Do you?

use strict;
use Devel::Size qw(total_size);

my %hash = ( 0 => 3,
             1 => 7,
             2 => 5,
             3 => 0,
             4 => 2,
           );

my @vals = qw(3 7 5 0 2);

print 'T1: ', total_size(\%hash),"\n";
print 'T2: ', total_size(\@vals),"\n";
[download]

30% less space for the array, same access speed. There need to be as many of these data structures in memory as possible. qed

And hashes can be more efficient if there are holes in the index.

use strict;
use Devel::Size qw(total_size);

my %hash = ( 0 => 3,
             1 => 7,
             2 => 5,
             3 => 0,
             20202 => 2,
           );

my @vals = qw(3 7 5 0);
$vals[20202] = 2;

print 'T1: ', total_size(\%hash),"\n";
print 'T2: ', total_size(\@vals),"\n";

[waswas:/var/tmp] waswas% perl hah
T1: 291
T2: 131228
[download]



-Waswas

PetaMem,
Yeah. I know they take up more space and can be slower than arrays WRT accessing. The work involved in making an array work when a hash makes natural sense almost always outweighs the benefit gained by not using the hash.

Let me put it another way. When you need to access something by key then a hash is almost always the way to go. I feel that with the amount of information originally given, it was not unexpected to question the motives of using an array over a hash. I am not saying there is never a reason not to use a hash - it is just good to get a sanity check first.

Cheers - L~R