The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.

NAME

GIS::Distance - Calculate geographic distances.

SYNOPSIS

    use GIS::Distance;
    
    # Use the GIS::Distance::Haversine formula by default.
    my $gis = GIS::Distance->new();
    
    # Or choose a different formula.
    my $gis = GIS::Distance->new( 'Polar' );
    
    # Returns a Class::Measure object.
    my $distance = $gis->distance( $lat1, $lon1, $lat2, $lon2 );
    
    print $distance->meters();

DESCRIPTION

This module calculates distances between geographic points on, at the moment, planet Earth. Various "FORMULAS" are available that provide different levels of accuracy versus speed.

GIS::Distance::Fast, a separate distribution, ships with C implementations of some of the formulas shipped with GIS::Distance. If you're looking for speed then install it and the ::Fast formulas will be automatically used by this module.

METHODS

distance

    my $distance = $gis->distance( $lat1, $lon1, $lat2, $lon2 );
    
    my $point1 = Geo::Point->latlong( $lat1, $lon1 );
    my $point2 = Geo::Point->latlong( $lat2, $lon2 );
    my $distance = $gis->distance( $point1, $point2 );

Takes either two decimal latitude and longitude pairs, or two Geo::Point objects.

Returns a Class::Measure::Length object for the distance between the two degree lats/lons.

See "distance_metal" for a faster, but less feature rich, method.

distance_metal

This works just like "distance" except for:

  • Does not accept Geo::Point objects. Only decimal latitude and longitude pairs.

  • Does not return a Class::Measure object. Instead kilometers are always returned.

  • Does no argument checking.

  • Does not support formula arguments which are supported by at least the GIS::Distance::GeoEllipsoid formula.

Calling this gets you pretty close to the fastest bare metal speed you can get. The speed improvements of calling this is noticeable over hundreds of thousands of iterations only and you've got to decide if its worth the safety and features you are dropping. Read more in the "SPEED" section.

ARGUMENTS

    my $gis = GIS::Distance->new( $formula );

When you call new() you may pass a partial or full formula class name as the first argument. The default is Haversive.

If you pass a partial name, as in:

    my $gis = GIS::Distance->new( 'Haversine' );

Then the following modules will be looked for in order:

    GIS::Distance::Fast::Haversine
    GIS::Distance::Haversine
    Haversine

Install GIS::Distance::Fast to get access to the Fast:: (XS) implementations of the formula classes.

You may globally disable the automatic use of the Fast:: formulas by setting the GIS_DISTANCE_PP environment variable. Although, its likely simpler to just provide a full class name to get the same effect:

    my $gis = GIS::Distance->new( 'GIS::Distance::Haversine' );

SPEED

Not that this module is slow, but if you're doing millions of distance calculations a second you may find that adjusting your code a bit may make it faster. Here are some options.

Install GIS::Distance::Fast to get the XS variants for most of the PP formulas.

Use "distance_metal" instead of "distance".

Call the undocumented _distance() function that each formula class has. For example you could bypass this module entirely and just do:

    use GIS::Distance::Fast::Haversine;
    my $km = GIS::Distance::Fast::Haversine::_distance( @coords );

The above would be the ultimate speed demon (as shown in benchmarking) but throws away some flexibility and adds some foot-gun support.

Here's a benchmarks for these options:

    2019-03-13T09:34:00Z
    GIS::Distance 0.15
    GIS::Distance::Fast 0.12
    GIS::Distance::Fast::Haversine 0.12
    GIS::Distance::Haversine 0.15
                                                                 Rate
    PP Haversine - GIS::Distance->distance                   123213/s
    XS Haversine - GIS::Distance->distance                   196232/s
    PP Haversine - GIS::Distance->distance_metal             356379/s
    PP Haversine - GIS::Distance::Haversine::_distance       385208/s
    XS Haversine - GIS::Distance->distance_metal             3205128/s
    XS Haversine - GIS::Distance::Fast::Haversine::_distance 8620690/s

You can run your own benchmarks using the included author/bench script. The above results were produced with:

    author/bench -f Haversine

The slowest result was about 125000/s, or about 8ms each call. This could be a substantial burden in some contexts, such as live HTTP responses to human users and running large batch jobs, to name just two.

In conclusion, if you can justify the speed gain, switching to "distance_metal" and installing GIS::Distance::Fast looks to be an ideal setup.

As always with performance and benchmarking, YMMV.

COORDINATES

When passing latitudinal and longitudinal coordinates to "distance" they must always be in decimal degree format. Here is some sample code for converting from other formats to decimal:

    # DMS to Decimal
    my $decimal = $degrees + ($minutes/60) + ($seconds/3600);
    
    # Precision Six Integer to Decimal
    my $decimal = $integer * .000001;

If you want to convert from decimal radians to degrees you can use Math::Trig's rad2deg function.

FORMULAS

These formulas come bundled with this distribution:

These formulas are available on CPAN:

AUTHORING

Take a look at GIS::Distance::Formula for instructions on authoring new formula classes.

SEE ALSO

SUPPORT

Please submit bugs and feature requests to the GIS-Distance GitHub issue tracker:

https://github.com/bluefeet/GIS-Distance/issues

AUTHOR

    Aran Clary Deltac <bluefeet@gmail.com>

CONTRIBUTORS

    Mohammad S Anwar <mohammad.anwar@yahoo.com>
    Florian Schlichting <fsfs@debian.org>

LICENSE

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.