I have a class with an attribute set up as follows:
has _data => (
is => 'ro',
lazy => 1,
builder => '_load',
);
sub _load {
my $self = shift;
return retrieve $self->_file;
}
However I now want to call a method already defined on the class before returning the data.
In old-school Perl OO, I'd be doing something like this:
sub _load {
# Assuming laziness is implemented somewhere else.
my $self = shift;
$self->{_data} = retrieve $self->_file;
$self->refresh; # which does something with $self->{_data}
return $self->{_data};
}
But I can't figure out a 'clean' way to do this in Moose.
I've considered the following, but think they are quite ugly, and that there must be a better way of doing this.
_data
read-write, I could potentially write the data to the accessor, call the method then return the value from the accessor for Moose to write back to the accessor._raw_data
, store the data in there, modify refresh()
to use that attribute, and everything else uses _data()
. $self->{_data}
directly.I tried an after '_load' => \&refresh;
, but that just created an endless loop.
This would be a nice use of triggers:
has _data => (
is => 'ro',
lazy => 1,
builder => '_load',
trigger => sub { shift->refresh },
);
Except that triggers don't work on default/built values - only on values passed to the constructor explicitly, or passed to a writer/accessor method. Sad face. :-(
One solution would be to rewrite your refresh
method so that instead of operating on $self->_data
, it can accept a parameter (perhaps falling back to operating on $self->_data
if no parameter is given.
sub _load {
my $self = shift;
my $tmp = retrieve $self->_file;
$self->refresh($tmp);
return $tmp;
}
sub refresh {
my $self = shift;
my $data = scalar(@_) ? $_[0] : $self->_data;
# do something with $data
}