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 2023-12-17 00:00:00Z and 2023-12-18 00:00:00Z

yeah there's no reason to add vector in it

00:01

its just 3 const char*

github-merge-queue[bot]

[ddnet:master] 2 new commits

67c101b Add /join [player]. - furo321 4429347 Merge pull request #7662 from furo321/more-team-commands - def-

00:02

[ddnet/ddnet] Pull request closed: #7662 Add /join [player]

00:02

[ddnet/ddnet] Issue closed: #3878 Add /team first ! To create teams faster.

#4100 was a mistake xddd

https://github.com/ddnet/ddnet/pull/4100

Cleaned AFK checking by gerdoe-jr · Pull Request #4100 · ddnet/ddnet

Using bool instead of int for m_SendNAfkMsg Deleted m_pAfkMsg from CPlayer declaration Cleaned last activity check code Please test this PR, because I don't know what can go wrong here :P Check...

What is a pod

https://stackoverflow.com/questions/146452/what-are-pod-types-in-c

What are POD types in C++?

I've come across this term POD-type a few times. What does it mean?

plain old data

a

00:12

Thank

Afaik it’s a term that has been deprecated in the standard

github-merge-queue[bot]

[ddnet:master] 2 new commits

c984c68 Fix clang warning -Wbitwise-instead-of-logical - furo321 4af28fd Merge pull request #7672 from furo321/bitwise-instead-of-logical - heinrich5991

00:29

[ddnet/ddnet] Pull request closed: #7672 Fix clang warning -Wbitwise-instead-of-logical

it’s not much of an explanation

00:32

chiller afaik pod structs are classes or structs which can only be comprised of c-compatible types (primitives and other POD structs) so they can be treated as a contiguous chunk of data and not something more conceptual like classes in cpp

00:33

so a POD struct won’t be decorated with things like automatic ctors and dtors

Ah no nesting hm

00:34

I thought it’s just about not having behavior/methods

yea

Ah said they can contain other pod structs

00:34

Red your message wrong

yea

00:35

pod is a myth anyway

Ok

you shouldn’t need to have your struct be represented in this way unless you’re working with c code which needs to be able to understand your structure

00:36

the representation in memory is just an added bonus to that

You want StandardLayoutType nowadays

i got my dog back from the groomer

00:40

she is a very sweet girl

Ewan

it’s not much of an explanation

thats why i bringed SO answer but whatever

nice this multialias workaround has no sense since name is first argument and we can't use variadic without messing up whole file to only get some fancy brackets

01:17

also brackets are disallowed in c macro

01:17

not the brackets itself tho

01:18

like preprocesser thinks that MACRO({"first", "second"}) has two arguments isntead of one

01:18

im giving up on this

01:18

2 hours of lurking xd

[ddnet/ddnet] Pull request opened: #7674 Disable cycling through command history when logging in to rcon.

As it makes no sense to allow the user to cycle through the command history while in the rcon login prompt.

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 of bounds array indexing
[ ] Changed no physics that affect existing maps
...

gerdoe

thats why i bringed SO answer but whatever

i didn’t see it

gerdoe

like preprocesser thinks that MACRO({"first", "second"}) has two arguments isntead of one

quotes around and escape the pair item quotes

03:44

it will get you somewhere with stringify

[ddnet-maps:master] 1 new commit

5d2b62d A Twin Electric Field - ddnet-maps

[ddnet/ddnet] Issue opened: #7675 Client crashed on Arch Linux (Vulkan & NVIDIA)

Client only crashed when use any fullscreen(Vulkan)

gerdoe

#4100 was a mistake xddd

why xd

jopsti where do u live

morning vietnam

09:38

https://www.youtube.com/watch?v=khZq6JKVnlk

Good morning vietnam original

chillerdragon: in almania

gerdoe

like preprocesser thinks that MACRO({"first", "second"}) has two arguments isntead of one

Protip: MACRO(({"first", "second"}))

time to write yet another proc macro

09:50

it will be my 10th proc macro. I am getting better in this

jopsti which city

good morning

09:53

jopsti which street

09:53

which house

all i will say is, that it is "Münsterland"

Embedded image

09:53

lerato!

09:53

im close to rq

09:54

next to bielefeld lmao

09:55

lerato whats your guess how many times or days we have been in the same city at the same time

yeah

09:56

teero or smth is from bielefeld

09:56

this nobo

xd

09:56

imagine nobody beliving you where you come from

if you want some interesting history

09:57

https://en.wikipedia.org/wiki/Battle_of_the_Teutoburg_Forest

Battle of the Teutoburg Forest

The Battle of the Teutoburg Forest, described as the Varian Disaster (Latin: Clades Variana) by Roman historians, was a major battle between Germanic tribes and the Roman Empire that took place at modern Kalkriese from September 8–11, 9 AD, when an alliance of Germanic peoples ambushed Roman legions and their auxiliaries, led by Publius Quinctil...

09:58

this is the most famous one

09:58

also one of the most beautiful places around here

o

10:04

im onto somsin

10:05

lerato u knew the full solution all along didnt u -,-

10:05

why am i coding this then

10:07

Embedded image

10:07

Embedded image

10:08

im sure heinrich loves the simplicity xd

10:09

im still not sure i like all the added ({})

why do you use a macro?

@ChillerDragon

#include <iostream>
using namespace std;

template<typename H, typename ... T> constexpr auto last_of(H const& h, T const&... t)
{
    if constexpr (sizeof...(t)) return last_of(t...); else return h;
}

template<typename H, typename ... T> constexpr void A(H const& h, T const&... t)
{
    cout << h << " " << last_of(t...) << endl;
    if constexpr (sizeof...(t) > 1) A(t...);
}

int main(int, char**)
{
    A("r","rescue", "command for ChillerDragon");
    A("top","top5", "mess with selfesteem");
    A("earrape", "fng only");
}

10:28

could be eased if the helptext comes first in the parameter list

<ChillerDragon> macros can be redefined as something else

10:33

<ChillerDragon> but i just realized my approach is fucked

Do. Not. Use. Macros. Where. You. Can. Avoid. Them.

<ChillerDragon> Can we avoid them?

not always ^^

10:34

but it's rare

<ChillerDragon> i mean in our case

10:34

<ChillerDragon> not in general

10:34

<ChillerDragon> Imo the macros bring something really nice for us here

I posted a solution

<ChillerDragon> what about the client side then?

I've to confess I don't have a clue what you mean

<ChillerDragon> the client also includes the file

10:36

<ChillerDragon> but tbh we are not really using the power of macros yet

10:36

<ChillerDragon> but i would like to do that

10:36

<ChillerDragon> to also generate the function definitions of the callbacks

macros are very difficult to debug

<ChillerDragon> https://github.com/ddnet/ddnet/blob/4af28fd2d87e8e1c5629874b2a0ee16d0860a18b/src/game/client/components/chat.cpp#L34

ddnet/src/game/client/components/chat.cpp at 4af28fd2d87e8e1c562987...

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

10:37

<ChillerDragon> here we want the macro to mean something else than on the server side

10:37

<ChillerDragon> tbh i am also pro removing that

Fussel

Do. Not. Use. Macros. Where. You. Can. Avoid. Them.

Check lua source code troll

troll

<ChillerDragon> @Fussel did anyone ever have to debug them?

You don't really need macros and templates here, just always use {"cmd"} also if there is only one command and only work with an initializer list

<ChillerDragon> @Robyt3 you also wanna remove all macros!?

he wanted to get rid of the {}, it was his choice

Embedded image

10:39

sos

sose

seems like the initializer list goes out of scope and then all breaks

<ChillerDragon> here we want the macro to mean something else than on the server side ← and this is a good example for where things easily go wrong but are difficult to debug

@Fussel i do not remember debugging being an issue there ever

10:40

also do you recommend to change the whole current implementation of 10yr old macro code just because it could be difficult to debug? xd

as long as a macro works as intended, you don't need to do ^^

10:41

no, but it's unnecessary to proceed with this

@Robyt3 i am not sure i understood what you mean

I imagine it would look like this, untested:

#define CHAT_COMMAND(names, params, flags, callback, userdata, help) \
for(const char *pName : names) \
    m_pConsole->Register(pName, params, flags, callback, userdata, help);

CHAT_COMMAND({"map"}, "?r[map]", CFGFLAG_CHAT | CFGFLAG_SERVER | CFGFLAG_NONTEEHISTORIC, ConMap, this, "Vote a map by name")

(edited)

<ChillerDragon2> wont work

i guess

10:44

because you didnt put (

10:45

CHAT_COMMAND({"map", "bar"}, "?r[map]", CFGFLAG_CHAT | CFGFLAG_SERVER | CFGFLAG_NONTEEHISTORIC, ConMap, this, "Vote a map by name")

10:45

here the , between map and bar ist a seperator for the macro args

10:45

so you need partenthesis around

Makes sense

which then breaks the : loop

10:47

and then the () are included

10:48

Embedded image

10:48

without it works fine

10:48

with it doesnt

10:48

so i decided to call std::initilaizer_list manually and let it use the parenthesis

10:48

problem is that variable i defined now goes out of scope and leaves a dangling pointer

10:49

in my test env i had warnings off and all worked fine but in ddnet code -Wdangling broke :c

ChillerDragon

lerato whats your guess how many times or days we have been in the same city at the same time

Once maybe

yea sounds realistic

10:49

chiler reply guessing pro

10:50

i srsly developed another sense

Robyt3

You don't really need macros and templates here, just always use {"cmd"} also if there is only one command and only work with an initializer list

Now that is nasty, + it still requires the extra parens too because macros don't care about braces

so how to fix

10:51

im out of ideas

Remove the macro entirely and pass an initializer list directly to an overloaded function pepeH

pepeH

Just mess around with it until you get bored. I'll give it a lookover if you don't find the solution

omg robsit

10:53

but then it breaks my epic macro plan

10:53

of deleting half of gamecontext.h

If you want to get rid of macros you can also use polymorphism to achieve the difference in behaviour

i do not

10:54

want

10:54

to

10:54

get

10:54

rid

10:54

of

10:54

the

10:54

macros

I love the macro too. So elegant

get rid of the macros

10:54

use rust

auh9wdh8oawd

use proc macros

Chat commands shouldn't really be registered for all mods as macros on the server side, if you want to support modding. This seems like a useless macro to me:

#define CHAT_COMMAND(name, params, flags, callback, userdata, help) m_pConsole->Register(name, params, flags, callback, userdata, help);

Jupstar ✪

use proc macros

those are cleaner

10:54

actually in rust

This code should be in the individual mod gamecontrollers

we could have made a macro with a custom DSL for declaring commands

10:55

we could also do the localization with macros

@Robyt3 it can be in gamecontrollers already

10:56

https://github.com/ZillyInsta/ddnet-insta/blob/17974abe3fe1922c7d4ad8655b77ea4619774fd2/src/game/server/gamemodes/base_instagib.cpp#L17-L19

ddnet-insta/src/game/server/gamemodes/base_instagib.cpp at 17974abe...

A teeworlds instagib (grenade/laser capture the flag) mod based on DDRaceNetwork (gctf/ictf) - ZillyInsta/ddnet-insta

10:56

userdata and callback can be something other than cgamecontext

AFAICT the only reason for the chat command macros to exist is because they are hard-coded for the client as well (#7665), which I agree should probably not be the case anymore (in some time when most mods had time to update to using chat command messages)

https://github.com/ddnet/ddnet/issues/7665

Do not let the client assume available chat commands · Issue #7665 ...

Since #7649 the server can inform the client about what chat commands it has. The ddnet client has a list of commands which matches the current ddnet version and it will always autocomplete those. ...

Now just expand the macro and the inclusion and it looks good, I don't see a reason for this indirection (edited)

and the next reason will be to generate this with macros https://github.com/ddnet/ddnet/blob/4af28fd2d87e8e1c5629874b2a0ee16d0860a18b/src/game/server/gamecontext.h#L369-L482

ddnet/src/game/server/gamecontext.h at 4af28fd2d87e8e1c5629874b2a0e...

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

10:57

wat does that mean robsti

10:58

which indirection

It means removing the macro and the file

o

And putting the actual C++ directly into the file

hm

10:58

bbut

Because it only makes it more complicated to go to other files again

imagine having to define functions in a header file

10:58

so annoying

10:58

when it can be macro generaterd

That's why command callbacks should be std::functions so you can pass lambda expressions when registering the command

oh

This additional indirection of command handler functions is even more annoying

ok you totally knocked me out now xd

11:00

yea sounds good but i suddenly lost motivation

Can multiple joins/leaves happen between 2 consecutive snapshots?

ChillerDragon: It's a huge change and I don't want you or me to do it right now, but that's the end result I would like to see

yea but then

11:01

what am i doing here

11:01

:D

11:01

i just close the pr and call it a day

11:01

sad that i didnt get to generate headers

In general I'd prefer less usage of the preprocessor instead of more

11:02

Hard-coded server console commands should also be removed from the client eventually

yes

11:02

eventually

[ddnet/ddnet] Pull request closed: #7657 Add aliases for chat commands

chiller why did u ask btw

i realized at 4am that i do not know where you live

11:07

which i really should know

11:07

so i waited until the morning to ask

are you a beautiful, intelligent, althetic woman?

no

ah too bad

xd

xd

i need to know where people live that i interact with in teeworlds

11:08

in case i am near by

11:08

to hookup

justatest

1

11:08

my address being public helped me meet zwelf for example

yeah

11:09

i remember your photo in a photo of a magnet

xd

barcelona

ik

11:10

ur

11:10

and i know lerato is homeless

lol

deen i forgot tbh xd

let's all go to barcelona

11:10

at least there is good weather

heinrich i already saw

rip

leak

11:11

was ccc around this dates

11:11

i wish they did smth similar here

ryo go organize

https://vxtwitter.com/spectatorindex/status/1736330277428113765 (edited)

The Spectator Index (@spectatorindex)

BREAKING: French diplomat has died of his wounds after being injured in Israeli airstrike on Gaza

178

64

11:12

rip

ohno politics

chiller

jopsi

can you do a video like this again? https://www.youtube.com/watch?v=wKV18A1DluU

Teeworlds in 2020

it was always planned as 2 episode thing

11:13

but i got a lot of requests i might do it

but with a world tour again

11:14

https://www.youtube.com/watch?v=C8YQP-YH-0c ah yeah this was the best

Teeworlds 0.7 is released

11:14

0.7

xd

11:14

jopsti 0.7 enjoyer

dead version

it's just so troll to play teeworlds somewhere in the world in the public

make a ddnet vid

such a influencer moment

imagine

11:15

seeing a guy

11:15

with his laptop on the ground

11:15

middle of a park

11:15

with tw

11:16

saying 0.7 released!

11:16

x

11:16

d

xd

im sad resamvi doesnt do vids

11:16

best youtuber

11:16

https://www.youtube.com/watch?v=FQIOXVUcNew

Teeworlds - ResamVi: Goodbye

chiller do u want to meet ryo?

11:18

he is a blond latino

yes

lol

11:18

not latino

next time im in barcelona

11:18

ima ping the shits out of ryo

nice

xd

my next planned meet is some 12yo kid from block

11:18

thats gonna be cring

lol

wtf

11:19

pedo

it indeed is weird

why u meeting him

we play a lot together and im in his home town

11:19

might as well say hi

and he's literally 12?

he might be 18 idk xd

uhh

xDD

maybe he is 40

12 is only learath

chiller pls

wot

11:20

@Axi

11:20

did the ping work?

yes

yea

he/him/bulgarian im a divorced 52 years old male and i have no life and friends

hi axi

@Axi how old r u

hey

XD

im fine

11:21

and you chiller?

XDDDDDDDDDDDDDD

11:21

he'll troll u to death

who who

11:21

bulgi biggest trol

11:21

@Axi how old

17

11:21

xd

yea still weird af

yeah

11:22

don't get raped by chiller

how old r u chiller

11:22

im 23

chiller is 25

age reveal

soon 30

sadboi

i know everything about chiller

ok afk

Ryozuki

im 23

i keep forgetting you still so young

11:25

why am i so old

how are u so old

i dunno when the fuck the time passed so fast

11:32

i swear yesterday i was like 20

i should make yt vids but lazy

cyberFighter

i should make yt vids but lazy

horse races or what are they about

Learath2

Protip: MACRO(({"first", "second"}))

i did it all, couldnt compile either

Jupstar ✪

why xd

brought protocol structure for some reason

ChillerDragon

my next planned meet is some 12yo kid from block

Don't if you don't want to become jail

gerdoe

i did it all, couldnt compile either

tried one more time - still doesn't work xd

Chiller are you still interested in the console command registration thing? If not I'll make template blackmagic

hello jail I'm Teero

gerdoe

tried one more time - still doesn't work xd

There is no way the preprocessor is allowed to split that into two, are you sure you aren't testing it wrong?

Robyt3

I imagine it would look like this, untested:

#define CHAT_COMMAND(names, params, flags, callback, userdata, help) \
for(const char *pName : names) \
    m_pConsole->Register(pName, params, flags, callback, userdata, help);

CHAT_COMMAND({"map"}, "?r[map]", CFGFLAG_CHAT | CFGFLAG_SERVER | CFGFLAG_NONTEEHISTORIC, ConMap, this, "Vote a map by name")

(edited)

tested, preprocesser thinks that {"map", /* <- this one */ "nomap"} has two arguments isntead of one

12:09

templates > macros

Skill issue

xd

12:13

i have forgotten why did i close #6517

https://github.com/ddnet/ddnet/pull/6517

Add CObjectPool, the replacement of alloc.h new overriding ma...

There's nothing special, switched from previous global pool to local one. Next step is unifying m_apPlayers calls by introducing new CObjectPool *PlayerPool() const or CPlayer *GetPlayer(int Cl...

seems to be messed up tho

ChillerDragon

Click to see attachment 🖼️

ah

12:17

real skill issue then

[ddnet/ddnet] Pull request opened: #7676 Minor refactoring of demo_extract_chat tool

Replace unnecessary gameclient.h include with more specific includes. Fix storage creation error message not being logged as the logger was initialized after checking for the failed storage creation. However, in this case we want to avoid non-error log messages so the tool's output is only the extracted demo chat, except in error cases. Rename Process function to ExtractDemoChat and make it static to avoid exporting it. Use log_error instead of dbg_msg.

Checklist

...

[ddnet/ddnet] Pull request opened: #7677 Refactor map_resave tool

Add log messages for errors and success. Extract function ResaveMap. Ensure reader is closed when writer could not be opened.

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 of bounds array indexing
[ ] Changed no physics tha...

I have up and are totally out. Feel free to do whatever c:

Replying to @Learath2 Chiller are you still interested in the console command registration thi…

13:05

Gave

13:05

Fuxk phones

github-merge-queue[bot]

[ddnet:master] 2 new commits

1f5a648 Minor refactoring of demo_extract_chat tool - Robyt3 88ce582 Merge pull request #7676 from Robyt3/Tools-Demo-Extract-Refactoring - Chairn

13:24

[ddnet/ddnet] Pull request closed: #7676 Minor refactoring of demo_extract_chat tool

13:27

[ddnet/ddnet] Pull request closed: #7677 Refactor map_resave tool

13:27

github-merge-queue[bot]

[ddnet:master] 2 new commits

a883018 Refactor map_resave tool - Robyt3 d3959e2 Merge pull request #7677 from Robyt3/Tools-Map-Resave-Refactoring - Chairn

nah its unreal for me to make it work

@TsFreddie im improving my flutter

13:46

dart is like modern java

13:47

Image attachment

Ryozuki

@TsFreddie im improving my flutter

make an AI Chat bot

that can be cool

so i can dump chillers chat history in there and have a mini chillerdragon in my pocket

:O gimme gimme

@TsFreddie do u know any public apis?

13:49

or run my own model

13:49

sounds expensive

RWKV can probably run on phones

https://www.rwkv.com/

RWKV Language Model

The RWKV Language Model

13:49

this?

ye

uh oh

13:50

but idk how to use this

idk either lol

13:51

i tried compiling rwkv.cpp and it ran fine on cpu only.

maybe something simpler as a first complex flutter app

13:51

xd

It’s probably just a couple python calls

actually ill use my own todo app as a list to buy groceries

13:51

i always wanted an add free one

so its probably okay to run a 1.4b models on phones with cpu only

Everything AI is made for the lowest common denominator. The new pRoGrAMmeRs

Learath2

It’s probably just a couple python calls

i prefer smth dart natvie

13:52

native

13:52

i need to see how dart can interface with python

13:52

https://dart.dev/

Dart programming language

Dart is a client-optimized language for fast apps on any platform.

Ryozuki

i always wanted an add free one

jtx Board is foss i think?

btw the hot reload is insane

13:53

i literally run the app on all time

13:53

modern tooling is smth else

Hot reload is such a pleasure to have

also using the app on my phone is even seamless

i mean, rwkv.cpp is native owo

owo

13:56

rwkv ai00 server also has a vulkan implementation

13:56

its pretty cool to not rely on cuda or python

also the refactor abilities are p good

14:07

on this editor

14:07

Image attachment

when do we get better antiping xd

get better internet

14:12

4Head

literally every game ever accounts for poor internet except ddnet

Ryozuki

4Head

wtf is this emote

4head

14:31

from twitch

Voxel

literally every game ever accounts for poor internet except ddnet

with a tradeoff

i thought it was stepfunn

the tradeoff is to give more control to client

14:31

which allows hacks

14:31

ddnet is completly server authoritive

14:32

and well it requires precise movements

Teero

when do we get better antiping xd

kog+bad internet+loud music

gerdoe

kog+bad internet+loud music

music not loud xddd

kog players cant live without their loud music

downloading gta6

k

do you really want to replay trailer forever

gta6 mid whens gd 2.2 gonna finish

nouis

1

Voxel

kog players cant live without their loud music

@Mixi they saw your videos MONKAS

MONKAS

i did not

Voxel

gta6 mid whens gd 2.2 gonna finish

never

14:47

its like teeworlds 1.0

oh yea i forgot teeworlds was forever under 1.0

and will be forever

14:49

dead franchise

teeworlds versions

Image attachment

i shouldve joined ddnet 5 years ago max

soon i'll know teeworlds for half of my life

pepeW

1

wow

cringe ur not from 2008

yeah i am pretty late

imagine u found the game after ddnet rls

i got into it after all the fun stuff happened

i think i almost started around when ddnet was released

Voxel

i got into it after all the fun stuff happened

but ur new

14:56

it cant be

14:57

0.4.3 was next gen

grandma coke

pretty new

Cøke

but ur new

ik but i heard old versions / events were better than this one

yes bcs nowdays nobody care

14:59

we had....wait....0 tourneys this year

dead franchise x2

only one will be a simple hh map that u can play one time

only reason its still gonna go next year is because heinrich paid like 600+ for the servers LOL

will be better when roles change

15:00

Voxel

only reason its still gonna go next year is because heinrich paid like 600+ for the servers LOL

u dont know hahahah

15:00

it was like never reached

damn so it Is dead

deen has a good heart

15:01

and know the game is the best

the best in...?

all categorys

false

u cant tell me a better game

i guess games just suck then

15:02

womp womp

no p2w, only own skillz

15:03

idk its the best game

15:03

but kids these days want to be the best after 3 hours

Cøke

no p2w, only own skillz

skillz and The Internet Speed of Jesus Christ

and good mouse

but i can imagine how easy ddnet is when you guys release the same gameplay 7 times

15:05

in a month

wat?

and 3080

now u confused me

Ryozuki

and 3080

no, most optimized for me GPU

15:05

xd

Jupstar ✪

now u confused me

current maps are a bore

15:05

its all just hh

play gores

15:06

then u have smth to do for the next 3 years

thats not any better

it can't be boring if u are not good enough to solo finish

15:06

guess i joined ddnet too late then

15:07

oh well

chill dude

15:07

just start game

15:07

play

15:07

die bcs u took it too serious

get bored quit

Voxel

but i can imagine how easy ddnet is when you guys release the same gameplay 7 times

u cant bcs u are not playing

15:17

u dont even know the basics

but do i want to

?

todo-flutter-ryo.apk

18.97 MB

16:06

anyone wanna try my todo app

16:06

owo

16:06

(i promise no sus code)

Sus

i made a google play console acc (edited)

16:08

i paid 25€ to the corpos

[ddnet/ddnet] Pull request closed: #7674 Disable cycling through command history when logging in to rcon.

16:11

github-merge-queue[bot]

[ddnet:master] 2 new commits

2f89297 Disable cycling through command history when logging in to rcon. - furo321 4ddf1df Merge pull request #7674 from furo321/history-login-prompt - Robyt3

Hi, guys. Can you help me? I need pos of my crosshair (edited)

16:15

it is pChr->m_Input.m_TargetX and m_TargetY or something another?

16:15

logistic_robot

banhammer

Cøke

0.4.3 was next gen

0.5 is mainstream then xd

Jupstar ✪

banhammer

i need to get is on the server side

16:15

poggers2

it should be the cursor pos yes

do i need to * 32?

16:16

or its correct already?

but u could easily check by printing these variables

16:16

mhh dunno rn

huh?

Ryzen

do i need to * 32?

You don't need to multiply it.

thx

[ddnet/ddnet] Pull request opened: #7678 add ui setting for cl_show_local_time_always

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 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-addresss...

have any of you ever had cracking, popping sounds when try to play audio on discord, zoom in firefox justatest

justatest

(also it can crash a page) (edited)

17:38

asking for a friend

sounds weird

17:57

maybe the sound cracking is a side-effect to a driver problem?

but spotify and youtube work fine (edited)

17:57

:\

what os?

gentoo

justatest

MilkeeyCat

asking for a friend

xD?!

justatest

17:59

ill try to install chrome

did you convert him to gentoo ala ryo style or is your friend you?

Mr.Gh0s7

did you convert him to gentoo ala ryo style or is your friend you?

2nd one is correct

18:00

and sooner or later ill install linux on my friend's pc

gigachad

18:01

can you backtrace it in any way?

18:04

like are there logs or smt?

there should be

18:05

and i installed chrome owo

owo

18:05

time to remember password for discord pepeW

pepeW

ez 5th try, holy hel

ez5thtry,holyh31 nice pwd

nothing on chrome

18:07

always knew firefox sucks

18:07

im chrome andy now

hi

18:17

where is server log file ?

18:20

wow , here is not much going on without me

18:24

the quitting. please wait is so annoying on windows "idk about linux tho" it take sometimes a minute to quit the game

yeah why does it do that

The client needs to wait for all background jobs to complete before it can quit or it may crash when quitting

do you suggest i don't alt+f4 my way out ?

18:52

is it the same on linux ?

რილია

do you suggest i don't alt+f4 my way out ?

Probably doesn't cause problems right now, but I wouldn't recommend it. You could also Alt+Tab out already.

რილია

is it the same on linux ?

Should be, it's the serverlist refresh jobs taking the longest

I can confirm the "bug" is on linux too

Robyt3

Should be, it's the serverlist refresh jobs taking the longest

it can't force jobs to end?

Doesn't work very well with the current curl interface we are using, would be improved by #5842

https://github.com/ddnet/ddnet/pull/5842

Use curl-multi by Learath2 · Pull Request #5842 · ddnet/ddnet

Supersedes #5092, I am still not very sure about this structure either, but I think it's time someone else takes a look. Remaining concerns: Should errors in the http module just be fatal? I...

1

pepeW

just client.cpp things (edited)

Image attachment

about render text

20:47

can i make somehow to rainbow only the text i want to be rainbowed (edited)

20:48

cause it looks like

20:48

Image attachment

20:48

eww

The text color is a global variable, you need to reset to the default text color after doing your colored text

1

works, thanks

i remember running into that problem when redoing the editor ui

1

1 more question

21:14

how can i put multiple names in strcmp?

21:14

Image attachment

What do you want it to do? Check if the name matches either or matches both?

put multiple names like -StormAx || melom etc

21:15

so i don't have to make 100500 new lines (edited)

Move the logical or outside the function call: str_comp(..., ...) == 0 || str_comp(..., ...) == 0

im getting special notice in your client ? monkaStop

monkaStop

ofc ;D

21:18

ugmmm

21:18

like this?

21:18

Image attachment

thats one way to do it, yes

ok cool

21:19

:D

21:19

thanks

Robyt3

The client needs to wait for all background jobs to complete before it can quit or it may crash when quitting

Out of curiosity, are those jobs network oriented? I did not seem to encounter that screen when having a pretty good internet

Devinci

Out of curiosity, are those jobs network oriented? I did not seem to encounter that screen when having a pretty good internet

Yeah, it's the jobs to determine the best master server. If you have a good connection to all master servers then it should be done quickly

21:35

If you don't quit right after refreshing the server list (or after starting the client) then it should also quit pretty fast

Oh that's very good to know. I'll keep in mind to not refresh before leaving, something that I do quite a lot xD. Thanks!

Can we cache that result across reboots?

Replying to @Robyt3 Yeah, it's the jobs to determine the best master server. If you have a g…

Possibly, but I'm almost done reworking #7376. Now that jobs can be cancelled, the client quits almost immediately.

https://github.com/ddnet/ddnet/pull/7376

Add IJob::Done callback called on main thread by Robyt3 · Pull Re...

We currently poll if specific background jobs are done in various places to perform some final computations on the main thread. To simplify job handling the virtual function IJob::Done is added to ...

chillerdragon

Can we cache that result across reboots?

You can create ddnet-serverlist-urls.cfg in the same directory as settings_ddnet.cfg and put a url there. https://master1.ddnet.org/ddnet/15/servers.json for example and it should be cached.

I don't think that prevents the requests to determine the best master though

It seems to, the game quits pretty much instantly now.

At least for quitting after initial start it doesn't seem to make a difference, i.e. with command line ddnet quit

It does for me?

With ddnet-serverlist-urls.cfg
DDNet quit  0.81s user 0.64s system 88% cpu 1.641 total

Without
DDNet quit  0.92s user 0.68s system 31% cpu 5.011 total

Maybe it simply doesn't try other master servers

22:25

Jobs should simply get a bool "canceable" and skip execution if true and in destructor

22:26

Why does it matter that the http request is not finished?

22:28

Or can one not cancel curl requests?

Jupstar ✪

Or can one not cancel curl requests?

Yes, but we are not doing that at the correct time right now, we are only cancelling jobs in destructors when it's too late. I'll probably update 7376 tomorrow, adding IJob::Cancel.

Robyt3

Yes, but we are not doing that at the correct time right now, we are only cancelling jobs in destructors when it's too late. I'll probably update 7376 tomorrow, adding IJob::Cancel.

So our problem is that we don't handle any curl error if curl was stopped and what ever blocking call returned an error? Or is that simply not possible? (edited)

The code is already there to properly abort the curl calls, but we only abort the HTTP requests when they are already done effectively

Is there no global instance or something for curl, that we have to stop?

22:33

Or can stop*

That's probably what curl-multi is for, it will make quitting even faster I assume

22:34

We neglect to call curl_global_cleanup

Mh OK. Didn't know curl is such low level api

Curl has a progressfunction callback that can tell curl to abort, but we don't tell curl to abort soon enough at the moment

I find it really overkill how hard it is to use, considering we do like 2 http requests lmao

theoretically we could just quit the process while the http requests are still running

22:38

but we don't have the signaling that the http requests are unimportant

22:38

and since we introduced important jobs, such as editor saving, we have to wait for some jobs, at least

Well abort sounds like something a async code could support. So i guess that would be one good way.

we're using the sync interface of curl (in a thread) (edited)

22:40

curl-multi would be the async interface

Yeah but apparently we can handle aborts already

22:41

As far as i understood rbyte

Robyt3

The code is already there to properly abort the curl calls, but we only abort the HTTP requests when they are already done effectively

Here

I'm not sure that actually works if no bytes are currently being received on the http stream

heinrich5991

I'm not sure that actually works if no bytes are currently being received on the http stream

Yeah, that's the issue with the current progress callback, it is only called when bytes are received

Maybe we could use a simpler http lib

22:42

There doesn't seem very high interest by any Dev to maintain good code

when "just quitting" works, what's the problem with libcurl?

22:43

curl is the de facto standard library for doing http(s) requests, don't htink we should switch to something else

But apparently we cannot just abort requests

22:44

And we de facto don't have curl multi

we can easily abort requests

22:44

(at the time where we want it)

22:44

by just not waiting for the curl threads to complete

Ok then apparently we can now

Seems to work well enough with what I have right now, but curl-multi would be even better (edited)

You mean a quick exit?

no

22:45

just no thread_join on the thread doing the http request

That sounds like a hack to me. Leave it to the OS. Im lazy

22:46

In theory it could wait for it

the OS is much better at cleanup anyway

heinrich5991

and since we introduced important jobs, such as editor saving, we have to wait for some jobs, at least

How would you model "important" jobs? bool IsImportant() with default false or bool IsCancellable() with default true?

bool ShouldWaitBeforeQuitting()? maybe even defaulting to true to opt out of it?

heinrich5991

just no thread_join on the thread doing the http request

Slightly annoying because the process will linger around for a moment

heinrich5991

the OS is much better at cleanup anyway

That's same thinking as global variables. It's not modular, it's not reusable. I dislike it

which process will linger around?

Jupstar ✪

That's same thinking as global variables. It's not modular, it's not reusable. I dislike it

it's also faster ^^

heinrich5991

which process will linger around?

Or does the process die immediately if only curl threads remain?

heinrich5991

it's also faster ^^

I can live with slower exit, if that means cleaner code

22:48

The Vulkan spec even forbids it

Robyt3

Or does the process die immediately if only curl threads remain?

it will exit immediately

22:52

since leaving main e.g. calls exit_group on Linux

22:52

which kills all other threads as well

Jupstar ✪

The Vulkan spec even forbids it

what does the vulkan spec forbid?

heinrich5991

what does the vulkan spec forbid?

Leaving resources alive

at process exit?

23:02

got a link?

heinrich5991

at process exit?

It doesn't know about what a process is

https://en.wikipedia.org/wiki/Very_long_instruction_word

Very long instruction word

Very long instruction word (VLIW) refers to instruction set architectures designed to exploit instruction level parallelism (ILP). Whereas conventional central processing units (CPU, processor) mostly allow programs to specify instructions to execute in sequence only, a VLIW processor allows programs to explicitly specify instructions to execute...

Jupstar ✪

It doesn't know about what a process is

would still be interested in a link

heinrich5991

would still be interested in a link

Well it's in the spec as said https://registry.khronos.org/vulkan/specs/1.3-extensions/html/vkspec.html#vkDestroyDevice

Prior to destroying a device, an application is responsible for destroying/freeing any Vulkan objects that were created using that device as the first parameter of the corresponding vkCreate* or vkAllocate* command.

23:09

is that what you're talking about? it seems to talk about not freeing parent objects before freeing all children

23:09

(like not calling curl_global_cleanup while stuff is still running)

Nope. Read the note

23:10

It does not give you that guarantee

Prior to destroying a device, an application is responsible for destroying/freeing any Vulkan objects that were created using that device as the first parameter of the corresponding vkCreate* or vkAllocate* command.

23:10

this?

23:11

it still seems to only say that you need to free children objects before the parent (edited)

This is also it. But you quoted the same thing again

ah. oops

heinrich5991

it still seems to only say that you need to free children objects before the parent (edited)

Why only?

23:11

You need to do that

it doesn't seem to be unambiguous to me

It's not a matter of opinion. The Vulkan spec says it. If you don't do it it's undefined behavior

it is critical to turn off the engine prior to leaving the car to avoid fuel wasting

23:13

this doesn't tell me to leave the car AFAICT?

If you don't clean it. It might leak

23:14

Just because our awesome operating systems will be more clever than us doesn't mean it's correct

that doesn't seem to be useful

23:15

the operating system always has to clean up this stuff to function correctly

23:15

the process might be killed, for example

Ok. I have sleep to catch. Not worth my time. Night

night

@Jupstar ✪ @heinrich5991 https://ryhl.io/blog/temporary-shared-mutation/

Temporarily opt-in to shared mutation – Alice Ryhl

The purpose of this blog post is to celebrate the anniversary of two really neat methods on the Cell type: from_mut and as_slice_of_cells. Both methods were released in version 1.37.0 of Rust, exactly one year ago from the date this post was published.

yea, seen it in the stdlib docs

https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/commit/?id=cbe0e415089636170aa6eb540ca4af5dc9842a60

23:28

net: phy: add Rust Asix PHY driver This is the Rust implementation of drivers/net/phy/ax88796b.c. The features are equivalent. You can choose C or Rust version kernel configuration.

interesting

23:31

https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/tree/drivers/net/phy/ax88796b_rust.rs?id=cbe0e415089636170aa6eb540ca4af5dc9842a60

23:31

https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/tree/drivers/net/phy/ax88796b.c?id=cbe0e415089636170aa6eb540ca4af5dc9842a60

Exported 676 message(s)