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 2022-07-21 00:00:00Z and 2022-07-22 00:00:00Z

Thats cool: https://teeworlds.leaningtech.com/teeworlds.html

01:38

they did that with cheerp

[ddnet-maps:master] 1 new commit

e90a154 A Dynamic Dragger - ddnet-maps

Patiga

@Jupstar ✪ is there a reason why I shouldn't use vertex buffers as uniform buffers? not sure what my thought process was there, but its what I did a few times

I guess doesnt matter too much. Maybe the driver puts uniform buffer in faster memory locations . but probably not since the limit is rather huge nowadays

06:57

Along wgpu allows this at least ^^

(Especially for an opengl backend which might care more about it)

[ddnet-web:master] 1 new commit

42cf57f Add 20€ donation by Chris ❆ - def-

@Learath2 Why dont we introduce new snap objects for entities we use in DDNet, that requires specific information? Why do we add an EntitiyEx, that requires another snap object and therefore fills up the already quite small snap very fast with many objects? We have the extended net message and object system, i think it would be the right way to go to actually use it for more things

11:22

I mean, I see the idea of entitiyEx of course, but for me it makes more sense to have 1 object, that includes all the information, rather than 2 objects that the client has to combine later on again.

11:22

Especially since we are not limited to new objects anymore

11:24

I'm thinking about it, because on my server for example I did an optimization to see more entities, that I only send entityEx for close objects, so that eventually prediction will work correctly. But this will need to be changed with the upcoming laser color change, because otherwise lasers further away wont have the correct color as I am then only sending the laser itself. Same goes for spectating (pause or spectator), I dont send entitiyEx there, because it doesnt predict anything anyways in this situation. But now for the laser color, it is indeed needed, also when speccing

11:25

All of this wont really be a problem, once we have extended snapshots tho, but I think we should consider taking one of those options sooner or later.

i wish i understood the predicition thing better feelsbadman

feelsbadman

prediction is basically just serverside stuff running on the client to predict it. In order to predict it correctly tho, the client needs some information from the server

11:26

for example, if you are on a laser door which is closed, your tee will look bouncy

11:26

because its not predicted

11:26

it thinks it can fall there, cuz there is no wall

11:26

but the server puts your position back up, because the server says nop, the door is closed, you are ON it and not falling

11:27

So the client for example needs information on whether the door is closed or not right now, in order to predict it properly

so the server says door closed and the client can then predict that it must be on top of the laser?* (edited)

basically yea, and the server also needs to send the client when the door is open again, because otherwise the client thinks its closed and it would think you have to be on top of it, it would create a small stutter when going through it otherwise probably (laserdoor prediction is not yet in anyways)

aight aight i get the concept

And antiping especially is a very extended prediction. It includes prediction for player inputs and weapons, for example on a server with 200-300+ ping, it takes quite some time until your shot grenade bullet actually spawns etc, with antiping, the client tries to predict it before it receives the actual bullet from the server. That works, because the client knows about tunings from the server, and it knows which weapon you're holding, and therefore can do some server calculations on the client

okay yeah i always wondered if predicition = antiping

And thats also, why players seem to jiggle a bit with antiping, cuz when it notices the predictions are wrong, and we have actually received info about the other player from the server now, it puts the player in the correct position where it belongs

happy

1

but predicition is not only used for antiping

11:31

okay

bluesky

but predicition is not only used for antiping

Exactly

11:31

hookthrough for example, also uses prediction, or stoppers, but its not antiping

11:32

being smooth on top of a stopper and not jiggling like on a laser door, thats prediction

11:32

and the hook going smoothly through the block, thats prediction

fokkonaut

And antiping especially is a very extended prediction. It includes prediction for player inputs and weapons, for example on a server with 200-300+ ping, it takes quite some time until your shot grenade bullet actually spawns etc, with antiping, the client tries to predict it before it receives the actual bullet from the server. That works, because the client knows about tunings from the server, and it knows which weapon you're holding, and therefore can do some server calculations on the client

and the better the prediciton, the smaller the stutter as soon as the packet arrives (edited)

11:32

thanks for explaining it <3333

Well, you cant have a perfect prediction because you dont know when the player will stop hold A for example

11:33

but as long as he holds A, you can predict its position basically

fokkonaut

Well, you cant have a perfect prediction because you dont know when the player will stop hold A for example

yes

11:33

cant do more than guess

bluesky

thanks for explaining it <3333

music

I once had the idea of a splitted antiping, basically so that it would only predict other players when interacting with them, and otherwise relying on the (maybe late) info from the server

bluesky

music

no problem

fokkonaut

I once had the idea of a splitted antiping, basically so that it would only predict other players when interacting with them, and otherwise relying on the (maybe late) info from the server

Because hooks, and grenade explosions, hammers, all of that, especially when its coming from your own tee, can be predicted, and also other players interacting with it. + it would stop the jittering of other players as we dont really predict their movements alone

happy

1

Stupid question, but in order to resolve merge conflicts with your repo and master without causing errors in your build, would you need to do a pull request from master to your repo to update all the files? Or would that kind of break the pull request in master? (edited)

Voxel

Stupid question, but in order to resolve merge conflicts with your repo and master without causing errors in your build, would you need to do a pull request from master to your repo to update all the files? Or would that kind of break the pull request in master? (edited)

Rebase your branch

11:37

to upstream master

checkout in your master branch, update it to the latest ddnet master, then checkout your working branch and do git merge master i believe

11:37

then it will tell you which files have conflicts

11:38

and you have to manually fix them

11:38

actually havent done this is quite a time

oh this is kind of easy

Image attachment

11:40

i know what to work around fixing these conflicts, i just needed to know how to do it

lmao i just googled for the wrong programming language and did only realize it 20mins later when closing tabs. I made the answer work without noticing its wrong language xxxxxxxxxxxxxD

12:05

next level brain rott

UGHHHHH WHY DOES C++ HAVE TO BE SO HARD i resolved conflicting merges and i guess because of that the entire thing gives me 30 different compile errors (edited)

Image attachment

justatest

I would say if branches have diverged a lot it's easier to rebase on master than to merge master in your branch

false alarm, i accidentally created a static int CButtonContainer

12:25

what now

Image attachment

12:25

i didnt even do anything this time wtf

k2d222

I would say if branches have diverged a lot it's easier to rebase on master than to merge master in your branch

but then how do i test it

12:29

lol i think this program is telling me to just sleep at this point

Image attachment

12:31

Everything's seemingly updated in https://github.com/ddnet/ddnet/pull/5639 ... I don't have the brain capacity to figure out what I did wrong

Add different laser colors for different types by VoxelDoesCode · P...

Since we're doing a push to assist newer players, I think it would be a good idea to differentiate the different types of lasers, so that non-moving laser entities aren't confused with door...

12:32

all the bugs are in RenderDropDown, for some reason

Voxel

what now

it tells you what's wrong

12:33

make sure those arguments match what you have in the header

well i fixed it but now im looking at my commits and it really showed this is my first time coding #(

Image attachment

don't worry, git merging is a pain whether youre a newbie or a pro even if it gets a bit easier

hello i write this bind and now i cant move; bind x "+toggle cl_mouse_max_distance 2 VALUE1; +toggle inp_mousesens 1 VALUE2"

kada_anes

hello i write this bind and now i cant move; bind x "+toggle cl_mouse_max_distance 2 VALUE1; +toggle inp_mousesens 1 VALUE2"

hello, VALUE1 and VALUE2 should be replaced with the values you use for mouse distance and sensitivity, default values are 400 for VALUE1 and 200 for VALUE2

like this : bind x "+toggle cl_mouse_max_distance 2 400; +toggle inp_mousesens 1 200"

kada_anes

like this : bind x "+toggle cl_mouse_max_distance 2 400; +toggle inp_mousesens 1 200"

yes that should work

[ddnet/ddnet] Pull request opened: #5647 Fix null pointer / out of bounds accesses in chat and maplayers

See #5646.

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

[ddnet/ddnet] Issue opened: #5648 Get away from building stuff in python, our C++ code uses

I think it's not really nice style, IDE's find the wrong include and that's confusing. We can do most stuff with constexpr or (c++20) consteval.

Voxel

well i fixed it but now im looking at my commits and it really showed this is my first time coding #(

don't worry, this is fine, you can also squash commits together if you want to reduce number of commit to look more "pro"

This is driving me insane

Image attachment

15:57

Lol, it was so bad it took the screenshot after discord stopped hogging the cpu

15:59

Discord and spotify just randomly decide to use 100% of an entire core and for whatever reason the kernel just can't handle it

Learath2

This is driving me insane

what kind of CPU do u even have?

Intel i7 4770k

sounds good enough

It is a little old, but it shouldn't crumble under spotify ffs

maybe caching or smth?, ever tried a clean reinstall or a different computer?

16:06

i mean that these electron apps suck compared to native programs is obvious, but i think i've never seen them using 100% of one core

For almost half a minute too, it just won't let go

Jupstar ✪

i mean that these electron apps suck compared to native programs is obvious, but i think i've never seen them using 100% of one core

I've seen it on discord on macOS too

16:07

but there the kernel managed to schedule other stuff even though one core was being completely hogged

16:08

I'm starting to wonder whether I misconfigured this kernel, but it used to happen on arch too with the precanned kernel

Learath2

I've seen it on discord on macOS too

does it maybe scan files on your computer?, or is it CPU only? discord employee: while(WaitForMessage) ++i

Jupstar ✪

does it maybe scan files on your computer?, or is it CPU only? discord employee: while(WaitForMessage) ++i

I was thinking about this, so I got iotop too so I can check io usage when it happens again

I loled at "precanned kernel"

you can try to denice it to lowest level to prevent system unresponsiveness

17:07

this is what i sometime use for vscode because it either hogs cpu or whole memory

[ddnet/ddnet] Issue opened: #5649 Server SIGSEGV while playing the map

Info: I created a Docker image for the DDNet server, but the server crashed because of SIGSEGV when going through a certain part of the map. I checked with the 16.2.2 and 16.2.1 versions in Docker and outside of Docker. Also, i tried to get useful info from gdb, but gdb gave me this:

Thread 1 "DDNet-Server" received signal SIGSEGV, Segmentation fault.
0x00000000004bec4d in ?? ()

Reproducting issue: 1. Setup a DDNet server with map "newbee" 2. Tp to checkpoint 11 3...

actually, its more their cppservtools stuff that probably handles c++ files that is really cpu and memory hungry

[ddnet:master] 3 new commits

b4eee12 Fix null pointer access in maplayers when the snapshot has no game info - Robyt3 95636da Fix out of bounds access with highlighted messages in server demos - Robyt3 b1b1803 Merge #5647 - bors[bot]

17:32

[ddnet/ddnet] Pull request closed: #5647 Fix null pointer / out of bounds accesses in chat and maplayers

[ddnet/ddnet] Issue opened: #5650 UBsan runtime error when using `super` command

``` src/game/server/teams.cpp:484:35: runtime error: shift exponent -1 is negative 0x55d065b241c4 in CGameTeams::TeamMask(int, int, int) src/game/server/teams.cpp:484 0x55d0657f92ed in CCharacter::TickDefered() src/game/server/entities/character.cpp:821 0x55d065a5d04c in CGameWorld::Tick() src/game/server/gameworld.cpp:283 0x55d0659327e6 in CGameContext::OnTick() src/game/server/gamecontext.cpp:852 0x55d0656e6a88 in CServer::Run() src/engine/server/server.cpp:2807 ...

[ddnet/ddnet] Issue closed: #5649 Server SIGSEGV while playing the map

fokkonaut

@Learath2 Why dont we introduce new snap objects for entities we use in DDNet, that requires specific information? Why do we add an EntitiyEx, that requires another snap object and therefore fills up the already quite small snap very fast with many objects? We have the extended net message and object system, i think it would be the right way to go to actually use it for more things

It seems like, with too many lasers for example on my server, the client just cant do all the work in time, it starts to lag if you zoom out a lot.

19:00

Weird is, sometimes it can sometimes it cant BLMODsad

BLMODsad

maybe system charges change, or like cpu heat can force throttle

It’s a bit cringe to use kali as a daily driver imo :DWhy do you ask? (@ᴿᵉˣᵉᴺ)

Is there anyone whose operating system is Kali Linux?

ye kali cringe

22:42

use arch

22:42

or gentoo if u have time

use whatever you like

use windows

build an os that only runs discord and ddrace

build an os that only runs ddrace

22:58

and connects to other apps thru ddrace client

ddnet window manager when

Exported 110 message(s)