[postgis-devel] st_pointonsurface fails

[Sandro Santilli]

On Fri, Feb 17, 2012 at 11:44:56AM +1100, James Sewell wrote:

> I've discovered a supposed valid (but very thin) geometry which ST_PointOnSuface fails on:
...
> I presume it has something to do with being so thin, but it comes up valid so it must have a least one internal point to be a polygon ... right?

How does it fail ? I guess it throws a TopologyException ?
The PointOnSurface algorithm intersects the longest bounding
box median with the geometry, which may be where the exception
raises (you can try that manually). Maybe trying the other
median in case of an exception would fix this case.

Feel like filing a ticket on geos trac ?
Use HEXWKB format for the geometry.

http://trac.osgeo.org/geos

--strk; 

  ,------o-. 
  |   __/  |    Delivering high quality PostGIS 2.0 !
  |  / 2.0 |    http://strk.keybit.net
  `-o------'