01:03 < bridge> <davide55> cl_predict_freeze seems a bit messy in the last releases, anyone else experimenting issues?
01:03 < bridge> <davide55> Or has it been changed recently?
01:04 < bridge> <pioooooo> Only visually I think
01:08 < bridge> <davide55> Yeah
01:18 < bridge> <davide55> it's so bugged now 😄
01:20 < bridge> <kebscs> Check nightly
07:56 < bridge> <chillerdragon> @kollpotato: 0.7 only flooder here u go! https://gitlab.com/teeworlds-network/twnet_parser/-/tree/5b54a9c2facfa0bba54adee2fc1a397657e70375/examples/07/flood
08:11 < bridge> <chillerdragon> Isolated enough that my PR got rejected because of it -.-
08:28 < bridge> <ryozuki> ill call my kernel libc "elibc"
08:30 < bridge> <Solly (bun bun)> https://www.desmos.com/calculator/wh6on32ml7
08:30 < bridge> <Solly (bun bun)> squishy quad
09:34 < bridge> <ryozuki> i achieved glory, had to add PIC support to the elf loader tho xd
09:34 < bridge> <ryozuki> https://cdn.discordapp.com/attachments/293493549758939136/1404005067723505675/image.png?ex=68999d1d&is=68984b9d&hm=b0c562868f4f339cd3c263d6433eaf42d3eead2bddcb95d00fe168882134a2ce&
09:46 < bridge> <ryozuki> i can load pic pie elf programs
09:46 < bridge> <ryozuki> idk if i need smth special for aslr
09:57 < bridge> <c0d3d3v> mh in https://github.com/ddnet/ddnet/pull/4980/commits/65b7380023f37210ef55b35b1dd4b6b2530214a6 I changed that turrets will fire to the closest player in team, and ignore solo players. But the same mechanic is also build in draggers, seems I should revert it, because it is a game mechanic that it draggs only the solo team mate if he is closer, instead of both (solo would get dragged anyway)
10:25 < bridge> <blaiszephyr> ChillerDragon:
10:25 < bridge> <blaiszephyr> 
10:25 < bridge> <blaiszephyr> why did you use `AUTOMATED` for #10665 - isnt that kind of misleading? i think `CONTROLLED` or something would fit better, the camera is not automated in a literal sense, is it?
10:26 < bridge> <ChnFreddie> I coined that word.
10:26 < bridge> <blaiszephyr> https://github.com/ddnet/ddnet/pull/9407#issuecomment-2764730661 for your comment btw
10:26 < bridge> <blaiszephyr> 
10:26 < bridge> <blaiszephyr> the AB will require a bump for it anyway
10:27 < bridge> <ChnFreddie> Also what happened to AUTOMATED, did chiller change what it does? It was supposed to mark dummy hammer input as AUTOMATED
10:28 < bridge> <blaiszephyr> looks like `AUTOMATED` simply tracks if camera is of type spec
10:28 < bridge> <blaiszephyr> ```cpp
10:28 < bridge> <blaiszephyr> {
10:28 < bridge> <blaiszephyr>             m_aLastPos[g_Config.m_ClDummy] = GameClient()->m_Controls.m_aMousePos[g_Config.m_ClDummy];
10:28 < bridge> <blaiszephyr>             GameClient()->m_Controls.m_aMousePos[g_Config.m_ClDummy] = m_PrevCenter;
10:28 < bridge> <blaiszephyr>             GameClient()->m_Controls.m_aInputType[g_Config.m_ClDummy] = CControls::EInputType::AUTOMATED;
10:28 < bridge> <blaiszephyr>             GameClient()->m_Controls.ClampMousePos();
10:28 < bridge> <blaiszephyr>             m_CamType = CAMTYPE_SPEC;
10:28 < bridge> <blaiszephyr>         }
10:28 < bridge> <blaiszephyr> ```
10:28 < bridge> <blaiszephyr> ```cpp
10:28 < bridge> <blaiszephyr> {
10:28 < bridge> <blaiszephyr>             m_aLastPos[g_Config.m_ClDummy] = GameClient()->m_Controls.m_aMousePos[g_Config.m_ClDummy];
10:28 < bridge> <blaiszephyr>             GameClient()->m_Controls.m_aMousePos[g_Config.m_ClDummy] = m_PrevCenter;
10:28 < bridge> <blaiszephyr>             GameClient()->m_Controls.m_aInputType[g_Config.m_ClDummy] = CControls::EInputType::AUTOMATED; <---------
10:28 < bridge> <blaiszephyr>             GameClient()->m_Controls.ClampMousePos();
10:28 < bridge> <blaiszephyr>             m_CamType = CAMTYPE_SPEC;
10:28 < bridge> <blaiszephyr>         }
10:28 < bridge> <blaiszephyr> ```
10:30 < bridge> <ChnFreddie> I did did that
10:30 < bridge> <ChnFreddie> Hmm
10:30 < bridge> <ChnFreddie> I'll just stop commenting about it. There's a reason I've given up on it lol
10:30 < bridge> <blaiszephyr> :justatest:
10:30 < bridge> <blaiszephyr> trying to understand it thats all xd
10:32 < bridge> <ChnFreddie> Ah I think automated is set when camera jumps, cuz it wasn't user input (and originally treated as absolutely but Heinrich wanted another flag)
10:32 < bridge> <ChnFreddie> Whenever camera type changes, input target was forcefully overridden
10:32 < bridge> <blaiszephyr> aaah, okay in that context automated may make sense, as its `automated absolute input`
10:32 < bridge> <ChnFreddie> I hate looking at these code tho
10:33 < bridge> <ChnFreddie> I'll still stop doing that
10:33 < bridge> <ChnFreddie> Also idk about the AB changes, ~~and kinda sounds like a implementation leak on the AB side oof. ~~
10:34 < bridge> <ChnFreddie> Cuz the ABI definitely didn't change.
10:35 < bridge> <ChnFreddie> Just a minor detail I guess but I'm pretty sure AB implementation should handle flags better if extra bits affect AB implementation even when ABI didn't change.
10:41 < bridge> <ChnFreddie> None of the admins acknowledged my comment or chillers regarding to ABI either.
10:44 < bridge> <ChnFreddie> Actually, I remembered what happened
10:44 < bridge> <ChnFreddie> I'll add more context on GitHub
10:44 < bridge> <blaiszephyr> ❤️
10:44 < ws-client> **<ChillerDragon>** sos im blind
10:45 < ws-client> **<ChillerDragon>** someone knows where client ids get chosen?
10:45 < bridge> <blaiszephyr> Maintainer btw
10:45 < ws-client> **<ChillerDragon>** @blaiszephyr not my choice read the old pr comments
10:45 < bridge> <Assa> Wtf is a weak bounce?
10:46 < ws-client> **<ChillerDragon>** @blaiszephyr fakof xd
10:46 < bridge> <blaiszephyr> have you ever tried dropping down from a fly-position and you had weak as the driver, and no matter how hard the bottom tee hammered you keep on dropping?
10:47 < ws-client> **<ChillerDragon>** skil issue
10:47 < ws-client> **<ChillerDragon>** but ye ik what you mean sometimes ships sinkin
10:48 < bridge> <ChnFreddie> Context added
10:50 < bridge> <blaiszephyr> heinrich seems to want that, so i guess that'll be done in a seperate PR at some point then
10:50 < bridge> <ChnFreddie> Want what
10:51 < bridge> <blaiszephyr> the input disctinction in the ab
10:51 < bridge> <ChnFreddie> It's in the ABI already
10:51 < bridge> <blaiszephyr> the input distinction in the ab
10:51 < bridge> <ChnFreddie> Via player flags
10:51 < bridge> <ChnFreddie> That's why I took it out?
10:51 < bridge> <ChnFreddie> Cuz it's in there
10:52 < bridge> <ChnFreddie> Fuck it, I'll look at the code.
10:52 < bridge> <blaiszephyr> :bruhh:
10:53 < bridge> <ChnFreddie> Looked it's in there. AB can use the new flags
10:53 < bridge> <ChnFreddie> And Heinrich personally removed the tag of ABI in the original PR
10:53 < bridge> <ChnFreddie> I'm pretty sure it's not needed anymore
10:54 < bridge> <ChnFreddie> Looked, it's in there. AB can use the new flags
10:55 < bridge> <Assa> ah this is player bouncing, and not laser bouncing :kek: thank you
10:56 < bridge> <blaiszephyr> do you know about the interaction of a bounced-off laser hitting a tee holding space+grenade vs a direct hit?
10:56 < bridge> <blaiszephyr> the game works in mysterious ways
10:57 < bridge> <manikoo> draggory's alter ego
10:58 < bridge> <Assa> @pioooooo I invite you to optimize setup time in render layers ^.^ namely [this](<https://github.com/ddnet/ddnet/blob/2934124cc9fa3f87d8bbd98193282ee2c9c0e5de/src/game/client/components/render_layer.cpp#L532>) function for tilelayers and [this](<https://github.com/ddnet/ddnet/blob/2934124cc9fa3f87d8bbd98193282ee2c9c0e5de/src/game/client/components/render_layer.cpp#L906>) function for quad layers. Most performance issues come probably from the quads.
10:59 < bridge> <Assa> I know of players shooting a laser and disconnecting making the projectile invalid. Maybe we should accept Chillers PR about that
10:59 < bridge> <ChnFreddie> Double checked again in case i'm just dumb. ye i'm 99% sure AB have the entire playerflags (which includes the two new flags) already. https://github.com/ddnet/ddnet/commit/8fd6d7da1051a51750477058ec28f8352e404269#diff-29ffe32916052e1066eed938021aadfeb29f855d7c63d9c1ca3350aceffb79f3R1355-R1369
11:04 < bridge> <Assa> maybe we could introduce yet another variable for quad render mode:
11:04 < bridge> <Assa> `sv_quad_render_mode 0 = grouping and auto clip activated, 1 = only grouping activated, 3 = all deactivated, low setup time, lowest performance`
11:09 < bridge> <Assa> does anybody know a good extension to find unused imports in vs code?
11:09 < bridge> <Assa> ---
11:09 < bridge> <Assa> does anybody know a good extension to find unused imports in vs code?
11:13 < bridge> <cinavarkeles> Hello
11:30 < ws-client> **<ChillerDragon>** found it
11:30 < ws-client> **<ChillerDragon>** this seems to determine the client id https://github.com/ddnet/ddnet/blob/2934124cc9fa3f87d8bbd98193282ee2c9c0e5de/src/engine/shared/network_server.cpp#L549-L561
11:36 < ws-client> **<ChillerDragon>** @robyt3 any chance i can frame debug dummies being replaced by humans on join as a bug so we can fix it? xd
11:37 < ws-client> **<ChillerDragon>** or is it a feature?
11:38 < ws-client> **<ChillerDragon>** need it for non debug dummies in forks -.- #8385
11:38 < chillerbot> https://github.com/ddnet/ddnet/issues/8385
11:46 < bridge> <robyt3> I think it would make sense if debug dummies would behave like normal clients when possible, if it's possible to do cleanly, ideally without changing `CNetServer`
11:48 < ws-client> **<ChillerDragon>** sounds tricky because the way i understand it the server allocates client ids based on ids not being used by a net connection here
11:48 < ws-client> **<ChillerDragon>** https://github.com/ddnet/ddnet/blob/2934124cc9fa3f87d8bbd98193282ee2c9c0e5de/src/engine/shared/network_server.cpp#L549-L561
11:49 < ws-client> **<ChillerDragon>** so the debug dummy would need a different state in net connection than offline which to me sounds like having to touch cnetserver and introducing a state bot
11:50 < ws-client> **<ChillerDragon>** https://github.com/ddnet/ddnet/blob/2934124cc9fa3f87d8bbd98193282ee2c9c0e5de/src/engine/shared/network.h#L226-L234
11:50 < ws-client> **<ChillerDragon>** patching that enum is what i did in my fork thats why i would like to move that low level network diff to ddnet
11:50 < ws-client> **<ChillerDragon>** but if u say thats bad im rip
11:52 < bridge> <robyt3> See https://github.com/ddnet/ddnet/issues/9554#issuecomment-2601810313
11:52 < ws-client> **<ChillerDragon>** or maybe state online would also work instead of state bot but im afraight we get timeouting debug dummies and the server sending network traffic to it self
11:53 < ws-client> **<ChillerDragon>** oh heino opinion :eyes:
11:59 < bridge> <Assa> @robyt3 this is the only open conversation on my huge PR right now: <https://github.com/ddnet/ddnet/pull/10630#discussion_r2264814631>
11:59 < bridge> <Assa> 
11:59 < bridge> <Assa> Do you prefer if I ping you or request review?
12:04 < bridge> <robyt3> Does it work if you remove the `const` or does clang-tidy complain?
12:05 < bridge> <robyt3> Otherwise add both a const function returning const pointer and non-const function returning non-const pointer
12:07 < bridge> <learath2> I like #10634 more than #10636. Can I just merge that or should we get one more person to weigh in?
12:07 < bridge> <DDNet> https://github.com/ddnet/ddnet/pull/10634
12:07 < bridge> <DDNet> https://github.com/ddnet/ddnet/pull/10636
12:13 < bridge> <Assa> What I mean is, that I need to drop all const modifiers on the functions using Graphics(), because they call non const functions, if I return a const IGraphics*
12:14 < bridge> <Assa> so there is no points to ever return a const IGraphics*
12:14 < bridge> <Assa> so there is no point to ever return a const IGraphics*
12:14 < bridge> <robyt3> Then only return a non-const pointer with a non-const getter
12:15 < bridge> <Assa> yes, but I don't need a const getter with a const pointer, if every function needs to drop const then
12:16 < bridge> <robyt3> Then also add a const function returning a const pointer, so the functions can be const when possible consistent with the discussion in #10555
12:16 < bridge> <DDNet> https://github.com/ddnet/ddnet/pull/10555
12:17 < bridge> <Assa> Here an example:
12:17 < bridge> <Assa> ```
12:17 < bridge> <Assa> void CRenderMap::RenderTile(int x, int y, unsigned char Index, float Scale, ColorRGBA Color) const
12:17 < bridge> <Assa> {
12:18 < bridge> <Assa>     if(Graphics()->HasTextureArraysSupport())
12:18 < bridge> <Assa> ```
12:18 < bridge> <Assa> 
12:18 < bridge> <Assa> `HasTextureArraysSupport()` is non-const. So if I have `const IGraphics* Graphics() const {...}` and `IGraphics* Graphics {...}` I need to drop the const modifier of RenderTile
12:18 < bridge> <Assa> I don't know what's the point of the const modifier then? It doesn't change anything on CRenderMap, only in the graphics backend
12:19 < bridge> <robyt3> Make the functions non-const for now, unless you want to also make the graphics functions const first 😅
12:19 < bridge> <Assa> okay :justatest:
12:22 < bridge> <ryozuki> every day  i think more
12:22 < bridge> <ryozuki> if ddnet had a account system
12:22 < bridge> <ryozuki> it would help proliferate a lot of ecosystems
12:22 < bridge> <ryozuki> because if u can verify a player is who he is u can do lot of services based on that
12:22 < bridge> <ryozuki> for example the wiki wouldnt need a captcha, simply a oauth from ddnet
12:23 < bridge> <ryozuki> or openid or whathever
12:25 < bridge> <Assa> yeah just dropped all const 🤷‍♂️
12:28 < bridge> <Assa> I would know how to do this with a federated system :justatest:  this would add yet another column in funding tho, because we'd need regional login servers
12:31 < bridge> <Assa> - one for each continent
12:31 < bridge> <Assa> - use the official ddnet servers themself as ISP using the database in the background
12:31 < bridge> <Assa> - forward login from continental federated AAI to ISP based on regional/config info
12:31 < bridge> <Assa> - round robin login attempts
12:31 < bridge> <Assa> 
12:31 < bridge> <Assa> legally you'd need a DPS and probably a TOS
12:34 < bridge> <Assa> - one for each continent
12:34 < bridge> <Assa> - use the official ddnet servers themself as ISP using the database in the background
12:34 < bridge> <Assa> - forward login from continental federated AAI to ISP based on regional/config info
12:34 < bridge> <Assa> - round robin regional login attempts on official regional servers
12:34 < bridge> <Assa> 
12:34 < bridge> <Assa> legally you'd need a DPS and probably a TOS
12:36 < bridge> <vahemaaa> use iFCity's account system
12:37 < bridge> <ryozuki> https://www.debian.org/News/2025/20250809
12:37 < bridge> <ryozuki> !!!
12:38 < bridge> <ryozuki> > i386 is no longer supported as a regular architecture: there is no official kernel and no Debian installer for i386 systems. The i386 architecture is now only intended to be used on a 64-bit (amd64) CPU. Users running i386 systems should not upgrade to trixie. Instead, Debian recommends either reinstalling them as amd64, where possible, or retiring the hardware.
12:44 < bridge> <Jupstar ✪> Wow so progressive. Impressive from ddnet pov 🤓
12:55 < bridge> <kollpotato> chillerdragon: yeah thanks, i already figured out a builtin way to flood
12:56 < bridge> <kollpotato> btw chiller i found ur coc account
12:56 < bridge> <kollpotato> https://cdn.discordapp.com/attachments/293493549758939136/1404055700342313100/IMG_20250809_222259.jpg?ex=6899cc45&is=68987ac5&hm=8b78655f3971bbfb035b1d85881cf181031c877745f3d3a2c8846df2e3097ef5&
12:58 < bridge> <filoqcus> xDDDDDDDDDD
12:58 < bridge> <filoqcus> :brownbear:
13:43 < bridge> <milkeeycat> https://cpu.land/ Interesting article
14:08 < bridge> <ryozuki> do a kernel and u learn that and more
14:10 < bridge> <milkeeycat> imagine doing a kernel in production ready language
14:29 < bridge> <jxsl13> is that arrogance, that I am smelling 👀
14:30 < bridge> <jxsl13> 👃
16:44 < bridge> <pioooooo> Sry I was just ranting as an player who was annoyed by the load time.  Currently I don't really have motivation for optimization of rendering stuffs...
16:45 < bridge> <Assa> I understand your frustration
17:57 < bridge> <robyt3> Should be easy to save like 25% by only performing the calculations in `GetRawData` once for every layer
18:09 < bridge> <Jupstar ✪> @pioooooo how high are load times for u?
18:10 < bridge> <robyt3> 16s before and 29s after according to #10663
18:10 < bridge> <DDNet> https://github.com/ddnet/ddnet/issues/10663
18:11 < bridge> <Jupstar ✪> 16s? Debug mode?
18:12 < bridge> <Jupstar ✪> I doubt cpus got 8 times faster since 2005 single threaded lol
18:12 < bridge> <robyt3> Was definitely faster than that for me in release with Vulkan
18:14 < bridge> <Jupstar ✪> @essigautomat how much processing even was added? Did u add like a new loop or smth?
18:14 < bridge> <Jupstar ✪> Almost double as slow sounds completely insane
18:19 < bridge> <Solly (bun bun)> so waht u do in this scenario is deepfly with hammer
18:19 < bridge> <Solly (bun bun)> B)
18:23 < bridge> <pioooooo> from ddnet.org/downloads
18:25 < bridge> <Jupstar ✪> On what setup
18:25 < bridge> <pioooooo> setup mean my pc spec?
18:25 < bridge> <Jupstar ✪> Yes
18:28 < bridge> <pioooooo> CPU
18:28 < bridge> <pioooooo> 
18:28 < bridge> <pioooooo>     Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
18:28 < bridge> <pioooooo> 
18:28 < bridge> <pioooooo> GPU 1
18:28 < bridge> <pioooooo> 
18:28 < bridge> <pioooooo>     NVIDIA GeForce GTX 1650
18:31 < bridge> <Jupstar ✪> And you are sure u didn't count the whole connecting phase? Only preparing map?
18:31 < bridge> <pioooooo> could be counting, but I don't think connection changed that lot between those 2 versions
18:32 < bridge> <Jupstar ✪> Sure But i don't want network weirdness be a problem
18:32 < bridge> <pioooooo> the loading time is consistently twice for me
18:34 < bridge> <Jupstar ✪> I'll look into it too when I'm home tonight
18:34 < bridge> <pioooooo> Is it possible that pcie speed also matters here since it looks like the uploaded data size increase?
18:34 < bridge> <pioooooo> idk
18:34 < bridge> <Jupstar ✪> Almost double is defs unacceptable, but 16s sounds comoletely unrealistic to me
18:34 < bridge> <Jupstar ✪> I dunno that map tho
18:35 < bridge> <pioooooo> large insane hammerhit maps be like
18:35 < bridge> <Jupstar ✪> I dunno.. No matter how bad the pcie is. Moving around 1gb of data should not take longer than 1-2s
18:36 < bridge> <Jupstar ✪> Generally i cannot imagine anything that could be that slow, except hdd and network
18:36 < bridge> <Jupstar ✪> But that shouldn't change between versions
18:37 < bridge> <Jupstar ✪> What can be slow ofc is smth like a swap
18:37 < bridge> <Jupstar ✪> But i assume you have like 8gb ram and 3gb vram
18:38 < bridge> <pioooooo> i got 16gb ram but could be almost full at all time :D
18:42 < bridge> <Jupstar ✪> I test on phone with my awesome internet connection, give me few minutes 😬
18:42 < bridge> <Jupstar ✪> https://cdn.discordapp.com/attachments/293493549758939136/1404142934529413310/Screenshot_20250810-184201_DDNet-18.8.1.png?ex=689a1d83&is=6898cc03&hm=a5c71befe56eb9272d86ab7a22e663364301d1ea41fec482e9ce552cf334dcf5&
18:44 < bridge> <pioooooo> :justatest:
18:51 < bridge> <Jupstar ✪> Yeah dunno, smth around 4-5s
18:51 < bridge> <Jupstar ✪> On phone
18:51 < bridge> <Jupstar ✪> Battery
18:56 < bridge> <pioooooo> maybe i should download more ram
18:57 < bridge> <pioooooo> or does phone gpu support the thing to do the tile buffer or something
18:58 < bridge> <pioooooo> i remember once checked the code and it's guarded by some if enabled stuff
18:58 < bridge> <pioooooo> i remember once checked the onInit code and it's guarded by some if enabled stuff
18:58 < bridge> <Jupstar ✪> Vk backend
18:59 < bridge> <Jupstar ✪> 12 gb ram. No vram since integrated
18:59 < bridge> <pioooooo> I don't know, I'd still be glad if you could look into it later
19:01 < bridge> <Jupstar ✪> Sure, but note I'm not complaining about that the new client sucks even more. I'd be interested how to similar strong computers can be so much off
19:02 < bridge> <Jupstar ✪> Sure, but note I'm not complaining about that the new client sucks even more. I'd be interested how two* similar strong computers can be so much off
19:03 < bridge> <Jupstar ✪> https://cdn.discordapp.com/attachments/293493549758939136/1404148172871438386/Screenshot_20250810-185854_DDNet-18.8.1.png?ex=689a2264&is=6898d0e4&hm=ee989b865f4932432960d8ea6b355a7533389a715bb31c7c46ba5493f5033680&
19:03 < bridge> <Jupstar ✪> See i use old ver too on phone
19:05 < bridge> <Solly (bun bun)> why does my bun bun have a chunk cut out of it, im just rendering a single freeform
19:05 < bridge> <Solly (bun bun)> https://cdn.discordapp.com/attachments/293493549758939136/1404148603357888594/image.png?ex=689a22ca&is=6898d14a&hm=8a94e5d9bccd139ca7cdd564cdc420474be52dfa144a3ff400cb5a9a6b3c1ee0&
19:05 < bridge> <Solly (bun bun)> ```cpp
19:05 < bridge> <Solly (bun bun)> const auto P1 = Transform({-0.5f, -0.5f});
19:05 < bridge> <Solly (bun bun)> const auto P2 = Transform({0.5f, -0.5f});
19:05 < bridge> <Solly (bun bun)> const auto P3 = Transform({0.5f, 0.5f});
19:05 < bridge> <Solly (bun bun)> const auto P4 = Transform({-0.5f, 0.5f});
19:05 < bridge> <Solly (bun bun)> const IGraphics::CFreeformItem Item = {P1.x, P1.y, P2.x, P2.y, P3.x, P3.y, P4.x, P4.y};
19:05 < bridge> <Solly (bun bun)> ```
19:06 < bridge> <Jupstar ✪> Texture coords
19:06 < bridge> <Solly (bun bun)> same hapens if i dont use texture and just solid color
19:06 < bridge> <Solly (bun bun)> is there an order i need to follow?
19:06 < bridge> <Jupstar ✪> Yes
19:07 < bridge> <Solly (bun bun)> would u happen to know what it is
19:07 < bridge> <Solly (bun bun)> or should i just keep swapping till it works xd
19:07 < bridge> <Jupstar ✪> 012023 i think
19:07 < bridge> <Solly (bun bun)> scuse me?
19:07 < bridge> <Jupstar ✪> Not sure by heart
19:08 < bridge> <Solly (bun bun)> thats how the triangles are formed? 012 and 023?
19:08 < bridge> <Solly (bun bun)> thats how the triangles are formed? 021 and 023?
19:08 < bridge> <Solly (bun bun)> so opposites then the rest
19:08 < bridge> <Solly (bun bun)> https://cdn.discordapp.com/attachments/293493549758939136/1404149545180336128/image.png?ex=689a23ab&is=6898d22b&hm=3c67896fb64b39389bcfa416be5da674e6f27ecc53fd1d172322da323ca36b35&
19:09 < bridge> <Solly (bun bun)> and flip everything
19:09 < bridge> <Solly (bun bun)> xd
19:09 < bridge> <Solly (bun bun)> maybe
19:09 < bridge> <Jupstar ✪> Yeah best look inside graphics threaded
19:09 < bridge> <Solly (bun bun)> i should do it will solid color lmao
19:09 < bridge> <Solly (bun bun)> okay
19:10 < bridge> <Solly (bun bun)> 013 032
19:11 < bridge> <Jupstar ✪> Look at the indices
19:11 < bridge> <Jupstar ✪> For free form
19:11 < bridge> <Jupstar ✪> Just switch p4 and p3 or smth in your code and test :brownbear:
19:12 < bridge> <Solly (bun bun)> sigh
19:12 < bridge> <Solly (bun bun)> https://cdn.discordapp.com/attachments/293493549758939136/1404150391326769172/image.png?ex=689a2475&is=6898d2f5&hm=23887f3202a246de095e84bf3c0c50ca291d55c0a7aeddbff12f9cfc36a1c7c6&
19:13 < bridge> <Assa> grouping + automatic quad clipping
19:14 < bridge> <Assa> it iterates twice through all quads
19:14 < bridge> <Assa> it iterates twice through all quads (in the worst case)
19:14 < bridge> <Assa> today I am going to make a map worse than the mona lisa
19:15 < bridge> <Solly (bun bun)> screw this im gonna make the triangles myself x-x
19:16 < bridge> <Solly (bun bun)> you cant?!?!?
19:16 < bridge> <Solly (bun bun)> epic ddnet moment
19:20 < bridge> <Solly (bun bun)> all orderings i try it keeps
19:20 < bridge> <Solly (bun bun)> adding a point in the middle
19:21 < bridge> <Solly (bun bun)> https://cdn.discordapp.com/attachments/293493549758939136/1404152804385226873/image.png?ex=689a26b4&is=6898d534&hm=2bafdb6c398678de8ab4e8efbc7e19979fc6b99174ae2a99653ac3162342b955&
19:22 < bridge> <Solly (bun bun)> this happens in the editor when the top 2 points are swapped
19:22 < bridge> <Jupstar ✪> And does that map have lots of quads
19:22 < bridge> <Jupstar ✪> Until Dawn
19:26 < bridge> <Jupstar ✪> Must be at least 200 million to take 10s longer xD
19:27 < bridge> <pioooooo> My speculation is map size (w*h) matters much
19:27 < bridge> <pioooooo> My speculation is map size (w*h) matters a lot
19:29 < bridge> <Jupstar ✪> Possible but then assa doesn't know his own code changes :frozen:
19:40 < bridge> <Assa> I don't experience loading times beyond 1 second.
19:42 < bridge> <soulyvevo> a bunch of 2048x2048 stars quads
19:42 < bridge> <soulyvevo> like a lot
19:42 < bridge> <Assa> Until dawn is really slow I must admit :justatest:  it probably is map size
19:42 < bridge> <kollpotato> i deleted them entirely xd
19:42 < bridge> <kollpotato> the loading speed felt the same
19:45 < bridge> <soulyvevo> idk when i removed them the fps went up and it didnt take long to load
19:46 < bridge> <robyt3> As I commented on the issue, `GetTileData` is very slow because it calls `GetRawData` every time.
19:46 < bridge> <soulyvevo> tho i guess the size is still the biggest factor
19:46 < bridge> <soulyvevo> abyss also takes some time to load
19:47 < bridge> <soulyvevo> but i never noticed any load time difference between versions
19:48 < bridge> <Jupstar ✪> Oh u already debugged lol
19:49 < bridge> <Jupstar ✪> Ofc doesnt explain freaking 16s xD
19:54 < bridge> <Assa> I give up, I can't get envelopes to sync with start at all 🙁
19:54 < bridge> <Assa> https://cdn.discordapp.com/attachments/293493549758939136/1404160971181785118/screenshot_2025-08-10_19-53-10.png?ex=689a2e4f&is=6898dccf&hm=9c9afc78f781a320c12c78b7931244b9347022194dfad348838e277e0c06e459&
19:56 < bridge> <Solly (bun bun)> almost there
19:56 < bridge> <Solly (bun bun)> https://cdn.discordapp.com/attachments/293493549758939136/1404161404079968376/image.png?ex=689a2eb6&is=6898dd36&hm=a10aeb28068d4017260bd51ffa36a8c438fbd0a6631ee18851043464e9e95584&
20:02 < bridge> <pioooooo> https://www.bilibili.com/video/BV1zJGjzoEdx/
20:02 < bridge> <Assa> there are multiple ways to archive this
20:03 < bridge> <Assa> mine is too expensive with 30 (fps) * 3,40 (minutes) * 60 (seconds) * ~300 (height) * 480 (width) = a lot of quads
20:03 < bridge> <robyt3> One quad and envelope per pixel?
20:04 < bridge> <Assa> one quad per pixel, one envelope per map, a lot of offsets per pixel
20:04 < bridge> <Assa> one quad per pixel, one envelope per map, one offset per quad
20:04 < bridge> <Assa> one quad per pixel, one envelope per map, one offset per quadlayer
20:04 < bridge> <Assa> one second per group
20:05 < bridge> <Assa> one editor crash after 20 seconds, 7 fps 😄
20:06 < bridge> <Assa> no I even optimize quads in width
20:15 < bridge> <Solly (bun bun)> somehow with the same code ive gotten it the other way around
20:15 < bridge> <Solly (bun bun)> https://cdn.discordapp.com/attachments/293493549758939136/1404166183569068112/image.png?ex=689a332a&is=6898e1aa&hm=b41a33ffd682afb66b98e59f6cba9fce4c39ece33f4c3865adc14831e52c7ee3&
20:15 < bridge> <Solly (bun bun)> i hate this
20:15 < bridge> <Solly (bun bun)> i hate this
20:17 < bridge> <robyt3> Never mind, maps can only contain 65536 envelopes
20:18 < bridge> <soulyvevo> whats that
20:18 < bridge> <Solly (bun bun)> squishy tees but for some reason my code is special
20:18 < bridge> <soulyvevo> @cyberfrighter worlds
20:19 < bridge> <Solly (bun bun)> my first try is the exact same as how quadcontaineraddquads does it
20:19 < bridge> <Solly (bun bun)> but takes out a chunk
20:19 < bridge> <Solly (bun bun)> after actually looking at quadcontaineraddquads ive got the same code again and it takes out a chunk in the opposite side
20:19 < bridge> <Solly (bun bun)> i dont fucking understand
20:23 < bridge> <Assa> only? well that sucks . /s
20:33 < bridge> <Assa> I am now "rendering" all of it, lets see if I can create same crashes 🙂
20:34 < bridge> <Assa> I wonder if this is the biggest ddnet map of all time
20:39 < bridge> <Solly (bun bun)> ):<
20:39 < bridge> <Solly (bun bun)> https://cdn.discordapp.com/attachments/293493549758939136/1404172236885327933/image.png?ex=689a38cd&is=6898e74d&hm=c3efe76b62182e3a9e0c87ab6b7de9743284576a6f4720ee13a5560e7ef7c1e5&
20:39 < bridge> <Solly (bun bun)> ```cpp
20:39 < bridge> <Solly (bun bun)> const auto PTL = Transform({-0.5f, -0.5f});
20:39 < bridge> <Solly (bun bun)> const auto PTR = Transform({0.5f, -0.5f});
20:39 < bridge> <Solly (bun bun)> const auto PBR = Transform({0.5f,  0.5f});
20:39 < bridge> <Solly (bun bun)> const auto PBL = Transform({-0.5f,  0.5f});
20:39 < bridge> <Solly (bun bun)> // TODO: Figure out what is up with texture mapping and draw a single quad instead of 2
20:39 < bridge> <Solly (bun bun)> const IGraphics::CFreeformItem Items[] = {
20:39 < bridge> <Solly (bun bun)>     {PTL.x, PTL.y, PTR.x, PTR.y, PBR.x, PBR.y, PBL.x, PBL.y},
20:39 < bridge> <Solly (bun bun)>     {PTR.x, PTR.y, PTL.x, PTL.y, PBL.x, PBL.y, PBR.x, PBR.y},
20:39 < bridge> <Solly (bun bun)> };
20:39 < bridge> <Solly (bun bun)> Graphics()->QuadsBegin();
20:39 < bridge> <Solly (bun bun)> Graphics()->QuadsSetRotation(pAnim->GetBody()->m_Angle * pi * 2.0f);
20:39 < bridge> <Solly (bun bun)> Graphics()->SetColor(pInfo->m_ColorBody.r, pInfo->m_ColorBody.g, pInfo->m_ColorBody.b, Alpha);
20:39 < bridge> <Solly (bun bun)> Graphics()->QuadsDrawFreeform(&Items[0], 1);
20:39 < bridge> <Solly (bun bun)> Graphics()->QuadsSetSubset(1, 0, 0, 1);
20:39 < bridge> <Solly (bun bun)> Graphics()->QuadsDrawFreeform(&Items[1], 1);
20:39 < bridge> <Solly (bun bun)> Graphics()->QuadsEnd();
20:39 < bridge> <Solly (bun bun)> ```
20:39 < bridge> <Solly (bun bun)> stupid
20:49 < bridge> <Solly (bun bun)> thi also breaks for transparent skins
20:56 < bridge> <cyberfrighter> ?
20:57 < bridge> <soulyvevo> cyberfighterworlds
20:57 < bridge> <cyberfrighter> why me
20:59 < bridge> <Assa> I actually wonder how I should improve this. GetRawData returns a void* and I need to cast it into a CTile/CTuneTile/CTeleTile ... somehow and extract sometimes more sometimes less information out of it
20:59 < bridge> <soulyvevo> egg
21:00 < bridge> <robyt3> You only need to perform all of the calls and checks in `GetRawData` once, then store the validated data pointer
21:01 < bridge> <Assa> oh yes
21:53 < bridge> <Assa> this also significantly improves ogl 1
22:00 < bridge> <Assa> @pioooooo can you test this?
22:29 < bridge> <chillerdragon> @kollpotato: please don’t leak my cock
22:33 < bridge> <kollpotato> :monkaS:
22:35 < bridge> <Assa> I don't feel like that does anything :omo:
22:49 < bridge> <Assa> ogl1 has some weird artifacts on the map Cyber Space :justatest:
22:52 < bridge> <Assa> ogl1 randomly crashed on me, but unfortunatly this was on a feature branch, but it crashed at an unrelated position to that :justatest:
23:15 < bridge> <pioooooo> This indeed decreased loading time like by half
23:15 < bridge> <Assa> really?
23:15 < bridge> <Assa> I see no benefits on my system
23:15 < bridge> <Assa> like none
23:16 < bridge> <pioooooo> 13s vs 25s
23:16 < bridge> <pioooooo> https://cdn.discordapp.com/attachments/293493549758939136/1404211916108337242/52b73f22a6392f34.png?ex=689a5dc1&is=68990c41&hm=3c170edfda50d44759a10e9031b23e4b9276630f759497c6004528b1b5d5c99c&
23:16 < bridge> <pioooooo> https://cdn.discordapp.com/attachments/293493549758939136/1404211916641009674/7b0067f6173c28e8.png?ex=689a5dc1&is=68990c41&hm=9890d6a9be9fee08f25e8c9081cafd37326a29a7ea617fa8b16dc2b4998e8fe6&
23:16 < bridge> <Assa> I guess you use vulkan backend?
23:16 < bridge> <pioooooo> yes vulkan
23:18 < bridge> <Assa> left is new and right is old?
23:18 < bridge> <pioooooo> yes, new read from old and old read from new
23:18 < bridge> <Assa> because left is 13 seconds and right is 25
23:18 < bridge> <pioooooo> :D
23:19 < bridge> <Assa> ah I see
23:19 < bridge> <Assa> I hope this is not just some caching effect
23:19 < bridge> <Assa> I wonder why I don't see any improvements, maybe some different bottlenecks
23:21 < bridge> <pioooooo> I tested two times for each, quite consistent
23:31 < bridge> <Assa> glad to help you btw (I am glad to help anybody who finished my map :brownbear: )
23:52 < bridge> <pioooooo> :D