DDraceNetwork - developer

ok tune lock tile is ready ill just wait until #9788 is merged

3

https://github.com/ddnet/ddnet/pull/9788

Fix jetpack tuning and incorrect tune zone override by KebsCS · Pul...

Fixes #9753 I don’t know why fake tuning was always sent instead of the real one, since the very first jetpack prediction commit: 38b16ee Now it matches other fake tunings and seems to work correct...

kebs

ok tune lock tile is ready ill just wait until #9788 is merged

01:24

we're getting tune lock before silksong i cant believe it

kebs

ok tune lock tile is ready ill just wait until #9788 is merged

it should also supercede #5732

https://github.com/ddnet/ddnet/pull/5732

Add tune-lock (closes #4651) by fokkonaut · Pull Request #5732 · dd...

This pull requests adds the tune-lock tile and a tune lock reset tile. Tune lock is index 69 and tune lock reset is index 70 from tune layer. Tune zones have a higher priority than locked tunings, ...

:D

I didn't see this before

thats.. old kekw

I've often tried to push things people wanted but devs/admins didn't see/wanted to, so yeah might not be best implementation and not 100% finished, but on my server it works really well since that PR is out

fokkonaut

I've often tried to push things people wanted but devs/admins didn't see/wanted to, so yeah might not be best implementation and not 100% finished, but on my server it works really well since that PR is out

we live in a society

wait what was the decided behavior for tune lock overrides?

02:30

i think tune lock should always be applied on top of the current tune zone if there is one

02:31

wow hecta dev role should've had it since a while ago xd

no, tune zones should have higher priority, as discussed with @Im 'corneum before. It's way more usable

louis

wow hecta dev role should've had it since a while ago xd

i wanted to be role-less ngl

fokkonaut

no, tune zones should have higher priority, as discussed with @Im 'corneum before. It's way more usable

how so?

fokkonaut

no, tune zones should have higher priority, as discussed with @Im 'corneum before. It's way more usable

oh yea i can kinda see why

02:46

so you be tuned but still have tune zones affect you

how would you make a tunezone in which players without a tunelock have a 0 force grenade, but the guy with tunelock has a normal nade

02:47

if tune zones have higher priority

they should both be added on top of eachother, but i think tunelock tunes for a specific specific_tune_setting should take priority over a tune zone's specific_tune_setting

louis

how would you make a tunezone in which players without a tunelock have a 0 force grenade, but the guy with tunelock has a normal nade

i almost said server tunes but im assuming the server should have normal grenade physics to begin with

server tunes should just be an abstraction for tunezone across the whole map

this is a very tricky situation because:

if tune lock has priority, entering a tune zone for the same tune doesn't do anything
if tune zones have priority, both tees (one tuned and the other not) will have the same exact tunings

05:06

*both tees inside it

i think tune lock prio makes sense

05:10

@Im 'corneum what parts were u trying to make where its worse?

ddnet-maps

[ddnet-maps:master] 1 new commit

2ad595b A Get2Carry - ddnet-maps

entasi

[ddnet/ddnet] Issue opened: #9790 Connection problems

Hello, i have an issue: While playing i receive Connection problems and freeze for 2-3seconds, extremely annoying, happens randomly between 15-30mins. It freezes only me, so its not ddos or server issue. Started ~month ago. Sometimes i get random timeout, predping increases to -927 and i cant connect to server again, reentering game helps enter server and restore session with timeout. Ill try to record Connection problem moment later, if needed. I have perfect lan connection, so i dont know w...

louis

@Im 'corneum what parts were u trying to make where its worse?

I dont remembee

bor for gor

Pioooooo

Click to see attachment 🖼️

this is cool i like this

Hecta

this is a very tricky situation because:

if tune lock has priority, entering a tune zone for the same tune doesn't do anything
if tune zones have priority, both tees (one tuned and the other not) will have the same exact tunings

Should new tune lock override old one, or only apply when no override? So in the second case have to unlock with the corresponding unlock tile and then tune lock works

ok i think ill rewrite to 2nd so this setup works just for one player

This is how slides would look like brownbear

arrows pointing in the opposite dir still triggers tho

Assa

This is how slides would look like brownbear

will this break old clients?

always

will this break old clients?

they wont have prediction for it

aa so itll break their antiping

12:23

will we need new entities?

always

aa so itll break their antiping

also the local character prediction will not look smooth

Jupstar ✪

arrows pointing in the opposite dir still triggers tho

@Hecta

always

will we need new entities?

speedup arrows are not part of the entities

always

will we need new entities?

no this is part of the speedlayer

12:24

you enter the angle in the ui, so the slides have 360 degrees of freedom, force and max speed is not needed

oo thats cool

only downside so far, they are skippable and there is nothing I can do against this, they are not slopes (edited)

2

12:28

I could implement true slopes, but the prediction isn't the only thing I'd break with true slopes justatest

vButtonsContainersSettingsControlsJoystickAbsolute

Java

im not allowed to use static anymore

12:35

a simpler time

12:40

wait does a vector of 0 size elements have no size per element? &vec[1] == &vec[2]

if anyone needs a review, ask me. I am constantly asking for reviews and want to give something back

Solly

wait does a vector of 0 size elements have no size per element? &vec[1] == &vec[2]

no its padded to 1 byte or more

doesn't a template vector have a type, shouldn't this be the size of the type?

Solly

no its padded to 1 byte or more

if you're using a struct, it's usually padded up to 4 bytes if you don't use the right pragmas for it

std::vector<CButtonContainer> vButtonsContainersNamePlateKeyPresses = {{}, {}, {}, {}}; class CButtonContainer{}; sizeof(CButtonContainer) != 0

12:43

friend mark

12:44

1

12:58

thanks git very co0ol

btw is there a reason we are not support VS 2022?

Assa

This is how slides would look like brownbear

waoh this is awesome

13:04

best new tile I've ever seen

Assa

btw is there a reason we are not support VS 2022?

we do?

13:05

I use vs 2022 exclusively

[cmake] CMake Error at CMakeLists.txt:26 (project):
[cmake]   Generator
[cmake] 
[cmake]     Visual Studio 16 2019
[cmake] 
[cmake]   could not find any instance of Visual Studio.
[cmake] 
[cmake] 
[cmake]

VS 2022 is installed?

13:05

I get this error in VS-Code

there is some way to pass an agurment to tell it to use vs 2022 but I don't use vscode so idk how

I see the variable, gonna search

13:07

-G "Visual Studio 16 2019"

13:09

just a configuration issue, when I switch to 2022 I get an error ...

what error

"EBUSY: resource busy or locked, unlink 'c:\Users\User\Desktop\workspace\ddnet\build\CMakeFiles\CMakeConfigureLog.yaml'".

(edited)

-# windows moment (edited)

if you're not using the vs 2022 IDE why not just install vs 2019 for the toolchain

13:11

then you won't have issues

or your cmake version is old

because windows doesn't like multiple vs versions

it would have to be pretty old

13:12

I have 4 VS versions...

i usually just keep one

>cmake --version
cmake version 3.31.0

so cmake just auto select the right one

maybe it's too new

Just generate the cmake manually?

maybe just cleanup and manually specify generator

There is a cmake gui

i have 3.30.5 works fine

13:13

in vs 2022 you just open cmake project and thats it

although to be fair i just used the vscode cmake extension it just list all the compilers and let me click on one

13:14

i haven't installed any actual visual studio for years. i only install vs build tools

I deleted the CmakeFiles directory, now I am just missing ffmpeg apprently

....

[cmake] CMake Error at CMakeLists.txt:1887 (file):
[cmake]   file COPY cannot find
[cmake]   "C:/Users/Marvin/Desktop/workspace/ddnet/ddnet-libs/ffmpeg/windows/lib64/avcodec-61.dll":
[cmake]   No error.

ffmpeg should be included in ddnetlibs

name leak!

oh that's on me

is the folder empty

hi Marvin

yeye I already leaked my proffesional github account, deal with it

13:15

hi

love you

I love me too

1

ok continue the troubleshoot. i'm just here to say hi

just a missing git submodule update

wait ddnet is supposed to handle ime?

13:20

darn

i mean ye..

13:20

chinese player base was big

an application can be completley blind to ime and work pretty functionaly

13:20

its nice to have text editing

pretty sure it is toolkits job

but ddnet stores the candidates, so i pressume they are rendered in ddnet

but we obviously aren't using a UI toolkit

13:21

ye

13:21

only on windows tho

not anymore B)

in SDL3?

yeah

13:21

now its cross platform

13:22

so bye bye windows restriction

it's not really a restriction

13:22

it's a exclusive fullscreen workaround

and you can should be able to choose between "native" (ibus) and ddnet

TsPiggy

it's a exclusive fullscreen workaround

oh right that

macos always composite so using a in-engine candidate is actually worse than letting system handle IME

xorg and wayland no longer have the exclusive fullscreen restriction either

not sure about linux. i think compositors always have a way to integrate with ibus or fcitx5

but i think you should be able to choose between ddnet and native

13:23

its not like ime candidates are that hard to render

so ye, better just kept it windows exclusive in that case. cuz people can really hate the in-game candidate list

13:23

you can already

TsPiggy

so ye, better just kept it windows exclusive in that case. cuz people can really hate the in-game candidate list

you can always turn it off

there is a setting to bypass the in-game one.

yes

but i do think default to the ime render player usually see is better

13:24

i kinda want to make ddnet also automatically give up on render candidates with windows mode

except on windows

I managed to get it working, I needed to manually delete cmake cache and all older build directories

in exclusive fullscreen

because windows moment

windowed fullscreen on windows is weird

13:25

the IME works on some machine but not others

its just a window without decoration at full res?

maybe sometime driver might just decide that it should take over the whole screen anyway

13:26

to reduce latency for windowed fullscreen

13:26

i'm guessing

well since "exclusive fullscreen" is no different for me anymore

13:26

i assume all issues with it are gone

13:26

not that i ever had any

debugging the server crashes immediatly ... ........

segfault?

accidental anti-debugger. ~~we should make that a feature and sell them to commercial project~~

DDNet-Server.exe (17248): Loaded 'C:\Users\Marvin\Desktop\workspace\ddnet\build\release\DDNet-Server.exe'. Module was built without symbols.
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\ntdll.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\kernel32.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\KernelBase.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\apphelp.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\shlwapi.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\msvcrt.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\ws2_32.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\rpcrt4.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\version.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\ole32.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\ucrtbase.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\combase.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\gdi32.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\win32u.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\gdi32full.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\msvcp_win.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\user32.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\shell32.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\advapi32.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\sechost.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\winbuilds\bin\sqlite3.dll'. 
DDNet-Server.exe (17248): Loaded 'C:\Windows\System32\bcrypt.dll'. 
The program '[17248] DDNet-Server.exe' has exited with code -1073741515 (0xc0000135).

13:28

what why does it use release?

13:28

WHAT

13:28

-DDEV=ON maybe?

13:28

even then debugger shouldn't crash a release build server tho

DDNet-Server.exe (2852): Loaded 'C:\Users\Marvin\Desktop\workspace\ddnet\build\debug\DDNet-Server.exe'. Symbols loaded. Still crashes the exact same way

omg

13:30

DeepSeek-R1 could be used for coding teeworlds code.

terminal time:

2025-03-02 14:30:17 E server: failed to load map. mapname='dm1'   
(base)

13:30

xD

Bamcane

DeepSeek-R1 could be used for coding teeworlds code.

I asked it something, and then it even know that needing to modify bam.lua

oh is that one of the

13:31

fully automated

probably not devin

However, sadly, it couldn't be used in my project about teeworlds.

Now I got it running

is it known, that IsGrounded is 1 pixel off?

13:47

I did:

if(IsGrounded())
  Vel.y = 0;

And I am flying 1 pixel in the air

something something gravity or something

13:54

Assa

is it known, that IsGrounded is 1 pixel off?

never hang in a 1tile gap? it even makes the joump sound even tho your feet are in air

#if defined(CONF_PLATFORM_ANDROID)
    // Translate the Android back-button to the escape-key so it can be used to open/close the menu, close popups etc.
    if(Key == KEY_AC_BACK)
    {
        Key = KEY_ESCAPE;
    }
#endif

ac back is not android specifoc and could be generated by a funky controller or keyboard https://usb.org/sites/default/files/hut1_5.pdf

Solly

#if defined(CONF_PLATFORM_ANDROID)
    // Translate the Android back-button to the escape-key so it can be used to open/close the menu, close popups etc.
    if(Key == KEY_AC_BACK)
    {
        Key = KEY_ESCAPE;
    }
#endif

ac back is not android specifoc and could be generated by a funky controller or keyboard https://usb.org/sites/default/files/hut1_5.pdf

but the feature is also only active on android xd

Jupstar ✪

but the feature is also only active on android xd

right now it is

14:11

but theres no reason not to handle it on other platforms

Solly

but theres no reason not to handle it on other platforms

ah so you want it to be used for controllers?

Jupstar ✪

ah so you want it to be used for controllers?

a funky controller could

14:11

i dont use controller

14:17

https://wiki.libsdl.org/SDL3/SDL_HINT_IME_IMPLEMENTED_UI candidates doesnt seem to disable native rendering

SDL_EVENT_TEXT_EDITING_CANDIDATES waiit no way, did sdl actually implement this?

1

14:33

I looked into this for several weeks seeing if I could implement it for SDL but there just was no way to get the candidate list out on macOS

14:33

Nor on linux

14:33

I wonder how they managed to do it

Pioooooo

[ddnet/ddnet] Pull request opened: #9791 Refactor math.h

Remove redundant 'inline' specifier on a constexpr function. Change in_range function to return boolean.

Checklist

[ ] 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 map...

Learath2

Nor on linux

all the ime methods probably have thier own ways, it just wasnt implemeneted

14:34

but now theres enough to do so

14:34

maybe macos is still borked

If they didn't implement it for anything except windows I'm very very disappointed

that could be the case because its not hiding the candidate ui for me with ibus (edited)

on macOS I managed to extract the candidates directly from memory by disabling many security measures and injecting myself into appkit, definitely not a proper solution

(edited)

on Linux there is actually 0 documentation, none of the used IMEs have any of their own documentation, all of their irc channels are dead too

14:38

at that point I was like damn, I'm not wasting more time on this shit

14:40

There is some standard for it inside X iirc but it wasn't enough

welp

14:43

@Learath2 do you happen to have a windows machine which can compile ddnet available to you

Solly

@Learath2 do you happen to have a windows machine which can compile ddnet available to you

sure, why?

can you try my sdl3 branch

14:44

and see if ime works

Tomorrow I can

https://github.com/SollyBunny/ddnet/tree/sdl3 sure