[postgis-users] PostGIS topology use ?

Stephen Woodbridge woodbri at swoodbridge.com
Thu Jul 1 16:33:14 PDT 2010

strk wrote:
> On Thu, Jul 01, 2010 at 09:43:33AM -0400, Stephen Woodbridge wrote:
>> 1. working with polygon coverages to maintain boundary coherence during 
>> edits
>> 2. simplify adjacent polygons without leaving gaps between them.
>> 3. editing road networks and maintaining coherence of joined segments 
>> and intersections
>> While these are more high level application ideas and I guess at the 
>> lowest level you could build a topology that would support any of these, 
>> where did not seem to be support for any higher level functions that 
>> would help support implementing these ideas.
> Right, not very management function reach at this time...
> For (2) the simplification part it wouldn't be too hard to do.
> Currently, if you use ST_Simplify(TopoGeometry) you'd get
> an automatic cast from TopoGeometry to Geometry, and _then_
> the Geometry would be simplified.
> A simple override of that function could instead _first_
> simplify components and _finally_ polygonize them togheter
> to get you the Geometry. Would that meet your use case 2 ?
> For (3) I don't think topology is for you. SQL/MM defined another
> model which was "network". Networks have different characteristics
> from topology. For example you may have intersections which do
> not form nodes (think bridges).
> Case (1) is the whole point of keeping the representation of
> your data as a topology, but I'm not sure I understand what
> you mean by 'during edits'. Postgis topology does have a
> validation function you could run after edits, but I guess
> your edits would be on the app level. For instance QGIS
> itself does use some form of topological representation when
> doing edits. Would be interesting to see it cope directly
> with topology in the backend.

Well, in general it a good start and I liked what I saw as far as it 
went. I think that the biggest problem with it at the moment is 
chicken-egg problem of not being mature enough feature wise to be useful 
and attract funding to add more to it and not funding or time to move it 
in that direction.

If you think about PostGIS and how it has matured over the years and how 
much integrating GEOS has done to increase its value as a tool, I think 
it is obvious in comparison that we need the topological equivalent of 
that effort to build out the topology engine.

The SQL/MM network is another interesting idea I need to look into 
because I do a lot with routing.


PS. It is great to see you actively involved in with PostGIS and GEOS 
again. Thanks for all you past and present efforts.

More information about the postgis-users mailing list