PageRenderTime 107ms CodeModel.GetById 13ms app.highlight 87ms RepoModel.GetById 1ms app.codeStats 0ms


Unknown | 566 lines | 428 code | 138 blank | 0 comment | 0 complexity | 9d07aad46bdae0ac59cba157ebd00506 MD5 | raw file
  1=head1 NAME
  3perlobj - Perl objects
  7First you need to understand what references are in Perl.
  8See L<perlref> for that.  Second, if you still find the following
  9reference work too complicated, a tutorial on object-oriented programming
 10in Perl can be found in L<perltoot> and L<perltootc>.
 12If you're still with us, then
 13here are three very simple definitions that you should find reassuring.
 15=over 4
 17=item 1.
 19An object is simply a reference that happens to know which class it
 20belongs to.
 22=item 2.
 24A class is simply a package that happens to provide methods to deal
 25with object references.
 27=item 3.
 29A method is simply a subroutine that expects an object reference (or
 30a package name, for class methods) as the first argument.
 34We'll cover these points now in more depth.
 36=head2 An Object is Simply a Reference
 38Unlike say C++, Perl doesn't provide any special syntax for
 39constructors.  A constructor is merely a subroutine that returns a
 40reference to something "blessed" into a class, generally the
 41class that the subroutine is defined in.  Here is a typical
 44    package Critter;
 45    sub new { bless {} }
 47That word C<new> isn't special.  You could have written
 48a construct this way, too:
 50    package Critter;
 51    sub spawn { bless {} }
 53This might even be preferable, because the C++ programmers won't
 54be tricked into thinking that C<new> works in Perl as it does in C++.
 55It doesn't.  We recommend that you name your constructors whatever
 56makes sense in the context of the problem you're solving.  For example,
 57constructors in the Tk extension to Perl are named after the widgets
 58they create.
 60One thing that's different about Perl constructors compared with those in
 61C++ is that in Perl, they have to allocate their own memory.  (The other
 62things is that they don't automatically call overridden base-class
 63constructors.)  The C<{}> allocates an anonymous hash containing no
 64key/value pairs, and returns it  The bless() takes that reference and
 65tells the object it references that it's now a Critter, and returns
 66the reference.  This is for convenience, because the referenced object
 67itself knows that it has been blessed, and the reference to it could
 68have been returned directly, like this:
 70    sub new {
 71	my $self = {};
 72	bless $self;
 73	return $self;
 74    }
 76You often see such a thing in more complicated constructors
 77that wish to call methods in the class as part of the construction:
 79    sub new {
 80	my $self = {};
 81	bless $self;
 82	$self->initialize();
 83	return $self;
 84    }
 86If you care about inheritance (and you should; see
 87L<perlmodlib/"Modules: Creation, Use, and Abuse">),
 88then you want to use the two-arg form of bless
 89so that your constructors may be inherited:
 91    sub new {
 92	my $class = shift;
 93	my $self = {};
 94	bless $self, $class;
 95	$self->initialize();
 96	return $self;
 97    }
 99Or if you expect people to call not just C<< CLASS->new() >> but also
100C<< $obj->new() >>, then use something like this.  The initialize()
101method used will be of whatever $class we blessed the
102object into:
104    sub new {
105	my $this = shift;
106	my $class = ref($this) || $this;
107	my $self = {};
108	bless $self, $class;
109	$self->initialize();
110	return $self;
111    }
113Within the class package, the methods will typically deal with the
114reference as an ordinary reference.  Outside the class package,
115the reference is generally treated as an opaque value that may
116be accessed only through the class's methods.
118Although a constructor can in theory re-bless a referenced object
119currently belonging to another class, this is almost certainly going
120to get you into trouble.  The new class is responsible for all
121cleanup later.  The previous blessing is forgotten, as an object
122may belong to only one class at a time.  (Although of course it's
123free to inherit methods from many classes.)  If you find yourself
124having to do this, the parent class is probably misbehaving, though.
126A clarification:  Perl objects are blessed.  References are not.  Objects
127know which package they belong to.  References do not.  The bless()
128function uses the reference to find the object.  Consider
129the following example:
131    $a = {};
132    $b = $a;
133    bless $a, BLAH;
134    print "\$b is a ", ref($b), "\n";
136This reports $b as being a BLAH, so obviously bless()
137operated on the object and not on the reference.
139=head2 A Class is Simply a Package
141Unlike say C++, Perl doesn't provide any special syntax for class
142definitions.  You use a package as a class by putting method
143definitions into the class.
145There is a special array within each package called @ISA, which says
146where else to look for a method if you can't find it in the current
147package.  This is how Perl implements inheritance.  Each element of the
148@ISA array is just the name of another package that happens to be a
149class package.  The classes are searched (depth first) for missing
150methods in the order that they occur in @ISA.  The classes accessible
151through @ISA are known as base classes of the current class.
153All classes implicitly inherit from class C<UNIVERSAL> as their
154last base class.  Several commonly used methods are automatically
155supplied in the UNIVERSAL class; see L<"Default UNIVERSAL methods"> for
156more details.
158If a missing method is found in a base class, it is cached
159in the current class for efficiency.  Changing @ISA or defining new
160subroutines invalidates the cache and causes Perl to do the lookup again.
162If neither the current class, its named base classes, nor the UNIVERSAL
163class contains the requested method, these three places are searched
164all over again, this time looking for a method named AUTOLOAD().  If an
165AUTOLOAD is found, this method is called on behalf of the missing method,
166setting the package global $AUTOLOAD to be the fully qualified name of
167the method that was intended to be called.
169If none of that works, Perl finally gives up and complains.
171If you want to stop the AUTOLOAD inheritance say simply
173	sub AUTOLOAD;
175and the call will die using the name of the sub being called.
177Perl classes do method inheritance only.  Data inheritance is left up
178to the class itself.  By and large, this is not a problem in Perl,
179because most classes model the attributes of their object using an
180anonymous hash, which serves as its own little namespace to be carved up
181by the various classes that might want to do something with the object.
182The only problem with this is that you can't sure that you aren't using
183a piece of the hash that isn't already used.  A reasonable workaround
184is to prepend your fieldname in the hash with the package name.
186    sub bump {
187	my $self = shift;
188	$self->{ __PACKAGE__ . ".count"}++;
189    } 
191=head2 A Method is Simply a Subroutine
193Unlike say C++, Perl doesn't provide any special syntax for method
194definition.  (It does provide a little syntax for method invocation
195though.  More on that later.)  A method expects its first argument
196to be the object (reference) or package (string) it is being invoked
197on.  There are two ways of calling methods, which we'll call class
198methods and instance methods.  
200A class method expects a class name as the first argument.  It
201provides functionality for the class as a whole, not for any
202individual object belonging to the class.  Constructors are often
203class methods, but see L<perltoot> and L<perltootc> for alternatives.
204Many class methods simply ignore their first argument, because they
205already know what package they're in and don't care what package
206they were invoked via.  (These aren't necessarily the same, because
207class methods follow the inheritance tree just like ordinary instance
208methods.)  Another typical use for class methods is to look up an
209object by name:
211    sub find {
212	my ($class, $name) = @_;
213	$objtable{$name};
214    }
216An instance method expects an object reference as its first argument.
217Typically it shifts the first argument into a "self" or "this" variable,
218and then uses that as an ordinary reference.
220    sub display {
221	my $self = shift;
222	my @keys = @_ ? @_ : sort keys %$self;
223	foreach $key (@keys) {
224	    print "\t$key => $self->{$key}\n";
225	}
226    }
228=head2 Method Invocation
230There are two ways to invoke a method, one of which you're already
231familiar with, and the other of which will look familiar.  Perl 4
232already had an "indirect object" syntax that you use when you say
234    print STDERR "help!!!\n";
236This same syntax can be used to call either class or instance methods.
237We'll use the two methods defined above, the class method to lookup
238an object reference and the instance method to print out its attributes.
240    $fred = find Critter "Fred";
241    display $fred 'Height', 'Weight';
243These could be combined into one statement by using a BLOCK in the
244indirect object slot:
246    display {find Critter "Fred"} 'Height', 'Weight';
248For C++ fans, there's also a syntax using -> notation that does exactly
249the same thing.  The parentheses are required if there are any arguments.
251    $fred = Critter->find("Fred");
252    $fred->display('Height', 'Weight');
254or in one statement,
256    Critter->find("Fred")->display('Height', 'Weight');
258There are times when one syntax is more readable, and times when the
259other syntax is more readable.  The indirect object syntax is less
260cluttered, but it has the same ambiguity as ordinary list operators.
261Indirect object method calls are usually parsed using the same rule as list
262operators: "If it looks like a function, it is a function".  (Presuming
263for the moment that you think two words in a row can look like a
264function name.  C++ programmers seem to think so with some regularity,
265especially when the first word is "new".)  Thus, the parentheses of
267    new Critter ('Barney', 1.5, 70)
269are assumed to surround ALL the arguments of the method call, regardless
270of what comes after.  Saying
272    new Critter ('Bam' x 2), 1.4, 45
274would be equivalent to
276    Critter->new('Bam' x 2), 1.4, 45
278which is unlikely to do what you want.  Confusingly, however, this
279rule applies only when the indirect object is a bareword package name,
280not when it's a scalar, a BLOCK, or a C<Package::> qualified package name.
281In those cases, the arguments are parsed in the same way as an
282indirect object list operator like print, so
284    new Critter:: ('Bam' x 2), 1.4, 45
286is the same as
288   Critter::->new(('Bam' x 2), 1.4, 45)
290For more reasons why the indirect object syntax is ambiguous, see
291L<"WARNING"> below.
293There are times when you wish to specify which class's method to use.
294Here you can call your method as an ordinary subroutine
295call, being sure to pass the requisite first argument explicitly:
297    $fred =  MyCritter::find("Critter", "Fred");
298    MyCritter::display($fred, 'Height', 'Weight');
300Unlike method calls, function calls don't consider inheritance.  If you wish
301merely to specify that Perl should I<START> looking for a method in a
302particular package, use an ordinary method call, but qualify the method
303name with the package like this:
305    $fred = Critter->MyCritter::find("Fred");
306    $fred->MyCritter::display('Height', 'Weight');
308If you're trying to control where the method search begins I<and> you're
309executing in the class itself, then you may use the SUPER pseudo class,
310which says to start looking in your base class's @ISA list without having
311to name it explicitly:
313    $self->SUPER::display('Height', 'Weight');
315Please note that the C<SUPER::> construct is meaningful I<only> within the
318Sometimes you want to call a method when you don't know the method name
319ahead of time.  You can use the arrow form, replacing the method name
320with a simple scalar variable containing the method name or a
321reference to the function.
323    $method = $fast ? "findfirst" : "findbest";
324    $fred->$method(@args);  	    # call by name
326    if ($coderef = $fred->can($parent . "::findbest")) {
327	$self->$coderef(@args);	    # call by coderef
328    }
330=head2 WARNING
332While indirect object syntax may well be appealing to English speakers and
333to C++ programmers, be not seduced!  It suffers from two grave problems.
335The first problem is that an indirect object is limited to a name,
336a scalar variable, or a block, because it would have to do too much
337lookahead otherwise, just like any other postfix dereference in the
338language.  (These are the same quirky rules as are used for the filehandle
339slot in functions like C<print> and C<printf>.)  This can lead to horribly
340confusing precedence problems, as in these next two lines:
342    move $obj->{FIELD};                 # probably wrong!
343    move $ary[$i];                      # probably wrong!
345Those actually parse as the very surprising:
347    $obj->move->{FIELD};                # Well, lookee here
348    $ary->move([$i]);                   # Didn't expect this one, eh?
350Rather than what you might have expected:
352    $obj->{FIELD}->move();              # You should be so lucky.
353    $ary[$i]->move;                     # Yeah, sure.
355The left side of ``->'' is not so limited, because it's an infix operator,
356not a postfix operator.  
358As if that weren't bad enough, think about this: Perl must guess I<at
359compile time> whether C<name> and C<move> above are functions or methods.
360Usually Perl gets it right, but when it doesn't it, you get a function
361call compiled as a method, or vice versa.  This can introduce subtle
362bugs that are hard to unravel.  For example, calling a method C<new>
363in indirect notation--as C++ programmers are so wont to do--can
364be miscompiled into a subroutine call if there's already a C<new>
365function in scope.  You'd end up calling the current package's C<new>
366as a subroutine, rather than the desired class's method.  The compiler
367tries to cheat by remembering bareword C<require>s, but the grief if it
368messes up just isn't worth the years of debugging it would likely take
369you to track such subtle bugs down.
371The infix arrow notation using ``C<< -> >>'' doesn't suffer from either
372of these disturbing ambiguities, so we recommend you use it exclusively.
374=head2 Default UNIVERSAL methods
376The C<UNIVERSAL> package automatically contains the following methods that
377are inherited by all other classes:
379=over 4
381=item isa(CLASS)
383C<isa> returns I<true> if its object is blessed into a subclass of C<CLASS>
385C<isa> is also exportable and can be called as a sub with two arguments. This
386allows the ability to check what a reference points to. Example
388    use UNIVERSAL qw(isa);
390    if(isa($ref, 'ARRAY')) {
391    	#...
392    }
394=item can(METHOD)
396C<can> checks to see if its object has a method called C<METHOD>,
397if it does then a reference to the sub is returned, if it does not then
398I<undef> is returned.
400=item VERSION( [NEED] )
402C<VERSION> returns the version number of the class (package).  If the
403NEED argument is given then it will check that the current version (as
404defined by the $VERSION variable in the given package) not less than
405NEED; it will die if this is not the case.  This method is normally
406called as a class method.  This method is called automatically by the
407C<VERSION> form of C<use>.
409    use A 1.2 qw(some imported subs);
410    # implies:
411    A->VERSION(1.2);
415B<NOTE:> C<can> directly uses Perl's internal code for method lookup, and
416C<isa> uses a very similar method and cache-ing strategy. This may cause
417strange effects if the Perl code dynamically changes @ISA in any package.
419You may add other methods to the UNIVERSAL class via Perl or XS code.
420You do not need to C<use UNIVERSAL> to make these methods
421available to your program.  This is necessary only if you wish to
422have C<isa> available as a plain subroutine in the current package.
424=head2 Destructors
426When the last reference to an object goes away, the object is
427automatically destroyed.  (This may even be after you exit, if you've
428stored references in global variables.)  If you want to capture control
429just before the object is freed, you may define a DESTROY method in
430your class.  It will automatically be called at the appropriate moment,
431and you can do any extra cleanup you need to do.  Perl passes a reference
432to the object under destruction as the first (and only) argument.  Beware
433that the reference is a read-only value, and cannot be modified by
434manipulating C<$_[0]> within the destructor.  The object itself (i.e.
435the thingy the reference points to, namely C<${$_[0]}>, C<@{$_[0]}>, 
436C<%{$_[0]}> etc.) is not similarly constrained.
438If you arrange to re-bless the reference before the destructor returns,
439perl will again call the DESTROY method for the re-blessed object after
440the current one returns.  This can be used for clean delegation of
441object destruction, or for ensuring that destructors in the base classes
442of your choosing get called.  Explicitly calling DESTROY is also possible,
443but is usually never needed.
445Do not confuse the previous discussion with how objects I<CONTAINED> in the current
446one are destroyed.  Such objects will be freed and destroyed automatically
447when the current object is freed, provided no other references to them exist
450=head2 Summary
452That's about all there is to it.  Now you need just to go off and buy a
453book about object-oriented design methodology, and bang your forehead
454with it for the next six months or so.
456=head2 Two-Phased Garbage Collection
458For most purposes, Perl uses a fast and simple, reference-based
459garbage collection system.  That means there's an extra
460dereference going on at some level, so if you haven't built
461your Perl executable using your C compiler's C<-O> flag, performance
462will suffer.  If you I<have> built Perl with C<cc -O>, then this
463probably won't matter.
465A more serious concern is that unreachable memory with a non-zero
466reference count will not normally get freed.  Therefore, this is a bad
469    {
470	my $a;
471	$a = \$a;
472    }
474Even thought $a I<should> go away, it can't.  When building recursive data
475structures, you'll have to break the self-reference yourself explicitly
476if you don't care to leak.  For example, here's a self-referential
477node such as one might use in a sophisticated tree structure:
479    sub new_node {
480	my $self = shift;
481	my $class = ref($self) || $self;
482	my $node = {};
483	$node->{LEFT} = $node->{RIGHT} = $node;
484	$node->{DATA} = [ @_ ];
485	return bless $node => $class;
486    }
488If you create nodes like that, they (currently) won't go away unless you
489break their self reference yourself.  (In other words, this is not to be
490construed as a feature, and you shouldn't depend on it.)
494When an interpreter thread finally shuts down (usually when your program
495exits), then a rather costly but complete mark-and-sweep style of garbage
496collection is performed, and everything allocated by that thread gets
497destroyed.  This is essential to support Perl as an embedded or a
498multithreadable language.  For example, this program demonstrates Perl's
499two-phased garbage collection:
501    #!/usr/bin/perl
502    package Subtle;
504    sub new {
505	my $test;
506	$test = \$test;
507	warn "CREATING " . \$test;
508	return bless \$test;
509    }
511    sub DESTROY {
512	my $self = shift;
513	warn "DESTROYING $self";
514    }
516    package main;
518    warn "starting program";
519    {
520	my $a = Subtle->new;
521	my $b = Subtle->new;
522	$$a = 0;  # break selfref
523	warn "leaving block";
524    }
526    warn "just exited block";
527    warn "time to die...";
528    exit;
530When run as F</tmp/test>, the following output is produced:
532    starting program at /tmp/test line 18.
533    CREATING SCALAR(0x8e5b8) at /tmp/test line 7.
534    CREATING SCALAR(0x8e57c) at /tmp/test line 7.
535    leaving block at /tmp/test line 23.
536    DESTROYING Subtle=SCALAR(0x8e5b8) at /tmp/test line 13.
537    just exited block at /tmp/test line 26.
538    time to die... at /tmp/test line 27.
539    DESTROYING Subtle=SCALAR(0x8e57c) during global destruction.
541Notice that "global destruction" bit there?  That's the thread
542garbage collector reaching the unreachable.
544Objects are always destructed, even when regular refs aren't.  Objects
545are destructed in a separate pass before ordinary refs just to 
546prevent object destructors from using refs that have been themselves
547destructed.  Plain refs are only garbage-collected if the destruct level
548is greater than 0.  You can test the higher levels of global destruction
549by setting the PERL_DESTRUCT_LEVEL environment variable, presuming
550C<-DDEBUGGING> was enabled during perl build time.
552A more complete garbage collection strategy will be implemented
553at a future date.
555In the meantime, the best solution is to create a non-recursive container
556class that holds a pointer to the self-referential data structure.
557Define a DESTROY method for the containing object's class that manually
558breaks the circularities in the self-referential structure.
560=head1 SEE ALSO
562A kinder, gentler tutorial on object-oriented programming in Perl can
563be found in L<perltoot>, L<perlbootc> and L<perltootc>.  You should
564also check out L<perlbot> for other object tricks, traps, and tips, as
565well as L<perlmodlib> for some style guides on constructing both
566modules and classes.