[postgis-devel] PIP short circuits

Paul Ramsey pramsey at cleverelephant.ca
Wed Nov 26 12:16:25 PST 2008


The most likely case is the one-member-multipoint, which probably does
occur quite often. (I'd have to test, but I think that's what the
shape importer creates) And would benefit, of course, from
acceleration.

P.

On Wed, Nov 26, 2008 at 11:34 AM, Kevin Neufeld
<kneufeld at refractions.net> wrote:
> As you all know, thanx to Paul and Mark, the short circuiting in PostGIS is
> now bug free (that we know of :) ) and works with MultiPolygons.
>
> Unless I'm reading the code base wrong, this logic does not yet include
> MultiPoints.  Does it make sense to include this as a todo item?  Is it
> faster to perform 100 quick PIP tests iterating over the points in a
> multipoint than it is to shift to GEOS and execute an actual intersects
> operation?  Obviously it depends on the number of points in a collection.  A
> collection of 2 points would probably benefit using PIP, 2000 points,
> probably not.  What is the threshold?  No idea.
>
> Next question.  It is worth it?  Do people even use MultiPoints?
>
> Just a thought.
> -- Kevin
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-devel
>



More information about the postgis-devel mailing list