DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.org/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 2023-12-23 00:00:00Z and 2023-12-24 00:00:00Z

Yes that much is obvious from the screenshot. But why? What are you wiresharking?

Replying to @Bota I pulled libtw2 and tried to build wireshark plugin

00:35

What you did looks good. No idea what the issue is. Maybe wrong rust version. Not sure if anyone ever tried to build that on windows you can try pinging Heinrich but he will read this eventually anyways I think

Replying to @Bota how i do that. I dont know rust

what toolchain are you using? msvc mangles symbol names differently from gcc and clang so it would expect different function names

don't forget the world depends on #7691

https://github.com/ddnet/ddnet/pull/7691

add rescue on kill tile by ewancg · Pull Request #7691 · ddnet/ddnet

Some people were tired of having to re-enter practice mode, teleport back to where they were, and situate their weapons/solo status when hitting a kill tile in practice mode. The game will now resc...

@mbvxoxo give me another idea

the world depends on monotonic buffer resource

true

07:23

@tela what macos ver u on

DDNet

https://github.com/ddnet/ddnet/pull/7691

@Ewan if its for me i would merge it but i want atleast another one to agree

yea

07:33

nobody looking at it

i gave my

of aproval

seal

07:33

i love seals

07:33

some ppl say my dog looks like a seal

07:33

but seals are just water dogs

07:34

give me an idea

xd

07:36

@Ewan did u see deen comment on my pr

07:36

should i make CHeap thread safe

Yes

the old way isnt and this isnt

that would be sick for when i have another go at async skins

but whats the point if the

is holding me back

i think when i do it again i will just batch them and not try and have them load in the bg with the game xd

07:36

no matter how i do that i will still get micro stutters

07:37

bcs it has to wait to upload to the gpu at an irregular time

Ryozuki

but whats the point if the

is holding me back

maybe it isn't

07:37

depends on how deen feels about dropping compilation host support for already very old macos (edited)

07:38

who is Shereef Marzouk

idk

07:39

why u ask

Ewan

depends on how deen feels about dropping compilation host support for already very old macos (edited)

but the ci on gh is using macos-latest iirc XD

yea

Ryozuki

why u ask

he's (c) at the top of ddracechat.cpp

07:39

maybe other files

07:39

never heard the real name before

Ryozuki

but the ci on gh is using macos-latest iirc XD

i'm sure you can still run the binary compiled by latest macos on <14

Ryozuki

should i make CHeap thread safe

Why?

Ewan

who is Shereef Marzouk

Greyfox

@Learath2 read deen comment

07:45

Image attachment

as long as CHeap doesn't have bad single thread perf because it's thread safe. which i can't see happening anyway

did u check my pr yet

07:46

https://github.com/ddnet/ddnet/pull/7692

simplify CHeap using std::pmr::monotonic_buffer_resource by edg-l ·...

https://en.cppreference.com/w/cpp/memory/monotonic_buffer_resource Checklist Tested the change ingame Provided screenshots if it is a visual change Tested in combination with possibly related c...

07:46

cheap is rarely used too

07:46

only for localisation and another thing

I think it should be made thread safe at the point of use. That's how most containers are unless they are specifically crafted to be threadsafe usually locklessly

apparently macos comes with gcc 13 in github ci?

07:46

can se use that?

07:47

https://github.com/actions/runner-images/blob/main/images/macos/macos-13-Readme.md

runner-images/images/macos/macos-13-Readme.md at main · actions/run...

GitHub Actions runner images. Contribute to actions/runner-images development by creating an account on GitHub.

Learath2

I think it should be made thread safe at the point of use. That's how most containers are unless they are specifically crafted to be threadsafe usually locklessly

i see

Do we even use it from multiple threads ever anyway?

Image attachment

07:49

Image attachment

07:49

those are the uses i found

07:49

is the serverbrowser MT?

07:50

Image attachment

07:50

is the console execution queue mt?

i believe so

07:51

dont take my word for it

in macos

07:51

CXX=/usr/bin/g++-13 would work?

no

Config using it now might explain some of the bizarre new bugs

github macos containers have gcc 13

maybe in the ci it might work

07:52

idk what they use

Clang/LLVM 14.0.3 Clang/LLVM (Homebrew) 15.0.7 - available on $(brew --prefix llvm@15)/bin/clang GCC 11 (Homebrew GCC 11.4.0) - available by gcc-11 alias GCC 12 (Homebrew GCC 12.3.0) - available by gcc-12 alias GCC 13 (Homebrew GCC 13.2.0) - available by gcc-13 alias

unrelated q but why is my chat command's pResult->NumArguments() only saying there is 1 arg when in fact there are 2

ddnet_lgbt

ddnet_lgbt

1

// execute config file
    if(pStorage->FileExists(CONFIG_FILE, IStorage::TYPE_ALL))
    {
        pConsole->SetUnknownCommandCallback(SaveUnknownCommandCallback, pClient);
        if(!pConsole->ExecuteFile(CONFIG_FILE))
        {
            const char *pError = "Failed to load config from '" CONFIG_FILE "'.";
            dbg_msg("client", "%s", pError);
            pClient->ShowMessageBox("Config File Error", pError);
            PerformAllCleanup();
            return -1;
        }
        pConsole->SetUnknownCommandCallback(IConsole::EmptyUnknownCommandCallback, nullptr);
    }

08:00

@Learath2 i see this but where is the callback called on execute file? xD

Look for a findcommand somewhere in executeline

Ewan

unrelated q but why is my chat command's pResult->NumArguments() only saying there is 1 arg when in fact there are 2

i guess it's always reporting as 2 when it's 3 or more

ah

is it not delimiting args by space after the actual name of the command?

08:01

do i have to do that myself

Ewan

is it not delimiting args by space after the actual name of the command?

What does your arg string look like?

08:02

If you have an r it won’t split anymore

but i dont see where a thread is started

an r?

08:02

an actual 'r'

When you register a command you give a definition of it’s arguments

08:03

Something like “sss” for a command that takes 3 strings e.g.

oic

08:03

i didn't realize that actually did anything

(edited)

r means rest, so it’ll give you all the rest of the string at that point

gotcha

we should speak british english

08:04

colour

08:05

now that i have lot of free time its time to do prs

08:05

and redeem myself as a ddnet dev

lol

(if macos doesnt hold me back)

holidays moment

08:05

anyone here ever rent a macos vps

i have a m1 myself

oh yeah i forgot

08:05

lucky

from work

08:06

@Learath2 did u check ewan pr

08:06

feel free to merge it

08:07

do u know what version of clang supports pmr

can CConsole::CResult::GetFloat ever return NaN

15? 16? 17?

no idea

well with brew u can get llvm 17

08:09

so

wanted?

08:14

08:15

i made it so u can /tp x y

08:15

/tp or /tp playername or /tp x y

@Learath2 why on gentoo i have gcc 12

08:16

and not 13

08:17

omg

08:17

i have it in eselect

08:17

nvm

i will make pr

of what

The thing i just sent

Ewan

can CConsole::CResult::GetFloat ever return NaN

NaN should not be possible

However you can get, HUGE_VAL on overflow and 0 on underflow

08:19

We don’t check or catch either

wow awesome

Oh now that I take a look at it we don’t check for invalid either, so you should get a 0 on things that can’t be converted to float too

08:20

Teeworlds moment

i will parse it myself then

/Users/runner/work/ddnet/ddnet/src/base/math.h:6:10: fatal error: algorithm: No such file or directory 6 | #include <algorithm>

08:20

tf why g++ 13 fails

was planning to prepend with + or - for relative coords anyway so i couldn’t use getfloat

Ewan

was planning to prepend with + or - for relative coords anyway so i couldn’t use getfloat

It should handle negatives just fine, no?

not what i mean

08:21

10 = absolute 10 +10 = current pos + 10 -10 = current pos - 10

Aha, yeah can’t do that with float

ya

08:22

or i could be more explicit and use tildes like minecraft

08:22

but doesn't seem needed

08:22

it's just ~ vs +/-0

08:23

but uhh actually can you place things at negative coords in a map

i feel like i've seen it before

08:23

in which case i would have to use a tilde or smth else

08:23

need to know real map boundaries

08:26

sounds like what i want is the entity clipping rectangle?

ok i surrender

08:27

fck macos

08:27

i hate ci driven development

xd

this also means i cant modernize other parts

08:34

cuz macos

08:34

imagine saying u support c++17 but not rly

yea

i was about to say smth

08:36

but it would piss u off

08:36

lol

08:38

go for it

Voxel really did rq :(. He was cool. Hope he comes back. Although he did say ddnet wasn't as fun anymore couple of times.

life is not daijoubu

08:40

08:40

i love this meme

why do all our artists rq xd

I still can’t believe he left because Ryozuki preached rust. It’s like getting angry at a priest for reading the bible

we already have too few

Learath2

I still can’t believe he left because Ryozuki preached rust. It’s like getting angry at a priest for reading the bible

i dont think its just that

08:41

he was frustrated with maps too

08:41

he said all new maps were unfun or smth

Jupstar ✪

we already have too few

i guess artists arent as keen to open source art as devs are for free software

08:42

we are more selfless

it's just bcs they are paid bad in RL

maybe

as dev u get a new job after 1 sec xD

I think artists want more recognition than programmers tbh

xd

...anyways

ye programmers stay in the shadows

08:43

normal ppl dont know john carmack

well he was not paid to be an artist afaik

true

08:44

idk what ravie does

db manager

i mean as job

xd

08:44

idk

guys what are the actual map boundaries

Ryozuki

idk what ravie does

He is a professional edgelord. Gets paid to have unwoke opinions

let's say for entities

08:45

i just need a number rn to test this

Plays the accordeon

-i16 +i16??

08:45

idk

08:45

xd

08:46

omw

where SHORT_MAX macro xd

08:46

feelsbadman

i just said it randomly idk tbh

Ewan

guys what are the actual map boundaries

I doubt there is any if you remove the death border, you just go out until floats become very inaccurate and bizarre things happen

08:46

Teeworlds farlands

k what is the death border

i think its 20 tiles away from the map size

ouu

or smth like that

can maps go negative

08:47

on either x or y

Learath2

Teeworlds farlands

basically if it goes to negative after hitting float limit, it wont render theorically

Cellegen

basically if it goes to negative after hitting float limit, it wont render theorically

u sometimes dont make sense xd

ig i should ask if maps can exist outside of quadrant 4

Cellegen

basically if it goes to negative after hitting float limit, it wont render theorically

It shouldn’t, float overflow works differently

float loses accuracy as it goes further from 0 iirc

so it would just reset (as in if its 64 or 128) hitting limit (edited)

so its not about not being able to render

float overflow is a myth

You’ll get stuck at HUGE_VAL

just use float128

1

08:48

big boi float

is huge val also for negatively huge numbers

08:48

tending towards - infinity

HUGE_VAL + d == HUGE_VAL forall d > 0 iirc

https://en.cppreference.com/w/cpp/numeric/math/HUGE_VAL

cpp/numeric/math/HUGE VAL

08:48

HUGE_VALF Expands to positive float expression that indicates overflow HUGE_VAL Expands to positive double expression that indicates overflow, not necessarily representable as a float HUGE_VALL Expands to positive long double expression that indicates overflow, not necessarily representable as a float or double

Ryozuki

i think its 20 tiles away from the map size

i guess i will use placeholders for now. i need to know how to get this info tho

https://doc.rust-lang.org/std/primitive.f64.html#associatedconstant.INFINITY

A 64-bit floating point type (specifically, the “binary64” type defined in IEEE 754-2008).

08:54

@Learath2 did u know arm uses 16 byte align

08:54

macos

Ryozuki

@Learath2 did u know arm uses 16 byte align

I did not know it as fact, but I’d guess it given thumb instructions are 16b

ah wait

08:55

its 16 for > 128bit

08:56

https://github.com/ziglang/zig/issues/2987 https://github.com/rust-lang/rust/issues/54341

make the alignment of >= 128 bit integer types 16 bytes to prevent ...

@cmpxchg on x86_64 supports u128 integers. However @alignOf(u128) reports 8, and without align(16) on the u128, the code segfaults at runtime. This alignment is reported by LLVM as the "ABI al...

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 {...

08:56

ok wait this is unrelated

08:56

xdd

08:56

but this shit came up at work

08:56

when making the compiler

damn

08:57

okay i added relative modes

08:57

i still need to know if maps can exist outside quadrant 4

08:57

so i can reserve -X.X for non relative purposes

08:57

absolute

08:57

sdiffjsd

but ye 64bit arm is 16

09:02

tbh it would be easy to just make it ~+ or ~- instead

09:08

im going with tildes for now

ur a tilde

wtf

09:18

rude

Ewan

Click to see attachment 🖼️

Ngl, but this seems more unpractical to use than current tp

yea i too love zooming in and inching towards the exact tenth place offset i need

09:45

???

09:45

it exists for a different purpose

09:45

also can be bound...

The only real application I see tp this it's if I wanted to go far to some direction to save me like 2 secs of dezooming. Otherwise most of tp needs too much precision for me to somehow guess the exact coordinates

well then i didnt make this for you

09:52

i think it would be quite helpful

Talking about tp, something that could really be useful is not falling just after tp (already happens when using it with spec)

how else would it work

09:53

i guess it wouldn't be hard to put the player in a limbo state but i dont want to mess with prediction code

Ewan

well then i didnt make this for you

^^ I was just giving my opinion. I'm sorry to hurt you? Idk. If you receice some positive feedback I'd like to know the use of it, because I can't really come up with anything other than fast travel to far distances

have you ever played a map and wanted to test a setup on the fly or something

09:54

you can iterate on starting coordinates for the setup using this without just tapping a and d until your fingers bleed

Oh, ok I see what you mean

it seems like you don't really play the game because it's really really common for me to be playing with other people & all of the sudden coordinates matter a lot

You want to use it to repeat a setup

or i guess you have a radically different playstyle

Ewan

it seems like you don't really play the game because it's really really common for me to be playing with other people & all of the sudden coordinates matter a lot

Well, I usually just accomplish whatever I try eithouth needing 99999 attempts

09:56

Ok I ll stop the beef here, but y I see

i just got trolled hard by vec4

10:15

spent about 30 min trying to figure out why i get sent out of bounds & subsequently killed only when my bounds checking is on

10:15

std::clamp(BaseY, Boundaries.y + 32.f, Boundaries.y + Boundaries.h - 32.f)

10:15

and i was doing it like that

10:15

but .h is the same as .x.... it's h for HSL

10:16

i needed .l (length)

TBH we could likely remove the color aliases at some point since we use ColorRGBA in most places

nice

10:21

i have sort of a niche need atm robyt

10:21

i am bounds checking /tp coordinates from my video up there but this requires getting the game layer & saving its boundaries

10:22

i don't want to do it every time the command is called but making it static would not be good because it would persist after a map change if it's only set once (edited)

10:22

should i check if the map changed inline or make it a member of CGameContext, and make it stick out like a sore thumb

Ewan

i am bounds checking /tp coordinates from my video up there but this requires getting the game layer & saving its boundaries

Doesn't seem very expensive to calculate the boundaries. Do you mean something like width+201?

yea

10:27

i suppose i will just do it every frame

CCollision keeps track of the game layer size, so if you wanted a cache then that would be the place, but just for width+201 it's not really worth it

10:27

I'd say profile first if it's an issue, otherwise it's premature optimization

k

10:27

Image attachment

10:28

this is all it is

10:28

i think it should be fine

@Ewan u can finally get dev role

nope

10:30

Image attachment

10:30

trolled again!

xd

these are issues from 2020

10:30

all of them complaints from heinrich

10:30

heinrich singlehandedly stopping progress. nice

And somehow we have two issues about the same thing

i really have not heard anyone but heinrich complain about that approach

If it generates lots of server messages then it's bad for the servers

10:32

The comment in #2221 explain a case in one map where it happens

https://github.com/ddnet/ddnet/pull/2221

Death tiles automatically /rescue when rescue is enabled or in prac...

Note I've already incorporated the conflicts from #2218, but that's not to say I incorporated #2218 in it's entirety.

what exactly is the issue rn?

Robyt3

The comment in #2221 explain a case in one map where it happens

can u eli5

kill will act like /r right? /r only works if u stand still somewhere, right?

yes exactly

then i dont see a problem with your design

the only place that you can get softlocked or repeatedly teleported without interfering has been removed in my pr

If you stand on a speeder and it moves you into a kill tile

then speeders shouldn't be safe

I only skimmed the comments

Robyt3

If you stand on a speeder and it moves you into a kill tile

but /r only works if u stand still

10:37

so this cant even happen

Then the comments still need to be "addressed" by extensive testing and explanation that this cannot happen

if not, then i'd claim that /r is bugged

10:38

yeah, the approach that heinrich wants is pretty overcomplicated tho. better try to make it easier by restricting it to our case

It could be possible to have a velocity of vec(0, 0) temporarily while inside a speeder

Robyt3

It could be possible to have a velocity of vec(0, 0) temporarily while inside a speeder

well then stand still and dont press any key

10:40

for 2 ticks

10:40

xdd

[ddnet/ddnet] Pull request opened: #7696 add x and y coords to /tp

Allows you to explicitly specify which coordinates to teleport to in the /tp and /teleport commands Syntax for the complete command is as follows /tp; teleport to the origin of your spectating perspective /tp arg'; teleport to this player /tp arg1 arg2; teleport to these coordinates (arg1 is x, arg2 is y) can also specify relative coordinates, like minecraft. these are all valid 10 (absolute 10.00) ~10 (current position + 10.00) ~+10 (current position + 10.00) `...

but pressing /r after died or not dieing, when you expect it defs sounds more unintuitive

GitHub

Click to see attachment 🖼️

@Ewan Can you avoid some of the unrelated style changes in your PR? Looks like you used clang-format-15 instead of 10 (edited)

Ewan

then speeders shouldn't be safe

it’s literally just like gravity

Robyt3

@Ewan Can you avoid some of the unrelated style changes in your PR? Looks like you used clang-format-15 instead of 10 (edited)

i’d like to but idk how to set it up with my ide

10:42

i will have to manually apply style changes inplace with the cli otherwise

You could simply not stage, commit and push those changes owo

owo

err we must not be talking about the same thing

10:43

anyway why does ddnet not use 15

10:43

quite annoying

Because the format looks trash

10:43

https://github.com/ddnet/ddnet/pull/7696/files#diff-29ffe32916052e1066eed938021aadfeb29f855d7c63d9c1ca3350aceffb79f3L1246-R1247

can’t you specify further

maybe we could invest some day in a definition about which tiles actively affect gameplay. would be intersting, e.g.:

air tiles
tele out
tune zones are all passive

Interfacing with clang-format is black magic

it’s literally just a file

10:44

and you write the rules you want

10:44

ez as pie

But I wish someone would do it of course

just ping deen, he fixes clang-format for you

He also didn't know how to get lines not to break

well sorry if you don’t agree but subjecting every contributor to use an old version of clang format when they do any more than 3 line changes doesn’t exactly seem logical

Ideally clang-format would either preserve our line lengths or at least not do something as stupid as the format I linked above

you can give it an explicit max line width

@Ewan u can take over #7034

https://github.com/ddnet/ddnet/pull/7034

Use clang-format 15 (fixes #7033) by def- · Pull Request #7034 · dd...

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 (especially base/) or add...

ehh

Ewan

well sorry if you don’t agree but subjecting every contributor to use an old version of clang format when they do any more than 3 line changes doesn’t exactly seem logical

The version and format should be upgrades yeah, but contributions shouldn't contain unrelated changes either

then u can use format 15 xd

should just cut your losses imo

10:47

impose clang format 15 changes on all ddnet source

10:47

without trying to preserve current style

I could agree with all previous clang-format decisions so far, but this is completely illogical formatting so I won't support it

right well it's definitely not hard to write your own'

10:47

rules

he have our own rules

10:48

we

10:48

sadly clang-format seems to have changed some things

it does not help that clang format 10 is not even in my pkg mgr

yep, but even clang-format15 should be pretty close. what even is your issue rn? does it create such a huge diff?

im finding so much stuff

10:49

prepare

#7034 only changed 300 lines

https://github.com/ddnet/ddnet/pull/7034

Use clang-format 15 (fixes #7033) by def- · Pull Request #7034 · dd...

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 (especially base/) or add...

mainly my problem is we are talking about whitespace here but now i have to build an old version of this tool from source because nobody took the time to modernize the CI

10:50

you could also consider having the CI in-place apply the proposed changes that make the build fail

The changes to comments I don't care about, but this is bad: https://github.com/ddnet/ddnet/pull/7034/files#diff-b3a45ccb488177dbd8a7327f6f67322eaf69a6f178a57680d9c6172f0a436b08L690-R692

yeah that's silly

10:50

hey ryo want another ddnet chore

10:51

ask in llvm server how to fix that bs

10:51

clang-format channel has seen enough of my face

ryo will tell you exactly one thing: use rust

10:51

and then i'll tell him, yeah but sometimes rustfmt also breaks, very annoying

o yea btw some help here pls https://github.com/ddnet/ddnet/pull/7696/files#diff-196a92816a2a9db961e4c9fe22ea5f2b66274242c6482955275ec818b5cb7dd1R1599

10:52

don't know what constant to use for these buffers

10:52

max console string length or whatever it's called seems absurd

10:52

not rly trying to allocate 8kb per arg

32 seems enough I guess

do u even need these tmp vars?

i suppose not

Doesn't look like it (edited)

changing real quick

Also GameLayer -> pGameLayer

o yea thx

Boundaries probably doesn't need to be static

10:54

Would also have the wrong name for that

wdym

10:54

oh yeha

10:54

s_

10:54

s

10:54

asjkbfkj

You are calculating the values every time, so you don't need static

10:55

One vec4 isn't very expensive to create/destroy either

my thinking was it helps to not allocate 4 floats again but i'm using aggregate init so w/e

Ewan

it does not help that clang format 10 is not even in my pkg mgr

It’s in none of ours either, you just install it once and it’s done

it's a real thorn in my side is all im saying

% clang-format --version                                            
Debian clang-format version 18.0.0 (++20231122111257+b3a9e8f7c0af-1~exp1~20231122111413.1710)

yea i was gonna say

10:56

surely debian has it

clang-format-15, what is that?

Ewan

it's a real thorn in my side is all im saying

there's not even a clang-format executable option in my ide so i really will have to manually inplace every comit

Ewan

mainly my problem is we are talking about whitespace here but now i have to build an old version of this tool from source because nobody took the time to modernize the CI

You don’t have to, this is a common hurdle so there are prebuilt versions of it. Everyone and their mothers uses this one https://github.com/muttleyxd/clang-tools-static-binaries

GitHub - muttleyxd/clang-tools-static-binaries: static-linked versi...

static-linked versions of clang tools for Linux, Mac OS X and Windows, available in AUR as clang-format-static-bin - GitHub - muttleyxd/clang-tools-static-binaries: static-linked versions of clang ...

oh cool

the vscode workspace also directly makes it easy to use if u add it to PATH

10:57

Anyway, yeah we should bump up to clang-format15 soon, if only we can figure out how to get rid of the ugly linebreaks

[ddnet/ddnet] Pull request opened: #7697 modernize some code

Some findings i found with clangd and clang tidy The explicitly defaulted function declarations enable more opportunities in optimization, because the compiler might treat explicitly defaulted functions as trivial.

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 (especially base/) or added coverage to integration test
[ ] Considered p...

owo

I usually just do python3 scripts/fix_style.py before committing

well theoretically we should simply use clang-format default values but i guess most ppl here would go on rampage

10:58

I DONT CARE aslong as i dont have to do anything manually

10:59

then i go on rampage

Jupstar ✪

well theoretically we should simply use clang-format default values but i guess most ppl here would go on rampage

It’s nasty. Idk what sort of brain worms do the people that make the style guide for clang format have

10:59

Just 0 sense of aesthetics

i write so much code inline and let the formater do it's job xD

ewan@machine /x/ddnet-practice-tp-coords (tp-coords)> clang-format-10 -i -style=file **/*.cpp **/*.h
exec: Failed to execute process '/usr/bin/clang-format-10': the total size of the argument list and exported variables (6kB) exceeds the operating system limit.

feelsbadman

if ... {bla} else {blam} whatever; ENTER

check my pr thanks

Learath2

I usually just do python3 scripts/fix_style.py before committing

i do ./scripts/fix_style

or CTRL+S

11:00

xd

Jupstar ✪

or CTRL+S

yes i use ur workspace

Ewan

ewan@machine /x/ddnet-practice-tp-coords (tp-coords)> clang-format-10 -i -style=file **/*.cpp **/*.h
exec: Failed to execute process '/usr/bin/clang-format-10': the total size of the argument list and exported variables (6kB) exceeds the operating system limit.

feelsbadman

Use the included script, it fixes the glob issue

yea it works

Ryozuki

yes i use ur workspace

based

swagular.

@Jupstar ✪ check my pr, theorically it enabled more optimization oportunities

11:01

default the world

it did Not Work

nice, 1ns better frame times

11:01

i take that

brb killing myself

Image attachment

also i found some interfaces that did not have a virtual destructor xd

Ewan

brb killing myself

thats good

yea well they dont like whitespace changes

Jupstar ✪

well theoretically we should simply use clang-format default values but i guess most ppl here would go on rampage

this is why rustfmt is best

11:02

u dont argue about style

11:02

u dont lose time

clang-format would benefit from some backcompat

11:04

but this sort of thing only happens after several versions of neglect

11:04

so i can't say i rly think it's a huge deal

clang-format also AFAIR profiles

LLVM profile

so u say i want "google-like style"

that's --style=google

11:05

or basedonstyle in your file

maybe one is close to ours

yea i always just use the llvm one in my own code

btw did anyone check if the dockerfile still works

nope

11:09

never used docker in my life

11:10

can't say i regret it

11:12

lmfao

11:13

hmmmm

11:13

this is a tricky thing to test here

11:13

trying to figure out exactly what would be considered out of bounds with the kill tile's hitbox

11:14

should i hardcode something or just round it to the next whole tile

You want to make sure the tee stays in bounds, right? Would be the easiest to make the checks a bit stricter so you don't need a lot of additional logic.

i just added 1 real coordinate (not 'display' coord)

11:16

works fine

11:16

though for some reason my bounds checking is screwing me over again

11:16

i didn't test this super thoroughly owo

owo

Can someone help me with a resolution problem I have with my ddnet?

wrong channel

11:28

#questions

oh sry

i forgot to transpose the game layer tile int coords to float coords

11:35

*32

the pr is winning now

11:48

i think adding some sort of 'limbo' state after a tp or rescue would satisfy both heinrich's and devinci's issues

11:48

where the tee is there but it's just not moving

11:48

and any movement input would make it move again

11:50

style check is failing after using the script so i really don't know

touching physics :justatest:

yea im not thrilled either

11:51

it'd need to interact with other entities so it'd have to put the whole team in limbo

justatest

justatest

justatest

maybe it's not the move

Ewan

style check is failing after using the script so i really don't know

It's a different style check that failed now

11:51

src/game/server/ddracechat.cpp:#include "game/mapitems.h"

11:52

Should use <> instead

oh that's clangd's fault

11:52

bleh

11:52

being able to dry-run the build-unrelated ci checks would be nice

i just added a method to IConsole & CConsole and it's undefined reference to vtable for IConsole

12:06

you don't use anything like IDL do you justatest

justatest

@Ewan you can dry run the ci? its a bunch of py scripts just run those

12:43

i have a shell script called ddlint that basically runs all of the CI stuff at once locally

12:43

https://github.com/ChillerDragon/crools/blob/5989f3e647ea1c7e9acfdbd78c0d5f3ccf30cbdc/ddlint

crools/ddlint at 5989f3e647ea1c7e9acfdbd78c0d5f3ccf30cbdc · Chiller...

Crap tools. Don't even bother looking. Contribute to ChillerDragon/crools development by creating an account on GitHub.

rad

12:53

when upstream

ChillerDragon: what is ctidy

13:06

clang-tidy? why do u call it ctidy

https://github.com/ChillerDragon/crools/blob/5989f3e647ea1c7e9acfdbd78c0d5f3ccf30cbdc/ctidy

crools/ctidy at 5989f3e647ea1c7e9acfdbd78c0d5f3ccf30cbdc · ChillerD...

Crap tools. Don't even bother looking. Contribute to ChillerDragon/crools development by creating an account on GitHub.

13:29

ah ye its another script

justatest

yea not sure if i can recommend my infrastructure for it

13:29

just wanted to show you that you can basically copy paste all the commands from the github actions into one script and then run it locally just fine

i might just get one of those local gha runners

omg they the worst

probably low overhead

fakin microsoft

13:30

just run native

13:30

its like 4 commands

x

Jupstar ✪

Could you explain why it was done like that? I mean i understand that this makes sense, in a way that it doesn't break any logic. But it's kinda weird xd

basically, since you can have something like

let param = match x.parse() {
    Ok(p) => p,
    Err(e) => {
        eprintln!("Error in param: {e}");
        continue;
    }
}

heinrich why u think it should be a separate command

you can model continue as ! which can be converted into any arbitrary type, thus the two branches of the match can return the same type

Ewan

heinrich why u think it should be a separate command

no perf gain at all + it was already multifunctional + more work to make it behave differently

Ryozuki

https://blog.rust-lang.org/2023/12/21/async-fn-rpit-in-traits.html

seems to be still mostly useless, as explained in the blog post :/

Ewan

i might just get one of those local gha runners

It's extremely finnicky to get working and it's slow. Can't recommend unless you are a container veteran and have a beastly pc

damn

13:41

https://github.com/ddnet/ddnet/pull/7691#pullrequestreview-1795439527 (edited)

13:41

i don’t understand what he means

13:41

tbh Heinrich i have seen 2 other ppl implement this on their own and many more expressing their desire for it

13:42

i don’t think your fundamental disagreement but unwillingness to actually code it yourself should stand in the way of what many players want

13:42

my implementation is brief, non-intrusive, works and fills the need that most ppl have

[ddnet/ddnet] Pull request opened: #7698 Use CEditorComponent methods instead of only CMapView

Also, small changes to CEditorObject. Previously only CMapView methods were used, but we should instead use the components methods by iterating over them in case we add more editor components.

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 (especially base/) or added coverage to integration test
[ ] Considered possible null pointer...

Bota

how i do that. I dont know rust justatest

justatest

I'll try to see where the error comes from. please report an issue on github

Ryozuki

Click to see attachment 🖼️

do we use it in threads?

Learath2

NaN should not be possible

strtod e.g. returns NaN if you input "nan" with any case

heinrich5991

strtod e.g. returns NaN if you input "nan" with any case

Oh yes, I totally forgot you could explicitly ask for a nan while focusing on the error states

Ewan

it does not help that clang format 10 is not even in my pkg mgr

you can keep it unformatted and ask someone else to format it

i dealt with it

[ddnet/ddnet] Issue opened: #7699 Forcing mods to be open-source to be included as a community

What would you think about fostering the open-source spirit of DDNet/Teeworlds and requiring all future communities that are included to be open-source? The KoG case is special IMO and should be grandfathered in for now, while also nudging them to be open-source in the future.

of course fokkonaut instantly responds

Cellegen

so it would just reset (as in if its 64 or 128) hitting limit (edited)

no, you would get stuck either at some big value if addition always rounds down to the same number, or you'll get stuck at infinity which is a valid value for floating points

Learath2

HUGE_VAL + d == HUGE_VAL forall d > 0 iirc

that sounds like infinity

Chairn

that sounds like infinity

yep, HUGE_VAL is defined as infinity on my machine

GitHub

Click to see attachment 🖼️

tricky

GitHub

Click to see attachment 🖼️

@Ryozuki

14:13

gog sorce

cyberFighter

of course fokkonaut instantly responds

Its stupid

it could be a good idea if all popular game modes but ddrace weren’t closed

noby will not have a community, bombay wont be official, blockZ cant be hosted on ddnet, and much more

fng as well

fokkonaut

Its stupid

u have instant alarm + notification when an issue has the words "mods" and "open-source"?

oh

14:15

i misread

14:15

xdd

then 1 year of "special case" handling for kog

yea that’s straight up ridiculous

I'm pretty sure we can get noby to open his sources. the only reason not to do it was laziness so far, AFAIK

all of this makes it very stupid to really force this

what’s the gain anyway

Pushing tw open sourceness

Ewan

what’s the gain anyway

longevity of the game

if ppl want to keep ripping each other off in their shitty block clients then let them

pushing open-sourceness in an open-source game

one must imagine sisyphus happy

14:16

block srvs i mean xd

@heinrich5991 even if it's lazyness from noby or whatever, i think with this approach communities will kinda lose their purpose (indicating official servers etc)

tbh it’s not necessarily against the spirit of teeworlds to have a closed game. it would be non permissive otherwise

as a closed src mod cant be a community and therefore is on the black market of servers

Something broke the rcon console:

Image attachment

well the point is to make you want to go oss

murpi

Something broke the rcon console:

LOL

Ewan

well the point is to make you want to go oss

I understood that

you

14:19

anyway Heinrich can u respond to smth

I just dont think it should be enforced

anythgng rly

14:19

waiting 2 sleep

new incremental mod added to ddnet where you gain lines in the rcon

I'm a fan of having discussion on github, discussions here will be forgotten

True

@fokkonaut has been the only one to actually comment on the issue I opened

well I responded to both of ur reviews

cyberFighter

u have instant alarm + notification when an issue has the words "mods" and "open-source"?

yes

Ewan

anyway Heinrich can u respond to smth

in both of the PRs I recently commented on, there's a comment by me that is more recent than a comment of yours, I think

watching ddnet repo would destroy my inbox

heinrich5991

in both of the PRs I recently commented on, there's a comment by me that is more recent than a comment of yours, I think

i blame this on the gh mobile app

14:21

i get no notifications and i had to manually refresh to see it

murpi

Something broke the rcon console:

Is that with nightly? Doesn't happen to me there. Which operating system?

Robyt3

Is that with nightly? Doesn't happen to me there. Which operating system?

Yes, nightly. I can't reproduce it. :/ Only restarting the client helped

14:28

Win11 64bit

Maybe something not initialized correctly

idk what im looking at anyway

murpi

Yes, nightly. I can't reproduce it. :/ Only restarting the client helped

Was this after joining a server or did you do something else like logout or clear the console?

murpi

Yes, nightly. I can't reproduce it. :/ Only restarting the client helped

Clearing the console seems to fix that, don't know what causes it though

Robyt3

Was this after joining a server or did you do something else like logout or clear the console?

I connected a dummy, that's all. (wasn't authed on my primary tee)

14:31

I never clear my console, but I did logout via bind

14:32

I used bind kp_1 "rcon_login <name> <accesskey>; rcon show_ips 1; rcon status; rcon dump_antibot; rcon logout" to be more precise

discuss with me fellas

[ddnet/ddnet] Issue opened: #7700 Remote console sometimes keeps scrolling up

Reported by @murpii: !DDNet_UjLn7tA8OX Happened when using bind to login and logout of rcon while on dummy but not authed in rcon with main tee. Doesn't seem easy to reproduce. Maybe a variable is not initialized properly. Clearing the console fixes it according to Skeith.

ayo once i said we need https://mclo.gs/ for ddnet

mclo.gs - Paste, share & analyse your Minecraft logs

Easily paste your Minecraft logs to share and analyse them.

15:01

i started with syntax highlight and probably will rq there but its kinda public and ready to use already

15:01

https://logs.zillyhuhn.com/?url=https://paste.zillyhuhn.com/jT

15:01

https://logs.zillyhuhn.com/?url=https://paste.zillyhuhn.com/dE

xD

15:04

such a chillerdragon feature

test

15:21

what the hell

15:21

Test

15:21

this is what i was trying to send

Image attachment

15:21

discord's not letting me for some reason

15:22

i super over-engineered it the first time

15:22

i liked the logic i thought it was clever

15:27

maybe one of my variable names is on some anti-injection blacklist

GitHub

Click to see attachment 🖼️

kog should definitly not be a exception and removed

removed from wat wowo

a

15:34

wowo

wowo

hot discussion foss

chuller

Uwun

true

i open my pc and put the same on gh in 5 mins

Wdym

so its not lost

15:35

i always said we shouldnt give visibility to closed source

15:36

we also shouldnt offer kog help with discord commands

15:36

a bot should silently delete the msg and send a private dm

15:36

its rly annoying

15:36

i ofc would change my mind if kog becomes fully open source, web and all

15:37

btw unique deserves more to be in the community list

1

15:37

we are fully open source, web too

open source btw

no pfp, fock feelsbadman

feelsbadman

Ryozuki

we are fully open source, web too

very true

15:38

tim has not been responding to my msgs

he is sadly not much into tw lately

15:39

life

usa servers db is not properly set up

15:39

idk since when

ddnet is not open source

but i turned race servers off about a month ago

it links spyware blobs

Ewan

but i turned race servers off about a month ago

still waiting for him to fix

ChillerDragon

ddnet is not open source

ddnet is open-source

15:40

check https://github.com/ddnet/ddnet

GitHub - ddnet/ddnet: DDraceNetwork, a free cooperative platformer ...

DDraceNetwork, a free cooperative platformer game. Contribute to ddnet/ddnet development by creating an account on GitHub.

https://github.com/ddnet/ddnet/blob/88ee1566463e9e377d34600b47fd4b6be28ca57a/CMakeLists.txt#L3148

15:40

where is the code for this dylib

15:40

https://github.com/ddnet/ddnet/blob/88ee1566463e9e377d34600b47fd4b6be28ca57a/CMakeLists.txt#L3156

that is not discord code

where is the code

is any open source code which relies on Microsoft’s CRT closed?

all of DDNet code is open source. you can also (and by default) build without linking to closed source code

https://www.gnu.org/philosophy/pragmatic.en.html

15:41

The GNU GPL is not Mr. Nice Guy. It says no to some of the things that people sometimes want to do. There are users who say that this is a bad thing—that the GPL “excludes” some proprietary software developers who “need to be brought into the free software community.” But we are not excluding them from our community; they are choosing not to enter. Their decision to make software proprietary is a decision to stay out of our community. Being in our community means joining in cooperation with us; we cannot “bring them into our community” if they don't want to join.

on linux, at least

ddnet.org ships code that is not public doesnt it?

15:41

or ships binaries

15:41

compiled by valve and discord

shut up

ddnet.org and steam ship binaries that are not open-source, yes

i dont like that

15:42

wouldnt call that foss

feel free to be this specific

15:42

but calling ddnet non-open source seems wrong to me

i am specific

no:

ddnet is not open source

out of context xd

lot of projects deliver the compiled software via a paywall but they host the full source code on a public site

15:43

that is still open and free software

xd

irc newlines be like

15:43

i followed up with why its not foss 1s after

tw should have used AGPL

pls no

15:44

exclusive licenses are evil

15:44

excluding*

but that is incorrect. ddnet.org/steam builds of ddnet are not open source

15:44

ddnet itself is very much open source

GPL is the most user respecting license

except the scripts repo :p

Ryozuki

we also shouldnt offer kog help with discord commands

What's this referring to?

and anitbot axaxax

murpi

What's this referring to?

$kog

KoG registration and login

First and foremost: DDNet and KoG aren't affiliated. You are not required to log-in on a DDNet server. If you need help on a server related to KoG, join their Discord server by clicking on the link below.

URL:

https://discord.kog.tw/

Registration process:

https://discord.com/channels/342003344476471296/941355528242749440/1129043200527569018

Migration process:

https://discord.com/channels/342003344476471296/941355528242749440/1129043332211945492

How to login on KoG servers:

https://discord.com/channels/342003344476471296/941355528242749440/1129043447517564978

Video Guide:

https://www.youtube.com/watch?v=d1kbt-srlac

and the other

15:44

which spams a wall

15:44

dont promote this

Ryozuki

GPL is the most user respecting license

no, it just prevents specific ppl to use it

but i mean it is embracing closed source software by using discord and steam api @heinrich5991

send a private dm and delete the message

and forcing that onto all users that download ddnet from official channels

Jupstar ✪

no, it just prevents specific ppl to use it

no, it ensures users have the freedoms everyone should have

15:45

fully owning a software u use

that downlad ddnet binaries from ddnet.org

yes

Ryozuki

fully owning a software u use

that is the case with other OS licenses too

15:45

one could argue, even more

Ryozuki

which spams a wall

implement it in a sane way, I think it'd be accepted

no because there will be devs using propietary licenses

I see it as hard to do though

this is why i link this article https://www.gnu.org/philosophy/pragmatic.en.html

15:46

u have to be pragmatic in this world

then remove license all together

15:46

that is pragmatic

15:46

you are just playing the game without breaking any rules

that requires us to overthro goverments and copyright laws which is impossible

15:46

u are not being realistic

Ryozuki

and the other

The 'other' no longer exists

while this license is

Ryozuki

that requires us to overthro goverments and copyright laws which is impossible

but it's governments and copyright laws that also have to protect your license in order to work

Jupstar ✪

but it's governments and copyright laws that also have to protect your license in order to work

yeah u cant change the game, u play the game smart

so you are stabilizing the rights of companies

its what copyleft does

15:47

tf are u saying

well your opinion

yes its mine

15:47

i use agpl everywhere now

good for you

15:47

sadly zwelf uses it too

good

sad

zwelf understands

15:48

he is also pretty based

15:48

he required the wiki to use a plugin which doesnt directly load the yt videos

15:48

for privacy reasons

15:48

i admire him

i think his ddnet physics have the potential to be used for a next gen ddnet

15:48

but with AGPL it doesnt have my support at least

15:49

that would be a step backwards compared to the cpp project

no

15:49

with agpl its even better

15:49

i hope he makes it rly good

15:50

this is how gpl thrives

15:50

why gcc had so much impact

ryo we trade reviews?

and now clang is the "better" compiler

Consider GNU C++. Why do we have a free C++ compiler? Only because the GNU GPL said it had to be free. GNU C++ was developed by an industry consortium, MCC, starting from the GNU C compiler. MCC normally makes its work as proprietary as can be. But they made the C++ front end free software, because the GNU GPL said that was the only way they could release it. The C++ front end included many new files, but since they were meant to be linked with GCC, the GPL did apply to them. The benefit to our community is evident.

or llvm generally

sadly llvm has made things like NVVM be born

15:50

its toxic af

only thing i still use gcc is to get most perf

i dislike llvm license the most

15:51

Consider GNU Objective C. NeXT initially wanted to make this front end proprietary; they proposed to release it as .o files, and let users link them with the rest of GCC, thinking this might be a way around the GPL's requirements. But our lawyer said that this would not evade the requirements, that it was not allowed. And so they made the Objective C front end free software.

nu obj c wat

15:52

xxxxxx'

https://en.wikipedia.org/wiki/Public_domain#/media/File:Cc-public_domain_mark.svg

The public domain (PD) consists of all the creative work to which no exclusive intellectual property rights apply. Those rights may have expired, been forfeited, expressly waived, or may be inapplicable. Because no one holds the exclusive rights, anyone can legally use or reference those works without permission. As examples, the works of Willia...

15:53

the best logo

15:53

Image attachment

15:53

agpl logo is cooler xd

15:53

ill add this logo in my web

add vulkan logo

lol

bcs vulkan API is royalty-free and fast af

15:58

that's how true kings live

15:58

u all so religious

@heinrich5991 pls dont ask me to remove comments

i hope to one day be as passionate about something as you are about rust and vulkan

Ryozuki

@heinrich5991 pls dont ask me to remove comments

on wiki?

15:59

or wat

its totally legit to say some private mods want monetary gain

Ewan

i hope to one day be as passionate about something as you are about rust and vulkan

if you are not religious about cpp, why don't you try rust?

Ewan

or wat

https://github.com/ddnet/ddnet/issues/7699

Forcing mods to be open-source to be included as a community · Issu...

What would you think about fostering the open-source spirit of DDNet/Teeworlds and requiring all future communities that are included to be open-source? The KoG case is special IMO and should be gr...

i did try rust

for over 1 week xd

no

15:59

too much brain power involved

Ryozuki

its totally legit to say some private mods want monetary gain

the "you" sounded to me like addressing Kaffeine

@Ryozuki i am suprised you with your agpl user rights whatever idiology are not bothered by discord

Jupstar ✪

for over 1 week xd

cpp flows thru my fingers naturally

and now you're telling him that he's doing it for possible monetary gain

Ewan

cpp flows thru my fingers naturally

and that's telling others what their intentions are

16:00

which is not nice

ok my you was like in 3rd person

16:00

as in the person doing a privative mod

can you replace it with "one"?

16:00

then it's clear

dude

ur so pedantic sometimes

why do u want to control what he saysa

16:00

it does not even reflect on u

16:01

let it go

because I want the discussion to go somewhere. insults or things perceived to be insults help make the discussion go nowhere

Ewan

cpp flows thru my fingers naturally

i've to admit, i much more rare need to open the rust "spec" (or google about rust behavior) than cpp behavior

16:01

even tho i code in cpp for 10 years

[ddnet/ddnet] Issue opened: #7701 Handling of maps with unknown UUID map items

This change silently corrupts maps with unknown future items instead of crashing. That seems like a downgrade. IMO, it should either show an error or try to re-add the future map items as well.

Originally posted by @heinrich5991 in https://github.com/ddnet/ddnet/issues/7693#issuecomment-1868297260 The same also applies to the editor, which was already doing what the tools are also doing now.

code in my brain is cpp

ChillerDragon: its true im sad i have to use discord, but i dont want to be invisible socially and i need to talk to friends i cant be like stalkman, but yeah it may be idiotic but i still want agpl

i see

16:02

thats fair

sad that drug dealers have ruined name of all the good open chat softwares

like which

Ryozuki

ChillerDragon: its true im sad i have to use discord, but i dont want to be invisible socially and i need to talk to friends i cant be like stalkman, but yeah it may be idiotic but i still want agpl

the EU might help with that

i am yet to discover good chat software @Ewan wat would u consider good?

heinrich5991

the EU might help with that

how so

Ewan

sad that drug dealers have ruined name of all the good open chat softwares

that doesn't sound like drug dealers, but rather someone trying to use drug dealers to fight against open chat software

telegram, signal

they suck ass

telegram is goated

telegram and signal are shit af

i know my views are probably impossible but its nice to have some ideals

telegram is closed src

16:03

signal is literal spyware

Ryozuki

how so

not discord specifically, but the dma forces some messengers to interoperate

yea

bound to your phonenumber

telegram is open-source

everything?

open client

open client

isnt serrvers closed

yea

Ewan

sad that drug dealers have ruined name of all the good open chat softwares

well vulkan is bad because it doesn't enforce me to write correct code, BUT it , IMO, makes 10000 more sense than OpenGL. It's much more clear to me what happens under the hood with it.

@heinrich5991 i think google play store doesnt allow GPL apps

thats not open source then

idk if they allow MIT

binding a chat protocol to a phone number is fucked

16:04

and should be illegal

Ryozuki

@heinrich5991 i think google play store doesnt allow GPL apps

apple does not, not sure about google

@Ewan join the vulkan force today not tomorrow

I think they're working on it, ChillerDragon

idek opengl

signal?

yes

thats amazing i didnt know that

16:04

holy

16:04

there might be hope

i think phone numbers should be more abundant so they not used anymore to identify someone

i only do 2d greaphics usuauaully

or just change phone number system

Ewan

idek opengl

16:04

vk has cool logos

i know

XD

it's awesome

[ddnet/ddnet] Pull request closed: #7698 Use CEditorComponent methods instead of only CMapView

16:05

github-merge-queue[bot]

[ddnet:master] 2 new commits

5599de3 Use CEditorComponent methods instead of only CMapView - archimede67 8f3d09e Merge pull request #7698 from archimede67/use-editor-components - Robyt3

good argument

16:05

rust has only ugly logos

16:05

vulkan has cool logos

golang has worst logos

rust has some cute logos

16:05

ok not logos

binding contacts/chats/2fa to some weird small device and a contract is such an unstable concept

who is this asshole

Image attachment

i want to flush him down a toilet

Ryozuki

rust has some cute logos

but not the crap crab

xd

Jupstar ✪

vulkan has cool logos

does vulkan have waifus?

Image attachment

@Jupstar ✪ u know there is a unsafe rust pet

lmao @Ewan xd

meet corro, the unsafe rust pet

Image attachment

16:06

Image attachment

kek

Embedded image

16:06

nimrocks

Embedded image

16:06

how did he create a black version

16:06

looks even cooler

Jupstar ✪

Click to see attachment 🖼️

i like more the one with text

16:07

"forging ahead"

16:07

or smth

16:07

Image attachment

i like the other more

@Jupstar ✪

Image attachment

when vulkan 1.4

vulkan is already too bloated

we need new API

@Jupstar ✪ but thanks to vulkan extensions u can easily add features right

Ryozuki

@Jupstar ✪ but thanks to vulkan extensions u can easily add features right

yeah, and validation layers etc. but they add like 3 new extensions per month xD

stonks

16:11

@heinrich5991 @Jupstar ✪ i think a override on a virtual with default is still virtual right?

Ryozuki

@heinrich5991 @Jupstar ✪ i think a override on a virtual with default is still virtual right?

from my understanding yes

wdym

and default also still implements the destructor chain

how would you mark it non-virtual then?

i made this changes with clangd suggestions

heinrich5991

how would you mark it non-virtual then?

override only works on virtual functions

Ryozuki

i made this changes with clangd suggestions

just remove the const stuff

Jupstar ✪

override only works on virtual functions

is there a difference between marking child class functions as virtual or not (e.g. if they're further derived)?

i agree with heinrich, even tho i'd argument from different perspective xd

override also makes it a error if the overriden signature doesnt match, while virtual doesnt and ucan get nifty bugs (edited)

16:12

override is safer and better

Ryozuki

override also makes it a error if the overriden signature doesnt match, while virtual doesnt and ucan get nifty bugs (edited)

I think override and virtual do different things

you can't finalize a virtual function

heinrich5991

is there a difference between marking child class functions as virtual or not (e.g. if they're further derived)?

no, u can always override, until u used final

not having anything to do with each other (edited)

yeah as long as its not marked final

16:13

u can override i think

override simply overrides any virtual fn, already implemented or not

In a member function declaration or definition, override specifier ensures that the function is virtual and is overriding a virtual function from a base class. The program is ill-formed (a compile-time error is generated) if this is not true. override is an identifier with a special meaning when used after member function declarators: it's not a reserved keyword otherwise.

16:13

see

16:13

it ensures

yes

heinrich5991

not having anything to do with each other (edited)

you are overriding the vptr what else you mean?

ensures that the function is virtual and is overriding a virtual function from a base class

ah, it makes it virtual

16:14

perfect

i also agree with jupstar gh comment

16:15

we should use refs wherever possible

16:15

so const matters

16:15

xd

well heinrich will say that it's less clear as func arguments

16:16

but anyway, let's not trigger this discussion again xD

its cuz cpp doesnt require u to do & right

16:16

xd

yes

also does cpp optimize refs better

16:16

?

16:16

since it knows its not null

16:17

we should also do noalias

dunno, i guess it's about the compiler

rust in cpp

yeah, if cpp would have move by default, it would be less of a pain

@heinrich5991 > There are advantages to keeping your code private, like discovering vulnerabilities but keeping them under wraps for personal benefit >> "discovering vulnerabilities but keeping them under wraps for personal benefit" does not sound like a benefit to the community at large at all. Sorry, I edited it my comment too late :| Edit: I changed it again so it makes more sense xD (edited)

@Ryozuki when do you start working for discord and add @ aliases so i can ping u with @rust

16:23

chiller once had it, sad it was removed

16:23

xd

Jupstar ✪

yeah, if cpp would have move by default, it would be less of a pain

No

Ewan

No

well that would at least eliminate the problem with const ref vs implicit copy

16:24

and many times you want a const ref as function parameter

16:24

but anyway this alone wont work anyway

16:25

there's a reason why rust kinda feels like the opposite syntax wise than cpp xD

when u move u deal with lifetimes

stuff u need to type out in rust, u dont need to do in cpp and vise versa (edited)

Ewan

when u move u deal with lifetimes

just avoid it if u want to stay sane

16:25

you do not need those 4 bytes

16:25

const ref is the way tho

Ewan

just avoid it if u want to stay sane

how do you want to approach insane gores, if u want to stay sane xd

i hate gores

ok ddnet calls it insane too

16:27

when nightmare category

insane is enough tbh

16:28

tho 50pts for a 5* is not nearly enough

we could also add a "free" category and shift stuff

all those pts are the difference between 20k and 30k players nowadays

16:28

the gap should be higher

so novice becomes free and moderate novice and hard moderate and insane hard and insane is free for harder stuff

idk

16:28

categories are fine as is

xd ofc they are

gonna make some eggs

egg

novice moderate brutal insane nightmare godlike supreme king SSGSSJ2 Ultra instinct

16:30

my unbiased list

uninstall

stronghold 4 release in half an hour! (edited)

16:32

everyone in?

atomic

heinrich5991

stronghold 4 release in half an hour! (edited)

ohh

thread-safe

lets do towers

i am so excited

heinrich5991

everyone in?

are u srs

16:34

i am actually in

im in

16:35

ger10

16:35

stormwind

Who will post the first stronghold 4 tower justatest

justatest

lets do a dev team

16:35

16:35

we wont pass

fine by me

16:37

u want to voice?

heinrich5991

yep, HUGE_VAL is defined as infinity on my machine

On all implementations that support infinities iirc

learath are u in

Learath2

On all implementations that support infinities iirc

that seems really irrelevant. all machines I know of support IEEE754

16:39

I'm sure we rely on IEEE754 somewhere more than "supporting infinities" somewhere

where's my team

Ryozuki

stormwind

dont see it

ger

im on 45.141.57.22:8336

45.141.57.22:8336 is an official DDNet server. https://ddnet.org/connect-to/?addr=45.141.57.22:8336/

Ewan

learath are u in

Am outside

come on in

whats going on vc

17:09

@cyberFighter yo

17:10

ill do a little trolling

hi

got trolled by cf himself

whats so special about stronghold 4

numbero 4 is kinda exciting

the final

as if we needed the fourth one

17:12

troll

gerdoe

whats going on vc

playing stronghold 4

@Ewan

17:52

ingame (edited)

what’s up

17:52

I have been waiting on my brother to exit his friend’s house for like 15 minutes

[ddnet/ddnet] Pull request opened: #7702 Refactor graphics SetColor functions

Reduce duplicate code. Replace clampf function with NormalizeColorComponent function that convert color component from float to unsigned char. Use size_t instead of int.

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 (especially base/) or added coverage to integration test
[ ] Considered possible null pointers and out ...

@Robyt3 come voice

does windows 11 have different emojis tf?

Image attachment

win10 has black outlined so seems yep

[ddnet/ddnet] Pull request opened: #7703 change cast of pointer to uintptr_t

1 liner cast from pointer to integer of different size

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 (especially base/) or added coverage to integration test
[ ] 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...

@Robyt3 come voice

in which update ninja was removed to frozen tee?

u mean the katana?

ye

-StormAx

in which update ninja was removed to frozen tee?

are you gonna fork it

ima gonna add oldshcool mode (edited)

so, does someone know?

-StormAx

in which update ninja was removed to frozen tee?

16.2.2? I think when the HUD was added?

@Cøke ingame

Ya, just checked, it was 16.2.2

Skeith

Ya, just checked, it was 16.2.2

yep

19:39

thanks

No problem

was fun but missed @Learath2

1

17:35:53 bridge | <ryozuki> we wont pass

yea sma

20:33

same

gg

gg

was funny run, happy end disney couldn't have written this story better (edited)

20:34

"funny run"

20:34

20:34

insider

xd

gg

the last sweaty moments

3

20:42

the hammer was so close to freeze xD

[ddnet/ddnet] Pull request closed: #7703 change cast of pointer to uintptr_t

justatest

1

20:58

[ddnet/ddnet] Pull request reopened: #7703 change cast of pointer to uintptr_t

justatest

1

enjoying your holiday ryo

thanks

@Ryozuki first time i heard you

normal

21:11

i dont talk much

i usually also don't talk much, but today dunno

same about u

i fell out of my role

xD

21:12

u showed ur gores skill

21:12

its so easy to notice

21:12

the only gores part

21:12

gj

xdd insane gores skills

xdd

21:12

i suck hard at them

21:12

they filler parts in ddnet

#include <stdio.h>
main(t,_,a)
char
*
a;
{
        return!

0<t?
t<3?

main(-79,-13,a+
main(-87,1-_,
main(-86, 0, a+1 )


+a)):

1,
t<_?
main(t+1, _, a )
:3,

main ( -94, -27+t, a )
&&t == 2 ?_
<13 ?

main ( 2, _+1, "%s %d %d\n" )

:9:16:
t<0?
t<-72?
main( _, t,
"@n'+,#'/*{}w+/w#cdnr/+,{}r/*de}+,/*{*+,/w{%+,/w#q#n+,/#{l,+,/n{n+,/+#n+,/#;\
#q#n+,/+k#;*+,/'r :'d*'3,}{w+K w'K:'+}e#';dq#'l q#'+d'K#!/+k#;\
q#'r}eKK#}w'r}eKK{nl]'/#;#q#n'){)#}w'){){nl]'/+#n';d}rw' i;# ){nl]!/n{n#'; \
r{#w'r nc{nl]'/#{l,+'K {rw' iK{;[{nl]'/w#q#\
\
n'wk nw' iwk{KK{nl]!/w{%'l##w#' i; :{nl]'/*{q#'ld;r'}{nlwb!/*de}'c ;;\
{nl'-{}rw]'/+,}##'*}#nc,',#nw]'/+kd'+e}+;\
#'rdq#w! nr'/ ') }+}{rl#'{n' ')# }'+}##(!!/")
:
t<-50?
_==*a ?
putchar(31[a]):

main(-65,_,a+1)
:
main((*a == '/') + t, _, a + 1 )
:

0&#60;t?

main ( 2, 2 , "%s")
:*a=='/'||

main(0,

main(-61,*a, "!ek;dc i@bK'(q)-[w]*%n+r3#l,{}:\nuwloca-O;m
.vpbks,fxntdCeghiry")

,a+1);}

21:20

merry C christmas

21:20

(try to run it)

smells like 1. Xmas.c, winner of the 1988 International Obfuscated C Code Contest (udel.edu) 279 points by evah 3 hours ago | flag | hide | 40 comments

indeed

21:22

u got some nice memory

21:22

or google

21:22

xd

let's test gpt xd

21:23

gpt 3 too weak

21:23

sadly no gpt-4 xd

xd

[ddnet/ddnet] Issue closed: #7644 ESC Bug (+scroll bug, password, editor,)

/_/\ ( o.o ) >^,^< meow!

i'm getting worse in cpp #7697 @Ryozuki ur fault

https://github.com/ddnet/ddnet/pull/7697

modernize some code by edg-l · Pull Request #7697 · ddnet/ddnet

Some findings i found with clangd and clang tidy The explicitly defaulted function declarations enable more opportunities in optimization, because the compiler might treat explicitly defaulted func...

too much rust

xDDD

it's funny how much stuff you can forget. i dunno if i remember anything useful from university anymore xD

anyway

21:46

c++ requiring so much complexity

21:47

is a plus for rust

21:47

in rust u dont deal with virtual destructor shenanigans

yeah cpp is very complicated indeed u only notice it if u havent used it for a bit

21:48

but i think it's still important to know that rust kinda still has vptr, just not stored into the object, but into the dyn ...

but dinamic dispatch is rarer in rust

but u kinda never have to think about it

can we have a dev channel that simply links every single message that mentions rust?

and idk it feels much simpler

21:49

in rust this shenanigans are actually called

21:49

"object safety"

21:49

trait is object safe if it has the following qualities (defined in RFC 255): All supertraits must also be object safe. Sized must not be a supertrait. In other words, it must not require Self: Sized .

21:49

Image attachment

BlaiZephyr | meloƞ

can we have a dev channel that simply links every single message that mentions rust?

you want to have a dev channel that copies the exact same content as this channel? rust

https://doc.rust-lang.org/reference/items/traits.html#object-safety

Traits - The Rust Reference

Jupstar ✪

you want to have a dev channel that copies the exact same content as this channel? rust

fair enough

Ryozuki

in rust this shenanigans are actually called

not only that. i'd say it even makes more sense like that. u can have POD data structures and still implement traits on it

21:52

this is huge struggle with cpp xd

21:52

u can ofc create a POD and put is as member field into your inherited class

21:52

but not enforced = u'll forget it the other day

yea the necessity for wrapper classes in cpp is a huge turnoff

22:12

it has relatively limited metaprogramming capabilities

merry christmas

github-merge-queue[bot]

[ddnet:master] 2 new commits

c31f82a Refactor graphics SetColor functions - Robyt3 47ccdd0 Merge pull request #7702 from Robyt3/Graphics-SetColor-Refactoring - def-

23:39

[ddnet/ddnet] Pull request closed: #7702 Refactor graphics SetColor functions

23:41

[ddnet/ddnet] Issue opened: #7704 displaying cl_nameplates_strong in locked spectator view

when having cl_nameplates_strong set to 1 and locking spectate on a person, i`d personally like to see the nameplate for the person i am spectating in terms of: for who does the spectating tee has strong/weak.

@Scrumplex again new launcher?

23:42

How many rebrands and forks does it need watafak

what new launcher

https://youtu.be/FyS8ELbafN0?si=qKDdtkubgBebly2S

Brodie Robertson

FOSS Minecraft Launcher Destroyed By Rogue Founder

23:57

Our boi scrumplex is being called by name at 6:28

23:58

One of our more famous contributors hehe being GitHub showcased in a 25k clicked video o.0

Exported 1,083 message(s)