[ REQ 5165]: qhull problem

[fabri at scs.carleton.ca]

Hello,

 I have a problem with the qhull algorithm. Even for extremely 
small examples in dimension 2, and with parameters which remove
the problems with collinear points, it produces an errormessage.
A typical run:

 rbox 5 D2 | qhull C-0 Qcm -f 
qhull internal error (matchnewfacets): 3 neighbors did not match up
hash 0 f4 v4 v0 
 neighbors: 2 -1
hash 1 f5 v4 v1 
 neighbors: 3 -2

While executing: rbox 5 D2 | qhull C-0 Qcm -f
Last point added to hull was p4.

At error exit:

Convex hull of 5 points in 2-d:

  Number of vertices: 4
  Number of facets: 4

Statistics:

  Number of points processed: 4
  Number of hyperplanes created: 6
  Number of distance tests for qhull: 10
  Number of merged facets: 0
  Number of distance tests for merging: 17
  CPU seconds to compute hull (after input):  0

precision problems
      1 flipped facets




Do you have any idea what I do wrong? Thanks in advance,

Andreas Fabri