Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | =head1 NAME |
d74e8afc | 2 | X<reference> X<pointer> X<data structure> X<structure> X<struct> |
a0d0e21e LW |
3 | |
4 | perlref - Perl references and nested data structures | |
5 | ||
a1e2a320 GS |
6 | =head1 NOTE |
7 | ||
8 | This is complete documentation about all aspects of references. | |
9 | For a shorter, tutorial introduction to just the essential features, | |
10 | see L<perlreftut>. | |
11 | ||
a0d0e21e LW |
12 | =head1 DESCRIPTION |
13 | ||
cb1a09d0 | 14 | Before release 5 of Perl it was difficult to represent complex data |
5a964f20 TC |
15 | structures, because all references had to be symbolic--and even then |
16 | it was difficult to refer to a variable instead of a symbol table entry. | |
17 | Perl now not only makes it easier to use symbolic references to variables, | |
18 | but also lets you have "hard" references to any piece of data or code. | |
19 | Any scalar may hold a hard reference. Because arrays and hashes contain | |
20 | scalars, you can now easily build arrays of arrays, arrays of hashes, | |
21 | hashes of arrays, arrays of hashes of functions, and so on. | |
a0d0e21e LW |
22 | |
23 | Hard references are smart--they keep track of reference counts for you, | |
2d24ed35 | 24 | automatically freeing the thing referred to when its reference count goes |
7c2ea1c7 | 25 | to zero. (Reference counts for values in self-referential or |
2d24ed35 | 26 | cyclic data structures may not go to zero without a little help; see |
2b4f771d | 27 | L</"Circular References"> for a detailed explanation.) |
2d24ed35 CS |
28 | If that thing happens to be an object, the object is destructed. See |
29 | L<perlobj> for more about objects. (In a sense, everything in Perl is an | |
30 | object, but we usually reserve the word for references to objects that | |
31 | have been officially "blessed" into a class package.) | |
32 | ||
33 | Symbolic references are names of variables or other objects, just as a | |
54310121 | 34 | symbolic link in a Unix filesystem contains merely the name of a file. |
d1be9408 | 35 | The C<*glob> notation is something of a symbolic reference. (Symbolic |
2d24ed35 CS |
36 | references are sometimes called "soft references", but please don't call |
37 | them that; references are confusing enough without useless synonyms.) | |
d74e8afc ITB |
38 | X<reference, symbolic> X<reference, soft> |
39 | X<symbolic reference> X<soft reference> | |
2d24ed35 | 40 | |
54310121 | 41 | In contrast, hard references are more like hard links in a Unix file |
2d24ed35 CS |
42 | system: They are used to access an underlying object without concern for |
43 | what its (other) name is. When the word "reference" is used without an | |
5a964f20 | 44 | adjective, as in the following paragraph, it is usually talking about a |
2d24ed35 | 45 | hard reference. |
d74e8afc | 46 | X<reference, hard> X<hard reference> |
2d24ed35 CS |
47 | |
48 | References are easy to use in Perl. There is just one overriding | |
49 | principle: Perl does no implicit referencing or dereferencing. When a | |
50 | scalar is holding a reference, it always behaves as a simple scalar. It | |
51 | doesn't magically start being an array or hash or subroutine; you have to | |
52 | tell it explicitly to do so, by dereferencing it. | |
a0d0e21e | 53 | |
903c0e71 PM |
54 | References are easy to use in Perl. There is just one overriding |
55 | principle: in general, Perl does no implicit referencing or dereferencing. | |
56 | When a scalar is holding a reference, it always behaves as a simple scalar. | |
57 | It doesn't magically start being an array or hash or subroutine; you have to | |
58 | tell it explicitly to do so, by dereferencing it. | |
59 | ||
60 | That said, be aware that Perl version 5.14 introduces an exception | |
61 | to the rule, for syntactic convenience. Experimental array and hash container | |
62 | function behavior allows array and hash references to be handled by Perl as | |
63 | if they had been explicitly syntactically dereferenced. See | |
64 | L<perl5140delta/"Syntactical Enhancements"> | |
65 | and L<perlfunc> for details. | |
66 | ||
5a964f20 | 67 | =head2 Making References |
d74e8afc | 68 | X<reference, creation> X<referencing> |
5a964f20 TC |
69 | |
70 | References can be created in several ways. | |
a0d0e21e LW |
71 | |
72 | =over 4 | |
73 | ||
74 | =item 1. | |
d74e8afc | 75 | X<\> X<backslash> |
a0d0e21e LW |
76 | |
77 | By using the backslash operator on a variable, subroutine, or value. | |
7c2ea1c7 GS |
78 | (This works much like the & (address-of) operator in C.) |
79 | This typically creates I<another> reference to a variable, because | |
a0d0e21e LW |
80 | there's already a reference to the variable in the symbol table. But |
81 | the symbol table reference might go away, and you'll still have the | |
82 | reference that the backslash returned. Here are some examples: | |
83 | ||
84 | $scalarref = \$foo; | |
85 | $arrayref = \@ARGV; | |
86 | $hashref = \%ENV; | |
87 | $coderef = \&handler; | |
55497cff | 88 | $globref = \*foo; |
cb1a09d0 | 89 | |
5a964f20 TC |
90 | It isn't possible to create a true reference to an IO handle (filehandle |
91 | or dirhandle) using the backslash operator. The most you can get is a | |
92 | reference to a typeglob, which is actually a complete symbol table entry. | |
93 | But see the explanation of the C<*foo{THING}> syntax below. However, | |
94 | you can still use type globs and globrefs as though they were IO handles. | |
a0d0e21e LW |
95 | |
96 | =item 2. | |
d74e8afc ITB |
97 | X<array, anonymous> X<[> X<[]> X<square bracket> |
98 | X<bracket, square> X<arrayref> X<array reference> X<reference, array> | |
a0d0e21e | 99 | |
5a964f20 | 100 | A reference to an anonymous array can be created using square |
a0d0e21e LW |
101 | brackets: |
102 | ||
103 | $arrayref = [1, 2, ['a', 'b', 'c']]; | |
104 | ||
5a964f20 | 105 | Here we've created a reference to an anonymous array of three elements |
54310121 | 106 | whose final element is itself a reference to another anonymous array of three |
a0d0e21e | 107 | elements. (The multidimensional syntax described later can be used to |
c47ff5f1 | 108 | access this. For example, after the above, C<< $arrayref->[2][1] >> would have |
a0d0e21e LW |
109 | the value "b".) |
110 | ||
7c2ea1c7 | 111 | Taking a reference to an enumerated list is not the same |
cb1a09d0 AD |
112 | as using square brackets--instead it's the same as creating |
113 | a list of references! | |
114 | ||
54310121 | 115 | @list = (\$a, \@b, \%c); |
58e0a6ae GS |
116 | @list = \($a, @b, %c); # same thing! |
117 | ||
54310121 | 118 | As a special case, C<\(@foo)> returns a list of references to the contents |
b6429b1b GS |
119 | of C<@foo>, not a reference to C<@foo> itself. Likewise for C<%foo>, |
120 | except that the key references are to copies (since the keys are just | |
121 | strings rather than full-fledged scalars). | |
cb1a09d0 | 122 | |
a0d0e21e | 123 | =item 3. |
d74e8afc ITB |
124 | X<hash, anonymous> X<{> X<{}> X<curly bracket> |
125 | X<bracket, curly> X<brace> X<hashref> X<hash reference> X<reference, hash> | |
a0d0e21e | 126 | |
5a964f20 | 127 | A reference to an anonymous hash can be created using curly |
a0d0e21e LW |
128 | brackets: |
129 | ||
130 | $hashref = { | |
131 | 'Adam' => 'Eve', | |
132 | 'Clyde' => 'Bonnie', | |
133 | }; | |
134 | ||
5a964f20 | 135 | Anonymous hash and array composers like these can be intermixed freely to |
a0d0e21e LW |
136 | produce as complicated a structure as you want. The multidimensional |
137 | syntax described below works for these too. The values above are | |
138 | literals, but variables and expressions would work just as well, because | |
139 | assignment operators in Perl (even within local() or my()) are executable | |
140 | statements, not compile-time declarations. | |
141 | ||
142 | Because curly brackets (braces) are used for several other things | |
143 | including BLOCKs, you may occasionally have to disambiguate braces at the | |
144 | beginning of a statement by putting a C<+> or a C<return> in front so | |
145 | that Perl realizes the opening brace isn't starting a BLOCK. The economy and | |
146 | mnemonic value of using curlies is deemed worth this occasional extra | |
147 | hassle. | |
148 | ||
149 | For example, if you wanted a function to make a new hash and return a | |
150 | reference to it, you have these options: | |
151 | ||
152 | sub hashem { { @_ } } # silently wrong | |
153 | sub hashem { +{ @_ } } # ok | |
154 | sub hashem { return { @_ } } # ok | |
155 | ||
ebc58f1a GS |
156 | On the other hand, if you want the other meaning, you can do this: |
157 | ||
158 | sub showem { { @_ } } # ambiguous (currently ok, but may change) | |
159 | sub showem { {; @_ } } # ok | |
160 | sub showem { { return @_ } } # ok | |
161 | ||
7c2ea1c7 | 162 | The leading C<+{> and C<{;> always serve to disambiguate |
ebc58f1a GS |
163 | the expression to mean either the HASH reference, or the BLOCK. |
164 | ||
a0d0e21e | 165 | =item 4. |
d74e8afc ITB |
166 | X<subroutine, anonymous> X<subroutine, reference> X<reference, subroutine> |
167 | X<scope, lexical> X<closure> X<lexical> X<lexical scope> | |
a0d0e21e | 168 | |
5a964f20 | 169 | A reference to an anonymous subroutine can be created by using |
a0d0e21e LW |
170 | C<sub> without a subname: |
171 | ||
172 | $coderef = sub { print "Boink!\n" }; | |
173 | ||
7c2ea1c7 GS |
174 | Note the semicolon. Except for the code |
175 | inside not being immediately executed, a C<sub {}> is not so much a | |
a0d0e21e | 176 | declaration as it is an operator, like C<do{}> or C<eval{}>. (However, no |
5a964f20 | 177 | matter how many times you execute that particular line (unless you're in an |
19799a22 | 178 | C<eval("...")>), $coderef will still have a reference to the I<same> |
a0d0e21e LW |
179 | anonymous subroutine.) |
180 | ||
748a9306 | 181 | Anonymous subroutines act as closures with respect to my() variables, |
7c2ea1c7 | 182 | that is, variables lexically visible within the current scope. Closure |
748a9306 LW |
183 | is a notion out of the Lisp world that says if you define an anonymous |
184 | function in a particular lexical context, it pretends to run in that | |
7c2ea1c7 | 185 | context even when it's called outside the context. |
748a9306 LW |
186 | |
187 | In human terms, it's a funny way of passing arguments to a subroutine when | |
188 | you define it as well as when you call it. It's useful for setting up | |
189 | little bits of code to run later, such as callbacks. You can even | |
54310121 | 190 | do object-oriented stuff with it, though Perl already provides a different |
191 | mechanism to do that--see L<perlobj>. | |
748a9306 | 192 | |
7c2ea1c7 GS |
193 | You might also think of closure as a way to write a subroutine |
194 | template without using eval(). Here's a small example of how | |
195 | closures work: | |
748a9306 LW |
196 | |
197 | sub newprint { | |
198 | my $x = shift; | |
199 | return sub { my $y = shift; print "$x, $y!\n"; }; | |
a0d0e21e | 200 | } |
748a9306 LW |
201 | $h = newprint("Howdy"); |
202 | $g = newprint("Greetings"); | |
203 | ||
204 | # Time passes... | |
205 | ||
206 | &$h("world"); | |
207 | &$g("earthlings"); | |
a0d0e21e | 208 | |
748a9306 LW |
209 | This prints |
210 | ||
211 | Howdy, world! | |
212 | Greetings, earthlings! | |
213 | ||
7c2ea1c7 GS |
214 | Note particularly that $x continues to refer to the value passed |
215 | into newprint() I<despite> "my $x" having gone out of scope by the | |
216 | time the anonymous subroutine runs. That's what a closure is all | |
217 | about. | |
748a9306 | 218 | |
5a964f20 | 219 | This applies only to lexical variables, by the way. Dynamic variables |
748a9306 LW |
220 | continue to work as they have always worked. Closure is not something |
221 | that most Perl programmers need trouble themselves about to begin with. | |
a0d0e21e LW |
222 | |
223 | =item 5. | |
d74e8afc | 224 | X<constructor> X<new> |
a0d0e21e | 225 | |
63acfd00 | 226 | References are often returned by special subroutines called constructors. Perl |
227 | objects are just references to a special type of object that happens to know | |
228 | which package it's associated with. Constructors are just special subroutines | |
229 | that know how to create that association. They do so by starting with an | |
230 | ordinary reference, and it remains an ordinary reference even while it's also | |
231 | being an object. Constructors are often named C<new()>. You I<can> call them | |
232 | indirectly: | |
233 | ||
234 | $objref = new Doggie( Tail => 'short', Ears => 'long' ); | |
235 | ||
236 | But that can produce ambiguous syntax in certain cases, so it's often | |
237 | better to use the direct method invocation approach: | |
5a964f20 TC |
238 | |
239 | $objref = Doggie->new(Tail => 'short', Ears => 'long'); | |
240 | ||
241 | use Term::Cap; | |
242 | $terminal = Term::Cap->Tgetent( { OSPEED => 9600 }); | |
243 | ||
244 | use Tk; | |
245 | $main = MainWindow->new(); | |
246 | $menubar = $main->Frame(-relief => "raised", | |
247 | -borderwidth => 2) | |
248 | ||
a0d0e21e | 249 | =item 6. |
d74e8afc | 250 | X<autovivification> |
a0d0e21e LW |
251 | |
252 | References of the appropriate type can spring into existence if you | |
5f05dabc | 253 | dereference them in a context that assumes they exist. Because we haven't |
a0d0e21e LW |
254 | talked about dereferencing yet, we can't show you any examples yet. |
255 | ||
cb1a09d0 | 256 | =item 7. |
d74e8afc | 257 | X<*foo{THING}> X<*> |
cb1a09d0 | 258 | |
55497cff | 259 | A reference can be created by using a special syntax, lovingly known as |
260 | the *foo{THING} syntax. *foo{THING} returns a reference to the THING | |
261 | slot in *foo (which is the symbol table entry which holds everything | |
262 | known as foo). | |
cb1a09d0 | 263 | |
55497cff | 264 | $scalarref = *foo{SCALAR}; |
265 | $arrayref = *ARGV{ARRAY}; | |
266 | $hashref = *ENV{HASH}; | |
267 | $coderef = *handler{CODE}; | |
36477c24 | 268 | $ioref = *STDIN{IO}; |
55497cff | 269 | $globref = *foo{GLOB}; |
c0bd1adc | 270 | $formatref = *foo{FORMAT}; |
55497cff | 271 | |
7c2ea1c7 GS |
272 | All of these are self-explanatory except for C<*foo{IO}>. It returns |
273 | the IO handle, used for file handles (L<perlfunc/open>), sockets | |
274 | (L<perlfunc/socket> and L<perlfunc/socketpair>), and directory | |
275 | handles (L<perlfunc/opendir>). For compatibility with previous | |
39b99f21 | 276 | versions of Perl, C<*foo{FILEHANDLE}> is a synonym for C<*foo{IO}>, though it |
277 | is deprecated as of 5.8.0. If deprecation warnings are in effect, it will warn | |
278 | of its use. | |
55497cff | 279 | |
7c2ea1c7 GS |
280 | C<*foo{THING}> returns undef if that particular THING hasn't been used yet, |
281 | except in the case of scalars. C<*foo{SCALAR}> returns a reference to an | |
5f05dabc | 282 | anonymous scalar if $foo hasn't been used yet. This might change in a |
283 | future release. | |
284 | ||
7c2ea1c7 | 285 | C<*foo{IO}> is an alternative to the C<*HANDLE> mechanism given in |
5a964f20 TC |
286 | L<perldata/"Typeglobs and Filehandles"> for passing filehandles |
287 | into or out of subroutines, or storing into larger data structures. | |
288 | Its disadvantage is that it won't create a new filehandle for you. | |
7c2ea1c7 GS |
289 | Its advantage is that you have less risk of clobbering more than |
290 | you want to with a typeglob assignment. (It still conflates file | |
291 | and directory handles, though.) However, if you assign the incoming | |
292 | value to a scalar instead of a typeglob as we do in the examples | |
293 | below, there's no risk of that happening. | |
36477c24 | 294 | |
7c2ea1c7 GS |
295 | splutter(*STDOUT); # pass the whole glob |
296 | splutter(*STDOUT{IO}); # pass both file and dir handles | |
5a964f20 | 297 | |
cb1a09d0 AD |
298 | sub splutter { |
299 | my $fh = shift; | |
300 | print $fh "her um well a hmmm\n"; | |
301 | } | |
302 | ||
7c2ea1c7 GS |
303 | $rec = get_rec(*STDIN); # pass the whole glob |
304 | $rec = get_rec(*STDIN{IO}); # pass both file and dir handles | |
5a964f20 | 305 | |
cb1a09d0 AD |
306 | sub get_rec { |
307 | my $fh = shift; | |
308 | return scalar <$fh>; | |
309 | } | |
310 | ||
a0d0e21e LW |
311 | =back |
312 | ||
5a964f20 | 313 | =head2 Using References |
d74e8afc | 314 | X<reference, use> X<dereferencing> X<dereference> |
5a964f20 | 315 | |
a0d0e21e LW |
316 | That's it for creating references. By now you're probably dying to |
317 | know how to use references to get back to your long-lost data. There | |
318 | are several basic methods. | |
319 | ||
320 | =over 4 | |
321 | ||
322 | =item 1. | |
323 | ||
6309d9d9 | 324 | Anywhere you'd put an identifier (or chain of identifiers) as part |
325 | of a variable or subroutine name, you can replace the identifier with | |
326 | a simple scalar variable containing a reference of the correct type: | |
a0d0e21e LW |
327 | |
328 | $bar = $$scalarref; | |
329 | push(@$arrayref, $filename); | |
330 | $$arrayref[0] = "January"; | |
331 | $$hashref{"KEY"} = "VALUE"; | |
332 | &$coderef(1,2,3); | |
cb1a09d0 | 333 | print $globref "output\n"; |
a0d0e21e | 334 | |
19799a22 | 335 | It's important to understand that we are specifically I<not> dereferencing |
a0d0e21e | 336 | C<$arrayref[0]> or C<$hashref{"KEY"}> there. The dereference of the |
19799a22 | 337 | scalar variable happens I<before> it does any key lookups. Anything more |
a0d0e21e LW |
338 | complicated than a simple scalar variable must use methods 2 or 3 below. |
339 | However, a "simple scalar" includes an identifier that itself uses method | |
340 | 1 recursively. Therefore, the following prints "howdy". | |
341 | ||
342 | $refrefref = \\\"howdy"; | |
343 | print $$$$refrefref; | |
344 | ||
345 | =item 2. | |
346 | ||
6309d9d9 | 347 | Anywhere you'd put an identifier (or chain of identifiers) as part of a |
348 | variable or subroutine name, you can replace the identifier with a | |
349 | BLOCK returning a reference of the correct type. In other words, the | |
350 | previous examples could be written like this: | |
a0d0e21e LW |
351 | |
352 | $bar = ${$scalarref}; | |
353 | push(@{$arrayref}, $filename); | |
354 | ${$arrayref}[0] = "January"; | |
355 | ${$hashref}{"KEY"} = "VALUE"; | |
356 | &{$coderef}(1,2,3); | |
36477c24 | 357 | $globref->print("output\n"); # iff IO::Handle is loaded |
a0d0e21e LW |
358 | |
359 | Admittedly, it's a little silly to use the curlies in this case, but | |
360 | the BLOCK can contain any arbitrary expression, in particular, | |
361 | subscripted expressions: | |
362 | ||
54310121 | 363 | &{ $dispatch{$index} }(1,2,3); # call correct routine |
a0d0e21e LW |
364 | |
365 | Because of being able to omit the curlies for the simple case of C<$$x>, | |
366 | people often make the mistake of viewing the dereferencing symbols as | |
367 | proper operators, and wonder about their precedence. If they were, | |
5f05dabc | 368 | though, you could use parentheses instead of braces. That's not the case. |
a0d0e21e | 369 | Consider the difference below; case 0 is a short-hand version of case 1, |
19799a22 | 370 | I<not> case 2: |
a0d0e21e LW |
371 | |
372 | $$hashref{"KEY"} = "VALUE"; # CASE 0 | |
373 | ${$hashref}{"KEY"} = "VALUE"; # CASE 1 | |
374 | ${$hashref{"KEY"}} = "VALUE"; # CASE 2 | |
375 | ${$hashref->{"KEY"}} = "VALUE"; # CASE 3 | |
376 | ||
377 | Case 2 is also deceptive in that you're accessing a variable | |
378 | called %hashref, not dereferencing through $hashref to the hash | |
379 | it's presumably referencing. That would be case 3. | |
380 | ||
381 | =item 3. | |
382 | ||
6da72b64 CS |
383 | Subroutine calls and lookups of individual array elements arise often |
384 | enough that it gets cumbersome to use method 2. As a form of | |
385 | syntactic sugar, the examples for method 2 may be written: | |
a0d0e21e | 386 | |
6da72b64 CS |
387 | $arrayref->[0] = "January"; # Array element |
388 | $hashref->{"KEY"} = "VALUE"; # Hash element | |
389 | $coderef->(1,2,3); # Subroutine call | |
a0d0e21e | 390 | |
6da72b64 | 391 | The left side of the arrow can be any expression returning a reference, |
19799a22 | 392 | including a previous dereference. Note that C<$array[$x]> is I<not> the |
c47ff5f1 | 393 | same thing as C<< $array->[$x] >> here: |
a0d0e21e LW |
394 | |
395 | $array[$x]->{"foo"}->[0] = "January"; | |
396 | ||
397 | This is one of the cases we mentioned earlier in which references could | |
398 | spring into existence when in an lvalue context. Before this | |
399 | statement, C<$array[$x]> may have been undefined. If so, it's | |
400 | automatically defined with a hash reference so that we can look up | |
c47ff5f1 | 401 | C<{"foo"}> in it. Likewise C<< $array[$x]->{"foo"} >> will automatically get |
a0d0e21e | 402 | defined with an array reference so that we can look up C<[0]> in it. |
5a964f20 | 403 | This process is called I<autovivification>. |
a0d0e21e | 404 | |
19799a22 | 405 | One more thing here. The arrow is optional I<between> brackets |
a0d0e21e LW |
406 | subscripts, so you can shrink the above down to |
407 | ||
408 | $array[$x]{"foo"}[0] = "January"; | |
409 | ||
410 | Which, in the degenerate case of using only ordinary arrays, gives you | |
411 | multidimensional arrays just like C's: | |
412 | ||
413 | $score[$x][$y][$z] += 42; | |
414 | ||
415 | Well, okay, not entirely like C's arrays, actually. C doesn't know how | |
416 | to grow its arrays on demand. Perl does. | |
417 | ||
418 | =item 4. | |
419 | ||
420 | If a reference happens to be a reference to an object, then there are | |
421 | probably methods to access the things referred to, and you should probably | |
422 | stick to those methods unless you're in the class package that defines the | |
423 | object's methods. In other words, be nice, and don't violate the object's | |
424 | encapsulation without a very good reason. Perl does not enforce | |
425 | encapsulation. We are not totalitarians here. We do expect some basic | |
426 | civility though. | |
427 | ||
428 | =back | |
429 | ||
7c2ea1c7 GS |
430 | Using a string or number as a reference produces a symbolic reference, |
431 | as explained above. Using a reference as a number produces an | |
432 | integer representing its storage location in memory. The only | |
433 | useful thing to be done with this is to compare two references | |
434 | numerically to see whether they refer to the same location. | |
d74e8afc | 435 | X<reference, numeric context> |
7c2ea1c7 GS |
436 | |
437 | if ($ref1 == $ref2) { # cheap numeric compare of references | |
438 | print "refs 1 and 2 refer to the same thing\n"; | |
439 | } | |
440 | ||
441 | Using a reference as a string produces both its referent's type, | |
442 | including any package blessing as described in L<perlobj>, as well | |
443 | as the numeric address expressed in hex. The ref() operator returns | |
444 | just the type of thing the reference is pointing to, without the | |
445 | address. See L<perlfunc/ref> for details and examples of its use. | |
d74e8afc | 446 | X<reference, string context> |
a0d0e21e | 447 | |
5a964f20 TC |
448 | The bless() operator may be used to associate the object a reference |
449 | points to with a package functioning as an object class. See L<perlobj>. | |
a0d0e21e | 450 | |
5f05dabc | 451 | A typeglob may be dereferenced the same way a reference can, because |
7c2ea1c7 | 452 | the dereference syntax always indicates the type of reference desired. |
a0d0e21e LW |
453 | So C<${*foo}> and C<${\$foo}> both indicate the same scalar variable. |
454 | ||
455 | Here's a trick for interpolating a subroutine call into a string: | |
456 | ||
cb1a09d0 AD |
457 | print "My sub returned @{[mysub(1,2,3)]} that time.\n"; |
458 | ||
459 | The way it works is that when the C<@{...}> is seen in the double-quoted | |
460 | string, it's evaluated as a block. The block creates a reference to an | |
461 | anonymous array containing the results of the call to C<mysub(1,2,3)>. So | |
462 | the whole block returns a reference to an array, which is then | |
463 | dereferenced by C<@{...}> and stuck into the double-quoted string. This | |
464 | chicanery is also useful for arbitrary expressions: | |
a0d0e21e | 465 | |
184e9718 | 466 | print "That yields @{[$n + 5]} widgets\n"; |
a0d0e21e | 467 | |
35efdb20 DL |
468 | Similarly, an expression that returns a reference to a scalar can be |
469 | dereferenced via C<${...}>. Thus, the above expression may be written | |
470 | as: | |
471 | ||
472 | print "That yields ${\($n + 5)} widgets\n"; | |
473 | ||
0a044a7c DR |
474 | =head2 Circular References |
475 | X<circular reference> X<reference, circular> | |
476 | ||
477 | It is possible to create a "circular reference" in Perl, which can lead | |
478 | to memory leaks. A circular reference occurs when two references | |
479 | contain a reference to each other, like this: | |
480 | ||
481 | my $foo = {}; | |
482 | my $bar = { foo => $foo }; | |
483 | $foo->{bar} = $bar; | |
484 | ||
485 | You can also create a circular reference with a single variable: | |
486 | ||
487 | my $foo; | |
488 | $foo = \$foo; | |
489 | ||
490 | In this case, the reference count for the variables will never reach 0, | |
491 | and the references will never be garbage-collected. This can lead to | |
492 | memory leaks. | |
493 | ||
494 | Because objects in Perl are implemented as references, it's possible to | |
495 | have circular references with objects as well. Imagine a TreeNode class | |
496 | where each node references its parent and child nodes. Any node with a | |
497 | parent will be part of a circular reference. | |
498 | ||
499 | You can break circular references by creating a "weak reference". A | |
500 | weak reference does not increment the reference count for a variable, | |
501 | which means that the object can go out of scope and be destroyed. You | |
502 | can weaken a reference with the C<weaken> function exported by the | |
503 | L<Scalar::Util> module. | |
504 | ||
505 | Here's how we can make the first example safer: | |
506 | ||
507 | use Scalar::Util 'weaken'; | |
508 | ||
509 | my $foo = {}; | |
510 | my $bar = { foo => $foo }; | |
511 | $foo->{bar} = $bar; | |
512 | ||
513 | weaken $foo->{bar}; | |
514 | ||
515 | The reference from C<$foo> to C<$bar> has been weakened. When the | |
516 | C<$bar> variable goes out of scope, it will be garbage-collected. The | |
517 | next time you look at the value of the C<< $foo->{bar} >> key, it will | |
518 | be C<undef>. | |
519 | ||
520 | This action at a distance can be confusing, so you should be careful | |
521 | with your use of weaken. You should weaken the reference in the | |
522 | variable that will go out of scope I<first>. That way, the longer-lived | |
523 | variable will contain the expected reference until it goes out of | |
524 | scope. | |
525 | ||
a0d0e21e | 526 | =head2 Symbolic references |
d74e8afc ITB |
527 | X<reference, symbolic> X<reference, soft> |
528 | X<symbolic reference> X<soft reference> | |
a0d0e21e LW |
529 | |
530 | We said that references spring into existence as necessary if they are | |
531 | undefined, but we didn't say what happens if a value used as a | |
19799a22 | 532 | reference is already defined, but I<isn't> a hard reference. If you |
7c2ea1c7 | 533 | use it as a reference, it'll be treated as a symbolic |
19799a22 | 534 | reference. That is, the value of the scalar is taken to be the I<name> |
a0d0e21e LW |
535 | of a variable, rather than a direct link to a (possibly) anonymous |
536 | value. | |
537 | ||
538 | People frequently expect it to work like this. So it does. | |
539 | ||
540 | $name = "foo"; | |
541 | $$name = 1; # Sets $foo | |
542 | ${$name} = 2; # Sets $foo | |
543 | ${$name x 2} = 3; # Sets $foofoo | |
544 | $name->[0] = 4; # Sets $foo[0] | |
545 | @$name = (); # Clears @foo | |
546 | &$name(); # Calls &foo() (as in Perl 4) | |
547 | $pack = "THAT"; | |
548 | ${"${pack}::$name"} = 5; # Sets $THAT::foo without eval | |
549 | ||
7c2ea1c7 | 550 | This is powerful, and slightly dangerous, in that it's possible |
a0d0e21e LW |
551 | to intend (with the utmost sincerity) to use a hard reference, and |
552 | accidentally use a symbolic reference instead. To protect against | |
553 | that, you can say | |
554 | ||
555 | use strict 'refs'; | |
556 | ||
557 | and then only hard references will be allowed for the rest of the enclosing | |
54310121 | 558 | block. An inner block may countermand that with |
a0d0e21e LW |
559 | |
560 | no strict 'refs'; | |
561 | ||
5a964f20 TC |
562 | Only package variables (globals, even if localized) are visible to |
563 | symbolic references. Lexical variables (declared with my()) aren't in | |
564 | a symbol table, and thus are invisible to this mechanism. For example: | |
a0d0e21e | 565 | |
5a964f20 | 566 | local $value = 10; |
b0c35547 | 567 | $ref = "value"; |
a0d0e21e LW |
568 | { |
569 | my $value = 20; | |
570 | print $$ref; | |
54310121 | 571 | } |
a0d0e21e LW |
572 | |
573 | This will still print 10, not 20. Remember that local() affects package | |
574 | variables, which are all "global" to the package. | |
575 | ||
748a9306 LW |
576 | =head2 Not-so-symbolic references |
577 | ||
903c0e71 PM |
578 | Since Perl verion 5.001, brackets around a symbolic reference can simply |
579 | serve to isolate an identifier or variable name from the rest of an | |
580 | expression, just as they always have within a string. For example, | |
748a9306 LW |
581 | |
582 | $push = "pop on "; | |
583 | print "${push}over"; | |
584 | ||
7c2ea1c7 | 585 | has always meant to print "pop on over", even though push is |
903c0e71 PM |
586 | a reserved word. In 5.001, this was generalized to work the same |
587 | without the enclosing double quotes, so that | |
748a9306 LW |
588 | |
589 | print ${push} . "over"; | |
590 | ||
591 | and even | |
592 | ||
593 | print ${ push } . "over"; | |
594 | ||
595 | will have the same effect. (This would have been a syntax error in | |
7c2ea1c7 | 596 | Perl 5.000, though Perl 4 allowed it in the spaceless form.) This |
748a9306 LW |
597 | construct is I<not> considered to be a symbolic reference when you're |
598 | using strict refs: | |
599 | ||
600 | use strict 'refs'; | |
601 | ${ bareword }; # Okay, means $bareword. | |
602 | ${ "bareword" }; # Error, symbolic reference. | |
603 | ||
903c0e71 PM |
604 | Similarly, because of all the subscripting that is done using single words, |
605 | the same rule applies to any bareword that is used for subscripting a hash. | |
606 | So now, instead of writing | |
748a9306 LW |
607 | |
608 | $array{ "aaa" }{ "bbb" }{ "ccc" } | |
609 | ||
5f05dabc | 610 | you can write just |
748a9306 LW |
611 | |
612 | $array{ aaa }{ bbb }{ ccc } | |
613 | ||
614 | and not worry about whether the subscripts are reserved words. In the | |
615 | rare event that you do wish to do something like | |
616 | ||
617 | $array{ shift } | |
618 | ||
619 | you can force interpretation as a reserved word by adding anything that | |
620 | makes it more than a bareword: | |
621 | ||
622 | $array{ shift() } | |
623 | $array{ +shift } | |
624 | $array{ shift @_ } | |
625 | ||
9f1b1f2d GS |
626 | The C<use warnings> pragma or the B<-w> switch will warn you if it |
627 | interprets a reserved word as a string. | |
5f05dabc | 628 | But it will no longer warn you about using lowercase words, because the |
748a9306 LW |
629 | string is effectively quoted. |
630 | ||
49399b3f | 631 | =head2 Pseudo-hashes: Using an array as a hash |
d74e8afc | 632 | X<pseudo-hash> X<pseudo hash> X<pseudohash> |
49399b3f | 633 | |
6d822dc4 MS |
634 | Pseudo-hashes have been removed from Perl. The 'fields' pragma |
635 | remains available. | |
e0478e5a | 636 | |
5a964f20 | 637 | =head2 Function Templates |
d74e8afc ITB |
638 | X<scope, lexical> X<closure> X<lexical> X<lexical scope> |
639 | X<subroutine, nested> X<sub, nested> X<subroutine, local> X<sub, local> | |
5a964f20 | 640 | |
b5c19bd7 DM |
641 | As explained above, an anonymous function with access to the lexical |
642 | variables visible when that function was compiled, creates a closure. It | |
643 | retains access to those variables even though it doesn't get run until | |
644 | later, such as in a signal handler or a Tk callback. | |
5a964f20 TC |
645 | |
646 | Using a closure as a function template allows us to generate many functions | |
c2611fb3 | 647 | that act similarly. Suppose you wanted functions named after the colors |
5a964f20 TC |
648 | that generated HTML font changes for the various colors: |
649 | ||
650 | print "Be ", red("careful"), "with that ", green("light"); | |
651 | ||
7c2ea1c7 | 652 | The red() and green() functions would be similar. To create these, |
5a964f20 TC |
653 | we'll assign a closure to a typeglob of the name of the function we're |
654 | trying to build. | |
655 | ||
656 | @colors = qw(red blue green yellow orange purple violet); | |
657 | for my $name (@colors) { | |
658 | no strict 'refs'; # allow symbol table manipulation | |
659 | *$name = *{uc $name} = sub { "<FONT COLOR='$name'>@_</FONT>" }; | |
660 | } | |
661 | ||
662 | Now all those different functions appear to exist independently. You can | |
663 | call red(), RED(), blue(), BLUE(), green(), etc. This technique saves on | |
664 | both compile time and memory use, and is less error-prone as well, since | |
665 | syntax checks happen at compile time. It's critical that any variables in | |
666 | the anonymous subroutine be lexicals in order to create a proper closure. | |
667 | That's the reasons for the C<my> on the loop iteration variable. | |
668 | ||
669 | This is one of the only places where giving a prototype to a closure makes | |
670 | much sense. If you wanted to impose scalar context on the arguments of | |
671 | these functions (probably not a wise idea for this particular example), | |
672 | you could have written it this way instead: | |
673 | ||
674 | *$name = sub ($) { "<FONT COLOR='$name'>$_[0]</FONT>" }; | |
675 | ||
676 | However, since prototype checking happens at compile time, the assignment | |
677 | above happens too late to be of much use. You could address this by | |
678 | putting the whole loop of assignments within a BEGIN block, forcing it | |
679 | to occur during compilation. | |
680 | ||
58e2a187 CW |
681 | Access to lexicals that change over time--like those in the C<for> loop |
682 | above, basically aliases to elements from the surrounding lexical scopes-- | |
683 | only works with anonymous subs, not with named subroutines. Generally | |
684 | said, named subroutines do not nest properly and should only be declared | |
685 | in the main package scope. | |
686 | ||
687 | This is because named subroutines are created at compile time so their | |
688 | lexical variables get assigned to the parent lexicals from the first | |
689 | execution of the parent block. If a parent scope is entered a second | |
690 | time, its lexicals are created again, while the nested subs still | |
691 | reference the old ones. | |
692 | ||
693 | Anonymous subroutines get to capture each time you execute the C<sub> | |
694 | operator, as they are created on the fly. If you are accustomed to using | |
695 | nested subroutines in other programming languages with their own private | |
696 | variables, you'll have to work at it a bit in Perl. The intuitive coding | |
697 | of this type of thing incurs mysterious warnings about "will not stay | |
698 | shared" due to the reasons explained above. | |
699 | For example, this won't work: | |
5a964f20 TC |
700 | |
701 | sub outer { | |
702 | my $x = $_[0] + 35; | |
703 | sub inner { return $x * 19 } # WRONG | |
704 | return $x + inner(); | |
b432a672 | 705 | } |
5a964f20 TC |
706 | |
707 | A work-around is the following: | |
708 | ||
709 | sub outer { | |
710 | my $x = $_[0] + 35; | |
711 | local *inner = sub { return $x * 19 }; | |
712 | return $x + inner(); | |
b432a672 | 713 | } |
5a964f20 TC |
714 | |
715 | Now inner() can only be called from within outer(), because of the | |
58e2a187 CW |
716 | temporary assignments of the anonymous subroutine. But when it does, |
717 | it has normal access to the lexical variable $x from the scope of | |
718 | outer() at the time outer is invoked. | |
5a964f20 TC |
719 | |
720 | This has the interesting effect of creating a function local to another | |
721 | function, something not normally supported in Perl. | |
722 | ||
cb1a09d0 | 723 | =head1 WARNING |
d74e8afc | 724 | X<reference, string context> X<reference, use as hash key> |
748a9306 LW |
725 | |
726 | You may not (usefully) use a reference as the key to a hash. It will be | |
727 | converted into a string: | |
728 | ||
729 | $x{ \$a } = $a; | |
730 | ||
54310121 | 731 | If you try to dereference the key, it won't do a hard dereference, and |
184e9718 | 732 | you won't accomplish what you're attempting. You might want to do something |
cb1a09d0 | 733 | more like |
748a9306 | 734 | |
cb1a09d0 AD |
735 | $r = \@a; |
736 | $x{ $r } = $r; | |
737 | ||
738 | And then at least you can use the values(), which will be | |
739 | real refs, instead of the keys(), which won't. | |
740 | ||
5a964f20 TC |
741 | The standard Tie::RefHash module provides a convenient workaround to this. |
742 | ||
cb1a09d0 | 743 | =head1 SEE ALSO |
a0d0e21e LW |
744 | |
745 | Besides the obvious documents, source code can be instructive. | |
7c2ea1c7 | 746 | Some pathological examples of the use of references can be found |
a0d0e21e | 747 | in the F<t/op/ref.t> regression test in the Perl source directory. |
cb1a09d0 AD |
748 | |
749 | See also L<perldsc> and L<perllol> for how to use references to create | |
82e1c0d9 | 750 | complex data structures, and L<perlootut> and L<perlobj> |
5a964f20 | 751 | for how to use them to create objects. |