[postgis-users] Topology performance test

[Sandro Santilli]

On Wed, Jun 27, 2012 at 11:54:56AM +1000, Luca Morandini wrote:
> On 06/22/2012 09:05 PM, Sandro Santilli wrote:
> >>
> >>Not at all ! Congratulations, you found a bug !
> >>Ouch, it hurts pretty much. Time to push a 3.3.5 to fix this mess.
> 
> Thanks Sandro, with the release of GEOS 3.3.5 things look in order:
> 
> select postgis_full_version();
> 
> POSTGIS="2.1.0SVN r9949" GEOS="3.3.5-CAPI-1.7.5" PROJ="Rel. 4.7.1,
> 23 September 2009" LIBXML="2.7.6" TOPOLOGY
> (1 row)
> 
> However, the processing time for building the topology has not changed (about 15m).
> If you remember, it was supposed that the latest and greatest GEOS
> and PostGIS would speed up things, which did not happen... what to
> do next ?

You could enable debugging and see if you can spot a pattern in how timing
goes. Does it slow down as geometries are added ? Is there a single or small
set of geometries that take a lot to import ? How does the time compares with
using ST_CreateTopoGeo (which takes the whole input at once, doesn't attempt
any snapping and doesn't construct the TopoGeometry objects) ?

--strk;

 Sent from our free software
 http://www.gnu.org/philosophy/free-sw.html