programming

[tender river]

[rigid timber]

ah, you can't then

[trim valve]

and my pc is not responding to remote desktop

[rigid timber]

rip

[solid bough]

I would personally go to LiPo if that is safe.

[olive sable]

i had game history class today, so like atari and magnavox odyssey, which was cool. i also had art history, which sucked a lot

im too colourblind to see the difference between primary red, secondary (not?)red, and ternary (not?)red

[glass flower]

" cargo:warning=cl : Command line warning D9002 : ignoring unknown option '-std:c23'"

[tender river]

cl

[rigid timber]

wat

[glass flower]

i tried adding the standard myself to the build script

[rigid timber]

how exactly?

[glass flower]

like it says in the docs

[rigid timber]

strange

[tender river]

try c11 maybe that should have threads.h

[obsidian mantle]

my clang failed on windows too, try gcc g++

[silent cloak]

clang acting like clank

[obsidian mantle]

i actually hasnt checked gcc on windows maybe it works

[glass flower]

no you don't get it... clang works. msvc doesn't

[clear sedge]

i love this

[obsidian mantle]

i see

[trim valve]

try the x86_64-pc-windows-gnu target maybe

[clear sedge]

msvc skill issue

[tender river]

its clatest on msvc

or c17

[glass flower]

still doesn't work

[glass flower]

now it can't find threads.h at all

[rigid timber]

msvc should support c23 just fine

[tender river]

depends on your build tools version

[rigid timber]

that is true

[tender river]

more of a windows skill issue

[rigid timber]

I mean... if you get an old version of gcc, it won't support new stuff either

more of a C stupidity

[tender river]

i'd agree if only there werent that many issues even with clang and gcc on windows

[opaque sigil]

Old msvc

[rigid timber]

I had zero issues with clang

worked first try

[tender river]

good for you

[opaque sigil]

clang/gcc tend to require a lot more work to get working because of all the windows specific things msvc does for you automatically
Worth it though

[warped narwhal]

[ivory plinth]

Concerned

[warped narwhal]

I don't think this is the place for you. You're looking for #neurotic-neurons

[glass flower]

this works now...

[opaque sigil]

Lucky

[glass flower]

but also.... board.get_piece_from_bitboard(made_move.to); it still crashes for me...

wait

i think i figured out why it crashes

i can't get the piece of a future move that hasn't been made yet....

[tender river]

no

[rigid timber]

wat

[tender river]

okay kinda

but not in the way you think

[glass flower]

it tried to get a piece from a bitboard. but if there is no piece there yet. because you haven't moved yet. so it crashes

also move.from.to_string() doesn't crash anymore

[tender river]

@trim valve get_piece_from_bitboard can return 0, so it should return Option<PieceType> not PieceType

[glass flower]

makes sense

[tender river]

same goes for get_piece_from_index

and Move should ideally use Option<PieceType> for promotion and the rust API should be responsible for converting it between your API and the C API's reprs

[glass flower]

but it works now. without me having to do weird things (besides having a local build.rs version)

[trim valve]

very fair, it's kinda hard to tell when stuff is mean to be nullable

[tender river]

[opaque sigil]

[clear sedge]

[fast pagoda]

damn

[torpid kelp]

quick CSS, html, javascriot(maybe)/ wikitext question
is it possible to creat a custom variable for template colors on wiki.gg , and if so, how do i make it ajustable with my themes?

[tender river]

you generally wont get an answer for hyperspecific things like that in here

[opaque sigil]

Just write into window

[torpid kelp]

worth a shot

[tender river]

your best bet is either asking chatgpt or researching the issue until you can reframe it in a way thats easier to answer for people who dont know anything about wiki.gg

[torpid kelp]

i shall never succumb to chatgpt
is it possible to create a css variable for color , that changes with themes ?

[tender river]

there are no "themes" (except dark and light mode but thats not theme, thats more like a thing the browser tells your website)

but css does have variables

you can define variables in one css file and the other css file can use these variables

then you can replace the first file with one with different variable definitions

i dont know if that answers your question

[torpid kelp]

oh i thought that was what the light/dark things were called
thanks you! that helps alot :D

[nocturne olive]

I just discovered GUMI singing Finnish and now I want NeuroSynth/JP to do it myself
So uh, any JP knowing people here that would be willing to do some stage 2 transcription?

[tender river]

you can find out whether the browser is set to prefer dark mode or light mode in css if you really need to

but you'd think wikis have some sort of builtin theming so you wouldnt have to do that

[torpid kelp]

they do!
i just wanna add a varible but i couldn't figure out how

i figured it out tho, ty :D

[fast pagoda]

https://chromewebstore.google.com/detail/custom-css-by-denis/cemphncflepgmgfhcdegkbkekifodacd?hl=en addon like this will let you do custom css on a per-domain basis

there's also stylish and similar addons

[trim valve]

ok rust bindings updated to address this

[fast pagoda]

and finally you can (temporarily i think? idk if it persists) edit the css in developer console (f12)

and then i read you figured it out lule

end me

[glad path]

hi

[glad path]

safari lets you set a global style sheet
does chrome not let you (ik that extension is per domain)

[opaque sigil]

Chromium does not no

[glad path]

darn

i still dont get why people hate safari so much

its really not that bad

[obsidian mantle]

why is my wsl acting different today

i cant access anything

yesterday it was fine

[glad path]

whats a wsl again

i forgot

[obsidian mantle]

linux subsystem for windows

[glad path]

k

[trim valve]

did you put your stuff in the windows file system?

[obsidian mantle]

it forgot it root or smth

[glad path]

rip

[obsidian mantle]

the hell

[trim valve]

cd /mnt/c/

that should put you in the windows file system

[obsidian mantle]

nice

i didnt have to do it yesterday for some reason

[trim valve]

how did you enter wsl this time?

[glad path]

hooray

i forced a steam cloud error so that i didnt loose progress

[obsidian mantle]

same as yesterday, through windows search here

anyway

draw

[glass flower]

doesn't crash anymore

[fast pagoda]

i always found easiest was just to open terminal from the folder i wanted and just wsl and then you're there but i dont think that changed much with it finding a root

[opaque sigil]

i always did wsl ~ because why the hell would i want a wsl shell in my windows home folder

[tender river]

i liked wsl 1 for how tightly integrated it was

but obviously microsoft didnt want to deal with compatibility issues and just went with virtualization instead

[opaque sigil]

[real sierra]

oops

[obsidian mantle]

it doesnt let me access board

it thinks this is it and doesnt see the actual definition in c file

i can probably work around with pointers

[tender river]

you dont need to

there are functions that use a Board*

thats all you need

[sage crag]

ye its intentional

typedef struct Board Board; is a form of encapsulation

[obsidian mantle]

i see

[sage crag]

prevents you from accessing struct fields

[obsidian mantle]

i can access it by reading it out of the pointer though

probably waste of tokens

[tender river]

you can but shiro can change the internals

and your code will break

[obsidian mantle]

🥺 shiro dont

[tender river]

shiro do

[obsidian mantle]

ok i will try with those functions then

so i have to get piece by index or smth huh

[tender river]

just read the entire header first to see what options you have

[sage crag]

shiro do

[tender river]

https://cdn.discordapp.com/attachments/1067638175478071307/1406337271241904158/ezgif-3f552200e37817.avif

[midnight sigil]

it's in the top of chessapi.c

but it's internal, all you can access is through the exposed functions

[midnight sigil]

merge pr and close corresponding issues

[real sierra]

time to merge a PR on mobile

merged

thank you

@nocturne olive your issue should be fixed now

[nocturne olive]

I saw

[midnight sigil]

thx Mr. Shiro

can I get extra salary before next paycheck?

[real sierra]

I'll give you half of my salary

[midnight sigil]

don't tell me it's all debt

[real sierra]

[nocturne olive]

IT WORKS

‼️

The illegal moves still are a thing though

[midnight sigil]

hmm plz provide more details, maybe send them in an issue chain

[nocturne olive]

I already gave Shiro logs from 2000 games

Some of which had the bug

[tender river]

shiro will fix it give him time

the deadline is november 1st anyway

[fierce ether]

bot is using 1788 tokens

so close yet so far

[midnight sigil]

2000 games

[tender river]

the competition is supposed to last for a month and you're given more than that

[midnight sigil]

wrong reply

[tender river]

so just wait until shiro fixes it

[tender river]

the issue was related to castling

[midnight sigil]

I see

[real sierra]

yes I need time to look into and fix

I have other priorities irl that stacked up at the same time so I gotta get those out of the way before I have more time for chess dev

my poor bot is just begging to be worked on

[sage crag]

phrr

[real sierra]

phrr

emote...

[sage crag]

[fierce ether]

1768 tokens

locked in

[real sierra]

koniivm

another

[sage crag]

koni always writing vms these days

koni made this one way faster than all previous ones at the cost of pre-execution translation

[real sierra]

h1vm still not done

[sage crag]

true

koni want to do this first

[real sierra]

📺

[sage crag]

you missed

[real sierra]

[tender river]

[real sierra]

---

📺

saw this for the first time today

number present at same time as letter

[sage crag]

#define ​ __builtin_printf("hello, world!");

[shadow sinew]

Could I request a C API feature/improvement? get_legal_moves() does its own memory allocation which requires tons of reallocs as well as bookkeeping in the calling code (since allocated memory needs to be freed).

Could we have a version that takes a pointer to an array of Move as input and writes into that? That way the calling code can take care of memory management. If the array is too small you could simply stop writing. (and just return the number of found moves)

Apparently the most known legal moves for a single chess position is 218. So if the calling code ensures to size the array accordingly you'd never have to worry about freeing stuff.

[solid bough]

Hmmm... That's some code:py for i in range(num_epochs): model.fit(x_train, y_train, batch_size=batch_size, epochs=1) mse = model.evaluate(x_train_mini, y_train_mini, batch_size=8, verbose=0) train_loss = model.evaluate(x_train, y_train, verbose=0) train_rmse = np.sqrt(train_loss) mse=model.evaluate(x_test, y_test, batch_size=batch_size, verbose=0) test_loss = model.evaluate(x_train, y_train, verbose=0) test_rmse = np.sqrt(test_loss) train_score.append(train_rmse) test_score.append(test_rmse) model.save('Model.h5') plotscrores(train_score, test_score, 'Scores.png', True)

And that is the most compact I was able to make it

That is what I started with:

[real sierra]

this could be done

[solid bough]

and yes, that is still about Deep doodle.

[real sierra]

if you'd like to make an issue for it on the GitHub, I'm trying to centralize bug reports and feature requests there

[shadow sinew]

I could try preparing a patch if you want.

[trim valve]

this is how I've had my keyboard since I changed to gboard

[real sierra]

that would also work if you think you have a good handle on how the legal moves function works

[shadow sinew]

Can't be that complicated right?

[fierce ether]

1744 tokens

i am unstopable

c quivers before me

[tender river]

not in XID_Start

[real sierra]

I would probably make the main legal moves function accept and populate an array, as well as a pointer to a length that begins initialized with the array max length, and then populates the array with moves up to that length

then add a wrapper on that where the supplied array is malloc'd, and returned afterward, to reimplement the original functionality

that would be the easiest way to integrate such a feature I believe

[sage crag]

nyan

byan

pyan

erfie

[solid bough]

Hi konii 2.0.1

[real sierra]

ie

[nocturne olive]

I wanted to run 10000 games of random moves vs my first bot vs my latest bot but somehow it turned into 30K games instead
Guess that'll take a while

[real sierra]

does your bot think that quickly

[twilit vector]

c language?

[tender river]

its c

[sage crag]

bad c

[tender river]

if (!strcmp(tok, fl_strnames[i]) && printf("instruction: %s\n", tok)) goto done;

tokens

[nocturne olive]

I think the Chess GUI is slowing it down more than its thinking speed

[sage crag]

ye unrelated to chess tournament

i can expend so many token

[nocturne olive]

It itself is also a bit on the slower side though, it recurses trough two turns worth of moves after all

[tender river]

tokens

[real sierra]

two turns

that makes more sense

my bot has a time budget and explores as much as it wants to within that

[nocturne olive]

Yeah it now looks at its own turn and the opponent's turn after that

[nocturne olive]

I don't know how to schedule stuff like that or how to make an efficient search algorithm
Today I saved almost 90 tokens though

[real sierra]

90 TOKENS

it's a good day if I can spare even 5

I think I currently have 2 tokens free

[nocturne olive]

Adding this after the while(true) alone saved 50 tokens

[real sierra]

insane

[nocturne olive]

It makes it so everything within the loop acts as if it was inside the ChessApi class so I can directly call its methods

There was also this slightly cursed thingy scoreCpy *= if (turn xor (from > to)) 3 else 1

[real sierra]

I remember that

[nocturne olive]

But yeah, lots of savings

My bot is currently 388 tokens

[sage crag]

[real sierra]

konii how shrink

how token

[nocturne olive]

Also there was this another thingy that slightly abuses the API scoreCpy *= move.promotion?.toCInt() ?: 1

[real sierra]

find ways

[tender river]

shiro you dont need tokens

[real sierra]

I do

[tender river]

just get rid of them

[real sierra]

[nocturne olive]

That removed a big when(move.promotion) block

[real sierra]

I need add feature to fix what transposition tables broke

[tender river]

anyway this is one example of token saving

[real sierra]

is goto better than if else

[tender river]

if it helps you avoid {} then probably?

[real sierra]

my if already has no ()

[tender river]

it wont save much tokens if one of the branches doesnt converge to the other one

or rather it wont save any tokens

[nocturne olive]

I don't know what I should implement next in my bot, I guess some recursive way to evaluate moves

[shadow sinew]

@real sierra created a pull request

[real sierra]

will review later

[nocturne olive]

So that I could go as deep as I want

Though the cost will be exponential as I go down the tree

[real sierra]

quiescence search

transposition tables

those would be my suggestions

[nocturne olive]

I'm clueless on what those are

[tender river]

well thats why you do research

[sage crag]

pphrr

[nocturne olive]

For MK1 I'm just trying to go off of what I can come up with myself, then I may do an MK2 where I look up algos on the internet

[real sierra]

https://www.chessprogramming.org/Quiescence_Search

[nocturne olive]

Just thought that would be a fun comparison

[real sierra]

https://www.chessprogramming.org/Transposition_Table

[nocturne olive]

So far I'm not done with MK1 yet, it's on subversion 8 so far

[real sierra]

CPW my beloved

[nocturne olive]

And I still have a ton of tokens left to go wild with

So who knows what garbage self-created algo I may end up with

And probably I'll end up with a hardcoded eval depth

[real sierra]

your algo is not too bad I think

I think iterative deepening is a good idea tho, that one is intuitive enough that you can probably avoid looking it up

after each depth you check the time remaining and decide to go one level deeper or not

[nocturne olive]

Well this is my current test going on, TestBot2 is while(true) random moves, Ver1 is the first variant of my bot, Ver8 is the latest variant

[real sierra]

I'd definitely suggest you to try to use the available time more, even if it's something as simple as a table of "if we have more than this much time left, search to this depth"

[tender river]

also shiro if you use gcc extensions you can goto between functions in certain cases

[clear sedge]

that would be such a mindfuck

[nocturne olive]

I'll see about that one, I'll probably either end up with a hardcoded depth , or something very scuffed
I have no clue how to recurse while checking the time just yet, the control flow for that sounds tough

[tender river]

i wonder if statement expresssions would save any tokens

[real sierra]

that's not too helpful, I don't have any functions that aren't inlined and aren't either recursive or terminal

[tender river]

(also clang supports gnu extensions too)

[clear sedge]

CLANGER

[real sierra]

gcc Gigachad compiler

[clear sedge]

GigaChad Compiler

it all makes sense now

[nocturne olive]

We will see
But either way so far I only have single-step recursion
I feel like recursing many steps while also accurately keeping track of steps and having the steps each be fully completed before starting a new one would require a very complex caching system to store moves

[tender river]

https://gcc.gnu.org/onlinedocs/gcc/Statement-Exprs.html see if this helps

[real sierra]

bfs, use FIFO queue

pop next node to search and when you search, add its children to fifo

keep popping until you reach a candidate of a new depth

then check time

[nocturne olive]

Oh so you mean construct a tree of moves in memory?

[real sierra]

that's what makes sense to me yes

[nocturne olive]

I guess that would work, but also would use a lot of memory to hold it all for large depths probably
My current code could also only do individual trees for every move as they're evaluated, rather than first caching all legal moves, because of how I use a forEach loop

[tender river]

oh i forgot, using auto instead of type* is a good way to save a token in some scenarios

[real sierra]

but that's C++

also I already macro'd every pointer

[tender river]

no its C

[real sierra]

it's c++

[tender river]

macros still take tokens

[tender river]

get in with the times

next you're gonna be surprised that C has function overloading

[real sierra]

no that one I knew

that's an actual C thing

[tender river]

auto is also an actual C thing

[nocturne olive]

At least as of now it would really only make sense to do a hardcoded recursion depth
Maybe if I rewrite most of the logic I could do a more robust system that uses a tree and timings

[real sierra]

While auto is a keyword in C, its common and powerful usage for type deduction, where the compiler automatically determines the variable's type from its initializer, is a feature introduced in C++11 and is not present in C.

50/50

[sage crag]

na

auto in c23

use it

[opaque sigil]

it's a thing in c23

[sage crag]

useful

[opaque sigil]

yeah

[tender river]

shiro relying on llm for info

[real sierra]

they added auto but didn't fix long

[sage crag]

yes because long is wontfix

[tender river]

it not broken

[real sierra]

bad behaviour is feature

[sage crag]

✅

[opaque sigil]

[sage crag]

bad behaviour intentional

[tender river]

welcome to <any language older than 10 years>

[real sierra]

Haskell

[tender river]

haskell was made in like 1990

well not that old but 1995 maybe

[shadow sinew]

Newer then Python

[tender river]

it has plenty of warts

[nocturne olive]

Now to decide if I should play osu or try to Chess

[real sierra]

wtf is this lecture

matrix factorization

[tender river]

[nocturne olive]

So apparently there's somehow a 1 in 26200 chance that the test bot fails to make a move in a game

I have no clue how that happened

And by test bot I mean the one that makes random moves

[tender river]

cause everybody's so scared

[nocturne olive]

No clue what happened

[sage crag]

I INTED

[tender river]

konii

wrong channel

[sage crag]

er

wasnt me

[tender river]

https://cdn.discordapp.com/attachments/1415898975235477524/1418753754538184774/cdawg-miniko-animation.gif

[sage crag]

https://cdn.discordapp.com/attachments/1407786158930399242/1407786160582955150/2968820-3.gif

transparent bg

[obsidian mantle]

my recursion is crashing

[sage crag]

you are using recursion in a chess bot?

how much stack do you think you have

[obsidian mantle]

i just want to check if the move doesnt give away my queen for example

[tender river]

its fine for tree search

[sage crag]

not trust it

[obsidian mantle]

its crashing though

[sage crag]

clang::must_tail

actually [[clang::musttail]]

[tender river]

cant always do

[final tinsel]

programming discussion in stream

[sage crag]

ye that why it good

must tail => no explode stack

[tender river]

what if i want to explode stack

[sage crag]

you have stack exploding function for that

[tender river]

[final tinsel]

rm -rf you know the rest

[tender river]

[sage crag]

int stack_explode(int *x) {
  int y = 1;
  return *x + stack_explode(&y);
}

it took me a while to make a function that the compiler doesnt optimise out

[obsidian mantle]

oh i free the board too early

[sage crag]

turns out, i forgot to explode the stack in the stack exploding function

[clear sedge]

..it has WHAT?

[tender river]

[clear sedge]

since when

[tender river]

c11 probably

[opaque sigil]

are you talking about _Generic

[tender river]

ye c11

[sage crag]

nope

[tender river]

i am

[clear sedge]

this is fascinating

[final tinsel]

neuro keeps going to the engineer applicating and then going to smthn else lol

[obsidian mantle]

i think i did something wrong

out of 10 games my recursion bot lost 2 games vs random bot, froze cute chess 4 times and drew 4 times by 3fold repetition

[sage crag]

study sama

[tender river]

https://cdn.discordapp.com/attachments/1371882394524712970/1394631865964691476/3dgifmaker20414.gif

[sage crag]

[real sierra]

I don't want to be an engineer

[sage crag]

real

[clear sedge]

holy fuck is she filling out job applications

she just like me fr fr

[tender river]

newero so true

[final tinsel]

they did it

[sage crag]

i see no problem

[rigid timber]

I wonder what this does, time to paste it in my terminal

[fast pagoda]

i can feel the productivity increasing

[sage crag]

coins

[sage crag]

no that's pixels

[tender river]

wrong this is

[sage crag]

wrong

[tender river]

[tender river]

liar

[sage crag]

[tender river]

https://cdn.discordapp.com/attachments/1067638175478071307/1406337271241904158/ezgif-3f552200e37817.avif

[obsidian mantle]

fixed evaluation function now it usually leaves random bot with bare king
but cannot finish

king mobility evaluation mhm

[real sierra]

if you want a very simple metric for that

give score for king's distance from center of board

will encourage moving the king to a corner

[obsidian mantle]

i was thinking more about attacking enemy king, not protecting my own. but thats good too

now whats this

[real sierra]

yes, I meant for attacking enemy king

[obsidian mantle]

7 turns 3fold

[real sierra]

much easier to checkmate on edge of board

in middle game you want your king safe in corner

in end game you don't want king in corner

[stark needle]

not this😭 😭

[opaque sigil]

is the android sdk that gigantic

[stark needle]

this screen goes with my brain surgery overlay

[real sierra]

i really should just write a token minimizer script

greedy macro solver

[tender river]

wouldnt it reduce to knapsack problem

[real sierra]

yeah

it's not hard to make

[tender river]

shiro knapsack problem is np-complete

[real sierra]

no actually it isn't quite knapsack

I don't have a limit on how many macros I can make

[fast pagoda]

AppCrashedFatalReport: getLastCrash not supported.

thanks

[tender river]

problem is assuming you have a b a b a, do you pick a b or b a for the macro? and this is an easy example

in an actual example there might be much worse cases

leading to exponential complexity

[real sierra]

I did say it was greedy

and I also didn't say the complexity matters at all to me

my code is pretty short to begin with

[tender river]

mhm if you just replace greedily it wont be optimal but it wont be slow

[real sierra]

algorithm I have in mind is O(n³) I think

doesn't need to be optimal, just needs to be better than handmade

[tender river]

use z function or prefix function

or dont if the speed is fine

[nocturne olive]

Instead of doing Chess stuff I may have spent my time today making Teto Finnish

[real sierra]

speed will be fine

[tender river]

here you go tho
https://cp-algorithms.com/string/prefix-function.html
https://cp-algorithms.com/string/z-function.html

absolute useless embeds

also if you start with longer sequences they might stop being optimal once you replace shorter sequences because of #define overhead being 2 tokens + 1 per use, but if you start with shorter sequences you may lose some longer repetition

the first issue sounds easier to work around

sorry i'll stop backseating

i just like string algorithms

[obsidian mantle]

this really slows down cutechess

if you check thousands of moves

or maybe my vm is super weak
will see later

Didnt check if its the same when messages arent rendered (with debug disabled)

[real sierra]

since it flushes each time, there will be some performance impact, but I found it useful for debugging

[tender river]

logs can hurt performance so much

but also i doubt cute chess's log handling is particularly optimized

[real sierra]

why would you put the most valuable question on the assignment last

now i dont have time for it

[tender river]

on my final math state exams in school i didnt even notice the last page

[real sierra]

hopeless to even try implementing this in the 40 minutes i have before the deadline

im just gonna derive the gradiant and hessian for part marks

[fast pagoda]

someone make it makes sense why i thought these were valid names for folders when i was sorting my backups to get off windows:

these contained: all my repos

all my configs

many of my inference related junk too

[opaque sigil]

speaking of valid names

proton created a couple c: folders on my ntfs drive, i wonder how windows would feel about that

[fast pagoda]

lol why did it put them on the ntfs volume

[opaque sigil]

i already had this massive ssd with games

cba to wipe it

[fast pagoda]

ntfs is p slow in linux overall even in a regular distro yea?

[opaque sigil]

nah it's fine

[fast pagoda]

i thought ntfs-3g was slow

[opaque sigil]

ntfs3 is pretty good

[fast pagoda]

i thought it was fine myself as well when i was on wsl but i just figured i wasnt pushing it or something

[opaque sigil]

some wine compat shenanigans

[fast pagoda]

i kept hearing how horribly bad it was supposed to be

[opaque sigil]

i haven't noticed anything horrendeous yet

well, besides some games flat out refusing to launch on that drive

[tender river]

wsl is a completely different matter

since its a vm accessing the host filesystem

[opaque sigil]

[tender river]

ntfs3 should be fast but i dont have any numbers

[fast pagoda]

yeah i just figured it was a wsl thing when i came to linux and i had no issues with my ntfs fs

on microsofts docs they still state that wsl2 io on windows fs is slower than that of wsl1

due to reasons

[opaque sigil]

you can thank the plan9 server for that

[tender river]

reasons being its now a vm

[fast pagoda]

the only real issue i had with ntfs on linux is that ntfs would break if i didn't shut down the way it wanted

it would say it was in some weird state and needed ntfs fix

to mount

[tender river]

shut linux or windows down?

[fast pagoda]

linux (this was after swapping, while i had ntfs drives mounted via ntfs3g)

[opaque sigil]

i had that once and it stopped my kernel from booting but i forgot why that happened

was a quick fix tho and hasn't happened since

[tender river]

you have boot relevant files on ntfs?

[opaque sigil]

nah

[fast pagoda]

yeah that's a no from me dawg i just had an external with it lmao oof that sounds scary

i did fuck up my fstab slapping my ntfs drive in there though

because it was unstable and needed ntfsfix before it would mount again

[opaque sigil]

something flagged the disk as dirty so the kernel freaked out trying to mount the drive

and died

[fast pagoda]

and of course then would fail to mount on boot

yeah that's any time ntfs shuts down "unsafely"

i had the same fstab sequence you did with that probably lmao

thank god for nano in a time like that

[keen hatch]

If you mount a whole disk there, it works relatively quickly. but of course if it's not NTFS

[fast pagoda]

the linux partition was fast

[tender river]

so the issue was fstab entries not being marked as optional

[fast pagoda]

yep

it was one of my first big fuck ups with linux lol

[tender river]

i think systemd initrd should take care of that

[fast pagoda]

not the way i jammed it in there i guess

cuz that thing was stuck as hell till i went into it w/ nano and commented the lines out

[tender river]

no arch uses scripted initrd by default

[fast pagoda]

is that complimentary to initramfs

or are they mutually exclusive

because every arch install i've had i've only noticed initramfs on boot but i could just be missing initrd

maybe because i'm on btrfs?

idk

[tender river]

same thing

[fast pagoda]

oh

well initramfs said it wanted no part in it

[keen hatch]

I have 3 4TB SSDs mounted in WSL2

[fast pagoda]

i need to just bite it and do that instead of my rusty spinnerz

well, i have a 1tb system m2 + a 4tb drive, that's my "faststorage" volume

then i have 4 tb of older drives in the case + another 8tb in an external hd enclosure

[tender river]

basically you'd add systemd hook to mkinitcpio and uuh replace some hooks maybe depending on your setup https://wiki.archlinux.org/title/Mkinitcpio

[fast pagoda]

to make it have a fallback?

[tender river]

systemd should be more resilient towards service failures than a script but honestly no idea

[fast pagoda]

you'd hope

[tender river]

(with systemd initrd it loads the fstab and converts it into systemd .mount units)

[fast pagoda]

although i hear shit talk about systemd still to this day

even tho the init warz were years ago

[tender river]

oh you can also just add nofail to mount options maybe?

[fast pagoda]

well, my mkinitcpio.conf is HIGHLY custom atm

[tender river]

another alternative is just not adding it to fstab

and using a systemd mount unit instead

[fast pagoda]

what i ended up doing was this

making them all btrfs

lol

but the ntfs drives were the ones in /mnt/ obviously

when they were still ntfs i just had them mount but with noauto

cuz half the time i needed to run ntfsfix to get them to mount no matter how i wanted to try and automount it

suppose there's probably a way to script it so it just auto ntfsfixes every boot lule

but yeah i basically said fuck that and backed up the drives and wiped back to btrfs -> put my stuff back on

i should've used like xfs or zfs for the storage probably but idk

btrfs is safe-ish except when it's pissing me off and refuses to increase its allocation

[tender river]

i once submitted a systemd pr to add a new mount option to systemd fstab

but when it released and i tried it it turned out it wasnt working for me

[fast pagoda]

[tender river]

useless pr hope at least one person somewhere finds that option useful

[fast pagoda]

users: 0

wait what about the person that submitted it, you're using it right?

nah

kek

all i know is i'm glad i didnt use bcachefs

[tender river]

imo, no reason to use bcachefs unless you have multiple drives

i use it on my server in a 3x2tb array and it's been constantly breaking but i havent lost any data... that matters to me at least

by breaking i mean refuse to mount and demand fsck

[fast pagoda]

sounds ntfs-y to me

[tender river]

there have been concerning issues with dir inode sizes becoming nonzero

which is surely not indicative of any corruption

[fast pagoda]

CoW scares me now that i know that it puts random writes all over the place when new changes are made to data

seems like a recipe for degredation to me but i can't really say why i feel that way

i know it can degrade write performance

[tender river]

but it gets you so much on non-db workloads, cant complain

also https://github.com/Zygo/bees

[fast pagoda]

yeah i mean btrfw is CoW so it's fine i just have to remember to kill that folder's CoW permission if im using it for like a VM and im happy

[fierce ether]

1638 tokens bot

[fast pagoda]

is that meant to supplement the built in btrfs balance/scrubs

because i already have those run on cron

well not cron it's systemd scheduler whatever that's called

[tender river]

btrfs balance/scrub dont do any deduplication

[fast pagoda]

they just move stuff around yeah

wasnt sure

cuz they do be saving "space" but i guess that's just """"occupied""""" but not

[tender river]

do they really save space afaik its mostly for multi device filesystems

[opaque sigil]

spread out writes are good

wasn't it just for error checking

[fast pagoda]

yes they do

[opaque sigil]

and i guess defrag

[tender river]

scrub is for error checking

[fast pagoda]

go run a balance

[tender river]

balance is for tree rebalance

theres a separate defragment command i think?

[fast pagoda]

pre scrub (im doing science on my system drive rn) gonna save the reported storage before and after scrub then before after balance

[opaque sigil]

that sounds like the opposite of defragmentation

[fast pagoda]

it sends them back thru the allocation

[fast pagoda]

and frees unallocated space that might be "allocated"

[tender river]

(about balance)

Extent sharing is preserved and reflinks are not broken. Files are not defragmented nor recompressed, file extents are preserved but the physical location on devices will change.

[fast pagoda]

yeah it just recovers space your system didnt think it could use but actually could really

there IS a btrfs-defrag

truly gargantuan savings from that scrub

balance usually finds more

[tender river]

this much fluctuation cant be attributed to anything at all

theres too much stuff running on modern pcs

[fast pagoda]

well it ran automatically like

6 hours ago

is the problem

i dont know if balance did or not

it didnt have a recent one sitting there so maybe not

[tender river]

arch wiki says

On a single-device filesystem, a balance may be also useful for (temporarily) reducing the amount of allocated but unused (meta)data chunks

sounds useless

[fast pagoda]

actually the only reason i know about this in depth is because it was causing my goddamn system to freeze

for 2mins

until it timed out

because the stupid btrfs allocator was looking for space to put shit in

it didnt want to expand

even tho i had like 600+ gigs free

[tender river]

i used to have btrfs on my laptop but then the fs ran out of metadata space and it just stopped working

[fast pagoda]

so it was running and would totally freeze the system for the full timeout timer

[tender river]

didnt want to do anything

no matter what i tried

i switched rootfs to bcachefs to try it out

its been working fine i guess

[fast pagoda]

this is what i saved back when that was an inssue

20 gigs of allocated lol

could have benefited from the dedup too back then obv

22 gigs of dup metadata kek

SO top tip if you're running btrfs and your system starts freezing for 2 mins, you see btrfs-cleaner in your processes hogging the IO, even if your drive isn't full sometimes btrfs is being stupid

[tender river]

i did report an issue that became the hallmark of "bcachefs still isnt stable" though

[fast pagoda]

in that case i ended up flooding the drive with fake empty data to force it to allocate more

i put like 100 gigs of empty .tmp files in

[fast pagoda]

you just ...switched it

just now?

[tender river]

[fast pagoda]

meant to reply to this

[tender river]

no a long time ago

[fast pagoda]

oh

[tender river]

.

[fast pagoda]

ahh

i thought you just offhand swapped it i was like damn

apparently it's not impossible and perhaps not even unheard of to in-place convert NTFS to ext4 with relatively low chance of issues

i wouldnt risk it tho

[tender river]

theres also btrfs-convert

but its only from ext{2,3,4}

[opaque sigil]

i tried that once and it broke on my laptop

[fast pagoda]

isnt that only ext and fuckin reiser fs

[tender river]

incredible chart

[fast pagoda]

f

[opaque sigil]

it creates a failsafe snapshot thankfully so i just restored that

my laptop shall stay ext4

[tender river]

this exists too

[fierce ether]

1627 tokens

i pray to cesus

(c jesus)

[fast pagoda]

i have no idea why my iotop doesnt read swapin or IO% when i have checked and i DO have those modules enabled in the kernel AND theyre not even deadge

she's balancin

[fast pagoda]

kent overstreet is like linus if he only made a fs and not the linux kernel

[opaque sigil]

kent

[tender river]

kent work in a team

[opaque sigil]

apparently impossible

i think they're now switching to distributing bcachefs via dkms right

[fast pagoda]

yep

man done unmerged himself from the kernel by just refusing to stop trying to push a bit further

seems like the personality type to just take a mile when given an inch until it boils over tbh

[opaque sigil]

oh well, he's clearly better off just developing it outside the kernel

[fast pagoda]

yeah everyone involved will probably be better off

[opaque sigil]

too bad file systems that aren't in the kernel don't exist

[tender river]

dkms is perfectly usable

[fast pagoda]

so it freed about 7 gigs of "allocated" space up

pointless as it will go back to being used anyways

but still it's not 0 if you're having like filesystem full issues i'd imagine

[tender river]

either way nixos maintainers will figure it out i can leech off them

[opaque sigil]

[fast pagoda]

i mean like nvidia drivers are dkms it's not like it's uncommon to use it

zfs is dkms too usually i think?

same with like virtualbox

[tender river]

isnt zfs custom kernel

maybe not

[opaque sigil]

dkms iirc

[fast pagoda]

https://aur.archlinux.org/packages/zfs-dkms

[opaque sigil]

with the option of using a custom kernel

[fast pagoda]

i know cachyos's cachy-sauce (whatever the fuck that means) kernel supports it ootb

[sage crag]

nyum

[fast pagoda]

they officially support ubuntu/slackware (based)/rhel/opensuse/nixos/gentoo/freebsd/fedora/debian/arch/alpine

i want to try slackware on my laptop

[tender river]

manage dependencies like its 1970

[fast pagoda]

http://www.slackware.com/~msimons/slackware/grfx/ this page is so real tho

beautiful

[tender river]

what is this

they should make "sigma powered"

[fast pagoda]

alpha was a cpu arch

DEC alpha

[tender river]

omegaverse linux

uuh

[fast pagoda]

the last hurrah of the great digital equipment corporation

first 64 bit processor mass produced :^)

some ancient texts (comments from the win2k source from people at microsoft working on the DEC alpha support)

 // the fucking alpha cpp compiler seems to fuck up the goddam type "LPITEMIDLIST", so to work
// around the fucking peice of shit compiler we pass the last param as an void *instead of a LPITEMIDLIST

 * !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
* !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
* !!!!!!!IF YOU CHANGE TABS TO SPACES, YOU WILL BE KILLED!!!!!!!
* !!!!!!!!!!!!!!DOING SO FUCKS THE BUILD PROCESS!!!!!!!!!!!!!!!!
* !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
* !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 

private\inet\mshtml\src\core\cdbase\baseprop.cxx:
// HACK! HACK! HACK! (MohanB) In order to fix #64710 at this very late


private\inet\mshtml\src\core\cdutil\genutil.cxx:
// HACK HACK HACK. REMOVE THIS ONCE MARLETT IS AROUND



private\inet\mshtml\src\other\moniker\resprot.cxx:
// <HACK>
goto EndHack;
// </HACK>



private\inet\mshtml\src\site\layout\flowlyt.cxx:
// God, I hate this hack ...



private\inet\wininet\urlcache\cachecfg.cxx:
// Dumb hack for back compat. *sigh*

private\inet\wininet\urlcache\filemgr.cxx:
// ACHTUNG!!! this is a special hack for IBM antivirus software



private\ispu\pkitrust\trustui\acuictl.cpp:
// HACK ALERT, believe it or not there is no way to get the height of the current
// HACK ON TOP OF HACK ALERT,



private\ntos\udfs\devctrl.c:
// Add the hack-o-ramma to fix formats.

private\shell\shdoc401\unicpp\sendto.cpp:
// Mondo hackitude-o-rama.

private\ntos\w32\ntcon\server\link.c:
// HUGE, HUGE hack-o-rama to get NTSD started on this process!

private\ntos\w32\ntuser\client\dlgmgr.c:
// HACK OF DEATH:



private\shell\lib\util.cpp:
// TERRIBLE HORRIBLE NO GOOD VERY BAD HACK



private\ntos\w32\ntuser\client\nt6\user.h:
* The magnitude of this hack compares favorably with that of the national debt.

 private\shell\ext\tweakui\genthunk.c:
* CallProc32W is insane. It's a variadic function that uses
* the pascal calling convention. (It probably makes more sense
* when you're *filtered*.) 

i dunno if that had to be filtered fwiw but i played safe

i think it was you chay that made me look into pascal a while ago because you were like 2 years old writing pascal or something

[tender river]

i just used line drawing functions

and made an infinite loop that prints numbers sequentially

[fast pagoda]

i dont think i would want to emerge or even ask 9999 bees to emerge

[fast pagoda]

prodigy

[tender river]

i got bored after that because i didnt really know what to do with it

[fast pagoda]

the only fun thing i did as a kid that i still get told about is my dad taking me to work (he's a meteorologist for noaa//govt weather forecaster in US) and when i was like 4-5 i got on the ASOS (weather observation system that sends out data continuously) and started sending messages to other offices in the surrounding states saying i was bored

like "hi help im bored"

and he got in so much shit

[tender river]

so did they help

[fast pagoda]

they called

and were like

hey uh

is everything alright there

at the office

getting some strange messages on the asos

i cant remember if it was asos or awips

it mightve been awips

[sick owl]

The mod bot for our uni soc server got into an argument with someone in a ticket

[fast pagoda]

stupid govt acronyms

[sick owl]

Mistral Small 3.2 if you were wondering

[fast pagoda]

neuro at her first job be like

imagine getting roasted by the moderator llm and then it bans you so you cant respond

that's tough

[fast pagoda]

okay how is bees --help a straight up "bees not found" but bees with nothing tries to run something but hangs

fish does NOT agree

[glad path]

* !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
* !!!!!!!IF YOU CHANGE TABS TO SPACES, YOU WILL BE KILLED!!!!!!!
* !!!!!!!!!!!!!!DOING SO FUCKS THE BUILD PROCESS!!!!!!!!!!!!!!!!
* !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
* !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! ```
might be my favorite one

[fast pagoda]

yeah this one is great

just threatening mortal punishment to your coworkers

[glad path]

someone tell that to freaking vscode

[fast pagoda]

classic issue as well

[glad path]

like it cannot understand to use tab indents

so annoying

[fast pagoda]

you can configure it

in vsc

[glad path]

it literalyl doesn't work

im not even joking it just still tries to use space indentation half the time at random

[fast pagoda]

override conflict or user/workspace/etc conflict maybe?

[glad path]

it is set to use tab in user
but for EVERY WORKSPACE it overrides it by default

or something like that

[opaque sigil]

[fast pagoda]

shouldn't user override unless it's explicitely set in a new workspace

that's weird

i blame steve jobs

[glad path]

I literally have to turn off detect indentation every time I think

[fast pagoda]

reee @tender river your bees require me to manually export a home dir then calculate the appropriate size of hash table to give this thing all because im too stubborn to use a helper script they provided since i didnt realize it was there till im into this

this is your fault

[fast pagoda]

hmmm

i wonder if there's some config file somewhere you can edit to stop that

probably

idk how dotfiles and stuff work for macos though

i must say

the fish autocompletion

is leagues ahead of zsh

[tender river]

[fast pagoda]

not even close

it just works

[tender river]

i like it for git too

git add would be considerably more annoying without it

[fast pagoda]

not me going git add . then git rm <bullshit i dont want to commit actually>

[tender river]

ye thats what i dont have to do

[fast pagoda]

and my all time favourite

git rm -rf --cached .

think it's --cached

[tender river]

yup --cached

[tender river]

i just use 256mb per tb i think

[fast pagoda]

it said to use a multiple of 128k for sure and then 128MB/tb is bad

err below that

this does it by volume i assume?

[tender river]

mhm run a separate instance per volume

[fast pagoda]

so a 256mb table for a 1tb drive, 1tb table for a 4 tb drive prob

[tender river]

1gb*

[fast pagoda]

no no i want the ful drive to be hash

lol

yeah 1gb

does it need a separate table per volume tho

or would they all just use a single 2gb hash table if my largest volume is 8gb

fk

8tb

[tender river]

1tb table for 4pb storage

[fast pagoda]

oh

[tender river]

[fast pagoda]

yeah

i got that

i am

google

just gotta go emerge some bees on my ironwood TPU cluster

[tender river]

separate tables

it stores it in drive root i think maybe

[fast pagoda]

fug so instead of beeshash.dat i should be naming these something not stupid to keep track of

i slapped it in my usr/bin lmao maybe i shouldn't

[tender river]

ye stores in /.beeshome

for me at least idk i use nixos

[fast pagoda]

usr/bin is better than my first attempt

which of course was blind copy paste

[tender river]

ah yeah BEESHOME is <drive>/.beeshome on nixos by default

or is it

[fast pagoda]

what about the whole /nix/store shit

[tender river]

ExecStart=bees-service-wrapper run 'UUID=fa754b1e-ac83-4851-bf16-88efcd40b657' 'verbosity=6' 'idxSizeMB=256' 'workDir=.beeshome' -- --no-timestamps --loadavg-target 8.0

thats how it starts for me

[fast pagoda]

arch packages are usually in /usr/bin but there's also /bin and /sbin

[tender river]

BEESHOME is not about path to the bees binary though

[fast pagoda]

beeshome is just the table

[tender river]

[fast pagoda]

i think

[tender river]

no sense to have it in /usr/bin

[tender river]

  bees_home="${bees_home:-$bees_mntDir/$bees_workDir}"

found the relevant line from bees-service-wrapper

[fast pagoda]

the arch PKGBUILD

prepare() {
  cd $pkgname-$pkgver
# Honor system build flags
  sed -e "s|^CCFLAGS =.*|CCFLAGS = $CFLAGS|" -i makeflags
}

build() {
  cd "${srcdir}/${pkgname}-${pkgver}"
  make BEES_VERSION="${pkgver}"
}

package() {
  cd "${srcdir}/${pkgname}-${pkgver}"

  make install DESTDIR="${pkgdir}" BEES_VERSION="${pkgver}"

  mv "${pkgdir}/usr/sbin" "${pkgdir}/usr/bin"
}

i would've been on a not horrible track if it were the binary

[tender river]

i dont understand the issue

[fast pagoda]

nah no issue im just yapping

efficiency is putting sudo before every one of these even tho sudo would've likely kept going

[tender river]

it wouldnt have

[fast pagoda]

see i knew i was smart

what bout sudo -v doe

[tender river]

nope

[fast pagoda]

ionno sudo -v ran without prompting a password right after

[tender river]

yeah but sudo only prompted for password once too

[opaque sigil]

You still need sudo

[tender river]

its ran as (sudo a) && (sudo b) && (sudo c)

each one between && is a separate command

[fast pagoda]

oh yeah i still need to TYPE sudo i just dont need another password

i forgor

[tender river]

sudo doesnt do anything to your session it just elevates a single command

[opaque sigil]

Do I care enough about fish to try and convert all my zsh specific stuff

[fast pagoda]

i have been having an aight time with it outside of having to fix bash scripts

[opaque sigil]

Actually

[fast pagoda]

ive been swapping back and forth without issues

[tender river]

you could've ran something like sudo bash -c 'truncate -s ........ && truncate -s .....'

[opaque sigil]

Does nix have a maintained package for zsh completions one could use

[tender river]

just programs.zsh.enable

it will then link /share/zsh-completions or whatever it is to your systemp ath

or do you mean a plugin for zsh itself

[opaque sigil]

Wait you're right I'm dumb

I guess there should be nothing to do then assuming programs actual ship their completions

[fast pagoda]

when i was on wsl

i used to use

sudo -i

every fucking shell session

every one

[opaque sigil]

I have this cursed script that does user -> root -> user for running stuff through my vpn and I got so tired of fighting with stuff it now needs sudo -E

[tender river]

i usually login as root to my server/router but only because whenever i have to do maintenance its always systemd or config file tweaking and it makes no sense to sudo for every single command

[sage crag]

https://cdn.discordapp.com/emojis/1276149117717970964.gif?quality=lossless&name=nopy&size=64

[tender river]

🐛

[fast pagoda]

oh yeah i acquired the TIDE

the defaults from garuda weren't bad tho

[tender river]

fun fact nopy was featured by google

[fast pagoda]

corpa

[opaque sigil]

Are you still trying to setup bees

[fast pagoda]

If you don't set BEESHOME, the path ".beeshome" will be used relative to the root subvol of the filesystem. For example:

yeah i got distracted fuckin with fish completions

when you talked about it

well

zsh

fish_config

i should use this prompt

[opaque sigil]

they want 4gb per tb for a 4k block hash table hmm

always satisfying

[tender river]

is this supposed to be readable

[fast pagoda]

not after you look at it once

[tender river]

who made this theme

[fast pagoda]

since you're now blind

[fast pagoda]

i'' just check this calculated css and OH MY GOODNESS

https://tenor.com/view/rainbow-plankton-under-the-sea-scared-shocked-plankton-gif-15504455

[sage crag]

https://pypp-docs.readthedocs.io/en/v1.0.1-alpha/

its not python, its not c++, let's call it py++

why though

this is useless?

[fast pagoda]

when u finally say fuck it and make the entire language a macro

[opaque sigil]

interesting, i wanted to try using the stable branch for my nixos base and then have some unstable packages but the nvidia kernel modules failed to build

[sage crag]

nrm

[fast pagoda]

https://www.tomshardware.com/video-games/pc-gaming/check-that-youre-ready-for-invasive-drm-with-new-feature-built-right-into-steam-secure-boot-and-tpm-status-now-listed-right-in-the-app

Every time I see stuff mentioning this goofy stuff i cringe

we need our EA JAVELIN SUPER ANTI HACKER TO RUN IN YOUR BOOTLOADER because... We we just do okay!!! You might get knifed by a cheater!!!!

Nope no thanks

[olive sable]

Goodmorning

I fell asleep at 4pm so now im a creature of the night again

Ill eat dinner and start working on my blender homework i guess

[amber fractal]

Who are we putting in the blender

[opaque sigil]

you slept for 11 hours?

[unique plinth]

Time zone?

[opaque sigil]

cest

it is currently 03:50

[unique plinth]

FOR BOTH OF YOU? (I presume we're talking about Sam)(I walked in on this conversation.)

[opaque sigil]

yes we're in the same time zone

[unique plinth]

that is for a long time,

Come on! get up do stuff Sam, there will be health detraments from this!

wait, is it 3:50am or pm, is this why I'm not getting replies for stuff?

(I'm in MDT)

[opaque sigil]

am

[amber fractal]

I would say classic chatters up at early times, but yeah I'm waking up at 3AM in my local time.

[fierce ether]

i played my bot in chess.com and it didnt make any blunders

just mistakes

[unique plinth]

You chatters need to get some sleep.

[glad path]

its not that late

[unique plinth]

(Early)

I'm yawning and it's 8:02 PM for me

[amber fractal]

I got my 8 hours, might have less tonight due to housework

[unique plinth]

alright, as long as your getting proper sleep.

but I hope the rest of you can say that too

[opaque sigil]

i always make sure to get my 8 hours

otherwise i can't function

[unique plinth]

https://tenor.com/view/palpatine-good-gewd-star-wars-episode-3-gif-15386400457909814305

https://tenor.com/view/palpatine-good-gewd-star-wars-episode-3-gif-15386400457909814305

[amber fractal]

No perms

https://tenor.com/view/palpatine-good-gewd-star-wars-episode-3-gif-15386400457909814305

[unique plinth]

how?

so confused

I shouldn't question it should I?

[opaque sigil]

you need to talk more to get embed perms via a role

[amber fractal]

pinged over in the bot channel

[unique plinth]

ah, I see thank you.

[glad path]

https://cdn.discordapp.com/attachments/1414311457574752296/1415716823936012339/NeuroCord_Member_Gen.gif

read this

[unique plinth]

stood for permissions.

[glad path]

^^

[olive sable]

i guess

[unique plinth]

off to discussion I go

[olive sable]

my watch says i wasnt actually asleep till about 5 pm

[glad path]

till about pm?

[olive sable]

so 9 hours and 40 min

[glad path]

?????

what am i misisng

oh

5pm