[postgis-users] simplify weirdnesses

[strk]

schabios wrote:
> Hi, strk,
> 
> On Tue, 17 Feb 2004 15:13:53 +0100
> strk <strk at keybit.net> wrote:
> 
> > > Is it desired that a PostGIS operation on valid geometries can make
> > > them non-valid?
> > It is *not* desired, but it would be very expensive to avoid this.
> 
> I was afraid to hear this :-(
> 
> > > And it seems that some (well, 65 thousand) Geometries even get
> > > deleted by simplify. 
> > Collapsed geometries do.
> 
> Yes, I know. But I suspect that the algorithm might collapse geometries
> that are larger than the simplification threshold.
> 
> > > As another example shows, the size of the deleted geometries can be
> > > much larger than the given simplification threshold:
> > > 
> > > logigis=# select box3d(geom) from admin_area_reconstruct where
> > > simplify(geom,0.05) is null;
> > > 
> > >                             box3d
> > > -------------------------------------------------------------
> > > BOX3D(5.95587 45.81802 1.7e-308,10.49203 47.80838 1.7e-308) 
> > > (1 row)
> > > 
> > > Can you explain this?
> > 
> > What kind of geometries are you simplifying ?
> 
> Polygons and Multipolygons. Water areas.

A multipolygon composed of 2 small polygons far away
from each other will have a huge box3d but could
be simplified to no polygons at all. 

--strk;