Consider the following:

use Benchmark;

my $foo = "bar";
my $test = new tester();

my ($buffer, $i);

my $a1 = new Benchmark;
for( $i=1; $i<1_000_000; $i++ ) {
    $buffer = $foo;
}
my $a2 = new Benchmark;


my $b1 = new Benchmark;
for( $i=1; $i<1_000_000; $i++ ) {
    $buffer = $test->report();
}
my $b2 = new Benchmark;

my $diff1 = timediff($a1, $a2);
my $diff2 = timediff($b1, $b2);

print "Variable copying: " . timestr($diff1) . "\n";
print "Object copying: " . timestr($diff2) . "\n";


package tester;

sub new {
    my $class = shift;

    my $self = { 'foo' => "bar" };

    bless $self, $class;

    return $self;

}

sub report {
    my $self = shift;

    return $self->{'foo'};

}
[download]

Running this gives (on a 1ghz pc hobbled by Win2k):

Variable copying: -1 wallclock secs (-0.81 usr +  0.00 sys = -0.81 CPU
+)
Object copying: -5 wallclock secs (-5.15 usr +  0.00 sys = -5.15 CPU)
[download]

Considering there are a million copies made of each the time difference isnt all that great. However, if all we wanted to do was code fast programs none of us would ever write OO :)

In other words, use a variable if calling the object is too much work :)

However there are many other reasons to code this way... have a look at This file and especially this file for some basic guidelines and explanation of OO programming.