Hi,<br><br>I have a query which on a notebook with Window7 - 64 bit ,<br>with postgres 8.4.4 32bit and Postgis 1.5.1 32 bit <br>work in about 10 minutes.<br><br>When I try the same query on a machine quite big then the notebook.<br>
And with <br>Linux RedHat 64 bit ,<br>but using Postgres 8.4.4 compile from source (64bit)<br>and Geos from trunk, and Postgis from trunk all compiled to 64 bit.<br><br>The same query sql run for a time really huge.<br>(some hours ..)<br>
<br>The query is this<br>INSERT INTO PUBLIC._TEST_CAMPIONAMENTO (id,idrtt,geom) (<br> select <br> a.oid::integer,<br> <a href="http://a.id">a.id</a>,<br> ST_Force_2D(ST_Multi(a.geom)) <br>
from <br> public.table1 as a,<br> public._sample_extent as b<br> where <br> ST_Intersects(a.geom,b.geom)=true<br>)<br>;<br><br>The only think I can do is passing from the trunk version 2.0.0 to the 1.5.1 version,<br>
but this mean to reload the table "table1". Because this table is quite huge (more the 9.500.000 records with many vertex on each records)<br>and reload it require a bit of time.<br>And I have some doubt that the 2.0.0 trunk was more slow than 1.5.1.<br>
<br>Before try this I have a question to ask:<br><br>the table" _sample_extent " is one only record with a multipolygon having about 20 part disjoint each other, but not near each other.<br>I don't know if the GIST index use only one unique extent of all the MultiPolygon or use more single extents one for each parts ?<br>
<br>And could be somethink is changed from 1.5.1 to 2.0.0 in the strategy of GIS index ?<br><br>Many thx,<br><br>Andrea<br><br>-- <br>-----------------<br>Andrea Peri<br>. . . . . . . . . <br>qwerty àèìòù<br>-----------------<br>
<br>