[postgis-devel] [PostGIS] #1798: Performance issue with ST_Intersects(geometry, geometry)

[PostGIS]

#1798: Performance issue with ST_Intersects(geometry,geometry)
---------------------+------------------------------------------------------
 Reporter:  nicklas  |       Owner:  pramsey      
     Type:  defect   |      Status:  new          
 Priority:  medium   |   Milestone:  PostGIS 2.0.1
Component:  postgis  |     Version:  2.0.x        
 Keywords:           |  
---------------------+------------------------------------------------------

Comment(by nicklas):

 "only cache on second occurence" will probably reduce the impact, but then
 prepared geometries are not used.

 I agree that it is a corner case.

 But what is important is if it can happen also with indexes.

 I have not tried and not looked at the code, but in the polygon-polygon
 case, do we have the "only cache on second occurence" in place? If not,
 what decides which of the polygons to be prepared? Can we in that case be
 sure that the index and the "prepared geometry" code agrees about which of
 them is in the "inner" and "outer" iteration. If not we might get the same
 problem also with indexes.

 In the polygon-point case I guess that it s always the polygon that
 searches for points in the index, is that correct, and not vice verse.

-- 
Ticket URL: <http://trac.osgeo.org/postgis/ticket/1798#comment:7>
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.