paradedb / pg_analytics

DuckDB-powered analytics for Postgres
https://paradedb.com
GNU Affero General Public License v3.0
184 stars 11 forks source link

Geoparquet support #153

Open haubourg opened 4 days ago

haubourg commented 4 days ago

What feature are you requesting?

Hi, I'm testing pg_analytics for mostly geospatial use cases. Does pg_analytics supports geoparquet format as Crunchy bridge does ? I saw only GeoJSON examples in the docs and failed to do geographic subset filters when trying to mimic this example : https://www.crunchydata.com/blog/vehicle-routing-with-postgis-and-overture-data

Why are you requesting this feature?

For a massive datawarehouse postgres with mostly geospatial data, we go toward analytics and explore all the available options to outsource storage of data to S3 geoparquet, and benefit from column storage features for analytics.
After having screened most available extensions, pg_analytics seems really advanced and easy to set up.
GeoParquet is now an Open Geospatial Consortium standard (OGC) and along with GeoPackage should be the first class citizens when dealing with spatial data.

What is your proposed implementation for this feature?

GDAL and duckDB both support GeoParquet reading and could provide the necessary bits.

Full Name:

Régis Haubourg

Affiliation:

CSTB

philippemnoel commented 4 days ago

Hey @haubourg! Welcome :)

I don't believe we have Geoparquet support, but it should be easy to add by following our Parquet and GeoJSON supports if duckdb-rs supports it. Would you like to try your hand at it? We probably won't have time on our team to do this for quite some time.

haubourg commented 4 days ago

Thanks for your feedback! I unfortunately don't have the coding skills and my benevolent time goes to QGIS.org 😅. Regards