CGAL users discussion list

[Marcel <>]

Looks like the root cause of my problem was me trying to use a 
CGAL::Point_3<CGAL::Epeck> the "wrong" way.
Manually initializing a CGAL::Lazy_exact_nt<CGAL::Gmpq> with a double and 
assigning it to that_point[0] results in strange behaviour. Using the (double 
x, double y, double z) constructor of said Point_3 works fine though.

This solves the biggest part of my problem, now I'm fighting with 
non-closedness of a torus - sphere and cube work fine.
Thanks for being my debugging rubber duck... I should've built a minimal 
example earlier. :)

Marcel

Am 08.04.2014 um 08:25 schrieb Sebastien Loriot (GeometryFactory) 
<>:

> I have trouble understanding what is your problem. Could you produce a 
> minimal example showing the problem with an assertion for example?
> 
> Thanks,
> 
> Sebastien.
> 
> 
> On 04/07/2014 12:45 PM, Marcel wrote:
>> Hello,
>> 
>> I've done further evaluations and noticed strange behaviour. The two 
>> points 'a' and 'b' I mentioned in the first post do print like this: 
>> (defined as 
>> CGAL::Point_3<CGAL::Exact_predicates_exact_constructions_kernel>)
>> 
>> a: 0.5 0.5 0.5 (cout << "a: " << a)
>> a: 0/1 0/1 0/1 (cout << "a: " << a.x().exact() << a.y().exact() << 
>> a.z().exact())
>> b: 0.5 0.5 0.5 (cout << "b: " << b)
>> b: 0/1 0/1 0/1 (cout << "b: " << b.x().exact() << b.y().exact() << 
>> b.z().exact())
>> 
>> The output values are clearly not the same (despite being represented 
>> differently).
>> 
>> The values being the same for both points looks fine as the points come 
>> from another visual mesh that actually has vertices in the same place 
>> (which is the reason why I am doing this).
>> Still, these points insist in not being equal (a == b results in "false") 
>> and I cannot seem to create a vector from a to b to measure their 
>> distance. (Resulting in the assertion error described in the first post) 
>> At least one of these circumstances is however necessary to decide if I 
>> can ignore that vertex.
>> 
>> So what might cause the different values where just representation is 
>> different?
>> And why won't CGAL create a vector from a to b for me?
>> 
>> CGAL 3.9 (Ubuntu 12.04 - unfortunately, I couldn't find a backport of CGAL 
>> 4.4)
>> 
>> Greetings,
>> Marcel
>> 
>> Am 01.04.2014 um 15:58 schrieb Marcel 
>> <>:
>> 
>>> Hello!
>>> 
>>> I'm trying to implement CSG operations with Nef_polyhedra on top of a 
>>> given framework that uses OpenSG for rendering. Now I can get all the 
>>> triangle data from OpenSG, but this results in 36 vertices for a cube as 
>>> every corner has 3 vertices (for 3 different normals). This is not a 
>>> closed mesh, therefore CSG operations on the Nef_polyhedron fail.
>>> Despite the problems that might arise from a fixed merging threshold with 
>>> complex models, I was trying to merge vertices in close proximity by 
>>> referencing the preexisting vertex in the indices array if there is one. 
>>> (When iterating over the given triangles)
>>> 
>>> So, first: Is there by chance a more or less convenient method for 
>>> automatically merging "close" vertices?
>>> 
>>> Second, the actual problem:
>>> When I try to compute the distance between two vertices, an assertion 
>>> fails:
>>> 
>>> CGAL/Interval_nt.h:89 "!(i>s) Variable used before being intialized (or 
>>> CGAL bug)"
>>> 
>>> in the last line of this snippet:
>>> 
>>> CGAL::Point a = positions->at(i); // Setting these equal manually really 
>>> makes them equal
>>> CGAL::Point b = newPoint;
>>> CGAL::Vector vecdiff(b, a);
>>> 
>>> I don't exactly understand what is wrong here. The points are certainly 
>>> close as they 'cout' as the same values. Still, they are not equal by 
>>> means of "a == b". (And I cross-checked by setting them to the same 
>>> values.)
>>> 
>>> So, is there either a way to completely avoid doing this manually or can 
>>> you tell me what I am doing wrong?
>>> 
>>> Greetings,
>>> Marcel
>>> 
>>> --
>>> You are currently subscribed to cgal-discuss.
>>> To unsubscribe or access the archives, go to
>>> https://sympa.inria.fr/sympa/info/cgal-discuss
>>> 
>>> 
>> 
>> 
> 
> 
> -- 
> You are currently subscribed to cgal-discuss.
> To unsubscribe or access the archives, go to
> https://sympa.inria.fr/sympa/info/cgal-discuss
> 
>