Paul's comment about constructed points not necessarily lying on lines due
to numerical rounding is 100% correct. However, in this particular case
it turns out that the fact that ST_Intersects reports false is a bug in
GEOS. See:
https://github.com/libgeos/geos/issues/565
The good news is
Hi Paul,
Thanks for the feedback.
I would expect, as a assertion:
ST_Intersects( ST_ClosestPoint(line, point), line) = true
In our current implementation, the ST_ClosestPoint(geometry g1, geometry
g2) can return a point that is not on g1.
If ST_ClosestPoint is returning a point that is not
If the shortest line runs vertext-to-vertext I would expect it to touch both
lines. If on the other hand it runs vertex-to-midsegment I would expect it is
possible it would not touch. Check the distance of the shortest line to the two
parent lines. It should be zero or very very very small.
Hi,
Help needed :-) I'm having trouble with lines returned by
ST_ShortestLine function.
The computed lines *should* start and end on the geometries [1], but it
is not happening.
In fact, the computed geometry sometimes crosses the original geometries
or does not touch the geometries at
