Vanity, Vanity, all is Vanity

Postscript: In 2005 I gave this class at OSCON
An audience member interrupted to say he had found an obvious way to speed up letter_histogram
- He had benchmarked it and found it substantially faster
His benchmark looked something like this:

        use Benchmark;
        my $t = "some reasonably long string here";
        timethese(-5, { orig => 'orig_letter_histogram($t)' ,
                        mine => 'my_letter_histogram($t)' ,
                      });

        sub orig_letter_histogram {
          my $strdex = (length $_[0])-1;
          $letter_hist{substr($_[0],$_,1)}++     for (0..$strdex);
        }
        sub my_letter_histogram { 
          $letter_hist{$1}++ while $_[0] =~ m/(.)//gs;
        }

The following week, I did it right
His suggestion is 250% slower:

          orig histo 11.42  0.03  11.45
          while //gs 37.28  0.03  37.31
                NULL  0.04  0.00   0.04