When You Can't Use Unreal Engine To Make A Game 

The more mature I become as a programmer, the better code I can make chatgpt generate XD

@@Evgeny1 I've found myself settling into descriptive, almost functional patterns, over time. Tends to read more like telling the computer what I want it to do, rather than how to do it.

It's the bell curve of experience wojack meme

Looking at my past work, yeah that tracks! LoC isn't a good measure turns out

Same happens to writers

Not really, unity uses more of an object oriented component system, with low separation of data and logic and higher complexity if you try to use composition. Unity DOTS however has a proper ECS system, but most unity developers don't know how to use it

@@ITR Ok, I stand corrected, I haven't used much of Unity, but that makes sense

It's a good advice for 99% of the people trying to make a game. Finishing a game is extremely hard, adding a game engine from scratch multiplies the time needed to finish it by a factor of 3 at least.

If you want to finish all your games or are trying to get a job in the industry then don't code an engine. If you like to tinker and have total control over your game then make an engine

@@ramoncf7 Yeah I wrote a 3d renderer and model loader in C++/OpenGL for an assignment for college (it had to have geo with normals, diffuse and specular textures, the world needed multiple light sources, a 3d user controlled camera) and just making that was a nightmare (we had 6 weeks and I had other classes and a job which didn't help!). Imo 99% of people who want to make a game would give up way before actually making anything besides a 3d camera in a blank void with some boxes and triangles cuz like... if that weren't a graded assignment I would of stopped after making a triangle spin around or something.

​@@orangy57Wouldn't having made a decent engine be good on a résumé? Ok, guess it depends on the capabilities, considering how simpler games can run on horrid spaghetti that makes god weep. (Note that I'm just a random person on the internet with very little game dev and zero industry experience.)

​@@Kenionatus (game developer here) Yeah, it would look nice on a resume. But depending on what you mean by "decent engine" it may also take at least several years of your life and never be finished. A simple, specialised engine is worth trying, and in fact in my university course we did just that. Making a well-performing engine for something like the N64 requires a lot more low level technical knowledge though, as compared to programming for PC.

Pointers are IDs the same way pointers are also optional types. The key difference is that if you hold a pointer, there is nothing stopping you from dereferencing it, whereas if you only hold an entity ID, you have to pass it to a system that enforces checking before you can anything with the referenced object if it even still exists.

@@MopeyMoon I never had problems with invalid pointers. I guess I used too much Java, C#, and smart pointers in C++. Often I end up with double links because I need the cascade effect right there. With IDs and an in memory database we can use a b-tree to manage memory and defragmentate it. In Java H2 is such a Database. For C there is SQLite? So IDs would live forever? Sounds like a memory leak to me.

​@@ArneChristianRosenfeldtIDs are just integers that don't have any lifetime associated with them, so I don't see how that could cause a memory leak. For example, say that entity 1 has an "attacking_id" field with an ID integer of 2, then it doesn't really matter whether entity 2 has already been killed or not in terms of memory leakage

@@sanderbos4243 but then you could just as well use pointers to attack types. Is this about serialisation? I mean, in code you try to avoid using “type1”, but you write “low poking attack” . The compiler then can either use enums to compile this to IDs , or static fields to compile to pointers.

@@ArneChristianRosenfeldt When I used "attacking_id" as an example, I meant "what entity ID am I attacking?", not "what type of attack am I performing?". It doesn't really matter whether it's an int or a pointer or a double, but the point is just that you don't want to have it be a pointer to a raw memory location that you dereference. The blog post "Handles are the better pointers" goes in much greater detail of the many advantages of it, so I highly recommend checking that :)

That is true

I would say that depends on what your game needs. I found it typically works fine for one actor to directly apply damage to another.

I'm not entirely sure what you are asking but I if there are two separate configurations for a block then you want two separate pieces of information on the block. I would also not use the entity/gameobject system for each individual block. I would have a single game object manage groups of blocks.

@@james.lambert I wrote the comment while in a hurry. What I meant was that you can glitch games into doing something else by doing certain actions. For example in super Castlevania, you can skip the whole game by performing a boosted jump and enter the next screen, if you do that the game checks the state of the character through a number. For example if the player is idle the number is 1, if he attacks 2 etc. The game is designed for the character to enter in the walking state but if you enter in the attacking state the game gets confused since the number happens to glitch the game. If it used classes (which was impossible) then this wouldn't have happened. It is very complicated to explain, I hope that I cleared things out. If not , just search super Castlevania Speedrun and I think you will understand that I'm saying.If you are reading this love your vids!

More like a reverse ASMR. The constant noise was very distracting thoughout the video.

My full time job is web development but I have been making games as a hobby for a very long time

This is the way. Just having a main job and making games on the side, web development specifically sounds like a drag, like working a job where you just make excell sheets all day.

Rust is amazing but it takes a long time to get good at managing all those types. Maybe start with C since its surface area is a lot smaller, then when you got the hang of pointers and stuff you can start with Rust. But of course it depends on what your level is now and how much time you got to spend. I would go straight to Rust again, coming from node/npm i feel right at home with cargo and crates.

If you're coming from high level languages, I would probably recommend jumping straight into Rust - it has one of the best type systems you will ever use. Rustlings is a great starter resource, though there are of course plenty of other ways to learn

I haven't used Rust, but I've heard from other game developers that it's slow to iterate and experiment in Rust because of things like the borrow checker that try to make you program things "correctly", when what you want in game development is the ability to program things stupidly at first so you can rapidly try things out. Someone wrote an in-depth article on this called "Leaving Rust gamedev after 3 years".

I would offer two alternative recommendations: Odin and Zig. Both are more powerful than C but not as complex as rust.

What do you want to Skript? Events in the level?

maybe possible but it would really be a better idea to create a custom scripting system for the specific purpose embedding lua into an n64 game is overkill imo

@@randomcatdude the Lua runtime is rather compact, plus it has a nice C interface. It would be much easier to make the mentioned entity system in a dynamic type lang, than in C

@@Raspredval1337 and we be slow as hell. Static types give you speed, especially in an Array.

@@ArneChristianRosenfeldt maybe, since the N64 bottleneck is memory access and dynamic type langs add significant memory overhead. Still, it might be ok for some tasks. Plus, it can be as performant as C code with a JIT compiler.

In the specific examples he showed here that's absolutely the way to go, assuming you won't need more than 64 bits. In terms of ECS, he explained tagging stuff, but you usually also want to be able to store information in the tag, which is called a component. So think of having a "health" component that you can slap onto anything and just contains a double.

Send me a message on discord and we can talk. I don't have a huge need for writing right now as I am still working on gameplay mostly but I wouldn't mind the help later on. lambertjamesd#3921

@@james.lambert Awesome! Thanks for responding! I couldn't find a way to message so I sent a friend request. I hope that's okay, I've never really used Discord before 😅

Here is what I have so far github.com/lambertjamesd/spellcraft I don't know how much of this will apply to your environment

@@james.lambert #include int main() { const struct { void (*log)(const char *); } console = { .log = printf }; console.log("Hello world from an edge of JavaScript"); } we have to get this lib to provide the ability to understand JavaScript. I want to use your lib in the future.

@@james.lambert can you make this lib an installer, please?

Why are your structs all the same size? Any strings?

Sounds a lot like an archetype ECS :P

Why do you think so? Genuinely curious.

Well, "keeping it simple" is hard. And "writing reusable code" is hard, so that you have then a simpler time ;)

I absolutely disagree. Writing reusable code is a habit, and you can develop an instinct for how to write pretty much anything in a reusable way. It's not really a case of trying to decide what you'll want to reuse and maybe getting it right or wrong.

game objects follow an oop behaviour tho. ecs is basically just game objects but without the class component. though they're inherently a completely terrible idea both implementations for a game on the n64 since memory bandwidth is awful at best.

@@snesmochaRamBus goes vroom. C structs and arrays are localised. Align to 16 byte cache boundaries. The bad thing about MIPS is that it blocks on a cache miss, while AtariJaguar only blocks when we access the register which we load to.

​​​@@snesmochaECS is not inherently OOP, nor is GameObjects. At least based on most metrics of OOP you can go by, clean code, solid, filled with inheritance etc. However, it is correct that ECS can have a performance impact as nothing beats a linear array of cache friendly memory. That's why Archetype based ECS stores things in section of linear memory (which his is currently sadly not which is okay because of what he calls rule 2). ECS is especially good for a middle strategy approach. That is, in the beginning linear array is easiest to program, then when it gets hard to maintain, go ECS. At some point performance concerns hits, and you pull some of it back. Thus the reason I'm calling it a middle strategy