CGAL users discussion list

["Fred Dorosh" <>]

> I have found and evaluated another GPL library that specializes in boolean
> operations on Polyhedra. This library (CARVE) performed on average 100
> (ONE
> HUNDRED) times faster than CGAL with the typical use-cases that apply to
> our
> application. It also proved completely computationably stable with our
> tests.

For the record, I also evaluated Carve for our project, and found similar
performance results (at least for low volumes of data). It's mostly
imputable to the overhead of using an exact arithmetic kernel.

On the other hand I rapidly ran into instability and crashes even with
some simple use cases. Our project required perfect robustness, and so
CGAL/Nef3 was ultimately retained for this reason.

Fred