FW: [postgis-users] Re: ERROR:XX Operation on two GEOMETRIES with different SRIDs (WAS:-->point_inside_circle(geometry,float,float,float) : units -> meters ?)

[David Blasby]

Martin Davis wrote:

>Dave, is this a GEOS problem?  I remember you were asking about why JTS required the same SRID for args to geometry methods.  It was a good point - upon reflection I decided this was too much handholding, so I dropped the requirement in 1.3.  It would be trivial to drop it from GEOS as well. In fact, we *should* drop it, as part of brining GEOS up to 1.3 compatibility.  (The requirement for identical precision models is also eliminated in 1.3, since it stores coordinates in different models compatibly. This probably isn't an issue for PostGIS, however).
>
This isnt GEOS related - I force SRID equivelence in PostGIS.  

The problem is an equivocation of "BOX".  There is an actual independent 
BOX3D type, a GEOMETRY has a built-in-but-hidden BOX3D, and there is a 
BOXONLY GEOMETRY-type used for indexing.

The BOX3D type doesnt have a SRID, but the BOXONLY GEOMETRY-type does. 
 This came from the time when GEOMETRY didnt have a SRID.  

Personally, it makes sense to get rid of the BOX3D type, but people love 
it too much.  Its too convient to say 'BOX3D(0 0, 10 10)' instead of 
 PolygonFromText('POLYGON((0 0,10 0, 10 10, 0 10, 0 0))').

dave