mingw/html/pod/perlobj.html

   1 <?xml version="1.0" ?>
   2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
   3 <html xmlns="http://www.w3.org/1999/xhtml">
   4 <head>
   5 <title>perlobj - Perl objects</title>
   6 <meta http-equiv="content-type" content="text/html; charset=utf-8" />
   7 <link rev="made" href="mailto:" />
   8 </head>
   9
  10 <body style="background-color: white">
  11 <table border="0" width="100%" cellspacing="0" cellpadding="3">
  12 <tr><td class="block" style="background-color: #cccccc" valign="middle">
  13 <big><strong><span class="block">&nbsp;perlobj - Perl objects</span></strong></big>
  14 </td></tr>
  15 </table>
  16
  17 <p><a name="__index__"></a></p>
  18 <!-- INDEX BEGIN -->
  19
  20 <ul>
  21
  22         <li><a href="#name">NAME</a></li>
  23         <li><a href="#description">DESCRIPTION</a></li>
  24         <ul>
  25
  26                 <li><a href="#an_object_is_simply_a_reference">An Object is Simply a Reference</a></li>
  27                 <li><a href="#a_class_is_simply_a_package">A Class is Simply a Package</a></li>
  28                 <li><a href="#a_method_is_simply_a_subroutine">A Method is Simply a Subroutine</a></li>
  29                 <li><a href="#method_invocation">Method Invocation</a></li>
  30                 <li><a href="#indirect_object_syntax">Indirect Object Syntax</a></li>
  31                 <li><a href="#default_universal_methods">Default UNIVERSAL methods</a></li>
  32                 <li><a href="#destructors">Destructors</a></li>
  33                 <li><a href="#summary">Summary</a></li>
  34                 <li><a href="#twophased_garbage_collection">Two-Phased Garbage Collection</a></li>
  35         </ul>
  36
  37         <li><a href="#see_also">SEE ALSO</a></li>
  38 </ul>
  39 <!-- INDEX END -->
  40
  41 <hr />
  42 <p>
  43 </p>
  44 <h1><a name="name_x_object__x_oop_">NAME
  45  </a></h1>
  46 <p>perlobj - Perl objects</p>
  47 <p>
  48 </p>
  49 <hr />
  50 <h1><a name="description">DESCRIPTION</a></h1>
  51 <p>First you need to understand what references are in Perl.
  52 See <a href="file://C|\msysgit\mingw\html/pod/perlref.html">the perlref manpage</a> for that.  Second, if you still find the following
  53 reference work too complicated, a tutorial on object-oriented programming
  54 in Perl can be found in <a href="file://C|\msysgit\mingw\html/pod/perltoot.html">the perltoot manpage</a> and <a href="file://C|\msysgit\mingw\html/pod/perltooc.html">the perltooc manpage</a>.</p>
  55 <p>If you're still with us, then
  56 here are three very simple definitions that you should find reassuring.</p>
  57 <ol>
  58 <li>
  59 <p>An object is simply a reference that happens to know which class it
  60 belongs to.</p>
  61 </li>
  62 <li>
  63 <p>A class is simply a package that happens to provide methods to deal
  64 with object references.</p>
  65 </li>
  66 <li>
  67 <p>A method is simply a subroutine that expects an object reference (or
  68 a package name, for class methods) as the first argument.</p>
  69 </li>
  70 </ol>
  71 <p>We'll cover these points now in more depth.</p>
  72 <p>
  73 </p>
  74 <h2><a name="an_object_is_simply_a_reference_x_object__x_bless__x_constructor__x_new_">An Object is Simply a Reference
  75    </a></h2>
  76 <p>Unlike say C++, Perl doesn't provide any special syntax for
  77 constructors.  A constructor is merely a subroutine that returns a
  78 reference to something ``blessed'' into a class, generally the
  79 class that the subroutine is defined in.  Here is a typical
  80 constructor:</p>
  81 <pre>
  82     package Critter;
  83     sub new { bless {} }</pre>
  84 <p>That word <code>new</code> isn't special.  You could have written
  85 a construct this way, too:</p>
  86 <pre>
  87     package Critter;
  88     sub spawn { bless {} }</pre>
  89 <p>This might even be preferable, because the C++ programmers won't
  90 be tricked into thinking that <code>new</code> works in Perl as it does in C++.
  91 It doesn't.  We recommend that you name your constructors whatever
  92 makes sense in the context of the problem you're solving.  For example,
  93 constructors in the Tk extension to Perl are named after the widgets
  94 they create.</p>
  95 <p>One thing that's different about Perl constructors compared with those in
  96 C++ is that in Perl, they have to allocate their own memory.  (The other
  97 things is that they don't automatically call overridden base-class
  98 constructors.)  The <code>{}</code> allocates an anonymous hash containing no
  99 key/value pairs, and returns it  The <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_bless"><code>bless()</code></a> takes that reference and
 100 tells the object it references that it's now a Critter, and returns
 101 the reference.  This is for convenience, because the referenced object
 102 itself knows that it has been blessed, and the reference to it could
 103 have been returned directly, like this:</p>
 104 <pre>
 105     sub new {
 106         my $self = {};
 107         bless $self;
 108         return $self;
 109     }</pre>
 110 <p>You often see such a thing in more complicated constructors
 111 that wish to call methods in the class as part of the construction:</p>
 112 <pre>
 113     sub new {
 114         my $self = {};
 115         bless $self;
 116         $self-&gt;initialize();
 117         return $self;
 118     }</pre>
 119 <p>If you care about inheritance (and you should; see
 120 <a href="file://C|\msysgit\mingw\html/pod/perlmodlib.html#modules__creation__use__and_abuse">Modules: Creation, Use, and Abuse in the perlmodlib manpage</a>),
 121 then you want to use the two-arg form of bless
 122 so that your constructors may be inherited:</p>
 123 <pre>
 124     sub new {
 125         my $class = shift;
 126         my $self = {};
 127         bless $self, $class;
 128         $self-&gt;initialize();
 129         return $self;
 130     }</pre>
 131 <p>Or if you expect people to call not just <code>CLASS-&gt;new()</code> but also
 132 <code>$obj-&gt;new()</code>, then use something like the following.  (Note that using
 133 this to call <code>new()</code> on an instance does not automatically perform any
 134 copying.  If you want a shallow or deep copy of an object, you'll have to
 135 specifically allow for that.)  The <code>initialize()</code> method used will be of
 136 whatever $class we blessed the object into:</p>
 137 <pre>
 138     sub new {
 139         my $this = shift;
 140         my $class = ref($this) || $this;
 141         my $self = {};
 142         bless $self, $class;
 143         $self-&gt;initialize();
 144         return $self;
 145     }</pre>
 146 <p>Within the class package, the methods will typically deal with the
 147 reference as an ordinary reference.  Outside the class package,
 148 the reference is generally treated as an opaque value that may
 149 be accessed only through the class's methods.</p>
 150 <p>Although a constructor can in theory re-bless a referenced object
 151 currently belonging to another class, this is almost certainly going
 152 to get you into trouble.  The new class is responsible for all
 153 cleanup later.  The previous blessing is forgotten, as an object
 154 may belong to only one class at a time.  (Although of course it's
 155 free to inherit methods from many classes.)  If you find yourself
 156 having to do this, the parent class is probably misbehaving, though.</p>
 157 <p>A clarification:  Perl objects are blessed.  References are not.  Objects
 158 know which package they belong to.  References do not.  The <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_bless"><code>bless()</code></a>
 159 function uses the reference to find the object.  Consider
 160 the following example:</p>
 161 <pre>
 162     $a = {};
 163     $b = $a;
 164     bless $a, BLAH;
 165     print &quot;\$b is a &quot;, ref($b), &quot;\n&quot;;</pre>
 166 <p>This reports $b as being a BLAH, so obviously <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_bless"><code>bless()</code></a>
 167 operated on the object and not on the reference.</p>
 168 <p>
 169 </p>
 170 <h2><a name="a_class_is_simply_a_package_x_class__x_package__x__isa__x_inheritance_">A Class is Simply a Package
 171    </a></h2>
 172 <p>Unlike say C++, Perl doesn't provide any special syntax for class
 173 definitions.  You use a package as a class by putting method
 174 definitions into the class.</p>
 175 <p>There is a special array within each package called @ISA, which says
 176 where else to look for a method if you can't find it in the current
 177 package.  This is how Perl implements inheritance.  Each element of the
 178 @ISA array is just the name of another package that happens to be a
 179 class package.  The classes are searched (depth first) for missing
 180 methods in the order that they occur in @ISA.  The classes accessible
 181 through @ISA are known as base classes of the current class.</p>
 182 <p>All classes implicitly inherit from class <code>UNIVERSAL</code> as their
 183 last base class.  Several commonly used methods are automatically
 184 supplied in the UNIVERSAL class; see <a href="#default_universal_methods">Default UNIVERSAL methods</a> for
 185 more details.
 186   </p>
 187 <p>If a missing method is found in a base class, it is cached
 188 in the current class for efficiency.  Changing @ISA or defining new
 189 subroutines invalidates the cache and causes Perl to do the lookup again.</p>
 190 <p>If neither the current class, its named base classes, nor the UNIVERSAL
 191 class contains the requested method, these three places are searched
 192 all over again, this time looking for a method named AUTOLOAD().  If an
 193 AUTOLOAD is found, this method is called on behalf of the missing method,
 194 setting the package global $AUTOLOAD to be the fully qualified name of
 195 the method that was intended to be called.
 196 </p>
 197 <p>If none of that works, Perl finally gives up and complains.</p>
 198 <p>If you want to stop the AUTOLOAD inheritance say simply
 199 </p>
 200 <pre>
 201         sub AUTOLOAD;</pre>
 202 <p>and the call will die using the name of the sub being called.</p>
 203 <p>Perl classes do method inheritance only.  Data inheritance is left up
 204 to the class itself.  By and large, this is not a problem in Perl,
 205 because most classes model the attributes of their object using an
 206 anonymous hash, which serves as its own little namespace to be carved up
 207 by the various classes that might want to do something with the object.
 208 The only problem with this is that you can't sure that you aren't using
 209 a piece of the hash that isn't already used.  A reasonable workaround
 210 is to prepend your fieldname in the hash with the package name.
 211  </p>
 212 <pre>
 213     sub bump {
 214         my $self = shift;
 215         $self-&gt;{ __PACKAGE__ . &quot;.count&quot;}++;
 216     }</pre>
 217 <p>
 218 </p>
 219 <h2><a name="a_method_is_simply_a_subroutine_x_method_">A Method is Simply a Subroutine
 220 </a></h2>
 221 <p>Unlike say C++, Perl doesn't provide any special syntax for method
 222 definition.  (It does provide a little syntax for method invocation
 223 though.  More on that later.)  A method expects its first argument
 224 to be the object (reference) or package (string) it is being invoked
 225 on.  There are two ways of calling methods, which we'll call class
 226 methods and instance methods.</p>
 227 <p>A class method expects a class name as the first argument.  It
 228 provides functionality for the class as a whole, not for any
 229 individual object belonging to the class.  Constructors are often
 230 class methods, but see <a href="file://C|\msysgit\mingw\html/pod/perltoot.html">the perltoot manpage</a> and <a href="file://C|\msysgit\mingw\html/pod/perltooc.html">the perltooc manpage</a> for alternatives.
 231 Many class methods simply ignore their first argument, because they
 232 already know what package they're in and don't care what package
 233 they were invoked via.  (These aren't necessarily the same, because
 234 class methods follow the inheritance tree just like ordinary instance
 235 methods.)  Another typical use for class methods is to look up an
 236 object by name:</p>
 237 <pre>
 238     sub find {
 239         my ($class, $name) = @_;
 240         $objtable{$name};
 241     }</pre>
 242 <p>An instance method expects an object reference as its first argument.
 243 Typically it shifts the first argument into a ``self'' or ``this'' variable,
 244 and then uses that as an ordinary reference.</p>
 245 <pre>
 246     sub display {
 247         my $self = shift;
 248         my @keys = @_ ? @_ : sort keys %$self;
 249         foreach $key (@keys) {
 250             print &quot;\t$key =&gt; $self-&gt;{$key}\n&quot;;
 251         }
 252     }</pre>
 253 <p>
 254 </p>
 255 <h2><a name="method_invocation_x_invocation__x_method__x_arrow__x_______">Method Invocation
 256     &gt;&gt;</a></h2>
 257 <p>For various historical and other reasons, Perl offers two equivalent
 258 ways to write a method call.  The simpler and more common way is to use
 259 the arrow notation:</p>
 260 <pre>
 261     my $fred = Critter-&gt;find(&quot;Fred&quot;);
 262     $fred-&gt;display(&quot;Height&quot;, &quot;Weight&quot;);</pre>
 263 <p>You should already be familiar with the use of the <code>-&gt;</code> operator with
 264 references.  In fact, since <code>$fred</code> above is a reference to an object,
 265 you could think of the method call as just another form of
 266 dereferencing.</p>
 267 <p>Whatever is on the left side of the arrow, whether a reference or a
 268 class name, is passed to the method subroutine as its first argument.
 269 So the above code is mostly equivalent to:</p>
 270 <pre>
 271     my $fred = Critter::find(&quot;Critter&quot;, &quot;Fred&quot;);
 272     Critter::display($fred, &quot;Height&quot;, &quot;Weight&quot;);</pre>
 273 <p>How does Perl know which package the subroutine is in?  By looking at
 274 the left side of the arrow, which must be either a package name or a
 275 reference to an object, i.e. something that has been blessed to a
 276 package.  Either way, that's the package where Perl starts looking.  If
 277 that package has no subroutine with that name, Perl starts looking for
 278 it in any base classes of that package, and so on.</p>
 279 <p>If you need to, you <em>can</em> force Perl to start looking in some other package:</p>
 280 <pre>
 281     my $barney = MyCritter-&gt;Critter::find(&quot;Barney&quot;);
 282     $barney-&gt;Critter::display(&quot;Height&quot;, &quot;Weight&quot;);</pre>
 283 <p>Here <code>MyCritter</code> is presumably a subclass of <code>Critter</code> that defines
 284 its own versions of <code>find()</code> and display().  We haven't specified what
 285 those methods do, but that doesn't matter above since we've forced Perl
 286 to start looking for the subroutines in <code>Critter</code>.</p>
 287 <p>As a special case of the above, you may use the <code>SUPER</code> pseudo-class to
 288 tell Perl to start looking for the method in the packages named in the
 289 current class's <code>@ISA</code> list.
 290 </p>
 291 <pre>
 292     package MyCritter;
 293     use base 'Critter';    # sets @MyCritter::ISA = ('Critter');</pre>
 294 <pre>
 295     sub display {
 296         my ($self, @args) = @_;
 297         $self-&gt;SUPER::display(&quot;Name&quot;, @args);
 298     }</pre>
 299 <p>It is important to note that <code>SUPER</code> refers to the <code>superclass(es)</code> of the
 300 <em>current package</em> and not to the <code>superclass(es)</code> of the object. Also, the
 301 <code>SUPER</code> pseudo-class can only currently be used as a modifier to a method
 302 name, but not in any of the other ways that class names are normally used,
 303 eg:
 304 </p>
 305 <pre>
 306     something-&gt;SUPER::method(...);      # OK
 307     SUPER::method(...);                 # WRONG
 308     SUPER-&gt;method(...);                 # WRONG</pre>
 309 <p>Instead of a class name or an object reference, you can also use any
 310 expression that returns either of those on the left side of the arrow.
 311 So the following statement is valid:</p>
 312 <pre>
 313     Critter-&gt;find(&quot;Fred&quot;)-&gt;display(&quot;Height&quot;, &quot;Weight&quot;);</pre>
 314 <p>and so is the following:</p>
 315 <pre>
 316     my $fred = (reverse &quot;rettirC&quot;)-&gt;find(reverse &quot;derF&quot;);</pre>
 317 <p>The right side of the arrow typically is the method name, but a simple
 318 scalar variable containing either the method name or a subroutine
 319 reference can also be used.</p>
 320 <p>
 321 </p>
 322 <h2><a name="indirect_object_syntax_x_indirect_object_syntax__x_invocation__indirect__x_indirect_">Indirect Object Syntax
 323   </a></h2>
 324 <p>The other way to invoke a method is by using the so-called ``indirect
 325 object'' notation.  This syntax was available in Perl 4 long before
 326 objects were introduced, and is still used with filehandles like this:</p>
 327 <pre>
 328    print STDERR &quot;help!!!\n&quot;;</pre>
 329 <p>The same syntax can be used to call either object or class methods.</p>
 330 <pre>
 331    my $fred = find Critter &quot;Fred&quot;;
 332    display $fred &quot;Height&quot;, &quot;Weight&quot;;</pre>
 333 <p>Notice that there is no comma between the object or class name and the
 334 parameters.  This is how Perl can tell you want an indirect method call
 335 instead of an ordinary subroutine call.</p>
 336 <p>But what if there are no arguments?  In that case, Perl must guess what
 337 you want.  Even worse, it must make that guess <em>at compile time</em>.
 338 Usually Perl gets it right, but when it doesn't you get a function
 339 call compiled as a method, or vice versa.  This can introduce subtle bugs
 340 that are hard to detect.</p>
 341 <p>For example, a call to a method <code>new</code> in indirect notation -- as C++
 342 programmers are wont to make -- can be miscompiled into a subroutine
 343 call if there's already a <code>new</code> function in scope.  You'd end up
 344 calling the current package's <code>new</code> as a subroutine, rather than the
 345 desired class's method.  The compiler tries to cheat by remembering
 346 bareword <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_require"><code>require</code></a>s, but the grief when it messes up just isn't worth the
 347 years of debugging it will take you to track down such subtle bugs.</p>
 348 <p>There is another problem with this syntax: the indirect object is
 349 limited to a name, a scalar variable, or a block, because it would have
 350 to do too much lookahead otherwise, just like any other postfix
 351 dereference in the language.  (These are the same quirky rules as are
 352 used for the filehandle slot in functions like <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_print"><code>print</code></a> and <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_printf"><code>printf</code></a>.)
 353 This can lead to horribly confusing precedence problems, as in these
 354 next two lines:</p>
 355 <pre>
 356     move $obj-&gt;{FIELD};                 # probably wrong!
 357     move $ary[$i];                      # probably wrong!</pre>
 358 <p>Those actually parse as the very surprising:</p>
 359 <pre>
 360     $obj-&gt;move-&gt;{FIELD};                # Well, lookee here
 361     $ary-&gt;move([$i]);                   # Didn't expect this one, eh?</pre>
 362 <p>Rather than what you might have expected:</p>
 363 <pre>
 364     $obj-&gt;{FIELD}-&gt;move();              # You should be so lucky.
 365     $ary[$i]-&gt;move;                     # Yeah, sure.</pre>
 366 <p>To get the correct behavior with indirect object syntax, you would have
 367 to use a block around the indirect object:</p>
 368 <pre>
 369     move {$obj-&gt;{FIELD}};
 370     move {$ary[$i]};</pre>
 371 <p>Even then, you still have the same potential problem if there happens to
 372 be a function named <code>move</code> in the current package.  <strong>The <code>-&gt;</code>
 373 notation suffers from neither of these disturbing ambiguities, so we
 374 recommend you use it exclusively.</strong>  However, you may still end up having
 375 to read code using the indirect object notation, so it's important to be
 376 familiar with it.</p>
 377 <p>
 378 </p>
 379 <h2><a name="default_universal_methods_x_universal_">Default UNIVERSAL methods
 380 </a></h2>
 381 <p>The <code>UNIVERSAL</code> package automatically contains the following methods that
 382 are inherited by all other classes:</p>
 383 <dl>
 384 <dt><strong><a name="item_isa"><code>isa(CLASS)</code>
 385 </a></strong>
 386
 387 <dd>
 388 <p><a href="#item_isa"><code>isa</code></a> returns <em>true</em> if its object is blessed into a subclass of <code>CLASS</code></p>
 389 </dd>
 390 <dd>
 391 <p>You can also call <code>UNIVERSAL::isa</code> as a subroutine with two arguments.  Of
 392 course, this will do the wrong thing if someone has overridden <a href="#item_isa"><code>isa</code></a> in a
 393 class, so don't do it.</p>
 394 </dd>
 395 <dd>
 396 <p>If you need to determine whether you've received a valid invocant, use the
 397 <code>blessed</code> function from <a href="file://C|\msysgit\mingw\html/lib/Scalar/Util.html">the Scalar::Util manpage</a>:
 398  </p>
 399 </dd>
 400 <dd>
 401 <pre>
 402     if (blessed($ref) &amp;&amp; $ref-&gt;isa( 'Some::Class')) {
 403         # ...
 404     }</pre>
 405 </dd>
 406 <dd>
 407 <p><code>blessed</code> returns the name of the package the argument has been
 408 blessed into, or <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_undef"><code>undef</code></a>.</p>
 409 </dd>
 410 </li>
 411 <dt><strong><a name="item_can"><code>can(METHOD)</code>
 412 </a></strong>
 413
 414 <dd>
 415 <p><a href="#item_can"><code>can</code></a> checks to see if its object has a method called <code>METHOD</code>,
 416 if it does then a reference to the sub is returned, if it does not then
 417 <em>undef</em> is returned.</p>
 418 </dd>
 419 <dd>
 420 <p><code>UNIVERSAL::can</code> can also be called as a subroutine with two arguments.  It'll
 421 always return <em>undef</em> if its first argument isn't an object or a class name.
 422 The same caveats for calling <code>UNIVERSAL::isa</code> directly apply here, too.</p>
 423 </dd>
 424 </li>
 425 <dt><strong><a name="item_version">VERSION( [NEED] )
 426 </a></strong>
 427
 428 <dd>
 429 <p><a href="#item_version"><code>VERSION</code></a> returns the version number of the class (package).  If the
 430 NEED argument is given then it will check that the current version (as
 431 defined by the $VERSION variable in the given package) not less than
 432 NEED; it will die if this is not the case.  This method is normally
 433 called as a class method.  This method is called automatically by the
 434 <a href="#item_version"><code>VERSION</code></a> form of <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_use"><code>use</code></a>.</p>
 435 </dd>
 436 <dd>
 437 <pre>
 438     use A 1.2 qw(some imported subs);
 439     # implies:
 440     A-&gt;VERSION(1.2);</pre>
 441 </dd>
 442 </li>
 443 </dl>
 444 <p><strong>NOTE:</strong> <a href="#item_can"><code>can</code></a> directly uses Perl's internal code for method lookup, and
 445 <a href="#item_isa"><code>isa</code></a> uses a very similar method and cache-ing strategy. This may cause
 446 strange effects if the Perl code dynamically changes @ISA in any package.</p>
 447 <p>You may add other methods to the UNIVERSAL class via Perl or XS code.
 448 You do not need to <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_use"><code>use UNIVERSAL</code></a> to make these methods
 449 available to your program (and you should not do so).</p>
 450 <p>
 451 </p>
 452 <h2><a name="destructors_x_destructor__x_destroy_">Destructors
 453  </a></h2>
 454 <p>When the last reference to an object goes away, the object is
 455 automatically destroyed.  (This may even be after you exit, if you've
 456 stored references in global variables.)  If you want to capture control
 457 just before the object is freed, you may define a DESTROY method in
 458 your class.  It will automatically be called at the appropriate moment,
 459 and you can do any extra cleanup you need to do.  Perl passes a reference
 460 to the object under destruction as the first (and only) argument.  Beware
 461 that the reference is a read-only value, and cannot be modified by
 462 manipulating <code>$_[0]</code> within the destructor.  The object itself (i.e.
 463 the thingy the reference points to, namely <code>${$_[0]}</code>, <code>@{$_[0]}</code>,
 464 <code>%{$_[0]}</code> etc.) is not similarly constrained.</p>
 465 <p>Since DESTROY methods can be called at unpredictable times, it is
 466 important that you localise any global variables that the method may
 467 update.  In particular, localise <a href="file://C|\msysgit\mingw\html/pod/perlvar.html#item___"><code>$@</code></a> if you use <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_eval"><code>eval {}</code></a> and
 468 localise <a href="file://C|\msysgit\mingw\html/pod/perlvar.html#item__"><code>$?</code></a> if you use <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_system"><code>system</code></a> or backticks.</p>
 469 <p>If you arrange to re-bless the reference before the destructor returns,
 470 perl will again call the DESTROY method for the re-blessed object after
 471 the current one returns.  This can be used for clean delegation of
 472 object destruction, or for ensuring that destructors in the base classes
 473 of your choosing get called.  Explicitly calling DESTROY is also possible,
 474 but is usually never needed.</p>
 475 <p>Do not confuse the previous discussion with how objects <em>CONTAINED</em> in the current
 476 one are destroyed.  Such objects will be freed and destroyed automatically
 477 when the current object is freed, provided no other references to them exist
 478 elsewhere.</p>
 479 <p>
 480 </p>
 481 <h2><a name="summary">Summary</a></h2>
 482 <p>That's about all there is to it.  Now you need just to go off and buy a
 483 book about object-oriented design methodology, and bang your forehead
 484 with it for the next six months or so.</p>
 485 <p>
 486 </p>
 487 <h2><a name="twophased_garbage_collection_x_garbage_collection__x_gc__x_circular_reference__x_reference__circular__x_destroy__x_destructor_">Two-Phased Garbage Collection
 488
 489   </a></h2>
 490 <p>For most purposes, Perl uses a fast and simple, reference-based
 491 garbage collection system.  That means there's an extra
 492 dereference going on at some level, so if you haven't built
 493 your Perl executable using your C compiler's <code>-O</code> flag, performance
 494 will suffer.  If you <em>have</em> built Perl with <code>cc -O</code>, then this
 495 probably won't matter.</p>
 496 <p>A more serious concern is that unreachable memory with a non-zero
 497 reference count will not normally get freed.  Therefore, this is a bad
 498 idea:</p>
 499 <pre>
 500     {
 501         my $a;
 502         $a = \$a;
 503     }</pre>
 504 <p>Even thought $a <em>should</em> go away, it can't.  When building recursive data
 505 structures, you'll have to break the self-reference yourself explicitly
 506 if you don't care to leak.  For example, here's a self-referential
 507 node such as one might use in a sophisticated tree structure:</p>
 508 <pre>
 509     sub new_node {
 510         my $class = shift;
 511         my $node  = {};
 512         $node-&gt;{LEFT} = $node-&gt;{RIGHT} = $node;
 513         $node-&gt;{DATA} = [ @_ ];
 514         return bless $node =&gt; $class;
 515     }</pre>
 516 <p>If you create nodes like that, they (currently) won't go away unless you
 517 break their self reference yourself.  (In other words, this is not to be
 518 construed as a feature, and you shouldn't depend on it.)</p>
 519 <p>Almost.</p>
 520 <p>When an interpreter thread finally shuts down (usually when your program
 521 exits), then a rather costly but complete mark-and-sweep style of garbage
 522 collection is performed, and everything allocated by that thread gets
 523 destroyed.  This is essential to support Perl as an embedded or a
 524 multithreadable language.  For example, this program demonstrates Perl's
 525 two-phased garbage collection:</p>
 526 <pre>
 527     #!/usr/bin/perl
 528     package Subtle;</pre>
 529 <pre>
 530     sub new {
 531         my $test;
 532         $test = \$test;
 533         warn &quot;CREATING &quot; . \$test;
 534         return bless \$test;
 535     }</pre>
 536 <pre>
 537     sub DESTROY {
 538         my $self = shift;
 539         warn &quot;DESTROYING $self&quot;;
 540     }</pre>
 541 <pre>
 542     package main;</pre>
 543 <pre>
 544     warn &quot;starting program&quot;;
 545     {
 546         my $a = Subtle-&gt;new;
 547         my $b = Subtle-&gt;new;
 548         $$a = 0;  # break selfref
 549         warn &quot;leaving block&quot;;
 550     }</pre>
 551 <pre>
 552     warn &quot;just exited block&quot;;
 553     warn &quot;time to die...&quot;;
 554     exit;</pre>
 555 <p>When run as <em>/foo/test</em>, the following output is produced:</p>
 556 <pre>
 557     starting program at /foo/test line 18.
 558     CREATING SCALAR(0x8e5b8) at /foo/test line 7.
 559     CREATING SCALAR(0x8e57c) at /foo/test line 7.
 560     leaving block at /foo/test line 23.
 561     DESTROYING Subtle=SCALAR(0x8e5b8) at /foo/test line 13.
 562     just exited block at /foo/test line 26.
 563     time to die... at /foo/test line 27.
 564     DESTROYING Subtle=SCALAR(0x8e57c) during global destruction.</pre>
 565 <p>Notice that ``global destruction'' bit there?  That's the thread
 566 garbage collector reaching the unreachable.</p>
 567 <p>Objects are always destructed, even when regular refs aren't.  Objects
 568 are destructed in a separate pass before ordinary refs just to
 569 prevent object destructors from using refs that have been themselves
 570 destructed.  Plain refs are only garbage-collected if the destruct level
 571 is greater than 0.  You can test the higher levels of global destruction
 572 by setting the PERL_DESTRUCT_LEVEL environment variable, presuming
 573 <code>-DDEBUGGING</code> was enabled during perl build time.
 574 See <a href="file://C|\msysgit\mingw\html/pod/perlhack.html#perl_destruct_level">PERL_DESTRUCT_LEVEL in the perlhack manpage</a> for more information.</p>
 575 <p>A more complete garbage collection strategy will be implemented
 576 at a future date.</p>
 577 <p>In the meantime, the best solution is to create a non-recursive container
 578 class that holds a pointer to the self-referential data structure.
 579 Define a DESTROY method for the containing object's class that manually
 580 breaks the circularities in the self-referential structure.</p>
 581 <p>
 582 </p>
 583 <hr />
 584 <h1><a name="see_also">SEE ALSO</a></h1>
 585 <p>A kinder, gentler tutorial on object-oriented programming in Perl can
 586 be found in <a href="file://C|\msysgit\mingw\html/pod/perltoot.html">the perltoot manpage</a>, <a href="file://C|\msysgit\mingw\html/pod/perlboot.html">the perlboot manpage</a> and <a href="file://C|\msysgit\mingw\html/pod/perltooc.html">the perltooc manpage</a>.  You should
 587 also check out <a href="file://C|\msysgit\mingw\html/pod/perlbot.html">the perlbot manpage</a> for other object tricks, traps, and tips, as
 588 well as <a href="file://C|\msysgit\mingw\html/pod/perlmodlib.html">the perlmodlib manpage</a> for some style guides on constructing both
 589 modules and classes.</p>
 590 <table border="0" width="100%" cellspacing="0" cellpadding="3">
 591 <tr><td class="block" style="background-color: #cccccc" valign="middle">
 592 <big><strong><span class="block">&nbsp;perlobj - Perl objects</span></strong></big>
 593 </td></tr>
 594 </table>
 595
 596 </body>
 597
 598 </html>