That's right; that's how you get the invocant in Perl.

(As a side note, beware of the indirect constructor call; my $object = Some::Class->new() is much more reliable than my $object = new Some::Class(). The latter is very sensitive to order of declaration during parsing; best to avoid it where possible.)

I would use Carp::confess in this case. And I probably wouldn't check if the method was overloaded on new(). Please consider the following example:

package MyPackage;

use strict;
use warnings;

use Carp qw(confess);

sub new {
    my ($class) = @_;
    my $self = {};
    bless $self, $class;
    return $self;
}

sub abstract_method {
    my ($self) = @_;

    my ($method) = ( caller(0) )[3];
    confess "${method} should be overloaded.";
}

package main;

my $p = MyPackage->new();
$p->abstract_method();
[download]

Don't you think it is cleaner?

I think the two primary pieces of information I would want in a "method should be overloaded" error message are the name of the method and the name of the class that didn't implement it. So I really think you need to report the latter (my $class= ref $self || $self;).

I'm not sure why you are writing code to look up the string "abstract_method" (perhaps so you can re-use this subroutine similar to *replace_me= \&abstrace_method;? -- no, doing so would still report "abstract_method" according to my testing, unfortunately). And, at least in this case, you don't even need to include the method name in the text since Carp::confess will already show that in the stack trace (though the error message is clearer if you do, of course). :)

- tye        

Actually the abstract_method name was used as example. Maybe was better if I used your_abstract_method_here or something.

Igor 'izut' Sutton
your code, your rules.