CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

Hamid G wrote:
> Hi
>
> What method would you suggest to use in order to write coordinates of 
> points from an exact kernel to a file on hard disk so that when we read 
> it later on we won't use any precision/information ?
Outputting everything in a file is generally not a good solution (but I 
understand that for testing purpose this is the fastest way, so the 
first thing to try is to increase the output precision of your stream 
using ios_base::precision.)

An alternative solution is to use the Polyhedron_incremental_builder
together with the Cartesian_converter (used on Point_3) to convert a
Polyhedron from one kernel to a Polyhedron from another kernel.

http://www.cgal.org/Manual/latest/doc_html/cgal_manual/Kernel_23_ref/Class_Cartesian_converter.html
http://www.cgal.org/Manual/latest/doc_html/cgal_manual/Polyhedron_ref/Class_Polyhedron_incremental_builder_3.html


This is not hard when you are used to, otherwise this is a good
CGAL exercise.

To use Cartesian_converter from Kernel K1 to Kernel K2, you need to do
something like this:
CGAL::Cartesian_converter<K1,K2> converter;

K1::Point_3 p_k1(1,2,3);
K2::Point_3 p_k2 = converter( p_k1 );

> I am using different modules from CGAL that don't support a common 
> kernel. So after using each module I write the results to a .OFF file 
> using provided iostream methods. And then I read that file in the next 
> module.
>
> typedef CGAL::Exact_predicates_exact_constructions_kernel Kernel;
> CGAL::Polyhedron_3<Kernel> shell;
> ofstream fid("out.off")
> fid << shell;
>
>
> I am getting random assertion failures here and there in the modules I 
> am using, similar to this:
>
> terminate called after throwing an instance of 
> 'CGAL::Precondition_exception'
>   what():  CGAL ERROR: precondition violation!
> Expr: ! k.is_degenerate_3_object()(t)
> File: /home/hamid/local/include/CGAL/Triangle_3_Segment_3_do_intersect.h
>
> This doesn't happen all the time for the same input and settings!
> So I am guessing assertion failures are somehow related to rounding off 
> the coordinates in file I/O operations.
Can the assertion happen at any step 1,2,3,4  ? or is it happening 
always at 3 or 4? does this happen while reading Polyhedra?

> This is a general view of which modules I am using from CGAL:
>
> 1- Surface_mesh with gray scale image as input (not sure which kernel 
> 'Surface_mesh_default_triangulation_3' is using)
>
> 2- Then the resulting surface mesh from 1 is used in Nef_3 for a 
> subtraction boolean operation with another surface mesh. (here I use 
> 'Exact_predicates_exact_constructions_kernel')
>
> 3- The resulting surface mesh from 2 is refined using 
> Exact_predicates_inexact_constructions_kernel in 'Subdivision_method_3'
>
> 4- The resulting surface from 3 is then fed to tetrahedral mesh 
> generator which uses Exact_predicates_inexact_constructions_kernel with 
> 'Robust_intersection_traits_3'
>
> Am I using the right kernel ? Any suggestion about which kernels are 
> better fit for this workflow assuming that I am using file I/O to send 
> the data between these modules.
>
> thanks in advance,
> Hamid G