DDraceNetwork - developer

02:39

wuh

02:39

how fix

02:45

ok it fixed itself

02:45

idk what was making it do that

02:46

there was nothing cargo running

Ewan

i figured it'd just take a blob of data and decode like sane people would make it do

You can do that

05:36

The graphics what take raw pointers in an overload

Ewan

i think it would be wise to bake the default skin into the game

No. If we don't aim for .exe only, then there will be other resources and dlls anyway. The data dir should be seen as read only anyway

05:38

You just increase exe file with that without any return

06:06

no, this is the only overload

06:06

i guess it's not an overload if it's the only one lol

Bro then a different function

06:31

Don't parse every conversation like a bot

06:32

I bet the name is 90% the same

beep bop

Jupstar ✪

Don't parse every conversation like a bot

hey, please be less insulting

heinrich5991

hey, please be less insulting

Ok seriously. Insult?

08:05

Heinrich sometimes i have the feeling u abuse ur powers here.

08:05

I kindly asked him to not parse every conversation like a computer

discuss moderation in DM

08:05

not kindly though

ChillerDragon: you need to rejoin your ws-client btw

@Ewan apparently the function you're looking for is called LoadTextureRaw

oh nice

08:32

sry i didn't mean to be so literal. i thought you meant there would be an actual overload of the function like there often is

08:33

a real overload is much easier to find than a similarly behaving function of a different name

I don't like overloads, especially not for different functionality like here

heinrich5991

discuss moderation in DM

with who?

08:39

everybody is afraid of u anyway

08:40

you have the power to remove others from staff

Ryozuki

everybody is afraid of u anyway

wtf are you talking about? You can talk to any admin in DM if you have complaints

i already complained several times

08:41

for example to learath

08:41

nothing is done anyway

I don't think Learath2 is afraid of me

08:41

if nothing is done, another explanation is that it was deemed to be no issue

its more like he rather not go against you i guess

08:43

@deen i think you will always side with heinrich so i never seen u as an option to talk to

08:43

i only considered learath but he kinda half ignores me

08:43

idk about murpi

08:43

he is newer admin so i think he would also not go against heinrich, even if my complains are valid

08:45

my complain is actually rly simple, just make heinrich be less stringent in this chat, i think he is too much and often kills good conversations, people use remarks, yeah you may not always be 100% the most polite, but humans interact this way, like jupstar case right now

08:46

i also dont think we act on bad faith when we are like this

08:47

and actually his really strict moderation brings bad vibes, and confrontation

08:47

its a repeated issue if u have seen, idk

08:48

thanks for coming to my ted talk

08:48

u can now ban me for breaking the rule of dicussing mod here

I also don't think you act in bad faith. I still think that people should change their ways to become more accepting to newcomers and people who make mistakes. don't ridicule them, don't tell people they're stupid

I think it's just a cultural thing, and some people are used to a different kind of conversation culture. But since you don't know who you're talking to here, it's better to err on the side of caution. So just be nice to others, seems easy enough.

thats the thing, to me, i never seen as not being that nice

08:52

i guess its truly cultural then

yeah, but it can seem that way from the outside, or to people who don't know you well

too bad jupstar is timeouted, from german culture, and cant give his points here

he was timed out for 5 minutes

08:55

i.e. he's not timed out anymore

i still think its too strict anyway

08:57

does ddnet have a mail provider?

08:57

gandhi decided to not give mailboxes anyway

08:57

and i use it for wiki

I think ddnet.org can receive mails

We also need to inform you of the end of the GandiMail service included with your domain (please note that the GandiMail service will end on November 30th, 2023). You can seamlessly convert to our new Mailbox service. In addition, we will introduce updates to our pricing policy effective July 13th 2023 for all our existing services and the new Mailbox service (EUR USD TWD ).

08:57

i need to send em

08:58

well i need mail for myself so ill search some

09:00

i wish i could just host it on my vps

09:01

the thing that sucks tho is that cloudflare doesnt protect mail anyway so i need it on a separate vps, even if it worked

i wasn't trying to pick apart your sentence jupstar i just want u to know

@Ewan isnt it 3 am there

Yea

09:16

I was playing halo

09:16

With a British buddy

Ryozuki

nothing is done anyway

What did you expect? Last time you complained I found it reasonable, I came here, we had a chat about it. You not liking how he wants to moderate the discord isn't like abuse for example

i cant expect anything since heinrich wants it like this

We aren't going to pull out the guillotine because you want a more casual discord environment while he wants a more wholesom (and imo stiff) one

Learath2

We aren't going to pull out the guillotine because you want a more casual discord environment while he wants a more wholesom (and imo stiff) one

Fwiw it feels stiff (and a little fake) to me because in my native language, especially between friends swearing is extremely common and sterile language is extremely uncommon. I also grew up in the unhinged era of the internet, so even the nice communities I remember had almost no rules relating to conduct

1

09:39

But it's not 2010 people don't come equipped with thick skin anymore, not every new person feels like they are our friends and not every language has people swearing like sailors

we are not swearing like sailors anyway

Nah but you can be very mean, even unintentionally

im passionate, and also i get frustrated, lack of change, things go the way one wants even if we discuss things

09:57

im stupid for caring too much about a kirby game

it happens

btw: https://github.com/ddnet/ddnet/issues/6753 discusses asset handling, I think others also had an opinion on that

Assets handling · Issue #6753 · ddnet/ddnet

#6724 moving around images has brought up a longstanding issue with futureproofing assets. As with all issues DDNet there is no way to make everyone happy. However, I think @heinrich5991's idea...

what, the embedded stuff?

10:10

Oh no the other stuff

10:10

Ok

waiting like a kid for the package to arrive

12:14

12:18

@Learath2 what is the bal thingy?

12:18

i only see "balanced output" references

12:19

i think its for long cables

Balanced output, each channel gets it's own ground

12:23

and usually more power

12:24

(the headphones you got don't have balanced input, so not important)

i see

whats that bass knob do

12:43

nvm google is a good tool

12:43

12:43

mr amazon hasnt arrived yet

12:44

it's just a preset eq, I don't like the one on the e10k so I'd guess the one on the k3 isn't much better

is software eq equivalent?

12:45

can i get the high khz on linux

12:45

i guess yes

Ryozuki

is software eq equivalent?

yes, digital signal processing before the dac might even be more accurate

nice

Ryozuki

can i get the high khz on linux

no idea, but who cares, you'll probably never have sources that high quality

are there no subs doing that? xd

you should be able to UAC2 is a standard

Ryozuki

are there no subs doing that? xd

subs?

@Learath2 in ddnet u can say the hz

12:48

anime subs

gotta hear the high quality oni chan

12:48

thought bro meant subwoofers for a sec

i prefer 8kb

8bit

my waifu better sound like an xbox cod player

lolno, 384 kBaud * 32 bit is 12kbps audio, the best bluray encode I can find for an actual movie is about 5kbps audio

f

12:54

good thing is its usb-c i can connect it to the mac and have actual aux ports

12:54

the dac

12:54

it powers itself with the usb-c?

384 kHz is actually way too high given the range of sounds humans hear

Ryozuki

it powers itself with the usb-c?

y

i wondewr if my phone can drive the headset xd

Learath2

384 kHz is actually way too high given the range of sounds humans hear

ryoz is a dog

Ryozuki

i wondewr if my phone can drive the headset xd

easily

rly?

12:56

Sound Loudspeaker Yes, with stereo speakers 3.5mm jack Yes 24-bit/192kHz audio

12:57

redmi note 11 pro

Voxel

ryoz is a dog

384kHz is enough to reproduce up to 192kHz, the range is so wide it covers every living species

Ryozuki

Sound Loudspeaker Yes, with stereo speakers 3.5mm jack Yes 24-bit/192kHz audio

An 60 ohm headphone is easily driven by any device nowadays, I find mine a little quiet on the phone but it still works just fine

i think the 599 is 50ohm

less ohms, the easier to drive

Frequency response (speaker) 12 Hz to 38,500 Hz Impedance 50 Ω Sound pressure level (SPL) 106dB SPL (1kHz/1Vrms) Total harmonic distortion (THD) < 0.1% (1kHz, 100dBSPL)

12:59

im afraid of carrying such expensive thing on the street anyways

12:59

xd

12:59

i just use my jbl wireless

Open backs are not enjoyable outside anyway

yeah

i think you should build your own speaker like diy perks

I guess 384kHz isn't quite enough if you need to generate bat calls

build ur own microphone

https://blog.rust-lang.org/2023/06/20/introducing-leadership-council.html

Introducing the Rust Leadership Council | Rust Blog

Empowering everyone to build reliable and efficient software.

why does this entire preview sound so demeaning

https://marabos.nl/atomics/

Rust Atomics and Locks by Mara Bos

Low-level Concurrency in Practice. This practical book helps Rust programmers of all levels gain a clear understanding of low-level concurrency. You'll learn everything about atomics and …

13:22

i need to read this

13:22

Mara Bos has the best book of Concurrency (even better than the Concurrency in action with C++) https://marabos.nl/atomics/

13:25

Jupstar ✪

here sneakpeak lol

fancy comparison :), I wonder where twgpu spends the most time. I assume it runs on the vulkan backend?

13:51

(however I try to ignore performance for now. its hard to make my brain not think about performance stuff all the time tho)

Patiga

fancy comparison :), I wonder where twgpu spends the most time. I assume it runs on the vulkan backend?

I dunno. But twgpu seems to load stuff before the window is created sadly. So I assume that's also the biggest bottleneck. Everything before the GPU is used

and I think I know now how I can approach mipmaps for the atlases. mipmaps should come after the rest of the gameplay rendering stuff though. the last 1-2 weeks I actually had 0 time for twgpu, buuuuuut surely the next few weeks will be less stressful :)

I think I was already "too fair" removing a tiny framespan for twgpu since window creating takes around 100ms

1

ah you are counting the time from window creation to a displayed image?

Patiga

ah you are counting the time from window creation to a displayed image?

Well not for twgpu. But i tested a small map like ctf where it was instant

13:54

So I assume whatever it does counts towards map loading

13:55

( the time spent before window creation)

is ddnet-pg ur thing?

yea window creation is the first thing it does, though that should be changed I guess. At adapter creation you can pass a compatible_surface parameter to ensure you can render on the window. I should instead pass None there and just quickly check manually later

Patiga

yea window creation is the first thing it does, though that should be changed I guess. At adapter creation you can pass a compatible_surface parameter to ensure you can render on the window. I should instead pass None there and just quickly check manually later

U sure about this?

which part of it :p

If the window is the very first thing created

Ryozuki

is ddnet-pg ur thing?

It's mine

13:56

Twgpu is his

right after device creation I think

Patiga

right after device creation I think

But some part of the map is loaded before the window is created?

13:57

That's what I experienced at least

ah the map gets loaded on the cpu, and basically right after the window is created https://gitlab.com/Patiga/twgpu/-/blob/8bb9c111af10cf85da90ab04ea856530fc2fbf35/twgpu-tools/src/bin/twgpu-map-inspect.rs#L45

Yeah

13:58

That makes sense

@Jupstar ✪ it sounds like ddnet-postgresql

And from that fact i removed like 200 Ms to compensate the loss while GPU is busy creating the surface

Ryozuki

@Jupstar ✪ it sounds like ddnet-postgresql

Deen said same

13:59

It's simply a shortcut for playground

13:59

BCS i am 5 and play with learath who is 12 there

13:59

The whole day

aaah I enjoy the comparison and the implied performance questions, but I gotta go back to work xd

14:01

I just gotta repeat

14:01

Patiga, you shouldn't care about twgpu performance right now

14:01

Patiga, you shouldn't care about twgpu performance right now

14:01

Patiga, you shouldn't care about twgpu performance right now

Xd

14:01

Well i can tell you something before u leave

14:01

Since I enabled benchmarks in my version

14:02

Loading the internal images from the zlib compression is the slowest part for me

14:02

Takes around 400ms

14:02

If it would be png, it would probably be double as fast

heh, yea I always hated that about map loading. its like >95% of the cpu time

Jupstar ✪

Takes around 400ms

those are part of the 600ms in your comparison?

Patiga

those are part of the 600ms in your comparison?

Yes

:o

It also includes connecting to a local server

14:04

Using quic

1

epic

14:04

quic

14:04

get it

14:04

@Jupstar ✪ when ready

Almost xd

i wanna play at 40k fps

Ryozuki

@Jupstar ✪ when ready

Well my goal is mainly to get to vanilla support. I'm not sure if I'm the best person to implement ddrace

okay I leave now, and this time for real :p

Build on vanilla, i want to create an fng mod

Patiga

okay I leave now, and this time for real :p

Cya

1

Ryozuki

https://blog.rust-lang.org/2023/06/20/introducing-leadership-council.html

thanks for the link, let's see how that turns out

heinrich5991

thanks for the link, let's see how that turns out

interesting bit from the rfc is, mod team or council team can decide to disband both at same time

14:31

As an absolute last resort, either the Council or the moderation team may choose to simultaneously dissolve both teams. Teams then select new representatives, and the contigent moderators become the interim moderation team and select successors.

def-

[ddnet-web:master] 1 new commit

328e298 Add 10 € donation by Dreamless aka thekid36 - def-

i got the packet, setup time

Ryozuki

interesting bit from the rfc is, mod team or council team can decide to disband both at same time

that's very interesting

15:05

this would have helped that one other time

15:05

when "the core team is unaccoutable"

Ryozuki

https://marabos.nl/atomics/

also interesting. not sure if I'll find the time to read it

Voxel

why does this entire preview sound so demeaning

wdym?

@Learath2 i got it workin

it rly sounds gooood

15:28

i guess its normal the dac heats up a bit xd

15:29

and these headphones are rly comfy

15:29

the previous ones i had rly put pressure

15:29

but they also passive closed sound

15:29

now i hear my keyboard lmao

heinrich5991

wdym?

the leadership council empowering everyone

Voxel

the leadership council empowering everyone

Empowering everyone to build reliable and efficient software

15:36

refers to the rust lang itself

15:36

its the motto

15:36

https://www.rust-lang.org/

A language empowering everyone to build reliable and efficient software.

15:37

its not specific to the article

yo my beautiful amigos, anyone know how setting CNetObj_PlayerInput works? It never seems to work properly for me and switches to some random weapon

1

what are you doing?

fooling around (edited)

15:46

trying to get weapon selection working on demand

15:47

Just getting this interface to work:

this kind of thing isn't allowed on ddnet servers :/ (edited)

i dont play on ddnet servers

15:47

for block

i dont think its allowed on any server besides lan

oh well then guess ill figure it out

15:48

just a shame a lot of code isnt documented well so i have to go on a rabbit chase

15:48

and find how it works by example

Jupeyy

[discord-skin-upload-bot:main] 1 new commit

9785596 Make the impl fail safe against discord character count limit - Jupeyy

oke figured it out i think, for some reason it doesnt use the weapons enums

BillionRay

just a shame a lot of code isnt documented well so i have to go on a rabbit chase

it would be a diferent story if u werent making a bot

15:56

if u were contributing a feature to ddnet u would get help

i may contribute in the future

15:56

but i dont really know the codebase well enough as of right now

u started the wrong path

it's fine, as long as you don't publish your bot or use it where it's not allowed

its kind of ironic to try to get help from here to do a bot, where we get the pains of defending against them

but we don't really like helping to create bots

i have some blog posts, they arent aimed at botting though, but u may learn something https://edgarluque.com/categories/ddracenetwork/

Home

My personal website.

fair enough, it is a shame though cause there is some stuff that isn't obvious in the source code without going down a rabbithole, like that array of size 200 with the predicted characters

Ryozuki

i have some blog posts, they arent aimed at botting though, but u may learn something https://edgarluque.com/categories/ddracenetwork/

nice, thanks :D

16:00

another botter helped !

BillionRay

fair enough, it is a shame though cause there is some stuff that isn't obvious in the source code without going down a rabbithole, like that array of size 200 with the predicted characters

i think it is obvious if u know more c++

16:00

do u know c++ at all?

yes obviously

by answering that u dont know c++

16:00

cuz nobody knows c++

16:00

in my opinion

16:01

trap question kek

well sure, but I know it well enough to be productive

16:01

i assume thats what you meant

hmm idk about

16:01

i remember u asking weird questions to learath

16:01

or maybe it was someone else

hmm

16:01

i did ask something a few months ago but im not sure what it was anymore

16:02

this?

idk

16:03

dont bot xd

no promises there

BillionRay

no promises there

i mean if you wanna not be able to play ddnet ever again thats on you

ok, if you can prove that i actually use my client unfairly on ddnet servers or released that client to the public

türk varmı

#off-topic

Hey @deen! I was messing around with the HTTP master server yesterday, and sending it "invalid" data. I set one of my clients name to the type null which I think broke your extract-master.py script as playtime for 2023-06-19 is 0 hours for everyone ;D sorry!

furo: the http master server info field is free-form json

17:44

I guess the extract-master.py script should be fixed

furo

Hey @deen! I was messing around with the HTTP master server yesterday, and sending it "invalid" data. I set one of my clients name to the type null which I think broke your extract-master.py script as playtime for 2023-06-19 is 0 hours for everyone ;D sorry!

Power saving day, wdym! Nobody played

furo: what are you interested in? :)

17:49

the client ignores what it considers malformed server info

yeah, I noticed that after some testing. I was mainly pointing out that it breaks that script and that it should be fixed.

looking for security vulnerabilities?

ChillerDragon

[ddnet/ddnet] Issue opened: #6754 Move players to new server on map vote

As discussed on irc @heinrich5991 suggested to move all players that voted yes on a map vote to a new server. This fixes the issue of players in the race being interrupted by map votes. And also solves the "coordination problem" at the same time. Which is moving all players together to the new map. Instead of them manually going together to a new server. Depends on https://github.com/ddnet/ddnet/issues/5662

I guess, just playing around wondering what would happen if I messed with the data. Conclusion? Just as you said, the client just ignores the server.

the code for that is here, btw: https://github.com/ddnet/ddnet/blob/e22e0c9777f1814e84fbe76cfdf3644676c6dc7f/src/engine/shared/serverinfo.cpp#L59-L178

would be awesome if u fix urself what u find

18:10

we should fuzz the script

what does that mean

a fuzzer?

18:11

A fuzzer is a program which injects automatically semi-random data into a program/stack and detect bugs. The data-generation part is made of generators, and vulnerability identification relies on debugging tools.

oh LOL

https://en.wikipedia.org/wiki/Fuzzing

Fuzzing

In programming and software development, fuzzing or fuzz testing is an automated software testing technique that involves providing invalid, unexpected, or random data as inputs to a computer program. The program is then monitored for exceptions such as crashes, failing built-in code assertions, or potential memory leaks. Typically, fuzzers are ...

18:12

An effective fuzzer generates semi-valid inputs that are "valid enough" in that they are not directly rejected by the parser, but do create unexpected behaviors deeper in the program and are "invalid enough" to expose corner cases that have not been properly dealt with.

18:12

key part

furo: He, thanks for reporting

i read its better to first max out volume in software and then the hardware knob, cuz it compresses less (edited)

18:33

is it true?

18:33

god i cant type

yes

ok nice to know

at least 100%

cuz before i used to have software at min

don't scale it up more than that

Robyt3

[ddnet/ddnet] Pull request opened: #6755 Remove individual RGBA sliders from editor color picker

Remove the individual RGBA sliders for editor color pickers and only show one button that opens the color picker popup instead. Decrease size of layer and point popups that previously had color properties which need less space now. Screenshots:

Before:

!screenshot_2023-06-20_20-39-43

After:

![screenshot_2023-06-20_20-40-00](https://github.com/ddnet/ddnet/assets/23437060/517f90cf-06a5-4332-9eb8...

GitHub

Click to see attachment 🖼️

@Ravie

@Robyt3 perfect

#how to go to part n parctice

19:23

got it

19:23

t/tp in spec

Ravie

@Robyt3 perfect

shouldnt we aim for usability and not for design for an editor?

19:40

sure it looks better inlined

Voxel

we could embed the useless images no one mods like the f1 and f2 decor stuff

i've seen people with modded f1 console though

but a html tag without extra click sounds useful

19:41

or "more usable"

19:41

color tag

We could allow shift+left-clicking the color picker button to immediately paste a color in RGB or RGBA hex format from the clipboard. And use shift+right-click to copy the current color to the clipboard. So it basically works like recipes in Factorio.

Jupstar ✪

shouldnt we aim for usability and not for design for an editor?

Robyt's previous PR improved the usability of the color picker and made the RGBA kinda useless

i mean that sounds useful too

you can now enter exact values as HSVA which is more user friendly

Ravie

Robyt's previous PR improved the usability of the color picker and made the RGBA kinda useless

but couldnt we inline the whole color picker

in the images it looks like we have plenty of space

19:48

but color to button and use the full color picker

tile layers also have color properties

19:48

yeah that dialog is a mess but with this it's already an improvement

19:50

maybe the layer shifting could actually go to the Tools tab? from my own experience it's pretty rarely used

1

Ryozuki

i think it is obvious if u know more c++

i kinda disagree, ddnet code base is kinda hard to get because some stuff are mixed

1

not only that, some stuff just isnt obvious without "Find in files" so you need to figure out by example or by the definition, and often times that process is recursive so its easier to ask one of you fellas (edited)

I also don't get why some data is duplicated in different structs, especially for the network part

Ryo u can use a high sample rate but you probably won't want to because the average person can't hear anything higher than 20,000Hz

20:19

44.1kHz and 48kHz are both already well above that

20:19

if u use 192 it will chew ur cpu

20:20

Sample depth is much more important than sample rate

there are more artifacts in 48kHz than 192kHz due to aliasing (edited)

Yea

20:21

Also consider that 99% of your input streams will be 44.1 or 48

20:21

so your cpu has to resample everything

20:22

which can introduce more artifacts in and of itself

Robyt3

We could allow shift+left-clicking the color picker button to immediately paste a color in RGB or RGBA hex format from the clipboard. And use shift+right-click to copy the current color to the clipboard. So it basically works like recipes in Factorio.

hmm

20:22

adding more keyboard and mouse* shortcuts makes it harder for new mappers (edited)

20:23

why not 2 small buttons below the colorpicker to paste/copy

Ewan

so your cpu has to resample everything

Pulseaudio would constantly crash for me on 96kHz but it was fine on 48

20:24

maybe fixed in pipewire & with a better cpu but idk

Ravie

maybe the layer shifting could actually go to the Tools tab? from my own experience it's pretty rarely used

to clean up more, maybe make Detail into a checkbox instead, and possibly hide Color TO unless Color Env is set to an envelope

20:27

and honestly the Game tiles tool promotes bad mapping xd

20:27

so maybe move it to tools as well

20:27

Name: should logically be at the top of the menu as well

Ewan

so your cpu has to resample everything

that's no problem as 192kHz is a multiple of 48kHz (not of 44.1 though)

21:14

and it's not resampling, more like interpolation

The editor code for these properties is rather rigid, so we can't even hide individual controls right now. That's why for the game layer the label for the property "Detail" is shown but the Yes/No buttons are hidden. (edited)

21:19

I added the shift+rightclick/leftclick hotkeys as described. The tooltip should hopefully be enough so mappers figure out that these hotkeys exist if they need them.

Chairn

that's no problem as 192kHz is a multiple of 48kHz (not of 44.1 though)

most media is 44.1

22:15

and from 44.1 -> 192 the interpolation is much more expensive than from 44.1 -> 48 (edited)

22:18

but it kind of depends on which resampling algorithm you choose and pulse/pipewire come with a lot of them already implemented

Chairn

and it's not resampling, more like interpolation

also i don't think this is technically any more accurate. yes, interpolation is usually a part of the resampling process when not converting between factors

22:20

but the process is still resampling

deen

furo: He, thanks for reporting

Fixed now