Here's what I do in all my scripts (for example, from Krang):

use File::Spec::Functions qw(catdir splitdir canonpath);
use FindBin qw($RealBin);
use Config;

BEGIN {
    # Find a KRANG_ROOT based on path to bin
    my @dir = splitdir(canonpath($RealBin));
    $ENV{KRANG_ROOT} ||= catdir(@dir[0 .. $#dir - 1]);

    # use $KRANG_ROOT/lib for modules
    my $lib        = catdir($ENV{KRANG_ROOT}, "lib");
    $ENV{PERL5LIB} = $ENV{PERL5LIB} ? "$ENV{PERL5LIB}:${lib}" : $lib;
    unshift @INC, $lib, "$lib/".$Config{archname};
}
[download]

This uses the location of the script to find a "lib" dir in the parent directory. It then does the equivalent of a "use lib" as well as setting up PERL5LIB so that any sub-processes will use the same modules. You may not need the PERL5LIB setup if you aren't going to run any sub-processes that need to find Perl modules.

This is better than putting a fully-qualified path in your script, in my opinion, because you can move your whole tree around and not change anything. This is particularly useful if you're working with other developers.

-sam