CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

Thanks for the update.

Can you try before the assertion

CGAL_assertion(new_rd >= copy_rd);

to change
new_rd=Orthogonal_distance_instance.new_distance(copy_rd,old_off,new_off,new_cut_dim);
into
new_rd=old_off==new_off?copy_rd:Orthogonal_distance_instance.new_distance(copy_rd,old_off,new_off,new_cut_dim);

I still suspect an optimization problem.

Sebastien.

On 03/06/2012 04:51 PM, Nicholas Mario Wardhana wrote:
> On 2 March 2012 20:18, Nicholas Mario 
> Wardhana<>
>   wrote:
> > To be more exact, the problem arises when incrementing
> > neighbourIterator (neighbourIterator++).
> >
> > Best regards,
> > Nicholas
>
> Hi all,
>
> I hope it is still fine for me to "resurrect" this few days-old thread.
>
> After a few days, I cannot generate a small point list which can
> result in assertion violation. But after playing around with my code,
> I detected a violation under Debug configuration with the involved
> values being
>
> new_rd = 3111.4645430113583 , and
> copy_rd = 3111.4645430113587 .
>
> I can conclude that my problem is with the floating point accuracy. So
> I changed lines 267 and 288 to
>
> CGAL_assertion(new_rd + 0.0001>= copy_rd);
>
> and the code works fine for me. Alternatively, I suppose I can just
> comment these lines, since these violation does not seem to be fatal.
> CMIIW
>
> Thank you very much for your help!
>
> Best regards,
> Nicholas