#!/usr/bin/perl use strict; use warnings; use Test::More tests => 3; BEGIN { use_ok('Class::C3'); # uncomment this line, and re-run the # test to see the normal p5 dispatch order #$Class::C3::TURN_OFF_C3 = 1; } =pod This tests a strange bug found by Matt S. Trout while building DBIx::Class. Thanks Matt!!!! / \ \ / =cut { package Diamond_A; use Class::C3; sub foo { 'Diamond_A::foo' } } { package Diamond_B; use base 'Diamond_A'; use Class::C3; sub foo { 'Diamond_B::foo => ' . (shift)->next::method } } { package Diamond_C; use Class::C3; use base 'Diamond_A'; } { package Diamond_D; use base ('Diamond_C', 'Diamond_B'); use Class::C3; sub foo { 'Diamond_D::foo => ' . (shift)->next::method } } Class::C3::initialize(); is_deeply( [ Class::C3::calculateMRO('Diamond_D') ], [ qw(Diamond_D Diamond_C Diamond_B Diamond_A) ], '... got the right MRO for Diamond_D'); is(Diamond_D->foo, 'Diamond_D::foo => Diamond_B::foo => Diamond_A::foo', '... got the right next::method dispatch path');