[postgis-users] Topology precision

Marc-Antoine Nüssli nuessli.ma at gmail.com
Tue Jul 10 02:22:27 PDT 2018

Thank you for your answer!
This is much clearer now. So, I will continue to live with floating
precision operations for the moment.

Best regards,

Le ven. 6 juil. 2018 à 22:34, Sandro Santilli <strk at kbt.io> a écrit :

> On Fri, Jul 06, 2018 at 11:25:16AM +0200, Marc-Antoine Nüssli wrote:
> > Hi there,
> >
> > First, I want to say a big thanks to Sandro Santilli and all people who
> > work on the topology features. It looks like it becomes now mature enough
> > to be used for production applications and this is really great to manage
> > geo-data much more consistently than with simple geometries (especially
> > when you deal a lot with territorial partitions like in my case)
> Thanks for the feedback!
> > I have a question regarding the precision value that we have to provide
> > when creating a topology. I am not sure to understand its purpose.
> > Somehow, I would have expected that it would be used when computing new
> > nodes positions, for example resulting from an intersection, but from my
> > tests it seems it is not the case (see my example below)
> > Do you have some links or other infos that would describe how & when this
> > global topology precision parameter is used?
> Using it for nodes position is planned. Doing so will need introduction
> of the concept of fixed-precision operations which are still unavailable
> in PostGIS. Such operations would first enforce snapping of input
> geometries to a fixed-precision grid and then ask GEOS to perform the
> operation using that grid.
> Right now PostGIS Topology only uses the "precision" value to snap
> incoming lines and vertices to existing nodes and vertices when using
> toTopoGeom or the underlying TopoGeo_addPoint, TopoGeo_addLinestring
> and TopoGeo_addPolygon and not overriding the global value.
> --strk;
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20180710/977e23b9/attachment.html>

More information about the postgis-users mailing list