paleolimbot commented on PR #45459: URL: https://github.com/apache/arrow/pull/45459#issuecomment-2673110425
@pitrou @wgtmac Thank you for the detailed reviews! I am working through these now but wanted to share my high-level plan just to make sure I'm headed in the right direction. Basically, I'd like to remove as much as possible from this PR while still allowing a route to read/write files that with GEOMETRY and GEOGRAPHY types. - Create a separate branch for writing example files with arbitrary CRSes ( https://github.com/apache/arrow/compare/main...paleolimbot:arrow:parquet-geo-write-files-from-geoarrow ). It's important to generate those files but the scope of this PR should shrink to a more reasonable scope based on your collective comments. Some of those files could also be used in tests here (or in a future PR)...the testing for some other types seems to use this mechanism. - Remove GeoArrow CRS support when writing Parquet (and thus remove the RapidJSON dependency). In this case we'd still be able to provide a mechanism for writing basically the "default CRS" case. I'll remove the `write_geospatial_extensions` writer option (it seems like we are in agreement on this although I don't feel strongly either way) - Keep support for reading into GeoArrow. This doesn't require a JSON parser and would still let us read arbitrary files (when opted in via `allow_arrow_extensions`. Most of this would ideally live in an extension type but removing it would make it difficult to test the PR. It would be opt-in, so would be difficult to accidentally depend on (it could be also marked experimental and could be changed pending the results of a proper mailing list discussion). - Raise the issue of `geoarrow.wkb` as a canonical extension type. I've spoken with Joris and Kyle about this but need to get their consensus on how to do this. -- 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]
