DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.tw/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 2020-03-19 00:00:00Z and 2020-03-20 00:00:00Z

[ddnet-maps:master] 1 new commit

225f7cc A Brosty - ddnet-maps

Does anyone even know how input works? How did things get so convoluted even? What is a OnPredictedEarlyInput?

Guess thats for syncing weapon input (edited)

11:39

with other inputs*

11:40

nuborn added that to ddnet

hmm nuborn isn't here anymore? It'd be nice if he could give me a quick runthrough of how everything ties together

Yes, sadly. why is he gone tho?

https://github.com/ddnet/ddnet/pull/1441

Sync weapon input with other input by trml · Pull Request #1441 · d...

Currently weapon input is applied immediately when it arrives at the server, while hook/movement inputs are buffered and not applied until the tick the client predicted them for. This pr makes the ...

11:57

I think this should be renamed to OnDirectEarlyInput

well, it is not direct anymore so the name is kinda okay

so #1441 made DirectInput be deferred like the predicted input?

yes

and OnClientPredictedEarlyInput sorta simulates the previous behaviour?

@Learath2 I'm still here sometimes

15:46

yes, it simulates it to keep rocket jump physics and similar the same

15:52

OnDirectInput applies weapon input immediately when it reaches the server. in theory you could just remove anything except OnPredictedInput and handle everything related to weapons there (then there would also be no desync). But that would change physics since ondirectinput also has other side effects

Huh, couldn't even highlight you, weird

Oh, I see, yes

I couldnt too, Learath

@Learath2 is ger2 hosted by you?

17:16

i think ger2 hates me, i have rly bad ping to it

Image attachment

3

3

3

17:16

ingame too

@Ryozuki @heinrich5991 hosts it

oh well ingame looks fine now

17:17

but one day it lagged a lot and other players were fine

17:17

and ger1 worked perfect

@heinrich5991 apparently the way we do threaded rendering is quite illegal

illegal? ^^

18:52

shh, don't tell the cops

@heinrich5991 well SDL doesn't really allow for the window/gl context to be created on a different thread then the one that does the rendering

19:42

turns out macOS is much stricter about it the last couple versions

@Learath2 u can create it on the main thread and on another thread call SDL_GL_MakeCurrent(wnd, ctx) to make it only be usable there

19:56

https://wiki.libsdl.org/SDL_GL_MakeCurrent

19:57

well thats what i've seen

That's what everyone seems to assume but I've never seen any SDL document that says it's allowed

19:58

and on macOS it's verynotallowed

isnt it more about opengl?

19:58

https://www.khronos.org/opengl/wiki/OpenGL_and_multithreading

OpenGL and multithreading

19:58

Can I create an OpenGL context for a control in a different thread than the one in which the was created on? Yes. According to the Microsoft documents (MSDN), there is no restriction. You can probably do the same on *nix systems as well.

19:59

looks like its valid

19:59

might be a macos thing

19:59

they are special

19:59

20:00

is there even a perfomance increase having the render calls in a thread?

20:00

brb eat

@heinrich5991 can you be here for a second?

I'm playing rn

20:08

can't really

sure

20:13

The CMakeLists on ddnet uses bundled libraries on mac for some reason and cmake is impossible to debug

sdl isn't available by default

teeworlds CMakeLists detects the one I installed just fine

ah

20:14

not with that attitude

I am extremely annoyed after digging through the mess that is SDL internals and reading objective C for the last 4 hours

Image attachment

Perfect, the bundled version of sdl2, through whatever miracle works just fine, the version I build and the version that ships with osx breaks

20:20

Actually nevermind, now I can't get that working either

20:21

I guess I'm resigning from all development duties because I can no longer get a working build of ddnet nor teeworlds on my computer

dont rq

20:22

building tw is hard

macos face rn troll

troll

It doesn't work, it's not even supposed to work, yet it worked 10 minutes ago

since when r u having build issues tho? werent u just debugging the resoltion bug?

20:23

oh since 10 m

that's a sign that I should just stop

ye take a break and watch some teeworlds programming beginner tutorials

20:24

https://www.youtube.com/watch?v=HB_aSIOmdu4

Teeworlds programming #001 - hello world server

20:24

lmao

Look what we are doing is wrong. A Cmd_Swap after a resize triggers a [NSOpenGLContext update] which is not allowed outside of the main thread, end of story

20:24

yet it worked with ddnet 10 minutes ago

20:25

there is no taking a break and coming back, it's not supposed to work, it's not working, it will never work

._.

https://wiki.libsdl.org/CategoryThread

20:40

NOTE: You should not expect to be able to create a window, render, or receive events on any thread other than the main one. For platform-specific exceptions or complicated options ask on the forums/mailing list.

CoreFoundation is a 15MB binary, it absolutely breaks radare2 :P

turns out doing SDL_GL_Swapwindow in a thread other then main is not ok on newer macOS, but our prebuilt clients are linked to an older sdk and thus AppKit suppresses the error to keep backwards compatibility

23:40

"2020-03-20 02:36:46.173 DDNet[19022:302085] NSOpenGLContextSuppressThreadAssertions=YES"

@ChillerDragon are you also on Catalina? can you please check whether it also behaves as such for you?

I do not have my Mac at home :/

23:42

But yes I’m Catalina

Well catalina and onward the way we do threaded rendering doesn't work unless the executable is linked to a SDK older then catalina

Sounds bad

"head over to your favorite browser" @ChillerDragon Kappa

Kappa

Image attachment

Exported 84 message(s)