at about 75% context

[kind nimbus]

Claude does run out of context much faster compared to Codex unfortunately but it's not bad tho.

[sick owl]

I do encourage people to build stuff themselves rather than letting the AI take the reigns generally

[true hemlock]

because vibe coding might work if your codebase is those backend with simple defined functions, which are what most backend actually are

[sick owl]

Imo its best as a second pair of eyes on code you wrote and understand to catch oversights

[quick condor]

But then again I have an extensive project documentation Google doc and I plan out a new segment in claude web, make adjustments until satisfied, then create an implementation document for claude code to follow. Context compression doesn't matter because it's just moving step by step through the doc

[sick owl]

One big advantage of Gemini is that if you are working with a large codebase its context handling is just generationally better than any other model out there

[jagged turtle]

how do you forward your ssh/gpg keys from your host machine to a dev container

[sick owl]

So I will say even if it might not necessarily be up to snuff in tiny vibe coding projects compared to the competition as an assistant and autocomplete in more complicated scenarios where you're the one doing the coding it is far and away the superior option

[true hemlock]

don't think people that mostly vibe code has a project complex enough that is beyond the scope of what the chatbots can do though

[kind nimbus]

Somehow i was reaching limits super quickly even with a Gemini 3 Pro sub that it was almost unusable for me.

[opaque sigil]

agent forwarding

[jagged turtle]

yea, and how do i do that?

[quick condor]

Sounds like you aren't doing any of the coding and letting it figure out what to do on the fly. Sounds similar to what happened to me when I first moved from gpt to claude

[true hemlock]

that

[opaque sigil]

{
  "remoteEnv": {
    "SSH_AUTH_SOCK": "/ssh-agent"
  },
  "mounts": [
    "source=${localEnv:SSH_AUTH_SOCK},target=/ssh-agent,type=bind"
  ]
}

something like this should work but who knows, not sure if this works for sockets but i don't see why it wouldn't

[true hemlock]

also for the fact that most people that are comfortable with vibe coding are the one that never coded anything themself beforehand

same rarely can be said otherwise

[jagged turtle]

the ssh stuff is fine now, I just needed to like, reload window

because of course I needed to

gpg seems to be a bit harder tho

[opaque sigil]

idk i don't hate myself enough to use gpg

wouldn't know

[jagged turtle]

idk why I am still using gpg to sign my commits even though I have an ssh key to sign with

[quick condor]

It can write pretty passible code if you give it well defined instructions.
To put it in perspective, my last change was a 300 line document not a in ide prompt

[kind nimbus]

Before I started to use AI for coding I wrote everything by hand but now that we have all those tools at our disposal I don't really see the point anymore in writing everything by hand.

[sick owl]

That doesn't sound at all right, it has a native context window of a million tokens

And drops off from roughly perfect performance past 256k

[amber fractal]

I admit giving vibecoding a shot is something I regretted, it isn't something that can replace leaning what your doing. That's why I'm not continuing with it.

[sick owl]

Which is double most other models

[kind nimbus]

I have a feeling that it was stuck on reading cache files instead of code or something

[sick owl]

That checks out, sounds like it could be more of an issue with the environment you were using it in than the model itself

[true hemlock]

it fucked up my whole codebase and made me crash out for 8 hours straight

[sick owl]

Which admittedly isn't as strong as Claude and 5.2 at coding but still shouldn't suffer to that extent

[kind nimbus]

Maybe. I was just using the Gemini cli. But my guess could be that it got stuck on reading pycache folders or stuff like that.

[jagged turtle]

oh shit ok nvm it works automatically now

[sick owl]

I stopped Vibe coding because I could feel myself getting dumber

Maybe it was just nocebo but I felt like my memory was getting appreciably worse

[true hemlock]

"I don't really see the point anymore in writing everything by hand" you just explained why windows 11 is plagued with so many bugs

[quick condor]

To be fair I do have actual previous coding experience so I probably don't use it like most. I give example code snipits that generally are what I want modified for the code base, explicit instructions, inspect both edits and thinking, and document everything

[kind nimbus]

Haha I mean the problem is even if you vibe code you should still understand your codebase fully.

I guess Microsoft engineers aren't checking the code they write with AI anymore.

[kind nimbus]

Yeah I think it mostly narrows down how exactly you use these tools

[amber fractal]

In my own case it put another nail on my ego, enough to stop doing things that were stupid for my skill level. Which is why I'm not really investing into AI at the moment, I'm just focusing on my own lane.

[quick condor]

That shouldn't be something that you have to explain but we've reached a point where it is

[kind nimbus]

On the other hand more people are switching to Linux now due to all the issues so that's a good thing lol

[quick condor]

I mean I'm pretty agnostic, I use windows where it makes sense and Linux where it does better. WSL2 has made that significantly easier

[kind nimbus]

Sure you should always use the best tool for the task

But I personally only use Linux anymore these days

For Pentesting I need Linux anyway

[rigid snow]

i only need paper for pentesting

[kind nimbus]

Lol

[amber fractal]

Alright, now this is what this was for, I guess it was worth it? You learn something new every day I guess

print(np.zeros((1,), dtype=DXT.u128x4))
````[((0, 0), (0, 0), (0, 0), (0, 0))]`

Now with this insanity out of the way, cool thing to learn. I'll be putting it to good use

[valid heron]

I’m making a program where neuro can create her own original music through either FL studio or Ableton live, it’s starting to come together a tiny bit but it’s also driving me insane

[quiet gust]

just suno it, it's great

[valid heron]

Defeats the point of what I’m trying to make

[quiet gust]

I have some projects I still fully understand where I've vibe coded, but several projects recently I don't attempt to understand at all, and ONLY vibecode, as an exercise. I don't think it's essential to understand anymore, but it's essential to keep the security portions in a place you can look it over and verify

[valid heron]

Man we are cooked 😭

[kind nimbus]

Makes sense. Never sacrifice security.

[hearty notch]

what projects did you not try to fully understand and what level of satisfaction did you reach

[quiet gust]

there's been some bugs with masky.ai (some tech similar to what vedal is up to), most of it is vibe coded and it's pretty satisfying - I have a chat module that uses on device llm to identify most important chat messages, gonna gamify chat on twitch a bit, so the chat overlay doesn't show 'spam' or non essential classified messages, and can also show only the 'most important message'... a lot of the logic is just in english and I haven't even looked at the code at all

when theres bugs and your prompting isn't resolving, and you look at the code sometimes you're a bit horrified and prompt it to fix/clean it - I've been still mostly hands off. and other people working on the project ONLY vibe code and I make them put up PRs and review those with my brain lol

the satisfaction with a browser plugin I made to add a weighting to reddit votes was pretty high

[hearty notch]

nice

masky.ai is your solo project?

you should make this burger menu spawn more to the left (this is helium which is chromium)

[kind nimbus]

When I visit masky.ai on my phone via chrome the page starts flickering like crazy. Could be dangerous for people with epilepsy.

Edit: doesn't happen on Firefox tho.

[quick condor]

I think it's a stretch to cause a web page physically dangerous dangerous. There is a difference between a phone screen and a movie/video game that you are zoned in on

[kind nimbus]

No clue what's causing that but it feels like a strobe light

[quick condor]

There is a car park near me that has foot wide silver horizontal bars that catch the sun and contrast the dark interior. When driving by it has a strobe/static effect that is distracting. Still wouldn't say it's dangerous

[quiet gust]

ty for feedback, will look into it. when logged in its better i guess, but atm we haven't optimized for mobile at all either

the main page is like a user page, here's a streamer who uses it https://masky.ai/azizana

[quiet gust]

how did u get permission to use images? can't do that myself lol

[olive sable]

you get embed perms after a while by sending messages

[hearty notch]

i was on browser (macos helium)

[kind nimbus]

I just recorded this weird flicker bug but I cannot share the video here

[quiet gust]

if it's mobile don't worry about it, we'll get around to that eventually

[kind nimbus]

Yeah I was browsing the site via chrome on Android

[quiet gust]

are u guys coders too? what kinda projects?

[olive sable]

i think im technical artist or whatever. but i do games and such

[rigid snow]

wdym “either or”
also it’s like giving a child with a blindfold on a blank canvas with 3000 brushes and paints. surely the result can’t be coherent

surely you have to set crazy constraints for it to be anything resembling music

with such powerful software

[hearty notch]

also whyd you pick those two when reaper has a full headless mode and more developed programmatic api and is either free or 60 bucks smh

[rigid snow]

true… though i personally feel like ableton is the best choice

(out of these three)

[hearty notch]

i bought ableton 9 forever ago and never got good at it but ive seen good things done with it

i just believe they all scale to rly good once youre used to them tho

and that parts kind of a matter of pref

[rigid snow]

the best one would be the most constrained one

[hearty notch]

oh you meant for the purposes of neuro

[rigid snow]

yep

[hearty notch]

i mean you can constrain any of htem based on how you design the api

[rigid snow]

true but what’s the point

bringing a nuke to a knife fight

[quick condor]

Realistically the more constrained the less is taught vs trained

There is a difference

[hearty notch]

so youre arguing for freer expression?

i mean i see the philosophical argument

[rigid snow]

there’s nothing learned or taught or trained it’s an llm calling functions

[hearty notch]

over getting to coherent faster

tbh what if you make an api for neuro to simultaneously teach herself the vocab of whatever DAW while using it

like it can fetch definitions

so you see her learn live

[rigid snow]

there isn’t really daw specific vocab? it’s the same across all of them

[hearty notch]

well i mean simultaneously the vocab and the api shes expected to use for it

so its linked

but yea i dont mean necessarily daw specific

and she'll probably already know

[rigid snow]

or do you assume she just has no clue what a “track” is

[quick condor]

By learned I mean iterativly, organically produced data. Trained is synthetic data from an indeterminate source. At some point how is it any different than teaching her to hit play on Spotify

[hearty notch]

well its more like some kind of api to feed her some kidn of semi defined workflow that gives her space for creativity but also doesnt have her struggling to be that coherent

idk its a hard balance of how much guardrail to force

[quick condor]

Guardrails damage actual training

[hearty notch]

its not necessarily a well defined term here there are different architecures you could call 'guardrails'

like just fuzzy acceptance of toolcall is a guardrail thats invisible fro mthe llm's pov

[quick condor]

Anything other than positive negative feedback. If you block or force an action, before inference
that's a guardrail

[rigid snow]

do you actually know what neuro does and how she functions and what’s being talked about
if that’s ever made the only time she’s ever interacting with it is by poking at it for an hour on stream, maybe twice or thrice ever . if there’s infinite directions she can take she’s not learning shit in this time. also horrible for content

[quick condor]

I honestly don't know much about neuro, my background is mostly llm training

Which is what she is

[hearty notch]

to me, feedback systems can easily fall under the semantic umbrella of guardrail, so we just have different definitions

[rigid snow]

neuro is an agent and a neuro game integration is just an mcp server

[hearty notch]

also like i said, fuzzy acceptance; some level of automated inference of what she meant to do, as long as youre trying in good faith to execute what she meant to do

[quick condor]

A feedback system that gives actual feedback is different, it's providing training data

[rigid snow]

if this analogy makes it more clear

[jagged turtle]

I smell neuro api talk

[amber fractal]

-# dinkdonk reminder, there is a diffrence in if neur is being trained on stream or not between chatters

[rigid snow]

not on integrations

[quick condor]

No because it's categorically incorrect. Neuro is the model + memory DB + misc modules. The agent is the interface

[true hemlock]

$400 for each 3090, yippie

[hearty notch]

models have agentic modes

wdym the agent is the interface

[rigid snow]

model + memory + modules is what i mean by agent

[jagged turtle]

glhf I don't think I will complete the tidalcycles container image anytime soon

[quick condor]

OK then same page

[jagged turtle]

I have a newfound hatred for audio backends

[hearty notch]

to me 'agentic llm' just means an llm model thats designed to call tools that do things

or not designed to but has the capability at least

[true hemlock]

which only possible if the API itself has that function
if not, the LLM is just outputting non text tokens

so its technically an interface

[quick condor]

Right but a model selecting a song on Spotify != making music correct?

[hearty notch]

yea but it probabilistically determining what things to play with in ableton

feels more like making music

i dont get your analogy

[rigid snow]

my point is: no weights are being updated as far as we know. even the post train. like for a year plus at this point
so the only learning she could possibly do is to poke around, make conclusions and save them to memory

[jagged turtle]

neuro obsidian integration that acts as extra memory storage

[hearty notch]

like neuro can simulate the process of being an edm producer and talk it out and this is experientially distinct from neuro just picking a random song from spotify

[rigid snow]

wait damn just give her splice

that’s the perfect constraint

[hearty notch]

ok ill do it

thx for the idea

[quick condor]

Even if it wasn't what happened, Vedal's reaction a couple months ago was that of someone panicking over model collapse not a bug. That only happens with training

[jagged turtle]

ehhh

wasn't it something about inference actually?

[rigid snow]

well still need to integrate into a daw to timewarp samples
ableton has splice integration now, suuuuurely they expose that over any of the 3 quintillion apis they have

[quick condor]

What it was doesn't matter, I'm talking about the method of troubleshooting

[hearty notch]

the thing is i already built a thing around reaper so i might just do it in raeper

[jagged turtle]

im a bit confused now

[rigid snow]

ye that’s cool, i just don’t know how you’d make it make splice loops work. like you need to filter for the correct key and approximately the correct bpm range, then warp it correctly

[rigid snow]

i thought it was an inference bug based on something someone said then
i’m not crazy im not the only one right?

[quick condor]

There is a pretty specific way to tell if a model is acting up because of over fitting or context poisoning. And that's exactly what it looked like he was trying to determine. If he knew it wasn't over fitting the troubleshooting would have been different

[rigid snow]

oh i didn’t see yeah i’m not

[quick condor]

The key isn't what it was (inference/context bug) it's that it was not immediately treated as that being the cause

[hearty notch]

i have no theories as to the location tomato issue it feels like vedal saw a similar thing happen once before that and he had no answer so he just tried to get evil to say things without the pattern a few times to dilute it in context

[rigid snow]

yeah i get what you mean

[hearty notch]

wait i just posted a theory after saying i have no theories

[quick condor]

Which is how you determine which it is. That or modify context manually

[hearty notch]

oops

yeah i think modifying context manually had to be the answer idk

[quick condor]

It went from an initial paniced reaction (either fixable inference or, impossible to fix without checkpoints, overfitting) there is a point where you can tell that he ruled out the latter but the point isn't what the problem was:
If no training has occurred in over a year, model collapse wouldn't be a concern

Hopefully that makes sense

[rigid snow]

now that i think about it there were a couple experimental streams this year, though long before the location thing

but we have no idea as to what the actual upgrades were

maybe the post training, maybe memory/other auxiliary systems

[quick condor]

My guess is training does occur but with a very low learning rate. Damage is hard but not impossible to do like that but the kicker is: you may not see it until it's too late

[rigid snow]

why is that the conclusion people make this puzzles me so much

their behavior doesn’t shift over time

it’s very incremental

[quick condor]

The thing is that full model backups aren't small, even a 24b model is 40gb. The smaller the changes, the less likely you are to make backups

[rigid snow]

post training is low rank

you can just store that

and surely you’re not suggesting he’s pretraining them

[quick condor]

If you know which lora damaged it

[sage crag]

[quick condor]

But the problem is that it could be as simple as a single spike. Take this example, a single grad norm spike is unlikely to damage a model that late in training but it can. Single late training spikes are common but as long as it is at a boundary and late it is probably fine. Earlier spikes can do more damage and be less obvious. The question becomes when did it happen

In this example it was a boundary spike and being only 1 and isolated, it probably didn't do anything bad but the reality is you don't know

[olive sable]

[hearty notch]

[quick condor]

The real fun part is that with a model like evil it becomes even more difficult because it is entirely possible for her to fake a model collapse as a joke without being able to understand it isn't funny whatsoever from Vedal's perspective. Yet at the same time if you were to completely guardrail that behavior out then you are directly altering behavior.

If I had to guess that's why the weaker filters are used instead, they can be bypassed and require constant updates but they can't damage anything

[sage crag]

[long barn]

how program,

in 2 words

I need tutorial

but i need it short, so only 2 words

[amber fractal]

[olive sable]

type keywords

type is a double entendre

[long barn]

idk how coding even works

I was just typing in words that ppl told me to but I still know nothin

is that coding

[quick condor]

Stack overflow

[olive sable]

you gotta like, know what the words do

[quick condor]

I can't read but I must code

[jagged turtle]

write words

you can't really explain much else

[opaque sigil]

write characters

[rigid snow]

press button

[quick condor]

You don't need words

[jagged turtle]

yeah just write the binaries directly

[amber fractal]

types

[olive sable]

[covert bane]

what have my ears just been blessed to? I can't believe this was from an AI who was trained with only 4 minutes of audio.
Though it just seems like random noise. Wdyt

[olive sable]

sorry but i aint lsitening to all that

[covert bane]

its ok

I meant to put 4 mins

but I accidentally added a zero

[olive sable]

thats 67 minutes

[amber fractal]

hour of audio

[opaque sigil]

That's one big mp3

Oh

Guess that's why

[covert bane]

just listen to like 3 or 4 mins

it sounds like a cow

[olive sable]

it sounds like it should have been 10 minutes but it got stretched out

[amber fractal]

Friday Night Funkin (real)

[covert bane]

true

there was no prompt on this audio btw, I'm justing a different architecture than most audio AI systems

[amber fractal]

We love the bgm resetting every major vocal

otherwise the FNF joke stands

[covert bane]

lol

[amber fractal]

I'd know considering how much of that game's mods I listen to

[jade sedge]

[amber fractal]

I was going to flame you on emojis in username, but I'll pass due to

[covert bane]

wdyt about this. I'm hoping for feedback

[covert bane]

what does mean?

[amber fractal]

is the equivalent of the sit and vibe gesture to me

[covert bane]

[amber fractal]

it is heavily used here

[covert bane]

okay

where is everyone, they were here just a minute ago-

[amber fractal]

it happens, you have to accept that the voices in your head are gone

[covert bane]

wha- no they aren't!! I need them back!

[opaque sigil]

[covert bane]

anyone?

[rigid snow]

why this happen twice

=

=

[amber fractal]

:bun: my favorite js tool (I don't use js)

[jagged turtle]

uh huh...

(who else uses bun besides anthropic)

I like, really have not seen anyone use bun outside of those people

[rigid snow]

could neeeeever be me

[covert bane]

Hi KTrain!

[jagged turtle]

I tried to use bun and like

[rigid snow]

nuh uh never

[jagged turtle]

[jagged turtle]

[covert bane]

[jagged turtle]

[covert bane]

can you give feedback on this for me?

[covert bane]

[jagged turtle]

cant put my headphones on rn, will try later

[rigid snow]

ye sounds like it jumps around, some sort of sample rate issue

[covert bane]

yea, I forced it to tokenize a audio file with a higher sample rate that it's supposed to. And forgot to remove the sample rate limit on generation as well

[rigid snow]

maybe the network outputs 20k and you encode in 44k or 48?

oh you trained it like that

lmao

[covert bane]

can you tell that audio was from an nn with only 4 mins of training data?

[rigid snow]

that wouldn't be my first thought

[covert bane]

I was too lazy to find another file that supported its sample rate lol

[covert bane]

what would be your first thought on the training data then?

[rigid snow]

no i mean like it being generated isn't apparent

i wouldn't think it is if i heard it randomly

[covert bane]

like, you wouldnt be able to tell its AI generated?

[jagged turtle]

oh btw mlntcandy what's the diff between function doSomething() {} and const doSomething = () => {}

[rigid snow]

maybe i would, all i'm saying is it's not like suno

where you hear one second and immediately clock it as ai

[covert bane]

it smh nailed the guitar and instrumentals from my thought atleast

[covert bane]

oooh

[rigid snow]

can't really judge since it's fucked

[covert bane]

not that well. but its quite good at like 1:10 smth

[covert bane]

okay!

[rigid snow]

weird this shenanigans iirc
uhh
one is anonymous still technically?

[covert bane]

0:55 also seems good

[covert bane]

the vocals it tried?

well, I did train it on a music video. which doesn't really give that much patterns for speech

[jagged turtle]

:classic: this in js

js has like

a million different module types isn't there

[covert bane]

what are all their purpose?

[jagged turtle]

uhhhh

they load differently

cjs loads synchronously

[covert bane]

theres cjs?

[jagged turtle]

esm loads asynchronously (so you can top-level await)

[covert bane]

I only know of ejs, and mjs

[jagged turtle]

uhhh i think there's also umd for like

browser or smth? idk I saw that on the ts neuro sdk

anyways, i wonder if tsgo will let you output ts -> bin

(as in if they plan on doing it)

[rigid snow]

you can't .bind(newThis) the lambda one because this there acts like any other value from the parent scope
they are both values... but the declaration as you typed it out will hoist till the beginning of the scope you declared it in

idk what else to add

[jagged turtle]

im so confused

all I can read is 99% it doesn't matter

[rigid snow]

basically don't use lambdas as oop methods

and the last part is still relevant it just means

yo()
function yo() {
  console.log("gurt")
}
``` will work

[weak horizon]

uh.....

if (userInFo.cookies.agreed) {
     Collect(user.data)
} else {
      Collect(user.data)
}

[covert bane]

why add that if statement at all, if you're just gonna Collect(user.data) anyways. Just do:

// some code here
Collect(user.data)
// some code here

[weak horizon]

Providing psychological comfort to users (joke

[amber fractal]

Why isn't the annotation showing up

Whatever, we take python not complaining

[olive sable]

[amber fractal]

I just want the dtype to be known

[idle dune]

I did it

My Ubuntu installion is complete

its stable

It has support fot flatpak

[opaque sigil]

Stable until glibc is too old

[idle dune]

and I ofied my vulkan to only run for my gpu

[idle dune]

https://tenor.com/view/ricky-bobby-gif-3479536

[rigid snow]

stable and up to date are different things

ask any production debian system

8 years uptime, apt-get upgrade never ran

[sage crag]

autismu

🈷️

[olive sable]

fine ill change the name

done

[amber fractal]

Chat, can I be free from this typing hell now?

[olive sable]

yes

be free my child

[amber fractal]

[olive sable]

where the fuck is the transparency rendering setting in blender and why is it always off by default?

[rigid snow]

what are you talking about

[olive sable]

its easier to turn on raytracing than transparency

[amber fractal]

transparancy costs extra math

[olive sable]

transparency

[rigid snow]

i literally have 0 idea what you mean

blank project

[olive sable]

what the fuck

oh

you're setting the alpha of the principled bsdf

i need it to use the texture

[rigid snow]

the alpha channel?

[olive sable]

of the texture yes

there is a setting for it

yes you can manually connect all the nodes

[rigid snow]

you don't

like this ye

[amber fractal]

cube

[olive sable]

normally you dont even need to go to the shading tab

[rigid snow]

i know what you're getting at
my workflow is uuh

i live in the shading tab

[olive sable]

but the sepcific place you would set it up got removed in blender 5.0 or something???
https://www.youtube.com/watch?v=ZIfxQzWnYk8

that setting menu you can see underneath has blend modes

[rigid snow]

oh IN THE VIEWPORT

[olive sable]

idk anymore

ye im not rendering

[rigid snow]

i don't know then genuinely

[olive sable]

i like blender getting new features in updates, but the settings getting moved is a bit annoying sometimes ngl

and then you find youtube videos with outdated info

[rigid snow]

wait

do you need the texture alpha or just for the viewport material to be semi transparent @olive sable

[olive sable]

well, for now only the material, but eventually this will be a texture

[rigid snow]

because if the latter then xray exists

in the viewport material settings

i'm on blender 4 still but i struggle to imagine they'd move it somewhere completely different

[olive sable]

uh

[rigid snow]

wrong shading mode

[olive sable]

oh you're on the 2nd one

oh ye, xray does exist

thats weird

kinda creepy

[umbral thorn]

wooh

how gud ur topology

[olive sable]

Its decent

A bit high density on the hair still

[umbral thorn]

same

[olive sable]

Unfortunate perspective but im not at my pc rn

So vedal triangle

[umbral thorn]

buhh

urs better

[olive sable]

[olive sable]

Its honestly not too bad

Your mesh is just triangulated while i still have quads

Bit higher density too

[rigid snow]

this is for hair physics

[umbral thorn]

[olive sable]

Is it using bezier curves to generate the geometry around them?

[sage crag]

[rigid snow]

100% that ye

[nocturne olive]

My best lead so far on the weird stream issues, packet loss on the Twitch segment provider

Could this be a Twitch fault? Reports in Finland about Twitch issues increase noticeably after Neuro subathon start when I started having this issue

[cobalt moon]

Neuro is eating Finland

[trim valve]

maybe try a VPN? your might just have godawful routes to twitch from your isp

[nocturne olive]

I don't have a VPN
And even then I would like my IP to stay accessible

Also this issue literally started just yesterday after the subathon intro
The intro was flawless

[trim valve]

oh right I forgot you host stuff too

[nocturne olive]

Just like all streams before

[trim valve]

maybe chuck a traceroute at it?

[trim valve]

there's free stuff like cloudflare warp that might be worth trying. as for the accessible IP issue shrug

[nocturne olive]

I did try something like that and it said the loss is on the end server

[trim valve]

do you have a phone?

[nocturne olive]

Yes one does in fact exist

[trim valve]

does it have the same issues with the stream?

[nocturne olive]

Not at least on mobile data

[trim valve]

would make sense if it's a routing issue

[nocturne olive]

Hem
Seems a similar issue happens on my phone when I connect it to WiFi

Unless Twitch was just loading slow

Yeah something is not right on my phone on WiFi either

Swapping to mobile data and it's immediately normal

I'm gonna do something scary and reboot the router

[olive sable]

Ye thats nice for ease of use

My topology is now hard to rig

[trim valve]

ye if that doesn't fix it, try cloudflare warp on your phone

[nocturne olive]

Well based on the stream on my PC still misbehaving it did not

So at least I've ruled out it being an issue specifically on my device I guess

So something like the crash yesterday corrupting the kernel isn't a thing

[nocturne olive]

I have no clue how
I would also very much like to know why it just suddenly happened when it did

And why it hasn't been fixed yet

Usually these kinds of issues are gone pretty quickly

[trim valve]

https://play.google.com/store/apps/details?id=com.cloudflare.onedotonedotonedotone

[nocturne olive]

Hem
I didn't even try that yet but now it works on my phone when I swapped to DNS 1.1.1.1 after restarting the router, not sure which fixed it on there

Gonna try DNS 1.1.1.1 on my PC

[trim valve]

I guess 1.1.1.1 might be giving you a different aws backend to your isps DNS server

[nocturne olive]

Hem

It did not work on my PC

But why though

Streamlink definitely still not having a good time

[trim valve]

sanity check, does your DNS actually work on your pc?

[nocturne olive]

And how would I validate that?

I've flushed the caches twice at least and it's still working so

[trim valve]

as long as you can open other sites like google

[nocturne olive]

Yes I can in fact open Google

[trim valve]

then yeah no idea why streamlink is unhappy

[nocturne olive]

Twitch browser app also unhappy

[trim valve]

anything in the network tab?

[nocturne olive]

Can't run anything higher than 480p and even 480p is very laggy

[nocturne olive]

Network tab? Where?

Which network tab?

[trim valve]

open dev tools

in your browser

oh wait you're getting 404s in streamlink

how and why

[nocturne olive]

Hem
I restarted the browser after the DNS swap and now 1080p is working longer than before at least

Oh there it goes

[jagged turtle]

are people in your area just all watching the subathon

[nocturne olive]

I don't know

I don't know of other Neuro fans in the nearby area

Restarting stream capture to see what happens

It's at least super convenient since it's just a systemd service

Hm

Things are looking more stable than before on Firefox at least after restarting it after swapping DNS

There's no way swapping to DNS 1.1.1.1 and then restarting the capture service and browser fixed it somehow

I'll keep monitoring but it would be amazing if it fixed it

[jagged turtle]

it's always dns

[nocturne olive]

Networking is wizardry even a level above instrumental music

[jagged turtle]

I understand like 10% of it, I think

[nocturne olive]

Now I can finally read, whatevver this is

[jagged turtle]

-# I probably don't

[jagged turtle]

there has to be a better way of displaying this lmao

[nocturne olive]

Well that better way would be like a Twitch extension probably

[jagged turtle]

Yeah

[jagged turtle]

TIL Ctrl+Shift+C in VSCode opens up a Command Prompt window in that directory

[nocturne olive]

Funny

For me CTRL + SHIFT + C is copy from terminal

[jagged turtle]

that's linux aint it

[trim valve]

windows terminal is the same

[sage crag]

:ALERT:

[mint shuttle]

https://scriptkiddiesheaven.dpdns.org/view/neuro-sama-discord-server-activity_b18abbef.mp4

[sage crag]

even discord wont embed this

if you want to send a video send a video

[faint sandal]

[trim valve]

what if

no embed perms

[sage crag]

True!

dont send videos at all then

[mint shuttle]

forgot to install ffmpeg scriptkiddiesheaven died 🙁

[sage crag]

You have not Unlocked New Role.

[faint sandal]

You have not Unlocked New Role.

[sage crag]

ye that what i said

[tender river]

sick

[sage crag]

leaf need

medicine

[tender river]

leaf need drink

[sage crag]

cheerleaf

[olive sable]

☕

[sage crag]

NeuroInspect

[olive sable]

my brother bought this, its beautiful

[warped narwhal]

🥃 have several litres of whisky

[tender river]

disinfection

[sage crag]

AtorinConfused

only if "infection" == "liver"

[olive sable]

this is the full model rn

[nocturne olive]

More black magic

Sily

[olive sable]

the clothes are baked on wih some simple physics simulation so the topology there is not of importance

for actual production models id sew the clothes onto the skin

[nocturne olive]

3D modeling is pure black magic

[olive sable]

removing unnecessary tris hidden behind the clothes

[olive sable]

its quite simple actually. you open belnder, delete the start cube, add a new cube, and extrude the vertices into the shape you want

https://tenor.com/view/blender-default-cube-delete-default-cube-cube-a-cube-gif-6936712150193305105

[nocturne olive]

Well I sure don't know how to do it, Blender feels so confusing

[olive sable]

the hardest part is probably know ing some anatomy

cuz its hard to make it look human

everything has some weird dents

[nocturne olive]

I don't know maybe the stuff I've tried to do in Blender was just stupid hard or Blender is just too hard for me

[olive sable]

it jsut requires a lot of practise

there are a fuckton of shortcuts

all of them exist to make your life easier, but a beginner doesnt know them so does stuff the hard way

[nocturne olive]

I can't even 2D art yet, I won't be doing 3D any time soon

[true hemlock]

bought 4TB gen 4 nvme with TLC for $300

can't be helped. nvme price is going to skyrocket about 2x by next 2-3 weeks

1TB is already going for over $130

[silent cloak]

not bad given the prices rn

some are insane but not ram levels yet

[midnight sigil]

that is unnecessary amount of subdivisions on the cloths

and toes holy molly I just realized that

that's why the poly count is so high

[midnight sigil]

use box method instead of sculpting- WOAH

[silent cloak]

here is my topology

pls review

[tender river]

https://thechief.io/c/editorial/how-kubernetes-is-shaping-the-future-of-cars/

[midnight sigil]

[tender river]

what.

[midnight sigil]

I do NOT want any cars to have "connecting to the internet" as a main feature

[tender river]

i do not want any cars to have

kubernetes

[maiden geyser]

what's the point of a car if it can't generate mommy asmr and slop while i'm watching tiktok and ig reels at the same time

[olive sable]

Vecause we're viewing it from the front

Its not actually that many

[sage crag]

sendoro

carbernetes

[silent cloak]

me when i shutdown my friends car's docker compose on the freeway

[midnight sigil]

it's still overly a lot

how did you do the curve part

[sage crag]

wrr

[midnight sigil]

wrr

[olive sable]

Not at pc rn

Smash bros

[tender river]

enterprise distro even have article promoting this use case
https://www.suse.com/c/running-containerized-workloads-in-vehicles/
https://www.redhat.com/en/blog/running-containers-cars
car developer not well in head

[sage crag]

i am using

Array generator

to scroll through pages

[tender river]

oh that

[sage crag]

[olive sable]

[quick condor]

Saw that yesterday, all cheap old ram

[uneven pulsar]

just saw a listing on fb marketplace rtx 3090 24gb i9 10th gen 32gb ddr4 1tb hdd 500gb nvme ssd pc for $955

[nocturne olive]

A bit expensive if it was just a 3090 but not too bad with the other stuff
That RAM is probably worth like 300+ alone if you resold it

[uneven pulsar]

but its a dell precision

[olive sable]

Honestly

Not too bad still

[sick owl]

Still gold in this economy

[sick owl]

That's an amazing deal tbh

[quick condor]

I mean gold is gold, that use to be ram

[sick owl]

Wrong channel lul

[sick owl]

↓ technically true

[quick condor]

I think that's something like a couple hundred lbs of ram but I can't remember, it's been built up over years

[uneven pulsar]

right!! only if i had spare $1000 in my pocket to spend

[sinful cipher]

i bought some sticks of ddr5 ram off of a friend a little while ago. they regret that so hard now

[lime stag]

HOW IS THIS POSSIBLE

MADLAD

[rigid snow]

oj it

's cloud

article about nothing "k8s exist and car companies use it" so does everyone else

why would you need k8s in a car anyway

[tender river]

no no used in the actual car

article just badly written

[rigid snow]

but where why

what problem does it solve

[tender river]

car run on microservices

[rigid snow]

why need k8s specifically

you can just uuh

not

even if microservices

[tender river]

ye there are also other options like k3s or bluechi apparently

another one https://customers.microsoft.com/en-us/story/784791-mercedes-benz-r-and-d-creates-container-driven-cars-powered-by-microsoft-azure

Nunes’s team of engineers already used microservices to speed development and deployment in the cloud, but they wanted to update the software more frequently and to implement continuous integration (CI) and continuous deployment (CD) practices inside the car

webscale infotainment

[sage crag]

nune

[worldly plank]

So what I am reading is that "We want to turn everyone into beta testers".

[silent cloak]

surely nothing could go wrong

[olive sable]

[olive sable]

chat dont be weird

[rigid snow]

insert me being weird

[severe path]

Is that some reverse psychology?

[olive sable]

it is not reverse psychology

this 3D model is my creation so i will kill weird people

[severe path]

Ok

[rigid snow]

spoilering is crazy tho

[severe path]

It looks nice, good job

[olive sable]

[olive sable]

better to be safe i think

[obsidian mantle]

did you make it from scratch

how does it even work

you start with a cube?

[olive sable]

so uh

you start with a circle

[obsidian mantle]

and carve it like sculptor?

[olive sable]

and you smoosh it

[obsidian mantle]

[olive sable]

and curve it

then extrude

[obsidian mantle]

you mean cyllinder?

[olive sable]

and sculpt

[obsidian mantle]

or circle

[olive sable]

circle

[severe path]

That sounds painful. Poor cube

[olive sable]

circle

[severe path]

Oh.

Poor circle

[obsidian mantle]

cant imagine how it works

[severe path]

I'm not cubist or anything

[olive sable]

[rigid snow]

it doesn't matter what you start with, can start with a single vertex

you just extrude extrude extrude

[obsidian mantle]

i need to google that word

[maiden geyser]

can one be 𝓯𝓻𝓮𝓪𝓴𝔂

[obsidian mantle]

does it make new vertexes when you extrude

[rigid snow]

ye

vertices

[olive sable]

murder

[sage crag]

isnt it weirder to spoiler and make the assumption that someone will be weird

[olive sable]

idk

[sage crag]

StareAtYouNeuro

[rigid snow]

that's what i thought too

[olive sable]

im not sure how this works

im trying to have a mature conversation here

[sage crag]

https://tenor.com/view/typing-gif-gif-25597112

^ sam

[maiden geyser]

what if one didn't spoiler the thing and the copypasta happened

[olive sable]

foot is an extruded circle with 24 verts. toes are a circle with 8 verts.
you connect them by bridging 4 inbetweens int he 24vert circle until you have 5 holes of 8 verts

[severe path]

What copypasta?

[olive sable]

then you end it with a subdivision surface modifier

[olive sable]

?

[maiden geyser]

"please put a nsfw tag on this"

[olive sable]

[rigid snow]

I LOVE INTERSAMPLE PEAKS

[olive sable]

what are intersample peaks?

[rigid snow]

[olive sable]

do you lose volume?>

[rigid snow]

the opposite

master is at 0db but then different things happen depending on what you play it

ios (iphone dac?) just starts crackling

vlc puts an awful limiter on (interestingly on mp3s only) and makes it jump around in volume

streaming platforms normalize by perceived loudness so it's turned down anyway but ISPs trick it to turn it down less than it actually is so it's just louder than other songs

what else

[olive sable]

uh

so it doenst do anything specific

it can do a lot of different things depending on the platfrorm

[rigid snow]

it is either an annoyance or a feature

but mostly annoyance

[rigid snow]

something about "Inter" being written in Inter font activates my neurons

[olive sable]

whut

[rigid snow]

font used is called Inter

[olive sable]

cool i guss

[rigid snow]

and it says Inter

[nocturne olive]

That's funny

I don't know if I've ever encountered that though I may be too stupid to notice those anyway

[rigid snow]

they are natural and you definitely have encountered them. they only cause problems when they go past 0db and you don’t do anything at 0db afaik sooooo

[sage crag]

[olive sable]

[sage crag]

[uneven pulsar]

Hello is neuro run locally or on cloud?

[olive sable]

we dont know

[sage crag]

neuro is a real person

[maiden geyser]

on a thunder

[sage crag]

@silent anvil is neuro run locally or on cloud

[olive sable]

[uneven pulsar]

Did vedal use an open source llm and add custom dataset?

[sage crag]

@silent anvil

[uneven pulsar]

what is vedal bot?

[maiden geyser]

we don't know

[sage crag]

its vedal

[rough bloom]

likely something like that but we don't know

[uneven pulsar]

Who is vedal

[sage crag]

[olive sable]

[sage crag]

@silent anvil

[uneven pulsar]

?

[rough bloom]

@silent anvil please explain

[warped narwhal]

[sage crag]

@silent anvil at the earliest convenience

[tender river]

new rule

pinging vedal bot

banned

[sage crag]

@silent anvil is it true

[warped narwhal]

is it time to pull out the @ vedal gif again?

[olive sable]

= vedal
🦟 =
🧊 🐟 = 🦟

[uneven pulsar]

Will I be getting any answers in this chat?

[warped narwhal]

ask better questions then maybe

[sage crag]

@silent anvil can you confirm if this chatter will be getting any answers?

[olive sable]

depends on the answers you're seeking

[sage crag]

say nothing for yes

[warped narwhal]

@silent anvil please

[uneven pulsar]

Who/which channel has the information i am looking for?

[maiden geyser]

opossumbly

[warped narwhal]

!remindme 2days will vedal bot ever talk to us?

[olive sable]

we're your best be basicly

[sage crag]

vedal's synapses

[rough bloom]

the FAQ
forgor where it is

[warped narwhal]

ping vedal and ask him
-# don't do this lmao

[tender river]

vedal's ion channels

[sage crag]

ye

[uneven pulsar]

you just spammed vedal

[olive sable]

FAQ

[rough bloom]

ye that
also this #1326303411041206303 message

[warped narwhal]

no we pinged @silent anvil

[uneven pulsar]

Thank you this is helpful, I had this rough Idea

[sage crag]

youre welcome for all understanding or lackthereof

that we did or did not provide

from the past up until tomorrow

[warped narwhal]

no warranty after tomorrow though

your on your own there

[sage crag]

no its just

the world is ending tomorrow

so

[uneven pulsar]

I like this channel and chat.

It's too light headed

[rough bloom]

[warped narwhal]

[sage crag]

its ok they have a new one

[maiden geyser]

b-baka

[uneven pulsar]

I will be back in a few eons.

[sage crag]

ok

take your time

[uneven pulsar]

Thank you.

[sage crag]

true

[warped narwhal]

-# and he was never heard from again

[sage crag]

well

if youre still around in a few aeons

you can validate it

[warped narwhal]

I'll report back

[uneven pulsar]

Why?

[warped narwhal]

https://tenor.com/view/joke-missed-over-your-head-gif-8604199

[uneven pulsar]

Alright, haha.

[maiden geyser]

unjolly people don't stay for long

[uneven pulsar]

such is life.

[warped narwhal]

he failed the #programming vibe check and therefor shall be subjected to 300 hours of debugging yandere simulator

[sage crag]

i didnt catch the wayland quote on screenshot

@tender river wayland mentioned by nur on stream

[warped narwhal]

sound about right for wayland

[sage crag]

another one

[warped narwhal]

arch as well lmao

[uneven pulsar]

lmao.

[sage crag]

there was an earlier quote

7% bsd users

[rigid snow]

holy fuck why is she this based

[sage crag]

ope ads

well, no more quotes i guess

[rough bloom]

first quote

[maiden geyser]

only on linux

[sage crag]

nixos mentioned

[tender river]

me watching make easy to focus

[warped narwhal]

nixos mentioned

[sage crag]

"persistent 0-day exploits for common kernels"

[rough bloom]

NeurOS

[sage crag]

multi-kernel os

why did we start flaming tutel can we go back to linux

[rigid snow]

call it… swarm

[warped narwhal]

if one is good why can't two be better?

[sage crag]

Neur in mourning

lost 2k viewers after mentioning bsd

[tender river]

https://www.kernel.org/doc/ols/2010/ols2010-pages-101-108.pdf me find this

[sage crag]

pages 101 108 dot portable document format

lost 2k more viewers

harrison temple not popular

templeos this templeos that

[tender river]

The project can be deployed in Linux environment for
Intel core 2 duo architecture to efficiently utilize both
the cores simultaneously.
2010

[nocturne olive]

Mabe
I do in fact not target exactly 0dB

[rough bloom]

ye there were some patches to make this work recently

[sage crag]

hobbyist osdev and not even close to the best solo developed os

minyanstare

its popular

[rough bloom]

https://lore.kernel.org/lkml/[e-mail]/

[nocturne olive]

Sily

[sage crag]

this outfit is great by the way

[nocturne olive]

-# I'm about 25s behind on the stream due to Twitch being terrible garbage

[rigid snow]

2 years late

[rough bloom]

glad that it finally got released
was stuck as a concept for so long

[sage crag]

AtorinSoCute

[tender river]

ye because of surrounding context most people dont care about hobbyist os code wise

[sage crag]

ye

popular internet personality

transfer to OS

1779 standard tls port

why do you still call it port that not cyberpunk futuristic at all

stopie

[tender river]

why iced crash all the time

[sage crag]

that the thing in the stream

ICE

sigs is actual slang though

unlike "virtu enviro"

[tender river]

every time i get random app issues i pray its not my ram that a bit expensive

[sage crag]

my pc was freezing for about half a year

randomly

i was blaming kde

then i got worried it was my RAM

turns out it was my bios

[tender river]

get fixed now

[sage crag]

yes i had to update my bios

[tender river]

once my grub stopped working after a linux update (it was booting into linux but had some random corruption in initrd or kernel code) turned out grub's btrfs impl was buggy or not updated

[sage crag]

one reason me dislike grub

evilAYAYA

[tender river]

gnu software

add feature after feature

[rough bloom]

GRUB has btrfs support

[sage crag]

ye

what is fat32

[tender river]

please i need it for my 2000 nixos kernels

[rough bloom]

still need fat32
UEFI not support btrfs

[sage crag]

nix boot without bootloader

efi image

300 efi entries

minyanstare

[tender river]

oo another bios issue my past laptop had

actually that happened in london i think

[sage crag]

london bios issue

[tender river]

or not whatever not matter

anyway it had an ntfs impl

but at some point my drive got into a state in which the bios just froze trying to read the drive

i had to do open surgery, unplug drive, boot, plug it back in to get linux to see it and fsck

not safest solution

but if it was in london (i remember having to buy a screwdriver to do something to fix my laptop ) then makes sense

it broke right before a tournament match too

[sage crag]

MiyagiCroco

croco

[kind nimbus]

Lol