Gunship

"The game engine for people who don't like making games."

Gunship is an experimental game engine being developed as an effort to learn more about the inner-workings of a modern game engine. It is being written in Rust in an attempt to break away from the game industry's unhealthy codependence on C++.

Design Goals

At this point Gunship is meant to be more of a learning project than a production-ready game engine. The idea is to build as much as possible from scratch in order to develop myself as a programmer and learn all the parts of a modern game engine. So far I've allowed myself to use the standard library and libraries for binding to platform-specific native libraries, but the intent is to keep dependencies to a minimum, even if there are relevant crates in the Rust ecosystem.

As for the engine itself high-level design is pretty nonexistent. I'm keeping development directed by building a game along with the engine, so for the time being the design motivations for Gunship are "whatever work best for making a game". At present the design is roughly as follows:

• The renderer is a separate crate from the engine core. It's potentially going to be usable on its own, though right now that's not a primary goal. Mostly the renderer's goal is to provide a high level, backend agnostic system for rendering that can be tested idependently but easily be plugged into the engine core.
• The engine core provides the scheduler, which makes it easy to write highly parallel, asynchronous gameplay and engine code.
• The engine core only provides the primitives for gameplay development (e.g. transforms, lights, meshes, etc.), but provides little-to-no framework for structuring gameplay code. Such frameworks will likely be provided as a layer on top of the engine core.

Current Features

The engine is very early in its development and so doesn't have much to show off as of yet. That said, it does have a few working features at this point:

• Multi-threaded, async engine and gameplay code makes it easy to write games while taking full advantage of multiple cores in modern CPUs.
• Basic (very basic) 3D mesh rendering support, just basic meshes and limited support for custom shading. No shadows, highly inefficient.

Broken Features

The following features have been implemented in the past but were broken at some point:

• 3D collision processing with sphere and box colliders. Actually pretty fast compared to the other collision systems, though very much lacking in functionality.
• Code hotloading, which allows you to recompile code without stopping and starting the game. Currently very hacky in its implementation, but very much functional.
• Super basic audio -- only one audio source at a time and pretty much hard-coded to work on my machine.

Planned Features

Beyond the obvious (making the existing systems more robust and actually usable) there are a few features that I plan to work on in the near future:

• Proper rendering system, including more efficient rendering through batching, skinned meshes, and particles.
• Cross-platform support for Linux and OSX at least, eventually I'd like to get to mobile support and consoles, though who knows how long that could take.
• Proper logging system.

Contributing

At this point the engine's not stable enough for anyone but me to attempt to use it to make a game -- if you want to use Rust to make a game then check out Piston. On the other hand if what you want is to build some engine subsystem from scratch but don't want to have to put together all of the other pieces, feel free to use Gunship as your baseline. Even then I don't recommend it, but I'm happy to help however I can (just send an email or open an issue to ask questions).