You called it.
Obe, Regina wrote:
...
Looking at Kevin's results I am beginning to think the bottleneck may be
in the GeomN type functions and maybe he is right that its doing a copy
call for each or something silly like that or a memory leak (I wasn't
looking is it linear or quadratic as the number of geometries
increases?).
Kevin,
What happens if you leave out the PointN call altogether just to see
what kind of penalty is being incurred there.
e.g.
SELECT t.gid, generate_series(1, ST_NumPoints(the_geom))
FROM kneufeld.tmp t
QUERY
PLAN
------------------------------------------------------------------------------------------------------------
Seq Scan on tmp t (cost=0.00..1.01 rows=1 width=339661) (actual
time=134.755..213.814 rows=21227 loops=1)
Total runtime: 280.351 ms
Total runtime: 148.297 ms
Total runtime: 147.929 ms
Total runtime: 147.780 ms
_______________________________________________
postgis-users mailing list
[email protected]
http://postgis.refractions.net/mailman/listinfo/postgis-users
