On Thu, Jun 15, 2006 at 11:48:37PM -0400, Tom Lane wrote:
> "Emilia Venturato" <venturato@faunalia.it> writes:
> > Postgis developper said it could be a postgresql bug.
>
> Or it could be a postgis bug. Without a test case we can use to
> reproduce the problem, it's all speculation. Please send a complete,
> self-contained test case...
This report resembles a message Emilia posted in postgis-users a
couple of weeks ago. The only public discussion is a request for
the PostGIS version and copy of the data:
http://postgis.refractions.net/pipermail/postgis-users/2006-June/012281.html
http://postgis.refractions.net/pipermail/postgis-users/2006-June/012282.html
Emilia, did you and Sandro (strk) have off-list discussion about
this problem? What do version() and postgis_full_version() return?
What happens if you select the geometry column without a join, i.e.,
"SELECT the_geom FROM wwf_terr_ecos_multigeom WHERE ..."? Do you
get the segmentation fault with the original query if you select
AsText(the_geom) or AsEWKT(the_geom) instead of just the_geom?
Did the segmentation fault leave a core dump in your $PGDATA directory
or somewhere beneath it? If not then you might need to adjust your
coredumpsize resource limit.
--
Michael Fuhr