[postgis-users] Trunk 2.0.0 seem slow on Linux 64 bit

[Andrea Peri]

>>Hallo

>>*
>*>*First your question about index-handling of multi-geometries.
*>>*The index handles the whole geometry as one big so in your case where
*>>*the multi-parts are spread you will get a lot better performance if you
*>>*use st_dump to split the multi to several rows of single polygons and
*>>*build an index on that.
*
>Hi,
>
>That work very well.
>
>Always do index on the dumped version of geometries
>
>Now the query time is down to few minutes. :))
>
>Dumping the geometry in single parts is definitively better for index.
>
>Now I do some tests, as you suggest, to reduce the number of unknown.


Hi,

(just for know)

Dumping the MultiPolygon to improve the performance of index, may give a new
trouble.
Infact before having 1 unique Geometry multipolygon when it was intersect
with the archive, it return exactly the geometry intersecting.
Now using a dumped version of MultiPolygon, the query return in output the
same geometry for every polygon it intersects.

I notice infact that a line (a river) that intersects many polygons will be
extract more than one time, one for every intersects polygon.

So is necessary to add the clausole "distinct" to the query, to avoid the
duplicates .

-- 
-----------------
Andrea Peri
. . . . . . . . .
qwerty àèìòù
-----------------
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20100822/bc80abb9/attachment.html>