I love custom game engines. Looks fantastic! Will you be sharing the source?
More than happy to talk about any specific part however (e.g. how scenes are handled, the code itself, or how particular features are implemented or optimized).
You said you didn't explicitly use simd, but did you do anything to help the optimizer autovectorize like float chunking
I only used traits to more easily implement the scenes; a Scene needs to implement a new(), a start() and an update(), so that I can put them in an array and call them like scenes[current_scene_idx].update() from the main loop.
Also, I used some short and simple closures to avoid repeating the same code in many places (like a scope-local write() closure for the menus that wraps drawtext() with some default parameters).
The vast majority of the time is spent in the triangle filling code, where probably some autovectorization is going on when mixing colors. I tried some SIMD there on x86 and didn't see visible improvements.
Apart from obvious and low-hanging fruit (keeping structs simple, keeping the cache happy, don't pass data around needlessly) I didn't do anything interesting. And TBH profiling it shows a lot of cache misses, but I didn't bother further.
drawmeshindexed(m: &Mesh, mat: &Mat4x4, tex: &Image, uv_off: &TexCoord, li: &LightingInfo, cam: &Camera, buf: &mut Framebuffer)
so there is also no global state/objects. All state is passed down into the functions.
There were some cases that RefCells came in handy (like having an array of references of all models in the scene) and lifetimes were suggested by the compiler at some other similar functions, by I ended up not using that specific code. To be clear, I have nothing against those (on the contrary), it just happened that I didn't need them.
One small exception: I have a Vec of Boxes for the scenes, as SceneCommon is an interface and you can't just have an array of it, obviously.
Another soft rule: no member functions (except for the Scenes); structs are only data, all functions are free functions.
Also no operator overloading, so yes, lots of Vec3::add(&v1, &v2). I was hesitant at first but this makes for more transparent ops (* is dot or cross?) and does not hide the complexity.
The whole thing is around 6-7kloc and I think it would be possible to rewrite in C++ in a day or two.
Here's a counterpoint: every time you write a for loop in Rust, you are using iterators.
(Note that it was a severe design flaw to make ranges like 0..10 iterators directly rather than just IntoIterator, because this means ranges can't be Copy and as such it's inconvenient to pass them around.. but fortunately they are going to fix that in a new edition)
But actually..
Do you mean you prefer writing for i in 0..myvec.len() and then accessing myvec[i], rather than using for x in &myvec or for x in &mut myvec, and using the x directly? But why?
To be honest, grepping the source, I found a couple of places with for x in &myvec. Probably tried them while learning the language, they worked and left them there.
But really, I am just more used to the old way. It visually feels more familiar, I have an explicit i to do things, and it's easier to delete/manipulate elements while in the loop and not invalidate anything. It's not that I am against an iterator if it matters and it's better, of course.
In my case, the important loops are like "for i in x0..x1 { draw(i, y); }". Is there a way to turn this into something that, for example, skips the bound checks or optimizes it in any other way?
Things like
enum Something {
One(String),
Two(i32),
}
I think this plus pattern matching is the foundation of Rust's superpowers. It's also very old tech and absolutely not Rust's invention, present in languages like OCaml and SML. Hence the early Rust slogan, "technology from the past, come to save the future from itself"
And of course, Options and pattern matching are easily the best part of the language and very powerful. I am obsessed with things like "let x = if {...}".