lat/lon vs lon/lat ordering of WKT
Different vendors interpret the OpenGIS standard differently on this point, and I'm curious about the MySQL GIS developers take on it.
The standard on the one hand says: "A Spatial Reference System, also referred to as a coordinate system, is a geographic (latitude-longitude), a projected (X, Y), or a geocentric (X, Y, Z) coordinate system." Implying that when WKT is used to store geographic data that a lat/lon approach is correct.
On the other hand, in an appendix on specifying Map Projections, it lists the following parameters to a map projection:
"central_meridian - the line of longitude chosen as the origin of x-coordinates"
"latitude_of_origin - the latitude chosen as the origin of y-coordinates"
Implying the opposite.
MS SQL does it lat/lon, Google Maps does it lat/lon.
Postgres and Oracle and lots of third party libraries do it lon/lat.
I'd like to say forget it, but it has huge implications for the current implementation of the GIS functions in MySQL.
Even though they are touted as 'precise', without a valid sense of what spatial representation is in play, they are only precise for planar data, and who is using the GIS for planar data? Practically no one, its for maps above all else.
For this, proper elliptical implementations of the GIS functions are needed, and elliptical functions require that a choice be made about the ordering, since latitudes and longitudes behave so differently when taken off the plane.
In this respect, MySQL is now years behind other database vendors, who not only have made these decisions, but have working elliptical models in play.
It concerns me that I don't see these issues being discussed, and that even the very minimal planar implementation has been languishing for nearly a year unchanged.