[geos-devel] ST_Snap

[Sandro Santilli]

[moved to geos-devel only]

On Wed, Feb 29, 2012 at 10:57:15AM +1100, James Sewell wrote:

> I was under the impression that ST_Snap(a, b, x) snapped geom a to
> geom b within precision x. My understanding was that in practice this
> means that all lines in geom a are snapped to geom b if they are within x.
> 
> Am I misunderstanding how this works, or am I misunderstanding
> snap-rounding? Comments?

The function snaps vertices and segments of A to _vertices_ of B.
A segment/segment intersection alone doesn't produce a snap.
Also the current procedure may return invalid geometries, which makes
things further complicated. There's a ticket about that.

I guess this is what Martin refers to when he talks about an
heuristic approach. I'm curious as well about Martin's definition
of "snaprounding".

--strk;