Actually it makes it ever so slightly slower, much to my surprise :-/

use Benchmark qw(cmpthese);

sub substrings_TheHobbit {
  my $string = shift;
  my @result = ();

  foreach my $length (1..length($string)) {
    foreach my $offset (0..length($string)-$length) {
      push @result,substr($string,$offset,$length);
    }
  }
  return @result;
}

sub substrings_thelenm {
  my $string = shift;
  my @result = ();
  my $strlen = length $string;

  foreach my $length (1..$strlen) {
    foreach my $offset (0..$strlen-$length) {
      push @result, substr($string,$offset,$length);
    }
  }
  return @result;
}


cmpthese(-10, {
    TheHobbit   => sub { substrings_TheHobbit("Just Another Perl Hacke
+r,") },
    thelenm     => sub { substrings_thelenm("Just Another Perl Hacker,
+") },
});

__output__

Benchmark: running TheHobbit, thelenm, each for at least 10 CPU second
+s...
 TheHobbit: 12 wallclock secs (10.51 usr +  0.03 sys = 10.54 CPU) @ 80
+4.55/s (n=8480)
   thelenm: 14 wallclock secs (10.44 usr +  0.02 sys = 10.46 CPU) @ 80
+9.37/s (n=8466)
           Rate TheHobbit   thelenm
TheHobbit 805/s        --       -1%
thelenm   809/s        1%        --
[download]

Is there any reason why a function call would be quicker accessing a lexical variable?
_________
broquaint