DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.org/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet

Between 2024-05-07 00:00:00Z and 2024-05-08 00:00:00Z

srjc6 or smth

00:00

i don’t remember

@heinrich5991 https://github.com/sjrc6

00:43

i was 1 letter off

Ryozuki

Click to see attachment 🖼️

There is actually a real, similar law in China about databases though. They have to support specific (Chinese) encryption schemes and have a minimum performance in a benchmark to be allowed to be sold in China

[ddnet/ddnet] Pull request closed: #8323 Fix smoke particles sometimes being stuck inside solid tiles

01:39

github-merge-queue[bot]

[ddnet:master] 2 new commits

6cbc85f Fix smoke particles sometimes being stuck inside solid tiles - Robyt3 ae2c9e2 Merge pull request #8323 from Robyt3/Client-Particles-Smoke-Nudge - def-

<ChillerDragon> Any reason why dbg_assert does not have vargs for the string parameter? I want to print values in the assert to see what went wrong. But I do not want to call str_format every time.

Tater

Could the ddnet license be changed to prevent the distribution of open source cheats? Obviously wouldn't apply to past versions but if it's allowed by the current license it would be nice

I don't think a mere license would magically stop anyone from distributing cheats. You'd have to fight in court and everyone knows that no one is gonna do.

Ryozuki

Click to see attachment 🖼️

why haven't I heard of that

monkaS

Image attachment

Tater

monkaS

maybe that's why I haven't heard of it, anyone who knows are all dead (edited)

surely they meant prosecuted?

that's probably just a meme

02:52

that's surely a meme actually. cuz websites doesn't exists in china anymore

what?

02:52

only apps?

it's all apps and miniapps now yep

yep

miniapps is so weird

02:53

they reinvented a web browser

ye

02:54

it's all about limitations, regulations and selling their existing monopolized userbase of the entire nation to others

02:54

if it is just a web browser it would be too open

02:54

catxplosion

02:55

in wechat miniapps you have to apply for notifications and user have to agree to receive each notification once and only once.

I guess it makes sense if you have an app with 500 million users to try to leverage your userbase for as much as possible

02:55

replace the entire OS with your app xd

TsFreddie

in wechat miniapps you have to apply for notifications and user have to agree to receive each notification once and only once.

honestly that's a pretty good idea to allow delivery notice but not notification spams

02:57

but in practice, it is really annoying to keep tapping allow and reject everytime you order something, and you have to be careful what to allow cuz it is usually not that clear which notifications is for what

do any of them have mini-miniapps?

you probably can't do mini-miniapps

thats what apple said about mini-apps the first time tho right?

does apple have miniapps now?

idk

i think miniapps are kickstarted by android, but wechat just abused them to hell. no idea what it is like on iOS, but on android miniapps shows up individually besides wechat in taskmanager.

oh weird

02:59

well on android you can just download more apks

02:59

it's not really that special?

well for miniapps you install nothing

ah

Image attachment

03:00

these are all wechat

03:00

a starbucks app, a food ordering app and wechat itself.

well most large enough websites are replaced by apps on mobile anyway, do people not use them on PC either?

no one owns a PC now?

lol ok

"Don't you guys have phones?" makes a lot of sense in this context (edited)

but surely people use PCs to write documents and do work atleast? but they don't own one at home?

03:02

how else do they play ddnet xd

they do

03:03

i mean i definitely have a gaming laptop, but i don't think most do.

03:04

gamers on steam sure

03:04

but not even every students have their own computers i think

03:05

my sister bought a macbook and never used it

is there a PC gaming marketplace which is bigger than steam? or people still use steam?

Tater

is there a PC gaming marketplace which is bigger than steam? or people still use steam?

steam is still the biggest

03:06

there is wegame (tencent one), steamchina (valve + perfectworld) and some others, but they don't really have a lot games, since you'd need to get a publishing certificate for each game which is insanely hard and slow to get

03:07

steamcommunity.com is blocked but steampowered.com the store isn't so it is unofficially but deliberately allowed.

03:08

yep

I think anything related to trades, even if it is only digital goods are not really blocked

buff163 is by far the largest 3rd party steam marketplace so I assume it's allowed lol

catxplosion

well like international trades, not like skin trading but sure.

ah

Tater

is there a PC gaming marketplace which is bigger than steam? or people still use steam?

Steam on the Top!

[ddnet-maps:master] 1 new commit

99b1d0c M Necron - ddnet-maps

how to play ddnet on android is one of the top google searches leading to ddnet.org. Anyone want to put in the work to make DDNet run on Android again? I think @Jupstar ✪ already got us most of the way there?

[ddnet/ddnet] Pull request closed: #8320 Enforce more styling with clang tidy

05:03

github-merge-queue[bot]

[ddnet:master] 2 new commits

b422b0e Enforce more styling with clang tidy - ChillerDragon edff784 Merge pull request #8320 from ChillerDragon/pr_tidy_more - def-

Ewan

i think ryo is not so versed in cpp tho?

its not like im a noob, i would just need to use it more, also they probs care more about MLIR xo

05:37

xp*

05:37

but i prefer rust

[ddnet/ddnet] Issue closed: #8267 Proposal for a New DDNet Game Mode

06:00

[ddnet/ddnet] Pull request closed: #8324 Add chat commands for toggling weapons in practice mode

06:00

github-merge-queue[bot]

[ddnet:master] 5 new commits

d644987 add weapon chat commands for practice mode - ewancg bf78d7f cleanup - ewancg ba5b39d change forgotten functions to use practice check - ewancg 87cab3e change GetPracticeCharacter() to not be static, fix solo checks - ewancg df30321 Merge pull request #8324 from ewancg/chat-practice - def-

06:00

[ddnet/ddnet] Issue closed: #8092 /practice commands /weapons

[ddnet/ddnet] Issue opened: #8325 Integration test fails on Error: near line 1: in prepare, database is locked (5)

https://github.com/ddnet/ddnet/actions/runs/8980020413/job/24662999972?pr=8318 https://github.com/ddnet/ddnet/issues/8261#issuecomment-2081216658

[*] Shutting down client1
[*] Shutting down client2
Error: near line 1: in prepare, database is locked (5)
[-] Error: no ranks found in database
[!] Warning: ./stderr_client1.txt

morning edlang

Image attachment

Image attachment

07:07

poggers2

Ryozuki

Click to see attachment 🖼️

It looks kinda satisfying

xD

Ryozuki

Click to see attachment 🖼️

just noticed the typo omg (edited)

07:34

morning coding be like

07:35

xd i also made a typo on that sentence kek

deen

how to play ddnet on android is one of the top google searches leading to ddnet.org. Anyone want to put in the work to make DDNet run on Android again? I think @Jupstar ✪ already got us most of the way there?

I'll try to build it, but it would require connecting mouse+keyboard to play because we don't have on-screen touch controls

<ChillerDragon> There used to be on screen touch controls right?

09:51

<ChillerDragon> That is the old client people still play with if i rember correctly

got a huge load about gsm lte 3gpp and 5g

I love to find older lines of code that I wrote, just to wonder wth was going through my head at the time

10:08

today I found this masterpiece if Some(true).is_some() {

10:09

I'm relatively sure that was meant as a placeholder, but a comment would've been appropriate ^^

10:12

a few days ago I found this in an unused function nouis

nouis

min_binding_size: Some(NonZeroU64::new(0).unwrap()),

https://gcc.gnu.org/gcc-14/

Patiga

a few days ago I found this in an unused function nouis

nouis

min_binding_size: Some(NonZeroU64::new(0).unwrap()),

hmmmm

Patiga

a few days ago I found this in an unused function nouis

nouis

min_binding_size: Some(NonZeroU64::new(0).unwrap()),

isnt this a panic

Patiga

a few days ago I found this in an unused function nouis

nouis

min_binding_size: Some(NonZeroU64::new(0).unwrap()),

assigning 0 to a "NonZeroU64" and unwrapping it - wont this fail AND panic at the same time? :D

yes, it's similar to calling panic!() directly

<ChillerDragon> patiga drunk dev

@Robyt3 i managed to crash the client by setting cl_refresh_rate and gfx_refresh_rate to 240 aswell as turning Vsync on and then turning cl_refresh_rate and gfx_refresh_rate to 1000 and turning Vsync off i dont get a crashdump, and i cant even debug it for some reason, can you see if you can reproduce it somehow? - really weird.

2024-05-07 12:58:03 I client: got pong from current server, latency=20ms
12:58:16: /home/melonarch/Projects/ddnet/build/Desktop-Debug/DDNet ist abgestürzt.

literally the only output i get lmao (edited)

meloƞ

@Robyt3 i managed to crash the client by setting cl_refresh_rate and gfx_refresh_rate to 240 aswell as turning Vsync on and then turning cl_refresh_rate and gfx_refresh_rate to 1000 and turning Vsync off i dont get a crashdump, and i cant even debug it for some reason, can you see if you can reproduce it somehow? - really weird.

2024-05-07 12:58:03 I client: got pong from current server, latency=20ms
12:58:16: /home/melonarch/Projects/ddnet/build/Desktop-Debug/DDNet ist abgestürzt.

literally the only output i get lmao (edited)

Does it happen again if you perform the same steps? I assume it's a random GPU driver bug

tried it 3 times in a row

11:04

if i have cl_showfps to 1, you can see it jump between 240-1000 repeatedly when setting the refresh rate on both cl and gfx and it crashes as soon as you turn off Vsync

Please open an issue, not sure if we can do anything about it though

i will play around with it a bit and see if i can tickle out a crashdump somehow lol - but i did get this:

Image attachment

Did you try running with a debugger attached?

I'll have to leave for 2 1/2 hours, i will Run gdb and Check

meloƞ

i will play around with it a bit and see if i can tickle out a crashdump somehow lol - but i did get this:

you can also check if you get coredumps, or enable them if you don't

11:16

the coredump contains the process at the segmentation fault time

11:16

e.g. check coredumpctl

[ddnet/ddnet] Pull request opened: #8326 Remove RGB images from the map file format

Cherry-picked from https://github.com/teeworlds/teeworlds/pull/2822. CC https://github.com/teeworlds/teeworlds/issues/2812 CC https://github.com/teeworlds/teeworlds/issues/962 Since DDNet never saved these kinds of images, this should cause no issues.

Checklist

[x] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverag...

[ddnet/ddnet] Pull request opened: #8327 Revert "Initialize all ghost item member (Closed #8297)"

This reverts commit f319ed239ad27ceda928b9bcd399ca6c6379c88b. Initializing these variables with junk data doesn't seem to be an improvement over not initializing them. It hides potential Valgrind warnings about data accesses to uninitialized memory though. https://github.com/ddnet/ddnet/pull/8309/commits/f319ed239ad27ceda928b9bcd399ca6c6379c88b#r1590801501

Checklist

[ ] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination ...

0x00005555556a04c0 in CCommandProcessorFragment_Vulkan::ClearFrameData (this=0x555555fb6660, FrameImageIndex=4) at /home/melonarch/Projects/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:2048
2048                if(BufferPair.m_pMappedData != nullptr)

dis what i got (edited)

Any idea how to resolve these linking errors with rust symbols when compiling for android? (these are when compiling for x86_64)

ld.lld: error: undefined symbol: std::allocator<char>::allocator()
>>> referenced by cxx.cc:9 (src/cxx.cc:9)
>>>               cxx.o:(cxxbridge1$cxx_string$init) in archive debug/libddnet_engine_shared.a
>>> referenced by cxx.cc:204 (src/cxx.cc:204)
>>>               cxx.o:(_ZNK4rust10cxxbridge16StringcvNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEEEv) in archive debug/libddnet_engine_shared.a
>>> referenced by cxx.cc:317 (src/cxx.cc:317)
>>>               cxx.o:(_ZNK4rust10cxxbridge13StrcvNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEEEv) in archive debug/libddnet_engine_shared.a

See full log attached. Error messages when compiling for x86:

ld.lld: error: debug/libddnet_engine_shared.a(ddnet_engine_shared-fa6cb16cd6c82918.3zc566cpdlohmk2v.rcgu.o) is incompatible with elf_i386

36.9 KB

meloƞ

0x00005555556a04c0 in CCommandProcessorFragment_Vulkan::ClearFrameData (this=0x555555fb6660, FrameImageIndex=4) at /home/melonarch/Projects/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:2048
2048                if(BufferPair.m_pMappedData != nullptr)

dis what i got (edited)

Seems like it could be an issue in our code, please open an issue

Robyt3

Seems like it could be an issue in our code, please open an issue

will do

[ddnet/ddnet] Issue opened: #8328 segfault on ClearFrameData

from discord: i managed to crash the client by setting cl_refresh_rate and gfx_refresh_rate to 240 aswell as turning Vsync on and then turning cl_refresh_rate and gfx_refresh_rate to 1000 and turning Vsync off 12:58:16: /home/melonarch/Projects/ddnet/build/Desktop-Debug/DDNet ist abgestürzt. gdb outputted the following: ``` 0x00005555556a04c0 in CCommandProcessorFragment_Vulkan::ClearFrameData (this=0x555555fb6660, FrameImageIndex=4) at /home/melonarch/Projects/ddnet...

What would happen if we promote DDNet on TikTok or Instagram?

fokkonaut

What would happen if we promote DDNet on TikTok or Instagram?

stepfunn got a tiktok clip that got a few thousand views iirc :P

lol nice

https://www.tiktok.com/@stepfunn/video/7338119477168491808

TikTok · Stepfunn

370 likes, 5 comments. “Two guys using one braincell #viral #insane #famous #fyp #mustsee #edit #explore #win #legend #winner #goat #affiliate #secretlab #teeworlds #twitch #livestream#ddrace#ddracenetwork #funny”

fokkonaut

What would happen if we promote DDNet on TikTok or Instagram?

https://www.tiktok.com/@joni_2210?_t=8m9ceur7x4b&_r=1 has multiply clips with 300k views

joni_2210 bei TikTok

@joni_2210 3188 Follower, 84 Folgen, 131.3k Likes – Schau dir tolle Kurzvideos von joni_2210 an

gm

Ryozuki

lol someone from intel (with a intel.com domain) contacted me

nice, try it and when u fame remember where you come from: teeworlds

Robyt3

I'll try to build it, but it would require connecting mouse+keyboard to play because we don't have on-screen touch controls

if u need help ping me. What i can defs tell is: the libs would require pinned versions instead of just using master xD

Jupstar ✪

if u need help ping me. What i can defs tell is: the libs would require pinned versions instead of just using master xD

AFAICT the last puzzle piece is getting rust to link correctly (https://discord.com/channels/252358080522747904/293493549758939136/1237410618789728303) unless more errors appear after that. Otherwise it mostly worked by following the README. Needed to switch to SDL2 branch because SDL master is now SDL3. And we renamed the DDNet cmake target to game-client at some point.

do you use latest ndk?

no, I used the versions described in the readme. you think newer versions fixed rust support?

My current changes: https://github.com/Robyt3/ddnet/tree/Android-Building-Fix

Robyt3

no, I used the versions described in the readme. you think newer versions fixed rust support?

i dunno if it's important that the rust toolchain doesn't go out of sync with the ndk compiler or how exactly the rust toolchain generally detects the compiler

16:18

so could at least be worth a try, but dunno

yeah, maybe my rustc is too recent for the rest of the toolchain that comes with NDK 23, I'll try with latest NDK later

Robyt3

I'll try to build it, but it would require connecting mouse+keyboard to play because we don't have on-screen touch controls

Yeah, the onscreen touch controls are vital and hard to get right

What do you guys use to write code in c++? I had to learn it to make physical simulations and mathematical models, but I use vi and gnuplot for that, is there an ide I should use?

16:46

Image attachment

if you're using gnuplot for visualization you can use literally anything, if you're used to vi, why not use nvim?

16:52

you can use something like this: http://www.lazyvim.org/ which makes setting up a LSP (Language Server Protocol) and TS (Treesitter) for code autocompletion and suggestion pretty easy (edited)

Getting Started | LazyVim

LazyVim is a Neovim setup powered by

lazy.nvim

funman

What do you guys use to write code in c++? I had to learn it to make physical simulations and mathematical models, but I use vi and gnuplot for that, is there an ide I should use?

oO gigachad

funman

What do you guys use to write code in c++? I had to learn it to make physical simulations and mathematical models, but I use vi and gnuplot for that, is there an ide I should use?

Nowadays I'm using neovim. But I've also used vscode for years. Can't go wrong with either

16:55

If you aren't very fluid with vim, I might even go as far as to say vscode might be better atleast in the mid-short term

neovim my beloved

you should first make your own OS then your desktop and then your editor, if u are feeling hardcore you can design ur own CPU architecture

16:57

and compiler

16:57

and language

16:57

and go live in a cave

anyone knows how to deal with this problem?

Image attachment

Ryozuki

you should first make your own OS then your desktop and then your editor, if u are feeling hardcore you can design ur own CPU architecture

It is known that you need to start from making a bootloader

true

アリヂン

anyone knows how to deal with this problem?

Split it into a couple more manageable pieces?

16:58

I have an issue with my neovim setup that I can't pinpoint. In large json files, it really craps out

my neovim setup is 40 different files lol

how many plugins do you have even

Mine is 3 files right now. I might split it into one more later. I kinda keep all my plugins and plugin configs directly in init.lua for now

22

I once had a setup where I had one file per plugin. But then I got bored of it

funman

What do you guys use to write code in c++? I had to learn it to make physical simulations and mathematical models, but I use vi and gnuplot for that, is there an ide I should use?

IMO most important is: use clangd and clang-format

My clangd once formatted entirety of ddnet's codebase to replace spaces with tabs

17:04

And then it refused to do the same thing for another project in which I actually needed it

17:05

clang-format*

17:05

Actually, why do people even use spaces for indentation

アリヂン

Actually, why do people even use spaces for indentation

i dunno, but that's exactly why i'd always use clang-format... to not care about tabs vs spaces

rustfmt best

17:12

nobody discusses it

1

17:12

yep

nope

it defs has better defaults than clang-format lol

lol

clang-format default is some weird 1990 code base style

アリヂン

Actually, why do people even use spaces for indentation

it's more consistent as it does not rely on a special "tab" character

zhn

you should figure it out yourself by lurking 0.7 skin renderer code in upstream

I figured out how to render 0.6 skins, now 0.7 shouldn't be too hard

Image attachment

archimede67

it's more consistent as it does not rely on a special "tab" character

tab allows everyone to define their tab width

2

How is it more consistent if you can leave 3 or 5 spaces on accident

Ryozuki

tab allows everyone to define their tab width

tab width is just a fancy name for number of spaces

It's like an old technology which no one dares to replace

アリヂン

How is it more consistent if you can leave 3 or 5 spaces on accident

if you don't have a formatter then yeah

I mean, by tabs have advantages on all sides againts spaces

archimede67

tab width is just a fancy name for number of spaces

with spaces u cant customize tab width

but i'm talking consistency when you go from one editor to another

Ryozuki

with spaces u cant customize tab width

you can tell how many spaces you want your indentation to be (edited)

Any usable editor should support custom width I believe

Ryozuki

with spaces u cant customize tab width

is that critique on rustfmt?

i think u dont get it

17:15

thats just adding spaces

17:15

with tab its always 1 tan

17:15

tab

i know that

Jupstar ✪

is that critique on rustfmt?

xd idk

you can adjust how the code base looks without your formatter changing 65159 lines in 1206 files xd

im just saying

but visually if you say your tab is 4 width, then it will render as if it was 4 spaces

Exactly

yea thats the point xd

And you have an option to change that, as opposed to spaces

yes, but it will still use \t which can vary from one text editor to another (edited)

vim default is tab width of 8

an issue that you don't have with spaces

Learath2

Nowadays I'm using neovim. But I've also used vscode for years. Can't go wrong with either

I heard that there are certain things you need an ide to do like if you are working on a bigger project it’s a lot more efficient

iirc

hence the consistency i'm talking about

funman

I heard that there are certain things you need an ide to do like if you are working on a bigger project it’s a lot more efficient

neovim can be an ide (edited)

what?

Honestly, that's a config issue, I always set my tab size to 4

funman

I heard that there are certain things you need an ide to do like if you are working on a bigger project it’s a lot more efficient

you dont need IDE anymore in the world of LSP

Teero

neovim can be an ide (edited)

Oh I assumes it was a text editor

LSP are portable Ides

I’ll look into that

what about text editors you don't have the control over ? like online code previews or idk

funman

I heard that there are certain things you need an ide to do like if you are working on a bigger project it’s a lot more efficient

I've been hearing these things for a decade now. Haven't needed one no matter the size of the project even back when I only had ctags to navigate with

17:18

These people honestly must be talking about some insanely massive projects. Far bigger than anything I have access to

Interesting

17:19

I’ll get neovim then

17:19

I want to try coding stuff for ddnet to practice c++

archimede67

what about text editors you don't have the control over ? like online code previews or idk

I mean, that's one of the problems with tabs, they are not adopted everywhere

Maybe you might have trouble browsing around chromium? Idk which part they claim doesn't scale though. Clangd can do a lot including renames

アリヂン

I mean, that's one of the problems with tabs, they are not adopted everywhere

how old must your editor be to not support tabs? XD

Learath2

I've been hearing these things for a decade now. Haven't needed one no matter the size of the project even back when I only had ctags to navigate with

relying on one ide imo isn't great anyway. at work there is a old codebase that uses an old IDE and the IDE sucks so fuckig hard lol

lol

never heard of one

https://twitter.com/Demilogic/status/1787883295507730618

(@Demilogic) on X

I’ve sold more copies of Terraria than all your games combined

feelsbadman

1

competition is good xd

Image attachment

Terraria 2 confirmed??!?!

アリヂン

I mean, that's one of the problems with tabs, they are not adopted everywhere

I don't think it's an adoption issue, but more of a "what's the default tab size" issue which isn't the same everywhere which can make code look very large or very tight, which does not happen with spaces. That's the only reason why I think people use spaces instead of tabs

Ryozuki

https://twitter.com/Demilogic/status/1787883295507730618

i'd not call that diss. that was humiliation

If you could set your own tab width everywhere, there would be no more issues tho

Yeah, but that would required you to set it manually

@funman I do however use an IDE for php and Java. The biggest thing I'm missing without them is boilerplate generators. Writing setters and getters by hand is just oh so boring

Don't you also need to manually adjust space indent size?

I know I can create snippets for them but it's just one more manual thing I need to think about

I mean, setting tab width is a once-in-a-lifetime task, I did it and will never do again

アリヂン

Don't you also need to manually adjust space indent size?

Only in your editor so it knows when you press tab how many spaces to add, but then the file will look the same everywhere

What is your favourite tabwidth?

4

2

17:26

xd

17:26

ok maybe 4

0

4

For me, it's mostly the matter of how many junk characters I have to delete

Btw google still hasn't fixed this https://issuetracker.google.com/issues/242221770 show me a better argument for abolishing capitalism

OH

17:27

imagine -2 tab width

17:27

code indents inwards

17:27

madmen

17:27

arab code

3 indents = 12 spaces, and i'd have to do 12x to delete it all

xDDDDDDDDDD

Ryozuki

imagine -2 tab width

You start at the middle and unindent as you nest

XD

would be super funny

17:27

for a stream

jesus christ

17:27

heresy

17:28

what if the code is already on the left side of the screen?

アリヂン

3 indents = 12 spaces, and i'd have to do 12x to delete it all

Most editors have some magic to make this just 3 backspaces

Learath2

Btw google still hasn't fixed this https://issuetracker.google.com/issues/242221770 show me a better argument for abolishing capitalism

isnt it open source?

You shift everything right just to emphasize one line? (edited)

Jupstar ✪

isnt it open source?

You need to root your phone and patch the kernel to fix it. Can't use your bank app and many other things afterwards

Learath2

You need to root your phone and patch the kernel to fix it. Can't use your bank app and many other things afterwards

so it already is patched or what

They are literally not fixing it so you are forced to buy either their dongle or a bluetooth headphone

just not deployed

Jupstar ✪

so it already is patched or what

The guy who opened the issue put out a hacky patch. That's all that's available

17:31

Tbf they support their own dongle using that exact same patch, so not sure if calling it hacky is fair. But the proper way to implement this starts with the kernel exposing the dac hardware volume control to userspace

Image attachment

17:32

Is this a real C++ code?

what?

c++23

Learath2

You need to root your phone and patch the kernel to fix it. Can't use your bank app and many other things afterwards

i just have a second phone for my bank xddd i only use that phone for bank nothing else

17:32

so my main phone can be free as a bird

17:32

epyc

I can not be assed honestly. Just spy on me in a walled garden. They probably already have 8 cameras in my house

I wonder in how many editions we will get borrow checker in c++

Jupstar ✪

i just have a second phone for my bank xddd i only use that phone for bank nothing else

But not the worst idea

Learath2

@funman I do however use an IDE for php and Java. The biggest thing I'm missing without them is boilerplate generators. Writing setters and getters by hand is just oh so boring

As of now ig I’m not gonna be doing any other languages besides python but that shit is easy af

アリヂン

I wonder in how many editions we will get borrow checker in c++

i think there were talks about "profiles" that allow such stuff

poggers2

1

I wonder if one can write an external borrow checker for C++

isnt that tsan

17:35

and ubsan

17:35

well its not as powerful

those are runtime checks

this

1

oh

17:36

tru

17:36

weak

17:36

rust strong

17:36

monka

17:36

17:36

c++ also needs a MIRI

17:36

to add UB to rust they must have a check for it in MIRI

17:36

otherwise they wont add it

it's time for a new language ++c++

C-- when

17:37

like C++ but without OOP

Could have like a set of types that wrap raw objects that provide the borrow checking

17:39

Compiled by clang they are just optimized out as they are nothing but empty tags. Compiled by c++borrowck you track allocations and references to objects wrapped in those types

17:40

It might end up very annoying without real language support as very early rust was though

Sans

I figured out how to render 0.6 skins, now 0.7 shouldn't be too hard

you can render any res?

アリヂン

like C++ but without OOP

its called c

pepeW

1

アリヂン

like C++ but without OOP

C?

thonk

zhn

you can render any res?

with a few changes yea, but rn I'm rendering at 96x64

Learath2

I wonder if one can write an external borrow checker for C++

pvs-studio thing?

https://www.apple.com/newsroom/2024/05/apple-introduces-m4-chip/

Apple introduces M4 chip

Apple today announced M4, the latest Apple-designed silicon chip delivering phenomenal performance to the all-new iPad Pro.

Ryozuki

https://www.apple.com/newsroom/2024/05/apple-introduces-m4-chip/

>on ipad

Also rendering 500-ish tees, with nodejs, is surprisingly fast

17:46

fyi

17:46

kekw

Jupstar ✪

clang-format default is some weird 1990 code base style

clang format still doesn't support multiline star-aligned comment blocks pepeW

pepeW

yo should i just use freetype for font rendering or some wrapper of it like https://github.com/ulrichard/ftgl?

it depends

17:48

what do you want to do

Ryozuki

https://www.apple.com/newsroom/2024/05/apple-introduces-m4-chip/

theyre gonna run out of numbers atp

zhn

what do you want to do

making a small game in c++&opengl

Teero

yo should i just use freetype for font rendering or some wrapper of it like https://github.com/ulrichard/ftgl?

looks bit outdated

Teero

making a small game in c++&opengl

prototype?

i personally think freetype is ez to use. but generally text rendering is hard/annoying

zhn

prototype?

then yup why not

zhn

prototype?

nop xd

I kinda wanna take a shot at 0.7 skins now, you know where might I find that on gh? @zhn

okay freetype it is thx

Teero

okay freetype it is thx

https://github.com/ocornut/imgui

GitHub - ocornut/imgui: Dear ImGui: Bloat-free Graphical User inter...

Dear ImGui: Bloat-free Graphical User interface for C++ with minimal dependencies - ocornut/imgui

Jupstar ✪

https://github.com/ocornut/imgui

no xd

Sans

I kinda wanna take a shot at 0.7 skins now, you know where might I find that on gh? @zhn

something on master tree on src/game/client/*/skin.* i think (edited)

Teero

no xd

why not xd

Jupstar ✪

https://github.com/ocornut/imgui

doing from scratch with glfw opengl xd

Teero

doing from scratch with glfw opengl xd

then use pure freetype

Jupstar ✪

why not xd

https://scratch.mit.edu/ troll

troll

/s (edited)

well then freetype it is indeed

xd

Teero

https://scratch.mit.edu/ troll

troll

/s (edited)

xddd

17:51

You can call me Master-GPT

17:51

i know the words

okey tell how to load font!

17:51

@Jupstar ✪ @Jupstar ✪

@Jupstar ✪ you could try to use std::file_system to load it

17:52

then create a freetype font face @Jupstar ✪

xddddd

17:52

JPT

JLM

OJ

zhn

JPT

thanks for asking

17:53

#include <ft2build.h>
#include FT_FREETYPE_H

int main() {
    FT_Library library;
    FT_Error error;

    // Initialize FreeType library
    error = FT_Init_FreeType(&library);
    if (error) {
        // Handle error
        return 1;
    }

    // Load the font file
    FT_Face face;
    error = FT_New_Face(library, "path_to_your_font.ttf", 0, &face);
    if (error == FT_Err_Unknown_File_Format) {
        // Handle unsupported file format
        return 1;
    } else if (error) {
        // Handle other errors
        return 1;
    }

    // Set character size
    error = FT_Set_Pixel_Sizes(face, 0, 24);
    if (error) {
        // Handle error
        return 1;
    }

    // Load character bitmap for a specific character
    error = FT_Load_Char(face, 'A', FT_LOAD_RENDER);
    if (error) {
        // Handle error
        return 1;
    }

    // Access the character bitmap
    FT_Bitmap bitmap = face->glyph->bitmap;

    // Now you can use the bitmap data as needed

    // Clean up
    FT_Done_Face(face);
    FT_Done_FreeType(library);

    return 0;
}

17:54

actually looks correct

17:54

but dunno if it compiles

17:54

but smth like that is indeed the process

huh looks odd

17:55

should try it myself later :p

17:55

never actually touched freetype

zhn

should try it myself later :p

epyc. now we all get freetype devs

time for https://learnopengl.com/In-Practice/Text-Rendering

LearnOpenGL - Text Rendering

Learn OpenGL . com provides good and clear modern 3.3+ OpenGL tutorials with clear examples. A great resource to learn modern OpenGL aimed at beginners.

17:57

anti-aliasing will be a pain too i think

Teero

time for https://learnopengl.com/In-Practice/Text-Rendering

nice one

17:57

download a pdf though xd

17:58

archoover

Teero

anti-aliasing will be a pain too i think

it's super super hard

17:58

in dd-pg i now have a mode to disable msaa exactly once per frame

17:58

and usually this is before UI code is rendered

17:59

so UI code is not affected by msaa (edited)

17:59

tbh i am not even sure if u can do that in opengl well in newer probably

17:59

at least not as i did it

17:59

maybe i'll just use a texturemap with characters. all of the fonts will be my own anyways :D

18:00

im not gonna try making ttfs

18:01

~~that would limit me to monospace tho if i don't include other info.~~ edit: it would not limit me to monospace xd (edited)

if text rendering would be ascii chars only, it would not suck

18:02

18:02

but with unicode and all the weird stuff it just sucks to implement it xD

quick question, where can I get the 0.7 client?

Sans

quick question, where can I get the 0.7 client?

https://teeworlds.com/

18:05

@Jupstar ✪ do you think a texture hashmap would be fast enugh?

18:06

does caching frequent phrases work?

Teero

@Jupstar ✪ do you think a texture hashmap would be fast enugh?

u want to create one texture per text?

Jupstar ✪

u want to create one texture per text?

one texture map with all the characters and maybe combine frequent combinations of characters and cache them

Sans

quick question, where can I get the 0.7 client?

https://github.com/teeworlds/teeworlds

GitHub - teeworlds/teeworlds: A retro multiplayer shooter

A retro multiplayer shooter. Contribute to teeworlds/teeworlds development by creating an account on GitHub.

Teero

one texture map with all the characters and maybe combine frequent combinations of characters and cache them

first sounds good

and releases

second sounds useless

Jupstar ✪

second sounds useless

and trouble to implement xdd

18:08

macro and template magic in my head

Jupstar ✪

second sounds useless

when having static gui that never changes it should be a good optimization rendering "Text" directly instead of "T" "e" "x" "t"

18:10

i don't know how much of a difference it makes but i have noticed that ddnets ways or rendering text is not insane fast

18:11

i dunno how ddnet does it tho

Teero

when having static gui that never changes it should be a good optimization rendering "Text" directly instead of "T" "e" "x" "t"

i dunno, it only creates a few more vertices

Teero

i don't know how much of a difference it makes but i have noticed that ddnets ways or rendering text is not insane fast

which one

18:11

the chat or the menu?

Jupstar ✪

which one

ingame xd

Teero

ingame xd

how is it not fast? xd

wait

scoreboard?

tele nums

tele nums should be fast on newer GL versions

18:12

you just use opengl 1.x all the time xD

Jupstar ✪

tele nums should be fast on newer GL versions

faster than anything u could achieve with freetype

18:13

bcs it's created at client start

Jupstar ✪

bcs it's created at client start

wha?

18:14

wait does client bake a whole font atlas on startup

zhn

wait does client bake a whole font atlas on startup

well only numbers 1-255

18:14

but yes

18:14

it does

nice approach

Jupstar ✪

well only numbers 1-255

i get around 180 fps on 1500x1500 map full of teles xdd

Image attachment

Teero

i get around 180 fps on 1500x1500 map full of teles xdd

with opengl 1?

vulkan xD

18:15

i forgor to change

no way

18:15

send map

Teero

i get around 180 fps on 1500x1500 map full of teles xdd

and that is ingame?

28.45 KB

Jupstar ✪

and that is ingame?

yea

18:17

okay with opengl 3.3 i get ~190

18:17

and its a little more stable

Teero

and its a little more stable

but just do cl_text_entities 0

18:18

then you notice the problem is not the text rendering

18:18

i get around 300-400 fps, depending on zoom lvl

oh. okay

18:19

i tried rendering some weight values some time ago for like a 300x300 map and it was pretty laggy

18:19

although that was maybe my fault

if u need high fps at high zoom levels u should use patigas rendering method

18:20

it scales better at high zoom values

18:20

note that it also uses lot of VRAM tho

18:20

if that is an issue for you

float ScreenX0, ScreenY0, ScreenX1, ScreenY1;
m_pClient->Graphics()->GetScreen(&ScreenX0, &ScreenY0, &ScreenX1, &ScreenY1);
float BorderBuffer = 100;
ScreenX0 -= BorderBuffer;
ScreenX1 += BorderBuffer;
ScreenY0 -= BorderBuffer;
ScreenY1 += BorderBuffer;

const int Width = m_pClient->Collision()->GetWidth();
const int Height = m_pClient->Collision()->GetHeight();
for(int x = 0; x < Width; ++x)
{
    for(int y = 0; y < Height; ++y)
    {
        if(x * 32 > ScreenX1 || x * 32 < ScreenX0 || y * 32 > ScreenY1 || y * 32 < ScreenY0)
            continue;
        char aBuf[8];
        int Index = y * Width + x;
        CUIRect Rect = {(float)x * 32, (float)y * 32, 32, 32};
        // evil top left
        str_format(aBuf, 8, "%d", m_pClient->Collision()->IsEvilTeleport(Index));
        m_pClient->Ui()->DoLabel(&Rect, aBuf, 15.f, TEXTALIGN_TL);
        // normal top Right
        str_format(aBuf, 8, "%d", m_pClient->Collision()->IsTeleport(Index));
        m_pClient->Ui()->DoLabel(&Rect, aBuf, 15.f, TEXTALIGN_TR);
        // check evil bottom left
        str_format(aBuf, 8, "%d", m_pClient->Collision()->IsCheckEvilTeleport(Index));
        m_pClient->Ui()->DoLabel(&Rect, aBuf, 15.f, TEXTALIGN_BL);
        // check normal bottom Right
        str_format(aBuf, 8, "%d", m_pClient->Collision()->IsCheckTeleport(Index));
        m_pClient->Ui()->DoLabel(&Rect, aBuf, 15.f, TEXTALIGN_BR);
    }
}

ah i was testing tps since smh they didn't work correctly and i had broke smth (edited)

18:23

this is old code please just imagine the aBuf is outside the loops and so is the Index and the CUIRect declaration (edited)

yeah dolabel is rip

18:28

@Patiga do u plan to support physics layers rendering?

str_format is also very slow, str_from_int is much faster if you just need to convert an integer to a string

poggers2

1

yea, will do at some point

Robyt3

Any idea how to resolve these linking errors with rust symbols when compiling for android? (these are when compiling for x86_64)

ld.lld: error: undefined symbol: std::allocator<char>::allocator()
>>> referenced by cxx.cc:9 (src/cxx.cc:9)
>>>               cxx.o:(cxxbridge1$cxx_string$init) in archive debug/libddnet_engine_shared.a
>>> referenced by cxx.cc:204 (src/cxx.cc:204)
>>>               cxx.o:(_ZNK4rust10cxxbridge16StringcvNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEEEv) in archive debug/libddnet_engine_shared.a
>>> referenced by cxx.cc:317 (src/cxx.cc:317)
>>>               cxx.o:(_ZNK4rust10cxxbridge13StrcvNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEEEv) in archive debug/libddnet_engine_shared.a

See full log attached. Error messages when compiling for x86:

ld.lld: error: debug/libddnet_engine_shared.a(ddnet_engine_shared-fa6cb16cd6c82918.3zc566cpdlohmk2v.rcgu.o) is incompatible with elf_i386

if it's too much work to get this to work, perhaps just remove the rust stuff from ddnet. it hasn't seen any use

Sans

I figured out how to render 0.6 skins, now 0.7 shouldn't be too hard

I heard @Patiga knows a lot about rendering skins

oh?

I actually haven't looked at 0.7 skins yet, cant help there :/

ah rip

Sans

ah rip

but maybe you could document your findings in a document similar to this: https://github.com/heinrich5991/libtw2/blob/ac6b4c422aebd43bc18a5f9a3526b999e62e0b5c/doc/tee_rendering.md

libtw2/doc/tee_rendering.md at ac6b4c422aebd43bc18a5f9a3526b999e62e...

Some Teeworlds stuff in Rust.™. Contribute to heinrich5991/libtw2 development by creating an account on GitHub.

19:26

so that we can also get documentation on 0.7 skin rendering

heinrich5991

but maybe you could document your findings in a document similar to this: https://github.com/heinrich5991/libtw2/blob/ac6b4c422aebd43bc18a5f9a3526b999e62e0b5c/doc/tee_rendering.md

That's hella useful how did i not know about this

(written by @Patiga btw ^^)

patiga my beloved

heartw

1

19:28

nod

19:29

Slight issue tho, I am wildly inexperienced in this stuff

19:29

I'm doing this caveman style until I get it working

19:29

then I'm never touching it again xd

my lib is public though, and I like to believe my code is easily understandable, so by all means check it out when I'm done :D

19:36

@Patiga btw, about 0.6 skins, how do I actually color them?

19:36

if the player has custom colors set

19:37

Do I just tint the image with the color or is there some other process I should do?

my code is not correct there currently

19:38

if you figure out the rgba color, you would only need to multiply it on top

I've got the hsl from the tw code

19:38

I can convert it

[ddnet/ddnet] Issue opened: #8329 Add Map client settings for each map

Like make it possible to have a manual or automatic saved map Setting per map, which gets loaded when u load it. cl_map_settings 1/0 Good for pvp: ,ctf4old | ctf5 | etc Disable ammo_healthbar ,angle, postion, speed, High Details Off/On, (refresh rates) Higher on PVP ,Assets DDRace: ON/Off Settings for: Entities, High Details, Load specific map backgrounds, Background colour, background music "On/Off" (Future: Cursor), Pistole Sound, Grenade Explosion Sound, (jetpack maps) ammo_healt...

but I meant applying the color (edited)

19:38

do I just do a simple tint?

multipy the pixel color by the tees color

19:39

where rgba 255 = 1

so non-transparent pixels

19:39

aight

you can do it for all ^^

19:40

for transparent it simply does nothing :)

ye but every pixel excused from processing matters

1

Do 0.7 color codes work differently than 0.6?

20:04

I can't seem to be able to convert properly

20:04

with 0.6 codes

There is a slight difference. The lightness clamping is different

Jupstar ✪

so could at least be worth a try, but dunno

Using latest NDK and Android API still results in the same linker errors

heinrich5991

if it's too much work to get this to work, perhaps just remove the rust stuff from ddnet. it hasn't seen any use

yeah, might be the best solution :/

Learath2

There is a slight difference. The lightness clamping is different

wdym

Lightness values from 0 to 255 don't map to real hsl lightness 0.0 to 1.0

20:09

For 0.6 they map to x to 1.0 and for 0.7 they map to y to 1.0. I don't remember x and y though, but I remember y < x

i really wish I could read c++

20:09

so I can ask where might I find that

20:09

(edited)

C++ doesn't look too far off from JS, does it?

uhh

20:10

it's basically tring to read chinese while being an american

The lightest value for 0.7 should be in teeinfo.h

20:12

The one for 0.6 should be iiin color.h I think

deathgood

try color&255, color/256&255, color/256/256&255

For 0.7 I used this kind sir's suggestion, lemme see if I can cook something similar for 0.6

It's the same for both 0.6 and 0.7. It's just that you interpret the L component differently. Teeworlds just does not allow L to be 0 so you cant have pure blacks

20:19

0.6 and 0.7 have different darkest colors allowed. 0.7 allows slightly darker ones

Robyt3

Any idea how to resolve these linking errors with rust symbols when compiling for android? (these are when compiling for x86_64)

ld.lld: error: undefined symbol: std::allocator<char>::allocator()
>>> referenced by cxx.cc:9 (src/cxx.cc:9)
>>>               cxx.o:(cxxbridge1$cxx_string$init) in archive debug/libddnet_engine_shared.a
>>> referenced by cxx.cc:204 (src/cxx.cc:204)
>>>               cxx.o:(_ZNK4rust10cxxbridge16StringcvNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEEEv) in archive debug/libddnet_engine_shared.a
>>> referenced by cxx.cc:317 (src/cxx.cc:317)
>>>               cxx.o:(_ZNK4rust10cxxbridge13StrcvNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEEEv) in archive debug/libddnet_engine_shared.a

See full log attached. Error messages when compiling for x86:

ld.lld: error: debug/libddnet_engine_shared.a(ddnet_engine_shared-fa6cb16cd6c82918.3zc566cpdlohmk2v.rcgu.o) is incompatible with elf_i386

mhhhh it often says cxxbridge.. maybe update the cxx dep?

20:25

but dunno i dont really understand why the allocator<char> fails

20:26

on the other hand side it works without problem on non android

heinrich5991

if it's too much work to get this to work, perhaps just remove the rust stuff from ddnet. it hasn't seen any use

how should have it been integrated?

20:26

as submodules?

not sure what you mean with 'should'

20:27

the rust integration works

uhm

20:28

i mean

20:28

all .rs files are just rewritten from cpp ones

ld.lld: error: debug/libddnet_engine_shared.a(ddnet_engine_shared-fa6cb16cd6c82918.3zc566cpdlohmk2v.rcgu.o) is incompatible with elf_i386

sounds like you're compiling the rust code for the wrong architecture. check its architecture with file maybe

doing them one by one is kinda not useful, by submodules i mean fully rewrite of src/base for example

20:29

from my point of view though

20:29

i just don't know what should be done to rustify codebase

https://gaultier.github.io/blog/how_to_rewrite_a_cpp_codebase_successfully.html

20:30

there's also the librsvg story

hm so problem is that rust bus factor for ddnet is 0 (1)

yea, I guess the gaultier story isn't applicable there

20:39

there are more developers writing c++ code than those writing rust code in the ddnet community, I think

ig problem is that current core code e.g src/base is maintainable enough to not be RIIR

@Robyt3 does it compile i386 on the x64 build or what?

20:53

or is that the x86 build

The long log in https://discord.com/channels/252358080522747904/293493549758939136/1237410618789728303 is from the x86_64 build, with the x86 build I get the ... is incompatible with elf_i386 errors instead, with arm/arm64 builds I also get other errors

heinrich5991

ld.lld: error: debug/libddnet_engine_shared.a(ddnet_engine_shared-fa6cb16cd6c82918.3zc566cpdlohmk2v.rcgu.o) is incompatible with elf_i386

sounds like you're compiling the rust code for the wrong architecture. check its architecture with file maybe

It outputs libddnet_engine_shared.a: current ar archive for the files from both the x86 and x86_64 builds

hmm, sorry, thought it'd output the architecture

When compiling for ARM I get errors like

ld.lld: error: debug/libddnet_engine_shared.a(ddnet_engine-97b91143faa7528b.zhg56s06o907mid.rcgu.o) is incompatible with armelf_linux_eabi
ld.lld: error: debug/libddnet_engine_shared.a(ddnet_base-98591c5ad164f713.1wywbl7wtpc263ng.rcgu.o) is incompatible with armelf_linux_eabi

that sounds like you're compiling the rust code for the wrong architecture

21:37

readelf -h <archive>.a | grep 'Class\|File\|Machine'

21:37

is what stackoverflow gives me

21:38

@Robyt3

This is for the x86 build:

File: libddnet_engine_shared.a(ddnet_engine_shared-fa6cb16cd6c82918.10dggtc7pn3aflxu.rcgu.o)
  Class:                             ELF64
  Machine:                           Advanced Micro Devices X86-64
...

x86_64 sounds like it compiled for the wrong architecture

I think this looks promising: https://github.com/rust-lang/rustup/issues/2872#issuecomment-1158238378

Can not install on android - target 'aarch64-linux-android' not fou...

Problem When trying to install rustup using the linux command on the rustup site, when proceeding with install it gives me the following error: Current installation options: default host triple: aa...

21:51

So uses of cargo would need to be replaced with cargo ndk -t aarch64-linux-android (with the respective arch) (edited)

<ChillerDragon> > str_format is also very slow, str_from_int is much faster if you just need to convert an integer to a string

23:10

<ChillerDragon> @Robyt3 woah i did not know that

23:11

<ChillerDragon> Can we add a grep for str_format "%d" to the CI to auto check that?

it's kinda sad that we can't simply optimize str_format…

How exactly does it go from pink to blue idk, and I've spent 2 hours trying to figure it out I need help ;-;

Image attachment

Image attachment

23:29

This is after converting the color code to hsl, and translating saturation and lightness to a percentage of 255

Exported 515 message(s)