04:10 < bridge> i have dummy connect added to my swap keybind so i can just swap and it connects 04:10 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314428835353198663/image.png?ex=6753bcbf&is=67526b3f&hm=795243a44239cd3e9361e275bf92391648165c36fbeccbf588a99c7c28ad74e0& 04:11 < bridge> it does however spam console which is a bit annoying 04:40 < bridge> demos are not teeworlds datafiles right? 06:24 < bridge> I did? 06:24 < bridge> oh didnt... commit anything 06:40 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314466567857242142/image.png?ex=6753dfe3&is=67528e63&hm=a4f378ff4c33d4e19a5203c4926fe8e0b2a8929ce7afbf366d844ce95037cab9& 06:41 < bridge> probably... should be a timeout on that 06:45 < bridge> wait no fucking way Localize is preprocessed by a python script 06:45 < bridge> 😢 06:49 < bridge> it's been like that since teeworlds i think 06:51 < bridge> surely it can just be a plain function 06:51 < bridge> wdym 06:51 < bridge> maybe inline stuff to make it figure out a hash 06:51 < bridge> Localize is a function 06:51 < bridge> it gets replaced by python? 06:51 < bridge> no? 06:51 < bridge> the script just pulls the string from the source code and put them in txt files 06:52 < bridge> ah okay thats fine 06:52 < bridge> :blobfacepalm: 06:52 < bridge> xd 06:52 < bridge> ~~at least question our code gen for network.py first~~ 06:53 < bridge> i already did that 06:53 < bridge> :kekw: 06:53 < bridge> but tbf code gen is pretty common tho 07:11 < bridge> how tf does `std:vector::embrace_back` crash 07:11 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314474210118340669/image.png?ex=6753e701&is=67529581&hm=1f8b821e1a78796cbcbf9c663774d78e4d8ce2d74519a8966a52285b2e3f2de6& 07:14 < bridge> you can make any memory access crash with enough effort. 07:18 < bridge> ah 07:19 < bridge> some LoadPng is threaded and if the texture produces warnings the string vector can crash 07:19 < bridge> probably race condition? 07:30 < bridge> I couldn't comment on security implications of this because I'm not qualified, but I have a hunch that this is not good for security? 07:30 < bridge> the threaded load for community icons are remote after all 07:31 < bridge> :greenthing: 07:39 < bridge> nah, i'm not adding mutex myself, i'll probably mess it up somehow. 07:39 < bridge> i should make lunch and do my own project probably 09:49 < bridge> @sollybunny 09:50 < bridge> afternoon 09:50 < bridge> You can write a script that just sends a broadcast every 3600 seconds that I specify 09:50 < bridge> ? 09:50 < bridge> edit it urself :D 09:51 < bridge> I won't be able to:( 09:51 < bridge> skill issue 09:51 < bridge> ask chatgpt 09:51 < bridge> My script no work 09:51 < bridge> It does not give out correctly 09:57 < bridge> <♂S1mple♂> Then learn the programming language and do it 09:57 < bridge> <♂S1mple♂> Or at least pay someone to do this silly stuff 09:58 < bridge> <♂S1mple♂> Weird how long chiller was doing this for free 10:00 < bridge> <♂S1mple♂> I would understand if you'd try to do something, ran into issue and specified it here, but asking debugging ai code or asking doing from scratch for free is just not polite 10:20 < bridge> this makes me so mad 10:20 < bridge> welcome to c++ 10:21 < bridge> :nouis: 10:30 < bridge> any bug like that is potentially a problem. an attacker just has to find a way to abuse it 10:30 < bridge> i wanted to fix it, but i'm bored 10:31 < bridge> i just wait for ryo to pop up telling me it wouldn't have happened with rust 10:31 < bridge> 10:31 < bridge> back when ryo had no gf, it took -15 seconds for him to say it 10:32 < bridge> he basically spammed that the whole day and once someone opened an issue, he could say: SEE 10:32 < bridge> i'm still strugging with shader depth 10:32 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314524866048823347/image.png?ex=6754162e&is=6752c4ae&hm=4e0611887338c389e60ec46ba401cf7cc98b50b8d4213bc80d37574bdb8c767b& 10:32 < bridge> what is your problem 10:34 < bridge> isn't blender one of these mfckers that use z as up coordinate 10:34 < bridge> im sorry to inform you its not just "One of" 10:34 < bridge> not that z 10:34 < bridge> it was actually minecraft who lied 10:34 < bridge> that should be illegal 10:35 < bridge> i agree 10:35 < bridge> if you tell me what your problem is, i might be able to give you a gpt like answer that sounds good but does not help at all 10:35 < bridge> not if you use cryengine 10:35 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314525672311623782/1_IWv0KVHPnf_xxL05Vh7oVQ.webp?ex=675416ee&is=6752c56e&hm=bf6f6163bb7e006e3e278fac383a0decd7cebb34342a0807fc04b05ea47d3e15& 10:35 < bridge> the real illegal one is unreal 10:36 < bridge> hm, I'm left handed :kek: 10:36 < bridge> i was tought in school z up and it screwed with me 10:36 < bridge> (im right handed btw) 10:36 < bridge> gotta use Unity, then 10:36 < bridge> no way 10:36 < bridge> u american? 10:36 < bridge> or where do they teach you that 10:36 < bridge> bri'ish 10:37 < bridge> a level further maths 10:37 < bridge> i'm trying to get a screenspace depth value manually by transforming a vertex position from view space to clip space. but the result doesn't match godot's result of depth value 10:37 < bridge> impressive 10:37 < bridge> ~~i dropped it i just wanted to feel cool~~ 10:38 < bridge> the application problem is that i want to try to make a inverted hull outline to separate between objects somehow, i haven't math'ed that part out tho 10:39 < bridge> otherwise it does this 10:39 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314526545280503848/image.png?ex=675417bf&is=6752c63f&hm=9fcc80a5651affd28505d3dfc557e7366e24af965153db2dca996be6a947b094& 10:39 < bridge> woaw 10:39 < bridge> tw tee has a leg 10:39 < bridge> its not just disembodied feet confirmed 10:39 < bridge> its not just disem~~bodied~~leggied feet confirmed 10:39 < bridge> still is, it's just the hull clipping into the body 10:40 < bridge> do you have an output of the value itself to compare it against what godot's output or smth. 10:40 < bridge> 10:40 < bridge> sometimes they flip the coordinate systems 10:40 < bridge> it's hard to check values for shader, i haven't dig through godot's code yet, so i should just do that 10:41 < bridge> what are you doing with the depth value 10:42 < bridge> good days 10:42 < bridge> can't you do that with a stencil just as fine? 10:42 < bridge> godot doesn't have stencil yet 10:42 < bridge> WAAT 10:42 < bridge> godot doesn't have Z offset either, which is pretty easy to do in unity lol 10:43 < bridge> wat 10:43 < bridge> well they do have stencils but only internally 10:44 < bridge> https://github.com/godotengine/godot/pull/60650 10:44 < bridge> sometimes there's some operation you need to do to linerize the depth 10:44 < bridge> cant u do it with this 10:44 < bridge> render priority 10:44 < bridge> idk if that's relevent for you 10:45 < bridge> https://github.com/eldskald/godot4-cel-shader 10:45 < bridge> hehehe 10:45 < bridge> https://github.com/eldskald/godot4-cel-shader/blob/main/src/outline.gdshader 10:45 < bridge> if i want to put the outline behind everything probably. currently i'm using depth so rendering order does not matter. 10:48 < bridge> that's just the hull with the grow width pre calculated in view space 10:48 < bridge> same thing 10:48 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314528918505525310/image.png?ex=675419f4&is=6752c874&hm=9bc1dac67f52be0067eacf00bf592a5ddad3b24be7e510973c75f67acb9cf086& 10:49 < bridge> the thing you want doesn't really sound possible? 10:49 < bridge> it is in unity, i've done it 10:49 < bridge> how are you supposed to know if the outline is behind or in front of something in an order-indepedant way? 10:49 < bridge> depth comparison, i just need to shift depth a bit 10:50 < bridge> depth comparing what to what 10:50 < bridge> but godot doesn't let you do that, so i'm hard writing into depth manually 10:50 < bridge> using a shifted depth value compared to the value in the depth buffer... 10:51 < bridge> but isn't there always going to be some depth where the clipping happens? 10:51 < bridge> yes, but i just need to shift it a little bit so it is not this blatant. :justatest: 10:53 < bridge> but if you offset it forward the ones that are behind stuff will be in front 10:53 < bridge> look, i've already done it in unity in a commercial product. i'll just try a bit more and if it doesn't work out i'll use post process or some other way. 10:53 < bridge> it will look like this? 10:53 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314530159524839464/image.png?ex=67541b1c&is=6752c99c&hm=6be78774bc30588f49d3c8a2f76450ea54df7bc5cc8493666d68610d6890e344& 10:53 < bridge> no 10:53 < bridge> no? 10:53 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314530261689438230/image.png?ex=67541b35&is=6752c9b5&hm=1b1aa68d6d9ba1b7628532f55015abed37210ef798035d01074a21848f0f0044& 10:53 < bridge> it will look like this 10:54 < bridge> ignore the color 10:54 < bridge> ddnet 3d update? 10:54 < bridge> doesn't it require sorting to do it that way tho? 10:54 < bridge> Ddnet 3d before GTA 6 gg 10:55 < bridge> TEEWORLDS VR 10:55 < bridge> i'd be so down bad for 3D ddnet 10:55 < bridge> guys, 3d rendering stopped sorting opaque objects since like forever. 10:55 < bridge> :justatest: 10:55 < bridge> Twinbop 3D 💥 10:55 < bridge> :twinbop: 10:56 < bridge> twinbop with boobs:poggers2: 10:56 < bridge> smh 10:57 < bridge> :greenthing: 10:58 < bridge> so does it work already or what is this screenshot? 10:58 < bridge> i moved the feet forward to show what i want 10:59 < bridge> :kek: 10:59 < bridge> idk how you can use just the depth buffer and the depth of the outline hull to figure out the render order without sorting, it sounds impossible 11:00 < bridge> i think he simply wants to create a second mesh moved a bit further into the back 11:00 < bridge> or not? 11:00 < bridge> xd 11:00 < bridge> he's doing that already, the issue the second mesh clips into the other ones 11:00 < bridge> I am going to LOSE MY MIND 11:00 < bridge> kinda but in screen space 11:01 < bridge> I have been at this for FIVE HOURS 11:01 < bridge> hence shader 11:01 < bridge> he's doing that already, the issue is the second mesh clips into the other ones 11:02 < bridge> I'm passing around a `([String], ((Int, Int), (Int, Int)))` because I'm not allowed to have proper state 11:03 < bridge> what is the intended behavior when the foot is intersecting the tee body, will the outline render above or below the tee? 11:03 < bridge> You know it's bad when your data types look like lisp 11:04 < bridge> yes, then i'd expect an output like 11:04 < bridge> https://discord.com/channels/252358080522747904/293493549758939136/1314528918514171949 11:04 < bridge> 11:04 < bridge> but if you don't get that, i assume you should compare if the screenspace coordinates are flipped again or smth. 11:04 < bridge> can't you like print the projection matrix and look if z is negative scaled or not 11:04 < bridge> Thanks functional programming, very cool 11:04 < bridge> i can't print in godot shader 11:05 < bridge> is there no renderdoc like debugger? 11:05 < bridge> I mean, destructuring that mess would be really easy if I was allowed to use pattern matching, but then I'd be binding variables, and that's a no no around here 11:05 < bridge> the renderdoc-like debugger is renderdoc 11:05 < bridge> i actually never knew how to print shader stuff. i only used renderdoc as a profiler before 11:05 < bridge> godot doesn't have built-in graphic debugger 11:05 < bridge> do any engines? 11:06 < bridge> how can they have 3 quadrillion contributors but not have smth like this xd 11:06 < bridge> unity has a frame analyzer that shows you buffers in each draw call and the uniform values 11:06 < bridge> that's pretty basic 11:06 < bridge> but godot has none 11:07 < bridge> I don't know of any engines that do anything close to what renderdoc does, I think you just use renderdoc 11:07 < bridge> yes 11:08 < bridge> you can still do visual debugging by outputting colors 11:08 < bridge> i am doing that 11:08 < bridge> This is a weird artifact, how did this happen 😄 11:08 < bridge> what artifact 11:09 < bridge> there is no artifacts, inverted hull just looks like that :brownbear: 11:10 < bridge> Ooooh, I see what's going on there 11:11 < bridge> just emulate a stencil yourself 11:11 < bridge> just rewrite godot at that point 11:12 < bridge> i think using stencil usually looks like this tho. 11:12 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314534913692667904/image.png?ex=67541f8a&is=6752ce0a&hm=353da7a3e9659f73dfb316717d4fda57fd9e89ec9b4fd945736de6b622ad87e7& 11:12 < bridge> So you are trying to fix this by messing with the depth buffer by hand? 11:12 < bridge> yep 11:13 < bridge> I guess that does sound doable, you should be able to just pull out the outline and render it in front of the tee closer to the camera 11:14 < bridge> Though idk how you'd hide the pieces that are supposed to be behind the solid parts like that though 11:14 < bridge> A curious problem, I wish you many luck 11:17 < bridge> i'm trying to shift depth towards object's local origin but without going over the hull growth length. 11:17 < bridge> just fireup renderdoc already 11:17 < bridge> so origin in front = outline more likely in front 11:17 < bridge> :pepeW: \ 11:17 < bridge> :pepeW: 11:18 < bridge> i've done nothing in the past 10 minutes, i'm just here chatting 11:18 < bridge> i will 11:18 < bridge> 11:19 < bridge> But the output here matches the exact shader? 11:19 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314524866048823347/image.png?ex=6754162e&is=6752c4ae&hm=4e0611887338c389e60ec46ba401cf7cc98b50b8d4213bc80d37574bdb8c767b& 11:19 < bridge> it is the shader yes 11:20 < bridge> Help 11:20 < bridge> i already don't understand how e.g. ungrow works. is it used somewhere magically? 11:21 < bridge> no 11:21 < bridge> it's just me trying stuff without deleting old stuff 11:21 < bridge> the actual shader in that screenshot is like 4 lines 11:21 < bridge> probably don't read into it too much 11:22 < bridge> i honestly don't understand your shader at all xD 11:22 < bridge> if you are wondering, this is literally inverted hull 11:22 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314537387254349835/image.png?ex=675421d8&is=6752d058&hm=04d449642837eabfebd05ea71f20cb48c5bc3963164f76b99bbe56f64acce8ca& 11:22 < bridge> so this is the shader in that screenshot 11:22 < bridge> wait, not that one 11:22 < bridge> oops 11:22 < bridge> but why is the projection matrix not fired in the vertex shader 11:23 < bridge> it was already done by godot, you just need to shift it 11:23 < bridge> oh ok 11:25 < bridge> i am surprised about the shader output tho 11:25 < bridge> maybe your problem lies somewhere else xd 11:25 < bridge> And it's normal that my server doesn't show the path of the grenade projectile , there's just a shot and then a second later the arrival 11:25 < bridge> it's not really a problem 11:25 < bridge> i'm just trying things 11:25 < bridge> the clipping is expected for inverted hull method 11:25 < bridge> pretty common 11:27 < bridge> burh, i could just shift fragcoord.z, what am i doing 11:27 < bridge> so 0.001 is still too big or what 11:27 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1314538815523917884/image.png?ex=6754232c&is=6752d1ac&hm=41b5b6948773dca4103b5d318f83fed52443aa41b903d24c49524b3a7bba3831& 11:28 < bridge> hot 11:36 < bridge> @tsfreddie are the normals multiplied with the inv projection matrix btw? 11:36 < bridge> 11:36 < bridge> or how does that work 11:36 < bridge> NORMAL is in model space 11:37 < bridge> would that need to be in screenspace too xd 11:37 < bridge> wouldnt 11:37 < bridge> VERTEX is also in model space i think. it get's converted into view space later by godot 11:37 < bridge> 😮 11:38 < bridge> > inout vec3 VERTEX Vertex in local coordinates. 11:38 < bridge> https://docs.godotengine.org/en/stable/tutorials/shaders/shader_reference/spatial_shader.html 11:38 < bridge> it is in view space in frag 11:38 < bridge> > in vec3 VERTEX 11:38 < bridge> > 11:39 < bridge> > Vertex that comes from vertex function (default, in view space). 11:40 < bridge> if you really don't want engine to do that, there's a flag for it `skip_vertex_transform` 11:40 < bridge> It's really sad how the brain forgets about lots of stuff xd 11:40 < bridge> i did all the 3d math once 11:41 < bridge> i don't actually do 3d math i think.... 11:41 < bridge> turn this to true.. 11:41 < bridge> 11:41 < bridge> Then you can do all the math 11:41 < bridge> ok u get all the matrices as globals 11:41 < bridge> boring xd 11:42 < bridge> at this point is just what space it is in, what do i want to do to it. i don't calculate anything mentally anymore. 11:45 < bridge> mat4 is for converting spaces, dot for checking direction differences, cross for getting an axis from two vectors, how they do it? no idea. what's the result? doesn't matter, i'll just plug them in and see. something doesn't look right? probably cuz i forgot a normalize somewhere.:kek: 12:13 < bridge> WOOO :Celebrate: 12:15 < bridge> I CAN DO BETTER 12:16 < bridge> I have the same problem! 12:16 < bridge> im setting a uniform per model to tell them the body's center. so i can shift depth just in front of or behind the body's sphere 12:16 < bridge> 12:18 < bridge> The dummy debug print update is quite nice and useful. But having no way to turn off the intense spam of a shared connect/swap bind is horrible. @sollybunny I was thinking of adding a connect\_dummy\_if\_not\_connected 12:18 < bridge> custom command to my client to solve it, 12:18 < bridge> aRe YoU ScArEd Of PRing CoFiG? 12:18 < bridge> I have the same issue for binding the exec command that also prints to the console. So I can not use it for some common binds 12:27 < bridge> How to speed up the level completion timer? 13:00 < bridge> can anyone answer me where located enums like "PLAYERFLAG_PLAYING" i found it only in build folder in generated folder. 13:00 < bridge> some python scripts 13:01 < bridge> search for PLAYING maybe 13:01 < bridge> hm, okay 13:04 < bridge> Should be in network.py 13:05 < bridge> yep, `datasrc/network.py:L7 PlayerFlags = ["PLAYING", "IN_MENU", "CHATTING", "SCOREBOARD", "AIM"]` 13:06 < bridge> dont we have 6 flags now 13:06 < bridge> :nouis: 13:07 < bridge> **insert fitting insult for my boss here** - brb 3hours, i actually need to do shit 13:07 < bridge> probably is getting reverted 13:07 < bridge> yes please. :kek: - dont want to hate on your work but playerflags are so damn precious 13:08 < bridge> tatar was thinking on some other ways to fix that coordinate switch 13:08 < bridge> We have 6 now? 13:08 < bridge> i used one 13:08 < bridge> What for? 13:09 < bridge> thanks 13:10 < bridge> #9318 13:10 < bridge> https://github.com/ddnet/ddnet/pull/9318 13:15 < bridge> it's crazy that i didn't think of that 13:15 < bridge> tbf if they are so valuable we'd have invented a way to add more of them 13:16 < bridge> there is a server to client only EXPLAYERFLAGS 13:16 < bridge> they are pretty much useless given we didn't use one in almost a decade 😄 13:17 < bridge> i only used it cuz it is just tied to the input which is kinda needed for that fix 13:18 < bridge> I remember investigating this one in the past and just deciding it was too much hassle to fix properly 13:18 < bridge> But IMO the proper way is to use an input flag, the entire issue is because spec state is not properly tied to input state 13:19 < bridge> i thought playerflag is a inputflag 13:19 < bridge> yes 13:19 < bridge> I just didn't use the correct terminology 13:20 < bridge> i think the point is that we don't want to use it 13:21 < bridge> Do we really not want to use it? We control both the client and the server nowadays, we can just bump up the size of the playerflags somehow, I'm pretty certain 13:22 < bridge> In the olden days it was a huge issue because we didn't control the client 13:22 < bridge> How to speed up the level completion timer? 13:22 < bridge> i think people are trying to find a better use for it before we commit to using up another bit 13:22 < bridge> also it messes up bot detection for now 13:22 < bridge> eh, bots who used playerflags for secret communication are a veeery 0.5 thing 13:22 < bridge> Hey 13:23 < bridge> Play faster? I don't understand what you want 13:23 < bridge> I wanna to make timer go faster 13:24 < bridge> funny thing is that with that pr noby servers straight up bans you. although i'm sure they can just fix it on their end. 13:24 < bridge> You can't without some very complex modification of the client and the server 13:24 < bridge> I saw map which make timer go from 5 seconds to 20 mins 13:25 < bridge> Did Tater have any ideas on how he'd fix it without using up a flag? 13:25 < bridge> We have an "add time" tile you can use 13:25 < bridge> It won't go faster but it will add some time to the timer 13:25 < bridge> we discussed a bit, i think there was a idea of sending a tick info along with the camera state using a netmessage 13:25 < bridge> Aaaa 13:25 < bridge> Ok 13:26 < bridge> my thought is that might be network dependent, i believe netmessages are different packets? 13:26 < bridge> Mhhhhhhhh, you hit the age old issue of snaps/inputs not being synchronized to netmessages at all 13:26 < bridge> there also might be a client side method hidden some where 13:27 < bridge> (The entire issue is this btw, the netmsg that puts the client into spec is not properly synchronized to the input/snaps) 13:27 < bridge> Then u need to configure both server and client? To make timer go faster 13:27 < bridge> Recompile 13:28 < bridge> Not just configure. You need to write a lot of code to modify that kind of stuff 13:28 < bridge> Server and client? 13:29 < bridge> oh also i think i had an idea of always using a single target space somehow 13:29 < bridge> I think so, but I'm not sure, I didn't ever try it before 13:29 < bridge> Server definitely 13:29 < bridge> like if your tee exists you always send the target relative to your tee. 13:30 < bridge> oh nvm, there still will be a jump after unpause nvm. 13:30 < bridge> oh nvm, there still will be a jump after unpause. 13:30 < bridge> then ye, that's about what we've discussed 13:31 < bridge> Well I don't mind the playerflag, but if you do find a way to avoid it I guess that's fine too 13:31 < bridge> Enjoy 13:31 < bridge> it won't be me finding it probably 13:31 < bridge> :kek: 13:31 < bridge> probably someone else finds it or just revert it and won't fix 13:35 < bridge> I would like to know how I can get the flag as in the gamemode ctf or as in the [video](https://youtu.be/-A9NTJ1P688), without changing the source code of the server and without installing other modifications for the server 13:37 < bridge> you can modify clients to render it 13:37 < bridge> otherwise it is a server mod 13:37 < bridge> no other way around it 13:39 < bridge> Thanks. I'll do it 15:05 < bridge> and still noby fng bans any non standard flags client xd 15:05 < bridge> k client used flags afair 15:09 < bridge> but imagine yourself making all in one c based game engine, i don't really see any other way to write such a huge thibg without using oo 15:10 < bridge> you will mimic them anyway, even if didn't want to do it 15:10 < bridge> I mean OOP mimics real life in a way, so you can say almost anything you do "mimics OOP" in a way 15:11 < bridge> My issue is with polymorphism and inheritence, not stuff like methods 15:11 < bridge> as data structure you mean? 15:12 < bridge> Wym? 15:12 < bridge> I'm back to torturing myself and now I'm failing to write a `foldr` such that it terminates given an infinite list... 15:13 < bridge> can you expand this a bit? i can't get it right 15:15 < bridge> I don't like inheritence and polymorphism. I don't want to see you inherit methods/fields from a child class. I don't want to see you calling virtual methods 15:16 < bridge> Today I unlocked a new crash error, bus error 15:18 < bridge> You can however have methods in your "class", you can have classes in classes. Those I don't mind 15:18 < bridge> how would you replace usage of polymorphism and inheritance? 15:18 < bridge> And for this you could just as well use structs and functions :) 15:18 < bridge> Design your program such that neither is relevant 15:19 < bridge> Well just don't use them? Idk how to explain something so natural to me 15:19 < bridge> so we can rewrite teeworlds in c without it being spaghetti right :justatest: 15:19 < bridge> Entity component systems avoid inheritence e.g. If you want to look at an example 15:20 < bridge> Another option is generics, I don't mind those 15:20 < bridge> Well, I'm not sure if we specifically could pull that off :justatest: 15:20 < bridge> yeah, i have seen generics impl in rust, cpp ecs are mostly inheritance or polymorphism based 15:21 < bridge> and flecs c ecs just reimplements inheritance and polymorphism afair :justatest: 15:21 < bridge> this can be false tho 15:23 < bridge> It boils down to avoiding dynamic dispatch, you know every component attached to an entity at compile time (not always but imagine if you did for a second). This means you no longer need virtuals 15:23 < bridge> And entities do not inherit fields from components, a character is not a physics object, it just contains a physics component. So no inheritence 15:32 < bridge> in my mind i have an example of teeworlda client components, i don't know how to em in a sane way without OO 15:32 < bridge> compose em* 15:32 < bridge> checked flecs, looks like full powered runtime db written in c to use as ecs 15:38 < bridge> That's the easiest example to replace since we don't add components at runtime. Just don't have virtual calls and instead of looping through an array of components just call the methods of each 15:38 < bridge> everyone just says ecs for gamedevs nowadays, but generally don't understand the motivation of that for games. maybe in bevy i can see it or a few genres. but for general purpose i honestly like objects / scripts, i do agree inheritance is terrible tho, you can still do components just by doing objects in objects. 15:40 < bridge> I don't do enough gamedev to have an opinion here. But if I were designing an engine today I'd probably have something like an ECS based on generics 15:41 < bridge> ecs can be hella fast tho 15:41 < bridge> Another interesting way of doing it is the data oriented approach of a pipeline of transformers. This has the benefit of being very cpu friendly 15:41 < bridge> if you do memory layout correctly 15:42 < bridge> never heard of transformers tbh 15:42 < bridge> ~~other than GPT~~ 15:44 < bridge> It might be terminology I invented. The idea is that you have pure functions without side effects that you chain together 15:44 < bridge> I guess you would usually still have something like an ECS with that approach too though 15:45 < bridge> i still don't understand lol 15:46 < bridge> maybe you are just describing ECS being set up in a OOP language? 15:47 < bridge> unity and some lua based framework does to call chains to setup the world. 15:47 < bridge> unity and some lua based framework does do call chains to setup the world. 15:47 < bridge> unity ecs and some lua based framework does do call chains to setup the world. 15:48 < bridge> No I was more talking about what you hinted at with ecs being fast. One way to achieve that is to have components that you transform (tick) using a chain of pure functions 15:49 < bridge> ye, still not sure 15:49 < bridge> i'm too game editor coded maybe 15:49 < bridge> Perhaps we just ended up talking about different things 15:50 < bridge> also as i said before, game devs are kinda dumb 15:50 < bridge> i do feel too far gone from actual programming concepts 15:51 < bridge> Doesn't Unity have an ECS nowadays btw? 15:51 < bridge> they do 15:51 < bridge> maybe it is finished in unity 6 15:51 < bridge> i doubt it tho 15:52 < bridge>