From: Paul Ramsey <pramsey@cleverelephant.ca>
Date: Wed, 11 Nov 2009 19:35:20 +0000 (+0000)
Subject: Slightly loosen equality tolerance check for point in cone... does this fix cunit... 
X-Git-Tag: 1.5.0b1~258
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=8d538ca180bdc9f109a975f871978df20c2c25b2;p=postgis

Slightly loosen equality tolerance check for point in cone... does this fix cunit regressions?


git-svn-id: http://svn.osgeo.org/postgis/trunk@4788 b70326c6-7e19-0410-871a-916f4a2858ee
---

diff --git a/liblwgeom/lwgeodetic.c b/liblwgeom/lwgeodetic.c
index 467975041..75219469e 100644
--- a/liblwgeom/lwgeodetic.c
+++ b/liblwgeom/lwgeodetic.c
@@ -450,7 +450,7 @@ int edge_point_in_cone(GEOGRAPHIC_EDGE e, GEOGRAPHIC_POINT p)
 	** However, tuning the tolerance value feels like a dangerous hack. 
 	** Fundamentally, the problem is that this test is so sensitive.
 	*/
-	if ( vp_dot_vcp > vs_dot_vcp || fabs(vp_dot_vcp - vs_dot_vcp) < 1e-17 )
+	if ( vp_dot_vcp > vs_dot_vcp || fabs(vp_dot_vcp - vs_dot_vcp) < 1e-16 )
 	{
 		LWDEBUG(4, "point is in cone");
 		return LW_TRUE;