3 perlboot - Beginner's Object-Oriented Tutorial
7 If you're not familiar with objects from other languages, some of the
8 other Perl object documentation may be a little daunting, such as
9 L<perlobj>, a basic reference in using objects, and L<perltoot>, which
10 introduces readers to the peculiarities of Perl's object system in a
13 So, let's take a different approach, presuming no prior object
14 experience. It helps if you know about subroutines (L<perlsub>),
15 references (L<perlref> et. seq.), and packages (L<perlmod>), so become
16 familiar with those first if you haven't already.
18 =head2 If we could talk to the animals...
20 Let's let the animals talk for a moment:
23 print "a Cow goes moooo!\n";
26 print "a Horse goes neigh!\n";
29 print "a Sheep goes baaaah!\n";
42 Nothing spectacular here. Simple subroutines, albeit from separate
43 packages, and called using the full package name. So let's create
46 # Cow::speak, Horse::speak, Sheep::speak as before
47 @pasture = qw(Cow Cow Horse Sheep Sheep);
48 foreach $animal (@pasture) {
60 Wow. That symbolic coderef de-referencing there is pretty nasty.
61 We're counting on C<no strict refs> mode, certainly not recommended
62 for larger programs. And why was that necessary? Because the name of
63 the package seems to be inseparable from the name of the subroutine we
64 want to invoke within that package.
68 =head2 Introducing the method invocation arrow
70 For now, let's say that C<< Class->method >> invokes subroutine
71 C<method> in package C<Class>. (Here, "Class" is used in its
72 "category" meaning, not its "scholastic" meaning.) That's not
73 completely accurate, but we'll do this one step at a time. Now let's
76 # Cow::speak, Horse::speak, Sheep::speak as before
81 And once again, this results in:
87 That's not fun yet. Same number of characters, all constant, no
88 variables. But yet, the parts are separable now. Watch:
91 $a->speak; # invokes Cow->speak
93 Ahh! Now that the package name has been parted from the subroutine
94 name, we can use a variable package name. And this time, we've got
95 something that works even when C<use strict refs> is enabled.
97 =head2 Invoking a barnyard
99 Let's take that new arrow invocation and put it back in the barnyard
103 print "a Cow goes moooo!\n";
106 print "a Horse goes neigh!\n";
109 print "a Sheep goes baaaah!\n";
112 @pasture = qw(Cow Cow Horse Sheep Sheep);
113 foreach $animal (@pasture) {
117 There! Now we have the animals all talking, and safely at that,
118 without the use of symbolic coderefs.
120 But look at all that common code. Each of the C<speak> routines has a
121 similar structure: a C<print> operator and a string that contains
122 common text, except for two of the words. It'd be nice if we could
123 factor out the commonality, in case we decide later to change it all
124 to C<says> instead of C<goes>.
126 And we actually have a way of doing that without much fuss, but we
127 have to hear a bit more about what the method invocation arrow is
128 actually doing for us.
130 =head2 The extra parameter of method invocation
136 attempts to invoke subroutine C<Class::method> as:
138 Class::method("Class", @args);
140 (If the subroutine can't be found, "inheritance" kicks in, but we'll
141 get to that later.) This means that we get the class name as the
142 first parameter (the only parameter, if no arguments are given). So
143 we can rewrite the C<Sheep> speaking subroutine as:
147 print "a $class goes baaaah!\n";
150 And the other two animals come out similarly:
154 print "a $class goes moooo!\n";
158 print "a $class goes neigh!\n";
161 In each case, C<$class> will get the value appropriate for that
162 subroutine. But once again, we have a lot of similar structure. Can
163 we factor that out even further? Yes, by calling another method in
166 =head2 Calling a second method to simplify things
168 Let's call out from C<speak> to a helper method called C<sound>.
169 This method provides the constant text for the sound itself.
172 sub sound { "moooo" }
175 print "a $class goes ", $class->sound, "!\n";
179 Now, when we call C<< Cow->speak >>, we get a C<$class> of C<Cow> in
180 C<speak>. This in turn selects the C<< Cow->sound >> method, which
181 returns C<moooo>. But how different would this be for the C<Horse>?
184 sub sound { "neigh" }
187 print "a $class goes ", $class->sound, "!\n";
191 Only the name of the package and the specific sound change. So can we
192 somehow share the definition for C<speak> between the Cow and the
193 Horse? Yes, with inheritance!
195 =head2 Inheriting the windpipes
197 We'll define a common subroutine package called C<Animal>, with the
198 definition for C<speak>:
203 print "a $class goes ", $class->sound, "!\n";
207 Then, for each animal, we say it "inherits" from C<Animal>, along
208 with the animal-specific sound:
212 sub sound { "moooo" }
215 Note the added C<@ISA> array (pronounced "is a"). We'll get to that in a minute.
217 But what happens when we invoke C<< Cow->speak >> now?
219 First, Perl constructs the argument list. In this case, it's just
220 C<Cow>. Then Perl looks for C<Cow::speak>. But that's not there, so
221 Perl checks for the inheritance array C<@Cow::ISA>. It's there,
222 and contains the single name C<Animal>.
224 Perl next checks for C<speak> inside C<Animal> instead, as in
225 C<Animal::speak>. And that's found, so Perl invokes that subroutine
226 with the already frozen argument list.
228 Inside the C<Animal::speak> subroutine, C<$class> becomes C<Cow> (the
229 first argument). So when we get to the step of invoking
230 C<< $class->sound >>, it'll be looking for C<< Cow->sound >>, which
231 gets it on the first try without looking at C<@ISA>. Success!
233 =head2 A few notes about @ISA
235 This magical C<@ISA> variable has declared that C<Cow> "is a" C<Animal>.
236 Note that it's an array, not a simple single value, because on rare
237 occasions, it makes sense to have more than one parent class searched
238 for the missing methods.
240 If C<Animal> also had an C<@ISA>, then we'd check there too. The
241 search is recursive, depth-first, left-to-right in each C<@ISA> by
242 default (see L<mro> for alternatives). Typically, each C<@ISA> has
243 only one element (multiple elements means multiple inheritance and
244 multiple headaches), so we get a nice tree of inheritance.
246 When we turn on C<use strict>, we'll get complaints on C<@ISA>, since
247 it's not a variable containing an explicit package name, nor is it a
248 lexical ("my") variable. We can't make it a lexical variable though
249 (it has to belong to the package to be found by the inheritance mechanism),
250 so there's a couple of straightforward ways to handle that.
252 The easiest is to just spell the package name out:
254 @Cow::ISA = qw(Animal);
256 Or declare it as package global variable:
259 our @ISA = qw(Animal);
261 Or allow it as an implicitly named package variable:
267 If the C<Animal> class comes from another (object-oriented) module, then
268 just employ C<use base> to specify that C<Animal> should serve as the basis
269 for the C<Cow> class:
274 Now that's pretty darn simple!
276 =head2 Overriding the methods
278 Let's add a mouse, which can barely be heard:
280 # Animal package from before
283 sub sound { "squeak" }
286 print "a $class goes ", $class->sound, "!\n";
287 print "[but you can barely hear it!]\n";
296 [but you can barely hear it!]
298 Here, C<Mouse> has its own speaking routine, so C<< Mouse->speak >>
299 doesn't immediately invoke C<< Animal->speak >>. This is known as
300 "overriding". In fact, we don't even need to say that a C<Mouse> is
301 an C<Animal> at all, because all of the methods needed for C<speak> are
302 completely defined for C<Mouse>; this is known as "duck typing":
303 "If it walks like a duck and quacks like a duck, I would call it a duck"
304 (James Whitcomb). However, it would probably be beneficial to allow a
305 closer examination to conclude that a C<Mouse> is indeed an C<Animal>,
306 so it is actually better to define C<Mouse> with C<Animal> as its base
307 (that is, it is better to "derive C<Mouse> from C<Animal>").
309 Moreover, this duplication of code could become a maintenance headache
310 (though code-reuse is not actually a good reason for inheritance; good
311 design practices dictate that a derived class should be usable wherever
312 its base class is usable, which might not be the outcome if code-reuse
313 is the sole criterion for inheritance. Just remember that a C<Mouse>
314 should always act like an C<Animal>).
316 So, let's make C<Mouse> an C<Animal>!
318 The obvious solution is to invoke C<Animal::speak> directly:
320 # Animal package from before
323 sub sound { "squeak" }
326 Animal::speak($class);
327 print "[but you can barely hear it!]\n";
331 Note that we're using C<Animal::speak>. If we were to invoke
332 C<< Animal->speak >> instead, the first parameter to C<Animal::speak>
333 would automatically be C<"Animal"> rather than C<"Mouse">, so that
334 the call to C<< $class->sound >> in C<Animal::speak> would become
335 C<< Animal->sound >> rather than C<< Mouse->sound >>.
337 Also, without the method arrow C<< -> >>, it becomes necessary to specify
338 the first parameter to C<Animal::speak> ourselves, which is why C<$class>
339 is explicitly passed: C<Animal::speak($class)>.
341 However, invoking C<Animal::speak> directly is a mess: Firstly, it assumes
342 that the C<speak> method is a member of the C<Animal> class; what if C<Animal>
343 actually inherits C<speak> from its own base? Because we are no longer using
344 C<< -> >> to access C<speak>, the special method look up mechanism wouldn't be
345 used, so C<speak> wouldn't even be found!
347 The second problem is more subtle: C<Animal> is now hardwired into the subroutine
348 selection. Let's assume that C<Animal::speak> does exist. What happens when,
349 at a later time, someone expands the class hierarchy by having C<Mouse>
350 inherit from C<Mus> instead of C<Animal>. Unless the invocation of C<Animal::speak>
351 is also changed to an invocation of C<Mus::speak>, centuries worth of taxonomical
352 classification could be obliterated!
354 What we have here is a fragile or leaky abstraction; it is the beginning of a
355 maintenance nightmare. What we need is the ability to search for the right
356 method wih as few assumptions as possible.
358 =head2 Starting the search from a different place
360 A I<better> solution is to tell Perl where in the inheritance chain to begin searching
361 for C<speak>. This can be achieved with a modified version of the method arrow C<< -> >>:
363 ClassName->FirstPlaceToLook::method
365 So, the improved C<Mouse> class is:
367 # same Animal as before
369 # same @ISA, &sound as before
372 $class->Animal::speak;
373 print "[but you can barely hear it!]\n";
377 Using this syntax, we start with C<Animal> to find C<speak>, and then
378 use all of C<Animal>'s inheritance chain if it is not found immediately.
379 As usual, the first parameter to C<speak> would be C<$class>, so we no
380 longer need to pass C<$class> explicitly to C<speak>.
382 But what about the second problem? We're still hardwiring C<Animal> into
385 =head2 The SUPER way of doing things
387 If C<Animal> is replaced with the special placeholder C<SUPER> in that
388 invocation, then the contents of C<Mouse>'s C<@ISA> are used for the
389 search, beginning with C<$ISA[0]>. So, all of the problems can be fixed
392 # same Animal as before
394 # same @ISA, &sound as before
397 $class->SUPER::speak;
398 print "[but you can barely hear it!]\n";
402 In general, C<SUPER::speak> means look in the current package's C<@ISA>
403 for a class that implements C<speak>, and invoke the first one found.
404 The placeholder is called C<SUPER>, because many other languages refer
405 to base classes as "I<super>classes", and Perl likes to be eclectic.
407 Note that a call such as
409 $class->SUPER::method;
411 does I<not> look in the C<@ISA> of C<$class> unless C<$class> happens to
412 be the current package.
414 =head2 Let's review...
416 So far, we've seen the method arrow syntax:
418 Class->method(@args);
425 which constructs an argument list of:
429 and attempts to invoke:
431 Class::method("Class", @args);
433 However, if C<Class::method> is not found, then C<@Class::ISA> is examined
434 (recursively) to locate a class (a package) that does indeed contain C<method>,
435 and that subroutine is invoked instead.
437 Using this simple syntax, we have class methods, (multiple) inheritance,
438 overriding, and extending. Using just what we've seen so far, we've
439 been able to factor out common code (though that's never a good reason
440 for inheritance!), and provide a nice way to reuse implementations with
443 Now, what about data?
445 =head2 A horse is a horse, of course of course, or is it?
447 Let's start with the code for the C<Animal> class
448 and the C<Horse> class:
453 print "a $class goes ", $class->sound, "!\n";
458 sub sound { "neigh" }
461 This lets us invoke C<< Horse->speak >> to ripple upward to
462 C<Animal::speak>, calling back to C<Horse::sound> to get the specific
463 sound, and the output of:
467 But all of our Horse objects would have to be absolutely identical.
468 If we add a subroutine, all horses automatically share it. That's
469 great for making horses the same, but how do we capture the
470 distinctions of an individual horse? For example, suppose we want
471 to give our first horse a name. There's got to be a way to keep its
472 name separate from the other horses.
474 That is to say, we want particular instances of C<Horse> to have
477 In Perl, any reference can be an "instance", so let's start with the
478 simplest reference that can hold a horse's name: a scalar reference.
483 So, now C<$horse> is a reference to what will be the instance-specific
484 data (the name). The final step is to turn this reference into a real
485 instance of a C<Horse> by using the special operator C<bless>:
489 This operator stores information about the package named C<Horse> into
490 the thing pointed at by the reference. At this point, we say
491 C<$horse> is an instance of C<Horse>. That is, it's a specific
492 horse. The reference is otherwise unchanged, and can still be used
493 with traditional dereferencing operators.
495 =head2 Invoking an instance method
497 The method arrow can be used on instances, as well as classes (the names
498 of packages). So, let's get the sound that C<$horse> makes:
500 my $noise = $horse->sound("some", "unnecessary", "args");
502 To invoke C<sound>, Perl first notes that C<$horse> is a blessed
503 reference (and thus an instance). It then constructs an argument
506 Now for the fun part: Perl takes the class in which the instance was
507 blessed, in this case C<Horse>, and uses that class to locate the
508 subroutine. In this case, C<Horse::sound> is found directly (without
509 using inheritance). In the end, it is as though our initial line were
512 my $noise = Horse::sound($horse, "some", "unnecessary", "args");
514 Note that the first parameter here is still the instance, not the name
515 of the class as before. We'll get C<neigh> as the return value, and
516 that'll end up as the C<$noise> variable above.
518 If Horse::sound had not been found, we'd be wandering up the C<@Horse::ISA>
519 array, trying to find the method in one of the superclasses. The only
520 difference between a class method and an instance method is whether the
521 first parameter is an instance (a blessed reference) or a class name (a
524 =head2 Accessing the instance data
526 Because we get the instance as the first parameter, we can now access
527 the instance-specific data. In this case, let's add a way to get at
532 sub sound { "neigh" }
539 Inside C<Horse::name>, the C<@_> array contains:
541 ($horse, "some", "unnecessary", "args")
543 so the C<shift> stores C<$horse> into C<$self>. Then, C<$self> gets
544 de-referenced with C<$$self> as normal, yielding C<"Mr. Ed">.
546 It's traditional to C<shift> the first parameter into a variable named
547 C<$self> for instance methods and into a variable named C<$class> for
550 Then, the following line:
552 print $horse->name, " says ", $horse->sound, "\n";
558 =head2 How to build a horse
560 Of course, if we constructed all of our horses by hand, we'd most
561 likely make mistakes from time to time. We're also violating one of
562 the properties of object-oriented programming, in that the "inside
563 guts" of a Horse are visible. That's good if you're a veterinarian,
564 but not if you just like to own horses. So, let's have the Horse
565 class handle the details inside a class method:
569 sub sound { "neigh" }
571 my $self = shift; # instance method, so use $self
575 my $class = shift; # class method, so use $class
577 bless \$name, $class;
581 Now with the new C<named> method, we can build a horse as follows:
583 my $horse = Horse->named("Mr. Ed");
585 Notice we're back to a class method, so the two arguments to
586 C<Horse::named> are C<Horse> and C<Mr. Ed>. The C<bless> operator
587 not only blesses C<\$name>, it also returns that reference.
589 This C<Horse::named> method is called a "constructor".
591 We've called the constructor C<named> here, so that it quickly denotes
592 the constructor's argument as the name for this particular C<Horse>.
593 You can use different constructors with different names for different
594 ways of "giving birth" to the object (like maybe recording its
595 pedigree or date of birth). However, you'll find that most people
596 coming to Perl from more limited languages use a single constructor
597 named C<new>, with various ways of interpreting the arguments to
598 C<new>. Either style is fine, as long as you document your particular
599 way of giving birth to an object. (And you I<were> going to do that,
602 =head2 Inheriting the constructor
604 But was there anything specific to C<Horse> in that method? No. Therefore,
605 it's also the same recipe for building anything else that inherited from
606 C<Animal>, so let's put C<name> and C<named> there:
611 print "a $class goes ", $class->sound, "!\n";
620 bless \$name, $class;
625 sub sound { "neigh" }
628 Ahh, but what happens if we invoke C<speak> on an instance?
630 my $horse = Horse->named("Mr. Ed");
633 We get a debugging value:
635 a Horse=SCALAR(0xaca42ac) goes neigh!
637 Why? Because the C<Animal::speak> routine is expecting a classname as
638 its first parameter, not an instance. When the instance is passed in,
639 we'll end up using a blessed scalar reference as a string, and that
640 shows up as we saw it just now.
642 =head2 Making a method work with either classes or instances
644 All we need is for a method to detect if it is being called on a class
645 or called on an instance. The most straightforward way is with the
646 C<ref> operator. This returns a string (the classname) when used on a
647 blessed reference, and an empty string when used on a string (like a
648 classname). Let's modify the C<name> method first to notice the change:
652 ref $either ? $$either : "Any $either";
655 Here, the C<?:> operator comes in handy to select either the
656 dereference or a derived string. Now we can use this with either an
657 instance or a class. Note that I've changed the first parameter
658 holder to C<$either> to show that this is intended:
660 my $horse = Horse->named("Mr. Ed");
661 print Horse->name, "\n"; # prints "Any Horse\n"
662 print $horse->name, "\n"; # prints "Mr Ed.\n"
664 and now we'll fix C<speak> to use this:
668 print $either->name, " goes ", $either->sound, "\n";
671 And since C<sound> already worked with either a class or an instance,
674 =head2 Adding parameters to a method
676 Let's train our animals to eat:
682 bless \$name, $class;
686 ref $either ? $$either : "Any $either";
690 print $either->name, " goes ", $either->sound, "\n";
695 print $either->name, " eats $food.\n";
700 sub sound { "neigh" }
704 sub sound { "baaaah" }
709 my $horse = Horse->named("Mr. Ed");
716 Any Sheep eats grass.
718 An instance method with parameters gets invoked with the instance,
719 and then the list of parameters. So that first invocation is like:
721 Animal::eat($horse, "hay");
723 =head2 More interesting instances
725 What if an instance needs more data? Most interesting instances are
726 made of many items, each of which can in turn be a reference or even
727 another object. The easiest way to store these is often in a hash.
728 The keys of the hash serve as the names of parts of the object (often
729 called "instance variables" or "member variables"), and the
730 corresponding values are, well, the values.
732 But how do we turn the horse into a hash? Recall that an object was
733 any blessed reference. We can just as easily make it a blessed hash
734 reference as a blessed scalar reference, as long as everything that
735 looks at the reference is changed accordingly.
737 Let's make a sheep that has a name and a color:
739 my $bad = bless { Name => "Evil", Color => "black" }, Sheep;
741 so C<< $bad->{Name} >> has C<Evil>, and C<< $bad->{Color} >> has
742 C<black>. But we want to make C<< $bad->name >> access the name, and
743 that's now messed up because it's expecting a scalar reference. Not
744 to worry, because that's pretty easy to fix up.
746 One solution is to override C<Animal::name> and C<Animal::named> by
747 defining them anew in C<Sheep>, but then any methods added later to
748 C<Animal> might still mess up, and we'd have to override all of those
749 too. Therefore, it's never a good idea to define the data layout in a
750 way that's different from the data layout of the base classes. In fact,
751 it's a good idea to use blessed hash references in all cases. Also, this
752 is why it's important to have constructors do the low-level work. So,
753 let's redefine C<Animal>:
758 ref $either ? $either->{Name} : "Any $either";
763 my $self = { Name => $name };
767 Of course, we still need to override C<named> in order to handle
768 constructing a C<Sheep> with a certain color:
772 my ($class, $name) = @_;
773 my $self = $class->SUPER::named(@_);
774 $$self{Color} = $class->default_color;
778 (Note that C<@_> contains the parameters to C<named>.)
780 What's this C<default_color>? Well, if C<named> has only the name,
781 we still need to set a color, so we'll have a class-specific default color.
782 For a sheep, we might define it as white:
785 sub default_color { "white" }
789 my $sheep = Sheep->named("Bad");
790 print $sheep->{Color}, "\n";
796 Now, there's nothing particularly specific to C<Sheep> when it comes
797 to color, so let's remove C<Sheep::named> and implement C<Animal::named>
798 to handle color instead:
802 my ($class, $name) = @_;
803 my $self = { Name => $name, Color => $class->default_color };
807 And then to keep from having to define C<default_color> for each additional
808 class, we'll define a method that serves as the "default default" directly
812 sub default_color { "brown" }
814 Of course, because C<name> and C<named> were the only methods that
815 referenced the "structure" of the object, the rest of the methods can
816 remain the same, so C<speak> still works as before.
818 =head2 A horse of a different color
820 But having all our horses be brown would be boring. So let's add a
821 method or two to get and set the color.
828 $_[0]->{Color} = $_[1];
831 Note the alternate way of accessing the arguments: C<$_[0]> is used
832 in-place, rather than with a C<shift>. (This saves us a bit of time
833 for something that may be invoked frequently.) And now we can fix
834 that color for Mr. Ed:
836 my $horse = Horse->named("Mr. Ed");
837 $horse->set_color("black-and-white");
838 print $horse->name, " is colored ", $horse->color, "\n";
842 Mr. Ed is colored black-and-white
846 So, now we have class methods, constructors, instance methods, instance
847 data, and even accessors. But that's still just the beginning of what
848 Perl has to offer. We haven't even begun to talk about accessors that
849 double as getters and setters, destructors, indirect object notation,
850 overloading, "isa" and "can" tests, the C<UNIVERSAL> class, and so on.
851 That's for the rest of the Perl documentation to cover. Hopefully, this
852 gets you started, though.
856 For more information, see L<perlobj> (for all the gritty details about
857 Perl objects, now that you've seen the basics), L<perltoot> (the
858 tutorial for those who already know objects), L<perltooc> (dealing
859 with class data), L<perlbot> (for some more tricks), and books such as
860 Damian Conway's excellent I<Object Oriented Perl>.
862 Some modules which might prove interesting are Class::Accessor,
863 Class::Class, Class::Contract, Class::Data::Inheritable,
864 Class::MethodMaker and Tie::SecureHash
868 Copyright (c) 1999, 2000 by Randal L. Schwartz and Stonehenge
869 Consulting Services, Inc.
871 Copyright (c) 2009 by Michael F. Witten.
873 Permission is hereby granted to distribute this document intact with
874 the Perl distribution, and in accordance with the licenses of the Perl
875 distribution; derived documents must include this copyright notice
878 Portions of this text have been derived from Perl Training materials
879 originally appearing in the I<Packages, References, Objects, and
880 Modules> course taught by instructors for Stonehenge Consulting
881 Services, Inc. and used with permission.
883 Portions of this text have been derived from materials originally
884 appearing in I<Linux Magazine> and used with permission.