[mapserver-users] Mapserver and PostgreSQL

[Paul Ramsey]

Arjen Vrielink wrote:
 > I still think Postgis doesn't really
> add much GIS functionality to Postgres other than the conversion thing and 
> some abstraction of Postgres built in functionality. But by being a key 
> player in the Mapserver - Postgres connection they hold some pretty strong 
> cards... hmm, I'm convinced after all. 

I hope to convince you more! Firstly, what we have now that native 
PostgreSQL does not have on its own:

- Real GIS objects (polygons with holes, aggregate types) not toy 
geometric objects
- The ability to spatially index spatial objects > 8K in size (if you 
think native PostgreSQL types are ready for real GIS, you need to stress 
it a little harder)
- OpenGIS standards compliance, that means standard function names, 
standard object representations (binary and ascii)
- Coordinate reprojection support in the database. (Transform() function)
- Mapserver connector and support from numerous other open source GIS 
applications (GRASS, QGIS, OpenEV, OGR)

Secondly, what we will soon have which native PostgreSQL will probably 
never bother with:

- Full robust binary predicate support. Touches(), Contains(), Relate(), 
etc etc etc, all implemented with robust algorithms. More unbreakable 
than commercial GIS databases.
- Full spatial operators support. Buffer(), Union(), Difference().
- Multikey indexes combining spatial and non-spatial columns.

Is PostGIS preferable to shape files? Depends on what your needs are. 
Jan gave a pretty good summary of the kinds of higher level things you 
can do with PostGIS which you cannot do with shape files. When it comes 
to performance, (spatially indexed) shapefiles are unbeatable. When it 
comes to application flexibility, a PostGIS-based application is hard to 
beat.

Paul

-- 
       __
      /
      | Paul Ramsey
      | Refractions Research
      | Email: pramsey at refractions.net
      | Phone: (250) 885-0632
      \_