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 2025-02-15 00:00 and 2025-02-16 00:00

ChillerDragon: Voldemort client dev played on your name and got banned lol

00:21

We do automated name bans

00:23

You're unbanned now. If you make an account only you can play on your name

00:27

sry for the inconvenience we'll change some stuff in the future

ws-client

<ChillerDragon> why does it advertise cheat clients instead?

That is a good Point i will Change that as soon as possible

ws-client

<ChillerDragon> @Daniel why does the error message not say that?

Which error message?

Daniel

Which error message?

The ban message xd

Teero

The ban message xd

Oh yea we could be more specific...

less*

01:01

If you ban for cheat then don’t name the client brand

But thanks for your feedback @ChillerDragon this will help us to improve

Yea the automated name bans sound dangerous

It's like dual linked... Every IP is connected to nickname and every nickname to IP and the second link is kinda dangerous as i see the last time that alot of people get banned for using the name '.'

Imo ip and name should have to match both

agreed

chillerdragon

Imo ip and name should have to match both

so they should be able to change their name and log back in? lol

lol

[ddnet-maps:master] 1 new commit

1cab4b0 M X-Cross, A Romantic, M Semang 3 - ddnet-maps

Yes better that than allowing every player to name ban whoever they want

Replying to @Tater so they should be able to change their name and log back in?

07:01

I mean ideally cheaters wouldn’t be banned at all.

07:01

If there are no bans then you do not have to worry about ban bypasses. Most of the ban bypass protections affect legit users negatively.

07:03

There should be automated anti cheat which kicks and does not ban. That does not exclude players from the server permanently that try cheats but force them to switch to a legit client.

agreed

07:06

also where is the downloaded maps stored on linux

07:06

wait nvm i dont need to know acsualy i read my code wong

07:06

(for context im going to fuzz the map parser) (edited)

@david loves Jellycar Worlds: do you play the game?

yes

07:10

some

07:10

why?

Just curious

fair.

07:10

also how are you talking though a bot lol. oh wait probs slack or whatever the outer comunicatson methed was

Doing fuzz and not knowing the map downloads sounds like some casual who never played the game and does not intend to play the game. But instead is a security researcher who fuzzes all open source projects. I want to know if someone doesn’t care about the game and doesn’t intent to stay around.

07:11

I am on matrix right now and sometimes irc

07:12

Fuzzing the map parser has been done already btw in case you didn’t know

07:12

Not that it hurts to do it again it’s nice that you do it

07:13

Might be interesting for you if you don’t know it already https://mmmds.pl/fuzzing-map-parser-part-1-teeworlds/

07:14

Now that I think about it that blog might fit into the ddnet wiki blog collection

yea i saw that from the docs about the code on the website

07:15

im also planning on fuzzing the server connects but in a difrent way from them

07:15

but yea i play the game some

07:16

though im quite bad at it lol

Make sure to sell me some 0days before you publish the results. I need to install permanent anti cheat software on the machines of my fng player base

lol

07:17

its my first time fuzzing funny enough

07:17

so dont exspect much

If you are real gangster you get fuzzing integrated into the official GitHub CI

nah i dont feel like lerning how GitHub CI works

07:18

mabey someday though

But tbh not sure if I actually want slower CI runtimes. Maybe some external project with scheduled CIs for ddnet would be nicer

what does CI stand for here anyways

07:19

never used it

Chiller Intelligence

1

07:20

Apple stole my slogan

07:20

jk continuous integration

makes sence

07:21

it does some checks when pushing right?

And CD is ChillerDragon

07:21

Yes

07:21

The syntax is easy to learn

07:21

It’s yaml engineering

and yea im gonna copy a lot of https://mmmds.pl/fuzzing-map-parser-part-1-teeworlds/ 's code lol. at least for the map fuzzing

You just put a Linux command into a text file and it runs. If it exits non zero pipeline fails

will likely make my own for the network fuzzing though as i have a difrent idea on how to do it

<ChillerDragon> okay i collected enough ideas for my antibot now i am finally starting the repo. Any suggestions for a badass project code name?

hmm hacker-catcher

07:28

or hacker-catchers

<ChillerDragon> sry but thats not badass enough

07:29

<ChillerDragon> i think its time to use a LLM

Here are some badass name ideas for an anti-cheat for DDNet: Teeguard – A mix of Tee (from Teeworlds) and Guard, simple and effective. IronTee – Like Ironclad protection for your game. TeeShield – A strong, straightforward name that conveys protection. SentinelTee – Inspired by vigilant guardians. GhostTrap – A reference to catching invisible cheaters. AegisTee – Aegis means divine shield, combined with Tee. TeeStrike – Because cheaters deserve a swift strike. PurgeTee – Cleansing the game of cheaters. OverwatchTee – Keeping a watchful eye on players. ShadowSentinel – A silent yet unstoppable force against cheaters. Let me know if you want something more unique or themed differently!

<ChillerDragon> thats the only thing they can do :D generate text

i already did lol

<ChillerDragon> thanks

yw

https://github.com/cordx56/rustowl

GitHub - cordx56/rustowl: Visualize Ownership and Lifetimes in Rust

Visualize Ownership and Lifetimes in Rust. Contribute to cordx56/rustowl development by creating an account on GitHub.

08:16

epyc

08:16

@Learath2

08:16

made by a weeb

08:16

Image attachment

08:17

green: variable's actual lifetime

blue: immutable borrowing

purple: mutable borrowing

orange: value moved / function call

red: lifetime error - diff of lifetime between actual and expected

Ryozuki

https://github.com/cordx56/rustowl

Oooh, that looks cool

Chiller: Teelted

Ryozuki

https://github.com/cordx56/rustowl

Sweet

how can i change someone's skin by server? help me on gamecontext.cpp

09:49

i added player_name and player_clan, which changes player info by admin or server

09:49

but stuck at player_skin

09:52

bool CServer::SetClientSkinImpl(int ClientId, const char *pSkinRequest, bool Set)
{
    dbg_assert(0 <= ClientId && ClientId < MAX_CLIENTS, "invalid client id");
    if(m_aClients[ClientId].m_State < CClient::STATE_READY)
        return false;

    // trim the skin
    char aTrimmedSkin[MAX_SKIN_LENGTH];
    str_copy(aTrimmedSkin, str_utf8_skip_whitespaces(pSkinRequest));
    str_utf8_trim_right(aTrimmedSkin);

    bool Changed = str_comp(m_aClients[ClientId].m_aSkin, aTrimmedSkin) != 0;

    if(Set)
    {
        // set the client skin
        str_copy(m_aClients[ClientId].m_aSkin, aTrimmedSkin);
    }

    return Changed;
}

-# i connected this codeblock to gamecontext.cpp and call it, but when i enter player_skin command, nothing happens

@MilkeeyCat https://github.com/jsinger67/parol

GitHub - jsinger67/parol: LL(k) and LALR(1) parser generator for Rust

LL(k) and LALR(1) parser generator for Rust. Contribute to jsinger67/parol development by creating an account on GitHub.

david loves Jellycar Worlds

and yea im gonna copy a lot of https://mmmds.pl/fuzzing-map-parser-part-1-teeworlds/ 's code lol. at least for the map fuzzing

They also wrote a blog post on fuzzing the Teeworlds network: https://logicaltrust.net/blog/2020/07/socketfuzzer.html Fuzzing the map format isn't really worth it unless for learning, you should easily find crashes everywhere because the format is mostly unchecked at the moment (edited)

LogicalTrust - [EN] A-Z: Introducing SocketInjectingFuzzer and its ...

The idea behind Socket Injecting Fuzzer is simple - fuzz applications working in the client-server architecture by mutating real data packets that are exchanged by them in real usage scenarios. Implementation is also simple because it is based on hooking network-related functions on one side (or both) of a connection and modifying outgoing data....

<ChillerDragon> @zhn when russians say "soft" they mean cheat right? Is it short for software? I have seen sentences such as "ddnet is not software" which really confused me

1

Ryozuki

@MilkeeyCat https://github.com/jsinger67/parol

Imagine using parser generators

chillerdragon: yeah, they often refer to cheats and bots as soft, like a short of software

Ryozuki

green: variable's actual lifetime

blue: immutable borrowing

purple: mutable borrowing

orange: value moved / function call

red: lifetime error - diff of lifetime between actual and expected

@Anime.pdf worths a try for you? justatest

justatest

How is software not ambiguous? How do you call non cheat software then?

[ddnet-web:master] 1 new commit

ae4e576 Anime.pdf stepped down, cleanup - murpii

"cleanup" is crazy :( (edited)

@Anime.pdf

ws-client

<ChillerDragon> @zhn when russians say "soft" they mean cheat right? Is it short for software? I have seen sentences such as "ddnet is not software" which really confused me

smartest russian player

chillerdragon

How is software not ambiguous? How do you call non cheat software then?

we don't call software "software" in regular life

<ChillerDragon> @zhn weird stuff ngl. how do you call software then? program?

12:50

<ChillerDragon> i am so fucking unhappy with test output

12:51

<ChillerDragon> its full of spam i spend half day scrolling to the failed tests and half of my debug logs just get swallowed wtf

12:51

<ChillerDragon> GTEST is the most useless shit writing a better test framework can be done in 10 lines

12:54

<ChillerDragon> ok the log swallow i take back that was me editing the wrong file but i still hate GTEST

12:55

<ChillerDragon> i should probably blacklist client/prediction from my file finder i keep editing client code when i search for server file names

ws-client

<ChillerDragon> @zhn weird stuff ngl. how do you call software then? program?

Yap

1 warning and 20 errors generated ah i thought sdl3 and sdl2 were similar enough xd

13:18

seeing as its 1 to 1 enough to make a compat library

13:19

that might actually be a good way to do conversion

[ddnet/ddnet] Pull request opened: #9679 Break physics so konsti can cheat some maps

It bothers me that SetPosition does not actually set the position. I keep writing pChr->SetPosition(Pos);pChr->m_Pos = Pos like a crazy person. It is not needed right now because the values are synced on tick. But to me it just seems wrong to not set the position in a method literally called SetPosition. As a easter egg the diff between syncing the postion on call vs on tick unlocks some interesting new teleportation exploits.

:ro...

wacky pr

chiller completely drunk again

gah sdl3 has missing docs

did chillerdragon click on one of those scam links and they stole his github account instead of steam

@ChillerDragon isnt there antibot for fng by n** already?

<ChillerDragon> @Daniel i do not have permission from onbgy to use it so i have to reinvent the wheel yay time spent wisely

13:34

<ChillerDragon> @Solly which docs missing? I found sdl3 docs pretty good last time i tried and that was before official release xd

to do with joysticks

13:34

most of the changes are simple renames

<ChillerDragon> ew sticks

yeah but it doesnt compile...

13:35

if i dont fix it

<ChillerDragon> git log -S is your best friend

i want sdl3 so i can copy files to clipboard

<ChillerDragon> who needs docs of changes when u can do this

ws-client

<ChillerDragon> @Daniel i do not have permission from onbgy to use it so i have to reinvent the wheel yay time spent wisely

im working on a teehistorian antibot for gores currently in rust based on tee-hee by @Zwelf what is based in tw-game by @Patiga

<ChillerDragon> git clone https://github.com/libsdl-org/SDL;cd SDL;git log -S$FuncNameYouMissFromSDL2

ws-client

<ChillerDragon> git log -S is your best friend

y

13:36

ah

13:36

very cool

<ChillerDragon> so not real time but can be applied to teehistorian files @Daniel ?

ws-client

<ChillerDragon> so not real time but can be applied to teehistorian files @Daniel ?

yep. i think using antibot.cpp could be also interesting but with that you can improve the detection and rerun it on all runs again.

twgame is from Zwelf! ^^

<ChillerDragon> yea ddrace can really be done after the fact

yet again i am cucked by window being hidden from everywhere

<ChillerDragon> its more about removing ranks

Patiga

twgame is from Zwelf! ^^

did i messed up and switched the repos?

<ChillerDragon> in pvp modes you want real time otherwise enemys suffer

13:39

<ChillerDragon> but yes interesting approach maybe i should start recording teehistorian too never did that

ws-client

<ChillerDragon> in pvp modes you want real time otherwise enemys suffer

you only host block servers, correct?

<ChillerDragon> then you can have automated testes for antibot without having to test it in production

13:40

<ChillerDragon> @Daniel incorrect

13:40

<ChillerDragon> zillyhuhn incorporated conglomerate hosts an array of different gamemodes since decades

13:41

<ChillerDragon> this is what is running on my vps right now

13:41

<ChillerDragon> https://zillyhuhn.com/cs/.1739626873.png

Embedded image

i dont do auto bans... the final approach is to write demos out of "sussy" gameplay and anonymize it to show it to legit players and trusted ones.. and they can get the demo or just a file with extra info and the demo itself to watch and say if they thing it a bot or not (kinda like in cs2/csgo the overwatch)

<ChillerDragon> yea sounds fun @Daniel

isnt the antibot based on teehistorian interesting for ddrace server hosted by yourself?

<ChillerDragon> i have no cheaters in ddrace at least none i care about

thankfully

Daniel

i dont do auto bans... the final approach is to write demos out of "sussy" gameplay and anonymize it to show it to legit players and trusted ones.. and they can get the demo or just a file with extra info and the demo itself to watch and say if they thing it a bot or not (kinda like in cs2/csgo the overwatch)

don't even need to deal with demo files necessarily, could have a server just replaying the events

<ChillerDragon> but i think it would be cool for fng too just to have something you can throw antibot against

risu

don't even need to deal with demo files necessarily, could have a server just replaying the events

you can host and provide like a website where they can do the overwatch and can vote... but this idea only work with active playerbase who support it

<ChillerDragon> i havent even started to write antibot not sure if i will ever do haha but i will start with live stuff and maybe one day ill look into teehistorian

13:44

<ChillerDragon> my fng playerbase is 99% botters

Daniel

you can host and provide like a website where they can do the overwatch and can vote... but this idea only work with active playerbase who support it

I'm just saying that manually dealing with demo files is more effort than connecting to special overwatch server

i need sdl_window m_pWindow from backend_sdl everywhere, should i 1. weezel it thru some init like other places do 2. just expose it in IGraphics (its not like theres gonna be another backend any time soon) (edited)

though it's obviously a lot fancier on the server end

CS's Overwatch stopped working after this "area" was also botted :)

<ChillerDragon> @Daniel i would never trust players with saying who is bot or not

13:45

<ChillerDragon> people throw out cheating accusations way too much

since teeworlds dont support dynamic tiles you cant trap like players for 1 tick to a freezebox to check they behauvior.. like every normal person shouldnt see big diff and just press "a" for example but every bot would trigger all there things to prevent to go into the freeze xD

<ChillerDragon> @Daniel interesting

i thought u could do something with the map in real time

13:46

like how infclass does the seelction wheel

like place freeze block under someone and if they insta jump and they do it multiple times you can say for sure they use avoid freez

Daniel

since teeworlds dont support dynamic tiles you cant trap like players for 1 tick to a freezebox to check they behauvior.. like every normal person shouldnt see big diff and just press "a" for example but every bot would trigger all there things to prevent to go into the freeze xD

also youll just get urself into a war if your detection methods are not stealthy

@Ryozuki

    let program = Program {
        globals: vec![],
        functions: vec![Function {
            registers: vec![Register {
                name: "1".into(),
                ty: Ty::I8,
            }],
            name: "main".into(),
            blocks: vec![BasicBlock {
                name: "0".into(),
                instructions: vec![Instruction::Copy {
                    operand: Operand::Const(ValueTree::Leaf(Const::I8(0x45))),
                    place: Place::Register(0),
                }],
                terminator: Terminator::Return(Some(Operand::Place(Place::Register(0)))),
            }],
            params: vec![],
            ret_ty: Ty::I8,
        }],
    };

I can finally generate assembly for dis

<ChillerDragon> infclass selection wheel uses client exploit based on map file format and live patches the client to support custom protocol

Solly

also youll just get urself into a war if your detection methods are not stealthy

yep

noice

you have to do something heuristic probably

but since i dont implement it in realtime i dont do it... hahahahha i waited and think about this like 1min and just say ahh dont care... you can improve and changed the logic all the time

if you want to stop the spread of hacked clients maybe ban the voldermort client message /:

its always a war between antibot and cheaters

<ChillerDragon> i censored it on my servers

yea i will now hide my thoughs in silence

<ChillerDragon> new trend now is to kick players and the vote reason is a bot ad

1

Daniel

its always a war between antibot and cheaters

fair, but if you are doing things where you are testing clients, youll start impacting normal players and not stop anyone for very long (a direction detection would be instantly fixed)

even my ddos protection approach was public telled by me all the time.. nobody was interessted so i done it myself and it kinda work

ws-client

<ChillerDragon> new trend now is to kick players and the vote reason is a bot ad

never works tho

<ChillerDragon> the best protection against ddos is empty servers

1

ws-client

<ChillerDragon> the best protection against ddos is empty servers

5h into a playthrough, gets ddosed

13:49

damage is done

13:49

\; (edited)

Solly

fair, but if you are doing things where you are testing clients, youll start impacting normal players and not stop anyone for very long (a direction detection would be instantly fixed)

since you dont impact them since its based on some ticks and you teleport them back to the old state its okay... you can even trigger if they on plat or whatever...

many servers with equal players works yeah, but rarely do multiple of the same map appear

13:50

except linear ):

Daniel

since you dont impact them since its based on some ticks and you teleport them back to the old state its okay... you can even trigger if they on plat or whatever...

itll flash peoples screens which is slightly annoying, although if you have some less good detection beforehand

13:50

will be lessso

13:50

try it and see how many complaints you get

yea the ban messages will be changes today... but the kick instead of the ban will done in the next days

Solly

gah sdl3 has missing docs

Did you read https://github.com/libsdl-org/SDL/blob/main/docs/README-migration.md ?

if you get flashed only if you sus and only sometimes its okay. even if you standing on plat or whatever

13:52

okay time to play some geogeussr

Robyt3

Did you read https://github.com/libsdl-org/SDL/blob/main/docs/README-migration.md ?

now i have, but not very useful

if something is interesting with antibot i would appreciate a ping

theres lots of functions which require the window object but previously didnt

13:52

and uhh

13:52

no one has the window object

13:53

modularity and all that, but everywhere is either implictly using the window or squirriling it through so i dont see the reason to keep it modular when its just a facade

13:53

^ cry about it

13:54

i mean i could dynamic_cast

13:54

might be considered witchcraft tho

Solly

theres lots of functions which require the window object but previously didnt

Which functions?

settings mouse mode

You could look at https://github.com/libsdl-org/sdl2-compat and see how that uses SDL3

i assume it just keeps the most recent window

13:56

or first

13:56

whichever was the bheaviour in sdl2

yeah, that's why I said to look at the code, then you don't have to assume anymore

it doesnt make any difference, it has to store the window somewhere (because it has to call the equivilnat sdl3 func) (edited)

13:57

and ddnet only uses a single window

[ddnet/ddnet] Issue opened: #9680 Add ddnet protocol extension for PlayerFlags that does not require an alive tee

There is a thing in the teeworlds protocol which is called a m_PlayerFlag but ironically its not a flag for players but just for characters. Should be called CharacterFlag if you ask me. Anyways a real playerflag would be nice. To know what for example spectators are doing. Are they in the menu? scoreboard? chatting? This then allows to show the typing chat bubble for spectators on the cl...

Solly

settings mouse mode

Which function exactly? These don't take a window

extern bool SDL_SetRelativeMouseMode(bool enabled);
extern bool SDL_GetRelativeMouseMode(void);

they do now?

14:00

bool SDL_SetWindowRelativeMouseMode(SDL_Window *window, bool enabled)

Looks like it previously used the Window returned by SDL_GetKeyboardFocus()

that still works, but it would be better to use just the window

I guess the graphics backend just has to expose the window (as a void *) and then you cast back to SDL_Window * to avoid the dependency on SDL in the engine API

<ChillerDragon> @Daniel nice!

Robyt3

I guess the graphics backend just has to expose the window (as a void *) and then you cast back to SDL_Window * to avoid the dependency on SDL in the engine API

i was thinking to dynamic case the graphics backend

14:04

then error if its not sdl (which it wont be)

14:04

so if there is another backend it wont segfault

14:04

dynamic cast*

14:05

(and also doesnt require squiriling sdl_window thru a bunch of inits which is ... messy)

dynamic_cast is usually wrong if you don't know what you are doing

Robyt3

dynamic_cast is usually wrong if you don't know what you are doing

all backends created will always be of sdl3, if its not then that should be an error

Better not expose the window pointer at all, add a function to the graphics backend to call this SDL function

rather than ssuming it is

14:06

thats fair

14:06

but there are so many sdl things called

Same as Graphics()->SetWindowGrab which is probably similar

that it would be ... massive

14:06

and id rather be it done in a seperate pr (edited)

14:07

(its kinda wack to just make the things which require window into graphics)

14:07

(instead of all sdl stuff)

14:07

(but then id avoid dynamic cast which u dont like)

14:08

(but there still alot of methods)

yeah, changing to SDL3 would be a massive change anyway, it's not going to be done in a few hours. we also have a PR already: https://github.com/ddnet/ddnet/pull/9624

sdl3 by edg-l · Pull Request #9624 · ddnet/ddnet

Work in progress, still doesnt compile, some stuff i dont know how it works like IME thats why im opening a draft pr if anyone wants to contribute further. https://github.com/libsdl-org/SDL/blob/ma...

ive got that pulled

14:08

there are an innumerate amount of errors

14:08

which im slowly going through

I'd rather have all SDL3 migration in one PR though, but separate commits

14:09

Ideally all the automatic changes with the renaming scripts from the migration guide would be separate commits

how would i add to the pr anyway since i cant write to it

14:09

make a pr to the source branch?

yeah, possible

rememeber when ddnet changed ID to Id (edited)

14:12

sdl has done the opposite (edited)

my run was canncelled here? https://github.com/ddnet/ddnet/actions/runs/13333782464/job/37244156201?pr=9670

Feature Request: New bugfree speedtiles · ddnet/ddnet@09eb4e6

DDraceNetwork, a free cooperative platformer game. Contribute to ddnet/ddnet development by creating an account on GitHub.

Image attachment

14:28

uhh

14:28

almost

Assa

my run was canncelled here? https://github.com/ddnet/ddnet/actions/runs/13333782464/job/37244156201?pr=9670

Maybe a network timeout, you should be able to re-run the failed jobs:

Image attachment

what a great day it is today owo

owo

Image attachment

Robyt3

Maybe a network timeout, you should be able to re-run the failed jobs:

I am not

ok, maybe only maintainers can do it, I restarted it

merci ❤️

it works

<ChillerDragon> @Robyt3 did you guess the commit here? or is it in the rtp? https://github.com/ddnet/ddnet/issues/9674#issuecomment-2659622079

Server crash using hot_reload · Issue #9674 · ddnet/ddnet

15:14

<ChillerDragon> omg yes its in the file name i see it now aaaaaaaaaaa

The commit hash is in the crash dump filename

<ChillerDragon> ye but not inside the rtp right?

15:14

<ChillerDragon> i got trolled looking into the file

15:14

<ChillerDragon> the trace confuses me

No, the author of Dr.Mingw was against adding a function for this I think

<ChillerDragon> troll ngl

15:14

<ChillerDragon> https://github.com/ddnet/ddnet/blob/c51ef72e59b4c962bba70936c16d5240e3dc2a79/src/game/server/save.cpp#L19

ddnet/src/game/server/save.cpp at c51ef72e59b4c962bba70936c16d5240e...

DDraceNetwork, a free cooperative platformer game. Contribute to ddnet/ddnet development by creating an account on GitHub.

15:14

<ChillerDragon> how is line 19 reached without crashing in 16 17

15:15

<ChillerDragon> maybe 17 already died but UB moment did not let it crash yet?

The lines don't seem to match exactly

15:16

It crashes in the str_copy call on line 17

<ChillerDragon> i see okay that makes sense

ive fixed few hot reload crashes with https://github.com/ddnet/ddnet/pull/9160

<ChillerDragon> @kebs can you consitently reproduce the crashes you fixed? maybe you can reproduce the issue in the version he reported and if it was fixed we can close it

yes it consistently crashed before my pr

15:33

dont know what that issue did so idk if its the same

15:34

crash 1: join team, join another team, hotreload = crash crash 2: dummy join team 1, main character join team 2, hotreload = crash

@ChillerDragon how does team colors nameplate no worky

16:11

[it works?]

Image attachment

Which client is that?

Replying to @Solly ChillerDragon how does team colors nameplate no worky

16:34

I linked a chillerbot-ux bug not ddnet

Updating arch did not go so well

<ChillerDragon> https://zillyhuhn.com/cs/.1739638601.png

Embedded image

Updating neovim plugins didn't go well

[ddnet/ddnet] Pull request opened: #9681 Fix uninitialized CCamera::m_IsSpectatingPlayer variable

Randomly caught by UBSan in the CI: ``` /home/runner/work/ddnet/ddnet/src/game/client/components/camera.cpp:271:5: runtime error: load of value 190, which is not a valid value for type 'bool' 0 0x561f7a6063ab in CCamera::UpdateCamera() /home/runner/work/ddnet/ddnet/src/game/client/components/camera.cpp:271:5 1 0x561f7ae91b0f in CGameClient::OnRender() /home/runner/work/ddnet/ddnet/src/game/client/gameclient.cpp:801:11 2 0x561f7a154053 in CClient::Render() /home/runner/work/...

@ChillerDragon was #9021 fixed? i cant reproduce

https://github.com/ddnet/ddnet/pull/9021

Fix hot_reload losing spectator state by ChillerDragon · Pull R...

This fixes the bug that moved spectators in game on hot_reload Checklist Tested the change ingame Provided screenshots if it is a visual change Tested in combination with possibly related confi...

github-merge-queue[bot]

[ddnet:master] 2 new commits

8322e0f Fix uninitialized CCamera::m_IsSpectatingPlayer variable - Robyt3 809b0df Merge pull request #9681 from Robyt3/Client-Camera-UBSAN-Fix - Chairn

17:51

[ddnet/ddnet] Pull request closed: #9681 Fix uninitialized CCamera::m_IsSpectatingPlayer variable

[ddnet/ddnet] Pull request opened: #9682 Fix editor ctrl+right click selecting hidden layers/groups

Fixes #9626

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 coverage to integration test
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/...

<ChillerDragon> it might have been fixed yes

18:12

<ChillerDragon> @kebs yes its fixed!

i guess that hotreload pr, or resetting persistentdata in persistent whisper pr fixed it

<ChillerDragon> very nice kebs

18:30

<ChillerDragon> arch is unstable they say but i have to admit reading the arch wiki was super smooth but getting a kernel panic on debian is more scary

<ChillerDragon> ip grabber at? 134.122.64.174:8303

19:01

<ChillerDragon> impersnating Cup Of The Day server

19:01

<ChillerDragon> collecting ips for ddosing the tournament thats starting now i assume can we do some master banning?

19:01

<ChillerDragon> @deen @heinrich5991

Why am I suddenly getting several valgrind and ubsan errors regarding uninitialized variables (ERROR SUMMARY: 4166 errors from 21 contexts (suppressed: 0 from 0)) from seemingly unrelated changes? pepeW

pepeW

Could changes in the overall memory layout cause this?

maybe valgrind and ubsan can detect new things

Looks like the new branches and old branches use the same versions of clang and valgrind though

19:21

Ohh, I added a CGameClient constructor. Maybe the default when no constructor is used defaults to zeroing the memory?

github-merge-queue[bot]

[ddnet:master] 2 new commits

2ad6e9c Fix editor ctrl+right click selecting hidden layers/groups - KebsCS d9a0704 Merge pull request #9682 from KebsCS/pr-fix-ctrl-rightclick-hidden - Robyt3

19:57

[ddnet/ddnet] Pull request closed: #9682 Fix editor ctrl+right click selecting hidden layers/groups

19:57

[ddnet/ddnet] Issue closed: #9626 Editor: ctrl+right click should not select hidden layers or groups

Ok, looks like a lot of variables are only initialized due to CGameClient's value initialization because there is no constructor

@kebs: epic fixes

ty

is #9512 ready to be merged btw?

https://github.com/ddnet/ddnet/pull/9512

Improving the usability of automappers by VoxelDoesCode · Pull Requ...

Summary As of: Feb. 8, 2025: Previously on Jan. 12, 2025 This PR focuses on making the automapping feature much more manageable and flexible for mappers and map designers. The features included/pla...

does this add automap selection* (edited)

no

santatrollet

20:54

but someday i'll find a way to add it

"pls decline pr, just make new layer and automap only the new stuff you made. smh average dev trying to bloat up ddnet!!!"

where to report fake servers? 46.174.48.103:53010 fake clients

46.174.48.103:53010[

] is an official TeeFusion (Type: Block) server. TeeFusion is NOT affiliated with DDNet. Quick join: https://ddnet.org/connect-to/?addr=46.174.48.103:53010/

fokkonaut

where to report fake servers? 46.174.48.103:53010 fake clients

hm, it seems there is something wrong with it's register code rather than it faking players

nah

21:45

why do you think so

21:45

?

Robyt3

They also wrote a blog post on fuzzing the Teeworlds network: https://logicaltrust.net/blog/2020/07/socketfuzzer.html Fuzzing the map format isn't really worth it unless for learning, you should easily find crashes everywhere because the format is mostly unchecked at the moment (edited)

yea i saw

@Souly automapper update is getting merged rn

23:22

next nightly will go hard af

1

github-merge-queue[bot]

[ddnet:master] 4 new commits

3946608 WIP on adding a reference layer to automapping - VoxelDoesCode b36a68e Add modulo function inside automap logic - VoxelDoesCode 425bbdc Big automap code touchups, several related bug fixes - VoxelDoesCode 658877e Merge pull request #9512 from VoxelDoesCode/better_automap - Robyt3

23:34

[ddnet/ddnet] Pull request closed: #9512 Improving the usability of automappers

23:34

[ddnet/ddnet] Issue closed: #9590 Automapper rules are duplicated when re-importing an image

Exported 330 message(s)

Timezone: UTC+0