[postgis-users] The operator && is not overlaps operator basedonthe OGC SF SQL

[Paul Ramsey]

Just to clarify Darko: you are aware that the && test applies *only* to 
the geometries *bounding boxes* not to the geometries themselves, right? 
We use the && operator only for a "first pass" thinning of candidate 
geometries, which are then passed to the more intensive GEOS testing. 
("geom && candidate AND intersects(geom,candidate)")

Is the exact semantic of the bounding box test still important to you?

Darko Androsevic wrote:

> Hello Nicolas,
> 
> Thanks for the quick answer. However, I have to ask you for one additional
> thing.
> 
> Nicolas wrote:
> "&&" tells if 2 geometries' extents overlap or not."
> Since we agree that it IS NOT the OGC overlap, I would appreciate if you
> could you please define the bounding boxes "overlap" in the postGIS context
> or point me where it might be precisely defined? 
> 
> 
> BTW, I still think, based on my small set of simple test queries, that it
> actually follows the OGC's "intersects" = "not disjoint" operators sematics.
> Why is this important to me? 
> I would like to use it in combination with GEOS extensions to get the OGC
> operator semantics.
> For example, if it is defined as the OGC intersect then one does not have to
> use the additional GEOS intersect extension in case of the OGC's intersect
> query and improve performance.