[gdal-dev] OGR and GEOS

[Frank Warmerdam]

Kevin Webb wrote:
> Greetings!
> 
> The documentation for the OGRGeometry class mentions the GEOS library in 
> a few of the
> method definitions, but I wonder if there is a "stronger" OGR dependency 
> on the GEOS
> library than the documentation suggests. Perhaps GEOS should be a 
> required library.
> 
> I had done some processing using OGRGeometry::Intersects() without the 
> GEOS library
> installed, I then installed GEOS to run the same tests using 
> OGRGeometry::Distance().
> After installing GEOS, but before running the OGRGeometry::Distance() 
> tests, I re-ran the
> code using OGRGeometry::Intersects() and noticed different intersection 
> results.
> 
> Before-GEOS, the Intersects() method indicated intersection with 2 layer 
> polygons. After-GEOS,
> the Intersects() method indicated intersection with 1 layer polygon. 
> Note: After-GEOS results
> for both Intersects() and Distance() == 0 were the same.
> 
> Should the GEOS library be a requirement for deterministic OGR processing?

Kevin,

The Intersects() is bounding box based without GEOS, and based on
actual geometry intersection if it is available.  That is presumably
why your results differ.

For a large number of people OGR is just fine without GEOS, so I see
no reason to make it a fixed dependency.  It might be prudent to better
document the caveats of operating without it.

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent