rtree-build — create a serialised R-tree from CSV rectangle data


rtree-build [-d dimension] [-f format] [-h ] [-n nodes] [-o file] [-s strategy] [-v ] [-V ] [file]


The rtree-build program reads CSV files specifying a collection of rectangles, cuboids or higher-dimensional hyperectangles and writes a serialised R-tree to output. Such R-trees can be loaded quickly into programs using librtree.


-d, --dimension dimension

Specify the dimension of the R-tree to build.

-f, --format format

Specify the format of the output, one of json or bsrt. See rtree-json(5) and rtree-bsrt(5) for descriptions of these formats.

If this option is not given, then json will be taken as the default.

-h, --help

List the program options.

-n, --nodes-per-page nodes

Specify the number of nodes (understood to include their branches) in a page of memory. This parameter can affect performance quite dramatically, a value of eight will result in a build which is better than twice as fast as a value of one (for dimension two on AMD64 at least). But a value which is too large will result in a branching factor less than two, which is infeasible for a tree: In this case the program will error.

A value of zero is permitted (and the default), in this case the program will choose a value which has been determined to work well. The value chosen will probably change between versions, so if stability of output is important to you, then some manual experimentation and a fixed value is advised.

-o, --output file

Write the output to file, rather than stdout.

-s, --split strategy

Specify the splitting strategy, one of linear, quadratic (the default) or greene.

The linear strategy is the fastest to build but will generally result in slower queries than the quadratic or greene strategies.

-v, --verbose

Verbose operation.

-V, --version

Version information.


J. J. Green, .