[postgis-users] error: by topology.topogeometry

[paul.malm at lfv.se]

Thanks, I will try your suggestions!
/paul

Skickat från min iPhone

> 27 mars 2019 kl. 15:44 skrev Sandro Santilli <strk at kbt.io>:
> 
>> On Tue, Mar 26, 2019 at 02:07:55PM +0000, paul.malm at lfv.se wrote:
>> 
>> UPDATE merged SET topo_geom = topology.toTopoGeom(the_geom, 'roads_topo',  1,  0.001);
> 
> [...]
> 
>> I get the following error:
>> ERROR: Error: SQL/MM Spatial exception - geometry crosses edge 8618 CONTEXT: PL/pgSQL-function topology.totopogeom(geometry,topology.topogeometry,double precision) line 112 at FOR over SELECT-rows PL/pgSQL-function topology.totopogeom(geometry,character varying,integer,double precision) line 88 at assignment
>> 
>> The error appears at different edges  when changing the tolerance on CreateTopology and toTopoGeom.
>> I added the ST_makeValid function call on the input data without any success.
>> Does anyone know what I can do?
> 
> My suggestion is to do the topology loading in chunks rather
> than all a once, like:
> 
>  ALTER TABLE merged ADD id serial PRIMARY KEY;
> 
>  UPDATE merged
>    SET topo_geom = topology.toTopoGeom(the_geom, 'roads_topo', 1)
>    WHERE id >= 0 AND id < 100;
> 
>  UPDATE merged
>    SET topo_geom = topology.toTopoGeom(the_geom, 'roads_topo', 1)
>    WHERE id >= 100 AND id < 200;
> 
>  ...
> 
> Note I've removed the tolerance parameter, don't use it unless the
> chunk fails and you're trying to move things a little bit.
> 
> Another thing to do, if a chunk fails, is to cleanup what you have
> in the topology already, before trying again to move on.
> 
> My preferred method would be to use a wrapper function catching
> exceptions and moving on if toTopoGeom fails. Often you'll find
> that other geometries will get converted fine, so at the end
> of the UPDATE call you'd have a number of "topo_geom" values
> begin NULL (for the imports that failed) and a number of valid
> ones, then you could try again to import the geometries after
> some cleaning (and some times you don't even need a new cleaning
> as the presence of more geometries would be enough to attract
> vertices for snaps..)
> 
> --strk;
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/postgis-users