kd

Class::Method::Modifiers versus $self->SUPER.

The exec summary is:

$ dumbbench perl test1.pl 
cmd: Ran 21 iterations (1 outliers).
cmd: Rounded run time per iteration: 2.0408e-02 +/- 3.6e-05 (0.2%)
$ dumbbench perl test2.pl 
cmd: Ran 22 iterations (2 outliers).
cmd: Rounded run time per iteration: 1.5050e-02 +/- 6.9e-05 (0.5%)

So Class::Method::Modifiers seems 30% faster. I'm guessing that this is the result of not having the subroutine overhead call on every invocation. Am I right? How does this work?

Here's the code:

• test1.pl

  package One;
  sub blah {
      my ($self) = @_;
      return 1;
  }
  
  
  package Two;
  use base qw/One/;
  
  
  sub blah {
      my ($self) = @_;
      my $orig = $self->SUPER::blah(@_);
      return ++$orig;
  }
  
  
  package main;
  for (1 .. 10000) {
      Two->blah;
  }
  

• test2.pl

  use warnings;
  use strict;
  
  
  package One;
  sub blah {
      my ($self) = @_;
      return 1;
  }
  
  
  package Two;
  use base qw/One/;
  use Class::Method::Modifiers;
  
  
  around  blah => sub  {
      my ($orig) = shift;
      my ($self) = @_;
      my $ret = $orig->($self, @_);
      return ++$ret;
  };
  
  
  package main;
  for (1 .. 10000) {
      Two->blah;
  }