To expand on merlyn's excellent <g> suggestion, the following is an example of using Algorithm::Diff for comparing paths. Of course, I'm not sure that this is what you mean by "compare", but since you show the use of diff, perhaps it is:

#!/usr/bin/perl -w
use strict;
use Algorithm::Diff;

die "usage: $0 path1 path2\n" unless $#ARGV == 1;
my @path1 = split('/', $ARGV[0]);
my @path2 = split('/', $ARGV[1]);

sub match { print "same:\t$path1[$_[0]]\t$path2[$_[1]]\n"; }

sub discA { print "omitA:\t$path1[$_[0]]\n"; }

sub discB { print "omitB:\t\t$path2[$_[1]]\n"; }

Algorithm::Diff::traverse_sequences( \@path1, \@path2,
 { MATCH => \&match, DISCARD_A => \&discA, DISCARD_B => \&discB });
[download]