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 2024-11-18 00:00 and 2024-11-19 00:00

what're you trying to do

00:31

i assume you're trying to create some new textures.cpp file?

Voxel: you might opt to instead make a separate array of constant strings containing filenames at the indices of their enum values

00:44

so you can iterate instead on ranges within that enum and do m_aEmoticonTextures[i] = Graphics()->LoadTexture(m_aEmoticonFilePaths[i], IStorage::TYPE_ALL);

00:45

this could be a 2d char array or a std array of constexpr std string_views

00:46

that looks fine though, just lots of boilerplate

Voxel

class IGraphics *CComponent::Graphics() const { return m_pClient->Graphics(); }

Vocel: regarding your exception thrown, i believe i ran into this same thing when making the multithreaded assets thing, you need to wait to use client object request stuff like Graphics() and the like until after the module is initialized, not constructed. move it to onInit or whatever they call it, you should be able to find it in CComponent

00:48

i cant spell

00:48

oh sell

Good idea, I'll try that.

Maybe there's a way where I could efficiently do something like this. Where, I can make an enil containing all the segments, and then create other enums using the same sort of info?

01:15

Like, every enum except for the emoticons are in its own enum (for example: m_GuiTextures), which incidates IDs for every image that's in the array m_aGuiTextures

01:15

If I were to store in filepaths too, it'd just be sort of redundant, don't you think?

any more redundant than this?

01:30

you can do like this (not having tested it) // or extern constexpr const static std::array<std::string_view, EMOTICON_SIZE> s_aEmoticonTexturePaths = { // template params might be switched [EMOTICON_OOP] = “images/emoticons/oop.png”, … } (edited)

01:30

and then just iterate on it with the enum constants

01:31

this is no more redundant than what’s in your screenshot, because i believe those all have to be distinct function calls in binary now due to the way they’re defined sequentially, and all of the index information & known filenames are still stored in the same place (the constant “data” section of the executable”)

01:31

im typing a little sloppy on phone but hopefuly you get the gist of why this is better

I guess so.

ChillerDragon Thank fuck thanks for the demo pistol bullet fix

This goes inside the .cpp file or the .h file?

04:24

Ewan

04:24

(idk how to ping in irc)

alright thats it

im in the discord

04:35

and i am ONLY here for development purposes

04:35

@Ewan Something like this?

Image attachment

04:35

Currently set in the .cpp file

Voxel

Maybe there's a way where I could efficiently do something like this. Where, I can make an enil containing all the segments, and then create other enums using the same sort of info?

Recently found out that you can't get enum names in c++, so that's a bummer.

04:38

Really feeling hopeless about this project, but if I can somehow make it work, that'd be huge for adding and sharing around.

Swarfey

used to work for me when i did

cmake .. -DMYSQL=ON -DMYSQL_CONFIG:FILEPATH="C:\Program Files\MariaDB 11.3\bin\mysql_config.pl" -DMYSQL_INCLUDEDIR:PATH="C:\Program Files\MariaDB 11.3\include\mysql" -DMYSQL_LIBRARY:FILEPATH="C:\Program Files\MariaDB 11.3\lib\libmariadb.lib"

That's pretty much how I tried it.)

louis

what're you trying to do

I'm trying to remove texture atlases from DDNet. My goal is to make it easier to add in content without awkwardly having to shove new things into the texture atlases, and having to completely rework already existing atlases for the most updated version. The cons are that it'll be more expensive on the io side of things, and the code is going to be extremely messy because of its dependence on the sprites I removed. But I think the endgoals and potentials would be much greater, and it would give the ability for custom servers to send over custom weapons and particles easier, since there's no big texture atlas.

04:42

I know for a fact Jupstar already has a better way of doing this, but I think this would be fun, since it gives me hope on making things for this game again!

yeah, good feature to take on

04:43

i was moreso asking what you weretrying to do in that specific screenshot

Just trying to make it actually work, honestly.

i'd imagine if we want sending of textures over network, then how we store textures might need to be a lil more fleshed then how it is right now (i have no knowledge of the code though)

I guess right now I'm trying to rearrange the heirarchy of sprites and making it actually, render.

04:46

The next step is to write a program that takes what atlases we have now, and then spits out the respective folders, so that we don't have to have ppl complaining about their gameskins.

04:46

The step after that is figuring out if we need a ZIP file for easier sharing.

04:47

My take is that we do something like this:

default.weapons.twtp
default.particles.twtp
Voxels Fun Particles.particles.twtp

Where .twtp is just Teeworlds Texture Pack (edited)

04:48

and the middle is what type of pack it is

Voxel

@Ewan Something like this?

Yes but you probably shouldn't need s_aEmoticonTexturePaths before every single [], and you should be able to replace the string view ctor with the ""sv string literal

04:48

also double check ddnet naming conventions

04:48

s might not be right

Ewan

s might not be right

i just followed what you wrote down tbh

yea

04:48

it's from memory

justatest

Voxel

and the middle is what type of pack it is

Why not just have one generic type of pack that can override whatever specific parts it pleases? Or is this only for converting existing textures to the new format?

Voxel

My take is that we do something like this:

default.weapons.twtp
default.particles.twtp
Voxels Fun Particles.particles.twtp

Where .twtp is just Teeworlds Texture Pack (edited)

hm i dont think non-standard file extensions is the move (edited)

and yeah if it's a .zip just call it a .zip

idk i think just having it be .ZIP feels a bit off

i still dont know why it should really be a zip if we can use a tar instead

04:51

uncompressed zip is somewhat nonstandard

04:51

windows wont be able to make them

You could include an extra file in the zip for metadata, for stuff like info about the author and maybe info about textures for content that doesn't exist in the base game, if that ever becomes a thing

maybe

can't people just extract the zip and have it as a directory anyways

unless you discard all the software representations after loading into vram then you need to store the entire thing twice

zip files must be kinda clonky to handle in code

this whole thing is extremely hard, because it's basically rewriting how the entire game treats its graphics outside of actually rendering them

risu

You could include an extra file in the zip for metadata, for stuff like info about the author and maybe info about textures for content that doesn't exist in the base game, if that ever becomes a thing

if you have a single custom texture (lets say a laser) i feel like the client should get all of the information needed just thru the png

i know voxel i wrote a pr that overhauled all of the skins loading logic

04:52

it's delicate stuff

04:52

but still a question that needs to be asked

04:52

why zip

time to rewrite all of asset handling and implement full on resource packs that include textures, sounds, hell maybe UI layouts and animations too :D

04:54

surely not a huge pain in the ass :)

04:54

omw to rewrite half the game

Ewan

why zip

because its what everyone else keeps saying

04:54

#6753

https://github.com/ddnet/ddnet/issues/6753

Implement zip/directory asset handling · Issue #6753 · ddnet/ddnet

DDNet should switch from texture atlasses to individual textures for each sprite. This will allow us to fall back to default textures if a game asset doesn't contain a new sprite. old text #672...

probably because zips are easy for everyone to manage

i guess it did say to use .tar here but oh well

then again, everyone should be expected capable of installing and using an archive manager application no matter how non-technical they are

04:57

so no real significant downside to a tarball, and they'd be significantly more based :p

risu

then again, everyone should be expected capable of installing and using an archive manager application no matter how non-technical they are

eeeh

04:57

the average player shouldn't have to worry about much

04:57

skin pack creators, sure

these assets stand to gain very little from zip compression because png is already a very high compression ratio format

louis

skin pack creators, sure

yeah I was thinking about artists

they dont need compression and tar aligns more with the in memory representation aspect

okayyy this doesn't seem right

Image attachment

load the thing and copy out easily identifiable chunks

04:58

instead of deflating the entire thing and then copying

Voxel

okayyy this doesn't seem right

does it error?

04:58

that might be a fairly new feature

lets see what vs code has to say about this

Image attachment

do u not have LSP

idk what that is

05:00

language server protocol?

ye

05:00

are u on windows

windows 10 yea

open the vscode workspace for ddnet

05:00

and install the extensions

05:00

your life will be so easy

05:00

it's in the repo somewhere extra

wait why did i say vscode LOL

05:01

what i showed you was just vs

okay so what does intellisense say

05:01

you might have to include string view if you havent already

05:01

std array might be out of the picture which is perfectly fine u just cant iterate on it the same way

05:01

because it's not a primitive array initializer or something

Ewan

okay so what does intellisense say

Image attachment

did you incluide string view

05:04

ddnet is on c++17 right..?

05:04

or newer

lemme look

it may have to look more like

constexpr static const std::string_view s_aEmoticonTexturePaths[NUM_EMOTICONS] = {
  [EMOTICON_OOP] = "skjandasijkn"sv,
  ...
}

right lemme try that

voxuis

headbot

Voxel

right lemme try that

internet is down so i gotta type here using mobile

05:06

nvm LOL

Voxel

I'm trying to remove texture atlases from DDNet. My goal is to make it easier to add in content without awkwardly having to shove new things into the texture atlases, and having to completely rework already existing atlases for the most updated version. The cons are that it'll be more expensive on the io side of things, and the code is going to be extremely messy because of its dependence on the sprites I removed. But I think the endgoals and potentials would be much greater, and it would give the ability for custom servers to send over custom weapons and particles easier, since there's no big texture atlas.

as in remove them entirely, or construct them on init

05:09

because i dont think either is that good of an idea

Just splitting all the main gameskins into separate images

gameskins?

game.png particles.png extras.png pepeW

pepeW

atlases are used for performance reasons aswell as cuz less files to worry abt

05:12

you would (or should) stitch them back together on init

05:12

like mc does

05:12

which is why it takes so god damn long to load

05:12

90% my ass x-x

oh i didn't know that mc stitched them together

have u never seen those graphical bugs where you can see the atlas

in minecraft? no

Image attachment

LOL

its a very easy bug to make

05:13

you just have a logical error in your tex coords then boom

05:13

you can see the atlas

honestly i should let someone who actually knows what they're doing be behind the wheel like i think it'd be cool to actually be able to add more things easier, but at the same time it really does feel like there's 1000 better ways for what i'm trying to achieve

the problem with trying to make things more "expandable" is your kinda doing the thing you do at the end at the start

05:14

there are ideas and proposals for making the entities side of teeworlds (that is tile config and all that) more customizable

05:15

that is too allow 1. more tiles 2. easily added tiles 3. tuning on tiles (see the ice pr)

05:15

and i assume particles would be goign down the same route

05:16

as has been discussed at length, "modding support" that is refactoring ddnet to make it to add things is not a main priority

05:16

and issues / prs relating to modded support wont be accepted

05:16

which is a huge shame

Solly

and issues / prs relating to modded support wont be accepted

well then they're going to have to remove the big, fat "feature accepted" tag on the zip handling request

thats not modding support

then what is it what do you need zip files for in ddnet

modding support is to make adding things easier, that is tiles, custom behaviour, custom gamemodes, and anything else

05:18

thats just to make downloading maps faster

05:18

which is a good feature as ddnet map downloads are normally capped at 500kib/s

05:18

or less

that's not what the feature said

05:18

but oh well

gimme link

DDNet

https://github.com/ddnet/ddnet/issues/6753

this

Solly

and issues / prs relating to modded support wont be accepted

not true

05:20

ddnet maintainers seem to not support prs that support modding just for the sake of supporting modding, but a lot of prs that are related to modding support get merged

1

sorry im high on kaffine talking at me about their lost dreams x-x

Voxel

that's not what the feature said

i thought u meant zip map downloads

no i was talking about gameskins

05:21

like the feature said

05:21

it's been accepted for a while, its just nobody except juppey is doing jackshit about it

ah fair

its like all the good features ( coughcoughtunelockcough ) is just sitting out to dry

Solly

as has been discussed at length, "modding support" that is refactoring ddnet to make it to add things is not a main priority

i think that because of <msg im replying to> no one can be bothered since its not really needed for ddnet

tune lock is quite a heavy feature so i don't blame them too much. but it's true that maintenance is not active enough to keep up with all the PRs and issues

i wish u luck @Voxel :D

i dont think im going to continue this

05:25

again someone else with more knowledge should do it

05:25

like juppey

05:25

they already have a better heirarchy

05:26

i will stay in this development chat to hand out spitball ideas and whatnot, because i still think ddnet needs someone in the creativity department

lmao

салам

06:37

heartw

<ChillerDragon> @Pathos :) tbh you uploading videos with the bug is what motivated me to do it

https://github.com/ddnet/ddnet/issues/9229 anyone have an idea of what causes this?

чаго

seems to not work whether cl_predict 0 is on or not

тут есть русские

🍞fill🍞

тут есть русские

go to #off-topic to write in russian (edited)

<ChillerDragon> @louis I assume somewhere in the clients laser render code there is a check for the laser owner. Maybe some prediction stuff or ddnet extension. I did not look at the code but i assume something like if(!laser.owner) return;

ChillerDragon: i was taking a look and it seemed much deeper than that pepeW

pepeW

07:20

unless its just too late here and i missed smth obvious

ByFox

That's pretty much how I tried it.)

Have you tried using mariadb 11.3? Because your screenshots indicate you are using some different connector Otherwise Look at what exactly your building Arguments were. Your errors Show it is expecting a File path instead of the lib64 directory path you gave it (edited)

Voxel

Click to see attachment 🖼️

The components are not initialized yet in the constructor, you need to load textures in OnInit

Can I compile my ddnet client where on a local server I can reduce the server tick speed?

Evelyn

Can I compile my ddnet client where on a local server I can reduce the server tick speed?

You can, but you would need to distribute this client to all players & it will break some stuff

no, I'm talking about using it on a local server.

(╯°□°)╯︵ ┻━┻

08:55

don't worry I just read some comments (not here) (edited)

I dont get the question then, sorry

09:10

Maybe Chiller knows pepeH

pepeH

1

Ok

<ChillerDragon> @Evelyn yea probably idk im not into TAS ask teero and the boys who like to mess with that kind of stuff

louis

https://github.com/ddnet/ddnet/issues/9229 anyone have an idea of what causes this?

Team based rendering

louis

https://github.com/ddnet/ddnet/issues/9229 anyone have an idea of what causes this?

the moment you shoot the laser entitiy is spawned with the ownerID, which is your player ID
Every Snap the laser checks if the owner has a player character
if the owner has no player character, it's not rendered
after a teamchange there might be teammask shenanigans

(edited)

10:12

After a git blame the code for this is 14 years old with the description "I think i fixed the crash" lol

10:13

This is honestly spaghetti there

chillerdragon: did you make evelyn rq ddpp?

greenthing

1

What?

PSA: Do not share your private TAS clients. Do not use them on anything but local servers. They are bots and you will get banned for it

@Kero also lets move over here, as

12:35

*as #questions is for ddnet client stuff ^^

Patiga

@Kero also lets move over here, as

could be windows specific errors not sure. I didnt saved the master branch errors :/ Just wanted to try the preview map locally. But after step 5 not sure how to go forward. i have localhost:8000/?map=default but doesnt do anything for me. Console throws a error which I send in question channel. I assume i have wrong folder structure? In which directory should start the server ( does that even matter?) (edited)

ws-client

<ChillerDragon> @Pathos :) tbh you uploading videos with the bug is what motivated me to do it

:(

@Kero you need to start the server in the map-inspect-web directory all directories that you should create according to the readme should also be inside of the map-inspect-web directory

Assa

the moment you shoot the laser entitiy is spawned with the ownerID, which is your player ID
Every Snap the laser checks if the owner has a player character
if the owner has no player character, it's not rendered
after a teamchange there might be teammask shenanigans

(edited)

hm

15:11

so if the laser is not snapped on serverside it still exists?

15:11

does snap just prepare everything to be sent to client

Learath2

PSA: Do not share your private TAS clients. Do not use them on anything but local servers. They are bots and you will get banned for it

no fun

louis

so if the laser is not snapped on serverside it still exists?

yes, snap just sends updates over the network (for example sprite position), but tick() is doing the calculations

15:57

if you set sv_destroy_laser_on_death this bug won't happen, because then the tick() also resets the laser

ah okay makes sense

16:02

annoying bug

16:03

why does snap re check what's already in the ticks

snap and tick can be out of sync

on the server side?

A snapshot is quite literally a "snapshot" of the current gameworld to be sent to a client, something being snapped means it's in the current state being sent to the client

16:06

Snaps don't have to accurately represent the gameworld at all times, many mods (including ddnet) use this to do some trickery. Like the teams implementation, all tees exist in the same gameworld but for you only the ones within your team are snapped

16:08

Fokkonauts 128p mod snaps players as other entities if there are more than 64 in the view

i guess in this case it'd just be spaghetti code -- since it doesnt make sense to check if(!pOwnerChar) return; inside the claser::snap if it's already checked in tick

Learath2

Fokkonauts 128p mod snaps players as other entities if there are more than 64 in the view

and "hacks" different client IDs, but yes

louis

i guess in this case it'd just be spaghetti code -- since it doesnt make sense to check if(!pOwnerChar) return; inside the claser::snap if it's already checked in tick

https://github.com/ddnet/ddnet/commit/0a0e1df557da4f2b8d7a149dbbf16cfa2ff124df removing that would scare me a little on account of it being added by this commit

i think i fixed the crash · ddnet/ddnet@0a0e1df

14 years ago, the rest of the code grew there

16:12

also this should be the wrong place of the crash

Surely nothing will go wrong

don't worry, there is a character check right behind this xD

16:14

that commit looks so different to the actual code today

Actually that code does look absurd, how did these changes even make it past review and into the code

16:14

Do stuff with pOwnerChar discard it, do it again

From reading it, it does seem like dead code. There are lots of m_Owner >= 0 checks which should always be true because //m_Owner = -1; was commented out and the owner is never set to -1 when the CLaser object is constructed

Yeah, removing the 5 lines makes sense

i guess from a generic point of view, is there anything incorrect about just sending info from the previous tick thru the snap (along with removing any unnecessary information etc.)

16:16

how would a snap and tick desync

on client side? network issues

Someone knows how to compile their ddnet client, where the server tick speed will be changed on the LOCAL server

Assa

on client side? network issues

yes but doesn't the server still send the correct information from the corresponding tick

louis

i guess from a generic point of view, is there anything incorrect about just sending info from the previous tick thru the snap (along with removing any unnecessary information etc.)

I do not quite understand what you mean, you tick, then you snap the current tick. Why would you re-use the previous tick?

louis

how would a snap and tick desync

It's not unintentionally de-sync'ing. Calling it de-sync is a bit of a misnomer there. The snap is very intentionally not accurately representing the current state of the gameworld

i guess i'm asking: was that 'i think i fixed the crash' because of spaghetti code or because snaps inherently need be re-checked

Oh, this is from the olden days where we just guessed at stuff. That commit doesn't look like it fixes any crash

as in can something ever happen between the tick and the snap-from-tick that would make the server error if attributes aren't re-checked

16:20

okay

louis

as in can something ever happen between the tick and the snap-from-tick that would make the server error if attributes aren't re-checked

During OnSnap() (and therefore during the Snap()s called by that) the state of the entities has already settled, nothing will change until the snapping is done

16:23

So technically, if the laser did have an alive owner during the tick it'll still have an alive owner during snap. However, that's not information we store, thus the owners condition needs to be checked again

16:24

You'd need to add a new bool to laser, something like, m_OwnerAlive, but that helps nothing, it maybe avoids one pointer dereference on a condition that probably gets branch predicted insanely well since most owners of projectiles are usually alive

yeah

16:25

well if the laser still exists it must either have an alive owner or there must be sv_destroy_lasers_on_death 0 i think

Is that something worth bothering to risk missing an edge case though? It should result in no difference in performance, nor is the logic clearer

16:29

e.g. what happens on the exact tick that the owner died? I don't know for sure if we already looped through the entities and removed their projectiles yet

[ddnet/ddnet] Pull request opened: #9263 Clean up laser snap code

Closes #9229 I tested a little with sv_kill_delay 0 and tune laser_fire_delay 0 but would appreciate if someone more experienced can confirm it won't break anything

Checklist

[x] Tested the change ingame
[ ] 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 exist...

GitHub

Click to see attachment 🖼️

Please don't be broken

change sv_tick_rate_multiplier

17:43

its very buggy when u change it

17:44

very wip x-x

GitHub

Click to see attachment 🖼️

LGTM

17:54

albeit terrifying

(edited)

i was talking abt this earlier before i got banished for saying voldermort

17:57

but would tassing tools be accepted?

Solly

Original message was deleted or could not be loaded.

that is slowdown, frame step forwards and backwards (probably adding a command to set player state), and prediction showing

Evelyn

Solly

Evelyn (edited)

Original message was deleted or could not be loaded.

How to install this

you gotta compile it youreslf

Clone and then make?

yah

And it will appear on my online server?

git clone -b custom_tickrate --depth 1 https://github.com/sollybunny/ddnet
cd ddnet
mkdir -p build
cd build
cmake ..
make
./DDNet-Server

(edited)

18:00

the client has to support the lower tickrate or its really buggy

18:00

so should only be used on ur own server

local?

yeah

and if I run the server on my hosting, with this modification?

it will work fine, but prediction will make every non supporting client really buggy

18:01

the client also has to be around when the tick rate is changed otherwise it wont get it

Solly

but would tassing tools be accepted?

What do you mean "accepted"? They are not allowed on any official server. Releasing one whether source or binary will get you the bot author treatment

Learath2

What do you mean "accepted"? They are not allowed on any official server. Releasing one whether source or binary will get you the bot author treatment

tassing tools would have to be supported by the server, similar to how /tc doesnt work unless the server lets you

18:02

and could similarly be allowed only in /practice

Nothing stopping you from creating the tas on your local server then replaying it on an official server

Original message was deleted or could not be loaded.

And how can I compile it on windows in order to run it normally?

Learath2

Nothing stopping you from creating the tas on your local server then replaying it on an official server

o thats true

18:03

darned stupid people who use cool tools for malicious intent

Solly

i was talking abt this earlier before i got banished for saying voldermort

you were warned twice greenthing

greenthing

Evelyn

And how can I compile it on windows in order to run it normally?

im not sure how to compile for windows, read the README

Okay

meloƞ

you were warned twice greenthing

greenthing

once, first time was intentional when i said it, second time was out of shock (that the client had ads) (edited)

Solly

that is slowdown, frame step forwards and backwards (probably adding a command to set player state), and prediction showing

I mean isn't this a TAS tool you are just making public right here?

i mean its not that much of a tas tool

publicly available tas tools before gta 6 - damn it.

and can be done by ~~censored~~ (edited)

meloƞ

publicly available tas tools before gta 6 - damn it.

voldermort has proper tas tools

Every day I'm feeling more and more like heinrich was right with saying we should just ban all mention and discussions of TAS

yeah if people are this buggy with the mention of it im better of not

18:05

xd

Solly

and can be done by ~~censored~~ (edited)

please dont mention how to do this kind of stuff - in fact, any kind of help within obvious bot client development isn't really tolerated on this discord.

18:06

its not a rule stated anywhere, but just dont try to do it - you dont really want the i'm a person involved in bot client development stamp

censored by the oppressive ddrace network administration, damn

kek

1

ikr ):

meloƞ

its not a rule stated anywhere, but just dont try to do it - you dont really want the i'm a person involved in bot client development stamp

the stamp gets you the prestigious "none of us will ever answer your questions ever again reward" (edited)

justatest

1

damn but when you see obviously not real runs a human could never pull off

18:07

its so cool and i want to see more of them

youtube has a ton

yeah but i want more

Solly

its so cool and i want to see more of them

me too, that's why I told heinrich I didn't really want to ban the mention/videos of TAS runs

also replaying a demo on a sever wouldnt work?

18:07

cuz lag

Solly

also replaying a demo on a sever wouldnt work?

I'm not telling you how but you definitely can make it work

yeah i realised

18:08

but im not gonna

18:08

complete

18:08

;---;

18:08

it

18:08

x-x

As evidenced by the dozens of replaybotters we are dealing with right now

really?

18:08

damn

18:08

cmon people cant we just

18:08

not hack

people hacked pong man...

x-x

Also TAS doesn't immediately imply tick-perfect runs, the tools can be used to create just regular looking runs too with basically /r enabled

yeah

18:11

is there not some way to fug up replayed runs

So yeah, I'm in a weird position where I think TAS runs are very cool and fun to watch but I also see how the tools becoming public would just destroy the leaderboards beyond fixing

by having something on ddnet servers which is slightly different

18:11

which is not noticable to normal players

18:12

because a single wrong input = cascading effect

Learath2

So yeah, I'm in a weird position where I think TAS runs are very cool and fun to watch but I also see how the tools becoming public would just destroy the leaderboards beyond fixing

Excuse me, but when will the rules for communities be ready?

like randomly add some y velocity down every so often but only a smidge x-x

Evelyn

Excuse me, but when will the rules for communities be ready?

Any ruleset will probably include "Your community needs to exist for X days" where X is larger than 90, so probably not all that important when we put it out

Anyway, I'm curious, I want to see

Solly

like randomly add some y velocity down every so often but only a smidge x-x

The players are very very sensitive to this kind of stuff and for any tiny physics change you can imagine there is some weird map that uses it. So it's really hard to get something like this in

Learath2

The players are very very sensitive to this kind of stuff and for any tiny physics change you can imagine there is some weird map that uses it. So it's really hard to get something like this in

yeah ):

Evelyn

Anyway, I'm curious, I want to see

All communities must obey masterserver rules at all times
Communities must on average have X players in the last Y days
Communities must have existed for longer than Z days
Communities must be moderated These are some rules that I think we all agree on, there are a couple more and the masterserver rules are also being discussed right now, so it'll happen at the same time probably

(edited)

the master server has rules?

Solly

the master server has rules?

https://teeworlds.com/?page=docs&wiki=rules/server_rules this is the ruleset we inherit + we also remove servers from hosters that are abusive (ddosers, etc...) + we remove hosters hosting illegal/nsfw content

18:18

The latter 2 rules aren't really documented, we just applied common sense for our additions until now, but we want to codify it so morons like that one person advertising a bot client openly and expecting not to get banned don't show up and rules lawyer for hours (edited)

1

sigh

18:18

people

18:19

i was realy excited to make tassing for ddnet ):

@Learath2 What about the fact that cheat is allowed on block servers in TeeSports?

Solly

i was realy excited to make tassing for ddnet ):

You can make your own client and put your runs up on youtube, we don't mind that

the thought didnt even cross my mind to use it maliciously

Learath2

You can make your own client and put your runs up on youtube, we don't mind that

im not skilled enough to do tas runs

18:20

im sure uve seen the shenanagans tas's pull (edited)

Evelyn

@Learath2 What about the fact that cheat is allowed on block servers in TeeSports?

Not something we decided on yet

Learath2

So yeah, I'm in a weird position where I think TAS runs are very cool and fun to watch but I also see how the tools becoming public would just destroy the leaderboards beyond fixing

accounts^{tm}

oh yeah accounts would help so much

18:21

xd

it wouldn't help people cheating, but you could have some level of verification

louis

accounts^{tm}

Accounts would make it easier to clean up after the fact, but detecting such runs is the time consuming process

18:21

but yes, if random new account has top of the leaderboard it's immediately sus

do staff still manually have to search thru all the runs

and how the heck do u detect cheating in demos?

louis

do staff still manually have to search thru all the runs

currently, yes, we have to watch every sus run

Solly

and how the heck do u detect cheating in demos?

looking at it reaaaaaally carefully

we need the csgo overwatch system

but if someone is, as u said, splicing so they basically have /r, if they just make a client scroll back the demo recording perfectly theres no way its detectbale

18:23

because its not illegal to stop moving

No comment

new leaderboard rule: must always be moving or youre hacking

that wouldn't hlep

yeah but it would be afunny rule

18:24

YOU CAN HAMMER THRU 2 BLOCKS OF FREEZE??

18:24

wow these plp are insane (edited)

Learath2

The latter 2 rules aren't really documented, we just applied common sense for our additions until now, but we want to codify it so morons like that one person advertising a bot client openly and expecting not to get banned don't show up and rules lawyer for hours (edited)

like that one person advertising a bot client openly and expecting not to get banned don't show up and rules lawyer for hours

ah.. good times kekw

kekw

Thankfully the people that have been TASing so far have been mostly responsible with their tools

Solly

YOU CAN HAMMER THRU 2 BLOCKS OF FREEZE??

this has been done in real runs btw

meloƞ

this has been done in real runs btw

wow

meloƞ

like that one person advertising a bot client openly and expecting not to get banned don't show up and rules lawyer for hours

ah.. good times kekw

kekw

"but the rules don't explicitly say that you can't advertise a bot client openly, wtf"

Learath2

"but the rules don't explicitly say that you can't advertise a bot client openly, wtf"

kekw

funny af

Learath2

Thankfully the people that have been TASing so far have been mostly responsible with their tools

changes with me

Most of the TAS runs on officials servers are probably from paid cheats who are very aware that it's cheating

guh so many people making videos on the 40x speedup of linux, but its really just a memory optimization which has not that much of an effect on day to day

Learath2

The players are very very sensitive to this kind of stuff and for any tiny physics change you can imagine there is some weird map that uses it. So it's really hard to get something like this in

More importantly this causes prediction desyncs so even if you adjust them by 0.01 units, if they have high ping they might experience a correction of many tiles which makes it unplayable (edited)

18:37

I guess if we add some sort of "I'm going to adjust you by 0.01 units in 2 seconds" message to the client this strategy would work actually against replay clients

it's probably too sensitive for high-level players unfortunately

I think it's more about certain maps being too sensitive

hmm

It's impossible to know when you'll do something that makes the map break

yeah true

18:40

i'd also bet you could measure a difference with double nades / hard gores / edgehooks

18:40

like if the tee getting edge-hooked moves 0.01 units right before the hook comes out

You can be more subtle than adjusting position, a small difference in acceleration or velocity would be better (edited)

18:42

Those don't get quantized I think

yeah

Solly

YOU CAN HAMMER THRU 2 BLOCKS OF FREEZE??

You can hammer through 187 blocks at max speed xd

justatest

3

1

Teero

You can hammer through 187 blocks at max speed xd

how does one find this out? or did you just math

I don't think it's true?

Max vel is 6000

19:16

and you freeze 1 tick late

19:16

So you can hammer through 187 blocks of freeze (edited)

Robyt3

The components are not initialized yet in the constructor, you need to load textures in OnInit

yea i moved it there if i were to continue id have to figure out why everything is a white square

Did you confirm (e.g. with dbg_msg) that your OnInit function is getting called?

yea, i remember having a log saying the emotes were loaded in

19:20

hold on let me retry to build it and see how i can work from there

19:21

clearly the string array is causing troubles right now

Tater

I guess if we add some sort of "I'm going to adjust you by 0.01 units in 2 seconds" message to the client this strategy would work actually against replay clients

what if they had 2000 ping

19:30

????

2000 ping is the limit

19:30

but you could use 10 seconds

Robyt3

Did you confirm (e.g. with dbg_msg) that your OnInit function is getting called?

okay the oninit() isn't even running

Voxel

okay the oninit() isn't even running

You need to add your component to the list in gameclient.cpp: m_vpAll.insert(m_vpAll.end(), {&m_Skins, ... Probably as the first element so the texture are loaded last (components are initialized in reverse order).

i can't believe that was literally all i needed to do

19:41

justatest

19:43

i set the hook textures to something different for debugging purposes

19:44

the next step is to find out why the particles are pure white

i didnt like yellow so i repainted them

bro what kinda hook

Voxel

i set the hook textures to something different for debugging purposes

It looks great

Souly

bro what kinda hook

noby kind

Voxel

i set the hook textures to something different for debugging purposes

LGTM

Voxel

i set the hook textures to something different for debugging purposes

Push! Lgtm

yea yea hold on lemme fix a few other bugs

https://github.com/VoxelDoesCode/ddnet/tree/asset_rework

GitHub - VoxelDoesCode/ddnet at asset_rework

DDraceNetwork, a cooperative racing mod of Teeworlds, now with experimental features - GitHub - VoxelDoesCode/ddnet at asset_rework

21:12

I didn't completely remove content.py dependancy yet, but I think I made enough progress where I can send this

21:12

You guys can look over it before I move onto the next step

21:13

not sending this as a PR just yet because it feels too aimless to even consider to be a draft

owo

https://github.com/ddnet/ddnet/compare/master...VoxelDoesCode:ddnet:asset_rework#diff-1e7de1ae2d059d21e1dd75d5812d5a34b0222cef273b7c3a2af62eb747f9d20aL1441

Comparing ddnet:master...VoxelDoesCode:asset_rework · ddnet/ddnet

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

Pioooooo

owo

https://github.com/ddnet/ddnet/compare/master...VoxelDoesCode:ddnet:asset_rework#diff-1e7de1ae2d059d21e1dd75d5812d5a34b0222cef273b7c3a2af62eb747f9d20aL1441

whats this

Voxel

whats this

Looks like you moved a font entry in CMakeLists, the CI should also tell you that

ohhh

21:23

for some reason, sorting alphabettically in VSCode has different rules than CMake sorting

vscode clang extention also incorrectly formats stuff, its probably not reading the formatting thing, but the only difference between what i pressume to be default and ddnet is that default restricts char/lines to 80 (edited)

fixed!

Voxel

https://github.com/VoxelDoesCode/ddnet/tree/asset_rework

I assume the next step would to figure out how to make assets adjustable again? Some things I noticed from reading it:

Replace the enums in textures.h with enum class then replace all GameClient()->m_Textures.EParticleTextures::PARTICLE_AIR_JUMP etc. with CTextures::EParticle::AIR_JUMP. You shouldn't need the member variable to access the enum definition. That way you can also shorten the enum and enum item names so it's easier to read and write.
Eventually all the CTextures member variables should be private and only getters should be used to access them.
We don't use exceptions anywhere, so trying to catch any is pointless. If you want to check for errors, loop over all textures handles and check if any are invalid.
No need to check if texture handles are valid before calling UnloadTexture.

(edited)

1

i reset my config for testing

22:52

and

22:52

communities disaperaed?

22:53

if((g_Config.m_UiPage == PAGE_INTERNET || g_Config.m_UiPage == PAGE_FAVORITES) && !ServerBrowser()->Communities().empty())
    {
        CUIRect CommunityFilter;
        ToolBox.HSplitTop(19.0f + 4.0f * 17.0f + CScrollRegion::HEIGHT_MAGIC_FIX, &CommunityFilter, &ToolBox);
        ToolBox.HSplitTop(8.0f, nullptr, &ToolBox);
        RenderServerbrowserCommunitiesFilter(CommunityFilter);
    }

22:53

so why aint it rendering x-x

[ddnet/ddnet] Pull request opened: #9264 Fix out-of-bounds reads during dummy connecting

The dummy is considered connected by the engine client (IClient::DummyConnected function returns true) after receiving NETMSG_CON_READY, which also causes cl_dummy to be set to 1. However, the game client does not have the dummy's client ID until a snapshot is received, which means m_aLocalIds[1] (i.e. m_aLocalIds[g_Config.m_ClDummy]) is still set to -1 (or an old value from when the dummy was last connected on the server) for a moment when connecting the dummy. This was no...

Solly

communities disaperaed?

You are on the DDNet tab by default

oh

22:59

x-x

22:59

silly me

23:00

@Robyt3 about the collapsable serverinfo/communtiies and then my counter-pr of sticking communities at the top of filters because either solution is a solution to it taking up space, but no solution is not good obviously i think putting it in filters is better (since i made the counter pr) but which one is going to be ... merged? (It doesnt actually affect me ill change it in my fork) (edited)

Solly

@Robyt3 about the collapsable serverinfo/communtiies and then my counter-pr of sticking communities at the top of filters because either solution is a solution to it taking up space, but no solution is not good obviously i think putting it in filters is better (since i made the counter pr) but which one is going to be ... merged? (It doesnt actually affect me ill change it in my fork) (edited)

So far, neither of them seem like they would get merged, at least by me

do you not think that its in the way?

No, it's supposed to be highly visible

but it is a filter?

23:03

and not ... that importnat

23:03

ddnet will be at the top anyway just due to player count

23:03

and kog

And it's not visible on the individual community tabs so it doesn't take up space there

i dont use them i generally just filter

23:04

(but thats me)

23:04

if u switch back and forth there is layout shift (edited)

Solly

and not ... that importnat

It's up there instead of on the filter tab because it was previously not clear for players

there was previously a community filter?

23:05

i didnt notice evidently

23:05

i thought it was added at the same time communities got got their icons

It went through some iterations in nightly or maybe one release version, not sure

i kinda skipped from no communities / verifications to where it is now

23:06

i like the icons

the communities definitely feels out of place but i think a better solution involves refactoring the entire filter section (i've never used Server Address, and now we have two selections for Game Types, and i think a lot of these filter boxes can go into a collabsible 'advanced features' tab)

23:22

as of now the communities box feels a little small and hard to navigate

i like them all being in one place w/o dropdowns or extra menus i think the first type selection shgould be removed

Solly

i like them all being in one place w/o dropdowns or extra menus i think the first type selection shgould be removed

ye, but theres a tradeoff between the space for these filters and the space for the community list

as it is rn i dont think things need more space

as of now it only shows 4 communities for me without scrolling. I'd rather it show more communities than have some useless boxes I never check

do u use more than the top 4 communities?

23:24

idea

i use communities outside of the top 4 and there's no way to sort them

the type search box is to string search the type field right

23:25

but then the type list is a preset sort of thing (and also some special behaviour)

23:25

why not merge them

23:25

such that the type search box searches the types (in server list) and also the preset types (where if you click on one its updated in the search box)

23:25

same can be done with communities

23:26

oh no doesnt work u can select multiple types

can you type your whole message before you press enter pepeW

pepeW

no

23:28

because i dont think my whole sentance at once, it comes to me in parts

23:28

if i had done so the whole sentnace wouldve had to be different, since i realised my idea was a bit silly. but i still think that the whole combining string search and presets is a thing to consider

23:28

and thats too much effort

23:29

also some people like multiple short messages, so.. im not changing. you should just acept me 4 who i am

(edited)

23:29

(that concludes my ted talk)

23:29

#off-topic

23:30

(my fingers go faster than my mind woaaw)

when you type a bunch of short messages then go back on what you said it makes it harder to read and it means I'm less likely to understand the point you're trying to make. i probably type faster than you but i still type out my entire message before i hit enter justatest

justatest

louis

when you type a bunch of short messages then go back on what you said it makes it harder to read and it means I'm less likely to understand the point you're trying to make. i probably type faster than you but i still type out my entire message before i hit enter justatest

justatest

holdon

the game types search box can probably be moved inside the types box at the top anyways (and be scrollable along with the type box)

i have an image of me doing 150wpm which im really proud of (i cant reproduce it) (edited)

23:32

and i did it on a school keyboard

louis

the game types search box can probably be moved inside the types box at the top anyways (and be scrollable along with the type box)

"oh no doesnt work u can select multiple types" - me (edited)

ye, it'd need a button to toggle

23:33

might be ugly

any1 know how to fix cmake errors

23:38

23:39

or warnings which make tests fail

23:44

i figured it out

23:44

i justhad to read a bit harder

23:50

[  5%] Building CXX object CMakeFiles/rust-bridge-shared.dir/src/rust-bridge/engine/shared/rust_version.cpp.o
<command-line>: error: macro names must be identifiers
<command-line>: error: macro names must be identifiers
<command-line>: error: macro names must be identifiers
<command-line>: error: macro names must be identifiers
<command-line>: error: macro names must be identifiers
<command-line>: error: macro names must be identifiers

23:50

uhhh

google broke firefox everyone responsible deserves capital punishment

Learath2

google broke firefox everyone responsible deserves capital punishment

google helped firefox stay alive firefox was mismanaged was my understanding

or atleast permajail

23:55

google services keep breaking on firefox (obviously to push people to chromelikes)

ah yes

23:55

they do that

23:57

the cmakefile is so poorly formatted jesus

23:57

4 space indent, 2 space indent 1 space indent 8 space indent?

23:57

space imbetween Not and If? idk man

23:58

turns out my cmake formatter also sucks

23:58

and unindenting things breaks stuff

23:58

dman i didnt know cmakefiles were indent sensative x-x

Exported 505 message(s)

Timezone: UTC+0