in reply to RFC: Practical closure example: Spreadsheet writer

Nice node, but why closures rather than light weight OO? Consider an OO variant of your first example:

use strict;
use warnings;

package Fruit;

sub new {
    my ($class, $fruit) = @_;
    
    return bless {total => 0, fruit => $fruit}, $class;
}

sub Add {
   my ($self, $number) = @_;
   $self->{total} = $self->{total} + $number;
   return "$self->{total} $self->{fruit}" . ($self->{total} != 1 ? 's'
+ : '');
};

package main;

my $apple = Fruit->new ('apple');
my $orange = Fruit->new ('orange');

print $apple->Add (1)."\n";
print $apple->Add (3)."\n";
print $orange->Add (1)."\n";
print $apple->Add (5)."\n";
print $orange->Add (1)."\n";

[download]

Prints:

1 apple
4 apples
1 orange
9 apples
2 oranges

[download]

Ok, a little more verbose, but now you can add more functionality to your fruit object by adding member functions or sub-classing Fruit.

DWIM is Perl's answer to Gödel

Replies are listed 'Best First'.
Re^2: RFC: Practical closure example: Spreadsheet writer
by clinton (Priest) on Aug 05, 2007 at 00:00 UTC
    Good point, and actually, while I was typing the node, I kept on writing $writer->write($value) rather than $writer->('write',$value). This made me ask myself the same question: why not OO?

    And really, no reason - I would say (and I stand under correction) that really it is just a question of scale. A small job may be more easily (less verbosely) achieved with closures, while OO gives you more room to expand. Also, I would guess that using references to anonymous subs avoids most of the penalty hit of using OO (micro-optimization anyone?).

    So I would say that it is probably a different technique to achieve the same thing.

    I'd be interested in differing opinions though

    Clint

[reply]
[d/l]
[select]
      Also, I would guess that using references to anonymous subs avoids most of the penalty hit of using OO (micro-optimization anyone?).

      There may be a tiny benefit from avoiding potential dispatch lookup, but the benefit is fairly small. If anything, you're more likely to see memory savings (in the order of a few kilobytes). (Closures save the same CV but use different lexpads. If you reuse the closure, all references share the CV rather than constructing their own.)

[reply]

In Section Meditations