For example, can I implement a Vector(n) to represent a mathematical n-dim vector?

var Vector(dim: u32) = struct {
    .data: [dim]f64
 }

fn wedgeProduct(vec1: Vector(m), vec2: Vector(k)) Vector(m*k) {
    //... Implementation that references m and k
}

In this Vector type, .dim is runtime-known and constant (i.e. Vectors don't change size)

I know I can return struct definitions from functions, but that doesn't allow me to write a function with a signature like wedgeProduct, or I haven't been able to figure out how.

Should I give up on the array notation, and have my structs hold a dim field and a pointer to an appropriately-sized array? How would I enforce that the array size cannot change?