Writing the Renderer
Before we are able to write some games, we need the ability to draw something on the screen.
The renderer shouldn't do anything fancy. At the very least it should be able to display 3D models(including animations), allow some post-processing effects via full-screen shaders, using a deferred rendering path, do some basic UI stuff, and work for Desktop and WebGL.
Advanced things like ray tracing, clustered/tiles rendering paths or all the ML-assisted render-features in modern games are beyond the scope of this little renderer and probably any game I am going to write in the next decade.
Everything else will just be added as we require for the games at hand.
I don't have a strict plan yet, but the goal is to split the Renderer at least into these packages:
Hardware abstraction layer that deals with the unsafe graphics API's like OpenGL, Vulkan, etc. It's basically "dumb" and just exposes safe function to create and manipulate the objects that are living on the graphics device and the actual draw calls. Things like vertex buffers, shaders, and textures.
Loads and creates different kinds of resources. Loading images(that are turned into textures), models of different formats, reloading on change, etc.
Data structures shared by the other packages. Things like Color, Materials, Camera, and common math structures.
The actual renderer, using the other crates to offer high level functionality. It deals with the render-paths, draw call sorting, etc. It's the only thing directly exposed to the game, and exports the API of the others where necessary.