CGAL users discussion list

[Marc Glisse <>]

On Thu, 22 Jan 2015, Marius Kintel wrote:

> Sounds very cool. The lack of thread safeness is a major concern for us 
> (OpenSCAD) when using CGAL.
> I’m not familiar with the inner workings of the shared objects in CGAL, 
> but do you think your patch will make it possible to execute multiple 
> Nef polyhedron CSG operations in parallel (i.e. multiple threads reading 
> from the same operands but producing different output) ?

I don't know what you are using exactly, but did you try a kernel like 
Simple_cartesian<mpq_class> ? It doesn't use reference counting, that 
should avoid a number of thread-safety issues. Or maybe 
Filtered_kernel<Simple_cartesian<mpq_class> > to speed up predicates; see 
CGAL/Exact_predicates_exact_constructions_kernel.h for details how to turn 
it into a "true" kernel (i.e. make CGAL::Point_3<K> the same type as 
K::Point_3) if you need that.

--
Marc Glisse