CGAL users discussion list

[Michael Hoffmann <>]

Hi Benoit,

>>> After some tests, I think there is a bug in the min_strip_2 function (cf. 
>>> "minStrip.cpp" to reproduce the bug).
>>> I compute the min strip of the quadrilateral shown in figure 
>>> "minStrip.png" and I get the following results:
>>> -0.5818 -0.0609 0.547501
>>> -0.5818 -0.0609 0.257432
>>> One of these lines is printed in red.
>>> In my point of view this result is not correct, the min strip of this set 
>>> of points should be the green line and the line parallel to the green 
>>> line passing through the point (0.8726,0.6539).
>>> Thank your in advance for your help,

thanks for your report. I think the patch below fixes the problem.

Best regards,
Michael


===================================================================
--- Min_quadrilateral_traits_2.h        (revision 59078)
+++ Min_quadrilateral_traits_2.h        (working copy)
@@ -215,9 +215,9 @@
     RT
     width_numerator(const Strip_2& r, Cartesian_tag) const
     {
-      return
+      return CGAL_NTS square(
         r.second.dx() * (r.third.y() - r.first.y()) +
-        r.second.dy() * (r.first.x() - r.third.x());
+        r.second.dy() * (r.first.x() - r.third.x()));
     }
     
     RT
@@ -227,16 +227,16 @@
     RT
     width_numerator(const Strip_2& r, Homogeneous_tag) const
     {
-      return
+      return CGAL_NTS square(
         r.second.dx() *
           (r.third.hy() * r.first.hw() - r.first.hy() * r.third.hw()) +
         r.second.dy() *
-          (r.first.hx() * r.third.hw() - r.third.hx() * r.first.hw());
+          (r.first.hx() * r.third.hw() - r.third.hx() * r.first.hw()));
     }
     
     RT
     width_denominator(const Strip_2& r, Homogeneous_tag) const {
-      return r.first.hw() * r.third.hw() *
+      return CGAL_NTS square(r.first.hw()) * CGAL_NTS square(r.third.hw()) *
         (CGAL_NTS square(r.second.dx()) + CGAL_NTS square(r.second.dy()));
     }