[postgis-devel] Boundary of closed linestrings

[strk]

While implementing the topology population functions
I've stomped of this odd situation:

 A closed line has no boundary

Does it seems correct to you ?

This bug is about one problematic consequence of that:
http://trac.osgeo.org/postgis/ticket/770

I'd think a closed line would have a single-point boundary.
PostGIS thinks ST_StartPoint exists, and ST_EndPoint is the same.

GEOS returns MULTIPOINT EMPTY for ST_Boundary, and considers
the endpoint as if they were interior when it comes to DE9-IM
computation.

Martin: does JTS agree with GEOS on this ? if so, what's the rationale ?

--strk; 

  ()   Free GIS & Flash consultant/developer
  /\   http://strk.keybit.net/services.html