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 2020-10-08 00:00:00Z and 2020-10-09 00:00:00Z

[ddnet-maps:master] 1 new commit

7073b2a A nugtohS - ddnet-maps

[ddnet:master] 2 new commits

51b9670 Fix button rendering in "Render demo" - def- 6376890 Merge #3035 - bors[bot]

05:12

[ddnet/ddnet] Pull request closed: #3035 Fix button rendering in "Render demo"

[ddnet/ddnet] Pull request opened: #3037 Init date for demo dirs

Just to not rely on uninitialized values

[ddnet/ddnet] Pull request opened: #3038 Use std::sort instead of bubble_sort

Noticed, that its really slow, when you have alot of auto demos

[ddnet/ddnet] Pull request closed: #3037 Init date for demo dirs

08:36

[ddnet:master] 2 new commits

a4b32d1 Init date for demo dirs - Jupeyy cbec1a5 Merge #3037 - bors[bot]

[ddnet:master] 2 new commits

52911e4 Use std::sort instead of bubble_sort - Jupeyy 2d0c65b Merge #3038 - bors[bot]

08:44

[ddnet/ddnet] Pull request closed: #3038 Use std::sort instead of bubble_sort

[ddnet/ddnet] Pull request opened: #3039 Show friends on spectate tab

!2020-10-08-13:22:48-screenshot !2020-10-08-13:23:14-screenshot

[ddnet/ddnet] Pull request opened: #3040 Revert "Reset m_ServerCapabilities on Disconnect"

This reverts commit a6d2a35e11d6fa6e222b9d57ce2841e43066a766. It's unclear how this should have fixed the mentioned bug.

[ddnet/ddnet] Pull request closed: #3040 Revert "Reset m_ServerCapabilities on Disconnect"

11:36

[ddnet:master] 2 new commits

eb4699d Revert "Reset m_ServerCapabilities on Disconnect" - heinrich5991 b2b3b86 Merge #3040 - bors[bot]

I want to write redirects from the maps filenames with CRC to the maps with SHA256

14:17

nginx's built in matching engine only supports case-insensitive matching or (probably costly) regular expressions

14:17

so I guess I'll have to write a service that nginx will then connect to

14:18

how should I implement this, so that this service can run 10 years without me touching it?

Are you sure regex is too expensive?

no, I'm not

I'd do that first. my assumption is the regex can be compiled once and should perform well enough for us for 10 years

14:20

if we see nginx cpu performance problems, we can investigate

I was thinking about my map collection, but can investigate for ddnet maps too

14:20

the documentation kinda implied that it matches the regexes linearly

@heinrich5991 that's so we can offer downloads for KoG?

good thinking

ah

I'll implement it for ddnet's maps too

14:21

the documentation kinda implied that it matches the regexes linearly

14:21

and at 50k regexes, that sounds too expensive

14:21

but I didn't try it. do you think I should give it a try? @deen

ah, you need one regex per map?

14:22

I thought just one regex for all, too bad

14:22

just make symlinks on filesystem?

I guess that could work, but that doesn't give me the nginx redirect I was looking for

14:23

(need to be more precise with my requirements ^^)

why a redirect specifically?

hm. good question

14:24

I didn't want to list them in the file listing anymore, but I guess that can also be achieved by rendering the file list myself

the case insensitivity is annoying with Windows too btw

14:28

uwsgi + python is what we generally use for stuff like that in ddnet, but I don't know how stable it is in the long run

nginx made a thing called nginx unit

14:38

https://unit.nginx.org/

NGINX Unit: About

A versatile open-source server for a diverse Web.

14:39

https://unit.nginx.org/configuration/#configuration-python monkalaugh

monkalaugh

NGINX Unit: Configuration

Create and maintain a working configuration using certificates, listeners, routes, apps, and upstreams.

14:40

justatest

Image attachment

14:40

if you use go u get even better perfomance

Image attachment

treads?

https://itnext.io/performance-comparison-between-nginx-unit-and-uwsgi-python3-4511fc172a4c

Concurrency comparison between NGINX-unit and uWSGI (python3)

This article will show performance of two web application servers nginx-unit (a new modern application web server) and uWSGI (the best one…

14:41

it probs is a typo

14:41

xd

Heh nginx unit supports assembly, wanna re do the ddnet site in x86_64 assembly?

bluekitty

would still like to know what caused these bugs

Wait that also includes writing it in c++ and disasemblying it right?

map download bug and the other one

yo @Soreu I replaced the grass_main tile bitmaps with svgs on https://zillywoods.github.io/ZillyWoods-web/ any comments/suggestions?

ZillyWoods Client

Custom client for Teeworlds version 0.7+ coded by ChillerDragon

ChillerDragon: You probably don't need to disassemble it, I'm sure they just run any binary you want

lol ok true

14:46

then lets write it in shell and compile to binary using the shellcompiler

Oh, they actually mean assembly lol

chiller and his shell addiction

ChillerDragon my only comment for now if that I seriously fcked something up xd

14:46

Image attachment

yikes

And no, it's not your fault

how?

14:47

is svg new for browsers? lmao

missing proper style for edge-cases

14:47

nah, it's not related to change to svg

@ChillerDragon where are the svg

14:47

i want them

https://github.com/TeeworldsDB/images/tree/master/realistic/transparent

TeeworldsDB/images

A collection of free to use and edit images for other art work. No actual game skins. - TeeworldsDB/images

xd also a full mapres?

i was thinking about it

14:48

I almost finished all parts of grass_doodads so lets see

14:48

4k remastered :D

i'd love to have a 4k feature for ddnet

so do i

bcs on 2k screens tw already looks slighty blurry

playing tw in 4k since years

14:49

well ddnet comapred to vanilla looks even more blurry :p

how so

14:49

we updated all mapres

skins and hud for example

14:49

probably the way its rendered

skins yeah, and hud ur pr will fix

no

14:50

wait what hud pr?

ur game.png

nah

14:50

it still looks worse

why not

idk

14:50

ur the gfx guy

bcs its wasnt dilated

tell me

14:50

oh maybe

it should look the same

i didnt check after dilate

14:51

so im kinda activley working on the svg remakes and will let you know if i finished a full tileset

yeah nice

so @Soreu what is going on? Even works on my safari on iOS

sad that tw is so closed source about that mapres and stuff

14:51

bit annoying

ye

14:51

super sad

14:52

even infecteus license

I'd be happy to see optional higher quality graphics too!

what res do u play deen?

i don't play

lol

he plays 5:4

It will work everywhere, mobile Chrome messes the website look after browser header gets hidden on scroll

I was once considering adding an actual vector based renderer in there

i don't play

14:53

justatest

vector render oof

librsvg and render graphics to a few sizes at start

slug is such a great idea, vector graphics in the GPU

but I watch in 1920x1080 usually

would that increase performance?

14:53

probably huh?

14:53

and likley nice for zoozing

probably not

@deen we could just render to the selected resolution

the opposite

Wait, you mean replacing all graphics with SVGs?

no game does that

14:54

for a reason i guess

browser gamesß

14:54

?

no they dont

Reason is probably that they don't want to give out their assets in infinite quality

rasterization is too expensive for soo many vertices

don't underestimate the greed of large corporations

@Learath2 I'm quite sure you're right

rendering to a texture at start is useless then too, could just pre do it

im not sure

fuck the large corps that made bomberman and all the other flash games i play in my browser

@Deleted User well you can render it at any resolution needed, instead of shipping MASSIVE texture files

14:55

The new call of duty no longer fits on a 250g ssd

yea, but space is too cheap

14:55

CPU and ram isnt

well you can ship svg and let the client compile it to bitmap on client start or res change lol

I have RDR2 on ps4 still undownloaded because it would take me like a week at my third world speeds

newgrounds good old times

14:55

i dont play flash games anymore

but u still have flash installed right?

14:56

i keep getting download popups for it

14:56

downloaded it twice today

The way modern programmers treat memory and CPU would suggest they are very close to free

yeah but not in games (edited)

bluekitty

Image attachment

not for tripple A titles

flight sym

yeah thats a good example

xd

14:57

its a meme tho

why u should still write good code

its like crysis 2

At startup they are quite cheap, it already takes a shit ton of time to load, no one will notice a couple extra seconds needed to render the svgs

@Learath2 i get ur point

eh factorio optimized their game quite a lot

14:57

factorio best

but we dont have todo the same mistakes xd

after ddnet

after teeworlds

nah

14:57

teeworlds at the bottom

I love the idea behind the new flight sim, it takes the old idea of streaming textures and does it so well

14:58

Didn't think you could stream geometry that well over a reasonable connection

geometry?

14:58

isnt the map at the client already

14:58

or am i missing smth

it loads stuff like google earth i guess

14:59

just higher res

The base of the map is there, they stream features for the map though

but why wouldnt that be on the client already

14:59

doesnt the game take 100gb

14:59

xd

bcs the world

14:59

is bigger than 100gb

14:59

;D

eh

15:00

we got games that are 250gb rn

15:00

like the modern warfare

15:00

https://www.gamespot.com/articles/cod-modern-warfare-doesnt-fit-on-a-single-250gb-ssd-anymore/1100-6482933/

CoD: Modern Warfare Doesn't Fit On A Single 250GB SSD Anymore

As the game continues to grow, Modern Warfare in its entirety won't fit on the drive size you may use for your entire operating system.

15:00

haha

15:00

hehe

15:00

uh

15:00

monkalaugh

well probs microsoft also wants dependency of bing

15:00

so ppl are aware of it

15:00

i mean microsoft gifts u media markt gift cards for using bing xD (edited)

FS2020 has most of the world in the 100g, they just stream extra geometry and extra textures

15:02

Anyway, I love to see an old idea reused like this

https://youtu.be/hEmLMpLAnM4?t=35 poggers

poggers

Microsoft Flight Simulator 2020 | Airbus A320 | Full Flight

absoluty no flight fan, but deffs has its charm

i think this guy is a real pilot

The handling is sadly a bit arcade, I guess to pump the sales up

15:05

It's very niche really. I flew on FSX a bit back in highschool

15:06

It might be much more fun on fs2020 with the world actually looking decent. On FSX there was no point to exploring, so you just did standard commercial routes and those are quiiiite boring

[ddnet/ddnet] Issue opened: #3041 Spawning is still shuffled when in a solo team

Bug scenario: | Sequence of events | Result | |------------------------------------------------------|--------------------------------------------| | Player A is alone in a server | Their spawn point remains fixed. | | Player B joins the server | Both players' spawns are now shuffled to prevent collisions. | | Player A joins a team alone and locks it | Both of their spawn p...

[ddnet-web:master] 1 new commit

43a46fe Quick Tournament #52 - def-

feelsamazingman

5

normal particles

Image attachment

17:49

weird color particles for this guy only

Image attachment

17:49

how is this possible in team mode? @Deleted User @Learath2 o-o

on newest client?

15005

17:52

two ppl reported it, other has recent version too

rainbow particles when?

dunno if i changed something in particles in the last 2 years

is that possibly a server side bug

no

17:56

team colors are calced in the client

ye right

let me see if skin changes can affect blood color

18:02

do they have the bug always

18:02

or just sometimes

@Deleted User it was always just for 1 tee

18:04

and i think the tee left already

mhh, maybe blacktee or smth?

18:05

or a whitetee

n

18:05

sec ill ask for demo

maybe a skin prefix thing xddd

18:06

would be funny

18:08

is it possible to get shot in the tick u are unfreezed @noby ?

18:08

maybe its the ninja skin

possible but i think its very unlikely to happen on the same tee more than once per round

oh yeah

18:09

but u dunno the tee name?

sec

18:10

sent demo

18:10

tee name Ӽӥϣңӥⱪ

18:11

look around 1:00 and idk his skin name

where is the demo xd

i sent u

i didnt get it

18:11

or its still uploading xd

18:13

defs interesting

18:13

its just that skin

also it happens when they get shot and when they get spiked

18:13

what skin is it?

yeah its the blood color

18:13

wait

18:15

saddo i guess

18:15

vanilla skin

i have a stupid suggestion, but i should to announce it maybe add tips below loading box?

funny not reproducable only in demo

18:17

but thats enough too see what the client does

im trying to reproduce it with that skin a

18:18

and cant (?)

18:18

the guy was using 0. 7 client if it matters

yeah, but in demo its always

18:19

yeah i was guessing that

18:19

i'll see, maybe its indeed server side, even if i dunno how

are the particles rendered with transparency

yes

maybe there was a dupe render that rendered over it

18:19

making it more red

everything is rendered with trans

oh i meant like are the particles not opaque

they saw it ingame

well look into the image

18:20

but i guess they are bit transparent in the borders

we cant reproduce it with that skin even

not even with 0.7 client?

18:22

well anyway i'll debug the demo soon

no

by biggest troll

3

@noby well seems like it uses the color that the server sends for the particles, but gonna test it now, just to be sure

19:07

it only overrides the skin color itself

oh weird

19:09

so its not a new bug?

i'll test it now, i'll say more about it then

19:13

ok

19:13

now im sure

19:13

@noby its server side

19:13

CPlayer::Snap(

ty

pClientInfo->m_ColorBody = 65280

19:13

will reproduce

is it new

i guess ddnet could just override it, bcs makes more sense actually

time to make rainbow effect

xDD

@noby yes pls

skin color is serverside, but if GAMEFLAG_TEAMS is activated I think the client has its own team values

yes but not for particles

19:28

it doesnt override them

https://github.com/ddnet/ddnet/blob/b2b3b866cdac343bef8c39ea83d3269980ebc125/src/game/client/components/effects.cpp#L169

DDraceNetwork, a cooperative racing mod of Teeworlds - ddnet/ddnet

19:31

yes it does

19:31

m_ColorBody gets set

yeah it clearly uses colorbody

to the team color

not renderinfo.colorbody

yes

dude i just tested it xd

19:32

it uses the info from the snapshot

19:32

    m_RenderInfo.m_ColorBody = color_cast<ColorRGBA>(ColorHSLA(TeamColors[m_Team]));

19:32

but effect does not use renderinfo

19:33

it uses a different var

true, then thats the issue xd

19:33

effects have to use renderinfo ofc

yes

[ddnet/ddnet] Pull request opened: #3042 Use renderinfo for player death effect

the render info color is created the same way, just that it is allowed to be overritten for team colors see OnNewSnapshot and UpdateRenderInfo

also here PSA: maps now use sha256 https://heinrich5991.de/teeworlds/maps/maps/

[ddnet:master] 2 new commits

94a20d1 Use renderinfo for player death effect - Jupeyy 0d581c4 Merge #3042 - bors[bot]

20:55

[ddnet/ddnet] Pull request closed: #3042 Use renderinfo for player death effect

it works lol

not anymore with new update kek

kek

(

)))

https://www.amd.com/en/products/cpu/amd-ryzen-9-5900x

AMD Ryzen™ 9 5900X

21:17

new cpu

21:18

better than i9 10900k

21:18

monkalaugh

[ddnet-web:master] 1 new commit

33c37ed Frontpage video: Pros Play TW 107 - def-

12 cores 24 threads with 3.7GHz Up to 4.8GHz

21:18

crazy

Exported 321 message(s)