CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

What kernel are you using internally to do the computation?
I guess you export the result into double/float which result in rounding
of the exact output, which result in points having the same coordinates
in your case.

Sebastien.


On 08/31/2014 02:51 AM, crobar wrote:
> Hi,
>
> I'm working on a computational solid geometry package for the Matlab/Octave
> programming language. This interface is based on the project pyPolyCSG,
> which makes use of CGAL to perform it's operations. I have forked and
> modified this project to make a more generally useful library that is a
> wrapper for CGAL. In testing, I have found that CGAL appears to make meshes
> with duplicate nodes under certain conditions which then lead to errors in
> subsequent operations.
>
> The particular shape that causes the problem for me is constructed from the
> difference between a sphere and a cube.
> <http://cgal-discuss.949826.n4.nabble.com/file/n4659760/fillet.png>
>
> This was constructed by subtracting this sphere:
>
> <http://cgal-discuss.949826.n4.nabble.com/file/n4659760/sphere_unrotated.png>
>
> from this cube:
>
> <http://cgal-discuss.949826.n4.nabble.com/file/n4659760/cube.png>
>
> The resulting shape has a duplicate node at the point (273.4741e-003,
> 0.0000e+000, 418.5832e-003) i.e. on a point on the z-y plane. Neither shape
> has duplicate nodels before the difference operation. And when I try to the
> difference this object with another cube, I get:
>
> Error !!!!!!!!!!!!!!!!!!!!!!!
> Error !!!!!!!!!!!!!!!!!!!!!!!
>
> from CGAL
>
> However, the problem goes away if I first rotate the sphere by 90 degrees
> around the y-axis. I suspect this is because one of the lines making up the
> sphere (which has 20 x 20 segments in these examples) is no longer lying in
> the same plane as the edge of the cube.
>
> is there a possibility of a bug here?
>
> To reproduce the issue, you can find my modified pyPolyCSG project here:
>
> https://github.com/crobarcro/pyPolyCSG
>
> however, you don't need the whole project, only the "libpolyhcsg"
> subproject, which does not use python or matlab. It is a cmake project and
> can be built and installed with:
>
> cmake .
> make
> sudo make install
>
> The main files of relevance are polyhedron.h/cpp, ane
> polyhedron_binary_op.h/cpp
>
> The actual issue is reproduced in a test program, which you can find in the
> "test/cornerrd" subdirectory here:
>
> https://github.com/crobarcro/pyPolyCSG/tree/master/libpolyhcsg/test/cornerrad
>
> This is not a cmake project, but is only a single file, main.cpp.  A
> code::blocks project is provided.
>
> If I run this I get the following exception thrown when trying to difference
> the polyhedron with the duplicate vertices with a cube:
>
> Debugger name and version: GNU gdb (Ubuntu 7.7-0ubuntu3.1) 7.7
> In __cxa_throw () (/usr/lib/x86_64-linux-gnu/libstdc++.so.6)
> #1  0x00007ffff799dea8 in CGAL::Uncertain<bool>::make_certain
> (this=0x7fffffffd8f0) at /usr/local/include/CGAL/Uncertain.h:125
> /usr/local/include/CGAL/Uncertain.h:125:2852:beg:0x7ffff799dea8
> At /usr/local/include/CGAL/Uncertain.h:125
>
> The backtrace:
>
> #0  0x00007ffff71a3a30 in __cxa_throw () from
> /usr/lib/x86_64-linux-gnu/libstdc++.so.6
> #1  0x00007ffff799dea8 in CGAL::Uncertain<bool>::make_certain
> (this=0x7fffffffd8f0) at /usr/local/include/CGAL/Uncertain.h:125
> #2  0x00007ffff7979636 in CGAL::Uncertain<bool>::operator bool
> (this=0x7fffffffd8f0) at /usr/local/include/CGAL/Uncertain.h:133
> #3  0x00007ffff799d927 in
> CGAL::operator==<CGAL::Simple_cartesian&lt;CGAL::Interval_nt&lt;false> > >
> (v=..., w=...) at /usr/local/include/CGAL/Cartesian/Vector_3.h:143
> #4  0x00007ffff7978e09 in
> CGAL::CommonKernelFunctors::Equal_3<CGAL::Simple_cartesian&lt;CGAL::Interval_nt&lt;false>
> > > ::operator() (this=0x7fffffffda11, v1=..., v2=...) at
> /usr/local/include/CGAL/Kernel/function_objects.h:2229
> #5  0x00007ffff795aa09 in
> CGAL::Filtered_predicate<CGAL::CommonKernelFunctors::Equal_3&lt;CGAL::Simple_cartesian&lt;CGAL::Gmpq>
> > ,
> CGAL::CommonKernelFunctors::Equal_3<CGAL::Simple_cartesian&lt;CGAL::Interval_nt&lt;false>
> > > , CGAL::Exact_converter<CGAL::Epeck,
> CGAL::Simple_cartesian&lt;CGAL::Gmpq> >, CGAL::Approx_converter<CGAL::Epeck,
> CGAL::Simple_cartesian&lt;CGAL::Interval_nt&lt;false> > >,
> true>::operator()<CGAL::Vector_3&lt;CGAL::Epeck>,
> CGAL::Vector_3<CGAL::Epeck> > (this=0x7fffffffda10, a1=..., a2=...) at
> /usr/local/include/CGAL/Filtered_predicate.h:214
> #6  0x00007ffff7941af0 in CGAL::operator==<CGAL::Epeck> (p=..., q=...) at
> /usr/local/include/CGAL/Kernel/global_functions_3.h:784
> #7  0x00007ffff792e540 in
> CGAL::Facet_plane_3::operator()<CGAL::HalfedgeDS_in_place_list_face&lt;CGAL::I_Polyhedron_facet&lt;CGAL::HalfedgeDS_face_base&lt;CGAL::HalfedgeDS_list_types&lt;CGAL::Epeck,
> CGAL::I_Polyhedron_derived_items_3&lt;CGAL::Polyhedron_items_3>,
> std::allocator<int> >, CGAL::Boolean_tag<true>, CGAL::Plane_3<CGAL::Epeck> >
> > > > (this=0x7fffffffdb60, f=...) at
> /usr/local/include/CGAL/Nef_3/polyhedron_3_to_nef_3.h:105
> #8  0x00007ffff791fd2c in
> std::transform<CGAL::internal::In_place_list_iterator&lt;CGAL::HalfedgeDS_in_place_list_face&lt;CGAL::I_Polyhedron_facet&lt;CGAL::HalfedgeDS_face_base&lt;CGAL::HalfedgeDS_list_types&lt;CGAL::Epeck,
> CGAL::I_Polyhedron_derived_items_3&lt;CGAL::Polyhedron_items_3>,
> std::allocator<int> >, CGAL::Boolean_tag<true>, CGAL::Plane_3<CGAL::Epeck> >
> > > ,
> std::allocator<CGAL::HalfedgeDS_in_place_list_face&lt;CGAL::I_Polyhedron_facet&lt;CGAL::HalfedgeDS_face_base&lt;CGAL::HalfedgeDS_list_types&lt;CGAL::Epeck,
> CGAL::I_Polyhedron_derived_items_3&lt;CGAL::Polyhedron_items_3>,
> std::allocator<int> >, CGAL::Boolean_tag<true>, CGAL::Plane_3<CGAL::Epeck> >
> > > > > ,
> CGAL::Iterator_project<CGAL::internal::In_place_list_iterator&lt;CGAL::HalfedgeDS_in_place_list_face&lt;CGAL::I_Polyhedron_facet&lt;CGAL::HalfedgeDS_face_base&lt;CGAL::HalfedgeDS_list_types&lt;CGAL::Epeck,
> CGAL::I_Polyhedron_derived_items_3&lt;CGAL::Polyhedron_items_3>,
> std::allocator<int> >, CGAL::Boolean_tag<true>, CGAL::Plane_3<CGAL::Epeck> >
> > > ,
> std::allocator<CGAL::HalfedgeDS_in_place_list_face&lt;CGAL::I_Polyhedron_facet&lt;CGAL::HalfedgeDS_face_base&lt;CGAL::HalfedgeDS_list_types&lt;CGAL::Epeck,
> CGAL::I_Polyhedron_derived_items_3&lt;CGAL::Polyhedron_items_3>,
> std::allocator<int> >, CGAL::Boolean_tag<true>, CGAL::Plane_3<CGAL::Epeck> >
> > > > > ,
> CGAL::Project_plane<CGAL::HalfedgeDS_in_place_list_face&lt;CGAL::I_Polyhedron_facet&lt;CGAL::HalfedgeDS_face_base&lt;CGAL::HalfedgeDS_list_types&lt;CGAL::Epeck,
> CGAL::I_Polyhedron_derived_items_3&lt;CGAL::Polyhedron_items_3>,
> std::allocator<int> >, CGAL::Boolean_tag<true>, CGAL::Plane_3<CGAL::Epeck> >
> > > > , int, int, int, int>, CGAL::Facet_plane_3> (__first=..., __last=...,
> __result=..., __unary_op=...) at /usr/include/c++/4.8/bits/stl_algo.h:4926
> #9  0x00007ffff791a764 in
> CGAL::polyhedron_3_to_nef_3<CGAL::Polyhedron_3&lt;CGAL::Epeck,
> CGAL::Polyhedron_items_3, CGAL::HalfedgeDS_default, std::allocator&lt;int>
> > , CGAL::SNC_structure<CGAL::Epeck, CGAL::SNC_indexed_items, bool> > (P=...,
> S=...) at /usr/local/include/CGAL/Nef_3/polyhedron_3_to_nef_3.h:198
> #10 0x00007ffff7917d2c in CGAL::Nef_polyhedron_3<CGAL::Epeck,
> CGAL::SNC_indexed_items, bool>::Nef_polyhedron_3<CGAL::Epeck,
> CGAL::Polyhedron_items_3, CGAL::HalfedgeDS_default, std::allocator&lt;int> >
> (this=0x7fffffffe270, P=...) at
> /usr/local/include/CGAL/Nef_polyhedron_3.h:566
> #11 0x00007ffff79112b8 in polyhcsg::polyhedron_to_cgal (p=...) at
> /home/rcrozier/src/pyPolyCSG-crobarcro/libpolyhcsg/source/polyhedron_binary_op.cpp:86
> #12 0x00007ffff7911bb4 in polyhcsg::polyhedron_difference::operator()
> (this=0x7fffffffe300, A=..., B=...) at
> /home/rcrozier/src/pyPolyCSG-crobarcro/libpolyhcsg/source/polyhedron_binary_op.cpp:140
> #13 0x00007ffff7acbe29 in polyhcsg::polyhedron::operator-
> (this=0x7fffffffe340, in=...) at
> /home/rcrozier/src/pyPolyCSG-crobarcro/libpolyhcsg/source/polyhedron.cpp:758
> #14 0x00000000004013e6 in main (argc=1, argv=0x7fffffffe5c8) at
> /home/rcrozier/src/pyPolyCSG-crobarcro/libpolyhcsg/test/cornerrad/main.cpp:16
>
>
> If you are interested in the matlab.Octave interface, it can be found here:
>
> https://github.com/crobarcro/mpolycsg
>
> It is quite functional, but not yet documented.
>
>
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/CGAL-difference-on-Nef-polyhedron-produces-shape-with-duplicate-vertices-tp4659760.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.