> "Regina Obe" <l...@pcorp.us> writes: > > > Paul and Sandro, > > > > Any thoughts on this. I know we went back and forth with "Why don't > > we just populate spatial_ref_sys using postgis_srs_all or some other > > such thing or why don't we just change spatial_ref_sys to a view that > > relegates to reading these from proj.db make spatial_ref_sys a view or > > why don't we make it a reduced set. > > Not sure where we landed on that discussion. > > > > Before the issue was because we supported such old proj versions, but now > our minimum for 3.5 is 6.1 so all the options we had are now doable with our > minimum proj version. > > I think part of this comes down to > > Are we just deferring to EPSG, or are we becoming an authority for > including and/or defining SRSes? > > I thought it was "postgis uses database in proj and is not an authority", but am > I off?
Well we are definitely not an authority and would rather not be. We tried to keep the spatial_ref_sys in sync with EPSG out of necessity, but since 2.5, all that is needed is a proj.db and we relegate all processing to proj.db metadata except where the record doesn't exist in proj.db. The only thing we rely on is the SRID, auth_srid and authority and when no authority, we assume EPSG. It is still convenient to have these in spatial_ref_sys but not a necessity anymore and in fact I think users might prefer a spatial_ref_sys that has just the spatial_ref_sys they need. The search would be slightly faster, not to mention a lighter table.
