paleolimbot opened a new pull request, #109: URL: https://github.com/apache/parquet-testing/pull/109
This PR adds some geography test files with statistics that may be useful in testing implementations. Notably, geography statistics can have `xmin > xmax` such that row group statistics can "wrap around" the antimeridian (e.g., so that ship position statistics in the pacific ocean, or a catalogue of wildlife in Fiji do not have longitude bounds that span the globe). I recently implemented this in SedonaDB ( https://github.com/apache/sedona-db/pull/805 ) based on the pluggable statistics writer in arrow-rs ( https://github.com/apache/arrow-rs/pull/8414 ). The underlying stats are coming from s2geometry's S2LatLngRectBounder ( https://github.com/google/s2geometry/blob/master/src/s2/s2latlng_rect_bounder.h ) via s2geography ( https://github.com/paleolimbot/s2geography/blob/main/src/s2geography/coverings.h#L13-L19 ). I'd love to simplify that and just have it all in a self-contained implementation but certain components of bounding on the sphere (e.g., if a polygon contains the north pole) are non-trivial. The files are basically uniformly distributed (on the sphere) points, segements (basically sequential points sorted on a hilbert curve), and polygons (buffered points, basically rectangles). Both lines and polygon have some geographies that cross the antimeridian, and all the files have at least two row groups with wraparound statistics. All the files have at least one geometry intersecting the north pole and one intersecting the south pole (for polygons, the geometry contains it). These aren't exhaustive cases for geographical testing but the addition of the wraparound statistics will hopefully help ensure pruning is correct. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
