Boost logo

Geometry :

Subject: Re: [geometry] Index distance predicates names
From: Adam Wulkiewicz (adam.wulkiewicz_at_[hidden])
Date: 2012-09-04 14:44:49

Barend Gehrels wrote:
> Agree, near/far should then be avoided.
> I find the name "centroid" not so intuitive in this context. You mean
> here: "with_respect_to_its_centroid", or "relative_to_its_centroid".
> Maybe "centroid_relative"? or "furthest_point_measured"? or simply "to",
> so "to_nearest"?

To keep things short. I've changed:
near() to to_nearest()
centroid() to to_centroid()
far() to to_furthest()

So I guess that would be all for version 1 of the R-tree. Finally ;)
Code, tests and docs may be found here:


R-tree is a spatial index - k-dimensional searching accelerating data

Key features of this implementation:

1. Currently data is stored only in memory.
2. This implementation allows to use various balancing/creating
algorithms. Currently: linear, quadratic, R*-tree.
3. It is possible to pass various different spatial query predicates to
one spatial query. This makes query faster since the tree is traversed
only once and additional conditions narrows the search.
4. It is possible to perform k-nearest neighbors query, define how to
calculate distances, distances range and to pass additional spatial
predicates (pt. 4).


Geometry list run by mateusz at