DDraceNetwork - developer

00:00

god damn

00:00

delish

looks like big meatball

its rib eye

rib meatball

why u show meat but no inside

00:24

gotta show the inside

You also need to stop shaking your phone fr pepeW

xd

00:29

bad phone

Ewan

why u show meat but no inside

i got one but

00:30

its shaky

00:30

delete him,...

:c

00:31

i also got some whiskey mixed thing

00:31

@Davide I don't think this ddos protection thing is working too well

Learath2

@Davide I don't think this ddos protection thing is working too well

What happened?

Davide

What happened?

idk it gets through

Saw it, should be fixed

00:52

It was a 1.1Tbps attack

00:52

Seems like the load balancing was not working correctly and one of the uplinks got saturated

Found the reason of the incorrect load balancing

01:00

There's an ongoing maintenance on my main transit provider

01:00

@Learath2

01:01

So my current uplinks capacity is just half as the normal, until the maintenance ends

love you guys btw

1

01:23

Davide

It was a 1.1Tbps attack

Unlimited vps I guess with such attack

Davide

So my current uplinks capacity is just half as the normal, until the maintenance ends

Ah, that makes sense

Davide

It was a 1.1Tbps attack

Im still wondering who pays for this HUH

meloƞ

Im still wondering who pays for this HUH

it's far cheaper than a connection that large

TsFreddie

love you guys btw

we love you too freddie

1

<ChillerDragon> ryo living the life

03:15

<ChillerDragon> hope your meatball is vegan.

I Hope so too HUH

<ChillerDragon> love you too @TsFreddie owo

Learath2

it's far cheaper than a connection that large

imagine wasting money to ddos a community based game

ws-client

<ChillerDragon> hope your meatball is vegan.

I hope his meatballs turn onto rust

04:36

04:36

out of context bs ^

ddnet-maps

[ddnet-maps:master] 1 new commit

b3cb10d A EasyRight - ddnet-maps

<ChillerDragon> @Cellegen which games do you suggest for ddosing? c:

06:28

I dont think I wanna answer

murpii

[ddnet-web:master] 1 new commit

a96913a Update Staff - murpii

TsFreddie

love you guys btw

No homo, love you too ❣️

morning

1

Cellegen

imagine wasting money to ddos a community based game

Some "people" are just disgusting wastes of oxygen

Chill

erDragon?

NoChill

Learath is too hot to chill

Learath is cool af

f3

So he chill

11:16

11:18

@Robyt3 Thursday for Vulkan. What vk question is on your heart today

i looked into how MIR works in Rust

11:36

its a CFG

11:37

rly close to what llvm ir looks like

11:37

which makes sense xd

11:38

@Jupstar ✪ my vulkan question: can you have security vulnerabilities / injections via layers in vulkan

11:38

if it makes sense xd

11:38

ok actually a layer is smth the driver implements right

11:38

nvm

11:39

maybe a better question, can u do privilege escalation via vulkan

11:42

this looks like a lot but its barely 20€ xd

Jupstar ✪

@Robyt3 Thursday for Vulkan. What vk question is on your heart today

what is the limiting factor to the amount of shaders you can upload?

i would guess memory or whathever int max value the driver uses (edited)

Ryozuki

maybe a better question, can u do privilege escalation via vulkan

It's in the user land, so only if the back end has a bug I guess..

Patiga

what is the limiting factor to the amount of shaders you can upload?

There is a limit? Xd

idk :d

Ryozuki

@Jupstar ✪ my vulkan question: can you have security vulnerabilities / injections via layers in vulkan

BTW this isn't even a bad question. Because layers must not be inside the driver. But without root rights u probably can't install them

11:45

But overlays can defs contain bugs

Patiga

idk :d

Well then it's mostly the amount of memory I guess. Ofc for stuff inside the shaders there are individual limitations too. Amount of uniform, amount of textures per stage etc.

11:47

If it makes you happy to know. 4000 shaders are not uncommon even for mobile games

since you do most memory management in vulkan by yourself, do you also need to manage where the shaders are placed?

11:48

they should be located in v-ram, no?

dont u make ur own memory allocators for everything in vulkan?

meanwhile unity compiling 30k shaders if you are not careful when using URP Lit

Patiga

since you do most memory management in vulkan by yourself, do you also need to manage where the shaders are placed?

You don't have any influence about that

1

@Jupstar ✪ if i break my computer with i hammer i have influence in that it wont be placed anywhere :)

Memory management is still more opaque than writing asm for cpus

2

11:49

asm for cpus is crystal clear isnt it

11:49

there is no amiguity

11:49

ambiguity

i cooked our game all day today

11:50

everything needed baking its so slow and stressful

was it tasty?

at least we saw gameover the first time today

xd

it had been crashing before we could finish a round for the first two play test sessions

@TsFreddie the company where i work is also into making games now

11:51

they use unity :d but the backend is in elixir

wdym into

to usse the erlang otp

11:52

by into i mean they making games too

oof

They use edlang

honestly i think i hate making games now

xd

11:53

im not in that department so

11:53

im making a compiler

11:53

more interesting

i used to think making gamn is just putting things in place and make them move

but in the office i can see

11:53

and talked to some artists xd

11:53

3d modelers

@Patiga for the host side of the vk driver u actually can pass custom allocators. So you can influence that. I wonder if it's ever used anywhere. I guess no device that implements vulkan doesn't also have a heap xd

apparently you need to make the exact correct design decisions for every single objects. and it is too hard to teach all designers and artists about the technical detail in such a short time

huh interesting

and we just let them do their thing

11:55

and remake the entire map again on our end

xd

and they came back with a map like this

@TsFreddie i saw u starred my lalrpop-json :o

TsFreddie

and they came back with a map like this

xd

Ryozuki

@TsFreddie i saw u starred my lalrpop-json :o

misclicked

jk

12:01

although no idea what it is. its just that i rarely check the linetime thing. and clicking stuff make me feel like i was participating

12:01

so i participated

@Learath2 i have something u may like

12:03

https://github.com/dtolnay/inventory

GitHub - dtolnay/inventory: Typed distributed plugin registration

Typed distributed plugin registration. Contribute to dtolnay/inventory development by creating an account on GitHub.

that's not yours

Examples Suppose we are writing a command line flags library and want to allow any source file in the application to register command line flags that are relevant to it.

12:03

remembered u wanted smth like this for discord commands (edited)

12:04

@Jupstar ✪ https://zed.dev/features

Zed - Code at the speed of thought

Zed is a high-performance, multiplayer code editor from the creators of Atom and Tree-sitter.

12:05

https://github.com/zed-industries/zed

GitHub - zed-industries/zed: Code at the speed of thought – Zed is ...

Code at the speed of thought – Zed is a high-performance, multiplayer code editor from the creators of Atom and Tree-sitter. - GitHub - zed-industries/zed: Code at the speed of thought – Zed is a h...

12:05

@Jupstar ✪ omg a new editor in rust

Jupstar ✪

@Robyt3 Thursday for Vulkan. What vk question is on your heart today

Is it worth it to fix the validation warnings? With dbg_gfx 4 you get like 4 MiB of validation warnings, mostly about us not freeing everything I guess

@Jupstar ✪ bro this editor is legit

12:09

im trying it

12:09

it feels nice

12:09

Zed is currently only available on macOS minimally requires Catalina (10.15).

12:09

oh rip

12:09

well ill use it on my m1

12:10

apparently because they use metal and not vulkan

12:10

i hope they switch

Robyt3

Is it worth it to fix the validation warnings? With dbg_gfx 4 you get like 4 MiB of validation warnings, mostly about us not freeing everything I guess

Some warnings will sadly increase complexity, but not freeing stuff should actually throw an error instead of a warning

Ryozuki

apparently because they use metal and not vulkan

Lmao

Ryozuki

@Jupstar ✪ bro this editor is legit

Ive seen it already, as long as it has debugger support it sounds cool (edited)

Ryozuki

https://github.com/dtolnay/inventory

I did find this back when I was trying to do something like it yeah

12:44

Surprised you remember

is it possible to make inclient queue for unfinished maps? you choose max ping limit, you choose a map difficulty and type limits, there should also be points limit so 200pts novicers wont match for insane5, you press button and just wait when it finds ppl, it chooses a server so ping will be about the same for all, and automatically connects to it for all players

13:40

it would be useful for ppl who play rare, because its hard to find teammates to instantly play

https://github.com/DDNet-Team-Searcher

dear devs! is it possible to involve a mechanic that make a weapon a second hook? there would be lot new possiblities with that, intresting new parts, solutions...

no

Bors Matyas

dear devs! is it possible to involve a mechanic that make a weapon a second hook? there would be lot new possiblities with that, intresting new parts, solutions...

go check vanilla teeworlds prs

15:11

https://github.com/teeworlds/teeworlds/pull/2950

3

1

15:16

i played vanilla but never saw this

15:22

are you gonna realease it for ddnet?

Smetanolub

is it possible to make inclient queue for unfinished maps? you choose max ping limit, you choose a map difficulty and type limits, there should also be points limit so 200pts novicers wont match for insane5, you press button and just wait when it finds ppl, it chooses a server so ping will be about the same for all, and automatically connects to it for all players

That has already been tried in different ways and has never really worked. There were: -discord servers -the discord thingy were you can create like conversation, sorry idk how they are called. -and some web stuff like what MilkeeyCat sent. A lot of players mostly play with the same people over and over. Unless it's t0. So even tho you may find someone "randomly" or just by asking, it's not super common. It might work well with lower points players? idk

Devinci

That has already been tried in different ways and has never really worked. There were: -discord servers -the discord thingy were you can create like conversation, sorry idk how they are called. -and some web stuff like what MilkeeyCat sent. A lot of players mostly play with the same people over and over. Unless it's t0. So even tho you may find someone "randomly" or just by asking, it's not super common. It might work well with lower points players? idk

My thing is obviously will be successful and not doomed at all brownbear

Devinci

That has already been tried in different ways and has never really worked. There were: -discord servers -the discord thingy were you can create like conversation, sorry idk how they are called. -and some web stuff like what MilkeeyCat sent. A lot of players mostly play with the same people over and over. Unless it's t0. So even tho you may find someone "randomly" or just by asking, it's not super common. It might work well with lower points players? idk

We should do it as button in client, so more ppl will use it, especially new players who cant even find something in settings

Smetanolub

We should do it as button in client, so more ppl will use it, especially new players who cant even find something in settings

i belive more that accounts will be in ddnet than this

(edited)

We need both of those

17:57

Waiting for accounts from 1800s justatest

@archi

1

Avolicious

There are kids playing this game

I'm playing this game. I don't want toxic chat

heinrich5991

I'm playing this game. I don't want toxic chat

ive seen you 2 times in game justatest

I've seen him 1 time in game justatest

nob

<BµmM>

@archi

oh no you found my secret account

@Jupstar ✪ Seems like #6578 is caused by the client starting minimized and inactive on Windows, because shell_execute always starts the process like that. It's a bug anyway that the client is not in the foreground after restarting, but weird that Vulkan can't create a swap chain if the window is started minimized

https://github.com/ddnet/ddnet/issues/6578

Game window is not shown after restarting game in Windowed and Wind...

After restarting the client from within the client itself (via auto-updater or restart command added in #6577), the game window is entirely missing with Fullscreen mode: Windowed or Windowed Fullsc...

Robyt3

[ddnet/ddnet] Pull request opened: #7859 Ensure client window is maximized and active after restarting, refactoring

Add parameter to shell_execute to either start the process in the foreground or background on Windows. Previously, all processes were started in the background, because this is desired when starting the server from the client. However, this causes the graphics initialization to fail when restarting the client after updating or with the restart command when using Vulkan with windowed and windowed fullscreen mode. Closes #6578.

Checklist

[X] Tested the change ingame
[ ] Pro...

Robyt3

@Jupstar ✪ Seems like #6578 is caused by the client starting minimized and inactive on Windows, because shell_execute always starts the process like that. It's a bug anyway that the client is not in the foreground after restarting, but weird that Vulkan can't create a swap chain if the window is started minimized

i think this was also a bug when we still showed the client console (not related to vk) and i think i also seen a sdl issue, where someone complained about it, but the sdl dev said, windows sucks xd

1

Robyt3

@Jupstar ✪ Seems like #6578 is caused by the client starting minimized and inactive on Windows, because shell_execute always starts the process like that. It's a bug anyway that the client is not in the foreground after restarting, but weird that Vulkan can't create a swap chain if the window is started minimized

how did u install validation layers on windows? is it vk sdk or some nvidia thing?

Jupstar ✪

how did u install validation layers on windows? is it vk sdk or some nvidia thing?

VulkanSDK I guess

Robyt3

VulkanSDK I guess

generally the validation is correct

20:45

to prevent that from happening when tabbing out the vk backend stops rendering completely while so

20:45

maybe we can trigger this state here too.. rn it waits for the SDL event to say that the window was minimized

20:46

could also try to just force a higher resolution if the width or height is reported as 0

20:46

this should™️ be legal and will only result in vk's surface reporting that the surface doesn't match the requirements

It possible to reproduce the bug by creating a desktop shortcut to DDNet.exe and setting the shortcut launch option to minimized

Robyt3

It possible to reproduce the bug by creating a desktop shortcut to DDNet.exe and setting the shortcut launch option to minimized

mh interesting.. why does sdl window creation not fail here then? don't we explicitly tell which size we want for the window?

It works with OpenGL, so something with SDL<->Vulkan or only Vulkan must be the issue

Robyt3

It works with OpenGL, so something with SDL<->Vulkan or only Vulkan must be the issue

well opengl doesn't have a concept like minimized at all

20:53

@Robyt3 in vk backend in GetSwapImageSize can u try to add

if (RetSize.width == 0 || RetSize.height == 0) {
            RetSize.width = 800;
            RetSize.height = 800;
        }

i wonder what will happen

20:53

add it before VkExtent2D AutoViewportExtent = RetSize;

Jupstar ✪

@Robyt3 in vk backend in GetSwapImageSize can u try to add

if (RetSize.width == 0 || RetSize.height == 0) {
            RetSize.width = 800;
            RetSize.height = 800;
        }

i wonder what will happen

does not change anything

Robyt3

does not change anything

mh interesting. is the maxextend in VKCapabilities 0?

yes, currentExtent and maxImageExtent width and height are both 0 (edited)

well it's defs a weird thing that this can happen under windows. without a swapchain you can't test if the surface was changed, without surface change event, you can't really know when the surface is not 0 anymore. i wonder if there is any useful sdl event triggered directly after window creation

21:06

@Robyt3 another try: in RecreateSwapChain after vkDeviceWaitIdle(m_VKDevice);

        VkSurfaceCapabilitiesKHR VKSurfCap;
        if(!GetSurfaceProperties(VKSurfCap))
            return -1;

        auto Caps = GetSwapImageSize(VKSurfCap);
        if (Caps.m_SwapImageViewport.width == 0 || Caps.m_SwapImageViewport.height == 0) {
            m_RenderingPaused = true;
        }

and in PrepareFrame() replace RecreateSwapChain() with

            if (RecreateSwapChain() != 0) {
                return PureMemoryFrame();
            }

i am not sure about the last one, but maybe worth a try. It will probably only work if SDL at some point sends a SDL_WINDOWEVENT_RESTORED event

21:08

maybe m_RecreateSwapChain = true; in Cmd_WindowCreateNtf needs to be moved out of the android-macro else it probably has no idea that the surface is now ready

21:09

this is defs messy, bcs normally on desktops surfaces are never non ready, like on android xd

21:09

so tabbing out would always trigger a rebuild of the swapchain :/

doesn't work

but does it crash?

I don't get window events until after init already failed

Jupstar ✪

but does it crash?

no, same as before

21:12

SDL_WINDOWEVENT event=7 d1=0 d2=0
SDL_WINDOWEVENT event=1 d1=0 d2=0
SDL_WINDOWEVENT event=10 d1=0 d2=0
SDL_WINDOWEVENT event=12 d1=0 d2=0
SDL_WINDOWEVENT event=3 d1=0 d2=0
SDL_WINDOWEVENT event=14 d1=0 d2=0

21:12

all after creating the swapchain fails

mh ok

Robyt3

all after creating the swapchain fails

in Cmd_Init?

21:15

ah wait

Jupstar ✪

@Robyt3 another try: in RecreateSwapChain after vkDeviceWaitIdle(m_VKDevice);

        VkSurfaceCapabilitiesKHR VKSurfCap;
        if(!GetSurfaceProperties(VKSurfCap))
            return -1;

        auto Caps = GetSwapImageSize(VKSurfCap);
        if (Caps.m_SwapImageViewport.width == 0 || Caps.m_SwapImageViewport.height == 0) {
            m_RenderingPaused = true;
        }

and in PrepareFrame() replace RecreateSwapChain() with

            if (RecreateSwapChain() != 0) {
                return PureMemoryFrame();
            }

i am not sure about the last one, but maybe worth a try. It will probably only work if SDL at some point sends a SDL_WINDOWEVENT_RESTORED event

after m_RenderingPaused there must be a return -1;

21:16

xd

21:16

else it tries to create the swapchain xD

21:20

if it's only in cmd_init and some sdl event triggers maximize or restore i'd say it's an easy fix

21:21

else i'll have to think about it

I added

if(m_VKSwapImgAndViewportExtent.m_SwapImageViewport.width == 0 || m_VKSwapImgAndViewportExtent.m_SwapImageViewport.height == 0)
    return false;

after m_VKSwapImgAndViewportExtent = GetSwapImageSize(VKSurfCap); and this seems to work

21:22

combined with all your other changes, I don't know how many were actually required (edited)

21:22

I'll do more testing another day

Jupeyy

[ddnet/ddnet] Pull request opened: #7860 Check if swapchain can be created, e.g. if windows starts minimized

@Robyt3 try this

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 memcheck](https://github.com/ddnet/ddn...

Jupstar ✪

w h a t i s t h a t f o n t s p a c i n g justatest

monospace

not the first time that happened to me on vscode but also only like third time i wonder if it's related to using 2 vscode instances with different extensions or smth xd or if it's simply a console bug (edited)

21:33

but i also didnt update in a while

21:35

for some reason clang-format also is broken if i have a second vscode instance open

21:35

xd

21:35

clang-format-10 to be precise

21:35

i guess it somehow loads the new and the old one or smth like that

21:35

very weird