This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Re: Should keys in pseudo-hashes -always- exist? [DOC PATCH]
[perl5.git] / pod / perlref.pod
CommitLineData
a0d0e21e
LW
1=head1 NAME
2
3perlref - Perl references and nested data structures
4
a1e2a320
GS
5=head1 NOTE
6
7This is complete documentation about all aspects of references.
8For a shorter, tutorial introduction to just the essential features,
9see L<perlreftut>.
10
a0d0e21e
LW
11=head1 DESCRIPTION
12
cb1a09d0 13Before release 5 of Perl it was difficult to represent complex data
5a964f20
TC
14structures, because all references had to be symbolic--and even then
15it was difficult to refer to a variable instead of a symbol table entry.
16Perl now not only makes it easier to use symbolic references to variables,
17but also lets you have "hard" references to any piece of data or code.
18Any scalar may hold a hard reference. Because arrays and hashes contain
19scalars, you can now easily build arrays of arrays, arrays of hashes,
20hashes of arrays, arrays of hashes of functions, and so on.
a0d0e21e
LW
21
22Hard references are smart--they keep track of reference counts for you,
2d24ed35 23automatically freeing the thing referred to when its reference count goes
7c2ea1c7 24to zero. (Reference counts for values in self-referential or
2d24ed35 25cyclic data structures may not go to zero without a little help; see
7b8d334a 26L<perlobj/"Two-Phased Garbage Collection"> for a detailed explanation.)
2d24ed35
CS
27If that thing happens to be an object, the object is destructed. See
28L<perlobj> for more about objects. (In a sense, everything in Perl is an
29object, but we usually reserve the word for references to objects that
30have been officially "blessed" into a class package.)
31
32Symbolic references are names of variables or other objects, just as a
54310121 33symbolic link in a Unix filesystem contains merely the name of a file.
7c2ea1c7 34The C<*glob> notation is something of a of symbolic reference. (Symbolic
2d24ed35
CS
35references are sometimes called "soft references", but please don't call
36them that; references are confusing enough without useless synonyms.)
37
54310121 38In contrast, hard references are more like hard links in a Unix file
2d24ed35
CS
39system: They are used to access an underlying object without concern for
40what its (other) name is. When the word "reference" is used without an
5a964f20 41adjective, as in the following paragraph, it is usually talking about a
2d24ed35
CS
42hard reference.
43
44References are easy to use in Perl. There is just one overriding
45principle: Perl does no implicit referencing or dereferencing. When a
46scalar is holding a reference, it always behaves as a simple scalar. It
47doesn't magically start being an array or hash or subroutine; you have to
48tell it explicitly to do so, by dereferencing it.
a0d0e21e 49
5a964f20
TC
50=head2 Making References
51
52References can be created in several ways.
a0d0e21e
LW
53
54=over 4
55
56=item 1.
57
58By using the backslash operator on a variable, subroutine, or value.
7c2ea1c7
GS
59(This works much like the & (address-of) operator in C.)
60This typically creates I<another> reference to a variable, because
a0d0e21e
LW
61there's already a reference to the variable in the symbol table. But
62the symbol table reference might go away, and you'll still have the
63reference that the backslash returned. Here are some examples:
64
65 $scalarref = \$foo;
66 $arrayref = \@ARGV;
67 $hashref = \%ENV;
68 $coderef = \&handler;
55497cff 69 $globref = \*foo;
cb1a09d0 70
5a964f20
TC
71It isn't possible to create a true reference to an IO handle (filehandle
72or dirhandle) using the backslash operator. The most you can get is a
73reference to a typeglob, which is actually a complete symbol table entry.
74But see the explanation of the C<*foo{THING}> syntax below. However,
75you can still use type globs and globrefs as though they were IO handles.
a0d0e21e
LW
76
77=item 2.
78
5a964f20 79A reference to an anonymous array can be created using square
a0d0e21e
LW
80brackets:
81
82 $arrayref = [1, 2, ['a', 'b', 'c']];
83
5a964f20 84Here we've created a reference to an anonymous array of three elements
54310121 85whose final element is itself a reference to another anonymous array of three
a0d0e21e 86elements. (The multidimensional syntax described later can be used to
184e9718 87access this. For example, after the above, C<$arrayref-E<gt>[2][1]> would have
a0d0e21e
LW
88the value "b".)
89
7c2ea1c7 90Taking a reference to an enumerated list is not the same
cb1a09d0
AD
91as using square brackets--instead it's the same as creating
92a list of references!
93
54310121 94 @list = (\$a, \@b, \%c);
58e0a6ae
GS
95 @list = \($a, @b, %c); # same thing!
96
54310121 97As a special case, C<\(@foo)> returns a list of references to the contents
b6429b1b
GS
98of C<@foo>, not a reference to C<@foo> itself. Likewise for C<%foo>,
99except that the key references are to copies (since the keys are just
100strings rather than full-fledged scalars).
cb1a09d0 101
a0d0e21e
LW
102=item 3.
103
5a964f20 104A reference to an anonymous hash can be created using curly
a0d0e21e
LW
105brackets:
106
107 $hashref = {
108 'Adam' => 'Eve',
109 'Clyde' => 'Bonnie',
110 };
111
5a964f20 112Anonymous hash and array composers like these can be intermixed freely to
a0d0e21e
LW
113produce as complicated a structure as you want. The multidimensional
114syntax described below works for these too. The values above are
115literals, but variables and expressions would work just as well, because
116assignment operators in Perl (even within local() or my()) are executable
117statements, not compile-time declarations.
118
119Because curly brackets (braces) are used for several other things
120including BLOCKs, you may occasionally have to disambiguate braces at the
121beginning of a statement by putting a C<+> or a C<return> in front so
122that Perl realizes the opening brace isn't starting a BLOCK. The economy and
123mnemonic value of using curlies is deemed worth this occasional extra
124hassle.
125
126For example, if you wanted a function to make a new hash and return a
127reference to it, you have these options:
128
129 sub hashem { { @_ } } # silently wrong
130 sub hashem { +{ @_ } } # ok
131 sub hashem { return { @_ } } # ok
132
ebc58f1a
GS
133On the other hand, if you want the other meaning, you can do this:
134
135 sub showem { { @_ } } # ambiguous (currently ok, but may change)
136 sub showem { {; @_ } } # ok
137 sub showem { { return @_ } } # ok
138
7c2ea1c7 139The leading C<+{> and C<{;> always serve to disambiguate
ebc58f1a
GS
140the expression to mean either the HASH reference, or the BLOCK.
141
a0d0e21e
LW
142=item 4.
143
5a964f20 144A reference to an anonymous subroutine can be created by using
a0d0e21e
LW
145C<sub> without a subname:
146
147 $coderef = sub { print "Boink!\n" };
148
7c2ea1c7
GS
149Note the semicolon. Except for the code
150inside not being immediately executed, a C<sub {}> is not so much a
a0d0e21e 151declaration as it is an operator, like C<do{}> or C<eval{}>. (However, no
5a964f20 152matter how many times you execute that particular line (unless you're in an
19799a22 153C<eval("...")>), $coderef will still have a reference to the I<same>
a0d0e21e
LW
154anonymous subroutine.)
155
748a9306 156Anonymous subroutines act as closures with respect to my() variables,
7c2ea1c7 157that is, variables lexically visible within the current scope. Closure
748a9306
LW
158is a notion out of the Lisp world that says if you define an anonymous
159function in a particular lexical context, it pretends to run in that
7c2ea1c7 160context even when it's called outside the context.
748a9306
LW
161
162In human terms, it's a funny way of passing arguments to a subroutine when
163you define it as well as when you call it. It's useful for setting up
164little bits of code to run later, such as callbacks. You can even
54310121
PP
165do object-oriented stuff with it, though Perl already provides a different
166mechanism to do that--see L<perlobj>.
748a9306 167
7c2ea1c7
GS
168You might also think of closure as a way to write a subroutine
169template without using eval(). Here's a small example of how
170closures work:
748a9306
LW
171
172 sub newprint {
173 my $x = shift;
174 return sub { my $y = shift; print "$x, $y!\n"; };
a0d0e21e 175 }
748a9306
LW
176 $h = newprint("Howdy");
177 $g = newprint("Greetings");
178
179 # Time passes...
180
181 &$h("world");
182 &$g("earthlings");
a0d0e21e 183
748a9306
LW
184This prints
185
186 Howdy, world!
187 Greetings, earthlings!
188
7c2ea1c7
GS
189Note particularly that $x continues to refer to the value passed
190into newprint() I<despite> "my $x" having gone out of scope by the
191time the anonymous subroutine runs. That's what a closure is all
192about.
748a9306 193
5a964f20 194This applies only to lexical variables, by the way. Dynamic variables
748a9306
LW
195continue to work as they have always worked. Closure is not something
196that most Perl programmers need trouble themselves about to begin with.
a0d0e21e
LW
197
198=item 5.
199
200References are often returned by special subroutines called constructors.
7c2ea1c7 201Perl objects are just references to a special type of object that happens to know
a0d0e21e
LW
202which package it's associated with. Constructors are just special
203subroutines that know how to create that association. They do so by
204starting with an ordinary reference, and it remains an ordinary reference
5a964f20
TC
205even while it's also being an object. Constructors are often
206named new() and called indirectly:
a0d0e21e
LW
207
208 $objref = new Doggie (Tail => 'short', Ears => 'long');
209
5a964f20
TC
210But don't have to be:
211
212 $objref = Doggie->new(Tail => 'short', Ears => 'long');
213
214 use Term::Cap;
215 $terminal = Term::Cap->Tgetent( { OSPEED => 9600 });
216
217 use Tk;
218 $main = MainWindow->new();
219 $menubar = $main->Frame(-relief => "raised",
220 -borderwidth => 2)
221
a0d0e21e
LW
222=item 6.
223
224References of the appropriate type can spring into existence if you
5f05dabc 225dereference them in a context that assumes they exist. Because we haven't
a0d0e21e
LW
226talked about dereferencing yet, we can't show you any examples yet.
227
cb1a09d0
AD
228=item 7.
229
55497cff
PP
230A reference can be created by using a special syntax, lovingly known as
231the *foo{THING} syntax. *foo{THING} returns a reference to the THING
232slot in *foo (which is the symbol table entry which holds everything
233known as foo).
cb1a09d0 234
55497cff
PP
235 $scalarref = *foo{SCALAR};
236 $arrayref = *ARGV{ARRAY};
237 $hashref = *ENV{HASH};
238 $coderef = *handler{CODE};
36477c24 239 $ioref = *STDIN{IO};
55497cff
PP
240 $globref = *foo{GLOB};
241
7c2ea1c7
GS
242All of these are self-explanatory except for C<*foo{IO}>. It returns
243the IO handle, used for file handles (L<perlfunc/open>), sockets
244(L<perlfunc/socket> and L<perlfunc/socketpair>), and directory
245handles (L<perlfunc/opendir>). For compatibility with previous
246versions of Perl, C<*foo{FILEHANDLE}> is a synonym for C<*foo{IO}>.
55497cff 247
7c2ea1c7
GS
248C<*foo{THING}> returns undef if that particular THING hasn't been used yet,
249except in the case of scalars. C<*foo{SCALAR}> returns a reference to an
5f05dabc
PP
250anonymous scalar if $foo hasn't been used yet. This might change in a
251future release.
252
7c2ea1c7 253C<*foo{IO}> is an alternative to the C<*HANDLE> mechanism given in
5a964f20
TC
254L<perldata/"Typeglobs and Filehandles"> for passing filehandles
255into or out of subroutines, or storing into larger data structures.
256Its disadvantage is that it won't create a new filehandle for you.
7c2ea1c7
GS
257Its advantage is that you have less risk of clobbering more than
258you want to with a typeglob assignment. (It still conflates file
259and directory handles, though.) However, if you assign the incoming
260value to a scalar instead of a typeglob as we do in the examples
261below, there's no risk of that happening.
36477c24 262
7c2ea1c7
GS
263 splutter(*STDOUT); # pass the whole glob
264 splutter(*STDOUT{IO}); # pass both file and dir handles
5a964f20 265
cb1a09d0
AD
266 sub splutter {
267 my $fh = shift;
268 print $fh "her um well a hmmm\n";
269 }
270
7c2ea1c7
GS
271 $rec = get_rec(*STDIN); # pass the whole glob
272 $rec = get_rec(*STDIN{IO}); # pass both file and dir handles
5a964f20 273
cb1a09d0
AD
274 sub get_rec {
275 my $fh = shift;
276 return scalar <$fh>;
277 }
278
a0d0e21e
LW
279=back
280
5a964f20
TC
281=head2 Using References
282
a0d0e21e
LW
283That's it for creating references. By now you're probably dying to
284know how to use references to get back to your long-lost data. There
285are several basic methods.
286
287=over 4
288
289=item 1.
290
6309d9d9
PP
291Anywhere you'd put an identifier (or chain of identifiers) as part
292of a variable or subroutine name, you can replace the identifier with
293a simple scalar variable containing a reference of the correct type:
a0d0e21e
LW
294
295 $bar = $$scalarref;
296 push(@$arrayref, $filename);
297 $$arrayref[0] = "January";
298 $$hashref{"KEY"} = "VALUE";
299 &$coderef(1,2,3);
cb1a09d0 300 print $globref "output\n";
a0d0e21e 301
19799a22 302It's important to understand that we are specifically I<not> dereferencing
a0d0e21e 303C<$arrayref[0]> or C<$hashref{"KEY"}> there. The dereference of the
19799a22 304scalar variable happens I<before> it does any key lookups. Anything more
a0d0e21e
LW
305complicated than a simple scalar variable must use methods 2 or 3 below.
306However, a "simple scalar" includes an identifier that itself uses method
3071 recursively. Therefore, the following prints "howdy".
308
309 $refrefref = \\\"howdy";
310 print $$$$refrefref;
311
312=item 2.
313
6309d9d9
PP
314Anywhere you'd put an identifier (or chain of identifiers) as part of a
315variable or subroutine name, you can replace the identifier with a
316BLOCK returning a reference of the correct type. In other words, the
317previous examples could be written like this:
a0d0e21e
LW
318
319 $bar = ${$scalarref};
320 push(@{$arrayref}, $filename);
321 ${$arrayref}[0] = "January";
322 ${$hashref}{"KEY"} = "VALUE";
323 &{$coderef}(1,2,3);
36477c24 324 $globref->print("output\n"); # iff IO::Handle is loaded
a0d0e21e
LW
325
326Admittedly, it's a little silly to use the curlies in this case, but
327the BLOCK can contain any arbitrary expression, in particular,
328subscripted expressions:
329
54310121 330 &{ $dispatch{$index} }(1,2,3); # call correct routine
a0d0e21e
LW
331
332Because of being able to omit the curlies for the simple case of C<$$x>,
333people often make the mistake of viewing the dereferencing symbols as
334proper operators, and wonder about their precedence. If they were,
5f05dabc 335though, you could use parentheses instead of braces. That's not the case.
a0d0e21e 336Consider the difference below; case 0 is a short-hand version of case 1,
19799a22 337I<not> case 2:
a0d0e21e
LW
338
339 $$hashref{"KEY"} = "VALUE"; # CASE 0
340 ${$hashref}{"KEY"} = "VALUE"; # CASE 1
341 ${$hashref{"KEY"}} = "VALUE"; # CASE 2
342 ${$hashref->{"KEY"}} = "VALUE"; # CASE 3
343
344Case 2 is also deceptive in that you're accessing a variable
345called %hashref, not dereferencing through $hashref to the hash
346it's presumably referencing. That would be case 3.
347
348=item 3.
349
6da72b64
CS
350Subroutine calls and lookups of individual array elements arise often
351enough that it gets cumbersome to use method 2. As a form of
352syntactic sugar, the examples for method 2 may be written:
a0d0e21e 353
6da72b64
CS
354 $arrayref->[0] = "January"; # Array element
355 $hashref->{"KEY"} = "VALUE"; # Hash element
356 $coderef->(1,2,3); # Subroutine call
a0d0e21e 357
6da72b64 358The left side of the arrow can be any expression returning a reference,
19799a22 359including a previous dereference. Note that C<$array[$x]> is I<not> the
a0d0e21e
LW
360same thing as C<$array-E<gt>[$x]> here:
361
362 $array[$x]->{"foo"}->[0] = "January";
363
364This is one of the cases we mentioned earlier in which references could
365spring into existence when in an lvalue context. Before this
366statement, C<$array[$x]> may have been undefined. If so, it's
367automatically defined with a hash reference so that we can look up
368C<{"foo"}> in it. Likewise C<$array[$x]-E<gt>{"foo"}> will automatically get
369defined with an array reference so that we can look up C<[0]> in it.
5a964f20 370This process is called I<autovivification>.
a0d0e21e 371
19799a22 372One more thing here. The arrow is optional I<between> brackets
a0d0e21e
LW
373subscripts, so you can shrink the above down to
374
375 $array[$x]{"foo"}[0] = "January";
376
377Which, in the degenerate case of using only ordinary arrays, gives you
378multidimensional arrays just like C's:
379
380 $score[$x][$y][$z] += 42;
381
382Well, okay, not entirely like C's arrays, actually. C doesn't know how
383to grow its arrays on demand. Perl does.
384
385=item 4.
386
387If a reference happens to be a reference to an object, then there are
388probably methods to access the things referred to, and you should probably
389stick to those methods unless you're in the class package that defines the
390object's methods. In other words, be nice, and don't violate the object's
391encapsulation without a very good reason. Perl does not enforce
392encapsulation. We are not totalitarians here. We do expect some basic
393civility though.
394
395=back
396
7c2ea1c7
GS
397Using a string or number as a reference produces a symbolic reference,
398as explained above. Using a reference as a number produces an
399integer representing its storage location in memory. The only
400useful thing to be done with this is to compare two references
401numerically to see whether they refer to the same location.
402
403 if ($ref1 == $ref2) { # cheap numeric compare of references
404 print "refs 1 and 2 refer to the same thing\n";
405 }
406
407Using a reference as a string produces both its referent's type,
408including any package blessing as described in L<perlobj>, as well
409as the numeric address expressed in hex. The ref() operator returns
410just the type of thing the reference is pointing to, without the
411address. See L<perlfunc/ref> for details and examples of its use.
a0d0e21e 412
5a964f20
TC
413The bless() operator may be used to associate the object a reference
414points to with a package functioning as an object class. See L<perlobj>.
a0d0e21e 415
5f05dabc 416A typeglob may be dereferenced the same way a reference can, because
7c2ea1c7 417the dereference syntax always indicates the type of reference desired.
a0d0e21e
LW
418So C<${*foo}> and C<${\$foo}> both indicate the same scalar variable.
419
420Here's a trick for interpolating a subroutine call into a string:
421
cb1a09d0
AD
422 print "My sub returned @{[mysub(1,2,3)]} that time.\n";
423
424The way it works is that when the C<@{...}> is seen in the double-quoted
425string, it's evaluated as a block. The block creates a reference to an
426anonymous array containing the results of the call to C<mysub(1,2,3)>. So
427the whole block returns a reference to an array, which is then
428dereferenced by C<@{...}> and stuck into the double-quoted string. This
429chicanery is also useful for arbitrary expressions:
a0d0e21e 430
184e9718 431 print "That yields @{[$n + 5]} widgets\n";
a0d0e21e
LW
432
433=head2 Symbolic references
434
435We said that references spring into existence as necessary if they are
436undefined, but we didn't say what happens if a value used as a
19799a22 437reference is already defined, but I<isn't> a hard reference. If you
7c2ea1c7 438use it as a reference, it'll be treated as a symbolic
19799a22 439reference. That is, the value of the scalar is taken to be the I<name>
a0d0e21e
LW
440of a variable, rather than a direct link to a (possibly) anonymous
441value.
442
443People frequently expect it to work like this. So it does.
444
445 $name = "foo";
446 $$name = 1; # Sets $foo
447 ${$name} = 2; # Sets $foo
448 ${$name x 2} = 3; # Sets $foofoo
449 $name->[0] = 4; # Sets $foo[0]
450 @$name = (); # Clears @foo
451 &$name(); # Calls &foo() (as in Perl 4)
452 $pack = "THAT";
453 ${"${pack}::$name"} = 5; # Sets $THAT::foo without eval
454
7c2ea1c7 455This is powerful, and slightly dangerous, in that it's possible
a0d0e21e
LW
456to intend (with the utmost sincerity) to use a hard reference, and
457accidentally use a symbolic reference instead. To protect against
458that, you can say
459
460 use strict 'refs';
461
462and then only hard references will be allowed for the rest of the enclosing
54310121 463block. An inner block may countermand that with
a0d0e21e
LW
464
465 no strict 'refs';
466
5a964f20
TC
467Only package variables (globals, even if localized) are visible to
468symbolic references. Lexical variables (declared with my()) aren't in
469a symbol table, and thus are invisible to this mechanism. For example:
a0d0e21e 470
5a964f20 471 local $value = 10;
b0c35547 472 $ref = "value";
a0d0e21e
LW
473 {
474 my $value = 20;
475 print $$ref;
54310121 476 }
a0d0e21e
LW
477
478This will still print 10, not 20. Remember that local() affects package
479variables, which are all "global" to the package.
480
748a9306
LW
481=head2 Not-so-symbolic references
482
a6006777
PP
483A new feature contributing to readability in perl version 5.001 is that the
484brackets around a symbolic reference behave more like quotes, just as they
748a9306
LW
485always have within a string. That is,
486
487 $push = "pop on ";
488 print "${push}over";
489
7c2ea1c7 490has always meant to print "pop on over", even though push is
748a9306
LW
491a reserved word. This has been generalized to work the same outside
492of quotes, so that
493
494 print ${push} . "over";
495
496and even
497
498 print ${ push } . "over";
499
500will have the same effect. (This would have been a syntax error in
7c2ea1c7 501Perl 5.000, though Perl 4 allowed it in the spaceless form.) This
748a9306
LW
502construct is I<not> considered to be a symbolic reference when you're
503using strict refs:
504
505 use strict 'refs';
506 ${ bareword }; # Okay, means $bareword.
507 ${ "bareword" }; # Error, symbolic reference.
508
509Similarly, because of all the subscripting that is done using single
510words, we've applied the same rule to any bareword that is used for
511subscripting a hash. So now, instead of writing
512
513 $array{ "aaa" }{ "bbb" }{ "ccc" }
514
5f05dabc 515you can write just
748a9306
LW
516
517 $array{ aaa }{ bbb }{ ccc }
518
519and not worry about whether the subscripts are reserved words. In the
520rare event that you do wish to do something like
521
522 $array{ shift }
523
524you can force interpretation as a reserved word by adding anything that
525makes it more than a bareword:
526
527 $array{ shift() }
528 $array{ +shift }
529 $array{ shift @_ }
530
531The B<-w> switch will warn you if it interprets a reserved word as a string.
5f05dabc 532But it will no longer warn you about using lowercase words, because the
748a9306
LW
533string is effectively quoted.
534
49399b3f
GA
535=head2 Pseudo-hashes: Using an array as a hash
536
7c2ea1c7 537B<WARNING>: This section describes an experimental feature. Details may
49399b3f
GA
538change without notice in future versions.
539
7c2ea1c7 540Beginning with release 5.005 of Perl, you may use an array reference
49399b3f
GA
541in some contexts that would normally require a hash reference. This
542allows you to access array elements using symbolic names, as if they
543were fields in a structure.
544
545For this to work, the array must contain extra information. The first
546element of the array has to be a hash reference that maps field names
547to array indices. Here is an example:
548
549 $struct = [{foo => 1, bar => 2}, "FOO", "BAR"];
550
551 $struct->{foo}; # same as $struct->[1], i.e. "FOO"
552 $struct->{bar}; # same as $struct->[2], i.e. "BAR"
553
554 keys %$struct; # will return ("foo", "bar") in some order
555 values %$struct; # will return ("FOO", "BAR") in same some order
556
557 while (my($k,$v) = each %$struct) {
558 print "$k => $v\n";
559 }
560
561Perl will raise an exception if you try to delete keys from a pseudo-hash
562or try to access nonexistent fields. For better performance, Perl can also
563do the translation from field names to array indices at compile time for
564typed object references. See L<fields>.
565
e0478e5a
MS
566There are two ways to check for the existance of a key in a
567pseudo-hash. The first is to use exists(). This checks to see if the
568given field has been used yet. It acts this way to match the behavior
569of a regular hash. For instance:
570
571 $phash = [{foo =>, bar => 2, pants => 3}, 'FOO'];
572 $phash->{pants} = undef;
573
574 exists $phash->{foo}; # true, 'foo' was set in the declaration
575 exists $phash->{bar}; # false, 'bar' has not been used.
576 exists $phash->{pants}; # true, your 'pants' have been touched
577
578The second is to use exists() on the hash reference sitting in the
579first array element. This checks to see if the given key is a valid
580field in the pseudo-hash.
581
582 exists $phash->[0]{pants}; # true, 'pants' is a valid field
583 exists $phash->[0]{shoes}; # false, 'shoes' can't be used
584
5a964f20
TC
585=head2 Function Templates
586
587As explained above, a closure is an anonymous function with access to the
588lexical variables visible when that function was compiled. It retains
589access to those variables even though it doesn't get run until later,
590such as in a signal handler or a Tk callback.
591
592Using a closure as a function template allows us to generate many functions
c2611fb3 593that act similarly. Suppose you wanted functions named after the colors
5a964f20
TC
594that generated HTML font changes for the various colors:
595
596 print "Be ", red("careful"), "with that ", green("light");
597
7c2ea1c7 598The red() and green() functions would be similar. To create these,
5a964f20
TC
599we'll assign a closure to a typeglob of the name of the function we're
600trying to build.
601
602 @colors = qw(red blue green yellow orange purple violet);
603 for my $name (@colors) {
604 no strict 'refs'; # allow symbol table manipulation
605 *$name = *{uc $name} = sub { "<FONT COLOR='$name'>@_</FONT>" };
606 }
607
608Now all those different functions appear to exist independently. You can
609call red(), RED(), blue(), BLUE(), green(), etc. This technique saves on
610both compile time and memory use, and is less error-prone as well, since
611syntax checks happen at compile time. It's critical that any variables in
612the anonymous subroutine be lexicals in order to create a proper closure.
613That's the reasons for the C<my> on the loop iteration variable.
614
615This is one of the only places where giving a prototype to a closure makes
616much sense. If you wanted to impose scalar context on the arguments of
617these functions (probably not a wise idea for this particular example),
618you could have written it this way instead:
619
620 *$name = sub ($) { "<FONT COLOR='$name'>$_[0]</FONT>" };
621
622However, since prototype checking happens at compile time, the assignment
623above happens too late to be of much use. You could address this by
624putting the whole loop of assignments within a BEGIN block, forcing it
625to occur during compilation.
626
627Access to lexicals that change over type--like those in the C<for> loop
628above--only works with closures, not general subroutines. In the general
629case, then, named subroutines do not nest properly, although anonymous
630ones do. If you are accustomed to using nested subroutines in other
631programming languages with their own private variables, you'll have to
7c2ea1c7 632work at it a bit in Perl. The intuitive coding of this type of thing
5a964f20
TC
633incurs mysterious warnings about ``will not stay shared''. For example,
634this won't work:
635
636 sub outer {
637 my $x = $_[0] + 35;
638 sub inner { return $x * 19 } # WRONG
639 return $x + inner();
640 }
641
642A work-around is the following:
643
644 sub outer {
645 my $x = $_[0] + 35;
646 local *inner = sub { return $x * 19 };
647 return $x + inner();
648 }
649
650Now inner() can only be called from within outer(), because of the
651temporary assignments of the closure (anonymous subroutine). But when
652it does, it has normal access to the lexical variable $x from the scope
653of outer().
654
655This has the interesting effect of creating a function local to another
656function, something not normally supported in Perl.
657
cb1a09d0 658=head1 WARNING
748a9306
LW
659
660You may not (usefully) use a reference as the key to a hash. It will be
661converted into a string:
662
663 $x{ \$a } = $a;
664
54310121 665If you try to dereference the key, it won't do a hard dereference, and
184e9718 666you won't accomplish what you're attempting. You might want to do something
cb1a09d0 667more like
748a9306 668
cb1a09d0
AD
669 $r = \@a;
670 $x{ $r } = $r;
671
672And then at least you can use the values(), which will be
673real refs, instead of the keys(), which won't.
674
5a964f20
TC
675The standard Tie::RefHash module provides a convenient workaround to this.
676
cb1a09d0 677=head1 SEE ALSO
a0d0e21e
LW
678
679Besides the obvious documents, source code can be instructive.
7c2ea1c7 680Some pathological examples of the use of references can be found
a0d0e21e 681in the F<t/op/ref.t> regression test in the Perl source directory.
cb1a09d0
AD
682
683See also L<perldsc> and L<perllol> for how to use references to create
5a964f20
TC
684complex data structures, and L<perltoot>, L<perlobj>, and L<perlbot>
685for how to use them to create objects.