[postgis-tickets] [PostGIS] #2534: st_distance is returning incorrect results for large geographies

[PostGIS]

#2534: st_distance is returning incorrect results for large geographies
------------------------------------+---------------------------------------
 Reporter:  samuelb                 |       Owner:  pramsey      
     Type:  defect                  |      Status:  new          
 Priority:  medium                  |   Milestone:  PostGIS 2.1.1
Component:  postgis                 |     Version:  2.0.x        
 Keywords:  st_distance, geography  |  
------------------------------------+---------------------------------------

Comment(by pramsey):

 It seems to be consistently not finding that the point is in the polygon,
 so the distance is the distance to the ring.

 http://www.gcmap.com/mapui?P=+20S+20W-
 60N+20W%0D%0A60N+20W-+60N+90E%0D%0A+60N+90E-
 20S+90E%0D%0A20S+90E-+20S+20W%0D%0A0N+0E&MS=wls&DU=mi

 Interesting, the distance POINT(-19 0) and POINT(-21 0) are the same,
 which means the ring calculation is consistent, but the point-in-polygon
 calculation is not. Reported on 1.5 and 2.0, still exists in 2.2, so not a
 regression.

 I wouldn't expect your desired use case (creating bounding "boxes" on a
 sphere) to work particularly well regardless, since you're expecting
 straight lines where they don't exist (see the great circle image above
 and look at the top and bottom lines)

-- 
Ticket URL: <http://trac.osgeo.org/postgis/ticket/2534#comment:1>
PostGIS <http://trac.osgeo.org/postgis/>
The PostGIS Trac is used for bug, enhancement & task tracking, a user and developer wiki, and a view into the subversion code repository of PostGIS project.