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 2022-03-17 00:00:00Z and 2022-03-18 00:00:00Z

[ddnet-maps:master] 1 new commit

f4dad7a A Short And Precise 4 - ddnet-maps

Yes hehe (@Ryozuki)

chillerdragon github stalker

sorgaxD

is it possible to make shotgun not automatic on a ddnet server?

yes

08:11

Well, with code changes

[ddnet-web:master] 1 new commit

a2cf198 New South Africa server - def-

Embedded image

09:04

what does the checkmark mean again? isnt that official ddnet servers or something?

yeah, means ddnet or kog

interesting

09:05

@fokkonaut ?xd

that's not one of ours, so it's kog

on kog tab i am

xxxD

[ddnet-discordbot:master] 2 new commits

ae2aade Add French flag - def- 4b3a401 Merge pull request #28 from def-/pr-fra - def-

09:25

[ddnet/ddnet-discordbot] Pull request closed: #28 Add French flag

09:26

[ddnet/ddnet-discordbot] Pull request closed: #27 new log channel, ignore $kick edits & admin, modc channels from being…

09:26

[ddnet-discordbot:master] 5 new commits

f212750 new log channel, ignore $kick edits & admin, modc channels from being logged - murpii 8260d47 add tester channel aswell - murpii d980ffa Don't log internal category, move moderator channels in it's own category - murpii dd5f381 missed on_message_delete - murpii 4efcfe3 Merge pull request #27 from murpii/patch-1 - def-

mhh so sad that c++ standard is still so hard behind with utf8 support

10:28

tried to make regular expressions with utf8 support, but not even c++20 adds it for std::regex

Image attachment

ouuu std::regex sounds cool tho

yeah especially multiple excludes are nice

my text parsing become really ugly latley i str_find the shit out of strings xd

[quakenet] ChillerDragon

ouuu std::regex sounds cool tho

haha std::regex is slower than starting php and using regex there

lmao

@Jupstar ✪ rust has first class utf8 support and fast regex

but isnt it implementation dependent?

10:31

or is it an API flaw?

i think it uses std::string right

10:32

it doesnt use string_view

10:32

so no optimizations

yeah they want to add string view support, soon™️

10:32

xd

https://stackoverflow.com/questions/52697265/why-is-c-regex-so-slow

Why is C++ regex so slow?

I am new to C++ and have to process a text File. I decided to do this with a Regex. The Regex I came up with: ((^\s^=]+)\s=\s)?"?([^"^\s^;]+)"?\s*;[!?\s* I have written my C++ ...

but honestly we have like 1000 entries, i dont really notice the slowness

10:32

vscode's regex is fast

10:32

i generally am impressed by its file search speed

Image attachment

10:33

the top

10:33

Image attachment

10:33

the bottom of the list

10:33

c++ stl slower than python

10:33

https://github.com/mariomka/regex-benchmark

xd

vscode on top

Image attachment

10:33

on the optimized rust wins

10:33

BASED

yeah nice

10:34

c++ has to hurry up a bit

10:34

not like its impossible to catch up

Ryozuki

c++ stl slower than python

Python has everything interesting like this implemented in C

10:34

so it mostly depends on what algo and optimizations, not language

i wonder if they include init time

10:35

in the bench

10:35

e.g starting python

10:35

xd

10:35

Measuring is done inside the programs to avoid include startup, reading and writing times on results.

10:35

oh it says it

how does :BASED: look like?

10:35

ryo send image

Image attachment

ty

xd

here u can find more https://betterttv.com/emotes/shared/search?query=based

BetterTTV - Shared Emotes:

Emotes from other users you can add to your chat on BetterTTV.

Embedded image

xd

https://benchmarksgame-team.pages.debian.net/benchmarksgame/performance/regexredux.html

regex-redux (Benchmarks Game)

regex-redux — How fast programs match DNA 8-mers and substitute magic patterns.

Embedded image

well if they use their own impl of regex i guess it will be faster

10:39

but c++ std regex is not

10:39

xddd

^^

10:39

well we could use boost regex with unicode support, but is always annoying to add dependencies

i wonder if C impl optimizes for ascii only sometimes

10:39

in rust its always utf8

but boost is really useful xd

isnt boost like giant

yes

like adding a second stdlib

10:39

xd

yeah

10:40

but a useful one xD

Image attachment

10:40

the c code looks like some over optimized stuff

10:41

i doubt it has real use case

i think all code looks like it

10:41

rust is also allowed to use unsafe operations

Typical problem with benchmarks: https://en.wikipedia.org/wiki/Goodhart%27s_law

Goodhart%27s_law

https://benchmarksgame-team.pages.debian.net/benchmarksgame/program/regexredux-rust-6.html

just always trade everything for performance xd

the top answer doesnt use unsafe

but still looks pretty similar to me

10:42

to the c code

i dont see a malloc with a explicit size

10:42

xd

10:43

let mut input = Vec::with_capacity(51 * (1 << 20));

10:43

oh well

10:43

i guess its the same

10:43

xd

i still think c++ should hurry up a bit, independent of good speed bad speed std::filesystem should also take u8strings and magically make utf8 work cross plattform

10:44

its kinda annoying to have so many #ifdef branches in our code

oh

10:44

i found an interesting rant about abi today

10:44

https://news.ycombinator.com/item?id=30704642

jasonpeacock

C Isn't a Programming Language Anymore

10:45

@Learath2 read this

10:45

xd

he talks about c? XD

@Jupstar ✪ about c being the protocol every language needs to be able to talk to

10:45

C is the lingua franca of programming. We must all speak C, and therefore C is not just a programming language anymore – it’s a protocol that every general-purpose programming language needs to speak.

thats kinda true yeah

read it

10:46

its a good read

10:46

xd

10:46

Image attachment

10:47

My problem is that C was elevated to a role of prestige and power, its reign so absolute and eternal that it has completely distorted the way we speak to each other. Rust and Swift cannot simply speak their native and comfortable tongues – they must instead wrap themselves in a grotesque simulacra of C’s skin and make their flesh undulate in the same ways it does.

Ryozuki

on the optimized rust wins

https://github.com/mariomka/regex-benchmark/compare/optimized the optimization step is outstanding xD (scroll to the bottom) (edited)

Comparing master...optimized · mariomka/regex-benchmark

It's just a simple regex benchmark of different programming languages. - Comparing master...optimized · mariomka/regex-benchmark

ah that also anwsers ryos question

Image attachment

10:53

you can just disable unicode

xd

Ryozuki

the c code looks like some over optimized stuff

it is part of the task https://benchmarksgame-team.pages.debian.net/benchmarksgame/description/regexredux.html#regexredux

regex-redux description (Benchmarks Game)

What the regex-redux benchmarks game programs should do.

Ryozuki

My problem is that C was elevated to a role of prestige and power, its reign so absolute and eternal that it has completely distorted the way we speak to each other. Rust and Swift cannot simply speak their native and comfortable tongues – they must instead wrap themselves in a grotesque simulacra of C’s skin and make their flesh undulate in the same ways it does.

This is quite silly, it's not that C was arbitrarily elevated to the position by some outside power. It is the de-facto protocol because it's simple and widely implemented. Not because a secret cabal of programmers conspired to enforce C supremacy

xDD

nobody said anything about a conspiracy

11:08

ur missing the point

it was just a language close to assembly that was successful

11:09

thats the simple reason

that post comes because https://thephd.dev/to-save-c-we-must-save-abi-fixing-c-function-abi

To Save C, We Must Save ABI

After that first Firebrand of an article on Application Binary Interface (ABI) Stability, I’m not sure anyone expected this to be the title of the next one, huh? It seems especially bad,

11:11

and the abi mess causes problems on other languages

11:11

like rust

11:11

https://github.com/rust-lang/rust/issues/54341

i128 / u128 are not compatible with C's definition. · Issue #54341 ...

While fixing various bindgen bugs related to long double, int128, (rust-lang/rust-bindgen#1370, etc). I realized that the following Rust program, in my x86_64 Linux machine: #[repr(C)] struct Foo {...

11:13

rust also has 170~ target triplets

11:13

due to different abis

11:13

the abi world is beautiful

No one is stopping the author for saving us from this lunacy by convincing 600 disperate parties to adopt a new sane abi that people definitely wont complain about in 50 years

where is the xkcd

11:14

https://xkcd.com/927/

xkcd: Standards

Learath2

No one is stopping the author for saving us from this lunacy by convincing 600 disperate parties to adopt a new sane abi that people definitely wont complain about in 50 years

yeah thats a true thing

11:15

when i look at GPUs... they give a shit about c alignment, you gotta request it for every device and code to work for every one

The thing about using C with it's technically undefined ABI and the mess of compilers and implementation specific magic as the FFI is just that it works. I doubt anyone actually actively likes the fact that we are stuck like this

@Learath2 https://en.wikipedia.org/wiki/Interface_description_language

Interface description language

An interface description language or interface definition language (IDL), is a generic term for a language that lets a program or object written in one language communicate with another program written in an unknown language. IDLs describe an interface in a language-independent way, enabling communication between software components that do not ...

11:17

monkalaugh

Maybe one day if someone manages to convince microsoft to drop almost 2 decades of backwards compatibility we can move to an IDL

dev once for android and you'll hate JNI xd

Learath2

Maybe one day if someone manages to convince microsoft to drop almost 2 decades of backwards compatibility we can move to an IDL

i'd already be happy if they finally allow c11 XD

One day hopefully

hi smart people

Hi smart guy

Hi non-rust user

11:20

BASED

Ryozuki

Hi non-rust user

Xenophobic

I just rusty at rust

Hi game developer

Hi vulkan developer

Hi rust webasm developer

monkaS

11:21

@Learath2 hi UB enjoyer

11:21

PepeA

I spend a day deciding how should i model friend relationship in mongodb

11:22

imagine using nosql

11:23

https://www.youtube.com/watch?v=b2F-DItXtZs

Episode 1 - Mongo DB Is Web Scale

I hate SQL with raging passion

listen to this based youuber

11:23

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

11:23

"shards are the secret ingredient in the web-scale sauce; they just work."

11:23

i love this parody

I'm scared of sharding tbh

11:28

Btw i submitted my system architecture to bilibili. They hit me back with a risk analysis. In the report they said we got a robust and experienced team with common and manageable stack.

11:29

While the team is me and I hate SQL

11:29

pepeH

doesnt it depend on the task anyway what you choose?

11:30

friend relationship probs makes sense in a relational database xd

But I really don't want sql justatest

justatest

just store the mongodb key in the sql database

11:31

then you have both xd

I just made an array of subdocuments describing all friends of a user

11:32

Need transaction to keep two users in sync, but the read is probably travial and fast enough.

Relationships in a non-relational db, that is quite ironic, don't you think? :D

11:33

But i hate relationship

11:33

That's why I'm still single

thats the spirit

11:33

dont use databases

11:33

use GPU VRAM

11:33

query over shaders

Fwiw I think your approach is what I'd do aswell. I don't really do web scale mongodb magic much anymore though so grain of salt required

I never made anything that scales.

Ryozuki

that post comes because https://thephd.dev/to-save-c-we-must-save-abi-fixing-c-function-abi

xD in https://thephd.dev/to-save-c-we-must-save-abi-fixing-c-function-abi#-4 it kinda looks like he is using comic sans in his editor xD

Do you actually need transactions? Is there really harm if things fall slightly out of sync?

Probably not

11:39

I manage blocked user in the same array btw

11:40

So if you block someone you are actually adding them as friends with a blocked state

I think(tm) leaving the update of the 2nd party on a "best effort" should be fine. Worst case on the next api call he will be informed that he got unfriended anyway

TsFreddie

I hate SQL with raging passion

oh no

here comes our SQL dev xd

I actually like sql, am I weird?

11:41

Thinking about things in terms of their relations with other things sounds like the sane approach to me :P

i think im the weird one from what i can tell

NoSQL is very common. It's web scale even

what does web scale even mean

No idea, it's what everyone says

just take it litteraly

watch the video to understand

xd

can you not spin up more mysql nodes to handle more loads greenthing

greenthing

joining can be slow

you dont need reliability, ACID is not web scale

Jupstar ✪

joining can be slow

xd

11:44

"it uses joints and writes to disks"

xdD

i have a question

i have many questions

11:45

xd

ok i don't have questions anymore

11:46

thanks bud, that helps a lot

@TsFreddie what is the physics engine doing tho?

which

11:48

where

11:48

when

thought u do client side prediction

11:48

of a physics engine

ok

11:49

then

11:49

i have no idea

11:49

unity does its thing

ez

and i hate it

just embedd electron and use a javascript physics engine

we did actually

11:50

we did have a v8 in unity

xd

for hot swap OTA update

did you host the engine online or what?

i mean, js are just text. if we can send text to an already running client. we can run it (edited)

interesting approach

11:52

arent you targeting mobile phones?

ye, chinese mobile games likes to do this

11:52

although most of them use lua

11:52

I did convince the designer guys not to update gameplay logic, so only the UI part are written in typescript

sound like a good place to do some RCE

well if i control the update server

11:53

I am the RCE

11:54

also we use v8 because i hate lua

11:54

as long as the core gameplay doesn't run on js, it should be fine.

11:55

stuttering menu isn't really a big deal

tru, nobody cares about UI anyway xd

and we are probably gonna limit the fps to 30 in menu anyway

are you like the lead developer, or are the other devs just nobos, that you decide so much

i'm the lead developer

PepeA

legend

because i'm the only developer for three months

11:56

then we hired more

oh

xDD

and they are stuck with my decision

then im a lead developer too

11:56

greenthing

kek

run electron inside electron

we don't really have good programmers tbh

11:57

the salary is too low

11:57

I'm just a 0 experienced wildcard and i don't want to work 15 hours a day to keep my job

what is wildcard in this context?

idk

xD

u mean u have no previous exp thats why ur in this job

since they all suck. i can do whatever i want

ah

even tho i have 0 experience

same here

11:58

xdddd

kinda sounds cool tbh

i made my company go from windows server 2008 to debian 11

money doesnt always matter

Ryozuki

i made my company go from windows server 2008 to debian 11

ez

monkaS

ye. but arrogance can get to you sometimes.

ye stay humble

trying

u can just curse them in ur inner voice

i don't curse irl so i'm fine

12:00

scratch that, i curse in english. and my brain doesn't hate it.

12:00

i forgot english curse words are curse words

do u bow to them 3 times

12:00

and how much deep u bow

12:00

xd

12:00

bow

i'm not apologizing japanese style all day

12:01

speaking of arrogance

12:02

i think sometime you need to act egomaniac to make people convinced of your decision kek

kek

12:02

as a 0 experienced worker

you are not that unexperienced

not keks

well if i'm doing interviews i'm probably worst than 0 exp

i did interviews and i convinced some ppl of the greatness of rust

12:05

but i didnt get chosen

12:05

so i probs will stop it

12:05

xd

huh

maybe they hate rust and choose not to hurt you (edited)

️

they were python devs

12:05

so

12:05

monkalaugh

so they probably hate rust

12:05

kek

justatest

TsFreddie

so they probably hate rust

they probs dont know it

fair enough

feelsbadman

Remember, most devs are average

xd

they dont do open source in their freetime

12:06

they work to work

thats a shocking true

12:06

truth*

man i wish the average is better than I think

thats why the ppl who look out for devs skip certain ppl with less xp

12:07

because they think (and probs right) that most normies only learn while working

12:07

so they miss on hidden gems like me

12:07

BASED

can we hire you

the language barrier is hard

12:08

otherwise if u paid me everything i would think it

12:08

xd

you'll only need to talk to me

but u use unity

12:08

i dont like that much c#

12:08

i only ever used to mod tshock

12:08

plugins

justatest

12:09

i didn't choose to use unity okay

https://github.com/TerraPlugins

12:09

that was so long ago

12:09

all these are mine

also question is if salaery is even enough to live in europe

12:09

europe probs a bit more expensive

just don't live in europe

ez

12:10

china is cool from culture

12:10

but i dont like the government

"git pull", 2 hours later oh it finished

it's probably not gonne be enough to live in beijing tbh

the great wall monkalaugh

monkalaugh

12:10

doesnt china have a silicon valley

12:10

?

Ryozuki

the great wall monkalaugh

monkalaugh

i don't think that can stop you

12:11

i'm still here afterall

bcs you are over average xD

12:11

most ppl couldnt do it xD

are you saying Ryozuki is average

monkaS

that's mean

ryo has rust-proxy

12:11

its fine

tf is that

12:12

xd

i dunno

i have ryo-os

a proxy written in rust

show me your kernel

ah

12:12

but im stuck cuz the next stuff is rly complicated

12:12

i need to implement context switches

12:12

and multitasking

12:13

Image attachment

12:13

Image attachment

12:13

i can write

12:13

with my keyboard

12:13

xd

12:13

monkalaugh

did you alt tab

alt g

12:13

so that qemu lets me out

europe

otherwise im trapped

12:13

alt+g

oh

not altg

i'm dumb

Ryozuki

Click to see attachment 🖼️

epic

i also want to implement a slab allocator

12:14

like linux

12:14

https://www.kernel.org/doc/gorman/html/understand/understand011.html

add a monitor/screen component

Image attachment

to render individual pixels

here's my password

leaks

thanks xd

TsFreddie

Click to see attachment 🖼️

"offenses"

12:16

is this the report to china police

nah

12:16

just ban and mute

PepeA

we can't even get user's actual id

12:17

bilibili does the verification and we only have an open_id and their access token, and their age range. not even the real age.

12:18

but the age range is probably good for us. cuz we can finally tell if the blockers are literal kids or maybe kids.

all my rust projects, 80% abandoned

Image attachment

I don't keep abandoned projects

12:19

which is probably a bad idea

Ryozuki

all my rust projects, 80% abandoned

what is the secret blured one xd

xd

12:49

from work

Ryozuki

Click to see attachment 🖼️

u should call it ryOS

deen

Anyone know why the github CI is missing on this change? https://github.com/ddnet/ddnet/pull/4829

hi @deen , just to tell you im the author of this PR if you might want to tell me something directly here on Discord. thank you.

new devs

poggers

hi @f and thanks for the change

translate to russ language (edited)

Image attachment

partial github outages be like ... while ! git push -f; do git push -f; done

[quakenet] ChillerDragon

partial github outages be like ... while ! git push -f; do git push -f; done

?

[ddnet/ddnet] Pull request opened: #4830 fix sound bug in back in times 3

I don't know if there is an issue for this but I think we all know the sound bug at least since the release of Back in Time 3. Namely when you are in certain parts of a large map or as a spectator to the right edge of a large map then it comes to integer overflow and you hear all players in full volume (as if you would create the sounds yourself e.g. by your own shooting). The integer overflow is even detected by our nice UndefinedBehaviorSanitizer: And can look like this: /home/dani...

lol

14:40

github down

14:40

Image attachment

14:40

never seen that before

Yup, everything is down except the webserver. Even the Markdown parsers are down. (edited)

14:43

https://www.githubstatus.com/

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

Jupstar ✪

Click to see attachment 🖼️

Then only the calculation in double? sqrt((double)dx * dx + (double)dy * dy)

yes

14:54

x87 uses 80 bit internally anyway

Do we want to do FalloffX and FalloffY directly in double as well?

i've nothing against it

I wonder why this is not already noticed when mapping the map or whether this was planned as a surprise for the devs xD

[ddnet-web:master] 1 new commit

8f245b7 Add Mexico - def-

1

1

f

hi @deen , just to tell you im the author of this PR if you might want to tell me something directly here on Discord. thank you.

damn you're not dead

@ReiTW do u know him?

who he

sure he's a developer of iF|City

ah uh

and the one who hosts iF|City

closed source boo

wats his name?

15:23

ffanta?

no

15:23

f. is f

just f?

f

yes

15:23

f

hmm idk him

press f

no front but sounds like this name does not meet the minimum requirements to be functional

its like ppl on irc who call themselves a common word

15:25

and they get highlighted all time

15:25

like someone named "is"

or oy like wtf

15:25

super annoying

15:25

im always loosing it when someone uses a smiley in tw as username and i try to address a message to him

15:26

"can you do this part please? :("

GitHub

Click to see attachment 🖼️

That CodeQL faild is because github is broken, right?

probs

[ddnet/ddnet] Issue opened: #4831 Team killed without message

``` [2022-03-17 15:34:48][chat-command]: 3 used /pause [2022-03-17 15:35:00][chat-command]: 5 used /pause [2022-03-17 15:35:09][game]: kill killer='0:' victim='0:' weapon=-2 special=0 [2022-03-17 15:35:09][game]: kill killer='1:Mitake Ran' victim='1:Mitake Ran' weapon=-2 special=0 [2022-03-17 15:35:09][game]: kill killer='2:cheeser0613' victim='2:cheeser0613' weapon=-2 special=0 [2022-03-17 15:35:09][game]: kill killer='3:' victim='3:' weapon=-2 special=0 [2022-03-17 15:35:09][game]: k...

Cipy29

.16 and .31 still work tho

Do we actually want to allow this kind of cheat? So that you can use your own position in Debug View to make certain jumps precise. If so, I could also output the 3rd and 4th decimal digits. There are actually only decimal numbers because we divide the actual integer position by 32 to make it easier to track the position on blocks. if we don't want to support cheats like that we could also just stop dividing by 32. I would be in favor of also outputting the 3rd and 4th digit after the decimal point. (edited)

c0d3d3v

Do we actually want to allow this kind of cheat? So that you can use your own position in Debug View to make certain jumps precise. If so, I could also output the 3rd and 4th decimal digits. There are actually only decimal numbers because we divide the actual integer position by 32 to make it easier to track the position on blocks. if we don't want to support cheats like that we could also just stop dividing by 32. I would be in favor of also outputting the 3rd and 4th digit after the decimal point. (edited)

its not really a cheat, it was discussed before that anything that can be achieved ingame without any modifications or external help is allowed

16:35

it's just a config that binds d to "+jump; +right", a to "+jump;+left"

16:35

and displays which coords work on a 3 tile freezejump

I mean the output of the player position in the debug HUD.

don't see a reason why it'd be a cheat personally. but not sure what the admins think about it

yeah, let's keep allowing that. It's too late to go back on that, even vanilla TW had that

16:37

and you can add two decimals

poggers

1

technically you can use coords to get an advantage, for example setting up towers and the freezejump, but anyone can have that advantage, it doesnt even take time to learn it. just just look at a number and line up

people have been using it for many years, yeah

You can even add a bind to send coords

really, how?

acqwerty has a modded client that shows the coordinates in green when its a correct value

no proof

I mean, it could be added, i think not exist at the moment

If u'll make debug show more digits and just leave debug as it is it'll probably make it more annoying to look at your x coords :P

16:42

Maybe it'll be good to generally clean up the debug info

16:43

I don't know if the stuff on there is useful

just make it display as a fraction

I don't even know peiple that know what half of it means xd

also, on the topic of debug menu, i'd also love to see this feature at some point https://discord.com/channels/252358080522747904/293493549758939136/927774860443275325 some users would definitely benefit from having their x coords, checkpoint, or angle pinned so that it shows even without having debug menu on

Cipy29

I don't even know peiple that know what half of it means xd

It is also meant for debugging and not as a game mechanic that should be used by players.

Do people find it useful for debugging?

idk u dont rly need any debug stuff 99% of the time

im sure @Im 'corneum checks coords when mapping his weird stuff

nah i dont know where stuff works different. its probably some kind of grid that loops but its probably also different for every kind of floating point part

bencie

also, on the topic of debug menu, i'd also love to see this feature at some point https://discord.com/channels/252358080522747904/293493549758939136/927774860443275325 some users would definitely benefit from having their x coords, checkpoint, or angle pinned so that it shows even without having debug menu on

could be added as an option for the nameplate xD

Def not nameplate

my nameplates would be way too crowded then haha

16:47

yeah

It's already too crowded

16:47

In the corner, could add the option of disabling the hearts from the HUD since they're useless in ddrace

16:47

I already have that in my client

Cipy29

In the corner, could add the option of disabling the hearts from the HUD since they're useless in ddrace

I'm currently working on the new HUD, so this will probably be included that the lagacy hud is hidden.

poggers

2

Sounds good :D ill look forward to seeing it

[ddnet/ddnet] Pull request opened: #4832 Add further contributors, thanks to all!

I hope I didn't miss anyone.

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 standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddne...

i use the debug graphs more than coordinate or angle information tbh

add rifle undeep

16:55

kek

[ddnet/ddnet] Pull request opened: #4833 predicted weapons now uses non predicted player positions if player p…

Motivation: current weapon prediction can make it look as if you hit stuff whilst the laser doesn't actually hit anybody on screen !image This should fix that issue by using the non predicted world;

Checklist

[X] Tested the change ingame
[X] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Writt...

another first time contributor

17:02

poggers

[ddnet/ddnet] Pull request opened: #4834 add link to https://wiki\.ddnet\.tw/wiki/Development on readme

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 standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--u...

lol

17:05

i broke the embed link

17:05

im a hacker!!!

17:06

oh, depends where u click on the title

[ddnet:master] 2 new commits

944ed54 fix sound bug in back in times 3 - C0D3D3V dc4c03d Merge #4830 - bors[bot]

17:07

[ddnet/ddnet] Pull request closed: #4830 fix sound bug in back in time 3

GitHub

Click to see attachment 🖼️

oh

17:08

is this the bug where u hear the sound

17:08

of everyone

yes

i remember it happening on the chill lets climb 3 tourny

should happen on large maps, integer overflow

nice to see it fixed

[ddnet:master] 2 new commits

2e5e077 add link to https://wiki.ddnet.tw/wiki/Development on readme - edg-l 31aba33 Merge #4834 - bors[bot]

18:06

[ddnet/ddnet] Pull request closed: #4834 add link to https://wiki\.ddnet\.tw/wiki/Development on readme

[ddnet:master] 2 new commits

1501c28 Add further contributors, thanks to all! - def- 971f839 Merge #4832 - bors[bot]

18:24

[ddnet/ddnet] Pull request closed: #4832 Add further contributors, thanks to all!

@Jupstar ✪ Is there any advantage to not using a quad for two different textures? Even if the dimensions and position are the same? E.g. in the HUD there are quads created for full and empty hearts and shields does that make sense? https://github.com/ddnet/ddnet/blob/971f839c1792563b52fe65bd5d6de4495374cdcd/src/game/client/components/hud.cpp#L706-L741 Should I use a separate quad for each sprite?

ddnet/hud.cpp at 971f839c1792563b52fe65bd5d6de4495374cdcd · ddnet/d...

DDraceNetwork, a cooperative racing mod of Teeworlds - ddnet/hud.cpp at 971f839c1792563b52fe65bd5d6de4495374cdcd · ddnet/ddnet

[ddnet/ddnet] Issue opened: #4836 Kill splash effect not respecting Teeworlds Team color

Currently when a game mode uses mode with Blue and Red team. The client automatically overwrites the tee colors for their respective teams. All effects and displays, whether in the kill history, in the tab and in the game respect the color of the current team. With the exception of one effect, the explosion effect on tee death. This effect continues to follow the skin color or custom tee color, rather than the team color. The teeworlds 0.6 server side, under the hood, updates tee colors...

c0d3d3v

@Jupstar ✪ Is there any advantage to not using a quad for two different textures? Even if the dimensions and position are the same? E.g. in the HUD there are quads created for full and empty hearts and shields does that make sense? https://github.com/ddnet/ddnet/blob/971f839c1792563b52fe65bd5d6de4495374cdcd/src/game/client/components/hud.cpp#L706-L741 Should I use a separate quad for each sprite?

that's your choice If you rework the hud: the idea generally is to return the quad offset and have member variables named HUDHealthHeartOffset so its nicely readable but yeah i havent done that yet and to answer your question You could argue that its nicer to split it for more modularity, but mathematically and GPU wise its not wrong to reuse the same quad

Jupstar ✪

that's your choice If you rework the hud: the idea generally is to return the quad offset and have member variables named HUDHealthHeartOffset so its nicely readable but yeah i havent done that yet and to answer your question You could argue that its nicer to split it for more modularity, but mathematically and GPU wise its not wrong to reuse the same quad

oh and i think it just recently added that feature in https://github.com/ddnet/ddnet/pull/4728/commits/1f5a5b038258dc56502c3f258ab6165ce0d4b7f6 guess you can cherry-pick it if you want to use it

Jupstar ✪

that's your choice If you rework the hud: the idea generally is to return the quad offset and have member variables named HUDHealthHeartOffset so its nicely readable but yeah i havent done that yet and to answer your question You could argue that its nicer to split it for more modularity, but mathematically and GPU wise its not wrong to reuse the same quad

As far as I understand it, the main reason to use a separate quad for each object in the HUD is that they do not change their position, right? Then the position data of the quad does not have to be recalculated with every render. With RenderQuadContainerAsSprite on a moving object more calculations would have to be done than with RenderQuadContainer only.

there is another way to reposition something

19:07

and that is by adjusting the projection matrix

19:08

that is always uploaded to the GPU anyway

19:08

but yeah you are right it doesnt need to reposition it if you directly map the vertex correctly

@ReiTW im always dead

yes I see that hollande

hollande

19:11

when noby's antibot on ifcity

@ChillerDragon you don't no me? i know you

19:11

don't be ashamed and confess to everybody here

19:11

you used to play ifcity

19:11

nub

c0d3d3v

As far as I understand it, the main reason to use a separate quad for each object in the HUD is that they do not change their position, right? Then the position data of the quad does not have to be recalculated with every render. With RenderQuadContainerAsSprite on a moving object more calculations would have to be done than with RenderQuadContainer only.

mh yeah seems like our current implementation doesnt allow it for non sprite like quad containers, so i guess thats the main reason then

19:12

tho it could be implemented without changing the backends

19:12

by moving the projection matrix

@f this ping wont work, he's on matrix room

nvm ^^

Jupstar ✪

mh yeah seems like our current implementation doesnt allow it for non sprite like quad containers, so i guess thats the main reason then

Ok

Then I'll just try around and as soon as I have something to show, we can discuss again whether I have done that well...

How can I read in render.cpp render function from a named pipe?

19:49

I tried this but works with a test.cpp snippet, but not in the gamecode. (Have no knowledge of cpp at all)

std::ifstream file{"/tmp/pipe"};
std::string line;

Not really the best idea to do a file read on the very very hot path

1

My ideas are always best

(edited)

Jupstar ✪

x87 uses 80 bit internally anyway

compilers use sse for float computation nowadays

then the registers are even bigger

ye thats why u disable sse and enable soft floats

21:28

when doing kernel

xd

you also have to disable the red zone

21:31

monkalaugh

21:31

https://os.phil-opp.com/red-zone/

Disable the Red Zone | Writing an OS in Rust

The red zone is an optimization of the System V ABI that allows functions to temporarily use the 128 bytes below its stack frame without adjusting the…

now we have chairn, we need cache optimizations

21:33

64bit probs slows the cache a bit xd

well, i'd say data cache is probably fine, however, instruction cache must have a substantially high miss rate i guess

21:57

ehm

$ perf stat -e instructions,L1-icache-load-misses,L1-dcache-stores,L1-dcache-store-misses,L1-dcache-loads,L1-dcache-load-misses ./DDNet

   259 648 927 688      instructions
     6 707 225 567      L1-icache-load-misses     #    2.58% of all L1-icache hits
    47 319 023 683      L1-dcache-stores
       700 778 077      L1-dcache-store-misses    #    1.48%
    71 666 451 733      L1-dcache-loads
     2 677 123 546      L1-dcache-load-misses     #    3,74% of all L1-dcache hits

     162,973244501 seconds time elapsed

(edited)

i think thats good

i didn't try to attach it to an already ongoin game

22:10

maybe its mostly initialization

Exported 557 message(s)