Teeworlds - general

Discord is being very intermittent

00:00

broken as hell.

00:01

they can't deal with the traffic

00:01

funny, better send screenshots of messages than actual messages

00:01

broken as hell.

high quality internets

00:02

00:03

also likely because

nicem broken as hell was sent twice

00:04

nice broken as hell was sent twice (edited)

https://status.discordapp.com/

Discord Status

Welcome to Discord's home for real-time and historical data on system performance.

00:04

where is irc2.0 with markdown support and code highlighting.

00:05

nicem broken as hell was sent twice

00:05

interesting

00:07

where is irc2.0 with markdown support and code highlighting.

00:07

inpphhheption

00:07

inpphhheption

00:07

inpphhheption

00:07

Discord is being very intermittent

xD rip.

00:07

inpphhheption

also likely because

finns brought us IRC swedes brought us teeworlds what did the norwegians do, @MertenNor

nicem broken as hell was sent twice

00:08

discord is completely dying down currently

00:08

holy shit, I get all the messages like 2 to 5 times

00:08

discord is completely dying down currently

00:10

norwegians should bring us IRC2.0

00:10

Well...

00:13

Idk

bawnkrz joined the server. 2020-02-19 00:45:53Z

Piotr joined the server. 2020-02-19 00:47:01Z

cheese joined the server. 2020-02-19 10:26:00Z

@heinrich5991 the banlist didn't take effect, 24h later. you have an idea why?

13:08

guess some masterservers are a bit slow reloading the cfg?

it's updated

13:29

https://status.tw/master/bans/

Masterserver bans - status.tw

A service for displaying and analyzing the status and statistics of Teeworlds servers and players.

13:29

here at least (edited)

13:30

but still online, tho

>Dune: modified physics on CTF eh, it's ban worthy?

yes

@Deleted User https://www.teeworlds.com/?page=docs&wiki=rules/server_rules

14:44

server admins can reach us on IRC or here to get unbanned

14:46

(you can't modify physics on CTF without removing the limit in the source code)

@Dune Do you know the tickrate of TW servers? Does sv_highbandwidth enhance tickrate? By how much?

src/engine/shared/protocol.h:81: SERVER_TICK_SPEED=50,

16:45

if(g_Config.m_SvHighBandwidth || (m_CurrentGameTick%2) == 0)
                    DoSnapshot();

@Captain_AMN if sv_high_bandwidth is enabled, the server sends snapshots all the time, all it can

thx

16:46

all it can... how much roughly?

depends on your specs

it's server limited?

the description says you're not supposed to run sv_high_bandwidth other than in LAN

yep ive seen that

16:47

on tee other hand, i think 50 tickrate is wayyy too low for good players

I'm not sure about ticks, I don't think it changes the tick rate though?

16:47

should still be 50Hz, could be wrong

what's the point in having a tickrate lower than a low-tier monitor's refresh rate

16:48

in a fast game like teeworlds, that's a lot of inaccuracy

what's your ping

16:49

if it's 20ms, that's 1/50s

Tickrate 50, snaprate by default 25, with high bandwidth 50

i tested sv_highbandwidth on a server. you can see the difference when you disable prediction on your client

twice the snap frequency

Increasing the tickrate would break most physics code

I mean, it shouldn't be that big when you compare it to the latency you have anyway which is higher than 1/50s (edited)

well every layer of delay adds to inaccuracy

welcome to online games

you can't lower ping, but you can improve game tickrate

if you can't lower ping, I think going under 1/50s is not going to make much of a difference while making it very much worse for players with a slightly bad connection

@redix why break physics?

I already can't play properly on any of @ChillerDragon 's french servers, because of packet losses

it's not meant for the average joe

16:53

just an option for tournaments

16:53

and people who can notice such things

The physics code just wasn't designed to have a variable tickrate =\

standard snap rate is every 40ms (edited)

But enabling high bandwidth should be okay for most connections

i dont even know what snaprate is btw

By default the server does not send all snapshots

16:56

Just every second snap is sent to the clients (to save bandwidth)

what does high bandwidth mode do?

It sends all snaps to the client -> 50 per second

16:58

By default it's 25

and doesnt change tickrate...

16:58

ok

Yep. But might lower the delay a bit since your information about the game is more up to date

there should be an experimental mode to try higher tickrate

17:01

maybe it's possible to adjust the physics so they stay unchanged

17:02

TW could then be a proper e-sports game

17:03

it's 2020 and there are 360hz monitors and fiber connections, playing with 50hz tickrate is lol

17:04

even on 5G people will have lower pings anyways

17:04

for an extremely fast game like teeworlds, it should really be considered

Right now when you just double the tickrate you cannot even move as far as I remember

17:06

And this would definitely need a compatibility break (version 0.8)

what do you get from a higher tickrate?

17:06

360hz screens have nothing to do with that, display is as fast as you want

you get a better and more precise gaming experience

please be precise

17:07

I don't understand what you mean with "better and more precise gaming experience"

17:07

that's just buzzwords to me

you can input more information to the game than it currently allows

17:07

and faster

17:08

also the tees positions are displayed better

17:08

right now it's pretty inaccurate

the tees positions are displayed accurately

17:08

tee's*

not when they are moving fast

what do you mean?

if you look at a movie at 5 FPS or a movie with 60 FPS, do you notice a difference?

yes. luckily teeworlds can render to 1000fps if you want

yes, but it's useless unless the server doesnt have 1000 tickrate, and your monitor doesnt have 1000 hz

why is it useless if the server doesn't have a tickrate of 1000? I don't think that follows

17:10

[6:07 PM] Captain_AMN: you can input more information to the game than it currently allows

17:10

that point sounds valid though

17:10

currently, the average input delay is 10ms

17:11

which is a lot compared to ping, I guess

your graphic card can only render the information it gets, so if that information isn't updated adequately, you cant take advantage of it

teeworlds can render more than 50fps

17:11

it interpolates perfectly well

17:11

as said, you can have your 1000 fps

look

17:12

those 1000 fps, 940 of them, you wont even see on your monitor man

yes

17:12

but you complain about not seeing enough updates

Nah... For other players it just does linear interpolation between the last two ticks

yes

17:12

but you can render as often as you want, and get a new view every time you render

17:12

also on your fancy 360hz screen

True

if you match the server tickrate, game FPS, monitor hz close enough together, you have better visual and input accuracy

can you support that hypothesis?

17:13

I see that a higher tickrate can improve input latency (as said above)

can you support that hypothesis?

@heinrich5991 it's pretty straight forward, if you eliminate bottlenecks, you get benefits. guess we have to try it out to see if good players can notice a difference

so far your argument has mostly been "buzzword buzzword". I don't think it's that straightforward

17:17

we could try it out though

17:17

I don't have the feeling that the discussion will be much productive after this

it's normal that these are just buzzwords to you, only pros (tm) can understand

17:19

if you can't notice a difference between a 240 fps monitor and playing 30FPS playstation, it sounds like rocket science to someone

Btw... Ticks are not always executed after 20 ms :O in average yes, but single ticks are somewhere between 15 and 25 ms. The idling code from ddnet fixes this. The gametime graph is smoother then and prediction is probably slightly lower

ddnet shits all over TW

the problem is more the prediction than actually the interpolation of old data.

17:24

more data might improve the prediction?

17:24

but clogg people's internet connection with bad connectivity?

17:25

when getting too much data?

Todays internet connections can probably handle the amount of data (at least snaprate of 50) but people with a high ping will hardly benefit from it

more packets arrive?

Prediction for the local tee is just fine. Prediction for everything else is missing

17:27

Yeah. More snap packets

17:28

I think we tested high bandwidth in a few games on aeon server. No issues :D

if you disable prediction, it's still choppy

17:29

but on other servers, it might not be

might test it on zCatch and ask people if they got a better experience?

come find me ingame now

17:30

test it on my server

Prediction is there for data that's about to arrive, disabling prediction will obviously make the experience worse.

17:31

I'm not at a pc.

it feels pretty inaccurate to me

17:32

disabling it you get better aim

17:32

crisper rocketjumps

For me tw always felt kinda responsive compared to everything else

1

No one asked you Chilli!!!

better prediction would help more

18:38

in one case you are right, teeworlds (client) is currently running on 50 updates per second

18:38

and I would be happy to get it up to 144 (for my monitor)

18:39

well, you could work arount that with a better client prediction <.<

i did some testing with amn his problem with prediction is that your tee is not really in sync with the nades so it can be hard to avoid nades when you activate high bandwidth (and have a low ping) you can even play without prediction, so your tee is actually in sync with the nades

18:46

without high bandwidth you definitely need prediction, otherwise the game feels choppy

is there some documentation for high bandwith?

not really. it just sends 50 instead of 25 snaps per second to the client

18:53

basically doubles bandwidth

i would really like it on client base

client base?

side

19:15

most likely requesting higher bandwidth from the server.

maybe dynamically

high bandwidth 0 + prediction off = game is choppy high bandwidth 1 + prediction off = game is great, with some minor visual bugs

19:41

i would really like to experience how higher tickrate would feel like, but redix says you pretty much can not alter the tickrate

visual bugs: hook still hits the predicted tee, airjump effect is also at the predicted position

19:41

i guess nobody ever tested this code

in the early days of programming things worked without tests, because people knew what they were doing.

with increasing complexity you have an increasing need for tests

21:30

the guy who implements unit tests for teeworlds would be my personal Hero

21:31

but it's unzumutbar

@Assa

21:31

I wasn't able to find a coorect english translation for this word

21:31

germans have words for anything

21:32

maybe unbearable

21:34

basically a job where you don't want anyone to do it because it sucks (edited)

there are unit tests in teeworlds

21:46

@heinrich5991 made some

21:46

that comes with downsides though, it makes things less flexible imo (edited)

only legends remember

Xd

22:02

Emote 16

22:08

Snitches ditched it

wasn't that fixed?

@Dune which tests in particular make things less flexible?

I meant in general, I haven't had any issues with teeworlds tests

23:03

just I'm not a big fan of test-driven development

23:04

some tests are good imo

having tests and doing test-driven development is not the same

23:04

I wouldn't develop a huge piece of software without tests

23:04

you're just scared to touch any line, because it might break something and you might not notice

23:04

tests help a bit with that

especially when everything can touch at everything, yeah

23:06

reading the messages again, it really sounds like a pitchfork against your tests in particular, which I have no problem with. sorry!

sorry for the fuss

you're just scared to touch any line, because it might break something and you might not notice

by the way, when in this situation, tests only make me marginally more confident

23:12

wasn't the promise of OOP to shield against these side effects?

23:13

clearly Teeworlds isn't really thought with encapsulation in mind, you can just access Client() in half of the code etc.

current programming: let's change that, side effects do not exist :D

23:16

tests: let's change that, tests will tell us what side effects occurred.

nah

23:17

e.g. tests let me refactor the CJsonWriter

>:o

that didn't have anything to do with side effects

integration tests : don't exist

not in teeworlds

:'(

considering tests... the json writer tests fail on windows due to line endings. gh actions just does not recognize =\

it doesn't fail on gh actions or the tests on gh action but it doesn't consider the run failed?

23:38

*the tests on gh actions fail

it fails but gh actions does not consider it. return code seems to be okay, even though the tests failed

@jxsl13 emote 16 was fixed, if i remember correctly both server and client had to be 0.7.2 or 0.7.1 for this to work

23:59

~~we need emote 17 now~~