Commit | Line | Data |
a0d0e21e |
1 | =head1 NAME |
2 | |
3 | perlobj - Perl objects |
4 | |
5 | =head1 DESCRIPTION |
6 | |
7 | First of all, you need to understand what references are in Perl. See |
8 | L<perlref> for that. |
9 | |
10 | Here are three very simple definitions that you should find reassuring. |
11 | |
12 | =over 4 |
13 | |
14 | =item 1. |
15 | |
16 | An object is simply a reference that happens to know which class it |
17 | belongs to. |
18 | |
19 | =item 2. |
20 | |
21 | A class is simply a package that happens to provide methods to deal |
22 | with object references. |
23 | |
24 | =item 3. |
25 | |
26 | A method is simply a subroutine that expects an object reference (or |
27 | a package name, for static methods) as the first argument. |
28 | |
29 | =back |
30 | |
31 | We'll cover these points now in more depth. |
32 | |
33 | =head2 An Object is Simply a Reference |
34 | |
35 | Unlike say C++, Perl doesn't provide any special syntax for |
36 | constructors. A constructor is merely a subroutine that returns a |
cb1a09d0 |
37 | reference to something "blessed" into a class, generally the |
a0d0e21e |
38 | class that the subroutine is defined in. Here is a typical |
39 | constructor: |
40 | |
41 | package Critter; |
42 | sub new { bless {} } |
43 | |
44 | The C<{}> constructs a reference to an anonymous hash containing no |
45 | key/value pairs. The bless() takes that reference and tells the object |
46 | it references that it's now a Critter, and returns the reference. |
47 | This is for convenience, since the referenced object itself knows that |
48 | it has been blessed, and its reference to it could have been returned |
49 | directly, like this: |
50 | |
51 | sub new { |
52 | my $self = {}; |
53 | bless $self; |
54 | return $self; |
55 | } |
56 | |
57 | In fact, you often see such a thing in more complicated constructors |
58 | that wish to call methods in the class as part of the construction: |
59 | |
60 | sub new { |
61 | my $self = {} |
62 | bless $self; |
63 | $self->initialize(); |
cb1a09d0 |
64 | return $self; |
65 | } |
66 | |
67 | If you care about inheritance (and you should; see L<perlmod/"Modules: |
68 | Creation, Use and Abuse">), then you want to use the two-arg form of bless |
69 | so that your constructors may be inherited: |
70 | |
71 | sub new { |
72 | my $class = shift; |
73 | my $self = {}; |
74 | bless $self, $class |
75 | $self->initialize(); |
76 | return $self; |
77 | } |
78 | |
d28ebecd |
79 | Or if you expect people to call not just C<CLASS-E<gt>new()> but also |
80 | C<$obj-E<gt>new()>, then use something like this. The initialize() |
cb1a09d0 |
81 | method used will be of whatever $class we blessed the |
82 | object into: |
83 | |
84 | sub new { |
85 | my $this = shift; |
86 | my $class = ref($this) || $this; |
87 | my $self = {}; |
88 | bless $self, $class |
89 | $self->initialize(); |
90 | return $self; |
a0d0e21e |
91 | } |
92 | |
93 | Within the class package, the methods will typically deal with the |
94 | reference as an ordinary reference. Outside the class package, |
95 | the reference is generally treated as an opaque value that may |
96 | only be accessed through the class's methods. |
97 | |
748a9306 |
98 | A constructor may re-bless a referenced object currently belonging to |
a0d0e21e |
99 | another class, but then the new class is responsible for all cleanup |
100 | later. The previous blessing is forgotten, as an object may only |
101 | belong to one class at a time. (Although of course it's free to |
102 | inherit methods from many classes.) |
103 | |
104 | A clarification: Perl objects are blessed. References are not. Objects |
105 | know which package they belong to. References do not. The bless() |
106 | function simply uses the reference in order to find the object. Consider |
107 | the following example: |
108 | |
109 | $a = {}; |
110 | $b = $a; |
111 | bless $a, BLAH; |
112 | print "\$b is a ", ref($b), "\n"; |
113 | |
114 | This reports $b as being a BLAH, so obviously bless() |
115 | operated on the object and not on the reference. |
116 | |
117 | =head2 A Class is Simply a Package |
118 | |
119 | Unlike say C++, Perl doesn't provide any special syntax for class |
120 | definitions. You just use a package as a class by putting method |
121 | definitions into the class. |
122 | |
123 | There is a special array within each package called @ISA which says |
124 | where else to look for a method if you can't find it in the current |
125 | package. This is how Perl implements inheritance. Each element of the |
126 | @ISA array is just the name of another package that happens to be a |
127 | class package. The classes are searched (depth first) for missing |
128 | methods in the order that they occur in @ISA. The classes accessible |
cb1a09d0 |
129 | through @ISA are known as base classes of the current class. |
a0d0e21e |
130 | |
131 | If a missing method is found in one of the base classes, it is cached |
132 | in the current class for efficiency. Changing @ISA or defining new |
133 | subroutines invalidates the cache and causes Perl to do the lookup again. |
134 | |
135 | If a method isn't found, but an AUTOLOAD routine is found, then |
136 | that is called on behalf of the missing method. |
137 | |
138 | If neither a method nor an AUTOLOAD routine is found in @ISA, then one |
139 | last try is made for the method (or an AUTOLOAD routine) in a class |
140 | called UNIVERSAL. If that doesn't work, Perl finally gives up and |
141 | complains. |
142 | |
143 | Perl classes only do method inheritance. Data inheritance is left |
144 | up to the class itself. By and large, this is not a problem in Perl, |
145 | because most classes model the attributes of their object using |
146 | an anonymous hash, which serves as its own little namespace to be |
147 | carved up by the various classes that might want to do something |
148 | with the object. |
149 | |
150 | =head2 A Method is Simply a Subroutine |
151 | |
152 | Unlike say C++, Perl doesn't provide any special syntax for method |
153 | definition. (It does provide a little syntax for method invocation |
154 | though. More on that later.) A method expects its first argument |
155 | to be the object or package it is being invoked on. There are just two |
156 | types of methods, which we'll call static and virtual, in honor of |
157 | the two C++ method types they most closely resemble. |
158 | |
159 | A static method expects a class name as the first argument. It |
160 | provides functionality for the class as a whole, not for any individual |
161 | object belonging to the class. Constructors are typically static |
162 | methods. Many static methods simply ignore their first argument, since |
163 | they already know what package they're in, and don't care what package |
164 | they were invoked via. (These aren't necessarily the same, since |
165 | static methods follow the inheritance tree just like ordinary virtual |
166 | methods.) Another typical use for static methods is to look up an |
167 | object by name: |
168 | |
169 | sub find { |
170 | my ($class, $name) = @_; |
171 | $objtable{$name}; |
172 | } |
173 | |
174 | A virtual method expects an object reference as its first argument. |
175 | Typically it shifts the first argument into a "self" or "this" variable, |
176 | and then uses that as an ordinary reference. |
177 | |
178 | sub display { |
179 | my $self = shift; |
180 | my @keys = @_ ? @_ : sort keys %$self; |
181 | foreach $key (@keys) { |
182 | print "\t$key => $self->{$key}\n"; |
183 | } |
184 | } |
185 | |
186 | =head2 Method Invocation |
187 | |
188 | There are two ways to invoke a method, one of which you're already |
189 | familiar with, and the other of which will look familiar. Perl 4 |
190 | already had an "indirect object" syntax that you use when you say |
191 | |
192 | print STDERR "help!!!\n"; |
193 | |
194 | This same syntax can be used to call either static or virtual methods. |
195 | We'll use the two methods defined above, the static method to lookup |
196 | an object reference and the virtual method to print out its attributes. |
197 | |
198 | $fred = find Critter "Fred"; |
199 | display $fred 'Height', 'Weight'; |
200 | |
201 | These could be combined into one statement by using a BLOCK in the |
202 | indirect object slot: |
203 | |
204 | display {find Critter "Fred"} 'Height', 'Weight'; |
205 | |
d28ebecd |
206 | For C++ fans, there's also a syntax using -E<gt> notation that does exactly |
a0d0e21e |
207 | the same thing. The parentheses are required if there are any arguments. |
208 | |
209 | $fred = Critter->find("Fred"); |
210 | $fred->display('Height', 'Weight'); |
211 | |
212 | or in one statement, |
213 | |
214 | Critter->find("Fred")->display('Height', 'Weight'); |
215 | |
216 | There are times when one syntax is more readable, and times when the |
217 | other syntax is more readable. The indirect object syntax is less |
218 | cluttered, but it has the same ambiguity as ordinary list operators. |
219 | Indirect object method calls are parsed using the same rule as list |
220 | operators: "If it looks like a function, it is a function". (Presuming |
221 | for the moment that you think two words in a row can look like a |
222 | function name. C++ programmers seem to think so with some regularity, |
223 | especially when the first word is "new".) Thus, the parens of |
224 | |
225 | new Critter ('Barney', 1.5, 70) |
226 | |
227 | are assumed to surround ALL the arguments of the method call, regardless |
228 | of what comes after. Saying |
229 | |
230 | new Critter ('Bam' x 2), 1.4, 45 |
231 | |
232 | would be equivalent to |
233 | |
234 | Critter->new('Bam' x 2), 1.4, 45 |
235 | |
236 | which is unlikely to do what you want. |
237 | |
238 | There are times when you wish to specify which class's method to use. |
239 | In this case, you can call your method as an ordinary subroutine |
240 | call, being sure to pass the requisite first argument explicitly: |
241 | |
242 | $fred = MyCritter::find("Critter", "Fred"); |
243 | MyCritter::display($fred, 'Height', 'Weight'); |
244 | |
245 | Note however, that this does not do any inheritance. If you merely |
246 | wish to specify that Perl should I<START> looking for a method in a |
247 | particular package, use an ordinary method call, but qualify the method |
248 | name with the package like this: |
249 | |
250 | $fred = Critter->MyCritter::find("Fred"); |
251 | $fred->MyCritter::display('Height', 'Weight'); |
252 | |
cb1a09d0 |
253 | If you're trying to control where the method search begins I<and> you're |
254 | executing in the class itself, then you may use the SUPER pseudoclass, |
255 | which says to start looking in your base class's @ISA list without having |
256 | to explicitly name it: |
257 | |
258 | $self->SUPER::display('Height', 'Weight'); |
259 | |
260 | Please note that the C<SUPER::> construct is I<only> meaningful within the |
261 | class. |
262 | |
748a9306 |
263 | Sometimes you want to call a method when you don't know the method name |
264 | ahead of time. You can use the arrow form, replacing the method name |
265 | with a simple scalar variable containing the method name: |
266 | |
267 | $method = $fast ? "findfirst" : "findbest"; |
268 | $fred->$method(@args); |
269 | |
a0d0e21e |
270 | =head2 Destructors |
271 | |
272 | When the last reference to an object goes away, the object is |
273 | automatically destroyed. (This may even be after you exit, if you've |
274 | stored references in global variables.) If you want to capture control |
275 | just before the object is freed, you may define a DESTROY method in |
276 | your class. It will automatically be called at the appropriate moment, |
277 | and you can do any extra cleanup you need to do. |
278 | |
279 | Perl doesn't do nested destruction for you. If your constructor |
280 | reblessed a reference from one of your base classes, your DESTROY may |
281 | need to call DESTROY for any base classes that need it. But this only |
282 | applies to reblessed objects--an object reference that is merely |
283 | I<CONTAINED> in the current object will be freed and destroyed |
284 | automatically when the current object is freed. |
285 | |
748a9306 |
286 | =head2 WARNING |
287 | |
288 | An indirect object is limited to a name, a scalar variable, or a block, |
289 | because it would have to do too much lookahead otherwise, just like any |
d28ebecd |
290 | other postfix dereference in the language. The left side of -E<gt> is not so |
748a9306 |
291 | limited, because it's an infix operator, not a postfix operator. |
292 | |
293 | That means that below, A and B are equivalent to each other, and C and D |
294 | are equivalent, but AB and CD are different: |
295 | |
296 | A: method $obref->{"fieldname"} |
297 | B: (method $obref)->{"fieldname"} |
298 | C: $obref->{"fieldname"}->method() |
299 | D: method {$obref->{"fieldname"}} |
300 | |
a0d0e21e |
301 | =head2 Summary |
302 | |
303 | That's about all there is to it. Now you just need to go off and buy a |
304 | book about object-oriented design methodology, and bang your forehead |
305 | with it for the next six months or so. |
306 | |
cb1a09d0 |
307 | =head2 Two-Phased Garbage Collection |
308 | |
309 | For most purposes, Perl uses a fast and simple reference-based |
310 | garbage collection system. For this reason, there's an extra |
311 | dereference going on at some level, so if you haven't built |
312 | your Perl executable using your C compiler's C<-O> flag, performance |
313 | will suffer. If you I<have> built Perl with C<cc -O>, then this |
314 | probably won't matter. |
315 | |
316 | A more serious concern is that unreachable memory with a non-zero |
317 | reference count will not normally get freed. Therefore, this is a bad |
318 | idea: |
319 | |
320 | { |
321 | my $a; |
322 | $a = \$a; |
323 | } |
324 | |
325 | Even thought $a I<should> go away, it can't. When building recursive data |
326 | structures, you'll have to break the self-reference yourself explicitly |
327 | if you don't care to leak. For example, here's a self-referential |
328 | node such as one might use in a sophisticated tree structure: |
329 | |
330 | sub new_node { |
331 | my $self = shift; |
332 | my $class = ref($self) || $self; |
333 | my $node = {}; |
334 | $node->{LEFT} = $node->{RIGHT} = $node; |
335 | $node->{DATA} = [ @_ ]; |
336 | return bless $node => $class; |
337 | } |
338 | |
339 | If you create nodes like that, they (currently) won't go away unless you |
340 | break their self reference yourself. (In other words, this is not to be |
341 | construed as a feature, and you shouldn't depend on it.) |
342 | |
343 | Almost. |
344 | |
345 | When an interpreter thread finally shuts down (usually when your program |
346 | exits), then a rather costly but complete mark-and-sweep style of garbage |
347 | collection is performed, and everything allocated by that thread gets |
348 | destroyed. This is essential to support Perl as an embedded or a |
349 | multithreadable language. For example, this program demonstrates Perl's |
350 | two-phased garbage collection: |
351 | |
352 | #!/usr/bin/perl |
353 | package Subtle; |
354 | |
355 | sub new { |
356 | my $test; |
357 | $test = \$test; |
358 | warn "CREATING " . \$test; |
359 | return bless \$test; |
360 | } |
361 | |
362 | sub DESTROY { |
363 | my $self = shift; |
364 | warn "DESTROYING $self"; |
365 | } |
366 | |
367 | package main; |
368 | |
369 | warn "starting program"; |
370 | { |
371 | my $a = Subtle->new; |
372 | my $b = Subtle->new; |
373 | $$a = 0; # break selfref |
374 | warn "leaving block"; |
375 | } |
376 | |
377 | warn "just exited block"; |
378 | warn "time to die..."; |
379 | exit; |
380 | |
381 | When run as F</tmp/test>, the following output is produced: |
382 | |
383 | starting program at /tmp/test line 18. |
384 | CREATING SCALAR(0x8e5b8) at /tmp/test line 7. |
385 | CREATING SCALAR(0x8e57c) at /tmp/test line 7. |
386 | leaving block at /tmp/test line 23. |
387 | DESTROYING Subtle=SCALAR(0x8e5b8) at /tmp/test line 13. |
388 | just exited block at /tmp/test line 26. |
389 | time to die... at /tmp/test line 27. |
390 | DESTROYING Subtle=SCALAR(0x8e57c) during global destruction. |
391 | |
392 | Notice that "global destruction" bit there? That's the thread |
393 | garbage collector reaching the unreachable. |
394 | |
395 | Objects are always destructed, even when regular refs aren't and in fact |
396 | are destructed in a separate pass before ordinary refs just to try to |
397 | prevent object destructors from using refs that have been themselves |
398 | destructed. Plain refs are only garbage collected if the destruct level |
399 | is greater than 0. You can test the higher levels of global destruction |
400 | by setting the PERL_DESTRUCT_LEVEL environment variable, presuming |
401 | C<-DDEBUGGING> was enabled during perl build time. |
402 | |
403 | A more complete garbage collection strategy will be implemented |
404 | at a future date. |
405 | |
a0d0e21e |
406 | =head1 SEE ALSO |
407 | |
cb1a09d0 |
408 | You should also check out L<perlbot> for other object tricks, traps, and tips, |
409 | as well as L<perlmod> for some style guides on constructing both modules |
410 | and classes. |