DDraceNetwork - developer

053c440 Add 150 € donation by SpeedyGER & Chillbreaker - def-

almost there: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/65 then i can finally use wayland, epic

Tearing updates protocol (!65) · Merge requests · wayland / wayland...

This protocol is meant to be used by drivers to be able to hint to the compositor if and when it should employ asynchronous page flips for presentation....

09:23

hope they get it done before the year ends

https://lore.kernel.org/lkml/202210010816.1317F2C@keescook/

10:01

Subject: [GIT PULL] Rust introduction for v6.1-rc1

Ryozuki

https://lore.kernel.org/lkml/202210010816.1317F2C@keescook/

i really want to see some first driver written in it. maybe a rewrite of an existing actually. good for comparision

@Jupstar ✪ they already did

10:46

the nvme driver

10:46

it perfoms the same as the c one

10:46

https://news.ycombinator.com/item?id=32820219

mfrw

NVMe driver in rust with similar perf to C impl (Linux Plumbers 2022)

ok but maybe something more cpu bound is more interesting

def-

[ddnet-scripts:master] 1 new commit

3afd55e Remove custom pip repos - def-

Ryozuki

it perfoms the same as the c one

Well similar, seems there are some cases where it performed measurably worse

if only llvm wasnt buggy with the noalias feature

10:49

rust would be faster in some places

10:49

(not saying this specific to this case)

Yeah could be llvm issue, or maybe a gcc frontend for rust might generate better code

but anyway its also not interesting only from performance point of view:

build times
quality of code (how many bugs will it contain)
debugging tools etc

its cuz llvm rly was only focused for C/c++ issues

10:50

rust made llvm better

10:50

by pushing bugs and fixes

Learath2

Yeah could be llvm issue, or maybe a gcc frontend for rust might generate better code

the noalias thing is a known issue

10:51

https://stackoverflow.com/questions/57259126/why-does-the-rust-compiler-not-optimize-code-assuming-that-two-mutable-reference

Why does the Rust compiler not optimize code assuming that two muta...

As far as I know, reference/pointer aliasing can hinder the compiler's ability to generate optimized code, since they must ensure the generated binary behaves correctly in the case where the two

Jupstar ✪

but anyway its also not interesting only from performance point of view:

build times
quality of code (how many bugs will it contain)
debugging tools etc

Quality of code is hands down better, build times probably are of the same order of magnitude should be similar, tools idk. Debugging rust with gdb hasn't always been the smoothest of experiences for me, though it might just be a pebcak.

the answer explains a lot

def-

[ddnet-scripts:master] 1 new commit

66c3d92 Use --upgrade in pip3 to get latest version - def-

Learath2

Quality of code is hands down better, build times probably are of the same order of magnitude should be similar, tools idk. Debugging rust with gdb hasn't always been the smoothest of experiences for me, though it might just be a pebcak.

mh ok, i often read on reddit or other forums, ppl will join development of the kernel as soon as it supports rust, bcs they feel more confident But i cannot really imagine why this should be the truth... you don't magically have more insight about how hardware works etc. Sure not having security issues creates kind of confidence, but if the kernel crashes.... it still sucks xD

you dont have to know the intriquacies of C and its possibles UB to contribute to the kernel

10:54

yes y ou need to know about hardware

10:54

and currently most good rust devs know a lot of c too

Jupstar ✪

mh ok, i often read on reddit or other forums, ppl will join development of the kernel as soon as it supports rust, bcs they feel more confident But i cannot really imagine why this should be the truth... you don't magically have more insight about how hardware works etc. Sure not having security issues creates kind of confidence, but if the kernel crashes.... it still sucks xD

They are being silly. Interacting with real hardware is still not safe. They won't contribute shit

Ryozuki

you dont have to know the intriquacies of C and its possibles UB to contribute to the kernel

In my experience C is not the problem, working with Linux core stuff like memory management is just complex and interconnected. Not sure about drivers

i think most ppl contribute to drivers

10:55

they make up the most

Learath2

They are being silly. Interacting with real hardware is still not safe. They won't contribute shit

Especially the part where rust is being introduced. They'd have to be the ones building safe abstractions over the hardware. You'd still be working in the very scary part where rusts safety guarantees havent come in yet

anyway rust is only for drivers yet

(I implemented Kernel Samepage Merging for anonymous pages once to get less memory consumption when running many similar VMs, but that's the extent of my Linux kernel development experienc)

https://stackoverflow.com/questions/64859526/does-linux-kernel-list-implementation-cause-ub

Does Linux kernel list implementation cause UB?

Prerequisites: As per C standard, pointer arithmetics that would yield an invalid pointer, cause undefined behavior. Linux source code seems to conform with C standard in a desire to be compatible...

10:56

xD

The kernel actually does invoke implementation defined behaviour a lot actually. That's why it only compiled on gcc for the longest time and still as of now

makes sense

UB per the standard doesn't always have to be UB per implementation. E.g. signed integer overflow, virtually every implementation overflows in a sane manner

deen

(I implemented Kernel Samepage Merging for anonymous pages once to get less memory consumption when running many similar VMs, but that's the extent of my Linux kernel development experienc)

Was it for work?

university project, probably back in 2011

11:04

oh, it was actually the opposite. Anonymous pages worked with KSM already and I tried enabling it for memory mapped files

Sounds like a fun project

I think it worked in the end, but I only tested with pretty small examples

2022-10-02 16

02 I discord: error initializing discord instance, error=4 can someone explain this error

Daniel

2022-10-02 16

02 I discord: error initializing discord instance, error=4 can someone explain this error

probably that it didnt find any discord running

12:21

on your computer

why you need to install discord to play ddnet 64bit? i mean 32bit start normally without problems

12:22

but skins loading insane long xD

Daniel

why you need to install discord to play ddnet 64bit? i mean 32bit start normally without problems

u dont, u can ignore that error

12:22

is your client not starting or what?

this is the only output i got from the console if i try to run 64bit version

Jupstar ✪

is your client not starting or what?

friend of my

Daniel

this is the only output i got from the console if i try to run 64bit version

probs bcs nobody cares about 32bit anymore, and discord doesnt provide an API for that

Daniel

friend of my

what operating system?

teeworlds 64bit over steam and download is working fine... ddnet 32bit working fine but ddnet 64bit not

GPU?

12:24

if he has an AMD card: https://ddnet.org/news/amd-beta-opengl-crash/ he should try vulkan or downgrade to a stable driver if he has an old computer, see pinned messages in #bugs

yea he got an amd

Right click DDNet in Steam, add "gfx_backend Vulkan"

can he change to vulcan in ddnet-settings.cfg

with "

12:26

yes

12:26

add gfx_backend Vulkan in the end

12:26

but pls with a K not C

all cool ❤️ i will try it now with him (edited)

thanks, it works

might go to japan the last week of the year and celebrate new year there

1

1

13:18

13:19

@Learath2 ill try some original ramen there BASED

Ryozuki

might go to japan the last week of the year and celebrate new year there

Japan is very very pretty and has amazing food. Good choice for a trip

14:07

Spent 2 weeks and didn't even get close to seeing everything I wanted to see

def-

[ddnet-web:master] 1 new commit

3638351 Update server stats - def-

pog

def-

[ddnet/ddnet] Pull request opened: #5905 Update Tutorial map by unique_2

We made a small update of the tutorial map because of the new freeze graphics. We also made the teleport tiles the airjump hints consistent. Idk how updating the map works on your end, ideally everyone keeps their time because the changes don't influence anyone's times.

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 (...

I just made a script to automate the procedure of having mutliple cursors/keyboards at the same time under X. https://wiki.archlinux.org/title/Multi-pointer_X

Multi-pointer X

16:42

Tell me if you want to take a look at the script :)

I would upload a vid of me playing with my brother from the same pc but ddnet doesn't capture the second mouse xD

Mr.Gh0s7

I would upload a vid of me playing with my brother from the same pc but ddnet doesn't capture the second mouse xD

did you mod the client for that? i mean i know that SDL sends a deviceID but it doesnt work out of box does it ? XD

no it didn't work

19:29

Atleast with the second mouse

yeah, when splitscreen support ^^

Else the keyboards were fine (edited)

ah ok, nice ^^

@ChillerDragon https://github.com/joschuck/matrix-webcam

GitHub - joschuck/matrix-webcam: Take your video conference from wi...

Take your video conference from within the matrix. - GitHub - joschuck/matrix-webcam: Take your video conference from within the matrix.

21:03

this is for you

def-

[ddnet/ddnet] Pull request opened: #5906 Revert ingame menu close button

Even I got confused by it and accidentally hit power off button. Left side seems not popular for close either (#5888) Since opening the menu requires Escape, players can probably figure out Escape to close it again.

Revert "Don't show home button ingame"

>

This reverts commit d3ff903d445a7c5d62f9ed55690546a4689a608d.

>

Revert "Move home button to very right so it is consistent with close button ingame (fixes #5881)"

>

This reverts commit 9cc08cc9d31820a57cc5e221dcad3785542...

22:01

def-

[ddnet/ddnet] Pull request closed: #5888 Move close button ingame to very left so old players don't use wrong muscle memory

https://lkml.org/lkml/2022/9/19/1105#1105.php