Details
-
Improvement
-
Status: Resolved
-
Major
-
Resolution: Workaround
-
None
Description
I was looking at adding null safety to ST_Point but stumbled upon some inconsistencies with Postgis and other Sedona constructors.
ST_Point in Postgis takes an optional srid - as most constructors in Postgis and Sedona.
ST_Point in Sedona takes an optional Z value - making the 3 argument version in Sedona very different from Postgis.
Postgis has ST_Point, ST_PointZ, ST_PointM and ST_PointZM for points with different dimensions. All take an optional srid.
Postgis also has ST_MakePoint which is similar to Sedona ST_Point.
See: https://postgis.net/docs/ST_Point.html
I see changing the semantics of ST_Point from x, y, z to z, y, srid as a no option.
One possible upgrade path, if we want to align with Postgis, would be to:
- Introduce ST_PointZ and possibly ST_MakePoint
- Remove the optional third argument from ST_Point, forcing people to use the above for 3D points.
- Once enough releases have passed and we're sure people have upgraded we can add an optional srid argument to ST_Point.
Should we keep ST_Point as it is or align with Postgis using the plan above?
Attachments
Issue Links
- links to
