Bug in ST_LineCrossingDirection ?

[strk]

The manual page about ST_LineCrossingDirection is not very clear on the matter, but I would expect this to return -1 (crosses left) instead of 1 (crosses right):

SELECT ST_LineCrossingDirection(
 'LINESTRING(20 37,23 37,20 34,20 37)', -- g1
 'LINESTRING(20 34,20.5 33)' --g2
);

[strk]

LineCrossingDirection inputs

[strk]

See image, the first geometry passed to the function is the blue triangle. The green line clearly stands on the *left* side of it, no ?

[pramsey]

Uh, not sure. Segment by segment, it's on the left of segment two and the right of segment three. A closed string has a "left" and a "right" deterministicly in a way an open string does not necessarily.

[strk]

I think the problem is actually well defined, even if the blue line was open. The green line keeps being on the left side.

Maybe ST_LineCrossingDirection is not the right function to use for solving the problem. Should the documentation is not clear about the behaviour: "hat kind of crossing behavior exists between them" - ​https://postgis.net/docs/ST_LineCrossingDirection.html

[robe]

In prep for 3.1.3 release

[pramsey]

I'm not up to change this as I don't think it's a problem.