Another bogus result from ST_Snap

[strk]

A: LINESTRING(765795.1 2949559.2,765795.2 2949559.2,765794.9 2949559.5,765795.1 2949559.2)
B: LINESTRING(765795.1 2949559.2,765794.9 2949559.2,765795.2 2949558.8,765795.1 2949559.2)
tolerance: 0.25

Bottom-right vertex of line A gets snapped to top-left vertex of line B but the two points are at a distance of 0.3.

See #1714 for the source of this

[strk]

I think I get it: first the bottom-left corner of A (top-right triangle in the picture, with red vertices) gets snapped to top-left corner of B (bottom green triangle in the picture), then the bottom-right corner of A gets snapped to the top-right corner of B. In this case both distances are minor than 0.25.

To be verified.

[strk]

Note that always snapping to the closest vertex would have resulted in a collapse of the ring into a line.

[strk]

I think a collapse would be a sensible result for this case.

What might help, as an analytical and debugging function, would be one reporting the smallest distance between any two vertices in a geometry.

[strk]

The result of this with latest changes in GEOS (upcoming 3.3.9) is that the bottom-left of A gets snapped to top-left corner of B and nothing more. The result is simple, valid and non-collapsed. To be closed once GEOS 3.3.9 is released.

REF ​http://trac.osgeo.org/geos/ticket/501

[strk]

GEOS-3.3.9 was released 3 months ago: ​http://trac.osgeo.org/geos/milestone/3.3.9