NAME

MooseX::Interface - Java-style interfaces for Moose

SYNOPSIS

package DatabaseAPI::ReadOnly
{
  use MooseX::Interface;
  requires 'select';
  one;
}

package DatabaseAPI::ReadWrite
{
  use MooseX::Interface;
  extends 'DatabaseAPI::ReadOnly';
  requires 'insert';
  requires 'update';
  requires 'delete';
  one;
}

package Database::MySQL
{
  use Moose;
  with 'DatabaseAPI::ReadWrite';
  sub insert { ... }
  sub select { ... }
  sub update { ... }
  sub delete { ... }
}

Database::MySQL::->DOES('DatabaseAPI::ReadOnly');   # true
Database::MySQL::->DOES('DatabaseAPI::ReadWrite');  # true

DESCRIPTION

MooseX::Interface provides something similar to the concept of interfaces as found in many object-oriented programming languages like Java and PHP.

"What?!" I hear you cry, "can't this already be done in Moose using roles?"

Indeed it can, and that's precisely how MooseX::Interface works. Interfaces are just roles with a few additional restrictions:

• You may not define any methods within an interface, except:

  • Moose's built-in meta method, which will be defined for you;

  • You may override methods from UNIVERSAL; and

  • You may define constants using the constant pragma.

• You may not define any attributes. (Attributes generate methods.)

• You may not define method modifiers.

• You can extend other interfaces, not normal roles.

Functions

extends $interface

Extends an existing interface.

Yes, the terminology "extends" is used rather than "with".

excludes $role

Prevents classes that implement this interface from also composing with this role.

requires $method

The name of a method (or attribute) that any classes implementing this interface must provide.

requires $method => \@signature

Declares a signature for the given method. This effectively creates an around method modifier for the method to check the signature.

As an example:

requires log_message => [qw( Str )];

If the log_message method above were called with multiple arguments, then the additional arguments would be tolerated; the only check is that the first argument is a string.

const $name => $value

Experimental syntactic sugar for declaring constants. It's probably not a good idea to use this yet.

test_case { BLOCK } $name

Experimental syntactic sugar for embedded test cases. This extends the idea that an interface is a contract for classes to fulfil.

The block will be called with an instance of a class claiming to implement the interface in $_ and should return true if the instance passes the test and false if it fails.

package CalculatorAPI
{
  use MooseX::Interface;
  
  requires 'add';
  test_case { $_->add(8, 2) == 10 };
  
  requires 'subtract';
  test_case { $_->subtract(8, 2) == 6 };
  
  requires 'multiply';
  test_case { $_->multiply(8, 2) == 16 };
  
  requires 'divide';
  test_case { $_->divide(8, 2) == 4 };
}

package Calculator
{
  use Moose;
  with 'CalculatorAPI';
  sub add      { $_[1] + $_[2] }
  sub subtract { $_[1] - $_[2] }
  sub multiply { $_[1] * $_[2] }
  sub divide   { $_[1] / $_[2] }
}

my $result = CalculatorAPI->meta->test_implementation(
  Calculator->new,
);

The result of test_implementation is an overloaded object which indicates success when evaluated in boolean context; indicates the number of failures in numeric context; and provides TAP-like "ok" or "not ok" in string context. You can call methods passed and failed on this object to return arrayrefs of failed test cases. Each test case is itself an object, with name, code and associated_interface attributes.

Do not rely on test cases being run in any particular order, or maintaining any state between test cases. (Theoretically each test case could be run with a separate instance of the implementing class.)

one

This function checks the integrity of your role, making sure it doesn't do anything that interfaces are not supposed to do, like defining methods.

While you don't need to call this function at all, your interface's integrity will get checked anyway when classes implement the interface, so calling one will help you catch potential problems sooner. one helpfully returns '1', so it can be used as the magical return value at the end of a Perl module.

(Backwards compatibility note: in MooseX::Interface versions 0.005 and below, this was performed automatically using Hook::AfterRuntime. From 0.006, the one function was introduced instead.)

BUGS

Please report any bugs to http://rt.cpan.org/Dist/Display.html?Queue=MooseX-Interface.

SEE ALSO

MooseX::Interface::Tutorial, MooseX::Interface::Internals.

Moose::Role, MooseX::ABCD.

AUTHOR

Toby Inkster <tobyink@cpan.org>.

COPYRIGHT AND LICENCE

This software is copyright (c) 2012 by Toby Inkster.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.

DISCLAIMER OF WARRANTIES

THIS PACKAGE IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE.