v3.1.0

Add fast kNN search (index.neighbors(x, y[, maxResults, maxDistance, filterFn])).

v3.0.0

• Improved internal storage format, making it ~10-15% smaller.
• Made it much easier to reconstruct an index from a raw array buffer: Flatbush.from(index.data).
• More resilient construction: Flatbush will catch more errors when given bad arguments or data.

Thanks to @tyrasd for the help!

v2.0.4

Fix infinite loop on new Flatbush(0) (throws an error now).

v2.0.3

Fixed search not throwing an error if called before finish.

v2.0.2

• Slightly improved search performance.

v2.0.1

• Improved search performance by ~30%.
• Slightly better packing for cases where item bbox sizes vary a lot.

v2.0.0

• ⚠️ Breaking: flatbush(...) should now be new Flatbush(...).
• Add optional data constructor argument for reconstructing an index after transferring.
• Use ES syntax in the exported module (doesn't affect node/browser builds).
• Expose numItems, nodeSize, ArrayType, minX, minY, maxX, maxY properties.
• Avoid worst case performance when data is already hilbert-sorted.

v1.3.1

Fixed a sorting bug, in turn significantly improving search performance.

v1.3.0

• Export flatbush as a proper ES module
• Publish browser builds

v1.2.0

Add an ArrayType argument to flatbush constructor for customizing underlying data storage type.