DDraceNetwork - developer

<ChillerDragon> https://zillyhuhn.com/cs/.1712709169.png average magnus code snippet

00:36

<ChillerDragon> The size of an item is obviously located at the index of some offset .. duh

00:36

<ChillerDragon> if that wasnt clear enough here the implementation of Offsets()

00:36

<ChillerDragon> https://zillyhuhn.com/cs/.1712709308.png

awesome

ws-client

<ChillerDragon> @Mr. Teetorial what are you building?

I’m trying to make dummy copy and dummy hammer work simultaneously.

dummy has nothing to do with input handling honestly.

TsFreddie

dummy has nothing to do with input handling honestly.

Yes but it would maybe make an easy workaround possible. If the left mouse button inputs would not be like 100ms long but instead 1 ms it barely stops dummy copy. Obviously it would be much better if you could really use copy and hammer at the same time but I think that solution would be far more complicated that’s why I’m trying to make smth easier work first. At the moment I’m trying to work with some of chillers tutorials on YouTube. Awesome videos man.

also I believe some are opposing the idea of expanding dummy ability further due to various reasons. but if you really want to or just curious. dummy inputs are handled directly in the network layer

05:47

https://github.com/ddnet/ddnet/blob/39dd2a552d814e45e92e71cf762437fe52353b9a/src/game/client/gameclient.cpp#L445

But I having struggles because it’s my first time in c++ and originally only coded in Java when I was younger

dummy is not a game design where it was just a entity that interact with a player. it is literally a spoofed client connection, so a dummy will send their input independently to the non-dummy player. any desync and delay is likely unsolvable unless you remade the network protocol to pack two players input as one single piece of information.

05:51

it is very likely that the dummy system is more rudimentary and unreliable than you thought

or you can make the server aware that the character belongs to your player as well and mirror the actions on its side

or that

05:52

server-side dummy is actually preferred by deen at some point (edited)

05:52

iirc

I get that dummy movement in theory is independent but as soon as dummy copy is on the main tee movement inputs are copied each frame into the dummy movement inputs. If the main tee +fire input is now shortened to 1ms the dummy’s input will be 1 ms long aswell. This would cause dummy copy to stop working for only 1ms. You get what I mean with that? Or is it dumb?

on a tick based system, movements are being process exactly the same rate, there is no milliseconds to speak of

06:01

by all mean look at the code

06:02

if there is one or two ticks of discrepancy that is definitely a problem worth fixing.

06:04

If you just want to lower the client input latency. then the best you can do is to lower it as early in the chain as possible (hardware, input handling, fps) so the input can maybe catch a tick earlier on average. but for anything going through the network. an average of 1/25th second of input latency and jittering is expected

06:05

that's just how multiplayer game works. you could tap into subticks like CS2 does, but that requires a significant amount of protocol redesign, it is definitely doable but expect great difficulty and community push back. again. I'm not saying don't do it. I'm just saying it might not be worth your time to dive this deep into

why is server-side dummy unrealistic?

Ewan

why is server-side dummy unrealistic?

it's not

06:18

just no one really did it to the point where it is worth shipping

well u can still easily use old system on old clients

there were server-side dummy servers. but for ddnet I guess you'd be really good at pushing the idea forward cuz it's generally a community push back thing.

but you can have the server do both and opt into server-side dummy

06:19

it's a benefit for both client and server

I'm 90% sure if you just made a PR for server-side dummy, I would 100% support it

06:19

lol

less traffic and your dummy copy inputs and such will be predictable even on lag

TsFreddie

I'm 90% sure if you just made a PR for server-side dummy, I would 100% support it

perhaps

at this point it is all about changes

nobody wants to change anything

06:20

heinrich will find some reason to veto a pr i make

you'd really need to emulate all the features to disgusting details where it can comfortable for people to say it is worth it

06:21

the thought of it is already made most people not wanting to develop

06:21

beside, it is just a lot of effort to spend on for not a lot of features.

06:22

not that we are pushing features like crazy anyway. but still

most changes are ui stuff im noticing

also why am I commenting. I don't even PR

06:22

that's just because Robyt3 is very passionate about UX and he's probably the most active right now

true

06:23

that's kinda my forte generally

06:23

but there is no need for me to do it in ddnet (edited)

even within teeworlds, it is Robyt3 pushing UXs

06:23

phew

xd

was about to say you can't just tell others to stop working cuz it is useless

yeah

TsFreddie

also why am I commenting. I don't even PR

tbh, even if I were to make PR, idk what I can do

06:25

I feel like the thing I care about are already taken care of

ya there aren't a whole lot of ideas worth pursuing that aren't too big or too small

06:25

either add some 5 second thing

06:25

or accounts

well account system maybe, but that's not just ddnet. infrastructure is a commitment

honestly i wish i saw a little bit more leadership from deen because if he had chosen even just his preferred method we would probably have some kind of accounts system by now

06:26

he doesn't even have to do it. just decide

someone within the admins really has to push it plus we have to address the additional cost

yeah

06:26

additional cost i don't see being a massive issue

06:26

you won't need all that much storage

06:27

it already shares playtime stats by name

06:27

if they have infrastructure for that then putting shit in dbs based on accounts isn't gonna be more demanding than putting shit in a db based on their name

hah I start to receive mails https://mas-bandwidth.com/xdp-for-game-programmers/

cool

Ewan

additional cost i don't see being a massive issue

it's probably more than just storage.

why

06:47

the auth itself should have very little traffic

you need people to properly manage all that. mods are cool and all, but when it comes to accounts, we are holding players information and should be properly handled.

okay but there's no cost in that

it will be if you find out that no one are willing to help

volunteer operation