GeoParquet spec version 0.1.0

[jorisvandenbossche]

Just a heads up that the main parquet specification has moved from https://github.com/geopandas/geo-arrow-spec/ to https://github.com/opengeospatial/geoparquet, and that we officially released it as v0.1.0 (https://github.com/opengeospatial/geoparquet/releases/tag/v0.1.0).

In practice, however, this package should already be compliant to this spec (if it followed the metadata spec as in https://github.com/geopandas/geo-arrow-spec/). We only added an additional "edges" field, but since this is optional, not having it is not violating the spec (and the default is probably correct in most cases).
The new version in https://github.com/opengeospatial/geoparquet also doesn't include the top-level "creator" field with library name and version. But since additional fields are allowed, this is strictly speaking also not violating the spec.

So you might not need to change much in practice (but could certainly update the readme to include a link to https://github.com/opengeospatial/geoparquet).

We are still keeping the https://github.com/geopandas/geo-arrow-spec/ repo for Arrow-specific standards (extension types, other ways to store the geometries than WKB, ..), while the main GeoParquet metadata spec is now covered in https://github.com/opengeospatial/geoparquet,

[wcjochem]

Thank you for the update. I've been following some of the discussion on the updated spec (though I haven't been participating much myself) and the new developments are exciting. I'll update the readme to point to the new discussion.

[jorisvandenbossche]

As was mentioned in geopandas/geopandas#2376, there was actually one change in the GeoParquet spec compared to the original one: schema_version -> version

It would be nice to see sfarrow updated! I think this line:

sfarrow/R/st_arrow.R

Line 34 in aff5dee

schema_version = "0.1.0",

can be changed to version = "0.4.0"

In addition, in version 0.4.0, we switched from WKT to PROJJSON for the crs field.

[wcjochem]

Hi @jorisvandenbossche thanks for the reminder. I got distracted with other projects. I should be able to get this updated this week.

[gmoore016]

It would be great to see this merged! I'm trying to use sfarrow for my project, but my downstream pipeline says the GeoParquet file is malformed since it doesn't include the "version" metadata. From what I can tell the commit above solves this problem, but hasn't yet been merged into the main branch.

Thanks for your hard work!

Edit to add: I attempted to use the dev branch instead, but that seems to have broken the write_parquet method, as it's now attempting to call na.rm as a function. Screenshot below.

[wcjochem]

Thanks. This package is overdue an update, in general. I'll look into this after the new year.