neighbors support? about static_aabb2d_index

[jbuckmccready]

It is definitely possible, just not yet implemented, I think it could be easily ported using the std::collections::binary_heap as the priority queue. I'll look at it sometime today.

[RReverser]

Ah awesome, thanks!

[RReverser]

Actually... I've realised that what I probably need instead is KD-tree as I'm working with point data, so I can keep the issue open but this is not urgent or anything.

[jbuckmccready]

Ah yeah, you can use this spatial index for point data as well (by having min_x == max_x and min_y == max_y for all items added) but it is a little bit less efficient in memory and comparisons (but may still be plenty fast for your use case). The algorithm used can definitely be applied to points efficiently, there is actually an issue in the flatbush library about it: mourner/flatbush#33.

In Rust it would likely be a different type for performance and Rust's static type system, I don't have a use for that now but maybe I'll add that to this library sometime in the future as well.

I committed changes to support nearest neighbor queries (commit: f2ea39f), the unit tests pass but I haven't done any benchmarking, optimizations, or manual graphical confirmation tests yet.

I also probably want to add some examples, but I did update the docs and you can look at the tests for example use.

[RReverser]

The algorithm used can definitely be applied to points efficiently, there is actually an issue in the flatbush library about it: mourner/flatbush#33.

Ah yes, I've posted that comment above after chatting with mourner - he said the changes to represent points are a bit tricky, so for now I've looked at KD-tree implementations and stopped on https://docs.rs/kd-tree which seems promising and works well.

I committed changes to support nearest neighbor queries (commit: f2ea39f), the unit tests pass but I haven't done any benchmarking, optimizations, or manual graphical confirmation tests yet.

Awesome, thanks. I'll close this issue for now then.