Dewey Dunnington

That's a good "point" (as it were)...I probably won't spend much time prototyping this to focus on other things, and we can circle back to allow that encoding if or...

I made #26 to officially propose this change...feel free to chime in there with any additional ideas in either direction!

> Orthogonal to that issue, there's the usual and recurring issue with coordinate order used in geometry encoding vs axis order of the CRS definition. EPSG:4326 says first axis is...

I like the idea of deferring to the GeoPackage wording. In the unlikely event that users are begging for the ability to specify an authority compliant CRS, that could be...

I've come around to mandating WKT2 as the `crs` field, because it makes it possible to write compute kernels like "ellipsoidal length" (as long as the reader can read WKT2)....

In support of a `crs_type` field, I wonder if this could be useful to communicate CRS information that was provided to be properly formatted as WKT2 elsewhere. The example that...

Since the last discussion here, GeoParquet has moved to PROJJSON to represent CRS and can use a `null` value for the CRS field to indicate that the writer has no...

> Non-CRS-aware implementations can check for a specific CRS quickly by only needing to parse JSON, while CRS-aware implementations have the entire geodetic definition. My hesitation on *requiring* PROJJSON in...

@achapkowski in the proposed encoding in #22 could you set the `crs` field to `ESRI:2948330` (or whatever the ID is)? Perhaps we could formalize a `crs_encoding` value for AUTHORITY:CODE since...

Yes, it would be metadata on the geometry column (and I will add some concrete examples regarding what the CRS would look like in the case of projjson, wkt2, and...