DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.tw/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet

Between 2023-08-01 00:00:00Z and 2023-08-02 00:00:00Z

Ryozuki

@Learath2 @Chairn did u know about https://en.wikipedia.org/wiki/B%C3%BCchi_automaton

think I've seen it before. not sure in which context

https://edgarluque.com/blog/daily-gentoo/

this time I decided it wasn't that worth it encrypting my whole disk, so I went a simpler way.

sounds like a bad idea

Gentoo as a daily driver

My experience using Gentoo

07:30

the only benefit of that is less complexity. if that complexity is harder in gentoo than in arch linux, then I'd advocate against using gentoo

[ddnet/ddnet] Pull request opened: #6957 Delete fishy-looking assignment operator/copy constructor of IJob

Since they're not called anyway, they can go away.

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 memc...

heinrich5991

the only benefit of that is less complexity. if that complexity is harder in gentoo than in arch linux, then I'd advocate against using gentoo

its not harder its the same

08:17

its just that i didnt want to bother at a time

08:17

and well its not my laptop its my home pc

08:17

idk i just thought it wasnt worth it

08:17

also i think there is slightly less disk perfomance

08:18

and one of the reasons i wanted gentoo too was to achieve maximum perf

https://blog.apnic.net/2021/06/17/how-a-small-free-ip-tool-survived/

icanhazip: How a simple IP address tool survived a deluge of users ...

Major Hayden made a simple and free tool to check the public-facing IP address of infrastructure. Its popularity nearly killed it.

one of my grips with c++ is that = can mean many different things in many contexts, and overloads right? (edited)

08:19

in rust u always know what it means

GitHub

Click to see attachment 🖼️

@heinrich5991 u sure they arent called?

Ryozuki

@heinrich5991 u sure they arent called?

it compiles

well i guess that makes sense then xD

08:26

@heinrich5991 what do i need to get asan ubsan etc? just clang? valgrind?

ubsan clang I think

08:27

not sure about asan etc.

08:27

ubsan: pass -fsanitize=undefined

08:27

https://master1.ddnet.org/cdn-cgi/trace

UBSAN_OPTIONS=suppressions=./ubsan.supp:log_path=./SAN:print_stacktrace=1:halt_on_errors=0 ASAN_OPTIONS=log_path=./SAN:print_stacktrace=1:check_initialization_order=1:detect_leaks=1:halt_on_errors=0 LSAN_OPTIONS=suppressions=./lsan.supp ./DDNet

08:29

hm i dont see any output

08:34

if it doesnt say anything and just ddnet logs it means all is ok?

08:34

xd

[ddnet/ddnet] Pull request opened: #6958 Fix stack overflow in destructor when dnsbl object is kept too long alive

shared_ptrs of CServer::m_pDnsblLookup[ClientID] are set, but not cleaned up when done. Therefore the Job is kept alive until the player disconnects and a new player joins on that slot. Currently this means that the full linked list of jobs is kept alive. When the Job is overwritten with a new job, all the remaining objects in the list can be dropped. With enough jobs, that is causing a stack overflow in the destructor. This patch fixes this overflow by making the lifetime independe...

doesnt show up on ldd

08:41

ok got it now xd

08:42

=================================================================
==22179==ERROR: LeakSanitizer: detected memory leaks

Direct leak of 72704 byte(s) in 1 object(s) allocated from:
    #0 0x558d9f5f9d1e in malloc (/data1/edgar/ddnet/build/DDNet+0x1325d1e)
    #1 0x7f62ff140f65  (<unknown module>)
    #2 0x7f62ff0c9caa  (<unknown module>)

Direct leak of 6360 byte(s) in 5 object(s) allocated from:
    #0 0x558d9f5f9e58 in calloc (/data1/edgar/ddnet/build/DDNet+0x1325e58)
    #1 0x7f63135f2150  (<unknown module>)

Direct leak of 1152 byte(s) in 1 object(s) allocated from:
    #0 0x558d9f5f9fe5 in realloc (/data1/edgar/ddnet/build/DDNet+0x1325fe5)
    #1 0x7f63135f28e6  (<unknown module>)

Direct leak of 368 byte(s) in 3 object(s) allocated from:
    #0 0x558d9f5f9d1e in malloc (/data1/edgar/ddnet/build/DDNet+0x1325d1e)
    #1 0x7f63000a1377  (<unknown module>)

Direct leak of 176 byte(s) in 1 object(s) allocated from:
    #0 0x558d9f5f9e58 in calloc (/data1/edgar/ddnet/build/DDNet+0x1325e58)
    #1 0x7f631e9bfbde  (/usr/lib64/libdbus-1.so.3+0x30bde)

Direct leak of 8 byte(s) in 1 object(s) allocated from:
    #0 0x558d9f5f9d1e in malloc (/data1/edgar/ddnet/build/DDNet+0x1325d1e)
    #1 0x7f63135f1824  (<unknown module>)

Indirect leak of 164592 byte(s) in 521 object(s) allocated from:
    #0 0x558d9f5f9e58 in calloc (/data1/edgar/ddnet/build/DDNet+0x1325e58)
    #1 0x7f63135f2150  (<unknown module>)

Indirect leak of 9650 byte(s) in 264 object(s) allocated from:
    #0 0x558d9f5f9d1e in malloc (/data1/edgar/ddnet/build/DDNet+0x1325d1e)
    #1 0x7f62ffe40e53  (<unknown module>)

Indirect leak of 6156 byte(s) in 1 object(s) allocated from:
    #0 0x558d9f5f9e58 in calloc (/data1/edgar/ddnet/build/DDNet+0x1325e58)
    #1 0x7f62ffe5b8f5  (<unknown module>)

Indirect leak of 5791 byte(s) in 430 object(s) allocated from:
    #0 0x558d9f5f9d1e in malloc (/data1/edgar/ddnet/build/DDNet+0x1325d1e)
    #1 0x7f63135f1824  (<unknown module>)

Indirect leak of 5665 byte(s) in 17 object(s) allocated from:
    #0 0x558d9f5f9d1e in malloc (/data1/edgar/ddnet/build/DDNet+0x1325d1e)
    #1 0x7f63000a1377  (<unknown module>)

Indirect leak of 752 byte(s) in 6 object(s) allocated from:
    #0 0x558d9f5f9fe5 in realloc (/data1/edgar/ddnet/build/DDNet+0x1325fe5)
    #1 0x7f63135f28e6  (<unknown module>)

Indirect leak of 409 byte(s) in 2 object(s) allocated from:
    #0 0x558d9f5f9fe5 in realloc (/data1/edgar/ddnet/build/DDNet+0x1325fe5)
    #1 0x7f631e9d10ac  (/usr/lib64/libdbus-1.so.3+0x420ac)

Indirect leak of 204 byte(s) in 1 object(s) allocated from:
    #0 0x558d9f5f9e58 in calloc (/data1/edgar/ddnet/build/DDNet+0x1325e58)
    #1 0x7f62ffe5b469  (<unknown module>)

Indirect leak of 40 byte(s) in 1 object(s) allocated from:
    #0 0x558d9f5f9d1e in malloc (/data1/edgar/ddnet/build/DDNet+0x1325d1e)
    #1 0x7f62ffe4d5e0  (<unknown module>)

SUMMARY: AddressSanitizer: 274027 byte(s) leaked in 1255 allocation(s).

08:42

leaks leaks

08:48

Tracer caught signal 11: addr=0x708 pc=0x55589917c477 sp=0x7f5b38bfed00 ==25949==LeakSanitizer has encountered a fatal error. ==25949==HINT: For debugging, try setting environment variable LSAN_OPTIONS=verbosity=1:log_threads=1 ==25949==HINT: LeakSanitizer does not work under ptrace (strace, gdb, etc)

08:48

odd

08:48

running the pr for font index

08:49

im not using gdb lol

08:49

WARNING: found 3 unrecognized flag(s): print_stacktrace halt_on_errors halt_on_errors

08:52

seems like a asan error

08:54

@Robyt3

Image attachment

08:54

seems like the text width isnt calculated nicely

08:54

for the bg

github-merge-queue[bot]

[ddnet:master] 2 new commits

c5e27b2 Delete fishy-looking assignment operator/copy constructor of IJob - heinrich5991 862929e Merge pull request #6957 from heinrich5991/pr_ddnet_rm_copyassign_ijob - edg-l

08:54

[ddnet/ddnet] Pull request closed: #6957 Delete fishy-looking assignment operator/copy constructor of IJob

@Ryozuki what mail server do you use? is it easy to set up?

i used to use postfix + dovecot, with mysql but nowadays i would use sqlite

08:57

but i dont self host since i had issues

08:57

with the wiki emails for chinese people

I wanna self host because I only send mails to myself

Selfhosting mail is fun to set up, but sadly big corpo mail hosters just filter all 3rd parties now

08:59

Cartel behaviour

heinrich5991

I wanna self host because I only send mails to myself

postfix + dovecot aint bad but if u never setup a email server maybe is a bit

And they will never tell you why they filtered your mail, so you can’t even fix it

I'm just sending mails to myself

postfix for smtp, dovecot for imap

Can’t go wrong with postfix dovecot combo

if u use unix users its easier

09:01

if u want virtual users u need more setup

09:01

opendkim if u want dkim

I want to not receive emails from the oustide

09:01

but send emails to the outside

postfix then

09:02

https://www.digitalocean.com/community/tutorials/how-to-install-and-configure-postfix-as-a-send-only-smtp-server-on-ubuntu-22-04

How To Install and Configure Postfix as a Send-Only SMTP Server on ...

Postfix is a mail transfer agent (MTA), an application used to send and receive email. It can be configured so that it can be used to send emails by local ap…

thanks

Still need the dovecot for imap, but yeah you can have postfix sendonly

@heinrich5991 beware cloudflare doesnt protect MAIL dns records

09:03

i forgot the name of them

heinrich5991.de isn't behind cloudflare

ok

09:03

just sayin

MX

09:03

btw, I think

ye

postfix on its own will not run a server for receiving mail from the outside? no matter how I misconfigure it?

i think the key is

09:05

inet_interfaces = loopback-only

09:06

hmmm

that sounds like also restricting what I send (edited)

inet_interfaces (default: all) The local network interface addresses that this mail system receives mail on.

09:08

it doesnt say that here

It only restricts where you send from

https://www.postfix.org/postconf.5.html#inet_interfaces

Learath2

It only restricts where you send from

you both say conflicitng things

i say what the doc says

That setting will make the smtp server only bind to loopback. So it’ll effectively disable inbound mail from outside, but it’ll also stop outbound mail from outside

Learath2

That setting will make the smtp server only bind to loopback. So it’ll effectively disable inbound mail from outside, but it’ll also stop outbound mail from outside

so it'll open another socket for sending mail to the outside?

i think he means u cant connect to the smtp server with ur mail client and send a email?

09:11

but not sure

09:11

there is also this https://www.postfix.org/postconf.5.html#smtp_bind_address

09:12

idk all tuts point to setting that

09:12

loopback

Ryozuki

i think he means u cant connect to the smtp server with ur mail client and send a email?

Yes, this is what I’m saying. Atleast that’s how I remember it

seems to work. thanks @Ryozuki @Learath2

https://github.com/ddnet/ddnet/pull/1264 https://github.com/ddnet/ddnet/pull/3954

Fix (partially) skipping through teles and death tiles by heinrich5...

Previously you could partially skip through teleporters: Line two teleporters up in a line, if you through both of them in a single tick, first, the game would teleport you to the first teleporter&...

Fix the possibility of tees stacking by joining teams by heinrich59...

This would make some startline skips harder to exploit. Checklist Tested the change ingame Provided screenshots if it is a visual change Tested in combination with possibly related configuratio...

09:21

we need to think about if we want to merge those

09:21

@Learath2 what u say

09:21

i would merge them, but idk xd

09:22

well i reviewed 2 prs now

09:22

gg

09:22

been long since i did ddnet stuff xd

09:23

Image attachment

09:23

22mins

github-merge-queue[bot]

[ddnet:master] 2 new commits

6f03aab Use std::shared_ptr for editor resources, fix crashes with brush - Robyt3 b909cfa Merge pull request #6943 from Robyt3/Editor-shared_ptr-Refactoring - edg-l

09:36

[ddnet/ddnet] Pull request closed: #6943 Use std::shared_ptr for editor resources, fix crashes with brush

09:36

[ddnet/ddnet] Issue closed: #6935 Editor crashes when creating new map while holding a brush of a tile layer that has a texture

github-merge-queue[bot]

[ddnet:master] 3 new commits

85566fb Add separate log levels for each output - heinrich5991 ee21fde Make default server start a little less noisy - heinrich5991 27e846f Merge pull request #5434 from heinrich5991/pr_ddnet_loglevels - edg-l

09:46

[ddnet/ddnet] Pull request closed: #5434 Add separate log levels for each output

github-merge-queue[bot]

[ddnet:master] 2 new commits

d682d24 Fix stack overflow in destructor when dnsbl object is kept too long alive - Zwelf 406bc26 Merge pull request #6958 from Zwelf/pr-fix-crash-dnsbl - heinrich5991

10:01

[ddnet/ddnet] Pull request closed: #6958 Fix stack overflow in destructor when dnsbl object is kept too long alive

10:01

[ddnet/ddnet] Issue closed: #6954 Crash in DNSBL on RUS

@Jupstar ✪ have you used lavapipe before?

Ryozuki

@Learath2 what u say

I say I'm scared

Learath2

@Jupstar ✪ have you used lavapipe before?

Yes

10:31

Even on windows

Jupstar ✪

Yes

Do you know if I can somehow spoof VkPhysicalDeviceType it returns?

10:32

This game refuses to use a device with type VK_PHYSICAL_DEVICE_TYPE_CPU

Yes it only uses it as last resort

10:32

Or by manually select in graphics settings

I'm forcing it properly, the game just refuses to use it

10:33

[MainThread     ]             vulkan.cpp:2442   ERR| The only available GPU is of type VK_PHYSICAL_DEVICE_TYPE_CPU. Early exiting

10:33

I guess I'll need to compile my own copy of lavapipe or mesa :/

If you even talking about ddnet xd

Learath2

[MainThread     ]             vulkan.cpp:2442   ERR| The only available GPU is of type VK_PHYSICAL_DEVICE_TYPE_CPU. Early exiting

Lol rip

Linux gaming is still not there ngl, this optimus setup is PAIN

10:36

I wonder if I can sneak in with frida and spoof the device type

Optimus is simply trash xd

Oh, I definitely did not know this simple fact. I'll get a better laptop next time

I hope u don't need more than 20fps in your game

10:38

Lavapipe isn't really fast

Jupstar ✪

Lavapipe isn't really fast

I just need to check if it runs

LOL, it does kinda run with DXVK -> Lavapipe

Mh, idk this is far more trouble than it's worth. I can't get ddnet running on the dgpu either

11:17

I can see how someone might just think linux gaming is bad because it doesn't work on their laptop

well i blame laptops being overly propietary

11:23

just dont get smth with optimus

Is also the question how old your laptop is. BCS normally a Intel GPU also offers vk

ChillerDragon: where can I download your client?

Ok, I got it working, now I no longer have time to play, but still

xd

12:23

today i used raw ptrs on rust

12:23

sad day

12:23

Jupstar ✪

Is also the question how old your laptop is. BCS normally a Intel GPU also offers vk

Haswell seems to have incomplete vk support, just couldn't get the game to run with it

12:24

and both it and my dgpu only have Vk1.2 but DXVK needs Vk1.3, so I couldn't use the dx version of the game either

Ryozuki

today i used raw ptrs on rust

what for?

to pass values to the MLIR JIT,

12:25

function call

nvm apparently I only got ddnet working

12:26

I'm about to start crying

cant u return the laptop

12:26

u will be in pains

Ryozuki

cant u return the laptop

It's an old laptop of mine, perfectly strong enough to casually game on

12:27

It's 10 years old but not like I'm trying to play god of war, I just wanted to play a bit of Hades

12:28

Though at the very least I'm happy to report ddnet works fine on either gpu

xd

12:31

cant it run opengl tho

For ddnet all of our backends run fine on either gpu

jupstar pro

This whole experience is very meh, how do you even do this on non optimus laptops? do you just run the dgpu all the time for everything?

12:38

The whole graphics stack is SOOO convoluted, especially when you introduce wine in the middle, I still have no idea where what goes

get a linux friendly laptop

Like for some reason the nvidia vk layer that forces their gpu does a GLX command, but why? What does GLX have to do with VK?

Ryozuki

get a linux friendly laptop

It's a T440p, I think it's pretty well supported

2

Image attachment

13:39

quality software

when codecov 100%

no u

13:48

@Jupstar ✪ do you have any idea why MESA has MESA_VK_DEVICE_SELECT? I was under the impression that mesa wasn't even supposed to be in the path when using a physical device

Learath2

@Jupstar ✪ do you have any idea why MESA has MESA_VK_DEVICE_SELECT? I was under the impression that mesa wasn't even supposed to be in the path when using a physical device

Maybe it just changes the order the devices are enumerated

But is mesa even in the path to do something like that? I thought we only used mesa when doing software rendering

I guess for Mesa drivers only. Because yeah normally it's job of the Vulkan loader

13:52

But Mesa is basically a default implementation for all drivers.. which u customize to your need or accelerate to what is supported hardware wise

13:53

That's also why all these VK drivers pop up lately

13:53

They don't have to implement a lot to have a working VK driver .. it's just not very optimized ^^

13:55

"Vulkan Runtime The Vulkan runtime and utility code in Mesa provides a powerful shared core for building Vulkan drivers. It’s a collection of base structures (think base classes in OOO) which allow us to implement a bunch of the annoying hardware-agnostic bits in common code. "

13:55

https://docs.mesa3d.org/vulkan/index.html

13:55

Go ahead built yet another driver xd

Learath2

But is mesa even in the path to do something like that? I thought we only used mesa when doing software rendering

mesa is a vulkan driver impl

13:56

oh

Ryozuki

mesa is a vulkan driver impl

It's quite hard to even understand what a "vulkan driver impl" is, it's not a kernel thing, is it an x driver?

like a opengl driver

13:57

xd

Xd

Jupstar ✪

But Mesa is basically a default implementation for all drivers.. which u customize to your need or accelerate to what is supported hardware wise

Hm, do all drivers go through MESA? I think nvidia and amd ship their own libGLs that have nothing to do with MESA, no?

Learath2

Hm, do all drivers go through MESA? I think nvidia and amd ship their own libGLs that have nothing to do with MESA, no?

OpenGL is bit different.. for Vulkan it's completely different anyway

13:58

Vulkan doesn't need to know about x or Wayland etc

13:58

Khronos gives surface extensions for these things

13:58

Core Vulkan theoretically doesn't work on surfaces

Focusing on VK, where does MESA come in then?

Mesa is the llvm for GPU stuff xd

That sounds wrong, llvm compiles stuff, mesa seems to provide implementations

I guess u could say it's just a standard to unify the drivers

It seems to me like mesa is directly the implementation for opengl/vk, but can also do hardware acceleration, for which it has drivers. Does that sound about right?

Learath2

It seems to me like mesa is directly the implementation for opengl/vk, but can also do hardware acceleration, for which it has drivers. Does that sound about right?

I think on it's own it doesn't directly do hardware acceleration.. it's basically the interface for that

14:03

But for Vulkan it's not need at all for Linux

14:03

BCS VK was designed to overcome these design flaws

14:03

There might be Mesa only VK extensions

14:04

But that doesn't imply it's needed for most stuff

14:04

OpenGL is more tricky BCS on how it's built around surfaces and stuff like vsync

14:04

Which isn't in opengls hand

Learath2

@Jupstar ✪ do you have any idea why MESA has MESA_VK_DEVICE_SELECT? I was under the impression that mesa wasn't even supposed to be in the path when using a physical device

So to your question. It will probably only work for drivers built on Mesa

14:06

Amdvlk will ignore it

So MESA in the case of VK would be just another vulkan driver?

The loader most probably too

Learath2

So MESA in the case of VK would be just another vulkan driver?

Well there are multiple mesa drivers

14:07

Radv would be one

14:07

Lavapipe another

but it seems the nvidia one has nothing to do with mesa, so some vulkan drivers use mesa, others don't?

Learath2

but it seems the nvidia one has nothing to do with mesa, so some vulkan drivers use mesa, others don't?

Yes

14:08

At least we don't know^^

I see, quite confusing

What even confuses you

Everyone quite loosely calls things "drivers", I'm not sure where what plugs. The drivers I know plug into the kernel DRM

Basically in Vulkan u never cominucate with anything that isn't Vulkan itself

14:10

Only exception is the surface which gives u a Vulkan texture

X seems to have drivers, Vulkan seems to have drivers, Mesa seems to have drivers

14:11

Nvidia seems to call it's vulkan driver libGLX

i am a rust driver

14:11

and soon car driver

X might be what ud can the surface extensions

14:11

It has to give u space to draw on

14:12

But it's not vks job to care about how it works . VK just expects it to work with Vulkan core

Who provides the vulkan core?

but yeah if u want to know how the user space communicates with kernel

14:13

Nvidia probably own kernel modules

14:13

Anything. Else drm

Even nvidia hooks into the drm nowadays

Learath2

Who provides the vulkan core?

Nobody directly

14:14

U can code your own

So as far as vk is concerned, x and wayland doesn't exist

Yes. VK first of all is the API to communicate with the gpu

14:15

That the GPU has a surface is an extension

14:15

Just like in real life xd

14:15

Optimus xdd

This is a nice image

Image attachment

14:28

u know what haunts me

14:28

(signal: 11, SIGSEGV: invalid memory reference)

14:28

justatest

use rust

(edited)

14:28

hmm

14:29

the https://www.rust-lang.org/ homepage no longer says "no segfaults"

A language empowering everyone to build reliable and efficient software.

Ryozuki

we need to think about if we want to merge those

both are physics change though...

heinrich5991

use rust

(edited)

ye im using rust

14:29

its because i do lot of unsafe

14:30

cuz i depend on software made in c++

14:30

ur lovely llvm

@Voxel when does ddnet get epic logo like VK xd

14:30

Embedded image

14:30

He already has a hammer

Image attachment

nvidia's vulkan driver is showed into their GLX implementation

14:30

shoved*

Image attachment

14:31

we need more logos

I don't even know what makes Nvidia drivers always be trouble on Linux.. I guess some specific extensions desktop environment have to use

Ryozuki

Click to see attachment 🖼️

Epic

Note that although both the GLVND and non-GLVND GLX client libraries share the same SONAME of libGL.so.1, only one of them at a time may be installed at a time. /usr/lib/libGL.so.367.57 is the non-GLVND GLX client library, and /usr/lib/libGL.so.1.0.0 is the GLVND GLX client library. This library is also used as the Vulkan ICD.

That logo for video rendering xd

This confused me so much today

Learath2

Note that although both the GLVND and non-GLVND GLX client libraries share the same SONAME of libGL.so.1, only one of them at a time may be installed at a time. /usr/lib/libGL.so.367.57 is the non-GLVND GLX client library, and /usr/lib/libGL.so.1.0.0 is the GLVND GLX client library. This library is also used as the Vulkan ICD.

OpenGL is mess anyway xd

Something that is supposed to have nothing to do with opengl being inside an opengl library

Not because opengl spec.. but everything around it

I think I have one last unresolved question, you said surfaces are extensions, where can I read about them?

Khr surface extension

https://registry.khronos.org/vulkan/specs/1.3-extensions/man/html/VK_KHR_surface.html

It's named differently per od

14:34

Os

The VK_KHR_surface extension is an instance extension. It introduces VkSurfaceKHR objects, which abstract native platform surface or window objects for use with Vulkan. It also provides a way to determine whether a queue family in a physical device supports presenting to particular surface. Separate extensions for each platform provide the mechanisms for creating VkSurfaceKHR objects, but once created they may be used in this and other platform-independent extensions, in particular the VK_KHR_swapchain extension.

14:35

https://github.com/KhronosGroup/Vulkan-Tools/blob/main/cube/cube.c

Vulkan-Tools/cube/cube.c at main · KhronosGroup/Vulkan-Tools

Vulkan Utilities and Tools. Contribute to KhronosGroup/Vulkan-Tools development by creating an account on GitHub.

14:35

c code

14:35

for u

Ryo the better Google

im chatgpt5-rust

I think I understand what the stack looks like for vulkan now, one day I'll spend some time figuring out what it looks like for gl

It's not entirely different. Just that it's not really standarized

14:38

And that makes it a mess

Well in that case it seems the X server is still in the way unless DRI is involved

In OpenGL u have glx being the loader.. or egl

14:39

Or Nvidia's thing

14:39

And all need to be implemented by e.g. sdl2

14:39

Vulkan mostly doesn't need sdl at all

Jupstar ✪

@Voxel when does ddnet get epic logo like VK xd

i dunno

Or better: u need it less, because more stuff is standardized xd

time for the 2nd coffee

What does glvnd do?

I think it's just vendor neutral loader

14:42

So basically OpenGL stub(?)

14:42

I'm not 100% sure but that's how I imagine it

14:43

So just that the shared object isn't vendor at a specific location

14:43

At a vendor

14:44

Anyway. This is simply a mess. I have no idea what the reasoning behind all this was.. it's probably like that before I and u were even born

14:44

Lmao

Jupstar ✪

Anyway. This is simply a mess. I have no idea what the reasoning behind all this was.. it's probably like that before I and u were even born

I know why there is a mess, it just developed over time rather than being designed properly

14:45

All GL calls used to go through X for a while there before DRI was a thing

But dri is dead isn't it

14:46

I could swear I compile mesa without it nowadays xd

Yep, it also died

14:47

Actually the open source drivers might still be using it

Let's see when Vulkan hits it's limit and we need yet another API xd

Learath2 and Jupstar are the same person

14:48

they're always confusing its and it's

heinrich5991

Learath2 and Jupstar are the same person

It'

I could imagine one problem is maybe the static pipeline stuff

Jupstar ✪

Let's see when Vulkan hits it's limit and we need yet another API xd

its*

Jupstar ✪

Let's see when Vulkan hits it's limit and we need yet another API xd

Blame my keyboard xd

It's my 5th personality trained only on graphics

I just swipe

just say its

14:49

and when someone says

14:49

its wrong

14:49

just say

14:49

im lazy to type the '

14:49

so ur always right

14:49

here i should have said "u r"

14:49

xd

Your

14:50

That's what my keyboard said

14:50

Your wife

14:50

You're wrong

14:50

Xd

https://nim-lang.org/blog/2023/08/01/nim-v20-released.html

Nim v2.0 released

The Nim team is proud and happy to announce Nim version 2.0.

14:51

NIMROCKS

14:51

Yet another useful language

14:52

Python like

https://news.ycombinator.com/item?id=36955934

rlprompt

Nvidia’s CEO Is the Uncle of AMD’s CEO

14:52

lmao

I always wondered xr

14:52

Xd

https://arxiv.org/abs/2307.16802

Semiconducting transport in Pb$_{10-x}$Cu$_x$(PO$_4$)$_6$O sintered...

The very recent claim on the discovery of ambient-pressure room-temperature superconductivity in modified lead-apatite has immediately excited sensational attention in the entire society, which is fabricated by sintering lanarkite (Pb2SO5) and copper(I) phosphide (Cu$_3$P). To verify this exciting claim, we have successfully synthesized Pb$_2$SO...

14:53

https://news.ycombinator.com/item?id=36951140

spekcular

Semiconducting Transport in LK99 reproduction attempt

14:54

For the non-chemists/physicists:

>

Recently[1], there was an exciting claim that a new type of material could conduct electricity perfectly at room temperature (this is what is meant by "superconductivity")[4]. This material was a version of lead-apatite [2], a type of mineral, that was altered with certain additions and made by combining two other materials, lanarkite and copper phosphide.

>

The researchers writing this report wanted to check if this claim was true. So, they made the same types of materials (lanarkite, copper phosphide, and the altered lead-apatite) and tested how they conducted electricity and reacted to magnets.

>

What they found was that lanarkite (Pb2SO5) didn't conduct electricity well at all, and copper phosphide Cu3P) conducted electricity similar to a semiconductor [3]. The altered lead-apatite, which was supposed to be the superconductor, behaved more like a semiconductor (a type of material that can sometimes conduct electricity, depending on conditions).

>

Also, a key property of superconductors is that they repel magnets. But when the researchers put a magnet near their lead-apatite, there was no repulsion.

>

Because of these tests, they're suggesting that the original claim about this room-temperature superconductor should be re-examined more carefully. It doesn't seem to behave like a superconductor in their tests.

14:55

https://en.wikipedia.org/wiki/LK-99

LK-99 is a potential room-temperature superconductor with a gray‒black appearance.: 8 It has a hexagonal structure slightly modified from lead‒apatite, by introducing small amounts of copper. The material was first discovered and manufactured by a team of researchers including Sukbae Lee (이석배), and Ji-Hoon Kim (김지훈) from Korea University (고려대학교...

14:55

lol

I hear these news every other week

the comments on hacker news says everyone needs to chill a little

14:56

while this is examined

14:57

basically the opposite of what ryo did :p

i mean

14:58

i just shared

14:58

xd

yes

14:58

precisely

It seems AMDGPU does use DRI3, so many things, so many apis, so many interfaces aaaaa

Learath2

It seems AMDGPU does use DRI3, so many things, so many apis, so many interfaces aaaaa

But u usually use the gallium stuff

15:00

https://upload.wikimedia.org/wikipedia/commons/c/cf/Gallium3D_vs_DRI_graphics_driver_model.svg

If it wasn't clear why this is a big deal, if successful LK-99 would be a watershed moment for humanity easily on-par with invention of the transistor.

15:03

https://twitter.com/Andercot/status/1686286684424691712

Andrew Cote (@Andercot)

First claimed successful replication of LK-99 Accomplished by a team at the Huazhong University of Science and Technology and posted 30 minutes ago. Why this is evidence: The LK-99 flake slightly levitates for both orientations of the magnetic field, meaning it is not simply a…

Likes

8404

Retweets

2586

Jupstar ✪

But u usually use the gallium stuff

Isn't gallium3d how you are supposed to write the mesa drivers but dri still the name of the interface? pepeW

pepeW

Ryozuki

If it wasn't clear why this is a big deal, if successful LK-99 would be a watershed moment for humanity easily on-par with invention of the transistor.

I hope we get maglevs

Learath2

Isn't gallium3d how you are supposed to write the mesa drivers but dri still the name of the interface? pepeW

pepeW

Gallium is all in one I think xd

AAAAAAAAAAA

pepeW

pepeW

https://blogs.igalia.com/itoral/2014/07/29/a-brief-introduction-to-the-linux-graphics-stack/ found a decent looking writeup, I think I'll just read someone elses research as researching these on your own is pepeW

pepeW

Maybe I should start being mesa dev

15:11

Then you have trusted source

Ryozuki

For the non-chemists/physicists:

>

Recently[1], there was an exciting claim that a new type of material could conduct electricity perfectly at room temperature (this is what is meant by "superconductivity")[4]. This material was a version of lead-apatite [2], a type of mineral, that was altered with certain additions and made by combining two other materials, lanarkite and copper phosphide.

>

The researchers writing this report wanted to check if this claim was true. So, they made the same types of materials (lanarkite, copper phosphide, and the altered lead-apatite) and tested how they conducted electricity and reacted to magnets.

>

What they found was that lanarkite (Pb2SO5) didn't conduct electricity well at all, and copper phosphide Cu3P) conducted electricity similar to a semiconductor [3]. The altered lead-apatite, which was supposed to be the superconductor, behaved more like a semiconductor (a type of material that can sometimes conduct electricity, depending on conditions).

>

Also, a key property of superconductors is that they repel magnets. But when the researchers put a magnet near their lead-apatite, there was no repulsion.

>

Because of these tests, they're suggesting that the original claim about this room-temperature superconductor should be re-examined more carefully. It doesn't seem to behave like a superconductor in their tests.

well, not really surprising

Jupstar ✪

Maybe I should start being mesa dev

Become mesa dev, convince them to drop support for everything but the last gen, then completely remake the entire stack so it's clean

Good idea. Probably works just as well as here in ddnet xd

Just too many things in the stack, they should reduce it all to one library exposing one function drawTriangle

15:17

I'll also allow the DRM to exist

The libraries are over complicated BCS they always have to present the idea of the render pipeline

15:17

In theory u only need compute shaders and nothing else

arent the graphics pipeline, specialized more optimized compute shaders?

15:18

cuz they are more limited

Yes

15:18

Basically

but the limitation allows them to be faster

If hardware doesn't hit its limits soon I could imagine this is also just a matter of time

15:18

Until u have actual full control over the full pipeline

Ryozuki

but the limitation allows them to be faster

Yep.. but u could say that about anything in programming

15:20

A pure 2d pipeline with constraints directly for ddnet tile rendering would probs make it faster xd

15:20

We need that xd

ehm, hi :D recently ive been doing something useless with demos and i came to chunk_header part. it says there're 3 types. i think i found something about snapshot and snapshot delta. but what to do when its message type :p

Image attachment

popcornMoai_pb

Image attachment

MilkeeyCat

ehm, hi :D recently ive been doing something useless with demos and i came to chunk_header part. it says there're 3 types. i think i found something about snapshot and snapshot delta. but what to do when its message type :p

Um, afaik they are just netmsgs, up to you what you do with them

i mean how to get info from that thing (edited)

Yek

popcornMoai_pb

Fine for me

MilkeeyCat

i mean how to get info from that thing (edited)

It's just a CNetChunk you'll need to parse it as such

ok, thanks =]

Oh wait no, it's not, it's the data in one directly

with weird padding at the end

15:57

after int compression or so

15:57

it was realllly weird

15:57

I don't remember

15:57

what are you doing @MilkeeyCat?

(edited)

So you can directly implement CUnpacker. After that you can check UnpackMessageId to see how we figure out the message id

heinrich5991

what are you doing @MilkeeyCat?

(edited)

maybe one day it will be web demo viewer but for now i just try to find out how demo files work

ah nice

MilkeeyCat

maybe one day it will be web demo viewer but for now i just try to find out how demo files work

can I see what you're programming, somewhere?

heinrich5991

can I see what you're programming, somewhere?

for now its just few structs :p

Image attachment

uh, rust

16:01

cool

i dont rly know rust but writing it in js rn will be even worse :\

what's bad about JS? ^^

16:02

I heard you can use TS to add types

then i would need to find some library to work with binary or write something my own

16:03

but with deku i can just add few attributes and it works

surely, there's also a library for JS ^^

do you know the name of it? (edited)

but it's fine if you want to choose rust

16:05

no. I wouldn't have known deku either, though

heinrich5991

but it's fine if you want to choose rust

sooner or later if i dont stop it i will have to rewrite in js

random result: https://github.com/jDataView/jDataView

GitHub - jDataView/jDataView: DataView. Extended. Anywhere.

DataView. Extended. Anywhere. Contribute to jDataView/jDataView development by creating an account on GitHub.

16:08

not sure if good

heinrich the js promoter

ill take a look at this, thanks

[ddnet/ddnet] Pull request opened: #6959 Redesign quad/env/quadenv points

Proposal for a new point design to try and make the editor look a bit more modern. I think especially in the envelope editor it already looks a lot cleaner. But Im not sure if the quadpoints are easy enough to see (mostly the green ones). !screenshot_2023-08-01_18-23-27 !screenshot_2023-08-01_18-24-01 ## C...

@AssassinTee hey Is there any progress?

MilkeeyCat

then i would need to find some library to work with binary or write something my own

you might find this useful to start with (ignore readString): https://github.com/spacem/dntviewer/blob/7df9ec57b1aa36d6e611bf7d5d545f3b648d271d/simplerreader.js

sry again for stupid questions, DDNet is not using the libmysqlclient library it seems? atleast its not shown when i compile and run ldd

u need to enable the cmake option

18:30

read the readme

Learath2

Fine for me

Okay_pb

My ubuntu server isn't able to connect to the outside world or to my router anymore (while the router still gave it an ip adress). SSH is working, but i can't ping 8.8.8.8. I can't think of anything anymore causing this, any ideas?

what says ip a

20:11

or ip link

20:11

dont leak ip tho

what about ip route

well ip a wont show public

20:11

i think

20:11

xd

everything pointed correctly?

20:12

or point it to the obvious:have you tried turning it off and on again? systemctl restart networking/service networking restart

network is state UP

20:13

I can't leak my ip this way, only local IPs

20:13

(except maybe ipv6)

can you communicate with local devices?

yes, i am on it with ssh

20:14

(but I can't ping my router)

could be a problem with either default gateway or DNS res

ping 8.8.8.8 shouldn't use DNS

oh right..

default gateway is router, you can check it with ip route

20:16

ip route                                                                                                                                                                                                                         default via 192.168.2.1 dev enp86s0 proto dhcp metric 20100

20:17

I restartet the router, the NetworkManager, the system itself, I really have no idea anymore

safely assuming that other connected devices function properly? maybe some kind of ip adress conflict?

20:18

very unlikely, but since you almost tried everything i would've, idk

After backing it up I resetted even iptables

20:19

Nope, it has a static ip address reserved for it

anything in syslog/messages? (/var/log/syslog - /var/log/messages

nothing I could see easily, anything I should grep for?

the only thing i know out of the blue is grep,ing your network interface in syslog (edited)

20:29

i should start wording things better before pressing enter

it doesn't contain anything supsicious, only infos and some addr changes due to router restarts

20:30

@Ryozuki any ideas?

you can also grep for "dhcp" "route" "icmp" "firewall" and "network" just to make sure

idk

20:31

probs u messed up with dhcp

networkd-dispatcher[826]: WARNING: systemd-networkd is not running, output will be incomplete. Huh?

"systemctl status/start/enable systemd-networkd (edited)

okay I have an idea now what's wrong, wtf ...

keep us updated @AssassinTee

!

I had a few dead virtual networks and removed them now

20:46

but I get the same error from docker and it didn't fix my issue

20:47

And now I am out of ideas again

your server is running in a docker? - if so you could check it with "docker network ls" (edited)

the host machine has the issue, not the docker container

20:49

so, the docker container has the issue too, but just by transitivity

ip route please

could it be something with apparmor/selinux?

output of ping 8.8.8.8 please, too, please @AssassinTee

100% package loss, nothing to output

ip route please

I was already resetting ip rout rn:

Ziel            Router          Genmask         Flags Metric Ref    Use Iface                                                                                                                                                                                                        
default         192.168.2.1     0.0.0.0         UG    0      0        0 enp86s0
192.168.2.1     0.0.0.0         255.255.255.255 UH    0      0        0 enp86s0

try ping 192.168.2.1

is there a way to change the size of weapons on tees and change the spritesheet

heinrich5991

try ping 192.168.2.1

nope, nothing

that sounds like your problem, you can't connect to your supposed router

21:06

what's your local IP address?

21:06

did you obtain it from dhcp?

how do I know if i got it from dhcp?

hmm. do you know what manages the network interface?

21:07

e.g. NetworkManager or systemd-networkd

NetworkManager

heinrich5991

what's your local IP address?

enp86s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether SOMEMAC brd ff:ff:ff:ff:ff:ff
    inet 192.168.2.162/24 brd 192.168.2.255 scope global dynamic noprefixroute enp86s0

do you have this NetworkManager symbol in the tray?

I have the network-connected-symbol, if you mean that (happy not to run headless here)

can you edit the wired connection there by right-clicking?

21:10

it might tell you something about whether dhcp is supposed to be active

isnt it safe to assume dhcp is working? "<BROADCAST,MULTICAST,UP,LOWER_UP>

ӎěłŏƞ

isnt it safe to assume dhcp is working? "<BROADCAST,MULTICAST,UP,LOWER_UP>

might be a statically assigned IP address, no?

could be.. noprefixroute

cyberFighter

is there a way to change the size of weapons on tees and change the spritesheet

if I got that right, you can just resize each weapon with something like Gimp, scale in a way that keeps each weapon at the same center but smaller I have 2 files from before that you can check (add the files to assets/game/ folder in your teeworlds/ddnet folder, then choose them in settings - assets - game) (edited)

21:13

Image attachment

Image attachment

or just modify render.cpp

21:14

if i understood his question with spritesheet correctly

No Prefix Route: noprefixroute - Indicates that there is no specific route defined for this IP address; it will use the default gateway.

I also have noprefixroute on my link

21:22

doesn't seem to be the problem

21:22

did you get anything from the NetworkManager settings?

21:22

if not, could you try checking the NetworkManager logs?

21:22

journalctl -ru NetworkManager (edited)

NetworkManager[751]: <info> [1690924335.6729] manager: NetworkManager state is now CONNECTED_GLOBAL

I'm interested in messages related to dhcp

heinrich5991

did you get anything from the NetworkManager settings?

this didn't work or you didn't try?

I got a warning there over 2 hours ago, restartet the service and everything seems fine

21:31

got lease messages from dhcp4 and dhcp6

sudo iptables -L -n -v

https://www.science.org/content/blog-post/room-temperature-superconductor-new-developments

21:35

https://news.ycombinator.com/item?id=36957678

nneonneo

A room-temperature superconductor? New developments

21:35

500 comments

21:35

we are truly

21:35

in a point of a new era

why's that?

I'm doing this as a follow-on to my previous post on this topic, since this is a fast-changing story. To recap, preprints appeared last week making the remarkable claim of a well-above-room-temperature superconducting material at ambient pressure, dubbed LK-99. This is one of the most sought-after goals in all of materials science and condensed matter physics, something that until now has only been found in (numerous!) science fiction stories. The potential applications of such a material almost can go without saying - depending on what current density it could accommodate, it could improve almost anything that uses electromagnetism.

21:38

, it could improve almost anything that uses electromagnetism.

21:38

timeline

21:38

First claimed successful replication of LK-99 - https://news.ycombinator.com/item?id=36953819 - Aug 1, 2023 (265 comments)

>

LK-99 crystal verified to be magnetically levitated (with video) - https://news.ycombinator.com/item?id=36953396 - Aug 1, 2023 (10 comments)

>

Room-Temperature Ambient-Pressure Superconductor LK-99 preprint revision 2 - https://news.ycombinator.com/item?id=36952894 - Aug 1, 2023 (272 comments)

>

Origin of correlated isolated flat bands in LK99 - https://news.ycombinator.com/item?id=36951815 - Aug 1, 2023 (196 comments)

>

Semiconducting Transport in LK99 reproduction attempt - https://news.ycombinator.com/item?id=36951140 - Aug 1, 2023 (245 comments)

>

LK-99: The live online race for a room-temperature superconductor - https://news.ycombinator.com/item?id=36940323 - July 31, 2023 (619 comments)

>

Ask HN: Room temperature superconductor: what to expect? - https://news.ycombinator.com/item?id=36930707 - July 30, 2023 (61 comments)

>

“We are not cheating” (YH Kwon superconductor comments) - https://news.ycombinator.com/item?id=36929386 - July 30, 2023 (61 comments)

>

Korea Superconductor Papers Published ‘Without Consent’ - https://news.ycombinator.com/item?id=36927216 - July 30, 2023 (47 comments)

>

Argonne National Lab is attempting to replicate LK-99 - https://news.ycombinator.com/item?id=36916254 - July 29, 2023 (201 comments)

>

LK-99 - https://news.ycombinator.com/item?id=36897300 - July 27, 2023 (58 comments)

>

Superconductor news: What’s claimed, and how strong the evidence seems to be - https://news.ycombinator.com/item?id=36881808 - July 26, 2023 (471 comments)

>

The first room-temperature ambient-pressure superconductor? - https://news.ycombinator.com/item?id=36864624 - July 25, 2023 (875 comments)

>

Edit: thanks to mystery user for emailing hn@ycombinator.com to suggest we include the days, not just month/year, for fast-moving stories like this.

AssassinTee

After backing it up I resetted even iptables

@heinrich5991 already did that

AssassinTee

@heinrich5991 already did that

the command lists the iptables, doesn't reset it

yes, but the guide I used does

21:40

-F flag

and you don't want to list your empty iptables?

21:43

@Ryozuki the hype seems real indeed. can this give some concrete examples?

Chain INPUT (policy ACCEPT 27673 packets, 2164K bytes)
 pkts bytes target     prot opt in     out     source               destination

Chain FORWARD (policy DROP 0 packets, 0 bytes)
 pkts bytes target     prot opt in     out     source               destination

Chain OUTPUT (policy ACCEPT 25692 packets, 4621K bytes)
 pkts bytes target     prot opt in     out     source               destination

Chain DOCKER (0 references)
 pkts bytes target     prot opt in     out     source               destination

Chain DOCKER-ISOLATION-STAGE-1 (0 references)
 pkts bytes target     prot opt in     out     source               destination

Chain DOCKER-ISOLATION-STAGE-2 (0 references)
 pkts bytes target     prot opt in     out     source               destination

Chain DOCKER-USER (0 references)
 pkts bytes target     prot opt in     out     source               destination

Chain f2b-sshd (0 references)
 pkts bytes target     prot opt in     out     source               destination

21:45

that FORWARD line looks wwrong

heinrich5991

@Ryozuki the hype seems real indeed. can this give some concrete examples?

idk rn

21:45

lets see how it goes xd

21:45

ima go sleep now

good night

u2

AssassinTee

that FORWARD line looks wwrong

but it also says 0 packets, so maybe it's not bad

21:45

can you tell me what the output of curl -i http://1.1.1.1/ is?

I am expecting a timeout

21:48

so nothing

@AssassinTee what are you trying to fix

21:53

forward chain won't be reached if you just send or receive packets that won't be forwarded (edited)

@ReiTW The internet on my server isn't working, while everything seems to be configured right and the network has internet and I can ssh into the machine

so locally everything is fine but not outside

21:55

did u try to ping ur router from ur server

I can't tell where the issue is, everything looks like I should be able to connect to the internet with it, but I can't even ping 8.8.8.8

21:56

yes I did

do you have traceroute installed

can't ping the router, but get an IP adress?

ah

21:56

what does ip a shows

no, and I can't install it because I have no internet

put the full output here

21:57

ah u did above

ip a shows my network being up, ip link shows my gateway, iptables is now empty/cleaned

careful with iptables there's multiple tables

21:58

but if you didn't touch anything should be empty

no i had to reset it

I meant

21:58

raw, mangle, nat, etc..

21:59

by default when you do iptables -nL you are looking at the filter table: iptables -t filter -nL

21:59

each table has its use case, but I liked to use mangle table as it was "executed" before filter table, so packets were dropped before

22:01

@AssassinTee can you show your netplan config file?

mangle table looks clean like baby skin

if you didn't touch any of those, they'll be empty

22:02

also when you fix your bug restart docker as it adds iptables rules

ReiTW

@AssassinTee can you show your netplan config file?

system uses the NetworkManager

lol do you have a gui

yes, the server has a usecase for it

heinrich5991

might be a statically assigned IP address, no?

dynamic means it obtained an IP through dhcp, noprefixroute afaik is for ipv6 but ipv6 sux

nvm dunno for what it is

It just randomly started working again

22:20

(╯°□°)╯︵ ┻━┻

All i know from investigating the logs is some server produced a segfault

hus3h

if I got that right, you can just resize each weapon with something like Gimp, scale in a way that keeps each weapon at the same center but smaller I have 2 files from before that you can check (add the files to assets/game/ folder in your teeworlds/ddnet folder, then choose them in settings - assets - game) (edited)

im talking way bigger than usual

23:06

no size limit

[ddnet/ddnet] Issue opened: #6960 Hammer while in super is sometimes uneffective

Not sure what caused this, but hammering frozen tees while in super and being in freeze too is uneffective. It does trigger single tick unfreeze but won't move the tee. Video here: https://discord.com/channels/252358080522747904/757720336274948198/1136083810124046356

lol, discord preview directly links the post

Exported 565 message(s)