Geometry Boost List: [ggl] ggl::equals

Subject: [ggl] ggl::equals
From: Barend Gehrels (Barend.Gehrels)
Date: 2009-07-11 07:56:59

Hi Chris,

>
> In the context of ggl::index, is this ggl:equals code right?
>
>     template <typename B1, typename B2, std::size_t D, std::size_t N>
>     struct box_box
>     {
>         static inline bool apply(B1 const& box1, B2 const& box2)
>         {
>             if (!math::equals(get<max_corner, D>(box1),
>     get<min_corner, D>(box2))
>                 || !math::equals(get<min_corner, D>(box1),
>     get<max_corner, D>(box2)))
>             {
>                 return false;
>             }
>             return box_box<B1, B2, D + 1, N>::apply(box1, box2);
>         }
>     };
>

What happened here... I was sleeping. It is totally wrong indeed. It 
should be:

        if (!math::equals(get<min_corner, D>(box1), get<min_corner, 
D>(box2))
            || !math::equals(get<max_corner, D>(box1), get<max_corner, 
D>(box2)))

Sorry for the error. We didn't have a testcase for it until now.

>
> If I understand well, 2 geometries are not considered equal if 
> *max*(box1) <> *min*(box2) || *min*(box1) <> *max*(box2) ????

GGL assumes "correct" boxes, where min <= max. So no, the above is not 
considered equal (one is an 'inside-out' box).
(There CAN be a situation where min >= max, being still correct, that is 
in a spherical box (e.g. 170..-165) )

>
> I'm asking because of this code in ggl::index::rtree
>
>                 if (leaf->elements() < m_min_elems_per_node &&
>     elements() > m_min_elems_per_node)
>                 {
>                     q_leaves = leaf->get_leaves();
>
>                     // we remove the leaf_node in the parent node
>     because now it's empty
>                    
>     leaf->get_parent()->remove(leaf->get_parent()->get_box(leaf));
>                 }
>
>
> where the leaf is not actually removed...
>
Is probably fixed now by fixing the ggl::equals, just checked it in.

Thanks, Barend

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/ggl/attachments/20090711/ce92bf34/attachment.html