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 2023-05-12 00:00:00Z and 2023-05-13 00:00:00Z

Learath2

Looks about right to me, what's wrong?

I must do a POST request on a link and save the returned json (edited)

What's this used for?

Image attachment

heinrich5991

tell me what you want to do and I might be able to give you some example code

Use POST and pass Json params (edited)

Voxel

some day im going to clean up the pngs of ddnet

anyways: this is the hypothetical i was talking about:

Image attachment

Would be cleaner yeah

Voxel

What's this used for?

Just look in the source. I doubt it's used in-game

05:28

Maybe in 07

Yeah it's unused. And omg the png would look so much cleaner without it,

Image attachment

05:30

wait whats that blue bar for

05:30

was it for the old ninja timer?

Yes

05:30

But not old

05:30

07

okay wait, let me sort everything out really quickly. some of these are UI elements, not map items

Updated:

Image attachment

05:43

Take 2 on HUD: much more optimized, less blank space

Image attachment

You should leave empty space

actually this may be doable, if we make a script to convert old type pngs to the new lauout

05:43

layout

ok fine old one is used then

Probably even mkre

05:44

because

05:45

if new icons are added, it would fuck up old pngs and it would also require the client to be modified to support the larger image

expanding the image down or right and supporting it is not hard

Thats true

05:45

But backward compatibility

is this better?

Image attachment

(@backwardserich5991)

@Voxel make the same pngs but with the grid info

05:47

and i make a rust script

05:47

gigachad

works for me

@Ryozuki rust fast?

ye

05:47

fastest

woaw

i want to make sure everything is finalized before i make coords

500gb/s

justatest

1

Image attachment

05:51

wait are all the ninja dash streaks the same side in code? (checking right now)

05:53

yeah they are

05:55

idk what happened but when i split the image theres this weird ending trail thats cut off in one of the ninja muzzles

Image attachment

05:59

Image attachment

05:59

this is awkward (edited)

Image attachment

bug

same case goes for another one, which also has a cut off piece

Image attachment

trigger warning

its ok, we can fix it when the svg gets accepted into the database ^^

Should I wait until we get more opinions before I make coords? And would the coords have to be formatted like in content.py?

Image attachment

you can’t fix it without changing the code. it looks for that resource in that place, the coordinates have to be in the same relative position

06:13

all the svg scripts do is export at a larger size

extras, game, hud, particles, strong_weak > hud, map_items, particles, weapons

fokkonaut

You should leave empty space

ok im starting to see what you mean

06:27

theres a ton of empty space in the original images

Voxel

Should I wait until we get more opinions before I make coords? And would the coords have to be formatted like in content.py?

What resolution do they have btw

lets make it 4k

06:33

who needs to support potatoes

It's even worse, I'm asking BCS i want to know if they are power of two textures

06:35

So actual potatoes xdd

06:36

1995 gang xddd

oh tru

06:41

@Voxel tex needs to be 2**x

06:42

size

06:42

128,256,512,1024

yeah yea

06:43

everything is on a 32x32 grid

the total image size i mean

Voxel

is this better?

are you trying to trace vanilla gameskin?

one second

Souly

are you trying to trace vanilla gameskin?

rearranging

Voxel

everything is on a 32x32 grid

Power of two, not multiply of two

@Voxel what for lmao

2^x

Jupstar ✪

Power of two, not multiply of two

one seconnndd im getting the image sizes for everything

Jupstar ✪

2^x

2**5

Jupstar ✪

What resolution do they have btw

weapons.png : 704 x 448 particles.png : 348 x 512 map_items.png : 256 x 384 hud.png : 384 x 384

Souly

@Voxel what for lmao

game.png kind of sucks right now

Well might not matter anyway. Forgot we split them anyway

07:01

For pixel perfection xd

Voxel

game.png kind of sucks right now

ye i kinda get the point

08:12

what annoys me the most is the fact that you need to update the whole gameskin even if you just want to change one single thing

Yep, I'd not use an texture atlas again for most stuff

08:16

Even at the cost of sightly higher load time

Voxel

idk what happened but when i split the image theres this weird ending trail thats cut off in one of the ninja muzzles

This was like that since I've made the templates for gameskin

Souly

what annoys me the most is the fact that you need to update the whole gameskin even if you just want to change one single thing

yeah

Voxel

its ok, we can fix it when the svg gets accepted into the database ^^

Once it is done (and hopefully Nexus comes back), we'll progressively accept SVGs to teeskins

would be cool if we could maybe overwrite some gameskin part with additional file or something

like making your own template?

like if u have laser.png next to game.png it would show the laser from the second file or something

If there would be a text file which you can read the size of each weapon and paste it into the game with the appropriate size and position, it would be possible

Voxel

Should I wait until we get more opinions before I make coords? And would the coords have to be formatted like in content.py?

i think this is a huge benefit for when we want to add more things like map items or particles it's healthily set up in a way where you can add more to the right or bottom

Souly

like if u have laser.png next to game.png it would show the laser from the second file or something

oh

i mean its not a big issue for me personally because i can use things like gimp

08:27

but it would make it easier for sure

of course i'll add a bit of whitespace but other than that i feel like this would be a game changer

Voxel

i think this is a huge benefit for when we want to add more things like map items or particles it's healthily set up in a way where you can add more to the right or bottom

Easier to manage storing Assets on databases too, as in we can just use an online editor API to edit SVGs (like color, size and outline) and let them implement their own

08:28

Damn I wish I knew typescript so that I can help Nexus

08:28

xd

Voxel

of course i'll add a bit of whitespace but other than that i feel like this would be a game changer

Problem would be having other client migrate to that version, but I think it's for the better

08:30

It opens an opportunity to implement many things which would've made older / other clients broken

probably would require some asset loading priority

08:30

so it would support both

08:30

but load the new one over the old

It would need to recreate the old Assets from the new ones and load it into memory, so that other clients can see what Assets they use

08:31

Other than that, It should be fine

08:33

But this is what I'm getting at right now, because of other clients that don't depend on the newer versions of DDNet, it's made in a limited or restricted way. DDNet cannot expand as is right now

HUD is now updated so that we don't have to rearrange everything, just add more things into it

Image attachment

Please also nice the weapon Shields there

yes, please nice them

I still don't understand why they are in game png

They are not HUD so why should they be in hud.png? They are part of the game like health pickups

I mean, aren't these rendered onto the map? Like the flags?

Image attachment

Arent entities' hearts for floating hearts?

08:39

same for shields?

Robyt3

They are not HUD so why should they be in hud.png? They are part of the game like health pickups

The. Extras png

08:40

But not game png. They are entities on the map

if we make this update i highly doubt we need an extras.png (edited)

08:40

i doubt this will ever see the light of day but extras is the most useless thing ever (edited)

Tbh just stop using texture atlasses all along

08:41

They make customization impossible

08:41

Tomorrow someone adds a new map item and all customized map item png get incslid

08:42

Invalid

why not just yoink the 0.7 asset system kek

kek

So what? Are you saying make each and every thing seperate?

Yes

08:42

Later in the client we do this anyway

Yeah why not, only load things which need to be loaded

Can't argue with that

08:43

I wanted to seperate weapons into their own things anyways

08:43

So I guess this is the only logical step after that

After what I experience in Teeworlds Data, changing one weapons and calling it a complete set, while the original is almost identical... It really makes you think that you can save a loooot of VRAM if we just separate weapons into their own category

Jupstar ✪

Yep, I'd not use an texture atlas again for most stuff

cant we make the atlast at runtime, or is it worthless at that point

Then also change the structure a bit. Weapons in weapons directory. Map items in Map items etc

Ryozuki

cant we make the atlast at runtime, or is it worthless at that point

We currently do the opposite xd

ye would be nice

We split them to render pixel perfect

@Jupstar ✪ i mean the point is to load just 1 texture so it uses less ram, and reuse parts of it for the weapons etc right

08:45

oh

ahm no

so u end up loading it as multiple images?

This also makes it easier to add more useless things! Like seperate explosion graphics for seperate objects! (This is a joke)

Image attachment

ye, and good thing is, most duplicate texture parts in Assets would not need to be loaded at all, saving resources

aka u upload 6 textures instead of 1

08:46

to gpu

f.e: if you load 100 Assets which 30 has the same hook, then separating them will make you load only 1 hook and any other original asset parts will be loaded

Voxel

This also makes it easier to add more useless things! Like seperate explosion graphics for seperate objects! (This is a joke)

why not

@heinrich5991 if i edit libtw2s gamenet loader.py how do i generate the json files?

08:48

./generate_all didnt do anything

Ryozuki

why not

because people would be nosy about it again

08:48

like it'd be a fun detail but is it really nessicary?

ignroe it

Voxel

like it'd be a fun detail but is it really nessicary?

depends on the community justatest

justatest

@Jupstar ✪ i get a vulkang segfault on closing client

08:49

[Switching to Thread 0x7fffed90e6c0 (LWP 7581)]
0x00007ffff4f0fc70 in _XSend () from /usr/lib64/libX11.so.6
(gdb) bt
#0  0x00007ffff4f0fc70 in _XSend () at /usr/lib64/libX11.so.6
#1  0x00007ffff4f0600a in XQueryExtension () at /usr/lib64/libX11.so.6
#2  0x00007ffff4ef96f2 in XInitExtension () at /usr/lib64/libX11.so.6
#3  0x00007ffff4ec536d in XextAddDisplay () at /usr/lib64/libXext.so.6
#4  0x00007fffedfc9e6f in  () at /usr/lib64/libGLX_nvidia.so.0
#5  0x00007fffedfeb665 in  () at /usr/lib64/libGLX_nvidia.so.0
#6  0x00007fffded4d499 in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#7  0x00007fffded4b86c in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#8  0x00007fffded4cf9f in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#9  0x00007fffded3f7fc in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#10 0x00007fffded3fdfe in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#11 0x00007fffded3ffc1 in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#12 0x00007fffedfea7bc in  () at /usr/lib64/libGLX_nvidia.so.0
#13 0x00005555557490b5 in CCommandProcessorFragment_Vulkan::DestroySwapChain(bool) (ForceDestroy=true, this=0x555555c7a630) at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:4224
#14 CCommandProcessorFragment_Vulkan::DestroySwapChain(bool) (ForceDestroy=true, this=0x555555c7a630) at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:4220
#15 CCommandProcessorFragment_Vulkan::CleanupVulkanSwapChain(bool) (this=0x555555c7a630, ForceSwapChainDestruct=ForceSwapChainDestruct@entry=true)
    at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:5451
#16 0x000055555574eb64 in CCommandProcessorFragment_Vulkan::CleanupVulkan<true>() (this=0x555555c7a630) at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:5462
#17 0x000055555575ee90 in CCommandProcessorFragment_Vulkan::Cmd_Shutdown(CCommandProcessorFragment_GLBase::SCommand_Shutdown const*) (pCommand=0x555556850f50, this=0x555555c7a630)
    at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:6649
#18 CCommandProcessorFragment_Vulkan::RunCommand(CCommandBuffer::SCommand const*) (this=0x555555c7a630, pBaseCommand=0x555556850f50)
    at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:6529
#19 0x00005555557672d6 in CCommandProcessor_SDL_GL::RunBuffer(CCommandBuffer*) (this=0x555555c7c580, pBuffer=<optimized out>) at /home/edgar/data/ddnet/src/engine/client/backend_sdl.cpp:335
#20 0x0000555555763186 in CGraphicsBackend_Threaded::ThreadFunc(void*) (pUser=pUser@entry=0x555555ad02a0) at /home/edgar/data/ddnet/src/engine/client/backend_sdl.cpp:63
#21 0x00005555556f4de7 in thread_run(void*) (user=<optimized out>) at /home/edgar/data/ddnet/src/base/system.cpp:764
#22 0x00007ffff53efaa3 in  () at /lib64/libc.so.6
#23 0x00007ffff5471b5c in  () at /lib64/libc.so.6

08:49

Image attachment

we have nvidia-driver-530 yk

colors better xd

If we end up splitting all the game images and reordering them in folders, then we should just do this with all images once to get a better structure (including stuff like file icons, demo buttons, emoticons, etc.)

yeah the rewor needs to be done in 1 go

08:50

for everything

08:50

with scripts to convert old pngs to new

08:50

so custom assets are kept

Ryozuki

with scripts to convert old pngs to new

or the other way around for compatibility

idc about backwards compat

me too

gigachad

then why u say that

08:51

xd

idk, I had a brain damage for this moment

08:51

latest rls gang

08:51

gigachad

Ryozuki

@Jupstar ✪ i mean the point is to load just 1 texture so it uses less ram, and reuse parts of it for the weapons etc right

It won't necessarily use less ram. Also more textures don't mean more bandwidth. However it might mean using memory less effectively BCS it's in different locations

got it @heinrich5991 python3 serialize.py --version 0.7 raw/teeworlds-0.7.5.py > spec/teeworlds-0.7.5.json

nobo chiller

08:52

nobody cares about 0.7

08:52

@Jupstar ✪ any hint on why the segfault happens?

Robyt3

If we end up splitting all the game images and reordering them in folders, then we should just do this with all images once to get a better structure (including stuff like file icons, demo buttons, emoticons, etc.)

Wait, so how would it work when you want to actually customize it? What would this look like?

Image attachment

Ryozuki

[Switching to Thread 0x7fffed90e6c0 (LWP 7581)]
0x00007ffff4f0fc70 in _XSend () from /usr/lib64/libX11.so.6
(gdb) bt
#0  0x00007ffff4f0fc70 in _XSend () at /usr/lib64/libX11.so.6
#1  0x00007ffff4f0600a in XQueryExtension () at /usr/lib64/libX11.so.6
#2  0x00007ffff4ef96f2 in XInitExtension () at /usr/lib64/libX11.so.6
#3  0x00007ffff4ec536d in XextAddDisplay () at /usr/lib64/libXext.so.6
#4  0x00007fffedfc9e6f in  () at /usr/lib64/libGLX_nvidia.so.0
#5  0x00007fffedfeb665 in  () at /usr/lib64/libGLX_nvidia.so.0
#6  0x00007fffded4d499 in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#7  0x00007fffded4b86c in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#8  0x00007fffded4cf9f in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#9  0x00007fffded3f7fc in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#10 0x00007fffded3fdfe in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#11 0x00007fffded3ffc1 in  () at /usr/lib64/libnvidia-glcore.so.525.116.03
#12 0x00007fffedfea7bc in  () at /usr/lib64/libGLX_nvidia.so.0
#13 0x00005555557490b5 in CCommandProcessorFragment_Vulkan::DestroySwapChain(bool) (ForceDestroy=true, this=0x555555c7a630) at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:4224
#14 CCommandProcessorFragment_Vulkan::DestroySwapChain(bool) (ForceDestroy=true, this=0x555555c7a630) at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:4220
#15 CCommandProcessorFragment_Vulkan::CleanupVulkanSwapChain(bool) (this=0x555555c7a630, ForceSwapChainDestruct=ForceSwapChainDestruct@entry=true)
    at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:5451
#16 0x000055555574eb64 in CCommandProcessorFragment_Vulkan::CleanupVulkan<true>() (this=0x555555c7a630) at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:5462
#17 0x000055555575ee90 in CCommandProcessorFragment_Vulkan::Cmd_Shutdown(CCommandProcessorFragment_GLBase::SCommand_Shutdown const*) (pCommand=0x555556850f50, this=0x555555c7a630)
    at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:6649
#18 CCommandProcessorFragment_Vulkan::RunCommand(CCommandBuffer::SCommand const*) (this=0x555555c7a630, pBaseCommand=0x555556850f50)
    at /home/edgar/data/ddnet/src/engine/client/backend/vulkan/backend_vulkan.cpp:6529
#19 0x00005555557672d6 in CCommandProcessor_SDL_GL::RunBuffer(CCommandBuffer*) (this=0x555555c7c580, pBuffer=<optimized out>) at /home/edgar/data/ddnet/src/engine/client/backend_sdl.cpp:335
#20 0x0000555555763186 in CGraphicsBackend_Threaded::ThreadFunc(void*) (pUser=pUser@entry=0x555555ad02a0) at /home/edgar/data/ddnet/src/engine/client/backend_sdl.cpp:63
#21 0x00005555556f4de7 in thread_run(void*) (user=<optimized out>) at /home/edgar/data/ddnet/src/base/system.cpp:764
#22 0x00007ffff53efaa3 in  () at /lib64/libc.so.6
#23 0x00007ffff5471b5c in  () at /lib64/libc.so.6

Enable Vulkan layers. Can't imagine why it crashes

crap forgot to untag robyt

@Jupstar ✪ odd, my ddnet from gentoo ebuild doesnt crash

Voxel

Wait, so how would it work when you want to actually customize it? What would this look like?

It would need a window rework pepekek

pepekek

rm -rf build/*

env CXXFLAGS='-march=native -mtune=native' cmake -S . -B build/ \
    -DCMAKE_BUILD_TYPE=RelWithDebInfo \
    -GNinja \
    -DDEV=ON \
    -DVULKAN=ON \
    -DCMAKE_EXPORT_COMPILE_COMMANDS=ON

ln -f -s build/compile_commands.json compile_commands.json

08:53

it uses mold by default, thats pog

Voxel

Wait, so how would it work when you want to actually customize it? What would this look like?

Maybe provide shortcut buttons that select entire sets that have the same name

Jupstar ✪

Enable Vulkan layers. Can't imagine why it crashes

how to do that

Ryozuki

how to do that

Start with dbg_gfx 4

Voxel

Wait, so how would it work when you want to actually customize it? What would this look like?

For a start we could use the directory name.. e.g. default is weapons/default/*.png We can display them similar to here in a group

08:55

That's also how entities do it btw

yeah, just show them the same as before grouped, so you can select all of them at the same time

2023-05-12 10:55:48 I vulkan: deallocated chunk of memory with size: 196608 for frame 1 (buffer) 2023-05-12 10:55:48 I vulkan: deallocated chunk of memory with size: 196608 for frame 1 (buffer) Segmentation fault

It must be Vulkan error if it's a detection by a VK layer

08:56

Also u need to install them

08:56

Dunno if they exist for Nvidia

i got vulkan-layers

Oh man, you know what would be great? I don't think this will be optimal for real time gameplay but what if you could choose random weapons for each tee for demo rendering/playback?

Voxel

ddnet

Suggestion ddnet

ddnet

What if there was a tick box where if you have more than 1 game skin or emoticon pack every tee except yours is assigned a random one to keep? Like, everyone on a team has a different hammer look or grenade but they all function the same.

i had this idea before

Yes it's a long time suggestion

08:58

Like with everything our network code sucks

08:58

Before someone won't convince Heinrich to finally break with it it will create bloat

08:59

I mean for non random

08:59

E.g. a player sends his selection

08:59

Random is weird xd

08:59

But i like the idea of customization

yeah

09:00

i need to add new netobjs for damage indicator, spawn particles, etc so i can apply opacity correctly

Jupstar ✪

Random is weird xd

imagine you can send over particles too custom grenade explosions

09:02

@Jupstar ✪ ok, it segfaults on master, but not on 16.9 tag

09:03

im gonna find the commit

I don't want to go into detail but i talked to Heinrich. It's not that he dislike suchs ideas. It's simply that he wants such features always be downward compatible. So nice dreams but probably not maintainable. Esp. Since most contributions are not high quality and/or not highly tested

yeah he wantsb ackwards compat

09:04

idk why so much

09:04

we could make so much improvements

09:04

way more modularity

09:04

no ugly code

Ryozuki

@Jupstar ✪ ok, it segfaults on master, but not on 16.9 tag

Oh weird indeed

https://github.com/ddnet/ddnet/pull/6560

Replace EntityEx with new netobjs by trml · Pull Request #6560 · dd...

This is a proposal for replacing EntityEx with new netobjects, like suggested in #5860. Adds a new DDNetPickup that includes switch number a new DDNetProjectile that replaces the old one (which is...

Why do you want to split game.png anyway?

❯ git bisect bad
c8b69dd8156a4a36656d0467845ca9b02d20fc18 is the first bad commit
commit c8b69dd8156a4a36656d0467845ca9b02d20fc18
Author: Robert Müller <robytemueller@gmail.com>
Date:   Wed May 10 20:51:25 2023 +0200

    Properly uninitialize all client components

    Track stack of cleaner functions that destruct client components so exactly the correct cleanup is performed in the reverse order of initialization.

    This allows performing the cleanup also in the early-return cases without introducing duplicate code.

 src/engine/client/client.cpp | 61 +++++++++++++++++++++++++++-----------------
 1 file changed, 38 insertions(+), 23 deletions(-)

09:14

@Jupstar ✪ found it

09:14

its this

09:14

causes segfault

09:14

@Robyt3 !!

```bash xd

Learath2

Why do you want to split game.png anyway?

because its trash + incompatable and hard to update

[ddnet/ddnet] Issue opened: #6581 Segfault using vulkan

Caused by #6577 ``` c8b69dd8156a4a36656d0467845ca9b02d20fc18 is the first bad commit commit c8b69dd8156a4a36656d0467845ca9b02d20fc18 Author: Robert Müller Date: Wed May 10 20:51:25 2023 +0200 Properly uninitialize all client components Track stack of cleaner functions that destruct client components so exactly the correct cleanup is performed in the reverse order of initialization. This allows performing the cleanup also in the early-return cases without intro...

added the trace there

09:16

xd

I'll take a look later

ok thx

Voxel

because its trash + incompatable and hard to update

Trash is hard to define. Incompatible with what? Why is 5 different images easier to update than 1? To me it seems that the steps would be about the same

Learath2

Trash is hard to define. Incompatible with what? Why is 5 different images easier to update than 1? To me it seems that the steps would be about the same

diferent images allows more custom options tho

did you read the whole convo

ye read backlog

Voxel

did you read the whole convo

Mostly skimmed through it where you were talking about how you can add cool new graphics and how it’d break backwards compatibility. Neither are very convincing reasons to do it

@Learath2 do u know why gdb says warning: Missing auto-load script at offset 0 in section .debug_gdb_scripts

Ryozuki

diferent images allows more custom options tho

This is about the only benefit I can think of

Ryozuki

@Learath2 do u know why gdb says warning: Missing auto-load script at offset 0 in section .debug_gdb_scripts

Nope, sorry boss

09:21

Sounds unimportant

@Learath2 any fancy plugin scripts for gdb u use?

09:22

i wanna make my gdb powerful

09:22

or is lldb better

I use GEF while reverse engineering

oh

09:23

lldb shows column too

09:23

not just line

Learath2

I use GEF while reverse engineering

It’s not as useful when debugging software you have the source to

Learath2

I use GEF while reverse engineering

What are u reverse engineering?

* thread #5, name = 'DDNet', stop reason = signal SIGSEGV: invalid address (fault address: 0x0)
    frame #0: 0x00007ffff4f0fc70 libX11.so.6`_XSend + 48
libX11.so.6`_XSend:
->  0x7ffff4f0fc70 <+48>: movq   (%rax), %rsi
    0x7ffff4f0fc73 <+51>: andl   $0x1, %edx
    0x7ffff4f0fc76 <+54>: movq   %rsi, (%rsp)
    0x7ffff4f0fc7a <+58>: jne    0x7ffff4f0fdb9            ; <+377>
(lldb) bt
* thread #5, name = 'DDNet', stop reason = signal SIGSEGV: invalid address (fault address: 0x0)
  * frame #0: 0x00007ffff4f0fc70 libX11.so.6`_XSend + 48
    frame #1: 0x00007ffff4f0600a libX11.so.6`XQueryExtension + 170
    frame #2: 0x00007ffff4ef96f2 libX11.so.6`XInitExtension + 50
    frame #3: 0x00007ffff4ec536d libXext.so.6`XextAddDisplay + 77
    frame #4: 0x00007fffedfc9e6f libGLX_nvidia.so.0`___lldb_unnamed_symbol3788 + 239
    frame #5: 0x00007fffedfeb665 libGLX_nvidia.so.0`___lldb_unnamed_symbol4144 + 69
    frame #6: 0x00007fffded4d499 libnvidia-glcore.so.525.116.03`___lldb_unnamed_symbol75337 + 313
    frame #7: 0x00007fffded4b86c libnvidia-glcore.so.525.116.03`___lldb_unnamed_symbol75283 + 124
    frame #8: 0x00007fffded4cf9f libnvidia-glcore.so.525.116.03`___lldb_unnamed_symbol75329 + 31
    frame #9: 0x00007fffded3f7fc libnvidia-glcore.so.525.116.03`___lldb_unnamed_symbol75086 + 316
    frame #10: 0x00007fffded3fdfe libnvidia-glcore.so.525.116.03`___lldb_unnamed_symbol75094 + 110
    frame #11: 0x00007fffded3ffc1 libnvidia-glcore.so.525.116.03`___lldb_unnamed_symbol75095 + 17
    frame #12: 0x00007fffedfea7bc libGLX_nvidia.so.0`___lldb_unnamed_symbol4127 + 60
    frame #13: 0x00005555558379ee DDNet`CCommandProcessorFragment_Vulkan::DestroySwapChain(this=0x0000555555f99f60, ForceDestroy=true) at backend_vulkan.cpp:4224:25
    frame #14: 0x000055555583bfda DDNet`CCommandProcessorFragment_Vulkan::CleanupVulkanSwapChain(this=0x0000555555f99f60, ForceSwapChainDestruct=true) at backend_vulkan.cpp:5451:19
    frame #15: 0x0000555555857c60 DDNet`void CCommandProcessorFragment_Vulkan::CleanupVulkan<true>(this=0x0000555555f99f60) at backend_vulkan.cpp:5462:27
    frame #16: 0x000055555583fd56 DDNet`CCommandProcessorFragment_Vulkan::Cmd_Shutdown(this=0x0000555555f99f60, pCommand=0x0000555557150d10) at backend_vulkan.cpp:6649:22
    frame #17: 0x000055555583f613 DDNet`CCommandProcessorFragment_Vulkan::RunCommand(this=0x0000555555f99f60, pBaseCommand=0x0000555557150d10) at backend_vulkan.cpp:6529:20
    frame #18: 0x00005555558d0a63 DDNet`CCommandProcessor_SDL_GL::RunBuffer(this=0x0000555555f8f6c0, pBuffer=0x00007fffffffd390) at backend_sdl.cpp:335:38
    frame #19: 0x00005555558cf7bb DDNet`CGraphicsBackend_Threaded::ThreadFunc(pUser=0x0000555555de32a0) at backend_sdl.cpp:63:34
    frame #20: 0x00005555557ceae7 DDNet`::thread_run(user=0x0000555555edd8d0) at system.cpp:764:12
    frame #21: 0x00007ffff53efaa3 libc.so.6`___lldb_unnamed_symbol3510 + 707
    frame #22: 0x00007ffff5471b5c libc.so.6`___lldb_unnamed_symbol3936 + 11

Ryozuki

or is lldb better

Mh, gdb I find easier to use, but they are about the same in capability

lldb shows more info for me

09:24

i think u can tell clang to make lldb optimized debug info

09:24

atleast thats a thing in llvm iirc

Oh I remember one feature in lldb that’s nice, it has full support for c++ syntax. So you can do method calls in expressions easily

Jupstar ✪

What are u reverse engineering?

Mostly ctfs, though it’s been a while since I had time to do one

can someone review https://github.com/ddnet/ddnet/pull/6560 too

09:27

i want this merged

09:28

i noticed we got 33prs there

09:28

starting to look like tw xd

09:29

and also my pr thanks https://github.com/ddnet/ddnet/pull/6580

Ryozuki

starting to look like tw xd

Yeah

AnyaSob

09:31

Maybe I’ll do a review day tomorrow

i would merge stuff, but im not confident in havent missed stuff

09:31

having not*

09:31

english broke

https://youtu.be/IfCRHSIg6zo

Matt's Ramblings

Quake's PVS: A hidden gem of rendering optimization

Image attachment

09:45

why does clang doesnt find it

09:45

its annoying sometimes

09:46

clangd*

Do you have your compile_commands set properly? Does clangd even use them?

same thing happens in vscode

09:47

Image attachment

09:47

i have -DCMAKE_EXPORT_COMPILE_COMMANDS=ON

09:47

and it knwos the type of functions

09:47

if i hover on em

09:47

so its like partially knows

09:51

idk

09:51

oh

09:51

it onl happens in .h files

09:51

The error occuring in an .h but not in a .cpp file may indicate that clangd is treating the .h file as a C header rather than a C++ header.

Ryozuki

Click to see attachment 🖼️

Maybe it's actually not included. I had such things too after the header include refactor

ok it already treats it as c++ header

!rust wot a function without body?

@Ryozuki

https://github.com/heinrich5991/libtw2/blob/0352e6f2f23e01e17c6212701f8faa166c524522/wireshark-dissector/sys/src/lib.rs#L2902-L2908

libtw2/lib.rs at 0352e6f2f23e01e17c6212701f8faa166c524522 · heinric...

Some Teeworlds stuff in Rust. Contribute to heinrich5991/libtw2 development by creating an account on GitHub.

09:53

how does that work

dude its a extern fn

pub fn create_dissector_handle() -> dissector_handle_t;

its just inside and extern block

so its a c func?

same as pub extern fn

09:54

ye

hm

linked

09:54

libtw2 uses old rust

09:54

for lot of stuff

09:54

and also heinrich style

ye libtw is old

i dont like it much

hater

09:54

we know

xd

stop bully heinrich

09:55

i his fan

Ryozuki

i dont like it much

xd

09:57

it it compiles it means vec2 ends up there right

09:57

but why clangd doesnt find it

09:57

smh

09:57

rust-analyzer best

once again im stuck at hello world in rust fml

Ryozuki

but why clangd doesnt find it

Try including the base vmath thing there and see if that helps?

10:00

Idk how these tools resolve things when they aren't included in this file but the previous one includes both

Learath2

Try including the base vmath thing there and see if that helps?

it helps

10:02

but heinrich will say its a useless include

10:03

Image attachment

10:03

same happens in otherp laces

10:03

sux

10:03

c++ tooling bad

skill issue

10:04

have you tried using eclipse c++ instead of vim?

10:05

@Ryozuki "i noticed we got 33prs there" where?

Ryozuki

but heinrich will say its a useless include

Is it really? I think it's fine, the include guard will stop it from being included and it's only strictly correct with the include no?

Do you happen to know if there is any possibility inside a .rs file to hide the keyword unsafe from a user? E.g. in c it's simply #define notunsafe unsafe

trol

This is a serious question if u mean me

i mean u

Learath2

Is it really? I think it's fine, the include guard will stop it from being included and it's only strictly correct with the include no?

but this happens in many header files

If you are looking for an alternative to C's #define, there is nothing that will allow arbitrary words to be used as keywords (which is probably a good thing).

cant we just use c++20 modules

10:06

or 22

10:06

i forgot

Ryozuki

cant we just use c++20 modules

Don't know if it helps here

ryo where we got 33prs?

I'd bet the tooling is even worse for those

c++ headers are one of the hardest things to parse on earth

10:07

in ddnet chiller

Ryozuki

c++ headers are one of the hardest things to parse on earth

I'd go so far to say that they are unparseable

who is we? xd

10:07

ah u mean open?

ye dumbo

fakof

10:07

!jup maybe a macro can hide it away

@Jupstar ✪

Ryozuki

ye dumbo

Reported for being rude. You'll be banished to the shadow realm

!jup cargo add custom-unsafe-macro then call that and slap ur code in

@Jupstar ✪

Learath2

Reported for being rude. You'll be banished to the shadow realm

i added a o to not be rude

i added a fakof to not sound serious

I bet it was just a typo

dumbo is the elephant

lol

Image attachment

10:09

im safe

I'll ban you to be safe

ChillerDragon mh yeah thought about that. Shit xddd

u tryna detect unsafe rust?

Jupstar ✪

Do you happen to know if there is any possibility inside a .rs file to hide the keyword unsafe from a user? E.g. in c it's simply #define notunsafe unsafe

there is with macros

10:09

but u can add a forbid unsafe_code in the top level

ChillerDragon

u tryna detect unsafe rust?

unsafe rust? from ryo?

10:10

impossible

also there is generated code that can introduce unsafe code

Cellegen

unsafe rust? from ryo?

i do it everyday at work

justatest

im sure rust macros could generate could that gets added to compilation

10:10

code*

I am thinking about a way to make sure unsafe is not used at all for an API. Maybe I should simply disallow extern creates

rust macros can do anything

10:10

in sqlx they verify a sql query is valid

10:10

by checking with the db

10:10

at compile itme

10:10

time

Ryozuki

rust macros can do anything

Yeah, they are a problem

10:11

Can u disable them xd

well the ones that can do anything are the procdural macros

yea so i can write a rust macro that generates rust code and includes it in the compiled end result using the unsafe keyword without using the string "unsafe"

Yep

i mean u cant avoid this

10:11

for example another attack vector is build.rs

Jupstar ✪

I am thinking about a way to make sure unsafe is not used at all for an API. Maybe I should simply disallow extern creates

Just don't expose any unsafe functions?

it executes on build time

10:12

forbid unsafe is the best bet

Learath2

Just don't expose any unsafe functions?

I cannot control the code directly

u mean a dependency?

I can only analyse it

safe rust can be evil too

But i could compile it. So maybe inject forbid unsafe... But how safe would that be

10:12

The user could simply allow it again

Who would lie to you about being unsafe anyway?

there are hacky macros to check for panics maybe there is something for unsafe too

forbid unsafe checks the current library has no unsafe usage

Learath2

Who would lie to you about being unsafe anyway?

Someone who executes code on my computer

it works for macros too

10:13

it detects everything

10:13

but only for ur crate

Jupstar ✪

Someone who executes code on my computer

You can execute evil code without unsafe too

yeah

10:13

safe rust is not sandboxed

Learath2

You can execute evil code without unsafe too

It has no access to system api

Ryozuki

safe rust is not sandboxed

Yeah it's sandboxed

@Jupstar ✪ use this https://github.com/rhaiscript/rhai

GitHub - rhaiscript/rhai: Rhai - An embedded scripting language for...

Rhai - An embedded scripting language for Rust. Contribute to rhaiscript/rhai development by creating an account on GitHub.

You can open files in safe rust. You mean no unsafe anywhere in any call chain?

Ryozuki

@Jupstar ✪ use this https://github.com/rhaiscript/rhai

I need good performance

Jupstar ✪

I need good performance

it should have good perf

10:15

Image attachment

Learath2

You can open files in safe rust. You mean no unsafe anywhere in any call chain?

I actually have code that is sandboxed completely except for a specific API. And all I need here is memory safety

10:15

The API must be robust anyway

!jup https://crates.io/crates/no-panic

@Jupstar ✪

its not unsafe related but panic

10:16

A Rust attribute macro to require that the compiler prove a function can't ever panic.

the panic case is hard

Panic is not a problem

maybe unsafe gets caught by that too

I can control abortion

panic is not unsafe

10:16

nobo

but is unsafe panic?

unsafe needs to deal with panics

10:17

but not the other way around

10:17

a panic is safe

10:17

unsafe needs to deal with panics in the way of assuming stuff

yes but can a unsafe not panic?

unsafe can panic

10:17

safe c an panic

will it always?

????

So you get a piece of code to execute in a sandbox. You already limited the api to the outside. Just disallow external crates and std in the 'sandboxed' code?

i mean the no-panic crate ensures that your program wont compile if you have the possibility to panic. if unsafe introduces that possibility then you catch those as well

Yeah but e.g. using a vec should be allowed again xD

related to unsafe and panick (unwinds): https://doc.rust-lang.org/nomicon/unwinding.html

Unwinding - The Rustonomicon

The Dark Arts of Advanced and Unsafe Rust Programming

10:18

Rust has a tiered error-handling scheme:

>

If something might reasonably be absent, Option is used. If something goes wrong and can reasonably be handled, Result is used. If something goes wrong and cannot reasonably be handled, the thread panics. If something catastrophic happens, the program aborts.

10:18

xd

I just need to make sure the vec is not nullptr for example once the API gets this vec

Mh expose vecs through your api

10:20

You can't let them use std, if they can use std they can do anything

Mhh I dunno

Ryozuki

Rust has a tiered error-handling scheme:

>

If something might reasonably be absent, Option is used. If something goes wrong and can reasonably be handled, Result is used. If something goes wrong and cannot reasonably be handled, the thread panics. If something catastrophic happens, the program aborts.

And if you get an our of memory then throw your hands up in the air since there is nothing you can do in Rust

Fwiw I think you should use something like wasm, it's performant enough nowadays

I guess the sandbox itself needs support for checking memory safety here :/

deen

And if you get an our of memory then throw your hands up in the air since there is nothing you can do in Rust

there is try_ stuff getting in

10:21

but yeah

Learath2

Fwiw I think you should use something like wasm, it's performant enough nowadays

Yes. But if I want the wasm module to call host functions

https://rust-lang.github.io/rfcs/2116-alloc-me-maybe.html

2116-alloc-me-maybe - The Rust RFC Book

10:21

https://docs.rs/fallible_collections/latest/fallible_collections/

fallible_collections - Rust

impl Fallible collections on allocation errors, quite as describe in RFC 2116 This was used in the turbofish OS hobby project to mitigate the the lack of faillible allocation in rust.

10:21

@deen

10:22

this got a push thanks to linux

Jupstar ✪

Yes. But if I want the wasm module to call host functions

You can export host functions to wasm. Just need to be selective with what you expose. No direct file opening mechanisms

Learath2

You can export host functions to wasm. Just need to be selective with what you expose. No direct file opening mechanisms

And how does it handle data types?

@Jupstar ✪ https://docs.wasmtime.dev/lang-rust.html#importing-host-functionality

Rust - Wasmtime

In a very ugly manner

Ryozuki

@Jupstar ✪ https://docs.wasmtime.dev/lang-rust.html#importing-host-functionality

Well this is with simple data types

10:26

Is there also a complex example?

u probs need to make them repr C

10:26

but idk

10:26

i never used this

Forget passing complex types, you have something like an array, something like a dict, bools, ints, doubles iirc

A dictionary? So a hashmap?

10:28

Mh ok anyway, guess have to think what exactly i want

I think they called them tables tho I'm not a web assembly expert

https://github.com/OfekShilon/optview2

GitHub - OfekShilon/optview2: User-oriented fork of LLVM's opt-viewer

User-oriented fork of LLVM's opt-viewer. Contribute to OfekShilon/optview2 development by creating an account on GitHub.

Guess I need to use json strings

javascript kiddos

utf-16 ftw

10:31

ah i read js

10:31

xd

10:33

@Learath2 how do u prevent oom or do something if a alloc fails in c++?

10:33

ah i guess malloc simply returns a nullptr? xd

Ryozuki

ah i guess malloc simply returns a nullptr? xd

Yes

but i bet 99% of ppl dont check nullptr on malloc

10:34

we certainly dont

And in C++ it's either nullptr or new throws

Ryozuki

but i bet 99% of ppl dont check nullptr on malloc

I always check mine religiously

xd

10:34

but malloc is tricky right

10:34

cuz overcommit

10:34

in linux

10:34

it can give u all u want

guys... already 500+ messages and it's not even noon...

well nobody forces u to read em kek

10:35

mainly a discussion about images and my segfault

10:35

which wouldnt happen in rust (probs)

Chairn

guys... already 500+ messages and it's not even noon...

But Friday

Ryozuki

it can give u all u want

Yep, return of malloc doesnt mean much nowadays

Ryozuki

i would merge stuff, but im not confident in havent missed stuff

same here

Chairn

same here

merge my pr it will be fine

10:42

kek

I'll merge it later

brownbear

1

on the image thing, i don't think it's a good idea to split to small files

https://www.twitch.tv/videos/1817384336 i did the fixes of the pr yesterday, it took me 1h cuz i forgot abotu the stupid m_Teams, im certainly rusty

we would lose thematic games doing so, but we can maybe split them in software and allow a lot more option directly in game (like a split game icons option)

the video has no voice btw xd

Chairn

we would lose thematic games doing so, but we can maybe split them in software and allow a lot more option directly in game (like a split game icons option)

We lost what?

10:46

U mean that it will not look fitting design wise?

Jupstar ✪

But Friday

sorry, im working hard on friday too

10:46

yes, like thematic game.png, such as Pipou's

Why would we loose it

10:46

He can create a game bundle just like now

we need to ask the resident designer ravie

Ryozuki

the video has no voice btw xd

Boring

im not sure it's more practical, gotta ask graphics people @Pipou @Voxel

Jupstar ✪

Boring

i know, hence the title

Ryozuki

we need to ask the resident designer ravie

He always has opposite opinion to mine

10 minutes before my 2nd audition

Very simple xd

xd

Chairn

10 minutes before my 2nd audition

audition for?

Ryozuki

we need to ask the resident designer ravie

@Ravie

Ryozuki

audition for?

associate professor position

ohh

10:48

idk what that means but nice, or my condolences

10:48

(insert meme here)

10:49

Image attachment

10:49

xd

Xd

Chairn

associate professor position

Are u then rich enough to buy win10?

my mind is not poor enough to do that

just use linux

10:50

if my teacher told me he uses windows

soon(tm)

i would respect him less

XD

justatest

i use linux for almost everything

10:50

except games

same

10:50

sadly

I use it for everything

pog

i teach RISC-V on linux

Games that don't work are not worth it from the start

we really need #dev-offtopic

true

10:51

but sometimes i wanna try those aaa games

to not pollute this channel

Chairn

i teach RISC-V on linux

and make use of my 3080

Ryozuki

and make use of my 3080

Steal 4090 from sis

@Jupstar ✪ i went to her house and installed the 4090 vertically

10:52

it was like usual before, b ut we couldnt close the case

10:52

this time it worked

10:52

its a massive brick

10:52

vertical i mean like this

Image attachment

10:53

@Jupstar ✪ 190 fps on cyberpunk with the new rtx mode

10:53

and the fans barely spinning

10:53

55c iirc

10:53

she has a slow ram i told her to change tho

Xd

xd

10:54

i hope nvidia releases a new gpu

10:54

so i can get this 4090

10:54

kek

Jupstar ✪

Games that don't work are not worth it from the start

But genshin has waifus

do u rly play genshin

10:54

i got bored in 10 mins

10:55

but genshin is played by all egirl weebs

10:55

well most egirls play valoran

10:55

xdd

10:55

valorant

10:55

boring game too

10:55

csgo is better

Ryozuki

do u rly play genshin

Ofc, the music is amazing and the story is pretty interesting

oh it has story

And it has waifus, it's good enough for me

11:00

@Patiga have you been writing an essay for the last hour or is mobile discord glitching?

scrolling through the chat, I blame discord bluekitty

bluekitty

11:03

@Jupstar ✪ btw what was your reasoning to drop the texture atlas? most of them do not have pixels at the borders of each segment. for example skin, emote, (most?) of game

Patiga

@Jupstar ✪ btw what was your reasoning to drop the texture atlas? most of them do not have pixels at the borders of each segment. for example skin, emote, (most?) of game

That still scales them a bit

what scales them?

Because u have to remove texeloffset

11:04

From the texture coordinates

11:04

If u don't do this, it will bleed

ow what is a texel-offset?

11:05

if the textures within the atlas don't touch the border, it shouldn't be a problem, right?

If ur IMG is 1024*1024 it's 1/1024

Patiga

if the textures within the atlas don't touch the border, it shouldn't be a problem, right?

But then it's luck

11:05

It only works under this circumstance

afaik we don't have anti-aliasing on guns and tees, right?

We don't do multi sampling for individual render calls

11:07

Only full screen

11:08

Or only framebuffer

11:08

Would be more correct

I'm confused. multi-sampling != mip-maps, right? did I falsely believe that ddnet uses mip-maps?

Ddnet uses mipmap

alongside multi-sampling, or are they used in different places?

Chairn

@Ravie

yeah?

Patiga

alongside multi-sampling, or are they used in different places?

We only use multi sampling on the global framebuffer

11:11

We don't implement any other kind of anti aliasing

so we render a larger image and then multi-sample it down?

Ryozuki

but i bet 99% of ppl dont check nullptr on malloc

Doesn't matter much for a game. For some DBMSes I consider it essential not to crash on OoM, but to handle it better

Patiga

so we render a larger image and then multi-sample it down?

I'm a bit confused rn. Why larger image

11:15

Ssaa, super sampling, does that afaik

11:15

Multi sample averages between neighbors pixels. Or is the name ambiguous?

We only use multi sampling on the global framebuffer

I didn't understand this. is the global framebuffer the frame that gets presented, or one step before that?

Patiga

We only use multi sampling on the global framebuffer

I didn't understand this. is the global framebuffer the frame that gets presented, or one step before that?

That's a hard question. It's actually multiple framebuffer images that combine into one in the last step xD

i'm sure this has been answered recently enough but i can't find it how do you get the check mark on your server? i see kog also has it so i know it's not ddnet exclusive, what hoops do you have to jump through to get "verified"

So u could say it's after the fragment shader but before written to the output image i think

11:18

@Patiga i feel like we drifting away, what was the original question?

11:18

Why I do pixel perfect

1

11:19

BCS without Texel offsets we can bleed

11:19

Even if just hypothetically

11:19

At least the hook would bleed

11:19

Since the chain hits the edge

true, I currently handle the hook differently because it has to be rendered alongside the tee

Still not too sure on how to read rust docs how do i create a CStr that is not nul terminated? https://doc.rust-lang.org/core/ffi/struct.CStr.html

CStr in core::ffi - Rust

Representation of a borrowed C string.

so you have a texture for each atlas segment if I understand correctly, does that mean you have one draw call for each body part of each tee?

ah this should do CStr::from_bytes_until_nul() sorry for rubberducking in this channel

cstr is null terminated

11:22

a str or string isnt

ChillerDragon: is there context to your message in the chatlog? I don't think you are supposed to have a CStr without a nul byte

no context no brain

11:23

i probably do it wrong xd

what r u doing

11:23

cstr is only useful for ffi

trying to print a message in rust xd

why using cstr

check the example code

usnafe

im segfaulting since 1h on hello world

cstr is safe

Patiga

so you have a texture for each atlas segment if I understand correctly, does that mean you have one draw call for each body part of each tee?

Yes

does it guarantee termination

define termination

\0 at the end

cstr guarantees the str has a nul byte

11:24

its the stupid C requirement

also for each single particle? seems not that trivial to group them @Jupstar ✪

in rust code itself u should use str though, its more optimized

to be fair the only other way was to keep track of string lengths directly which would be even more of a pain

Patiga

also for each single particle? seems not that trivial to group them @Jupstar ✪

No

but yes i hate c string termination too

Ewan

i'm sure this has been answered recently enough but i can't find it how do you get the check mark on your server? i see kog also has it so i know it's not ddnet exclusive, what hoops do you have to jump through to get "verified"

anyone know

Patiga

also for each single particle? seems not that trivial to group them @Jupstar ✪

It gets grouped as much as possible

11:25

But the particle groups are new draw calls indeed

chiller check rust_version.rs

ah no ryo in tw codebase im rust wizzard dont worry

11:27

im trying to log to console in libtw2 wireshark dissector

11:27

i kinda manage but it segfaults or doesnt print at all

Patiga

also for each single particle? seems not that trivial to group them @Jupstar ✪

But since u ask I'd only allow squares for particles here and use a texture array rather than atlas

Embedded image

11:27

F

in ffi u have to be careful on the rust side

11:28

if u pass a object to cpp

11:28

cuz rust drops

11:28

use leak to give ownership to cpp side

if i def it in place?

mem forget

hm k

drop happens at the end of scope

11:28

https://doc.rust-lang.org/std/mem/fn.forget.html

forget in std::mem - Rust

Takes ownership and “forgets” about the value without running its destructor.

ye got it

11:29

thanks

11:29

ima try

u can also use https://doc.rust-lang.org/std/mem/struct.ManuallyDrop.html

ManuallyDrop in std::mem - Rust

A wrapper to inhibit compiler from automatically calling T’s destructor. This wrapper is 0-cost.

11:30

https://doc.rust-lang.org/std/mem/fn.forget.html#relationship-with-manuallydrop

forget in std::mem - Rust

Takes ownership and “forgets” about the value without running its destructor.

Jupstar ✪

But since u ask I'd only allow squares for particles here and use a texture array rather than atlas

ah do you actually group the entire particles png together in an texture-array, or only the same-sized ones? if you do actually group the entire thing together, would you consider also grouping the entire game.png together?

still abort ._.

11:32

maybe its wireshark commiting suicide because i use fatal log level

1

try debug level

Patiga

ah do you actually group the entire particles png together in an texture-array, or only the same-sized ones? if you do actually group the entire thing together, would you consider also grouping the entire game.png together?

Currently it's not done at all. I don't think that our particle system is build in a intelligent way anyway I think we already talked about pros and cons Generally batching is obv better of the runtime. However it also means that u either need own sampling as for texture atlasses. Or u need texture arrays with equally sized images. Game PNG is simply too diverse for that

11:33

If we wouldn't care about ogl1.x I'd do the first

ok got it finally it was be not being able to rtfm https://zillyhuhn.com/cs/.1683891254.png

Embedded image

Ewan

anyone know

Mh the checkmark thing was supposed to be ddnet only. But then people started aggressively ddosing the masters, so we had to add some 3rd party servers to info.json, those ended up also getting the checkmark

@heinricj5991 did you code ur dissector without print driven debugging? xd

Learath2

Mh the checkmark thing was supposed to be ddnet only. But then people started aggressively ddosing the masters, so we had to add some 3rd party servers to info.json, those ended up also getting the checkmark

So there is no process for it since it's not even intended as far as I know

I do realize we were here before, but we do come from the direction off ddnet this time ^^. well if you would choose batching, would you go the texture atlas or texture array path? :)

@Patiga u can also do multi texturing if u write a pipeline for all game elements anyway

11:35

U should have a guarantee of 16 images per shader

11:36

Maybe in fact I'd do that

hm that does seem a bit extreme, even to me :p. The different images do seem mostly separated in a sensible way (except that game png and skin png mix during tee rendering)

Patiga

hm that does seem a bit extreme, even to me :p. The different images do seem mostly separated in a sensible way (except that game png and skin png mix during tee rendering)

What is extrem?

well I feel like I try to batch together much more stuff than ddnet currently does

11:38

(and at times am struggling coz of this a little ^^)

The only downside is that it doesn't lay in continuous memory which might hurt the cache a tiny bit

and throwing literally everything into the same shader goes quite extreme in that direction

Patiga

well I feel like I try to batch together much more stuff than ddnet currently does

And how does multi texturing prevent that

Learath2

So there is no process for it since it's not even intended as far as I know

how did KOG and noby's get it?

Mh thinking of it, maybe only newer ogl versions allow index accessing a shader array tho

with multiple images attached you do get the problem of not being able to use the simple textureSample, because it isn't safe to use when in the same execution unit you might need to sample different textures (in wgpu its called non-uniform control flow)

Sampler array*

Patiga

with multiple images attached you do get the problem of not being able to use the simple textureSample, because it isn't safe to use when in the same execution unit you might need to sample different textures (in wgpu its called non-uniform control flow)

Even for indexing?

Ewan

how did KOG and noby's get it?

The second sentence. During the big ddos incident the masters were completely unusable, so we had to put some people in the ddnet list

11:42

(the ddnet list used to be over http even before the http masters)

I don't get it. Were they ddossing based on the check mark?

Jupstar ✪

Even for indexing?

no, not for indexing. I thought you wanted to have multiple textures in your bind group and choose dynamically which one to sample (your number 16, led me to believe that somehow)

Ryozuki

nobody cares about 0.7

please do not lazily diminish other people's work

Ewan

I don't get it. Were they ddossing based on the check mark?

Servers inside ddnet-info.json get the checkmark. When the master servers were down, people could only play on ddnet servers. People asked us to add other servers to our list so they can show up aswell even when the masters are down. We add servers to the list, they get the checkmark as a side-effect

Oh I see

Patiga

no, not for indexing. I thought you wanted to have multiple textures in your bind group and choose dynamically which one to sample (your number 16, led me to believe that somehow)

Yes I want that over an index xd

xd

11:45

I don't think that is a thing

I dunno

11:45

I am sure desktop GPUs should support it

Is there a way to get put into ddnet-info.json or do we just have to deal with being unavailable when the masters are down (and showing up lower in the server browser :/)

heinrich5991

please do not lazily diminish other people's work

https://www.merriam-webster.com/dictionary/banter

Definition of BANTER

to speak to or address in a witty and teasing manner; delude; challenge… See the full definition

and if it was, the problem with the normal texture-sampling function should still be tehre

Patiga

and if it was, the problem with the normal texture-sampling function should still be tehre

Why

Ryozuki

and also my pr thanks https://github.com/ddnet/ddnet/pull/6580

why do you add default params if you pass it all the time anyway?

cuz robyt asked

Ewan

Is there a way to get put into ddnet-info.json or do we just have to deal with being unavailable when the masters are down (and showing up lower in the server browser :/)

This was back in the day when the masters weren't http, current masters proved pretty resilient so there isn't need for that anymore. We have an issue somewhere github on how to revamp the internet tab #6506

it's good practice

https://github.com/ddnet/ddnet/issues/6506

Debloat the Internet tab by hiding DDNet and KoG servers · Issue #6...

Thoughts? Two positives I can think of: Minimal Moderation Needed: Official servers are exclusively listed under their designated tabs. People will always be able to locate our servers and refrain ...

Ewan

it's good practice

no

11:47

default params are bad

is that ahot take

i can remove em if u want @heinrich5991

Ewan

is that ahot take

no its as cold as ice

Ewan

Is there a way to get put into ddnet-info.json or do we just have to deal with being unavailable when the masters are down (and showing up lower in the server browser :/)

Fwiw the checkmark has nothing to do with sorting in the server browser

1

I'd use structs to do what default parameters do

11:48

That at least doesn't lead to typing the default value all the time

Learath2

This was back in the day when the masters weren't http, current masters proved pretty resilient so there isn't need for that anymore. We have an issue somewhere github on how to revamp the internet tab #6506

Oh actually it's #5654

https://github.com/ddnet/ddnet/issues/5654

Rename Kog Tab to Community and add trusted servers · Issue #5654 ·...

The current situation is unfair to good faith server owners whose servers don't have the checkmark, a more healthy way is to provide a way to apply to have the checkmark or be added to this new...

I'd also would preferred not passing the value explicitly if it's equal to the default. Default parameters there would mean that forks don't need to change as much when merging this

Jupstar ✪

Do you happen to know if there is any possibility inside a .rs file to hide the keyword unsafe from a user? E.g. in c it's simply #define notunsafe unsafe

open /proc/self/mem

Jupstar ✪

Why

ah you seem to correct with the vulkan feature SampledImageArrayNonUniformIndexing

11:49

hmm wait I should look further first

funny this is actually exactly what i'm asking about

Ewan

funny this is actually exactly what i'm asking about

for unique servers

the places where alpha is epxlicitly and equal to default are mostly because we dont have the mechanisms to get whether our tee and the other differs in team, etc

11:49

less the ones for ur own effects

Patiga

ah you seem to correct with the vulkan feature SampledImageArrayNonUniformIndexing

Oh so an extension?

yeah I think so

i think the registration system is the best bet

But yeah I think that would be the coolest solution

ChillerDragon: to log to the console, use println!? ^^

But since extension not practical for us xd

Ewan

for unique servers

I agree the current situation is less than ideal but there are 2 issues: 1) Figuring out what the new internet tab should look like 2) Finding a developer that has the time

why can't internet tab be everyone

Ryozuki

use leak to give ownership to cpp side

probably not needed if you just call a function. it only needs to live as long as teh function call

let trustworthy parties have the checkmark, don't change how it appears

11:51

would work fine

@Patiga the simplest solution would be if textureOffset would also have a parameter for size

11:52

So textureoffsetanssize

so just servers in peoples' closets and fakes won't have the check

heinrich5991

probably not needed if you just call a function. it only needs to live as long as teh function call

yeah, but i said if you want to give ownership :p

Then atlasses would be useful in an easy way

Ryozuki

https://www.merriam-webster.com/dictionary/banter

this is unclear from the outside. people see people bashing others for doing 0.7 work. this doesn't set a good tone for this chat. please stop doing that, even if it's "just a joke"

god

11:52

xd

11:52

Pray

Ewan

let trustworthy parties have the checkmark, don't change how it appears

It's hard to gauge who is trustworthy, it's very subjective. I'm pretty sure @Ryozuki wouldn't like it if his issue was resolved by me sitting here and handing out checkmarks to people I trust

well it wouldn't just be you

11:53

you elect mods that way don't you?

11:53

clean track record and has been around

11:53

do the same for servers

Jupstar ✪

@Patiga the simplest solution would be if textureOffset would also have a parameter for size

Ok NVM it would cause problems with mipmaps anyway

Learath2

It's hard to gauge who is trustworthy, it's very subjective. I'm pretty sure @Ryozuki wouldn't like it if his issue was resolved by me sitting here and handing out checkmarks to people I trust

CEO Learath gives blue checkmarks out

perhaps we can just get rid of the tabs altogether

oh i like that idea

heinrich5991

this is unclear from the outside. people see people bashing others for doing 0.7 work. this doesn't set a good tone for this chat. please stop doing that, even if it's "just a joke"

i would say the outside is empty, i doubt anyone works for 0.7 at this point, just chiller

Ryozuki

i would say the outside is empty, i doubt anyone works for 0.7 at this point, just chiller

there are a lot of people watching this chat

we should break compatibility, with 0.7, and with old clients

they see that it's apparently okay to be lazily negative

and rework

Ryozuki

we should break compatibility, with 0.7, and with old clients

(and what I see is not just banter here)

well i said banter to chiller

11:55

im not bantering with u

11:56

one can say smth as a joke and still hold it in a serious discussion

Chiller don't be sad, honey

or smth like that

what I mean is that you actually argue your point; i.e. now it seems like you actually meant it when you said that to chillerdragon

well i meant

11:56

it

11:56

nothing wrong with it?

yea, anyway. stop bantering in this chat

yep lets be robots

i don't understand why it's that serious, 0.7 is objectively a much less popular version of the game, therefore in comparison, nearly nobody cares about it

speak formally

Jupstar ✪

Ok NVM it would cause problems with mipmaps anyway

yeah so I would still be interested if you yourself would go with a 1. texture atlas, 2. texture array and stretching all segments to the same size, 3. texture array and some only take up a part of the texture size ^^

Ewan

i don't understand why it's that serious, 0.7 is objectively a much less popular version of the game, therefore in comparison, nearly nobody cares about it

daring today arent we

i've never touched it. i have no bias for or against it

11:57

i'm just saying

I think u just have to write it like this @Ryozuki . 0.7 is not interesting for the future of our project in the current state

0.7 is not interesting for the future of our project in the current state

lmfao

lol no

heinrich5991

perhaps we can just get rid of the tabs altogether

This is what I would do. Remove kog tab (maybe keep ddnet? we do have some extra functionality that idk how to expose to others, like finishes and region filtering), let everyone pass a "group_name", "group_key" in their server info, master server sends sha256(group_key) in the info, client groups by sha256(group_key) under name "group_name"

Learath2

This is what I would do. Remove kog tab (maybe keep ddnet? we do have some extra functionality that idk how to expose to others, like finishes and region filtering), let everyone pass a "group_name", "group_key" in their server info, master server sends sha256(group_key) in the info, client groups by sha256(group_key) under name "group_name"

if we remove kog tab, we remove the ddnet tab IMO

The only issue is people faking entire networks, for which idk what to do

the way to go is using the http master, enable advanced filter options

11:58

like ddnet has

11:58

for everyone

I guess we could let all servers send region, so we can do the region filtering for everyone, the finishes idk how to do

Ryozuki

0.7 is not interesting for the future of our project in the current state

you can say that you're not interested in 0.7 stuff, and you can say that you think it's holding us back

11:58

why would you use such corporate speak? it just looks like lies

heinrich5991

you can say that you're not interested in 0.7 stuff, and you can say that you think it's holding us back

i say both

heinrich5991

why would you use such corporate speak? it just looks like lies

well u dont want banter

11:59

u rly make it seem like we should be corporate speakers

no. I don't want to do that (edited)

11:59

you're allowed to say negative things

11:59

just not lazily

whathever lazily means here

11:59

i have put lot of effort into ddnet

11:59

where is the laziness

did you work on 0.7 heinrich

Patiga

yeah so I would still be interested if you yourself would go with a 1. texture atlas, 2. texture array and stretching all segments to the same size, 3. texture array and some only take up a part of the texture size ^^

I'm honestly struggling with an answer rn. E.g. if u'd write it I'd probably benchmarking it against using more draw calls. For particles it's defs interesting. But for tees that max at like 64 it might not be worth it. But generally spoken I'd probably then accept the badness of texture atlasses and write the sampling itself even if it's harder generally. Texture arrays are simply only interesting for similar sized stuff.. or split the draw calls with same sized stuff

lazy means "not putting work into the criticism"

being emotionally invested in ddnet, and seeing work on something that most likely, in my honest opinion, will end up not relevant or unused, is a bit sad

Ryozuki

whathever lazily means here

I guess "your work is shit" vs "your work is shit because you failed to do x, y and z"

we can just improve ddnet more

he's not even saying it's shit. just that people don't care about it

true, i originally said nobody cares about it

12:01

which is not a total lie

12:01

nobody is exaggerated, there will always be 1 like chiller

12:01

its a way of speaking

Ryozuki

which is not a total lie

for civility, it doesn't really matter whether something is true or not

12:02

the way it's said matters

12:02

people are free to invest their time into whatever they want. time spent on 0.7 is not necessarily time lost for ddnet

12:02

it might just be time lost for playing some game

@heinrich5991 "<heinrich5991> ChillerDragon: to log to the console, use println!? ^^"

he is free to do whathever he wants yes

OMG YES I JUST FIGURED OUT

and me to say that

Anyway, I need to do real world stuff like doing the dishes. @Ewan if you have any ideas about merging all lists put it in the issue, I think the desirables are -> some sort of trust structure -> some way to group servers -> some way to allow advanced filtering for all

awesome Chiller

i fuckin reimplemented all wireshark loggy all day to realize println! wasnt priting because the code was not even compiled xd

wat

12:03

how can println not be compiled

he didn't recompile his code

ah

and didn't sanity-check it

12:03

always try to check whether your changes are actually there

12:04

maybe try to crash at the earliest possible opportunity

Jupstar ✪

I'm honestly struggling with an answer rn. E.g. if u'd write it I'd probably benchmarking it against using more draw calls. For particles it's defs interesting. But for tees that max at like 64 it might not be worth it. But generally spoken I'd probably then accept the badness of texture atlasses and write the sampling itself even if it's harder generally. Texture arrays are simply only interesting for similar sized stuff.. or split the draw calls with same sized stuff

thanks for answering so many questions even forcing an answer to a very opinionated question unprepared heartw

heartw

I should probably try to understand the ddnet rendering code some time and not only small extracts of it happy

happy

heinrich5991

open /proc/self/mem

How does that work btw. Is that something the rust compiler offers?

Jupstar ✪

How does that work btw. Is that something the rust compiler offers?

no, the linux kernel

12:04

you can simply edit your own memory using that file

https://github.com/ddnet/ddnet/issues/5654#issuecomment-1198010235 I think I like this here, maybe we can make a milestone or something so someone can start implementing it?

Rename Kog Tab to Community and add trusted servers · Issue #5654 ·...

The current situation is unfair to good faith server owners whose servers don't have the checkmark, a more healthy way is to provide a way to apply to have the checkmark or be added to this new...

and easily do unsafe code

with root right?

heinrich5991

you can simply edit your own memory using that file

Ah yeah I see

Ryozuki

with root right?

no, for your own program

why is that allowed tho

I think I read that before

Ryozuki

why is that allowed tho

Why not?

Ub in safe rust xd

Ryozuki

why is that allowed tho

why not? the program can already do it via pointers

this

1

Learath2

Anyway, I need to do real world stuff like doing the dishes. @Ewan if you have any ideas about merging all lists put it in the issue, I think the desirables are -> some sort of trust structure -> some way to group servers -> some way to allow advanced filtering for all

i think the checkmark's presence should be based on some kind of vote, whether a server is verified or not be determined by what the master server says & not any local file. if the server is more robust now then i see no issue in this. as for advanced sorting, idk what that entails so i have no comment for now

i guess

now that i built the logger would you be interested heinrich?

Couldn't the rust Api prevent it xd

Embedded image

The standard impl

ChillerDragon: if you integrate it with the rust logging system, yes

wot is rust logging system

Jupstar ✪

Couldn't the rust Api prevent it xd

If you want to expose fs to your user, you need to make your api forbid these things

ChillerDragon

wot is rust logging system

check tracing

12:06

https://docs.rs/tracing/latest/tracing/

A scoped, structured logging and diagnostics system.

I'd probably go with log here, but they're interoperable

log is old news

12:07

tracing is the future

log is math

https://docs.rs/log/latest/log/

A lightweight logging facade.

idk what u guys talk about

chiller u know what a logging facade is

my client just crashed when tabbing into your server ChillerDragon

whats the crash log

12:07

what server?

Learath2

If you want to expose fs to your user, you need to make your api forbid these things

you basically can't

my coding skill is a facade ryo

symlinks will fuck up your day

dont be rude to urself chiller

will i get in trouble with heinrich for being rude?

nvm i'm silly

idk

i mistook one of these ki-o ones for one of urs

Learath2

If you want to expose fs to your user, you need to make your api forbid these things

I'd not allow any Io directly

anyways heinrich here is what i have what u said sounds like work i will leave it here https://github.com/heinrich5991/libtw2/compare/master...ChillerDragon:libtw2:pr_add_ws_log

Comparing heinrich5991:master...ChillerDragon:pr_add_ws_log · heinr...

Some Teeworlds stuff in Rust. Contribute to heinrich5991/libtw2 development by creating an account on GitHub.

12:09

yo @heinrich5991 you got motivation to help me debug https://github.com/heinrich5991/libtw2/pull/78 ?

gamenet: Add 0.7 connless inf3 (closed #77) by ChillerDragon · Pull...

Generated using cd gamenet/generate python3 serialize.py --version 0.7 raw/teeworlds-0.7.5.py > spec/teeworlds-0.7.5.json

12:10

it doesnt show up in the dissector somehow ;_;

ChillerDragon: you only need to do this: https://docs.rs/log/latest/log/#implementing-a-logger

A lightweight logging facade.

12:10

you'll learn a bit about rust

i rust pro no need to learn

Chairn

we would lose thematic games doing so, but we can maybe split them in software and allow a lot more option directly in game (like a split game icons option)

inkscape allows you to export in batches

i would also make dbg_log a safe wrapper, somehow

@heinrich5991 i dislike learning i am a result driven developer i only try to unserstand or learn something if I REALLY have to. otherwise its try and error until i get my gaming moment results

I see

@Patiga do u still use texture arrays for tee rendering? That's a part where I'd say it's more complex, and I'd love to see the performance cost of updating the array ^^

ChillerDragon

anyways heinrich here is what i have what u said sounds like work i will leave it here https://github.com/heinrich5991/libtw2/compare/master...ChillerDragon:libtw2:pr_add_ws_log

permalink btw: https://github.com/heinrich5991/libtw2/compare/heinrich5991:0352e6f...ChillerDragon:c35fe8c

E.g. downloaded skins

ye the link i sent is totally scuffed xd

heinrich5991

permalink btw: https://github.com/heinrich5991/libtw2/compare/heinrich5991:0352e6f...ChillerDragon:c35fe8c

u can use <> to not embed btw

yea… unfortunately it changes the message I send to IRC

but irc should have no problems tho

12:13

hmm now i wonder

so wen i compile the dissector shouldnt it also compile gamenet/*/src/msg/connless.rs ?

alacritty highlights the link correctly for me

12:13

no <> in the url

ChillerDragon

so wen i compile the dissector shouldnt it also compile gamenet/*/src/msg/connless.rs ?

that's not where the dissector gets its data from

12:14

it gets it from the json

12:15

https://github.com/heinrich5991/libtw2/blob/0352e6f2f23e01e17c6212701f8faa166c524522/wireshark-dissector/src/tw7.rs#L24

Ravie

yeah?

read log !! it's about deleting images atlas and splitting it into n files with one doodle per file

yes but i generated the json too

Chairn

read log !! it's about deleting images atlas and splitting it into n files with one doodle per file

I'd say we should put it into a zip file then

12:16

so it can still be shared as a whole

Chairn

read log !! it's about deleting images atlas and splitting it into n files with one doodle per file

so like splitting game.png into gun.png, bullet.png, grenade.png, hook.png... ?

heinrich5991

symlinks will fuck up your day

refuse to follow any symlinks

symlinks are a security hazard

Learath2

refuse to follow any symlinks

/home/user/linktoprocself/mem

Jupstar ✪

Tbh just stop using texture atlasses all along

@Ravie start from here

can you ask the linux kernel not to follow this?

eeeh idk if linux has a way to avoid TOCTOU there but you can stat it first to see if any path component is a symlink

12:18

but I guess then hardlinks fuck up your day

12:19

(though I think in this case the person who writes the potentially malicious code has access to the underlying system, so no need to worry about symlinks if you don't allow them to create them)

can't hardlink directories

How does java prevent it?

Jupstar ✪

How does java prevent it?

it doesn't

Mh ok

12:20

Why does Linux offer it at all xd

probably useful for crashdumps or so

Jupstar ✪

Why does Linux offer it at all xd

As said before, if you can access proc mem self you can also just use pointers

or any kind of meta functionality

Learath2

As said before, if you can access proc mem self you can also just use pointers

Yeah but java is meant to run in a VM for a reason xD

@Learath2 assuming no symlinks, how do you detect whether the file being opened is something like /proc/self/mem?

Oh, actually why not run it in a jail?

Jupstar ✪

Yeah but java is meant to run in a VM for a reason xD

you can also call native code in java

Ravie

so like splitting game.png into gun.png, bullet.png, grenade.png, hook.png... ?

exactly

heinrich5991

you can also call native code in java

Yeah but that's easy to prevent

u can also call java code from rust

I'd say

heinrich5991

@Learath2 assuming no symlinks, how do you detect whether the file being opened is something like /proc/self/mem?

I can think of one silly way, but it's just a thought experiment. Just generate a pattern in memory at runtime, then scan whatever you opened for it

Learath2

I can think of one silly way, but it's just a thought experiment. Just generate a pattern in memory at runtime, then scan whatever you opened for it

heh, nice

Jupstar ✪

@Patiga do u still use texture arrays for tee rendering? That's a part where I'd say it's more complex, and I'd love to see the performance cost of updating the array ^^

yup still on it :D

i read recently valve tracked memory access at places where it was never meant to access to detect cheats

Chairn

exactly

but why tho? it would break all existing skins and make new ones an even bigger pain to create

Ravie

but why tho? it would break all existing skins and make new ones an even bigger pain to create

script to port

12:23

since skins are structured its easy

yeah ok

12:24

but also if you make a gameskin then you have to hope that people install all your files and not turn it into some disgusting mix with parts from other ones

why?

12:25

do u care what others do

zip files:

Xd

Ravie

but also if you make a gameskin then you have to hope that people install all your files and not turn it into some disgusting mix with parts from other ones

I think this could be handles by zip files

but why limit

12:25

the idea is customization

U remove Atlas to use zip

12:25

Lmao

ah u mean for shipping

Jupstar ✪

U remove Atlas to use zip

this way we can name the parts and add/remove them without backcompat problems (edited)

we all had the same idea but different ways of saying it

backcompat again justatest

justatest

FBI

without future backcompat problems

open up!

XD

it does make it a little more annoying to share skins though

12:26

it's no longer a single picture

12:26

like in #📬submit-skins

make a vote

12:27

separated skins + more customizatio or single file skins and status quo

i guess but at the same time, its muuuch easier to organize and add new items

Jupstar ✪

U remove Atlas to use zip

remove atlas so you can upload more textures in hopes of breaking older computers bigbrain

bigbrain

(stating that old skins are not lost)

Learath2

remove atlas so you can upload more textures in hopes of breaking older computers bigbrain

bigbrain

the way

let's be honest, "more customization" will just make people turn their game into an even bigger visual cacophony

is it even more customization?

12:28

people can edit a skin file today

u can mix parts easily in the client

heinrich5991

is it even more customization?

it is easier customization, not more

yea

Ryozuki

u can mix parts easily in the client

ah, that's a different thing you're asking for though

Ryozuki

u can mix parts easily in the client

i think thats what ravie is arguing against

we don't need to change the file format for this

yeah but why not

Ryozuki

yeah but why not

"why not" is not a reason to change something in my book

12:30

especially user-facing

my book is the holy bible so it must be respected

for things that are not user-facing, maybe…

Yeah, change for changes sake isn't that great of an argument

well i think its better organized

12:30

i forgot what other reasons are there

It is better UX if people want to mix and match assets all the time

12:30

but I'm not sure that is a common demand even

Ravie

let's be honest, "more customization" will just make people turn their game into an even bigger visual cacophony

people already mix & match

Ewan

people already mix & match

mix & ~~match~~

if anything i just wanted to remove useless pngs

kind of odd to limit others on how they wanna mix the skins

12:31

why do u care

Ravie

mix & ~~match~~

deep

Ryozuki

kind of odd to limit others on how they wanna mix the skins

depends on whether this shows on other people's screens

Ryozuki

why do u care

hes a perfectionist

I care about the fact that people opening the default installation of the game see something like tees

12:32

when joining a server

walking hammers

yeah but default will look the same

well, it depends whether you want to propagate your visuals to other players

12:33

it sounds like you want to?

12:33

i.e. that everyone sees how you mixed your skins?

by skins i was talking about hud elements

12:34

too

12:34

but yeah for skins that would happen

12:34

for example mixing laser from x and grenade fro my

12:34

from y

Ryozuki

kind of odd to limit others on how they wanna mix the skins

cause I think changing the look of the game in thematic batches is better than pure chaos, you can already do it if you edit the png's but the game isn't actively supporting and encouraging you to do it

yea, I don't really care how people view the game oneself

i was mainly talking about that

12:34

didnt think about skins

12:35

tees

heinrich5991

yea, I don't really care how people view the game oneself

people who make custom assets do, you don't make it with the purpose of being split for parts and mixed in weird ways that don't look good, it defeats the purpose of even making a set

let's say "you do"?

12:40

that's fine, I think we should take that into account before changing anything, since you're obviously one of the people creating custom assets

yeah I know it's a minority but you should still consider what the purpose of making custom assets really is

i disagree with the statement but meh

12:41

its like saying i coded X in a way, and it has a permissive license, but i dislike you doing Y with it

Ravie

yeah I know it's a minority but you should still consider what the purpose of making custom assets really is

its for users to use them in whathever way they want

1

12:42

in my opinion

Learath2

remove atlas so you can upload more textures in hopes of breaking older computers bigbrain

bigbrain

Btw we do that anyway internally ^^

Jupstar ✪

Btw we do that anyway internally ^^

Good good, now we need to find a way to up the vram usage

Learath2

Good good, now we need to find a way to up the vram usage

True. 1gb on my 16gb limit is not enough

12:44

That's less than two tabs of Firefox xd

Ryozuki

its for users to use them in whathever way they want

well for me personally if my work isn't gonna be used how I intended then there's no point in even making or sharing it

we differ then

12:45

for me the amazing thing is how people find ways to use stuff in unexpected or never thought of ways, such as making a programming language

also it's up to the creator how he licenses his work and what he wants users to do or not do with it

insert alan turing quote here

programming is much less subjective than art

@Learath2 i just want to make one thing clear. I am not wasting memory, it has a purpose. GC languages that want to be fast kinda have to waste memory by design

12:48

so i'm new hardwarephil, because it offers more opportunities for better software design and more performance besides the performance increase by the generational improvement that comes anyway

12:48

isn't that awesome

imho if more memory usage leads to more perf, its not that wasted memory is it

that's awesome — as long as we don't push users away from their old hardware IMO

Ryozuki

imho if more memory usage leads to more perf, its not that wasted memory is it

depends on whether the increased perf has a positive effect or is just neutral

Ravie

programming is much less subjective than art

but the art here is used in a program

for me personally, it probably doesn't matter, because I don't run the game on a lot of FPS. for others, it does matter because they want to have very high FPS

anyway art benefits a lot from what i said too, derivative works and such

12:50

fanart

12:51

mix match xd

Jupstar ✪

so i'm new hardwarephil, because it offers more opportunities for better software design and more performance besides the performance increase by the generational improvement that comes anyway

hardwarephil is not a word sadly

heinrich5991

for me personally, it probably doesn't matter, because I don't run the game on a lot of FPS. for others, it does matter because they want to have very high FPS

but e.g. m1 is a good example, more energy efficient than your laptop, while still having more fps, more RAM etc. just that apple sucks

12:52

and their drivers xD

Ryozuki

hardwarephil is not a word sadly

the word would also have to be new_hardwarephil xd

12:52

bcs the new is important

new_safe_hardwarephil

Ryozuki

anyway art benefits a lot from what i said too, derivative works and such

I have yet to see a "derivative work" that comes anywhere close to the original in our community justatest

justatest

justatest

1

maybe we idealize teeworlds original art too much

12:53

cuz nostalgia or smth

12:53

its just a ball

Ravie

I have yet to see a "derivative work" that comes anywhere close to the original in our community justatest

justatest

most tee skins are derivative works of the original tee

Ryozuki

maybe we idealize teeworlds original art too much

I don't mean just original teeworlds art, I often see people mixing skins from Miper, Whis or Demix and it always looks terrible

well its hard to please u, but i understand, just like i find windows terrible

12:56

@Learath2 r u still interested in https://github.com/ddnet/ddnet/pull/2754

Check for the integrity of the data directory by Learath2 · Pull Re...

Isn't finished yet but it'd be nice if you could take a look before I spend more time on it. Fixes #2748

12:57

@heinrich5991 https://github.com/ddnet/ddnet/pull/1264#issuecomment-1408523015

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&...

12:57

the evening is lasting long xd

Ryozuki

well its hard to please u, but i understand, just like i find windows terrible

just speaking from the perspective of a huge perfectionist, and I can understand how much work other artists also put into making every pixel look just right, and then it gets twisted by some kid with gimp

dont insult my gimp

12:58

BASEDDEPT

I love gimp

poggers2

1

12:58

I use all the open source alternatives to adobe products gigachad

gigachad

chad

12:59

but u use windows?

12:59

frozen

yes but not 11

12:59

when they discontinue win10 I go the linux way saddo

saddo

Ryozuki

@Learath2 r u still interested in https://github.com/ddnet/ddnet/pull/2754

I'm not sure. I don't like how storage works at all. I have a feeling this will break some setups because of how we find data directories

i forgot, does realloc zero memory?

every sewcond windows is unusable

Ravie

just speaking from the perspective of a huge perfectionist, and I can understand how much work other artists also put into making every pixel look just right, and then it gets twisted by some kid with gimp

Glow skins lets go

even more unusable than it already is xd

Ryozuki

i forgot, does realloc zero memory?

No

ok

13:01

i guess i need to get memzero in too

true designers use adobeOS over windows

wait libc has memzero?

memset(a, 0, sizeof(a))

1

ah ok

note the funny parameter order

note that it "has" to use int as parameter

13:02

even tho we talk about bytes

heinrich5991

memset(a, 0, sizeof(a))

(dst, src, dst_size) isn't it like this for all mem functions?

Learath2

(dst, src, dst_size) isn't it like this for all mem functions?

I'd have thought it's the src_size

why are memory functions in string.h?

13:03

xd

13:03

anyway the header doesnt matter cuz im on llvm ir

Ryozuki

why are memory functions in string.h?

Shhhhhh

because strings are just arrays of bytes

13:03

mem* functions operate on strings that are not zero-terminated

Ryozuki

anyway the header doesnt matter cuz im on llvm ir

Doesn't llvm have a way to set some memory without calling into libc?

str* functions operate on strings that are zero-terminated

Learath2

Doesn't llvm have a way to set some memory without calling into libc?

to set memory allocated by malloc?

13:04

maybe there is a intrinsic

Memory allocated by malloc isn't any different to any other memory

https://llvm.org/docs/LangRef.html#llvm-memset-inline-intrinsic

13:05

oh pog

Perfect

@Ryozuki name ur wrapper crealloc

its ok

13:06

just need to get my open addressed linear probing hashmap done

recalloc

13:06

sounds funny too

i recall freeing this memory

13:06

said every c dev

but i'll do it again just to be safe

13:09

✅

declare void @llvm.memset.p0.i64(ptr <dest>, i8 <val>,
                                 i64 <len>, i1 <isvolatile>)

13:09

seeing this signature reminded me

13:09

why memset accepts a int (32 bits)

free(malloc(0)); // you are free now, c is freedom

and not a char

it does? >_>

#include <string.h>
void *memset(void *s, int c, size_t n);

yea why thoe

13:11

share with the class

13:11

i thought it was char

idk thats why am askin

oh

Ryozuki

why memset accepts a int (32 bits)

i thought it was a statement lol

13:11

mayeb just for filling a pattern

Ryozuki

and not a char

not unsigned char u meant xd

ye

13:11

i forget about C char stuff

It couldn't be a char it would be a unsigned char

xd

correct but signedness is not the question here

there are many questions to which there aren't answers

it seems char literals are actually ints in plain C

@Ryozuki what's worse, a C program that generates rust code or a rust program that generates C code?

maybe saves a cast

Ravie

@Ryozuki what's worse, a C program that generates rust code or a rust program that generates C code?

rust program that generates c

13:14

cuz the rust program is safe (if no unsafe)

"memset predates (by quite a bit) the addition of function prototypes to C. Without a prototype, you can't pass a char to a function -- when/if you try, it'll be promoted to int when you pass it, and what the function receives is an int."

1

it may also be a speed thing. sometimes smaller data is slower to access

13:14

eg why some bools used to also be ints

13:14

despite only needing 1 bit

Ryozuki

cuz the rust program is safe (if no unsafe)

but what about the program that it generated

i think it depends man

"prototypes were added during the C89 standardization. Before that, there were only function declarations, which specified the return type, but not the number or type(s) of parameter(s). "

13:15

ez

a rust program that generates c could realistically describe any of these new programming languages

13:16

x new lang is like zig in the fact that it makes c, but it's written in rust

13:16

could happen any second...

Jupstar ✪

"prototypes were added during the C89 standardization. Before that, there were only function declarations, which specified the return type, but not the number or type(s) of parameter(s). "

What are the quotes for? is that GiPiTed?

zogtib

What are the quotes for? is that GiPiTed?

i teach ryo some awesomeness of c

Ewan

it may also be a speed thing. sometimes smaller data is slower to access

idk

by copying stackoverflow answers

i think you google it xd

the thing is in llvm ir, you can have a data type of arbitrary bits

13:17

and llvm can use that to optimize better

13:17

i6 is valid

yeah

llvm also have infinite registers

bit packing is done under the hood

its the perfect machine

13:17

xd

std bitset is similar

13:17

a non 8*x bitfield

Ryozuki

and llvm can use that to optimize better

does it actually do that or do the non-register widths perform worse?

im sure they perform worse usually

13:18

but it can make optimizations to non 8*x size ints that it couldn't to 8*x ints because they're already 'optimized'... (edited)

Ewan

im sure they perform worse usually

bitset performs worse or the inifinite registers?

heinrich5991

does it actually do that or do the non-register widths perform worse?

i would say it probs treats the underlying data as register widths, but it can make assumptions about the contents (?) and do optimizatiosn thanks to that

13:19

but dont quote on me

13:19

i dont know for sure

Jupstar ✪

bitset performs worse or the inifinite registers?

i6

ah

13:21

here ryo more c awesomeness struct whatever { char a:3; char b:4; char c:1; };

like im sure in an application where you're only using 6 bits, using an i8 would still be faster but i6 would be the better candidate because under the hood bit packing can be done to reduce memory footprint

Jupstar ✪

here ryo more c awesomeness struct whatever { char a:3; char b:4; char c:1; };

ye i know xd

but i'm not sure because then pointers become impossible

Ewan

like im sure in an application where you're only using 6 bits, using an i8 would still be faster but i6 would be the better candidate because under the hood bit packing can be done to reduce memory footprint

i4 is used a lot in machine learning iirc

machine learning has gpu usually

13:22

idk how those work too well

13:23

magic devices

well llvm targets gpus too

yea

13:23

i have cpu bound brain

Ewan

like im sure in an application where you're only using 6 bits, using an i8 would still be faster but i6 would be the better candidate because under the hood bit packing can be done to reduce memory footprint

i wouldn't count on it xd

i can't see how i6 would be faster even theoretically (edited)

Ewan

i can't see how i6 would be faster even theoretically (edited)

well theoretically it surely can bcs of the memory footprint, but practically u are probably right in almost all cases

13:26

but somewhere someone finds one case where its not the case also depends if u mean that it uses less memory in total

13:26

else ofc not xd

alignment is the thing here

13:26

we are speaking in bits

yeah thats the question did u mean packing multiple data efficiently

13:27

or only a single i6

multiple

then hard to say, probably its slower, but maybe there is some edge case

single wouldn’t really make a different usually because it would probably just be given those extra 2 bits

13:27

cuz you cant pack the next data

similar to if loading a zip and uncompress it might be faster than loading the whole file as is

Ewan

single wouldn’t really make a different usually because it would probably just be given those extra 2 bits

but the calculations with it would be a problem

in practice accessing a non-multiple of 8 piece of data is slower because the cpu will have to first read that data then derive from it the relevant bits

but that mostly happens on the cpu

13:29

with registers

Ryozuki

idk thats why am askin

Ancient signature, probably from back before c89 even, so it'd be promoted to int by default

anyway before this leads to useless discussions i dunno, but would be interested with some benchmarks for some weird edge cases i cant think of rn

Fwiw "char" literals are just ints aswell for the same ancient reasons

in C (edited)

13:30

not in C++

^

int is_cxx = sizeof(char) == sizeof('C');

lol

xd

sizeof (char) is defined to be 1 anyway

i understand unsigned char is akin to "bytes" in sane languages signed char for the numeric type, and char?

So might aswell check sizeof 'c' == 1

char for text? xd

char is randomly signed or unsigned xdd they roll some dices

Ryozuki

i understand unsigned char is akin to "bytes" in sane languages signed char for the numeric type, and char?

Eeeeeeeeeh

bytes being u8

arm got unsigned x86 got signed

&[u8] rusty bytes

unsigned char is for bytes. char is for text. Try not to think too much of how it's handled internally

char8_t is for utf8 (edited)

i love my &str

xd

char is neither the same as signed char nor unsigned char

ohno

3

char8_t doesnt ensure its valid utf8 tho

13:35

the epic thing in rust is str is valid utf8 always

13:35

or ub

Jupstar ✪

char is randomly signed or unsigned xdd they roll some dices

plus this fact

utf8 supremacy

in a perfect world, there wouldnt be any utf

plus char/signed char/unsigned char could be non 8-bits

was there ever a CPU without 8 bits per byte?

13:37

probably

13:37

back in the ages

13:37

194x

i remember hearing such a thing

why does memset return the passed pointer?

13:37

should i use the new pointer

it's the same

Jupstar ✪

was there ever a CPU without 8 bits per byte?

The issue isn't the cpu, I think the issue was how the data was to be encoded in transmission

whats wchar_t

ohno

3

wide char

13:38

2 byte

13:38

for non ascii

The wchar_t type is an implementation-defined wide character type. In the Microsoft compiler, it represents a 16-bit wide character used to store Unicode encoded as UTF-16LE, the native character type on Windows operating systems

13:38

oh god

Ryozuki

whats wchar_t

xD

oh my god have u never messed with this

i know windows api loves it

Ryozuki

why does memset return the passed pointer?

maybe for chaining or something, like char *p = memset(alloc(...), ...)

try fucking specifying DoShitW on every win32 function ever

and thats why i never touched it

yea

13:38

painful

Just use utf8

Robyt3

maybe for chaining or something, like char *p = memset(alloc(...), ...)

makes sense

New winapi can mostly handle it too

wdym

if we only wanted to support Windows 10 and newer then we could also just use UTF-8 with the Windows API

Robyt3

if we only wanted to support Windows 10 and newer then we could also just use UTF-8 with the Windows API

LETS GO

without converting between char and wchar_t for every function

this is how they want you to do it

#ifndef UNICODE
#define UNICODE
#endif

DoSomething(); // is a macro; resolves to DoSomethingW, a function

#ifdef UNICODE
#undef UNICODE
#endif

DoSomething(); // is a macro; resolves to DoSomethingA, a function

(edited)

Robyt3

if we only wanted to support Windows 10 and newer then we could also just use UTF-8 with the Windows API

the conversion happens on the other side then, no performance on

wmemset ftw

13:40

wchar_t *wmemset(wchar_t *wcs, wchar_t wc, size_t n); (edited)

pain

Ewan

this is how they want you to do it

#ifndef UNICODE
#define UNICODE
#endif

DoSomething(); // is a macro; resolves to DoSomethingW, a function

#ifdef UNICODE
#undef UNICODE
#endif

DoSomething(); // is a macro; resolves to DoSomethingA, a function

(edited)

you can also (and maybe should) just call DoSomethingW directly

memset w/ sizeof type * count

heinrich5991

you can also (and maybe should) just call DoSomethingW directly

depends

This is one thing I like about newer languages. Proper unicode support

yeah, we always call the W variant explicity even though we define UNICODE and _UNICODE

Learath2

This is one thing I like about newer languages. Proper unicode support

yep yep

13:41

@Learath2 but some chose to go the evil way and use utf 16

Robyt3

yeah, we always call the W variant explicity even though we define UNICODE and _UNICODE

yeah this is good the only time this'll bite you in the ass is if using a system that somehow doesn't support unicode (old)

Windows doesn't even use utf16 technically. They use UCS2

Ryozuki

wchar_t *wmemset(wchar_t *wcs, wchar_t wc, size_t n); (edited)

u know whats funny? str functions accept the number of characters istead of the buff size so wchar_t a[10]; wsprintf(a, std::size(a), L"test %s", L"test"); would be correct while with sizeof(a) would be wrong

Ewan

yeah this is good the only time this'll bite you in the ass is if using a system that somehow doesn't support unicode (old)

lol, that's win98?

yea

13:42

note somehow and (old)

Learath2

Windows doesn't even use utf16 technically. They use UCS2

no, not UCS-2. they use possible-invalid UTF-16

When migrating to utf16 they left some bizarre quirks that make using winapi pain

i blame c++20

13:43

why didnt they take the chance with char8_t and extend the std::filesystem to allow utf8 characters

13:43

and what happens under the hood is not our problem

utf8_t

13:44

c# calls it Rune

https://devblogs.microsoft.com/commandline/windows-command-line-unicode-and-utf-8-output-text-buffer/ I probably sent this blog post a while back but until like 2018 the console was extremely bizarre

Windows Command-Line: Unicode and UTF-8 Output Text Buffer

In this post, we’ll discuss the improvements we’ve been making to the Windows Console’s internal text buffer, enabling it to better store and handle Unicode and UTF-8 text. Posts in the Windows Command-Line series: This list will be updated as more posts are published: Command-Line Backgrounder The Evolution of the Windows Command-Line Inside th...

https://learn.microsoft.com/en-us/dotnet/api/system.text.rune?view=net-7.0

Rune Struct (System.Text)

Represents a Unicode scalar value ([ U+0000..U+D7FF ], inclusive; or [ U+E000..U+10FFFF ], inclusive).

@heinrich5991 sori for all the oopsie daisys you probably could have done it twice in the time you spent reviewing my mess

std::filesystem is the worst

13:45

because it works pretty much only in theoreticals

13:45

that's probably the one thing i genuinely prefer doing the C way

what r u even working for? r u some secret microsoft dev? xd

I never used std::filesystem. What's wrong with it?

it's just strings

TIL about https://linux.die.net/man/3/iob_new

iob_new(3): create new I/O batch - Linux man page

iob_new creates a new I/O batch with enough space allocated for hint_entries entries (buffers or files). This is purely a performance hint, if you are unsure just pass 1.

@Ewan are u playing ddnet?

13:46

or just chilln here

everyone with less than 1k ddnet points should be read only in this channel

xdd

xDD

i dont wanna hear this rando robyt talk about a game he doesnt play

im trying to play more lately, but i got 7k points anyway

13:48

lmao

Ryozuki

TIL about https://linux.die.net/man/3/iob_new

Ooh, I hadn't seen this before

robyt is our saviour

!rob its a joke ofc we all love you <3

13:48

!robstr

13:48

!robster

13:48

omg xdf

13:48

!roby

13:48

fuck

bot is dead

how to ping roby

13:48

!ping

pong

good question

13:48

!robstar

!bugfixer

!roberster

weeb pfp = good dev

13:49

@Jupstar ✪ get a vulkan waifu pfp

!robster

13:49

WTRF

u dont even know ur bot

Ryozuki

@Jupstar ✪ get a vulkan waifu pfp

give me midjourney access

13:49

and i generate one

13:49

!robster

ok wait

!jup

@Jupstar ✪

https://github.com/ChillerDragon/ddnet-ircbot/blob/adfc747e1eb186fce903fa977939e6d50386b5ac/ping_pong.csv#L9

ddnet-ircbot/ping_pong.csv at adfc747e1eb186fce903fa977939e6d50386b...

irc bot for ddnet#developer. Contribute to ChillerDragon/ddnet-ircbot development by creating an account on GitHub.

13:50

it says here ffs

stable diffusion tho

chiller forgot to reload?

maybe

13:50

lemme rebuild docker image

13:50

docker moment

Ryozuki

stable diffusion tho

sadly its too bad :/

kio

Use POST and pass Json params (edited)

we don't have anything in the source code to build JSON, we always build it manually using str_format

it gets better and beter

13:50

but still too bad

heinrich5991

we don't have anything in the source code to build JSON, we always build it manually using str_format

oh jesus really?

13:51

they should use the lib then

heinrich5991

we don't have anything in the source code to build JSON, we always build it manually using str_format

I still need to port the CJsonWriter from upstream

@kio other than that, there's code in src/engine/server/register.cpp that uses HTTP to post some json. search that file for Json

Yeah, I wanted to do a json builder too for a while now

heinrich5991

we don't have anything in the source code to build JSON, we always build it manually using str_format

yes but the problem was do it with post

heinrich5991

@kio other than that, there's code in src/engine/server/register.cpp that uses HTTP to post some json. search that file for Json

oh nice

13:51

Image attachment

13:51

i did this work around ahah

!rob i added upstream json writer in #5660

https://github.com/ddnet/ddnet/issues/5660

[DRAFT] Start working on 0.7 skins menu by ChillerDragon · Pull Req...

I started working on this a while ago and it is still far from being finished. Just wanted to share that I am working on this and get some feedback if it even has a chance to be merged when finishe...

13:52

@Robyt3

EZ

what does the json writer do?

13:52

does it use rtti or what

it probably just formats

13:52

using variant union thing

oh shit nvm, it requires python 3.10 but i removed it on gentoo

it does system("jq %s > %s")

ChillerDragon: I'm not sure if the constants SERVERINFO_LEVEL_MIN/SERVERINFO_LEVEL_MAX add something. I'd either remove them or make them into an enum that names the variants, I think

13:53

ChillerDragon: after this is resolved (e.g. by discussing it), could you squash your PR?

Ryozuki

oh shit nvm, it requires python 3.10 but i removed it on gentoo

u can use it on huggingface for free

ChillerDragon: nice. if you have the time, extract just the json writer in another PR already, so we can merge it separately

constants is the first thing i found to name ranges

its sadly even faster than my PC XD

ChillerDragon

constants is the first thing i found to name ranges

why do you want a name though? it's 0-2 and cannot be changed

idk i thought its good style

13:54

but the info got lost in the rendered json anyways so we can also get rid of it

then let's get rid of it

oke sir

OMG

13:54

we need to embrace this

13:54

https://linux.die.net/man/3/bstring

bstring(3): byte string operations - Linux man page

The byte string functions perform operations on strings (byte arrays) that are not necessarily null-terminated. See the individual man pages for descriptions of each function.

Ryozuki

https://llvm.org/docs/LangRef.html#llvm-memset-inline-intrinsic

i wouldn't call that intrinsic, it's still target dependent. Real intrinsic would be 0 page on DRAM directly

squash into one commit with generated code?

Chairn

i wouldn't call that intrinsic, it's still target dependent. Real intrinsic would be 0 page on DRAM directly

its a llvm intrinsic, it handles the target dependent thing for u

Ryozuki

https://linux.die.net/man/3/bstring

also known as mem*

Does anyone know if it's possible to edit the data of a demo ?

Chairn

i wouldn't call that intrinsic, it's still target dependent. Real intrinsic would be 0 page on DRAM directly

it also allows u to mark volatile

it took me so long to get the constants working :( but ye i see ur point

13:56

waste of time

you can also put it into an enum, if you want names, ChillerDragon

Jupstar ✪

char is randomly signed or unsigned xdd they roll some dices

biggest troll ever, already got beaten by that several times

its also not obvious when looking at the json where the number comes from tho right?

Jupstar ✪

was there ever a CPU without 8 bits per byte?

first intel CPU were 4 bits iirc

Chairn

biggest troll ever, already got beaten by that several times

it's really good fun. sometimes you have to say < 0 and sometimes you have to say > 127 to check for ascii

Pipou

Does anyone know if it's possible to edit the data of a demo ?

I have a problem where I see teams that I don't want to see in the rendered video

ChillerDragon: the JSON tries to be machine readable. so it will resolve any constants

btw, lots of circuits are also 1 bit technically but very fast

yea

13:57

so lets get rid of it

13:57

otherwise it has no readibility improvements anyways

13:57

nobody looks into loader.py

okay

13:58

The memfrob() function encrypts the first n bytes of the memory area s by exclusive-ORing each character with the number 42. The effect can be reversed by using memfrob() on the encrypted memory area.

>

Note that this function is not a proper encryption routine as the XOR constant is fixed, and is only suitable for hiding strings.

Pipou

Does anyone know if it's possible to edit the data of a demo ?

its not impossible but there isnt such tool. if u lucky replace the game tee skins with transparent skins (edited)

how did this function get into libc?

heinrich5991

The memfrob() function encrypts the first n bytes of the memory area s by exclusive-ORing each character with the number 42. The effect can be reversed by using memfrob() on the encrypted memory area.

>

Note that this function is not a proper encryption routine as the XOR constant is fixed, and is only suitable for hiding strings.

no way xD

@Robyt3 i wanted to do that at first to make step by step stuff easier and less git conflicty but thought nobody would merge a unused upstream feature

heinrich5991

The memfrob() function encrypts the first n bytes of the memory area s by exclusive-ORing each character with the number 42. The effect can be reversed by using memfrob() on the encrypted memory area.

>

Note that this function is not a proper encryption routine as the XOR constant is fixed, and is only suitable for hiding strings.

This is not standard C, right??

Jupstar ✪

its not impossible but there isnt such tool. if u lucky replace the game tee skins with transparent skins (edited)

That sounds like a clever solution, will do (can't get rid of other players' particles but it should be good enough already)

Pipou

I have a problem where I see teams that I don't want to see in the rendered video

does /specteam works in demo?

Learath2

This is not standard C, right??

no, just a libc function

14:00

glibc, probably

Seems like gnu c yeah

Pipou

That sounds like a clever solution, will do (can't get rid of other players' particles but it should be good enough already)

maybe create an issue about this as new idea/feature

Chairn

does /specteam works in demo?

Nope, I can't even see the team

14:01

Both are currently in team

Image attachment

yeah

14:02

the network code is weird

14:02

it only sometimes sends the information

heinrich5991

The memfrob() function encrypts the first n bytes of the memory area s by exclusive-ORing each character with the number 42. The effect can be reversed by using memfrob() on the encrypted memory area.

>

Note that this function is not a proper encryption routine as the XOR constant is fixed, and is only suitable for hiding strings.

xd

Ah

did someone say network code?

14:02

i became actual tw netcode enjoyer

no

14:03

i think its overcomplex

it is

14:03

demo code is all over place

ye xd

ChillerDragon: wait. you learned the netcode without any immediate benefit. I thought you did not like learning

i had to learn tw netcode which was a very long and frustrating process to get my desired outcome of uniting all tw playerbases 0.6 and 0.7 and be able to play ctf again with a client that allows me to control the mouse

ChillerDragon

i had to learn tw netcode which was a very long and frustrating process to get my desired outcome of uniting all tw playerbases 0.6 and 0.7 and be able to play ctf again with a client that allows me to control the mouse

I think it might have been easier to fix whatever mouse trouble you have with ddnet

teeworlds*

14:08

i need ddnet client to work on 0.7 to play ctf with OG ctf players

14:08

because teeworlds client is weird

does your server not support both 0.6 and 0.7?

some of my servers yes

14:09

but no ctf server

I see

especially not vanilla ctf servers

14:09

and also i think it makes a lot of sense to not split the game into ddnet and teeworlds

14:10

i dont wanna miss out on whats happening in 0.7 or switch clients

[ddnet/ddnet] Pull request opened: #6582 Shutdown SDL after kernel and client

Register SDL for cleanup before creating the kernel and client, so SDL is shutdown after kernel and client. Otherwise the client may crash when shutting down after SDL is already shutdown. Closes #6581.

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...

[ddnet:master] 3 new commits

0db2f62 fix nameplate related indicators opacity - edg-l 1e319fa fix opacity for jumps - edg-l ddf57c9 Merge #6580 - bors[bot]

14:26

[ddnet/ddnet] Issue closed: #6520 Make extras invisible with cl_show_others_alpha 0

14:26

[ddnet/ddnet] Pull request closed: #6580 Fix cl_show_others_alpha not used in places where it can be

    %3 = llvm.mlir.constant(296 : i64) : i64
    %4 = llvm.mlir.null : !llvm.ptr
    %5 = llvm.call @realloc(%4, %3) : (!llvm.ptr, i64) -> !llvm.ptr
    %6 = llvm.mlir.constant(0 : i32) : i32
    %7 = llvm.call @memset(%5, %6, %3) : (!llvm.ptr, i32, i64) -> !llvm.ptr

ez (edited)

3 people carrying ddnet gigachad

gigachad

Image attachment

[ddnet:master] 2 new commits

82fb94a Shutdown SDL after kernel and client - Robyt3 d873095 Merge #6582 - bors[bot]

15:49

[ddnet/ddnet] Issue closed: #6581 Segfault using vulkan on exit

15:49

[ddnet/ddnet] Pull request closed: #6582 Shutdown SDL after kernel and client

[ddnet/ddnet-web] Pull request opened: #251 Update libtw2 doc

https://github.com/heinrich5991/libtw2/commit/0352e6f2f23e01e17c6212701f8faa166c524522

16:10

[ddnet/ddnet-web] Pull request closed: #251 Update libtw2 doc

16:10

[ddnet-web:master] 1 new commit

e2ee41b Update libtw2 doc - Zwelf

@deen i think for https://github.com/ddnet/ddnet/issues/6450 we need you to enable smth in steam

16:27

im looking for something ez to do next xd

I already did that

16:33

(implement it)

[ddnet/ddnet] Issue closed: #6172 hang/deadlock on macos silicon when changing windowed to any window type

oh nice xd

I usually use the github search

is:issue sort:updated-desc is:open -label:needs-discussion -label:to-reproduce -label:incomplete -label:"3rd party" -label:discussion

(with a bookmark) to find the issues to fix

oh nice

16:35

issue fix speedrun xD

Ryozuki

@deen i think for https://github.com/ddnet/ddnet/issues/6450 we need you to enable smth in steam

But I think this issue is not solvable anyway, as I explained in the issue "You probably need to select each individual language that the game supports, so unsupported languages don't appear in the list. However this could mean that if a user has Steam configured with a language that's not supported by the Steam API, then the GetCurrentGameLanguage API might default to English, so we cannot use the Steam API to first determine the language, as this would override the user locale, which would have been more accurate. The user could also later change the language in the Steam game properties and would be confused, as this language is only applied once when the game is first started, but the language selection in the game properties is likely intended so that the language can be changed at any time."

I can't find this setting in the steam settings

Robyt3

But I think this issue is not solvable anyway, as I explained in the issue "You probably need to select each individual language that the game supports, so unsupported languages don't appear in the list. However this could mean that if a user has Steam configured with a language that's not supported by the Steam API, then the GetCurrentGameLanguage API might default to English, so we cannot use the Steam API to first determine the language, as this would override the user locale, which would have been more accurate. The user could also later change the language in the Steam game properties and would be confused, as this language is only applied once when the game is first started, but the language selection in the game properties is likely intended so that the language can be changed at any time."

the latter part could be fixed by setting the language to "autoselected" instead of setting it to a concrete thing

maybe we can just use the UI language instead of the configured game language

16:38

the docs say you usually want the game language and not the UI language, but I think it makes more sense here

16:38

because we don't change the language in the client later, after the game language is changed in the steam settings

Robyt3

because we don't change the language in the client later, after the game language is changed in the steam settings

but that could be fixed by marking the language as autoselected

16:39

also, the game language is the same as the UI language by default, I think

yeah, game language just uses the UI language if it's not set specifically for that game

16:40

what is "autoselected" though?

16:40

is that a setting you can use?

ye

16:40

basically like the nickname

16:41

if you don't set one explicitly, it's updated every time you change your steam nick

16:41

but if you set it explciitly, it'll no longer be updated

https://github.com/ddnet/ddnet/issues/6139

Possible to join not started team but with practice enabled · Issue...

First tee joins /team 1 and use /practice Second tee joins /team 1 and is stuck there because of practice Practice should probably act as if the team started

16:49

i dont understand this issue

16:49

First tee joins /team 1 and use /practice Second tee joins /team 1 and is stuck there because of practice

>

Practice should probably act as if the team started

16:49

what does it mean stuck

16:49

i tested it and i dont see anything wrong

https://github.com/rust-lang/rust/pull/111111

16:53

big numbers

xd

Ryozuki

First tee joins /team 1 and use /practice Second tee joins /team 1 and is stuck there because of practice

>

Practice should probably act as if the team started

I think this complains about the fact that it is currently possible to join a /practice team

16:54

because you can't leave a /practice team without dying

16:54

not sure if making /practice teams unjoinable is really the best way to solve this

@heinrich5991 awaiting ur input https://github.com/ddnet/ddnet/issues/2762

Fullscreen setting consistently breaks my screen resolution on i3 ·...

Setup: 2 monitors, 1920x1080, 2560x1440. DDNet 14.6.2 (Steam version). $ i3 --version i3 version 4.18.2 (2020-07-26) © 2009 Michael Stapelberg and contributors When starting DDNet with fullscreen e...

ryo tryhard mode

ye

17:03

somebody has to do the dirty work

17:03

kekCapture

17:03

(joke i do nothing)

17:03

https://github.com/ddnet/ddnet/issues/3922

SQlite can't load savegame on CentOS · Issue #3922 · ddnet/ddnet

@Zwelf any idea? Reported by Dan_cao. Works for me on Arch Linux.

17:03

wh ocares about centos

17:03

close this

17:03

gigachad

17:03

right @Jupstar ✪

tru

@Jupstar ✪ https://github.com/ddnet/ddnet/issues/5699

17:05

is this true

i dont think it can cause such insane lags

17:05

but indeed its not the most efficient implementation

@Jupstar ✪ https://github.com/ddnet/ddnet/issues/6134

Render tune settings when using entities mode · Issue #6134 · ddnet...

It's kinda annoying when there's a tune zone and you don't know what it does. I think we could render the tune settings when using entities mode in a not too annoying way rather than ha...

17:06

can we render the tune zone config text in place?

17:06

xddd

mhh sounds hard to do nicely

also rly niche

the question is where do you even render the text exactly? would be distracting if it keeps moving while the tee moves

yeah^^

close as wontfix?

17:10

xd

17:10

i vote+ 1

17:10

+1

17:10

god i cant type

we can close 90% as wont fix xDDDD

xdd

17:11

i got a idea for a feature

17:11

but maybe its too ddnet centric

17:11

a tab to show info about ur name, since we can query the json api

17:11

points etc

[ddnet/ddnet] Issue opened: #6583 lockdown mod command to prevent troll raids

its common for trolls to constantly join and spam with multiple ips add a command that locks down the server without needed to set a password, for the given minutes maybe another approach is better

Do any of you use MATLAB?

Learath2

Do any of you use MATLAB?

why do u ask if we use non-free propietary software

Ryozuki

why do u ask if we use non-free propietary software

Because we atleast have one academic here and academics love matlab

only when I'm forced by university st(u|a)ff

and there may be people using octave

18:32

Anyway, any idea proprietary software enjoyers?

Image attachment

ask chatgpt

1

actually not a bad idea

its sad u cant take a academic course and not be bombarded by propietary non free software

18:34

public teachings ought to be free as in freedom

18:34

i had a teacher who was very into pushing this

18:34

he wanted to make my high school change all windows into linux

18:34

and replace all the programs

18:34

gigachad

18:35

he was nearing the retarement age too xd

18:35

retirement

Ryozuki

its sad u cant take a academic course and not be bombarded by propietary non free software

they are trying to upgrade the curriculum here but the professors are a bit resistant

18:36

circuit theory moved to python but they still use proprietary spice instead of ngspice

@Learath2 if gov made a law to only use fsf licensed software

18:37

it would push and motivate to make better

linear algebra is moving to python next year I think

Learath2

Anyway, any idea proprietary software enjoyers?

not sure if I parse correctly, are you trying to zip the two cell arrays together?

Patiga

not sure if I parse correctly, are you trying to zip the two cell arrays together?

Yes, I want to end up with one 2x2 cell array of 1x2 cell arrays

18:40

I think this control theory class will forever be in matlab tho. They use so many proprietary things, I can't imagine it existing without simulink and matlab

https://stackoverflow.com/questions/26165932/how-do-i-zip-multiple-cell-arrays-together while I don't know enough matlab to understand the solution, this looks good

ChillerDragon

true designers use adobeOS over windows

i bet it crashes daily and has like a thousand memory leak issues

ew

Patiga

https://stackoverflow.com/questions/26165932/how-do-i-zip-multiple-cell-arrays-together while I don't know enough matlab to understand the solution, this looks good

Someone on the matlab discord found me something better cellfun(@(x, y) {x, y}, n, d, 'UniformOutput', false)

sera all

@Learath2 @Jupstar ✪ @heinrich5991 my heros i need help

kio

@Learath2 @Jupstar ✪ @heinrich5991 my heros i need help

dont ask to ask

21:33

crazy

xd

i need to share an object

21:33

no wait

Voxel

xd

so, let me explain better, I have my own class, and from there I want to manage everything, be able to hook, send messages in chat, move, jump, disconnect from the server, ... (these are random examples that came to my mind ) but I don't understand how to make my class globally shared in other classes or have access to other classes directly from mine, because I guess I can't reinstantiate a character or player or client object or whatever, I guess I need the object originally created to be able to access its data inside, so the question is: is there a way to share all objects to my class or my class to all objects?

friend classes i suppose

how it works

in c++ you can access your friends privates

21:58

or so I heard

xd

21:58

oke but i never saw a similar thing in other languages

it's in every oop language you can think of

for example, this fucking m_pClient where can i find it without pass that from fucntion params

Learath2

it's in every oop language you can think of

have another name?

it's called friend in every language I can think of

wtf

22:00

im nub

22:00

very nub

22:00

stop programming

22:00

lets learn latin

kio

for example, this fucking m_pClient where can i find it without pass that from fucntion params

You either pass it from parameters or you keep your own, that's how it works

22:02

client components e.g. each keep their own m_pClient

@Learath2 so, if im in my sweety class and i wanna use m_pClient->m_Chat.AddLine(int ClientID, int Team, const char *pLine) without pass m_pClient by params of my function how can i do that

Learath2

You either pass it from parameters or you keep your own, that's how it works

what u mean "keep your own"

Your class should be initialized by another class that already has a pointer to CGameClient, which it can pass to you in the constructor of your class, which you can store in your class

ok this was my idea, but I wanted to put as many things as possible in the constructor, so where do you recommend me to instantiate my class?

22:05

gamecore, client or where

What is your class even?

game>client>kehub/

It all depends on context. When should your class be created?

God object troll

troll

and if then after instantiating it where I want and passing it what I want I want to call a method of my class from CChat::OnMessage or idk and then I still want to use the reference to my original object, how does it get to the CChat class? IT'S ALL SO CONFUSED HELP

Then CChat needs a reference to your object through some means

22:08

Look there is no hidden magic or anything, if you want a pointer you need to get it there somehow, usually during initialization or for one offs usually in function parameters

Learath2

It all depends on context. When should your class be created?

right now I use it as a util that I create whenever I need it and call the method, but now I want to do it better to implement functions that remain in the background called in a loop to check something I don't know yet, for example if a player he is so close to you he writes him to leave in chat

Learath2

Then CChat needs a reference to your object through some means

Normally I would do this in the constructor of the class, but I never see anywhere an object created by passing it parameters, although I see Inits wrapped with parameters never passed...

kio

right now I use it as a util that I create whenever I need it and call the method, but now I want to do it better to implement functions that remain in the background called in a loop to check something I don't know yet, for example if a player he is so close to you he writes him to leave in chat

I think what you are looking for is kinda like a component, components already have hooks everywhere so you don't have to think so much about where to initialize/call stuff

kio

Normally I would do this in the constructor of the class, but I never see anywhere an object created by passing it parameters, although I see Inits wrapped with parameters never passed...

I know how things work, I've been programming for a long time, but CPP always manages to confuse me

Learath2

I think what you are looking for is kinda like a component, components already have hooks everywhere so you don't have to think so much about where to initialize/call stuff

i saw this too but what should i do?

kio

Normally I would do this in the constructor of the class, but I never see anywhere an object created by passing it parameters, although I see Inits wrapped with parameters never passed...

Then do it in Init? It's heritage from the porting of teeworlds from C to C++.

kio

i saw this too but what should i do?

Make a class that inherits from CComponent, override one of the many fine hooks you can find in src/game/client.component.h to do your things, register your component with all the other components

Learath2

Make a class that inherits from CComponent, override one of the many fine hooks you can find in src/game/client.component.h to do your things, register your component with all the other components

I had done it initially, but since it inherits all methods like onRender and others, it seemed ugly to me

22:14

xd

They are empty by default, you only have to override the ones you want to change

sexy

I'll go practice alcoholism, have fun making what I suspect is a bot that we'll have to ban you for

so my class can be considered a component?

Learath2

I'll go practice alcoholism, have fun making what I suspect is a bot that we'll have to ban you for

ahahaha no no my man

kio

so my class can be considered a component?

It can be one, I don't see why not. Even if it's not you don't seem to know how best to hook into our code so it's best you stick to a component where all that stuff is already thought about for you

I want to make a serious client with nice graphics and some interesting features, but I want to fully understand the code

"components" are loosely defined anyway, most anything can be a component

22:17

they usually are rendered, but that's not strictly necessary I guess

kio

I want to make a serious client with nice graphics and some interesting features, but I want to fully understand the code

today i managed to apply the ddrace function when you finished a map without going into it for kog servers

22:19

question, is it legit for you to convert bees that are not mine made with post to bees of mine with get because they are more convenient? haha

Learath2

I'll go practice alcoholism, have fun making what I suspect is a bot that we'll have to ban you for

you look like a studied mind, don't drink, it hurts

kio

you look like a studied mind, don't drink, it hurts

The aim is to unstudy the mind Prayge

Prayge

if you're happy... haha (edited)

22:23

'Member access into incomplete type' beautiful

@Learath2 why the fuck does everyone access m_pClient which is protected as they want while it tells me that I don't have permission to access the variable? >:

They either access their own copy, (a lot of classes have m_pClient pointers of their own) or they are friends with the class they access it thru

but what role do you have on this customer? how do you know all these things haha

22:36

by "their copy" what do you mean? shouldn't they be pointing the variable at something?

Image attachment

I only spent a decade of my youth learning C++ and contributing to this codebase

Learath2

I only spent a decade of my youth learning C++ and contributing to this codebase

oh

22:37

very nice

22:37

a chad

kio

by "their copy" what do you mean? shouldn't they be pointing the variable at something?

Those are the classes that have a pointer to m_pClient, thus they can all access m_pClient as every class can access all their members

so do you recommend me to make my own copy of m_pClient?

kio

a chad

No, no, no, very not recommended

Learath2

No, no, no, very not recommended

what? you are a giga chad man the most powerful man in the world

kio

so do you recommend me to make my own copy of m_pClient?

I recommend you just make a component, you are clearly lost, it's much easier with all the boilerplate

gigachad

Learath2

I recommend you just make a component, you are clearly lost, it's much easier with all the boilerplate

eh eh the problem is that im retarded

Doubtful, you did manage to get it compiling

class KeHub : public CComponent{ public: m_pClient-> //error there (obv in 2 different files)

22:41

I saw I don't remember where a very long list of components, do you intend to register it there?

i might be stupid but are you adding #include <path/to/class.h>?

If you took a closer look at CComponent, you'll notice that while m_pClient is protected GameServer() isn't

Voxel

i might be stupid but are you adding #include <path/to/class.h>?

its different

o

Voxel

i might be stupid but are you adding #include <path/to/class.h>?

the problem I think is that the variable is protected

Image attachment

GameClient() sorry, brain slow

22:43

Oh, that's not the issue at all

Learath2

GameClient() sorry, brain slow

in fact, I randomly tried to put GameClient as a friend of my class but it didn't work (I didn't quite understand how or when to use friend xd)

Learath2

Oh, that's not the issue at all

sad

Inside your component you need to be including game/client/gameclient.h as the compiler says it doesn't know what a CGameClient is so it can't dereference it

AAAAAAAAAAAAAAAAAAAAA I THOUGHT IT WAS RECURCIVE

22:46

the include

22:46

lmao

component.h doesn't include it either

but actually it would be a disaster to inherit the imports hahaha

Learath2

component.h doesn't include it either

right, but there was an init, doesn't matter im stupid

22:47

i love u

22:47

❤️

22:48

i'm actually learning c++ haha

same

22:49

im rly bad at it but ive made a few good changes to this game (edited)

me too xd

I still suggest you grab a book rather than mess around with the game, but y'all do y'all

what is a book?

It's like a collection of written tutorials from a bygone era

ohhh

22:50

if they are not made by an Indian I don't care

They even used to print them onto thin sheets of bleached carbon using "ink" and bound them together. Truly fascinating stuff

sexy

22:52

can i zoom on it?

No, but they used to have this ancient invention called a "magnifying glass", it used a specially cut piece of glass to optically enlarge things

why not fuck! no accessibility for the most needy, what a monstrosity this is (edited)

Learath2

No, but they used to have this ancient invention called a "magnifying glass", it used a specially cut piece of glass to optically enlarge things

wtf dark magik

Learath2

I still suggest you grab a book rather than mess around with the game, but y'all do y'all

its actually more fun to mess with something already made than it is to start from scratch

Voxel

its actually more fun to mess with something already made than it is to start from scratch

my study style (edited)

It's much more fun to mess with it when your tools aren't working against you and the very language you try to express your ideas in isn't alien

22:56

But that's just me, as I said, you do you

trollet

22:56

Learath2

It's much more fun to mess with it when your tools aren't working against you and the very language you try to express your ideas in isn't alien

exactly

who is a kog player there?

i grew up using scratch lolol

Visual programming languages are I guess good places to mess around with stuff

cammostripes

22:58

teaching basic programming i make students use flowgorithm, really easy

You teach programming??

basic (edited)

22:59

im 18

Learath2

Visual programming languages are I guess good places to mess around with stuff

it really is fun

m_aInputData

Idk if microsoft still offers it but Kodu Game Lab was so much fun

where does this bastard come from how do i import it to my class >:(>:(>:(>:(>:(

Learath2

Idk if microsoft still offers it but Kodu Game Lab was so much fun

OHHHHH

23:02

crazy

kio

where does this bastard come from how do i import it to my class >:(>:(>:(>:(>:(

I'm all out of spoilers for tonight. You'll have to figure that one out for yourself

❤️

23:02

how old are u?

25

Learath2

I'm all out of spoilers for tonight. You'll have to figure that one out for yourself

if in two days I don't understand it I'll come back and ask you haha

Learath2

25

ok then getting married is legit

You can ask tomorrow too. Spoilers ought to be refilled by then

did you go or are you going to university?

But pro-tip. Now that you are in a component. Look at other components, if they do something similar to you, maybe you can steal some code

im only 20 wtf

Learath2

You can ask tomorrow too. Spoilers ought to be refilled by then

I gotta try to figure it out, let's see how fast I give up

kio

did you go or are you going to university?

Going to uni rn. I'll finish my bachelors hopefully this september. I'm just an awful student, way too lazy

Learath2

But pro-tip. Now that you are in a component. Look at other components, if they do something similar to you, maybe you can steal some code

pro-tips are always still something xd

Voxel

im only 20 wtf

I'm an old man now

Learath2

I'm an old man now

granny

Learath2

Idk if microsoft still offers it but Kodu Game Lab was so much fun

i guess but at the same time, scratch is like, half programming, half animation software

Learath2

Going to uni rn. I'll finish my bachelors hopefully this september. I'm just an awful student, way too lazy

I don't know what to do with my life

Voxel

i guess but at the same time, scratch is like, half programming, half animation software

Kodu is like 30% programming 79% game

kio

I don't know what to do with my life

Don't ask me. I'm sitting alone in the dark having a drink while talking C++ for the maybe 100th weekend in a row. I clearly did something wrong

Learath2

Don't ask me. I'm sitting alone in the dark having a drink while talking C++ for the maybe 100th weekend in a row. I clearly did something wrong

AHAHAHAHAH

23:07

If that's what you like, that's fine

No, that is most definitely not what I like

poggers2

2

I'm really stupid, I've seen now that certain types of variables are structs, so CNetObj_PlayerInput rappresents the input

Learath2

No, that is most definitely not what I like

shit

Lol, spam gets deleted so fast

i saw that xd

23:10

anyways when i grow up and still have the will to, i want to program useless and niche interactive art ptojects

like?

Voxel

anyways when i grow up and still have the will to, i want to program useless and niche interactive art ptojects

Conventional wisdom suggests that this is a good idea. Apparently chasing your dreams brings with it much satisfaction

kio

like?

https://discord.com/channels/252358080522747904/252358080522747904/1102119337923461191

Exported 1,691 message(s)