This is a post about Moose the post-modern Object-Oriented Programming system for Perl 5. Until a few days ago, I've learnt most of what I knew about Moose by reading some presentations about it and experimenting. However, a few days ago, I've decided that maybe there was a more-Moosey way of doing the following pattern that I've been doing in my classes:
# Super-base class.
sub new
{
my $class = shift;
my $self = {};
bless $self, $class;
$self->_init(@_);
return $self;
}
# In derived classes:
sub _init
{
my $self = shift;
my $args = shift;
$self->_init($args);
# Initialise from $args.
return;
}
Turns out there was - you can use the BUILD method to provide initialisation for each part of the class hierarchy (it gets calls in a walk-method style - not only for the most qualified sub-class). Only in my case in my en-Moosification of XML-Grammar-Fiction, I only needed a 'default' callback to the class members, and to make them 'lazy' by default.
This made me conclude that I have to learn Moose top-down to become more familiar with it. So I started reading the chapters of Moose::Manual and have already learned that I could encapsulate push() and shift() out of an array ref using its delegation, which also proved useful in XML-Grammar-Fiction. After I'm done with the manual, I'm planning to go over the Moose cookbook.
Hopefully, this will give me a lot of ways to avoid repetitive code in my CPAN modules. Of course, now, I'm facing the classic dilemma of whether to make something a base class or a role… decisions, decisions.
