Thread: difference between overlap and intersect using geometric types of postgresql

Can somebody help me. I can't understand the difference between overlap
of polygon and intersect of path.
Here are the results of comparison :
geo=# \d geom
            Table "public.geom"
 Column |         Type          | Modifiers
--------+-----------------------+-----------
 parc   | character varying(10) |
 poly   | polygon               |

geo=# select * from geom;
 parc |                 poly
------+---------------------------------------
 A135 | ((0,2),(1,6),(2,7),(5,5),(5,2),(3,1))
 A52  | ((4,1),(3,3),(7,4),(8,1))
 A221 | ((5,6),(4,8),(7,9),(8,7))
 A222 | ((5,7),(4,8),(7,9),(8,7))
(4 rows)

geo=# SELECT a.parc, b.parc, a.poly as polya, b.poly as polyb , (a.poly
@ b.poly) as cont, a.poly && b.poly as over, (a.poly &> b.poly)as
dr,(a.poly &< b.poly)as gh, (a.poly::path ?# b.poly::path) as inters
from geom a, geom b ;
 parc | parc |                 polya                 |
polyb                 | cont | over | dr | gh | inters

------+------+---------------------------------------+---------------------------------------+------+------+----+----+--------
 A135 | A135 | ((0,2),(1,6),(2,7),(5,5),(5,2),(3,1)) |
((0,2),(1,6),(2,7),(5,5),(5,2),(3,1)) | t    | t    | t  | t  | t
 A135 | A52  | ((0,2),(1,6),(2,7),(5,5),(5,2),(3,1)) |
((4,1),(3,3),(7,4),(8,1))             | f    | t    | t  | t  | t
 A135 | A221 | ((0,2),(1,6),(2,7),(5,5),(5,2),(3,1)) |
((5,6),(4,8),(7,9),(8,7))             | f    | t    | t  | t  | f
 A135 | A222 | ((0,2),(1,6),(2,7),(5,5),(5,2),(3,1)) |
((5,7),(4,8),(7,9),(8,7))             | f    | t    | t  | t  | f
 A52  | A135 | ((4,1),(3,3),(7,4),(8,1))             |
((0,2),(1,6),(2,7),(5,5),(5,2),(3,1)) | f    | t    | t  | t  | t
 A52  | A52  | ((4,1),(3,3),(7,4),(8,1))             |
((4,1),(3,3),(7,4),(8,1))             | t    | t    | t  | t  | t
 A52  | A221 | ((4,1),(3,3),(7,4),(8,1))             |
((5,6),(4,8),(7,9),(8,7))             | f    | f    | t  | t  | f
 A52  | A222 | ((4,1),(3,3),(7,4),(8,1))             |
((5,7),(4,8),(7,9),(8,7))             | f    | f    | t  | t  | f
 A221 | A135 | ((5,6),(4,8),(7,9),(8,7))             |
((0,2),(1,6),(2,7),(5,5),(5,2),(3,1)) | f    | t    | t  | t  | f
 A221 | A52  | ((5,6),(4,8),(7,9),(8,7))             |
((4,1),(3,3),(7,4),(8,1))             | f    | f    | t  | t  | f
 A221 | A221 | ((5,6),(4,8),(7,9),(8,7))             |
((5,6),(4,8),(7,9),(8,7))             | t    | t    | t  | t  | t
 A221 | A222 | ((5,6),(4,8),(7,9),(8,7))             |
((5,7),(4,8),(7,9),(8,7))             | f    | t    | t  | t  | t
 A222 | A135 | ((5,7),(4,8),(7,9),(8,7))             |
((0,2),(1,6),(2,7),(5,5),(5,2),(3,1)) | f    | t    | t  | t  | f
 A222 | A52  | ((5,7),(4,8),(7,9),(8,7))             |
((4,1),(3,3),(7,4),(8,1))             | f    | f    | t  | t  | f
 A222 | A221 | ((5,7),(4,8),(7,9),(8,7))             |
((5,6),(4,8),(7,9),(8,7))             | t    | t    | t  | t  | t
 A222 | A222 | ((5,7),(4,8),(7,9),(8,7))             |
((5,7),(4,8),(7,9),(8,7))             | t    | t    | t  | t  | t
(16 rows)

You can see that A135 and A221 for example don't intersect and don't
overlap but the results of postgresql say they overlap :/

Please help me !

Didier

dhoubrechts <dhoubrechts@cybernet.be> writes:
> Can somebody help me. I can't understand the difference between overlap
> of polygon and intersect of path.

A little looking in the code finds:

/*-----------------------------------------------------------------
 * Determine if polygon A overlaps polygon B by determining if
 * their bounding boxes overlap.
 *
 * XXX ought to do a more correct check?
 *-----------------------------------------------------------------*/
Datum
poly_overlap(PG_FUNCTION_ARGS)


Feel free to submit a better implementation ...

            regards, tom lane