From: www.itworld.com
June 20, 2001 —
Last week, we started our Slot module and defined a constructor. We
tested it by creating a new object and accessing its "credits" key,
which we said was a Bad Thing(TM). Let's add an object routine to our
module that returns the current value of the "credits" attribute (this
is called an accessor method):
sub credits {
my $self = shift;
return $self->{credits};
}
Accessing the data structure directly inside the object module is OK,
but the outside world should get its information from object methods.
This method shifts off the first argument, which will be the object
itself (a reference to the hash) and returns the value of the "credits"
key in that hash. That's it. We can now modify the little test script
from last week to use this method rather than accessing the value
directly:
#!/usr/bin/perl -w
use strict;
use Slot;
my $slot = Slot->new(100);
print $slot->credits(),"\n";
Why is this better than directly accessing the hash key itself?
Because, as a user of this module, we aren't supposed to know (or care)
that the object is a hash. It could have been an array reference, or it
could be in the next release. If we wrote code that accessed the object
as a hash, and then we downloaded the latest version and the author
changed it to use an array reference underneath, then all of our
current code would break. Just to show you, let's rebuild the module
using an array:
package Slot;
use strict;
sub new {
my $class = shift;
my $credits = shift || 100;
my $self = [ $credits, # credits
1, # bet
0, # win
0, # paid
undef, # spin
];
return bless $self, $class;
}
sub credits {
my $self = shift;
return $self->[0];
}
1;
__END__
Maybe the module author decided that array accesses are faster (not
enough to warrant using them in this case for sure, but it's an
example), and changed the module to this version. Our new code that
uses the credits() method still works just fine, but our old code that
used $slot->{credits} won't work at all now.
And that's a very important point about objects -- the object
encapsulates the data and provides an interface for us to use. Just
like we don't need to know how a real slot machine (or radio) works
inside, we shouldn't need to know how our object is built on the inside
(unless we are the ones building it of course). That means, as an
object builder, you need to provide any methods the user might need to
operate the object. This means you will be free in the future to change
the underlying object and the user never needs to know and doesn't have
to worry about a new version breaking old programs that use it.
Let's quickly add another attribute method, the bet() method:
sub bet {
my $self = shift;
$self->{bet} = shift || $self->{bet};
}
This is a dual purpose method --- we not only have a way to retrieve
the current betting amount, we can also set it at the same time. This
is a useful way of creating accessor methods instead of creating
separate get_bet() and set_bet() methods. If an argument is passed to
this method and it is not 0, we set the bet attribute to the new value
and return it, otherwise we use the old value. We haven't done any
error checking to ensure an integer was passed in, that is an exercise
for the reader.
We haven't yet actually defined how our slot machine will work, and
that will be the subject of next week's discussion.
Next Week: Constructing an Object (class data and more methods)
ITworld