CGAL users discussion list

[Andreas Fabri <>]

Camille Wormser wrote:
> Bernhard Kornberger a écrit :
> > This solves the problem. Sure, this is documented, but a programmer
> > can not expect that behavior. 
> I agree with you.
> When I checked your code, I assumed that it was a bug (and not a 
> feature...), and I corrected it. I attach modified versions of the 
> HalfedgeDS_items_decorator and the Polyhedron_builder_3 which work "as 
> expected" (that is, when the Supports_face_plane flag is true, the 
> equation is computed).

Hi Camille,

Note that it is a feature not a bug to not compute the plane.
So please don't hastily commit it to svn, but first discuss it
on the developer mailing list.

Note that the tag is a tag and not a variable which can change
the state at runtime, and it has "Supports" in the name, and
not "Computes".
> With these files, your program works fine (without valgrind...).
> > And even worse, the function returns
> > a wrong plane instead of some error message. Shouldn't the function
> > have an assertion like assert(plane_is_valid()); ??
> The problem is that the Supports_face_plane flag triggers a curious 
> behavior: when it is true, a Plane data member is present in the HDS 
> Face base, BUT it is not computed. Even stranger is the fact that 
> Supports_face_plane is true by default. If it were false, you would have 
> had a compile time error, as expected.

I guess what we might do is to add a boolean variable to the polyhedron
class or the face class that is true if the plane is computed, and when
one wants to access it, one checks the flag.

best,

andreas