Qhull/convex hull non-functional

[Ben West]

The output of qhull is incorrect, or does not match up with that of IDL. It doesn't seem that qhull is functioning properly at all: in a test set of data I made, half of the identified faces to the convex hull of the three-dimensional data set have two (of three) vertices being identical. Furthermore, only 8 faces (4 of which are actually faces) whereas IDL identifies 54 faces, and the faces that are actually proper triangles don't seem to be the same (as in, don't appear in the output from IDL).

What's interesting is that triangulate (which, I believe, works using qhull) does match up with IDL, so the issue is probably with qhull itself.

Attached is the set of data I am talking about, which was randomly generated and then tested in both FL 0.79.40 and IDL 8.5.1, along with the outputs each generated.

[Lajos Foldy]

Thanks for the report. It's a stupid error on my part, qhull was always called with ndim=2. Fixed for the next snapshot.

[Lajos Foldy]

Closed.