Keyde aims to provide minimal yet fast implementations of spacial query structures.
Currently, keyde provides a: - Kd-tree
Keyde provides a Point trait that is implemented for arrays of sizes 1 to 4,
tuples of 2 and 3 dimension as well as for all the basic 1D types (u8, i8, isize..)
By enabling optional features such as glam, you can get an implementation for that crate's
default Vec3, Vec4, Vec2 and Vec3A types.
Keyde wants to suppoort more linear algebra crates, so feel free to make a PR and add your favorite one.
See src/point_implementations.rs for inspiration.
Key things that differ keyde's kd-tree implementation from others:
- No recursion, only iterative implementations
- No cloning of your data, everything is refered to by indices into your data
- Provides KdTreeStrategy to choose sorting strategy which might help you find a
creation/querying-strategy that is more optimal for your particular data layout