Kyle Barron

An empty point has two values `NaN, NaN` from #81

Notes: - Probably makes more sense to wrap [`earcutr`](https://github.com/frewsxcv/earcutr/) directly, since then you won't go through geo objects at all. geo's impl is [here](https://github.com/georust/geo/blob/3f9e12f19493527afaf5330bfc3209364dad8d7a/geo/src/algorithm/triangulate_earcut.rs#L122-L130) - For the return type, maybe...

Also consider: - earcutr takes as input a flat array of vertex coordinates So... in preparation for earcut we can do: 1. First convert the coordinates to interleaved coords if...

In theory we could bind to earcutr here, but it won't be that useful for deck.gl-layers. The output of earcut is a different length of data than the number of...

earcut should be on a separate worker by default https://github.com/geoarrow/deck.gl-layers/blob/25fdeafe2d9c591570bab975ebea43ae82e97a8d/src/solid-polygon-layer.ts#L69-L75.

Spatial joins are already half-implemented in https://github.com/geoarrow/geoarrow-rs/pull/580. https://github.com/msalib/spatial-join is a useful reference but we can't use it directly because it uses geo objects. Additionally, the spatial joins here will use...

Initially intersects but should be all predicates eventually

Hello 👋 That sounds like you probably ideally want to use nearest neighbors instead of `within`. This is tracked by https://github.com/kylebarron/geo-index/issues/97. The RTree implementation already has the nearest neighbor implementation,...

A PR to add the iterator approach of nearest neighbors to the KDTree would be most welcome!

👍 I think this makes sense. Would you be able to put up a PR?