A couple of friends got together and, waddaya know, a new programming language project was born. We have no idea where it's going, or how feasible it is, but we're all pretty excited. We wanted to see where we could go from Rust, and there's some pretty wild ideas being thrown about. It's really fun!
While I haven't contributed much in terms of language design, I do have strong feelings about compiler architecture, and I pushed hard for making sure the compiler is as efficient as possible. I don't mean that we should be optimizing memory layouts and writing vectorized code today, but rather that we leave the right gaps in the compiler for us to improve performance later.
There are two hard cores that we had to consider how to optimize: type-checking (including inference and trait solving) and code generation. We actually have some really interesting ideas for the former (that may even apply to the latter in the future), but testing them is going to be really hard. If only we had some massive source of code written in our language that we could use for testing and benchmarking...
Well, I'm just going to squint a bit and call our language Rust. If I can write a relatively simple (and massively incomplete!) frontend for Rust, I can play with it and test it out against the millions of lines of Rust code available online. This is the perfect way to try out a radically different type-checking implementation, and to find out just how fast it can be (relative to 'rustc').
And thus 'krabby' was born. Its goal is to jump from source code to type-checked IR as quickly as possible, and maybe it'll turn into a whole Rust compiler one day. We'll see.