Thread 106189131

Embrace the Complexity Edition!!!

/gedg/ Wiki: https://igwiki.lyci.de/wiki//gedg/_-_Game_and_Engine_Dev_General
IRC: irc.rizon.net #/g/gedg
Progress Day: https://rentry.org/gedg-jams
/gedg/ Compendium: https://rentry.org/gedg
/agdg/: >>>/vg/agdg
Graphics Debugger: https://renderdoc.org/

Requesting Help
-Problem Description: Clearly explain your issue, providing context and relevant background information.
-Relevant Code or Content: If applicable, include relevant code, configuration, or content related to your question. Use code tags.

picrel sauce : https://www.gameenginebook.com/

Previsous: >>106120386

>>106189131 (OP)
These threads and /dpt/ have been failing to hit bump limit lately. Where did all the programmers that used to post here go?

what direction are you supposed to start from when developing an engine like that? anyway i was thinking about working on my own in vulkan but it's daunting as hell and i've debated on whether or not using an existing engine to prototype ideas first is more efficient before trying to tune my own system.

>>106189481
I would at least get experience using another hackable engine so you can get an idea of the pitfalls of bad design decisions.

>>106189369
I haven't been keeping tabs on /dpt/ for a long while now so idk about that but I'd /gedg/ is doing fine.
Its hard to make progress with game engine shit. Its also very common to give up on game engine shit. \
So not hitting bump limit is normal. What matters is(imho) the quality of progress that /gedg/ is making and it has been quite good, so all good, I'd say.
>>106189481
>what direction
you just Fuck Around and Find Out.
there is no right answer for that question as mileage varies from developer to developer.

>>106189369
I want to implement networking
For that I had to create some suitable commands that I can send over the network
For that I had to map commands to players
So I had to create a player component that can manage different players
The command was turn on/turn off which meant I head to deallocate it
I had a bug in my red black tree removal method, I fixed that
I had to update my data structure test project to a new version, turns out I hat two more bugs in my red black tree
Turns out there was another bug in my memory manager using my red black tree
I had to update my memory manager test project
I had to redirect user input to the network layer

Right now i am implementing the threads for managing UDP sockets.

glTF is too bare bones and USD support is spotty at best. It seems there's no good extensible 3D interchange format. UE5 project files will probably just become the default as Unreal continues to take over everything 3D related.

Why are windows and surfaces seperate objects?

>>106191529
glTF is for da web
OpenUSD is for digital content creation softwares(a must in vfx industry)
Both are very likely to stay forever. Gotta see.
>>106191734
because windows are a part of the operating system, collectively, and every operating system handles them as they see fit without a care for compatibility. Its fucked up.
So there was a need for having something like "surfaces" to abstract the windows away and deal with them somehow.

>>106191734
offscreen surfaces are useful

I'm going to have to take a break from my web based game engine development because I got a real job.

>>106193041
>web based game engine development
webgl or wgpu?

>>106193041
I stopped being a neet 8 months ago, adjusting to the lack of time took a while

Realized working solo might be futile
How do I join a team? I.e get a (remote) job as part of an indie dev team

>>106194517
>small indie dev team job
If that even exists, they are going to want portfolio of previous work

>Try to delegate small task to AI
>It is completely stumped by the borrow checker and goes in circles endlessly
Rust is job safety against LLMs.

do you guys care about vr?

>>106196124
does anyone?

Been working on client database sync stuff so far today, no progress webm

>>106196528
Too much eyestrain, I will personally never get in to it

>>106194517
>I.e get a (remote) job as part of an indie dev team

Work solo for a bit and release something or make tools.

>>106189369
I’ve been busy with other projects and my actual job. It sucks because I haven’t had the time to even shitpost here.

>>106190867
why not just use Steam Networking library which, among other things, gives you better latency because of their private ISP routing and gives you a free proxy that masks the real IP of your services to protect you from DDoS attacks
all for free

>>106197136
Because I am writing as much as possible if my engine against a memory save vm.

>>106197766
Also I don't think valve will let me use their relaying for free when I don't plan to be steam exclusive.

>>106197836
They do, as long as you release on Steam as well, Valve allows you to release on all platforms.

>>106196760
Here is a little on how that works
- When a client logs in to a prev. known server, the client shares its last sync epoch time with the server
- Server will provide updates since that epoch, so the client can front load that to its local database
Bedtime now, too tired to explain much else. Other than that, added event scheduling to the servers message queue. This is for things like chopping a tree, the entity we are interacting with has a definition which might queue the tree respawn after a few minutes.

bomp

An anon in another thread suggested I stop by here.
I'm working my way up to a legitimate game that I'm thinking I might want to have on Steam. But for now I am just slowly building competency with proof of concepts.
I am working in pygame because I have the best memory of python. I've tried C# before and it's a pain in the butt. Does it make more sense to stick to python to build competency, or am I only ever building things for myself using that langauge?

>>106200443
Build competency in what? Programming ability? Game design? Art? Also, what's the scope of your "legitimate game"? If is simple enough to be made using Pygame you may as well stick with Pygame

>>106200471
Programming ability.
The intended project months from now is a relatively simple dungeon-crawler physics game.
My concern is that, because it's *possible* to create my idea in pygame, I'll work ten times as hard to realize the vision instead of just doing it in a simpler way in unity/godot/etc because I have little formal training in programming. I'm worried I'll hit a wall and will have wasted my time. I have a day job so it's just a fun hobby at this point, but it'd be a shame to run into a wall. That's why I'm asking here.

>>106200508
NTA, but have you looked at Lua + Love2D. Lua is even easier to learn than Python and is much more popular in game dev.
My experience from 10+ years ago is that Pygame is horrible for anything non-trivial and it's only useful for learning, but that may have changed.

>>106200672
Thanks. I'll have a look.

I made a c++ json serializing using statically defined pointer to members maps with supported nesting and I'm bothered with the amount of templated code that has to be compiled for each structure and even encapsulated type combination. I'm tempted to just cast this shit to unsafe, raw untyped data to get rid of the bloat.

>>106202369
this kind of shit is why i cant wait for c++26 reflections. c# really has it on easy mode

runtime reflection adds overhead

I need to add some in game automation mechanics. Probably some "magic" powered tiles and levers for feeding ground items on a track. But I also think an npc task/slavery system would be nice, and makes more sense for "go mine iron ore" delegation vs something "magic" mine ore spawn?

>>106202924
you don't need to wait. just generate reflection code.
i'm using c# source generators to generate my own reflection, because the built-in is not good for serialization/games.

>>106198519
Progress for today:
- Created a calendar system class so I can track a synthetic "date" MM/DD/YYYY, with a 24 hour clock I have set for 1 second = 1 minute by default.
- I have 4 seasons, where I set the available types of "weather" for the season (eg: Winter can only have a snow, and blizzard). For the day/night cycle I will apply a filter to "darken" things on the screen overall (not ui elements). I don't have anything for dynamic lighting right now, maybe one day. The weather effects themselves are all shaders, which I had ready previously.

>>106193150
WebGL 1
I'm making improvements to do a sequel to a 3D web game I made that has ~100 daily users and they're all on chromebooks and tablets. Gotta get that lowest common denominator platform support.
According to web3dsurvey.com there's even a handful of extensions I can use that everybody has.

>>106189131 (OP)
genie 3 just destroyed this general. 1 year from now WoW will be made with 1 prompt.

https://x.com/GoogleDeepMind/status/1952732150928724043#m

>>106207522
See you next year then, maybe you can make a game!

>>106207522
>just destroyed this general
how?
I don't see any games published by that slop shit
>1 year from now
AGI was supposed to be 3 years ago though.
Not everyone is fine with slop shit like you, saar.
Accurate lighting, physics and detailed animations matter for good shit. AI could never perfect that.
Keep on coping.

>>106207522
also, how much does this shit cost?
I bet even AAA studios with over a day or two's worth of playtime in their games can't afford this shit.
Even (You) can't afford to play a simple game like flappy bird on this shit for more than a few minutes.
Imagine getting filtered for being poor for playing a flappy bird clone.
Truly hilarious hype.

>>106207965
>>106208012
why do you constantly have to respond to the AI shitposters?

>>106208045
I have to, to lessen the slop shill spam.

>>106197836
Tgnore what the other anon said, the networking library that is part of steamworks is proprietary, but the open source GameNetworkingSockets release is not parts of steamworks, it's a standalone library under BSD (but it has the same API as steamworks, if you wanted to opt-in).
But if you want your game to run on the web (because 90% of gen alpha prefer to play games on phones instead of PC's, and getting your game running on the web would make it easier for users to play and share in general). You need to stick to TCP because websockets are the only option.
>webrtc supports UDP like packet loss
Stick with a reliable protocol then worry about UDP (it's not that hard to support TCP and UDP). Webrtc is neat for peer to peer but if you aren't going to utilize that it's pointless (plus peer to peer shares IP addresses which means streamers are not going risk playing your game, and streamers are probably the only way to make a game viral by accident).
GameNetworkingSockets also supports peer to peer, but if you were sending a lot more data (microphone audio), you would probably benefit from webrtc's congestion algorithms (I don't think it would help with latency during optimal conditions, also webrtc has forced encryption with no way to disable it on the web, and websockets has a unencrypted mode). But if you wanted microphone audio on a native build (lets say you give people the option to play on the web AND native, because you need to debug the native build since wasm is harder to debug), you would need google's libwebrtc library which is probably going to make you realize that cmake is actually a good for C++, and google is going to show you that build systems could be much, much worse.

>>106208247
*also the main reason why you would want to switch to Steamwork version (you might be thinking, why not just release on steam using the open source library)
is because steam will protect the IP address for peer to peer games (I think steam offers it for free, but not for everyone?, but I haven't checked since then).

>>106208012
cost will be subsidized you moron. we don't need agi to generate game engines that have millions of open source repositories they can learn from. it takes ai 10 minutes to generate what you can do in months, let that sink in. google + xAI winning massively.

>>106208671
>cost will be subsidized you moron
link me google announcing that shit or fock off like a true retard
>it takes ai 10 minutes to generate what you can do in months
We know you and your AI can't do any real shit.
You keep on coming up with random stuff to cope about. Keep on coping.
Feel free to create a 3d game engine by using your google + xAI combo to prove me wrong.
Nobody's stopping you.
come on, go ahead.

>>106208752
I thought a lot of these posts praising LLMs and making big claims were just bait posts. But I took a look on YouTube without being logged in. There are countless videos about how some zoomer replaced his "team of developers" with "AI", selling courses and even prompts/workflows. There seems to be a direct need to exaggerate claims to make sales or at the very least get more clicks.

I use these tools every day. On a good day, it can make me about 30-40% more productive. But that's because I know how to use it well and most importantly know when NOT t. I have, several times, tried various "vibe coding" options (like Cursor). I've even paid money to try them out. They really like to generate code. Which is great for the boilerplate and basic stuff at the beginning. But once your project starts to get closer to the context size it starts to fall apart. You have to start being very selective, only let it see certain files and give it very precise instructions (like a very small and specific task). At that point you might as well just be pasting files into a regular language model, it doesn't give you much of an edge.

Now maybe in the future they'll figure out how to solve this. Some approaches like getting it to do a kind of Socratic dialog with itself or first produce a spec, then it reviews its own spec, then implements and so on. They seem promising but for the most part they just burn tokens. And all these tools start out great with "no limits" but even the paid tools are all putting limits (and they all seem to do that when they hit enough users). From what I can see the "agentic" stuff is just a meme. There's no actual hard data that shows getting it to talk to itself in a loop improves performance. But it costs tokens though. The best workflows I see are ones that have been really fine tuned for specific micro tasks (like run test coverage for what you changed and then set a new goal of adding tests automatically). But you can just do this manually.

>>106208752
He will show you next year, just you wait, he will finally make something

>>106208834
AI was already as good as it could be.
Just generating real time videos by slapping in more GPUs isn't an advancement but an application of the tech.
All the data has been scrapped. Training on more data will oversaturate and make the model perform much worse.
its over. Just a matter of time.
>>106208843
good for him.
I'm not a cripple that can't do shit without AI.
I can survive well without it so no waiting and pure action for me.

>>106209794
Don't forget that when these models also use their own outputs as training data they get much worse too. Even at a glance in things like image searches I see absolute slop results. AI maps and even anatomical diagrams that don't make any sense are likely being trained on by these newer models. And that's just at a glance with a handful of searches. I'm sure text is worse.

Even places like github are being shitted up with 250k-500k line projects of absolute slop. I don't know if you saw the chink's PR to some video editing tool? Hundreds of commits and pure slop. Critical stuff was removed in addition to lots of absolute nonsense code being added. "but it kind of does what I want" - even though it's a giant mess.

>>106209794
>>106209827
spoken like a true simpleton, elon is training grok 5 with synthetic data. don't presume you understand ai better than the richest man on the planet with the second best ai.

>>106209962
how can anyone be this subservient? are you indian?

>>106209962
>spoken like a true simpleton
I am a simpleton though, just not incompetent.
>elon
choke on his balls and donate all of your wealth and future earning to him so that he could build it much faster.
>don't presume you understand ai better
I didn't. I only presumed that I understand AI better than you.
>second best ai
and now come the same question second time
You've made a joke out of your AI shills last time by pointing out pricing and "Elon", both of which are highly irrelevant for attaining the title "best".
Do enlighten me on why it is second best, what is the first best and what determined this "best" thing.
>>106210122
>are you indian?
if Elon Mollosk was paying that anon, everything'd totally make sense of this fellow pajeet.

>>106211617
biting the baits if fun than just bumping

>>106211275
https://x.com/tetsuoai/status/1954429989618844066#m

not mine, but I run a AI twitter account which was personally retweeted by elon himself. state of the art AI is google for now due to their inherent advantage of not using nvidia gpus, they probably have the smartest team deepmind.

how the fuck are old gba games so perfectly crafted...................

>>106212111
small incremental improvements over decades

>>106212111
Fixed point numbers, son

>the only good dds converter is locked behind registering an account with nvidia
i fucking hate how everything requires an account these days. back in my day we could just fucking download things

>>106212111
checked
the GBA is probably the programmer's perfect console. it and the virtual boy are a world ahead from everything that came after architecturally, simple and solid and massively effective.

>>106212791
It is quite annoying. At this rate even my compile is going to start demanding a username and password.. maybe I should write my own language to be safe

>>106189131 (OP)
I have a desire to make a very realistic tank driving and shooting simulator from a 3rd person perspective (no tank interior). Very similar to World of Tanks, but more detailed. War Thunder physics is bullshit, M3 Stuarts slide like crazy carpets on ice.

I've tried with Godot, but don't really jive with it. It felt like a lot of clicking and tedious tasks. Maybe I need to learn how to mirror actions so I don't create tank tracks two times - once for each side of the tank... Godot also forces you to use RigidBody3D, and other Nodes, and click through all their parameters. Too much navigating a UI and clicking to configure everything. I experimented with native VehicleBody3D and WheelBody3D, but am unhappy with their capabilities. It's not realistic enough and it required too much tweaking of their native parameters (which is a lot of UI navigation and clicking).

Does anyone know of a game framework that is more developer friendly, i.e., you can make models and give them physics programatically rather than slogging through a UI?

I'm open to using anything (You) have made if it allows me to create a realistic tank easily.

Added random effect selection with weights so I can have a pool of animations with the same start/end keyframes and play them in a loop (the breathing/blinking while the menu is up in this example).

What safety precautions can I add to my game if I’m making something like Inzoi/Sims but online multiplayer? I never understood roblox and never bothered to see what’s it about but there’s this youtuber that showed how bad it is for youth from creepy users.

For context I don’t have a messaging system in the game and it’s only based on emotes (like what Nintendo does)

Continued progress on laying track

>>106207019
6/6/06 Easter Egg is possible now

>>106214008
>It felt like a lot of clicking and tedious tasks
got filtered by editor's UI. I know its annoying but gotta bear with it.
>Godot also forces you to use RigidBody3D
so will unreal and unity. If you want to simulate tanks with accurate physics, rigid body is where all for newton's law of motion are at.
>Too much navigating a UI and clicking to configure everything
that's inevitable with any game engine
>I'm open to using anything (You) have made if it allows me to create a realistic tank easily.
sorry, what you are asking is something that requires a Ph and a capital D in physics and a lot of experience with programming and software development. Not something a single man can pull off as an amateur hobby project. It requires serious money and a bunch of highly skilled people to perfectly create what you are asking and its out of poor /gedg/'s scope.
>>106214873
noice

>>106214873
cute clown, nice smooth animations anon

Decided to take an old dead SFML project and move it to Raylib. A lot of cleanup needed, but maybe 3D is going to be easier than I expected. Happy with Raylib so far

>>106211617
taking bait head-on and successfully dismantling both it and the baiter is far better than ignoring bait.

>>106216188
no

>>106216201
you're as bad as baiters.

>>106216188
>dismantling bait
Yeah he's so fucking dismantled by your facts and logic. Totally destroyed by you showering him with (You)s.

Fucking retard.

>>106216222
no that'd be the guy who writes up pages of text in respond to low effort AI baiters
although this thread is fucking dead anyway so I guess it doesn't matter

>>106211673
>did not take any money
>got bought by a single retweet
A third rate whore is more expensive and interesting than you.
Let go of AI and find cling onto something worthwhile.
>>106216262
I was just bumping with replies at one point.

>>106216258
the point isn't directly contending with the baiter, you massive retard. it's about undermining the effect they're trying to have on the thread. if you quickly and succinctly do so, you made the bait have the opposite effect the baiter intended. if they start seething (like you) it's a bonus.

>>106216370
>early 2000s forum neckbeard tier idiocy
An achievement, really.

>>106216389
>continues to prove my point
>complains about the thread worsening while being the main cause

>>106189369
Too busy fixing boring bugs at the moment. I'll have some noteworthy progress in a few weeks maybe.

3d modeling is such a gooner hobby, im at half mast the whole time. also why didnt anyone tell me to use a base mesh at the start, learning how to just make clothes is pretty fun

>>106217608
mm yes ugly middle-aged women

>>106217693
yeah, i think those guys are funny. i keep getting older yet the age i find attractive is always the same

>>106217693
milfs are hot, chud

>>106217850
not that one

>>106216288
https://x.com/tetsuoai/status/1954524767534411829#m

keep coping, grok just one shotted doom 3d fps.

>>106217944
>git clone github.com/DOOM
Damn look at the one shot ability

>>106189369
People are programming instead of posting

Created various weapon sprites. and these rocks just now

>>106217944
you are the one coping with "AI will take over in a year. All of you are bums for not using AI"
I am just chilling without AI because I'm not AI crippled.
having more GPUs and more talent behind a product doesn't make it the "first best".
Your ranking of AI is as delusional as how much you believe that I'm coping.
>grok just one shotted doom 3d fps
so what?
>me opens unreal
>me creates a new project with 3rd person template
I zero-shotted a 3d fps game.
It looks like its not easy to make grok generate code for that, took several attempts and is also a very challenging task that requires getting the prompt right all the time while expecting the AI wouldn't fuck it up midway.
>>106217855
nah, that one too.
Its too early to that without looking with the textures on. Mesh without textures can be very misleading.
>>106218626
noice!

>>106218626
why would I prospect that crap? I see the color right there
same problem runescape had and they removed that option

>>106219391
Your default interaction key will do the first action by default. As for that, runescape did the wrong thing (as usual)

I posted in the bad thread, forgive me

>>106219320
grok writes the engine dumbass, ironic you would cite unreal. you belong in agdg.

Time to finally get rid of vertex attributes and switch to a vertex pull setup, will also implement some amount of dynamic state. Pipeline should becomes significantly easier to wrangle. Although if we’re beyond honest it’s the layout’s that are a pain to deal with.

>>106220792
>grok writes the engine dumbass
grok wrote three.js?
So grok AI has been around since 2010 and created engines back then but struggles to create a native engine now, after tow and half decades?
>you belong in agdg.
I belong in the asylum and you truly deserve a permanent rest 6 feet under for your brain dead AI shills.
Kindly do the needful, saar.

>>106214998
some kind of multi-layer challenge/response that locks being able to persistently play with the same person behind some kind of parental controls
the emote only thing is good though

>I never understood roblox and never bothered to see what’s it about
4chan but if each thread was actually a free to play game, you could pay extremely disproportionate amounts of real world money for crypto you could only spend to look gayer, and the mods openly paid people who made "good" posts in the gay crypto for significant amounts of actual labor with the promise they wouldn't plunge its value

literally all you have to do is have a moderation team at all and you'd be miles ahead
like aside from the constant reports of users being groomed, it's my understanding that organized crime in some third world countries would occasionally exploit young lua devlopers economically and made them work on roblox games
presumably it's venezuala since they were already doing something with runescape gold farming

>>106220818
>grok wrote three.js?
did you write your engine in assembly? what a self own lmao

>>106221133
everybody knows grok didn't write tree.js, you brain dead retard.
Thanks for showing that AI is shit and its shillers are shilling hype based on lies. Thank you for your retarded efforts, nobody will seriously rely on AI for game engines now.
gork only vomited three.js code. If you are too retarded to think that an engine's code base would fit inside a single tweet with a couple of scrolls length then...
Kindly do the needful, saar.

>AI game engine in JavaScript
where do you retards come from

>>106222411
this one spawned as a by product of Elon Mollosk's retweet of his tweet.

>>106222411
Retard tripfag and some others (some/all of which might just be the tripfag without his trip) keep feeding him.

>>106223603
>some/all of which might just be the tripfag without his trip
lmao. Thanks for letting me inside your head, saar. Its very comfy.

After 36 hours of vibe coding I finally did it, I can count the distance between 2 points. Its not 100% accurate though, but close enough.

>>106223634
>36 hours to do (x2,y2) - (x1, y1)?
you are cooked saar

>>106223649
before anyone tries to correct me, I know where exactly I'm wrong. Its completely intentional.

>>106223649
I cant read code, Ill have to copy paste this when musk gives me more credits

>>106223692
sadge

Should I use two sockets - one for sending and one for receiving - in my UDP Host?

>>106225499
what are you trying to do?
You will always need two sockets for a duplex communication.

>>106225535
I am trying to implement a dedicated server that should handle multiple clients.

>>106225560
you'll need two sockets.
If we consider a typical game scenario, you will have two sockets at each end, one for the user and another for the server. Initially, the user will be the sender that sends game data about entity position and properties to the server and the server will be the receiver that receives the entity's data and processes it for validation against cheating or storing it.
Once the server is done processing, the server will be the sender and the user will be the receiver when the server sends the acknowledgement that this entity with these properties is legit and the user and all the other users are allowed to interact with it.
So depending on from where to where you are transmitting the message, the sender and receiver changes. You only need one socket at each end and since its UDP, your server can have only one socket and that can be used for transmitting message over to all the clients. The server having to deal with multiple client requests at once will be annoying. Just implement a simple message queue and chill for starters.

>decide to make my own ECS because flecs/EnTT/etc are too complex and bloated
>it's harder than expected

>>106226322
std::unordered_map entities_in_level

ez

This honestly isn't that complex it's just pedantic. Like some of these things can be just a header file or function or skipped entirely.

Message bump

>>106229277
>hashing an integer
add a specialized hasher that does nothing

How are you guys still failing with gpt 5 holy shit you guys are retarded

>>106231929
it gives you an even distribution

>>106232117
yes, but it fucks the locality

>>106232139
why would you want locality?

>>106232185
Locality, and its cache friendliness, is the killer feature of ECS.

>>106232185
why would you *not* want locality?

>>106232318
because it makes hash table lookups faster? what do you think the point of hashing is?

>>106232644
If you structure your ECS properly most of your lookups won't hash table lookups but direct array index access, which is much much faster.

>>106232801
who the fuck said anything about ECS

>>106232818
Follow the reply chain.

Project hopping time.

I wanna make a lowrez game for the lowrez jam.

What should I make?

Optional themes:
>Is Physics Real?
>Opposite Day
>It Happens Again
>The Player Is Replaced By AI
>Liberate
>Just One Button
>Undo
>The Floor Is...
>There’s Always A Bigger Fish
>Left Behind

>Try to learn OpenGL or DirectX11
>Every book is using 10+ helper utilities splitting code in 10 different files, blah blah just to get a simple cube on the screen
>let's learn what "everything" does and then we are going to use a subset of that to draw a box
>then just teach me how to draw a box and then explain what each line does
What the fuck do I do?

>>106235441
1. Start with the absolute basics, draw a solid colour triangle
2. then change the colour of the vertices
3. now draw another triangle to make a square
4. draw two squares perpendicular to each other
5. complete the rest of your cube

You can do that all as just lines of code if you want or add whatever level of abstractions you want as you go (like function to draw square or cube).

Once you have your cube you can look at the other drawing functions, get into texture mapping or shaders and so on.

If you're completely new to programming play with a higher level language to draw your cube, you can always port it to something like C or C++ later.

Oh and you can absolutely ask something like ChaGPT to help you with this. Don't get it to just generate the code all for you, you won't learn much that way, but tell it at a high level what you want to do (e.g. OpenGL with C++). You can even take the 5 steps I gave you and tell ChatGPT to generate a tutorial for you, add in terms like "minimal examples". Type out the code and run it.

>>106235322
"Left behind on the surface of the sun again, help the ai jump to safety (to its death liberating us, since ais are evil)"
that sound cover the themes

>>106235441
in all honesty, just give up, genuinely, just fuck off somewhere else

>>106235441
everyone here that uses opengl/vulkan/whatever and got something working and cool has been fighting with either C/C++ or opengl for more than a few years (4 years is probably the average time it takes to "figure out" opengl / vulkan or C/C++ if you are just starting).
also 4chan is for boomers, and slowly shrinking, discord is unfortunately where smart people go to make friends with other smart people (AKA high functioning autism, same 4chan but more friendlier / not so schizo because they actually take their meds).

game jams are for unity slop. how the fuck am I going to plop the 100k lines of boilerplate in my engine into something else.

Anybody else here is using Rust Rapier2d? I am having a problem with the Query Pipeline, the example in the user guide doesn't work. The compiler says that QueryPipeline::new(); doesn't exist anymore.

How do I solve this?

>>106208834
this lol. idk much about AI but have they figured out how to get it to actually problem solve yet? I understand how the generative side works, but I would love if I could have a computer actually solve complex tasks. I feel like a problem with requirements can be abstracted fairly easy. Hell I bet DARPA already did this years ago.

>>106237740
Game jams are stupid
Prototypes are stupid
Vertical slices are stupid

Something went terribly wrong with this terrain generation code

>>106238651
prototypes are important

>>106237979
Read the docs? Look at examples?

>>106241354
The examples are wrong. The docs are sparse and not helpful
https://docs.rs/rapier2d/latest/rapier2d/pipeline/struct.QueryPipeline.html

>>106241354
>>106241423
this is where you use ai to search to entire internet for you and output something you want.

>>106238651
@grok is this true?

>>106241423
Well, all the fields are pub so you're probably supposed to construct it manually. Everything looks like it implements Default except forQueryDispatcher which was doubtless always going to be &parry2d::query::DefaultQueryDispatcher.

>>106239396
If you made the mistake I did your indexing optimization is wrong.

Guys. Open GL 3.3 or Vulkan for an Orthographic 3D Game with Pre-rendered LudoKino Isometric look?

My main worry with Vulkan is that the minimal NVIDIA GPU is a GTX 650 and it feels a bit strange when my games look retro.

>>106241874
>13 year old GPU
I doubt anyone still running with GPUs that old is really playing games.
But if you want to go full lowest common denominator (without going pure software), OpenGL ES 2.

>>106241898
yeah I am thinking this conclusion as well.
Steam Hardware Survey affirms what you just said.

And Vulkan does have a lot of control benefits.

>>106241903
In a lot of cases, Vulkan really isn't "necessary". But I use it in my own (currently procrastinated) projects because I just think it's more fun to write.

Don't make the mistake of targeting pure 1.0. A bunch of new APIs that make everything less tedious (e.g. dynamic rendering) were implemented as compatibility layers on top of the old API, so shouldn't have any affect on hardware compatibility.
Although I took the "i don't care" approach and am just targeting the latest one.

>>106241874
There's no practical reason to be using Vulkan

>>106189131 (OP)
Is it okay to program in Godot using only C++? I don't want to learn another language just for 1 thing.

>>106242115
nah not really

>>106242115
>learn another language
kill all summerfags

Is it one of those unspoken things that everyone cheats in game jams?

does any of you guys have any experience with the steamworks c++ api?

>>106238132
>how to get it to actually problem solve yet?
It depends on the problem. It can do simple "chores" on your codebase well enough. That's what I like to use it for. Those kind of "shallow" problems where you're telling it to refactor or fix some minor problems the interpreter or compiler is complaining about. Those I would call "shallow" problems where you're maybe thinking for 5-10 seconds and then you're typing.

But when it comes to more "deeper" problems, that require more thought, unless there are examples out there on the internet of your problem then it can struggle. But it will never tell you "I don't know", it really wants to generate text/code so it'll give you something even if that doesn't make sense. So there are times where maybe I'll paste in some code and an error and try it a few times and know it's nonsense it's spitting out and that it's not suitable for this type of problem.

I've also used it for a quite few small toy projects, like using PyBoy. And it will generate code that's for an old version (because there are many more copies of that on the internet). So it's like, yeah, I could convert the docs to markdown and paste it in and then my instructions, that'll work, but the time wasted looking at the deprecation warnings and errors from the old code, you could've just skimmed the docs and wrote it yourself.

Also it's hit or miss depending on the amount of examples on the internet, like it's going to do a lot better with something like React components because of the sheer amount of that all over the web. If you're going for some niche programming language/library then it's not that "it'll struggle" it's worse, it's that it'll just make up stuff. Stuff that even looks plausible on a skim of the code but then you add it and it doesn't even work.

Maybe in the future they can work in some kind of "confidence" into its answers where it'll act like a human and tell you it's not sure but for now there's no distinction.

>>106189131 (OP)
I know this has been asked a million times but wheres the definitive resource to go for learning graphics. I want to build some shitty neural network simulation.
Should I start lower level or just jump to raylib?

>>106235508
>>106235931
>>106237709
Alright finally got it nailed down.
https://pastebin.com/iB8w7h7Y

TLDR:
Vertex Shader -> Performs Math on Vertices (Object->World->View is just a special case)
Geometry Shader -> Performs Math but can also manipulate data (creating/deleting vertices)
Fragment Shader -> Performs Math on Fragments (Lighting, Materials is just a special case)
Shaders are just callbacks that the GPU calls while iterating over vertices/fragments/etc respectively.

>>106246272
The vertex shader puts your vertices (through whatever math you do) into screen coordinates (0.0-1.0 x and y). The rasterizer then translates those coordinates into fragments(pixels) which is whatever your framebuffer size is, for example 1920x1080. The fragment shader is where you choose the color of the fragment (or discard) it via whatever math you’re doing such as phong shading.

Anyone tried making a renderer that doesn't use UV coordinates? Something like Ptex.
>that's not a renderer level thing
If you want decent performance out of it, it will have to be.

>>106245237
learnopengl.com

Bump. Got a ton of sweet progress done lately.

I'm designing the on-disk format for my resource database. It's pretty difficult to avoid serious flaws that lots of other resource databases have. For instance in Unreal they identify resources by their paths, so moving a resource inherently breaks things and they've made it so you can't do that without using the editor. Unity requires you to sync meta files in source control along with the asset files otherwise things will break on someone else's computer when the editor auto imports the assets with new GUIDs. I'm more interested in a simple layout of text files that isn't super brittle. Tricky thing to design, though...

>>106249358
meta is more than just guid. it's the "asset" part pf the content. import settings, for example. it can be used to create multiple assets from a single content file.

>>106249553
Yup, I took a bunch of notes on how various resource databases store the different parts of the information (identifier, metadata, raw data, cooked data, dependencies, etc.) and I was simplifying. I'm kind of surprised that no unix-y plaintext files model seems to exist from what I've found

>>106249358
My asset format will just be like .Net, You have namespaces and types and they are stored in dynamic asset libraries and shit.

>begun integrating the steam api
>the steam overlay is using d3d 11on12
>the debug layer warns that their implementation uses the wrong initial resource state for buffers and old ass apis
well, so much for wanting to catch all warnings, have to disable one of them just to get it working

>>106207522
AI increases productivity of those who're passionate and talented enough to use them right. Even when AI can generate interactive games well, many still need stories, creative ideas etc. AI will support this too, but for now we should assume that AI + human(s) beats AI alone when it comes to games.
Most people (I included) are consumers, they don't have (many) good ideas. Give them a prompt to generate "their" game and it'll be lame and predictable after a while even for them.

>>106215908
This looks cool as fuck, reminds me of the rts game I’m working on in raylib although I’ve found raylib to be a little inconvenient at times

>>106250701
AI increases productivity of beginners who can't program

>>106243218
How so?

>>106252052
I know AI can't code anything more complicated than boilerplate, but is it a good idea to use ChatGPT-5 to learn how to code? I like being able to ask questions and get answers without waiting 5 years just to get a condescending gatekeeping reddit reply. I feel like things are clicking and my shit is working but I can't shake the worry I'm going to develop bad habits.

i almost figured out how to implement prefab variants.

>>106243218
Yes. Game jams are specially made so that game devs can cheat to their heart's content. Everyone will not have the same starting point because all of them don't have PC with same specs and experience. It's basically pay to win when it comes to putting good assets in your game. Confining our game development to a small time frame will only force us to create slop shit. Only good and fair game jams are the long duration ones. Shirt duration ones are not worth it.

>>106214873
why's she so winded like she just climbed 20 floors up stairs

>>106254273
Yes that's what AI is best for. A replacement for google, reddit, stackoverflow, and random discords.

>>106257258
Who the fuck thinks PC specs have anything to do with game development lmao

>>106257805
I had to learn and use SHITdot for 3 years because my laptop couldn't ran unity or unreal

is it just me or nearly every new indie game about deck shuffling now?

>>106258811
Society is all gambling, prostitution and crime now

>>106243218
yes, the point is the deadline, not how you get there

>>106246638
>Anyone tried making a renderer that doesn't use UV coordinates
no because it's standard and has the most documentation

>>106246272
>shaders are callbacks
they're programs that run a jabillion times in parallel
>geometry shader
Don't use these anymore, performance can be shit on a lot of platforms where they're implemented in software (on mobile in particular), compute shaders have superseded them.

>>106242115
If you're a C++ pro yeah, it can be messy

>>106258811
lowest common denominator has always been a success in any entertainement field

Learning how to use clay in my freetime before work.

>>106260968
The learning curve is a bit weird but I think I'm getting the hang of it.

>>106257566
And remember to trust it only as much as you would trust a google, reddit, stackoverflow and random discord.

>>106189131 (OP)
Posted over in aggydaggy too.

>>106261434
none, download a C compiler and SDL

>>106261471
Didn't ask about that. I want to make a game, not circle jerk about reinvented wheels that are only real in my mind.

>>106259368
Yeah that's why I'm looking at alternatives.

>>106246638
I have been thinking of this a bit.

You could store matcaps in an array texture.
Either pass the array index as shader or instance uniform or replace the model UVs with index.
This should be relatively performant for models but the environment might look better with triplanar texture mapping.
Oh and if you want to support older graphics cards then instead of an array texture use a texture atlas and replace the 1D index value with a 2D index.

Both of these techniques are relatively well known. Matcaps in mobile dev and triplanar in AAA.
There are probably extension to for godot and unity too if you don't want to implement them yourself.

>>106261998
Use case?

This glitchy Visual Studio icon is stuck on my screen as I type, it happens on a daily basis, how do I turn it off?

>>106261486
kill yourself then, everyone else has already done everything you could possibly do in life

>>106262876
post code.

>>106262482
NTA but not having to care about UVs, especially on runtime generated geometry, would be pretty nice.

does your game includes cute females with very cute and hot legs?

>>106262524
You just accept weird bugs with visual studio, every version has it's own quirks in it, you could make a bug report if you really care but then when that version comes out it will just be something else, only solution that worked for me is find a version where the bugs don't actually annoy me that much and then just never upgrade it

>>106258811
I literally think about adding a card mechanic to my RTS game a few times a month just because I feel like I am missing out

>106263584
This place is gayer than reddit.

>>106263640
Welp, that sucks. But I'm still glad I switched my version a while ago, the new font is way easier on the eyes. Thanks, Microsoft.

>>106257258
What do you count as long duration?

>>106258656
Just use Defold
>NOOOOO 1 INDEXING AAAAAHHHHH

cross platform is a meme right?
a friend got my win32 d3d11 winsock game running on his linux machine simply by adding it to his steam library.
yet for some reason I still feel autistically drawn to opengl and glfw

>>106263792
But can it run on the browser?

>>106263998
I started with the browser but when my game captured the mouse there was this persistent unremovable popup going "press esc to regain mouse control" and that's a dealbreaker

>>106263792
It only works on Steam because of Proton and the API translation layers, glfw and OpenGL are already crossplatform

SDL3 / SDL_GPU is the way to go if you want something highly portable but you have a finite amount of time

>>106263584
Get out of this general, you are not welcome

>>106263584
stay in this general, you are welcome

>>106264366
Are you also a pedophile like him?

>>106263584
Howdy do there Woodchipper-American.

phew, finally out of ban
>>106261148
looking good. Noice to know you are still kickin'
With clay, UI styling becomes very manageable to an extent, good enough for our toy projects.
>>106263792
that's proton+wine magic.
You can play all of your rpg maker and unity games made for windows on linux without any issues by just opening them directly now.
Wine really is making windows obsolete. Or already has and we were too late to notice.

I'm mixing colors both in shader code and higher level code that puts the results in attributes. The results look different. Is this because of color spaces? Do I need to call a bunch of pow() every time I mix colors? Is this required in the shader code or in the application code? Isn't this wasteful? Help me frens

>>106264079
wine API is better and more stable than "native" linux API
there is literally no reason to use native linux API unless you want to cause yourself suffering

>>106263156
But you will have to care in some other way.

>>106246638
>Something like Ptex
Looked it up and seems like it uses UVs.
It doesn't assign UVs manually per vertex like the traditional way but generates them per face procedurally so that we don't have to deal with the first handed UV assignment torture.

>>106263584
it will, as a main character

page 10 Progress deadline

>>106269527
are you mocking us?

bevybros
https://x.com/MalekiRe/status/1926803230643671373

what do you guys use for music creation?

>>106271603
I use reaper daw with free vst plugins
Its more than good enough for a low skilled person like me.

>>106269527
WHAT YEAR IS IT!

Alright, I have learned the basics of 3D pipeline and it's time to commit to one of them. OpenGL is the obvious choice right? Is there anything decent about DirectX11/12 apart from easy integration which is intentionally done by microsoft by not providing modern OpenGL headers in the default SDK.

>>106272610
d3d has explicit pipeline state management as opposed to opengl's global state machine model, so it's harder to make weird bugs using d3d. go with opengl though, fuck microsoft

>>106272610
OpenGL is kind of crusty these days. Even if you don't give a shit about supporting Apple. OpenGL 4.6 is limiting. Vulkan is the best choice, or a (extremely thin) wrapper like SDL3GPU, D3D12 is only worth targeting if you care about xbox, same with Metal. Vulkan will run on Switch 1 / Switch 2 / Apple w/ MoltenVK (and soon KosmicKrisp), Windows, Linux…

D3D12 and Vulkan have about the same level of verbosity, and in practice you write a thin rendergraph so that you abstract the boilerplate at level that makes sense for your application.

>>106271357
>me, a Jai GOD, when plebs with crabs have to worry about compile times

is having a strong background of mathematics beneficial for making a game engine?

>>106273807
Better than not having one but most of the things you'll be doing are high school maths and maybe year 1 uni.

Good mathematical knowledge however will help you find better solutions than the normie shit gamedev youtubers would peddle.

>>106273325
Any day now its going to actually be released to the public. I cant wait to give it a try desu

>>106189131 (OP)
Is 2048 /gedg/ or /dpt/? Doing it just to learn and to get back into the habit of programming.

>>106273807
It's a head start but it's applied math so they stuff you learn in school mostly doesn't apply, maybe 6 months

>>106273807
yes
https://www.youtube.com/watch?v=YNtoDGS4uak&t=872s
>>106274902
both?

>>106275043
Ok, it's very bare bones so far, I'll replace things with loops using a constant for grid size instead of hardocding 4, but so far, it's this:

\ 2048.fth

\ Make Grid
variable grid 15 cells allot
grid 16 cells erase

\ Init Grid

: initGrid
grid
dup 0 cells + 1 swap !
dup 1 cells + 2 swap !
dup 2 cells + 3 swap !
dup 3 cells + 4 swap !

dup 4 cells + 5 swap !
dup 5 cells + 6 swap !
dup 6 cells + 7 swap !
dup 7 cells + 8 swap !

dup 8 cells + 9 swap !
dup 9 cells + 10 swap !
dup 10 cells + 11 swap !
dup 11 cells + 12 swap !

dup 12 cells + 13 swap !
dup 13 cells + 14 swap !
dup 14 cells + 15 swap !
dup 15 cells + 16 swap !

drop
;

\ Draw Grid
: drawGrid
grid
dup 0 cells + @ 4 .r
dup 1 cells + @ 4 .r
dup 2 cells + @ 4 .r
dup 3 cells + @ 4 .r
cr
dup 4 cells + @ 4 .r
dup 5 cells + @ 4 .r
dup 6 cells + @ 4 .r
dup 7 cells + @ 4 .r
cr
dup 8 cells + @ 4 .r
dup 9 cells + @ 4 .r
dup 10 cells + @ 4 .r
dup 11 cells + @ 4 .r
cr
dup 12 cells + @ 4 .r
dup 13 cells + @ 4 .r
dup 14 cells + @ 4 .r
dup 15 cells + @ 4 .r
cr
drop
;

: getRow ( i -- a b c d )
grid swap
( grid i )
4 *
( grid 4i )
cells +
( gridi0 )
dup ( gridi0 gridi0 )
@ swap ( a gridi0 )
1 cells + ( a gridi1 )
dup ( a gridi1 gridi1 )
@ swap ( a b gridi1 )
1 cells + ( a b gridi2 )
dup ( a b gridi2 gridi2 )
@ swap ( a b c gridi2 )
1 cells + ( a b c gridi3 )
@ ( a b c d )
;

drawGrid
initGrid
cr
drawGrid
cr

0 getRow .s cr
drop drop drop drop .s cr
1 getRow .s cr
drop drop drop drop .s cr
2 getRow .s cr
drop drop drop drop .s cr
3 getRow .s cr
drop drop drop drop .s cr

.s cr bye


The goal is to do it like in a loop: getRow/Col (invert optionally) fall (invert back) putRow/Col.

>>106275095
Forth?
now I want to make a 2048 in forth too.
Maybe I'll give factor a serous try now.

>>106275121
Yes. It's fun and simple. The invert word (which I renamed revert because there as already a invert) is just:
: 4revert ( a b c d -- d c b a )
swap 2swap swap
;

But that will obviously have to change when it accepts arbitrarily sized grids.

>>106273807
no

Elliptic curves are being more complicated than I expected, but now I only need to finish a few bugs before having ed25519 ready

>>106276067
Geez I posted on the wrong thread, though I was in /dpt/

bomp

progress report:
>implemented nanosleep in win32, no longer does the cooler fans sound like a jet engine

>>106266261
>Is this because of color spaces?
most likely.
You did not provide much info here.
In Vulkan, if you set your present image format to srgb, colors are implicitly converted from srgb to linear, all calculation within the shaders will be done in linear and then this linear is converted into srgb before presenting to the renderer image(that means you can avoid doing pow() as the drivers automatically do it for you).
If your final present image is of linear format, then you doing manual corrections with pow() will be necessary if you want to present srgb colors on a linear image.
Check what format the image you are present to is.

>>106217944
>>106220792
this doesn't mean anything. there's a gorillion threejs shitty FPS tutorial code to choose from on github. the point of programming is being able to ITERATE on something and AI is pretty awful at that. at some point you will ask it to do something it doesn't quite understand and it will get stuck in a terrible loop of just making the code worse and more broken. the only way to circumvent that issue is to understand the code yourself and hold the LLM's hand.

Testing scroll. Made a log which is just a ring buffer.

Forgive the message, it was prepared for the subhumans in the other thread.

>>106267265
You pretty much assign a tiny texture to each triangle instead of mapping it via UVs yeah. Obvious advantage of never having to care about UVs, especially for runtime generated geometry.

Drawbacks are, well, one texture per triangle (though that's the most naive implementation), difficulty managing triangle boundaries with texture filtering, and you obviously can't use common UV tricks (e.g. panning one of the coordinates for scrolling textures).

If you want it to run at runtime with decent performance you probably need some pretty low level renderer optimizations. The texture filtering issue is easy: don't use texture filtering.

>>106280129
noice!
>>106280165
yeah their approach works when performance and memory are not of high concern, like in the vfx industry.
But for us game devs, sticking to assignment of UVs per vertex gives the most performance with lesser memory usage while providing high control, so we will have to deal UVs directly. The only way to make this less annoying is by making compute or fragment shader generate UVs for us from the input info buffer that the CPU passes over. Maybe bindless will give better performance for a Plex implementation, gotta see.

How do you do navmeshes? I'm thinking of using recast/detour to do it but how do you even go about defining a floor for it? Do I import a separate model for the floor?

>>106280129
You should really set up a hotkey to start/stop recording. That pause/break key sure is looking lonely.

>>106280548
find all triangles that are obviously a floor (normal facing upward at a not too much big angle from straight line) then find it's neighbours and construct a graph from them so it can be traversed by whatever pathfinding algorithm you want.

>>106271357
>https://x.com/MalekiRe/status/1926803230643671373
looks neat is pic how you do nearest neighbor search in bevy? I never understood how they do that in ecs seems like a massive pain in the ass for their desired data layout.

>>106280129
nodev? im working, im just taking a break right now. i promise

>>106274901
Try the leak.

>>106281524
Feels rude desu. I'm going to wait until he releases it properly

>>106189131 (OP)
AI block is too small. AI should encompass all of that. AI is the future. Use AI or you're hurting yourself and everyone around you.

>>106282051
this diagram was made before the term 'AI' was coopted to mean hypeware, similar to 'crypto'.

>moved game to singleplayer instead of multiplayer
>things start moving
i felt for the multiplayer friendslop meme

>>106189369
I've moved onto electrical engineering.

Hytale is getting resurrected and the ECS engine is getting nuked. Thoughts?

>>106282996
im thinking in hot anime girls

>>106282996
For a game that size, it's either ECS or OOP.

inb4 "just like put the struct in the array and iterate them bro"

>>106282996
>ECS engine is getting nuked
i predict they will actually ship it this time

>>106282996
ECSlop

>>106283020
absolute terminally online brainrot

just write the code to make it do the thing you want it to do and when the code gets too messy think about how to make it less messy, bro

>>106283047
>just write the code to make it do the thing you want it to do and when the code gets too messy think about how to make it less messy, bro
Yeah and that's exactly how you end up with ECS in any project bigger than simple platformers. You don't think "ECS" is one extremely specific pattern with specific features right? Because it isn't, and it especially isn't some overcomplicated 100k+ LOC bullshit.

If you have entities that have components that are operated on by systems, it's an ECS. At the end of the day it's really just about a 3D storage structure and that you filter and select stuff based on components. Almost all complex games end either in that or OOP.
>I add torches to my platformer
>they need to heat up the player if he's nearby because I have snow levels so I add the torch struct to an array and iterate on it to do logic
>I add campfires which do the same so I just add a type member to the torch struct and call it "fire object" instead
>oh wait campfires also let you cook shit on them
>oh wait that's a lot of extra data to add to each torch in my level and why iterate over all torches too when I only need campfires?
>ok so I put campfires in another array
>now the player can build stoves that heat the player, can be cooked on, and can be upgraded too
>ok so I add another array... wait what if I just pull the common data for heating the player, cooking, and upgrading into their own structs and then I add them as members to the torch, campfire, and stove structs and then I have these functions that iterate over them based on which one of these structs they have and...
Bam, you now have an ECS, just less structured than if you had realized you could have planned for one from the get-go. Go look at Muratori's 2 hour talk on the history of OOP and you'll notice ECS is just a formalization of the design pretty much everyone since the 60s gravitated towards if they didn't go OOP.

>>106283155
your homemade "ECS" that you let develop organically as your code naturally matures will be 10x better and more ergonomic than using entt or some shit

>>106283180
No shit, that's exactly what I'm doing, but it's still an ECS. Just not one with 150k LoC of useless shit I don't need. I can't even add or remove components at runtime because I won't need it.

I'm prejudiced against OOP because no new language supports inheritance. Not Rust, not Go, not Zig.

>>106283259
you're limiting yourself by calling it an ECS. the problem with saying it's either OOP or ECS is that people will look up "ECS" and then just copy what some generic library is doing and then wonder why it's a pain in the ass to write gameplay code.

>>106263584
>>106282931
If anyone is not familiar with the Tachibana poster lore, over the years he posts this anime child in lewd depictions and constantly says "sex with" in replies to it in agdg. When someone replies about the "cartoon age" or whatever, he says that is specifically why he likes it. He is not welcome in this general, do not welcome him.

>>106283283
People thinking ECS must be something on the level of flecs or EnTT isn't really my concern, I'm still calling a spade a spade. ECS isn't really the only way to go about it either, but 90% of solutions end up close enough. I'm also making an engine, but I don't worry about people reading "engine" and thinking I must be making UE6 or Unity 2.

If anything it's important to decouple these terms from the pre-made, massively overcomplicated solutions people associate them with or it's just going to get worse. Seriously, how did ECS come to mean something complex? More complex than just an array of structs certainly, but that doesn't mean tens of thousands of lines of bullshit.

>>106283180
>>106283259
>>106283438
What's wrong with Entt / Flecs? I've wanted to try both but they seem polarizing

>>106283020
>For a game that size, it's either ECS or OOP.
what an ignorant statement

>>106283155
>You don't think "ECS" is one extremely specific pattern with specific features right?
thats exactly what it is
you wrote a lot of shit about something you dont understand

>>106283585
Massive, complicated systems that 99% of games don't need. You can make an ECS in 2k LoC or so, flecs I think is pushing 180k.
>>106283802
>hats exactly what it is
No, it isn't. Do your entities have generic components that encapsulate a specific functionality? Does your logic pick what entities to operate on based on the configuration of components of the entities? You have an ECS. You don't even need it to be stored in an array of structs honestly. 75% of the definition of an ECS is literally in the name.
>b-but my favourite youtuber made a 500 part series on how to make an ECS with archetype migration and and and
I don't give a shit, shove your youtuber e-celeb up your ass. People have been making games with (proto-)ECS since before you were born, they just hadn't all agreed on what to call it.

>>106283802
>something you dont understand
Also the fact that you think there is something TO understand about ECS shows you're a dunning-kruger retard. Stop making thinks complicated to appear smart you mong, ECS can be explained to a 5 year old child with 2 sentences.

>>106283967
>Do your entities have generic components that encapsulate a specific functionality? Does your logic pick what entities to operate on based on the configuration of components of the entities? You have an ECS
thats the extremely specific definition

>>106283967
>People have been making games with (proto-)ECS since before you were born
also no its not a common pattern even now, definitely not in the past because programming languages didnt have the generic programming features to support it

>>106189131 (OP)
i'm getting absolutely filtered trying to make pixel art bros.. should i just use ai to generate all the sprites? i can't draw for shit.

>>106283996
In broad terms yeah, when you look at everything, that's very specific since it excludes OOP and a bunch of other things. However it also doesn't imply anything beyond t hat. A system like that isn't inherently complex at all.

Every complex game gravitates towards this design pattern because you want to group things that have similar behavior (same reason OOP was created). Oh sorry you used "tags" instead of components, big fucking whoop you checked if a bit was flipped instead of if a component was present (which was likely found out by... checking if a bit was flipped).
>>106284007
Sketchpad was about as close as you get to an ECS back then, and it was written in ASM in the early 1960s.

>>106284045
ah yes all the ASM games written in the 1960s
wtf are you talking about

>>106284045
>it excludes OOP
It doesn't exclude OOP at all
>Every complex game gravitates towards this design pattern
It doesn't, because it turns into a spaghetti mess, hierarchies are a better way of handling most things

>It's another episode of ECSniggers proving to be even more retarded than pOOPjeets
ECS is retarded cultist garbage pushed by crab fags to cope with their chastity cage

>>106282996
only 10 more years and $50mil more

>>106283011
based and cute pilled

>>106251575
Thanks anon. Ya, im not sure how I feel on raylib yet. The lack of proper documentation is a pain, I've actually sometimes been looking at the docs Odin has for its binding, even though I'm using C++ lol.

Facking hell, exporting to IOs is a pain in the ass. at least I know the touch inputs and screen size settings work nice.

>>106282996
>Just one more rewrite bro

>>106272610
anything other than d3d11 is a waste of time. d3d12, vulkan introduce a lot of complexity you don't need, and opengl is buggy across diferent gpu vendors.

>>106286543
this. d3d11 runs on laptops from 2009 and on modern linux machines with proton. it just werks

>>106189131 (OP)
Thinking of trying out beeflang by reading example code. My first impressions on the demo game that the IDE ships with:
- Why is all the data and the functionality the game works with sprinkled across 13 files? I'd have structured the application such that one section loads everything (or quits otherwise), one section polls input, one section handles the game logic, and the last section gracefully cleans everything up.
- The amount of ceremony seen to express functionality is very high. For example, why bother with spreading out data and functionality related with the player over 3 files (Entity.bf, Hero.bf and GameApp.bf)? Is sticking to the inheritance model really nessecary?
- Magic numbers everywhere.
Overall I give the code an arbitary rating of cotton candy/tungsten cube
https://github.com/beefytech/Beef_website/tree/master/Samples%2FSpaceGame

>>106288508
your impressions show your inexperience, you should have some more humility
that code is much simpler than the code you'd see for a real game

>>106288508
>Why is all the data and the functionality the game works with sprinkled across 13 files?
>Is sticking to the inheritance model really nessecary?
Yes. That is how you are supposed to write "clean code".
Modern OOP design principles highly encourage that you to do this.

>>106288591
>that code is much simpler than the code you'd see for a real game
I really wish you stupid fuckers stopped accepting mediocrity because some retarded, overworked AAA faggots did it so that has to be how games are made.

Yeah most software is written like shit, so what? Not everyone is a starving jeet from the slums of Lucknow who only cares to shit out something on Steam hoping to escape his current life. You can hold your game to higher standards.

>>106288633
Who said it was shit? You clearly have little programming experience so you should abstain from making these assumptions

>>106288656
I don't care about Beef but if any of what he said is true, it's shit.
>magic numbers
>splitting everything into dozens of files
>OOP(s)

>>106288667
refer to the previous post about little programming experience

>>106288685
>I am a buckbroken human-shaped slop factory and anyone who disagrees with what I do just doesn't know as much as I do!

>>106288633
Chill, saar.
Poor fag jeets are just adhering to the white man's "clean code" philosophy and everyone seems to be happy.
>>106288667
that is exactly how you are supposed to code OOP in modern C++ though.
Its all OOP design principles, not completely Beef lang's fault for incorporating them.

>>106288691
Complaining about magic numbers is a classic noob complaint from people just repeating what their heard at school
And if you think a project with 12 files is confusing you just have no experience, that's tiny, that project is tiny

>>106288694
>not completely Beef lang's fault for incorporating them.
No, I know that, but anon was complaining about the general structure of the code, and he was right.
>>106288699
>Complaining about magic numbers is a classic noob complaint from people just repeating what their heard at school
Yeah I love returning to my code 6 months later and finding a random 5344.36122565 with no hints to what it is.
>just comment it bro
Why not just not use a magic number instead?
>actually it's just a 3.14 and very obvious
Then it's not a magic number by definition. Magic numbers are obscure, if they are obvious (e.g. the -1 in array.count-1) it's not a magic number.
>And if you think a project with 12 files is confusing you just have no experience, that's tiny, that project is tiny
That's the problem isn't it? A tiny project is already 12 files. How many files will a big one be?

>>106288749
the magic numbers in there are self explanatory
saying all numbers should be named consts is like saying all code should be commented
a big project could have thousands of files, really just comes down to your personal preference and what's easiest to organize
I don't do one file per class but you aren't wrong if you do that

>>106285562
The cheat sheet is enough for me most of the time, sometimes I go read the code directly.
My biggest issue with Raylib at the moment is text rendering, I need it more versatile and more optimised so I had to re-implement library functions.

>>106288749
>the general structure of the code
there was a time when I followed a C++ vulkan tutorial on youtube and the dude created a new file along with a new class for every vulkan object that he came across.
While the vulkan-tutorial was only one big arse file, his tutorial is a giant mess of OOP classes that made everything unnecessarily complicated and hard to follow.
That was all it took for me to realize that OOP isn't for me but for OOP fags, beef seems to be an amazing lang.

anyone have a good resource on using some kind of spatial partition for ray marching? Everything I find online is a bit vague making a simple sdf modeller but so far its shits itself at 50 shapes.

>>106288508
its a demo its supposed to be easy to understand. Just read the ide code if you want production code.

>>106289361
i have some raytracing code but it's with hardware apis and those handle building the acceleration structure for you. you just feed it geometry.

Is it cool for me to post here if I'm using AI for my project?

>>106290847
it's not really that different from following tutorials or copying code from stackoverflow
keep it up

>>106290847
It would be pretty dumb nowadays to not use AI specially for coding. Your time and the final result will be always more important.

>>106290857
>>106290858
Nice
Thoughts on hud and overall aesthetic?
it's wip but i like it so far, for a text based adventure.

>>106290847
I like it. Easy to read and looks like simple fun, gj.

>>106291028
yeah, anon saar did great job with typography

have any of you got any experience with the steam api?
if i understand the docs correct then ISteamUserStats::IndicateAchievementProgress does not do anything but show a toast with a progress bar specified by the nCurProgress and nMaxProgress parameters. however it only shows 0 out of n, completely ignoring the parameters. any ideas what im doing wrong? i cant seem to find any examples of IndicateAchievementProgress being used

>have to reinstall OS cause update is broken
>now have to re-install all the shit I need to compile my code
>The latest version of my libraries changes shit, so now I have to fix my code
I hate computers so much bros

>>106292484
bro your backups?

>>106292484
Time to move building tools to docker or something.