| Subject: | great_circle_direction gives wrong answer if 5th argument is supplied |
Distribution: Math-Complex-1.54
Module: Math-Trig 1.18
Perl Version: 5.8.8
Operation System: Linux 2.6.20-16-server (Ubuntu 7.04)
Math::Trig::great_circle_direction will give a wrong answer if the
caller accidentally supplies a non-null value as a 5th argument to the
subroutine call.
Example:
###########################################################
!/usr/bin/perl
use strict;
use Math::Trig qw(great_circle_direction pi);
print great_circle_direction(0, pi/2, 0, pi/4) . "\n";
print great_circle_direction(0, pi/2, 0, pi/4, 2) . "\n";
###########################################################
Results:
0 - correct
0.785398163397448 - incorrect
This can be the cause of obscure bugs in the caller's code -
particularly because when the value of the 5th argument is close to 1,
the error in the value returned is correspondingly small.
The bug occurs because on line 169 the 5th argument is accidentally
forwarded to &great_circle_distance:
166: sub great_circle_direction {
167: my ( $theta0, $phi0, $theta1, $phi1 ) = @_;
168:
169: my $distance = &great_circle_distance;
This could be fixed by passing an explicit argument list to
&great_circle_distance:
169: my $distance = great_circle_distance( $theta0, $phi0, $theta1,
$phi1 );
or
169: my $distance = great_circle_distance(@_[0..3]);