I'm partial to array backed objects where applicable. But I've never actually tried a blessed code ref... until 5 minutes ago after reading your reply! It might be a good solution to my application.

Out of interest, do you have insight on how perl knows to call the underlying code ref and not a method that may or may not be in the package? An extra lookup/check must be done when called?

#eg
$code_ref_obj->();#calls code ref directly
$code_ref_obj->a_method(); #Call a method on the object
[download]

> do you have insight on how perl knows to call the underlying code ref

I'm not sure I understand, your demo nailed it.

• $code_ref->(ARGS) without method name is de-referencing the code-ref.
• $obj->method_name(ARGS) with method name is looking up the "blessed" package inside the structure of the scalar and will try to resolve the method there.

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery}

My question wasn't very directional... I was wanting to know the relative performance of a blessed code ref vs a unblessed code ref

So I put together a little benchmark comparing:

1. calling a code ref directly
2. blessing a code ref and calling it directly
3. blessing a code ref and calling another method in the object

Blessing the code ref does have a performance impact when using it as a code ref compared to an unblessed code ref. However it is cheap compared to calling a method on the object

use strict;
use warnings;
use feature ":all";

use Benchmark qw<cmpthese>;

my $count=$ARGV[0]//1;

package SUB_TEST {
    sub new { bless sub {1}, __PACKAGE__}
    sub method { 1 }
}

my $obj=SUB_TEST->new();

my $sub=sub {1};

cmpthese($count, 
        { 
            blessed=>sub {$obj->()},
            method=>sub{$obj->method()},
            sub=>sub{$sub->()}
        }
    );
[download]

Output:

                 Rate             method    blessed     sub
method  18939394/s      --           -30%     -39%
blessed 27173913/s      43%            --        -12%
sub        30864198/s     63%          14%       --
[download]