Moose::Manual::Classes - Making your classes use Moose (and subclassing)
Using Moose is very simple,
package Person; use Moose;
That's it, you've made a class with Moose!
There's actually a lot going on here under the hood, so let's step through it.
When you load Moose, a bunch of sugar functions are exported into your class. These include things like
with, and more. These functions are what you use to define your class. For example, you might define an attribute ...
package Person; use Moose; has 'ssn' => ( is => 'rw' );
Attributes are described in the Moose::Manual::Attributes documentation.
Loading Moose also enables
warnings pragmas in your class.
When you load Moose, your class will become a subclass of Moose::Object. The Moose::Object class provides a default constructor and destructor, as well as object construction helper methods. You can read more about this in the Moose::Manual::Construction document.
As a convenience, Moose creates a new class type for your class. See the Moose::Manual::Types document to learn more about types.
It also creates a Moose::Meta::Class object for your class. This metaclass object is now available by calling a
meta method on your class, for example
The metaclass object provides an introspection API for your class. It is also used by Moose itself under the hood to add attributes, define parent classes, and so on. In fact, all of Moose's sugar does the real work by calling methods on this metaclass object (and other meta API objects).
Moose provides a simple sugar function for declaring your parent classes,
package User; use Moose; extends 'Person'; has 'username' => ( is => 'rw' );
Note that each call to
extends will reset your parents. For multiple inheritance you must provide all the parents at once,
extends 'Foo', 'Bar'.
You can use Moose to extend a non-Moose parent. However, when you do this, you will inherit the parent class's constructor (assuming it is also called
new). In that case, you will have to take care of initializing attributes manually, either in the parent's constructor, or in your subclass, and you will lose a lot of Moose magic.
Moose also allows you to remove its sugar functions from your class's namespace. We recommend that you take advantage of this feature, since it just makes your classes "cleaner". You can do this by simply adding
no Moose at the end of your module file.
package Person; use Moose; has 'ssn' => ( is => 'rw' ); no Moose;
This deletes Moose's sugar functions from your class's namespace, so that
Person->can('has') will no longer return true.
Moose has a feature called "immutabilization" that you can use to greatly speed up your classes at runtime. However, using it does incur a cost when your class is first being loaded. When you make your class immutable you tell Moose that you will not be changing it in the future. You will not be adding any more attributes, methods, roles, etc.
This allows Moose to generate code specific to your class. In particular, it creates an "inline" constructor, making object construction much faster.
To make your class immutable you simply call
make_immutable on your class's metaclass object.
If you override
new() in your class, then the immutabilization code will not be able to provide an optimized constructor for your class. Instead, you should use a
BUILD() method, which will be called from the inlined constructor.
Alternately, if you really need to provide a different
new(), you can also provide your own immutabilization method. Doing so requires extending the Moose metaclasses, and is well beyond the scope of this manual.
Dave Rolsky <email@example.com>
Copyright 2008-2009 by Infinity Interactive, Inc.