1+1=0

Here's Jon's benchmark:

     use Benchmark;

     sub one_by_one {
         my (@c);
         for (my $i; $i < 100000; $i++) {
             $c[$i] = rand;
         }
     }

     sub preallocate {
         my (@c);
         $#c = 99999;
         for (my $i; $i < 100000; $i++) {
             $c[$i] = rand;
         }
     }

     timethese (100, {    'preallocate' => 'preallocate()',
                          'one_by_one'  => 'one_by_one()'
                     } );

     Benchmark: timing 100 iterations of one_by_one, preallocate...
     one_by_one: 111 secs (50.85 usr  0.52 sys = 51.37 cpu)
     preallocate: 148 secs (67.13 usr  0.57 sys = 67.70 cpu)