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 2021-09-10 00:00:00Z and 2021-09-11 00:00:00Z

Moving from Eclipse to Visual studio

‍♂️ Seriously the progam feels so bad to me (edited)

Bruh eclipse

bruh eclipse

Bruh vs studio

Ryozuki

Bruh vs studio

Visual studio studio troll

troll

justatest

06:23

https://xkcd.com/2513/

xkcd: Saturn Hexagon

06:23

monkalaugh

‍

ahahahaha (@Ryozuki)

https://www.youtube.com/watch?v=b2F-DItXtZs&t=19s

xd

ChillerDragon u must web scale

07:04

"mongodb handles web scale"

07:04

xdd

07:05

"Mongo DB handles web scale. You turn it on and it scales right up"

Moving from one of the most annoying IDEs ever made to the worst IDE ever made sounds like a good way to quit programming

So you switched from emacs to Visual Studio ?

Learath2

Moving from one of the most annoying IDEs ever made to the worst IDE ever made sounds like a good way to quit programming

kek

I think I'm slowly losing my mind. How can I possibly keep failing this dumb computer networks exam ...

you study networks?

09:56

teach me pls

09:56

do you now how network namespaces work?

09:57

man ip-netns

09:57

dat boi

network namespaces give you a place where you can configure separte network views for separate programs

yas!

09:58

i want to use two interfaces at the same time

Sus

no proof

no talking about breaking the rules chillerdragon

ok i want to use two ips for two tw clients ( FOR EDUCATIONAL PURPOSES )

I'm done helping you with that (and people writing "for educational purposes" usually have something bad in mind)

yea educational purposes is a big meme :D

10:00

i actually want to use it to by pass max clients ip limit on fokkonaut server

You were so close too

?

10:01

did you reply to some msg?

10:02

can we ban discord replys?

Nah I replied to you. You were so close to learning how to do it but you just had to tell your reason for wanting to learn it :P

oh

omg matrix ded

10:04

otherwise you would have teached me? senpai learath :)

Ofc not, I'm scheduled to sulk for the rest of the day

troll

10:05

intransitive verb

10:05

: to be moodily silent

10:05

sulk

schmollen?

ah lol

10:05

What an absolute joke. I actually administer networks, how can I not pass a computer networks exam? Pure insanity

go teach me netns u will learn a lot

10:06

https://zillyhuhn.com/cs/.1631268342.png

10:06

zwelf unban me from matrix

10:06

aaa

10:06

omg my screenshots are ded

Everything is breaking down

ez fix

10:07

permissions master dragon

a matrix bacc

10:07

thanks zwelf

10:08

if it helps learath i still love you even if you failed ur exam :)

10:08

:3

10:10

im sure the students that only learn for tests and pass the exam could not administer networks in the real world like you my lord

chiller u need to learn to use google more

ikr

idk if its actually ur lazyness or that u dont know

10:11

but honestly

bit of both

most stuff u ask i always find with 1 google

thats why i ask u

10:11

u so good

10:11

wat u found now?

twintri

10:11

idk i didnt search that

10:11

actually i dont know much about that

nob

and even if i knew heinrich is here to use the banhammer on me on the asap

10:12

monkalaugh

i just fixed my mail server feels so good

postfix?

yas

10:12

more dovecot

10:12

was the issue

10:13

big pain in the ass

the next time i setup a email server ill use https://maddy.email/

10:13

im just to lazy to switch over rn

10:13

It replaces Postfix, Dovecot, OpenDKIM, OpenSPF, OpenDMARC and more with one daemon with uniform configuration and minimal maintenance cost.

the next time i setup a mail server ill dont

10:13

axaxax

10:14

i am a cheap maintainer

10:14

i do not get paid

Learath2

What an absolute joke. I actually administer networks, how can I not pass a computer networks exam? Pure insanity

are exams always close to reality?

10:16

maybe thats it

Ryozuki

It replaces Postfix, Dovecot, OpenDKIM, OpenSPF, OpenDMARC and more with one daemon with uniform configuration and minimal maintenance cost.

Sticker

monkaS

What is this the anti-unix newfangled web 69.0 software?

@Learath2 the kernel doesnt follow the unix philosophy

10:18

its a monolithic kernel

Thus we should make everything else monolithic too, got it

but anyway u dont know the pain of configuring all that shit

10:18

well maybe u do

I have the same setup my guy

10:19

Well not anymore because big-email doesnt let you host your own mail now

i still host it

10:19

gmail accepts my emails but oddly they take 10 mins more

10:19

to receive them

Got bored of all my mail ending in spam when I satisfy all the requirements on googles site. They just dont like my hoster

10:22

Fwiw if there was an open source movement to create a nice kernel based on microkernel architecture I'd support it. Alas it's too big of an undertaking now

Ryozuki

are exams always close to reality?

Not even close a lot of outdated shit. Like who even has CSMA/CD on their network? We have full duplex links and bus topologies aren't a thing anymore. It's not 1980

monkalaugh

Anyway, I'll just go cry for a bit. Then I'll think of wtf keeps going wrong when I take this shit

@Learath2 when are you going to merge?

Not like we can get a release until @heinrich5991 decides on what to do with the startline thing

what if they skip start line drag the 1-2 tees to finish(when they started) go over start and then kill going back a map is mostly faster than the opposite bcs of the teleporters

11:57

still a semi startline skip xd

12:01

maybe spawns should just always be protected by startlines, and disallow startline skips at all, instead of making it more complicated xd

12:02

where are patigas tools, that can detect it xd

@Deleted User pretty sure that you need to touch the start line before the finish line, else it won't count

yes

12:14

u skip with 10 ppl (in team) -> you drag 2 from start to finish -> the remaining 8 tees go to the start and kill

12:14

still have a fast finished team

12:14

its the same skip as fixed by the patch just with going back to startline

12:16

would be easier to define a non broken spawn, e.g. the area around a spawn should have solids(unskippable), or a startline and if you can get ninja(e.g. afk king) it requires the teleoutcome or startline to have shield before it

12:16

or if there is a tele gun, disallow shooting through start

Deleted User

u skip with 10 ppl (in team) -> you drag 2 from start to finish -> the remaining 8 tees go to the start and kill

do the two tees that get dragged touch the start?

yes

I dont understand the skip yet

do you understand the patch?

12:18

it does the same except the last part

12:18

10 tees, all skip 2 get dragged, the remaining kill

ah, haven't seen it yet

that was fixed

12:18

it its still atleast doable semi, depending on the map (edited)

I'll look into implementing https://github.com/ddnet/ddnet/issues/4088#issuecomment-915394258

Improve start line skip patch · Issue #4088 · ddnet/ddnet

A non started tee leaving the team makes the team unfinishable. This made a couple teams fail some with multiple hours in their timers. Without #3091 system messages are unreadable to some of our p...

[ddnet/ddnet] Pull request opened: #4148 Kill unfinishable teams if they don't enter /practice within a minute

This is to avoid players playing in unfinishable teams for a long time if they don't understand the system messages. Fixes #4088.

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 if it works standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect exis...

https://www.theverge.com/2021/9/10/22662320/epic-apple-ruling-injunction-judge-court-app-store

Apple must allow other forms of in-app purchases, rules judge in Ep...

The app store model on trial.

Hey. Today I was coding some Rust and hooking up a specific web rendering framework to a GUI framework. The web framework renders HTML to a framebuffer and I pass the framebuffer to the GUI framework to render it as part of the GUI. I could just pass the pointer to the framebuffer in memory but the problem is, the web framework outputs BGRA but the GUI framework only has options of RBG and RBGA. So I wrote some Rust to allocate a 500x500x3 vector of u8's and do the conversion in a loop. My current code has it set to 500x500 but it's supposed to be dynamic later. But the loop is super CPU heavy! How hard can it be to flip those bytes around? Do I need some from of SIMD? Never dealt with something like this. Thought someone here might have useful input

19:08

I need to do this each frame

BGRA and RBGA.. really weird formats Only chance todo it fast is on GPU probably, if you need it per frame

19:10

but reading from gpu memory might also be slow on dedicated cards

Yeah I'd need to get the input to the GPU and retrieve the results from the GPU. Sounds kinda heavy

19:12

I mean the whole web rendering is happening on the CPU

19:12

I don't get why my simple byte flipping thing should be the bottleneck here

bcs math for single pixles is the most expensive thing

I guess I could multithread it xD

i mean if you only target fast computers it could be fine

I want a feature that notifies on desktop whenever a friend goes online. For how much can anyone do this?

install steam

Need to notify when a teeworlds friend goes online

install steam

Actually it turns out if I build the Rust with --release everything runs MUCH faster. Might even be LLVM doing some SIMD magic behind the scenes

timakro ah, tight loops benefit a lot from --release even without simd

DMT: Not so hard actually with https://master1.ddnet.tw/ddnet/15/servers.json

Quite hacky, but something like this

while true; do
  friends=$(cat ~/.teeworlds/settings_ddnet.cfg | grep "^add_friend " | sed -e "s/^add_friend \"\([^\"]*\)\".*/\1/" | while read line; do echo -n ".==\"$line\" or "; done; echo)
  curl -s https://master1.ddnet.tw/ddnet/15/servers.json | jq ".servers[] | . as \$server | .info.clients[].name | select($friends false) | \"\(.) is on \(\$server.info.name), connect via steam://run/412220//\(\$server.addresses[0] | sub(\"tw-0.6\\\+udp://\"; \"\"))/\""
  sleep 10
  clear
done

[ddnet/ddnet] Pull request opened: #4149 Always send correct zoom dimensions

It is required to reset the Target after it being used, in order to not trigger this code, which would always send the m_LastZoom. This fixes it and always sends the correct zooming dimensions.

Checklist

[ ] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test if it works...

@heinrich5991

22:44

what do we do about it?

22:45

So, generally, I want to know the clients "real" dimensions, when at zoom level 10 (1.0f) by not sending the allowzoom flag for a short time, in order to calculate the cursor position while zooming in/out to match the actual position on the screen in the client

22:45

it works, just the fetchting of default show distance is not working properly

22:48

Also, when joining a server it doesnt send the zoom level at all, if it was the same on the other server you just came from.

sounds like a bug

22:50

I commented on the PR

(did you see the first line as well?)

yes, testing it right now seems to work

23:04

looking at the code, you can probably extract the original zoom coordinates without going to zoom level 10

how?

23:04

that would be amazing

23:05

i doubt, because i dont know the client's resolution

but I don't think the client sends its resolution (even on zoom level 10)?

23:06

I see no references to the actual screen size in CRenderTools::CalcScreenParams

it doesnt, but thats fine, i just need the show distance x and y at the level, where the input target matches the actual cursor pos on the screen

heinrich5991

I see no references to the actual screen size in CRenderTools::CalcScreenParams

How would i calculate it back?

I'm trying to figure out the formula rn

oh, okay

divide width by height and pass that to the CalcScreenParams function, zoom=1

23:08

aspect = width/height

what? xxd

23:09

ah

23:09

so i need that function on the server?

you can copy paste it

23:09

it doesn't have dependencies

let me try

23:14

WHAT

23:14

it really works

23:14

you are a magician

23:14

no wait

23:15

how is this possible

how is what possible?

fokkonaut

you are a magician

math!

i see, your math studying really pays off

23:17

xdf

[ddnet:master] 2 new commits

7aa4652 Always send correct zoom dimensions - fokkonaut 1feea5f Merge #4149 - bors[bot]

23:20

[ddnet/ddnet] Pull request closed: #4149 Always send correct zoom dimensions

heinrich5991

sounds like a bug

should be fixed i guess

23:36

i think resetting m_LastZoom on connect is okay?

hmm, it's already reset in OnReset()

23:39

is that not being called?

no its not set there, m_LastZoom belongs to CGameClient

23:39

since its only for whether to send the zoomstate or not

CGameClient::OnReset: m_LastZoom = .0;

also, i noticed, when holding bind for zooming in, it will send the new zoomstate when key is released, but when holding zoom out, it will send each step in between

heinrich5991

CGameClient::OnReset: m_LastZoom = .0;

mh

when zooming in, it only sends the new zoom state when it finishes zooming

23:41

when zooming out, it immediately has to send something, because more stuff gets into the view port

if you want to avoid potential tees jumping in, yes

yes, that's intended

23:43

theoretically it could try to merge these zoom requests until the they're sent out

Can the server detect dyn cam?

23:45

because dyn cam does not work with this cursor hack

23:45

why tho, lol

23:46

as soon as the camera starts moving away from the tee, so the tee is not in the center anymore, the cursor position will start to be further away than the displayed clientside cursor

23:47

Image attachment

23:47

Image attachment

Exported 216 message(s)