This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
some method calls not autoloaded
[perl5.git] / pod / perlsub.pod
CommitLineData
a0d0e21e
LW
1=head1 NAME
2
3perlsub - Perl subroutines
4
5=head1 SYNOPSIS
6
7To declare subroutines:
8
09bef843
SB
9 sub NAME; # A "forward" declaration.
10 sub NAME(PROTO); # ditto, but with prototypes
11 sub NAME : ATTRS; # with attributes
12 sub NAME(PROTO) : ATTRS; # with attributes and prototypes
cb1a09d0 13
09bef843
SB
14 sub NAME BLOCK # A declaration and a definition.
15 sub NAME(PROTO) BLOCK # ditto, but with prototypes
16 sub NAME : ATTRS BLOCK # with attributes
17 sub NAME(PROTO) : ATTRS BLOCK # with prototypes and attributes
a0d0e21e 18
748a9306
LW
19To define an anonymous subroutine at runtime:
20
09bef843
SB
21 $subref = sub BLOCK; # no proto
22 $subref = sub (PROTO) BLOCK; # with proto
23 $subref = sub : ATTRS BLOCK; # with attributes
24 $subref = sub (PROTO) : ATTRS BLOCK; # with proto and attributes
748a9306 25
a0d0e21e
LW
26To import subroutines:
27
19799a22 28 use MODULE qw(NAME1 NAME2 NAME3);
a0d0e21e
LW
29
30To call subroutines:
31
5f05dabc 32 NAME(LIST); # & is optional with parentheses.
54310121 33 NAME LIST; # Parentheses optional if predeclared/imported.
19799a22 34 &NAME(LIST); # Circumvent prototypes.
5a964f20 35 &NAME; # Makes current @_ visible to called subroutine.
a0d0e21e
LW
36
37=head1 DESCRIPTION
38
19799a22
GS
39Like many languages, Perl provides for user-defined subroutines.
40These may be located anywhere in the main program, loaded in from
41other files via the C<do>, C<require>, or C<use> keywords, or
be3174d2 42generated on the fly using C<eval> or anonymous subroutines.
19799a22
GS
43You can even call a function indirectly using a variable containing
44its name or a CODE reference.
cb1a09d0
AD
45
46The Perl model for function call and return values is simple: all
47functions are passed as parameters one single flat list of scalars, and
48all functions likewise return to their caller one single flat list of
49scalars. Any arrays or hashes in these call and return lists will
50collapse, losing their identities--but you may always use
51pass-by-reference instead to avoid this. Both call and return lists may
52contain as many or as few scalar elements as you'd like. (Often a
53function without an explicit return statement is called a subroutine, but
19799a22
GS
54there's really no difference from Perl's perspective.)
55
56Any arguments passed in show up in the array C<@_>. Therefore, if
57you called a function with two arguments, those would be stored in
58C<$_[0]> and C<$_[1]>. The array C<@_> is a local array, but its
59elements are aliases for the actual scalar parameters. In particular,
60if an element C<$_[0]> is updated, the corresponding argument is
61updated (or an error occurs if it is not updatable). If an argument
62is an array or hash element which did not exist when the function
63was called, that element is created only when (and if) it is modified
64or a reference to it is taken. (Some earlier versions of Perl
65created the element whether or not the element was assigned to.)
66Assigning to the whole array C<@_> removes that aliasing, and does
67not update any arguments.
68
69The return value of a subroutine is the value of the last expression
70evaluated. More explicitly, a C<return> statement may be used to exit the
54310121
PP
71subroutine, optionally specifying the returned value, which will be
72evaluated in the appropriate context (list, scalar, or void) depending
73on the context of the subroutine call. If you specify no return value,
19799a22
GS
74the subroutine returns an empty list in list context, the undefined
75value in scalar context, or nothing in void context. If you return
76one or more aggregates (arrays and hashes), these will be flattened
77together into one large indistinguishable list.
78
79Perl does not have named formal parameters. In practice all you
80do is assign to a C<my()> list of these. Variables that aren't
81declared to be private are global variables. For gory details
82on creating private variables, see L<"Private Variables via my()">
83and L<"Temporary Values via local()">. To create protected
84environments for a set of functions in a separate package (and
85probably a separate file), see L<perlmod/"Packages">.
a0d0e21e
LW
86
87Example:
88
cb1a09d0
AD
89 sub max {
90 my $max = shift(@_);
a0d0e21e
LW
91 foreach $foo (@_) {
92 $max = $foo if $max < $foo;
93 }
cb1a09d0 94 return $max;
a0d0e21e 95 }
cb1a09d0 96 $bestday = max($mon,$tue,$wed,$thu,$fri);
a0d0e21e
LW
97
98Example:
99
100 # get a line, combining continuation lines
101 # that start with whitespace
102
103 sub get_line {
19799a22 104 $thisline = $lookahead; # global variables!
54310121 105 LINE: while (defined($lookahead = <STDIN>)) {
a0d0e21e
LW
106 if ($lookahead =~ /^[ \t]/) {
107 $thisline .= $lookahead;
108 }
109 else {
110 last LINE;
111 }
112 }
19799a22 113 return $thisline;
a0d0e21e
LW
114 }
115
116 $lookahead = <STDIN>; # get first line
19799a22 117 while (defined($line = get_line())) {
a0d0e21e
LW
118 ...
119 }
120
09bef843 121Assigning to a list of private variables to name your arguments:
a0d0e21e
LW
122
123 sub maybeset {
124 my($key, $value) = @_;
cb1a09d0 125 $Foo{$key} = $value unless $Foo{$key};
a0d0e21e
LW
126 }
127
19799a22
GS
128Because the assignment copies the values, this also has the effect
129of turning call-by-reference into call-by-value. Otherwise a
130function is free to do in-place modifications of C<@_> and change
131its caller's values.
cb1a09d0
AD
132
133 upcase_in($v1, $v2); # this changes $v1 and $v2
134 sub upcase_in {
54310121
PP
135 for (@_) { tr/a-z/A-Z/ }
136 }
cb1a09d0
AD
137
138You aren't allowed to modify constants in this way, of course. If an
139argument were actually literal and you tried to change it, you'd take a
140(presumably fatal) exception. For example, this won't work:
141
142 upcase_in("frederick");
143
f86cebdf 144It would be much safer if the C<upcase_in()> function
cb1a09d0
AD
145were written to return a copy of its parameters instead
146of changing them in place:
147
19799a22 148 ($v3, $v4) = upcase($v1, $v2); # this doesn't change $v1 and $v2
cb1a09d0 149 sub upcase {
54310121 150 return unless defined wantarray; # void context, do nothing
cb1a09d0 151 my @parms = @_;
54310121 152 for (@parms) { tr/a-z/A-Z/ }
c07a80fd 153 return wantarray ? @parms : $parms[0];
54310121 154 }
cb1a09d0 155
19799a22 156Notice how this (unprototyped) function doesn't care whether it was
a2293a43 157passed real scalars or arrays. Perl sees all arguments as one big,
19799a22
GS
158long, flat parameter list in C<@_>. This is one area where
159Perl's simple argument-passing style shines. The C<upcase()>
160function would work perfectly well without changing the C<upcase()>
161definition even if we fed it things like this:
cb1a09d0
AD
162
163 @newlist = upcase(@list1, @list2);
164 @newlist = upcase( split /:/, $var );
165
166Do not, however, be tempted to do this:
167
168 (@a, @b) = upcase(@list1, @list2);
169
19799a22
GS
170Like the flattened incoming parameter list, the return list is also
171flattened on return. So all you have managed to do here is stored
17b63f68 172everything in C<@a> and made C<@b> empty. See
13a2d996 173L<Pass by Reference> for alternatives.
19799a22
GS
174
175A subroutine may be called using an explicit C<&> prefix. The
176C<&> is optional in modern Perl, as are parentheses if the
177subroutine has been predeclared. The C<&> is I<not> optional
178when just naming the subroutine, such as when it's used as
179an argument to defined() or undef(). Nor is it optional when you
180want to do an indirect subroutine call with a subroutine name or
181reference using the C<&$subref()> or C<&{$subref}()> constructs,
c47ff5f1 182although the C<< $subref->() >> notation solves that problem.
19799a22
GS
183See L<perlref> for more about all that.
184
185Subroutines may be called recursively. If a subroutine is called
186using the C<&> form, the argument list is optional, and if omitted,
187no C<@_> array is set up for the subroutine: the C<@_> array at the
188time of the call is visible to subroutine instead. This is an
189efficiency mechanism that new users may wish to avoid.
a0d0e21e
LW
190
191 &foo(1,2,3); # pass three arguments
192 foo(1,2,3); # the same
193
194 foo(); # pass a null list
195 &foo(); # the same
a0d0e21e 196
cb1a09d0 197 &foo; # foo() get current args, like foo(@_) !!
54310121 198 foo; # like foo() IFF sub foo predeclared, else "foo"
cb1a09d0 199
19799a22
GS
200Not only does the C<&> form make the argument list optional, it also
201disables any prototype checking on arguments you do provide. This
c07a80fd 202is partly for historical reasons, and partly for having a convenient way
19799a22 203to cheat if you know what you're doing. See L<Prototypes> below.
c07a80fd 204
09bef843 205Functions whose names are in all upper case are reserved to the Perl
19799a22
GS
206core, as are modules whose names are in all lower case. A
207function in all capitals is a loosely-held convention meaning it
208will be called indirectly by the run-time system itself, usually
209due to a triggered event. Functions that do special, pre-defined
f2fc0a40
AB
210things include C<BEGIN>, C<CHECK>, C<INIT>, C<END>, C<AUTOLOAD>,
211C<CLONE> and C<DESTROY>--plus all functions mentioned in L<perltie>.
5a964f20 212
b687b08b 213=head2 Private Variables via my()
cb1a09d0
AD
214
215Synopsis:
216
217 my $foo; # declare $foo lexically local
218 my (@wid, %get); # declare list of variables local
219 my $foo = "flurp"; # declare $foo lexical, and init it
220 my @oof = @bar; # declare @oof lexical, and init it
09bef843
SB
221 my $x : Foo = $y; # similar, with an attribute applied
222
a0ae32d3
JH
223B<WARNING>: The use of attribute lists on C<my> declarations is still
224evolving. The current semantics and interface are subject to change.
225See L<attributes> and L<Attribute::Handlers>.
cb1a09d0 226
19799a22
GS
227The C<my> operator declares the listed variables to be lexically
228confined to the enclosing block, conditional (C<if/unless/elsif/else>),
229loop (C<for/foreach/while/until/continue>), subroutine, C<eval>,
230or C<do/require/use>'d file. If more than one value is listed, the
231list must be placed in parentheses. All listed elements must be
232legal lvalues. Only alphanumeric identifiers may be lexically
09bef843 233scoped--magical built-ins like C<$/> must currently be C<local>ize
19799a22
GS
234with C<local> instead.
235
236Unlike dynamic variables created by the C<local> operator, lexical
237variables declared with C<my> are totally hidden from the outside
238world, including any called subroutines. This is true if it's the
239same subroutine called from itself or elsewhere--every call gets
240its own copy.
241
242This doesn't mean that a C<my> variable declared in a statically
243enclosing lexical scope would be invisible. Only dynamic scopes
244are cut off. For example, the C<bumpx()> function below has access
245to the lexical $x variable because both the C<my> and the C<sub>
246occurred at the same scope, presumably file scope.
5a964f20
TC
247
248 my $x = 10;
249 sub bumpx { $x++ }
250
19799a22
GS
251An C<eval()>, however, can see lexical variables of the scope it is
252being evaluated in, so long as the names aren't hidden by declarations within
253the C<eval()> itself. See L<perlref>.
cb1a09d0 254
19799a22 255The parameter list to my() may be assigned to if desired, which allows you
cb1a09d0
AD
256to initialize your variables. (If no initializer is given for a
257particular variable, it is created with the undefined value.) Commonly
19799a22 258this is used to name input parameters to a subroutine. Examples:
cb1a09d0
AD
259
260 $arg = "fred"; # "global" variable
261 $n = cube_root(27);
262 print "$arg thinks the root is $n\n";
263 fred thinks the root is 3
264
265 sub cube_root {
266 my $arg = shift; # name doesn't matter
267 $arg **= 1/3;
268 return $arg;
54310121 269 }
cb1a09d0 270
19799a22
GS
271The C<my> is simply a modifier on something you might assign to. So when
272you do assign to variables in its argument list, C<my> doesn't
6cc33c6d 273change whether those variables are viewed as a scalar or an array. So
cb1a09d0 274
5a964f20 275 my ($foo) = <STDIN>; # WRONG?
cb1a09d0
AD
276 my @FOO = <STDIN>;
277
5f05dabc 278both supply a list context to the right-hand side, while
cb1a09d0
AD
279
280 my $foo = <STDIN>;
281
5f05dabc 282supplies a scalar context. But the following declares only one variable:
748a9306 283
5a964f20 284 my $foo, $bar = 1; # WRONG
748a9306 285
cb1a09d0 286That has the same effect as
748a9306 287
cb1a09d0
AD
288 my $foo;
289 $bar = 1;
a0d0e21e 290
cb1a09d0
AD
291The declared variable is not introduced (is not visible) until after
292the current statement. Thus,
293
294 my $x = $x;
295
19799a22 296can be used to initialize a new $x with the value of the old $x, and
cb1a09d0
AD
297the expression
298
299 my $x = 123 and $x == 123
300
19799a22 301is false unless the old $x happened to have the value C<123>.
cb1a09d0 302
55497cff
PP
303Lexical scopes of control structures are not bounded precisely by the
304braces that delimit their controlled blocks; control expressions are
19799a22 305part of that scope, too. Thus in the loop
55497cff 306
19799a22 307 while (my $line = <>) {
55497cff
PP
308 $line = lc $line;
309 } continue {
310 print $line;
311 }
312
19799a22 313the scope of $line extends from its declaration throughout the rest of
55497cff
PP
314the loop construct (including the C<continue> clause), but not beyond
315it. Similarly, in the conditional
316
317 if ((my $answer = <STDIN>) =~ /^yes$/i) {
318 user_agrees();
319 } elsif ($answer =~ /^no$/i) {
320 user_disagrees();
321 } else {
322 chomp $answer;
323 die "'$answer' is neither 'yes' nor 'no'";
324 }
325
19799a22
GS
326the scope of $answer extends from its declaration through the rest
327of that conditional, including any C<elsif> and C<else> clauses,
457b36cb
MV
328but not beyond it. See L<perlsyn/"Simple statements"> for information
329on the scope of variables in statements with modifiers.
55497cff 330
5f05dabc 331The C<foreach> loop defaults to scoping its index variable dynamically
19799a22
GS
332in the manner of C<local>. However, if the index variable is
333prefixed with the keyword C<my>, or if there is already a lexical
334by that name in scope, then a new lexical is created instead. Thus
335in the loop
55497cff
PP
336
337 for my $i (1, 2, 3) {
338 some_function();
339 }
340
19799a22
GS
341the scope of $i extends to the end of the loop, but not beyond it,
342rendering the value of $i inaccessible within C<some_function()>.
55497cff 343
cb1a09d0 344Some users may wish to encourage the use of lexically scoped variables.
19799a22
GS
345As an aid to catching implicit uses to package variables,
346which are always global, if you say
cb1a09d0
AD
347
348 use strict 'vars';
349
19799a22
GS
350then any variable mentioned from there to the end of the enclosing
351block must either refer to a lexical variable, be predeclared via
77ca0c92 352C<our> or C<use vars>, or else must be fully qualified with the package name.
19799a22
GS
353A compilation error results otherwise. An inner block may countermand
354this with C<no strict 'vars'>.
355
356A C<my> has both a compile-time and a run-time effect. At compile
8593bda5 357time, the compiler takes notice of it. The principal usefulness
19799a22
GS
358of this is to quiet C<use strict 'vars'>, but it is also essential
359for generation of closures as detailed in L<perlref>. Actual
360initialization is delayed until run time, though, so it gets executed
361at the appropriate time, such as each time through a loop, for
362example.
363
364Variables declared with C<my> are not part of any package and are therefore
cb1a09d0
AD
365never fully qualified with the package name. In particular, you're not
366allowed to try to make a package variable (or other global) lexical:
367
368 my $pack::var; # ERROR! Illegal syntax
369 my $_; # also illegal (currently)
370
371In fact, a dynamic variable (also known as package or global variables)
f86cebdf 372are still accessible using the fully qualified C<::> notation even while a
cb1a09d0
AD
373lexical of the same name is also visible:
374
375 package main;
376 local $x = 10;
377 my $x = 20;
378 print "$x and $::x\n";
379
f86cebdf 380That will print out C<20> and C<10>.
cb1a09d0 381
19799a22
GS
382You may declare C<my> variables at the outermost scope of a file
383to hide any such identifiers from the world outside that file. This
384is similar in spirit to C's static variables when they are used at
385the file level. To do this with a subroutine requires the use of
386a closure (an anonymous function that accesses enclosing lexicals).
387If you want to create a private subroutine that cannot be called
388from outside that block, it can declare a lexical variable containing
389an anonymous sub reference:
cb1a09d0
AD
390
391 my $secret_version = '1.001-beta';
392 my $secret_sub = sub { print $secret_version };
393 &$secret_sub();
394
395As long as the reference is never returned by any function within the
5f05dabc 396module, no outside module can see the subroutine, because its name is not in
cb1a09d0 397any package's symbol table. Remember that it's not I<REALLY> called
19799a22 398C<$some_pack::secret_version> or anything; it's just $secret_version,
cb1a09d0
AD
399unqualified and unqualifiable.
400
19799a22
GS
401This does not work with object methods, however; all object methods
402have to be in the symbol table of some package to be found. See
403L<perlref/"Function Templates"> for something of a work-around to
404this.
cb1a09d0 405
c2611fb3 406=head2 Persistent Private Variables
5a964f20
TC
407
408Just because a lexical variable is lexically (also called statically)
f86cebdf 409scoped to its enclosing block, C<eval>, or C<do> FILE, this doesn't mean that
5a964f20
TC
410within a function it works like a C static. It normally works more
411like a C auto, but with implicit garbage collection.
412
413Unlike local variables in C or C++, Perl's lexical variables don't
414necessarily get recycled just because their scope has exited.
415If something more permanent is still aware of the lexical, it will
416stick around. So long as something else references a lexical, that
417lexical won't be freed--which is as it should be. You wouldn't want
418memory being free until you were done using it, or kept around once you
419were done. Automatic garbage collection takes care of this for you.
420
421This means that you can pass back or save away references to lexical
422variables, whereas to return a pointer to a C auto is a grave error.
423It also gives us a way to simulate C's function statics. Here's a
424mechanism for giving a function private variables with both lexical
425scoping and a static lifetime. If you do want to create something like
426C's static variables, just enclose the whole function in an extra block,
427and put the static variable outside the function but in the block.
cb1a09d0
AD
428
429 {
54310121 430 my $secret_val = 0;
cb1a09d0
AD
431 sub gimme_another {
432 return ++$secret_val;
54310121
PP
433 }
434 }
cb1a09d0
AD
435 # $secret_val now becomes unreachable by the outside
436 # world, but retains its value between calls to gimme_another
437
54310121 438If this function is being sourced in from a separate file
cb1a09d0 439via C<require> or C<use>, then this is probably just fine. If it's
19799a22 440all in the main program, you'll need to arrange for the C<my>
cb1a09d0 441to be executed early, either by putting the whole block above
f86cebdf 442your main program, or more likely, placing merely a C<BEGIN>
cb1a09d0
AD
443sub around it to make sure it gets executed before your program
444starts to run:
445
446 sub BEGIN {
54310121 447 my $secret_val = 0;
cb1a09d0
AD
448 sub gimme_another {
449 return ++$secret_val;
54310121
PP
450 }
451 }
cb1a09d0 452
19799a22 453See L<perlmod/"Package Constructors and Destructors"> about the
7d30b5c4 454special triggered functions, C<BEGIN>, C<CHECK>, C<INIT> and C<END>.
cb1a09d0 455
19799a22
GS
456If declared at the outermost scope (the file scope), then lexicals
457work somewhat like C's file statics. They are available to all
458functions in that same file declared below them, but are inaccessible
459from outside that file. This strategy is sometimes used in modules
460to create private variables that the whole module can see.
5a964f20 461
cb1a09d0
AD
462=head2 Temporary Values via local()
463
19799a22 464B<WARNING>: In general, you should be using C<my> instead of C<local>, because
6d28dffb 465it's faster and safer. Exceptions to this include the global punctuation
cb1a09d0 466variables, filehandles and formats, and direct manipulation of the Perl
19799a22 467symbol table itself. Format variables often use C<local> though, as do
cb1a09d0
AD
468other variables whose current value must be visible to called
469subroutines.
470
471Synopsis:
472
473 local $foo; # declare $foo dynamically local
474 local (@wid, %get); # declare list of variables local
475 local $foo = "flurp"; # declare $foo dynamic, and init it
476 local @oof = @bar; # declare @oof dynamic, and init it
477
478 local *FH; # localize $FH, @FH, %FH, &FH ...
479 local *merlyn = *randal; # now $merlyn is really $randal, plus
480 # @merlyn is really @randal, etc
481 local *merlyn = 'randal'; # SAME THING: promote 'randal' to *randal
54310121 482 local *merlyn = \$randal; # just alias $merlyn, not @merlyn etc
cb1a09d0 483
19799a22
GS
484A C<local> modifies its listed variables to be "local" to the
485enclosing block, C<eval>, or C<do FILE>--and to I<any subroutine
486called from within that block>. A C<local> just gives temporary
487values to global (meaning package) variables. It does I<not> create
488a local variable. This is known as dynamic scoping. Lexical scoping
489is done with C<my>, which works more like C's auto declarations.
cb1a09d0 490
19799a22 491If more than one variable is given to C<local>, they must be placed in
5f05dabc 492parentheses. All listed elements must be legal lvalues. This operator works
cb1a09d0 493by saving the current values of those variables in its argument list on a
5f05dabc 494hidden stack and restoring them upon exiting the block, subroutine, or
cb1a09d0
AD
495eval. This means that called subroutines can also reference the local
496variable, but not the global one. The argument list may be assigned to if
497desired, which allows you to initialize your local variables. (If no
498initializer is given for a particular variable, it is created with an
499undefined value.) Commonly this is used to name the parameters to a
500subroutine. Examples:
501
502 for $i ( 0 .. 9 ) {
503 $digits{$i} = $i;
54310121 504 }
cb1a09d0 505 # assume this function uses global %digits hash
54310121 506 parse_num();
cb1a09d0
AD
507
508 # now temporarily add to %digits hash
509 if ($base12) {
510 # (NOTE: not claiming this is efficient!)
511 local %digits = (%digits, 't' => 10, 'e' => 11);
512 parse_num(); # parse_num gets this new %digits!
513 }
514 # old %digits restored here
515
19799a22 516Because C<local> is a run-time operator, it gets executed each time
cb1a09d0
AD
517through a loop. In releases of Perl previous to 5.0, this used more stack
518storage each time until the loop was exited. Perl now reclaims the space
519each time through, but it's still more efficient to declare your variables
520outside the loop.
521
f86cebdf
GS
522A C<local> is simply a modifier on an lvalue expression. When you assign to
523a C<local>ized variable, the C<local> doesn't change whether its list is viewed
cb1a09d0
AD
524as a scalar or an array. So
525
526 local($foo) = <STDIN>;
527 local @FOO = <STDIN>;
528
5f05dabc 529both supply a list context to the right-hand side, while
cb1a09d0
AD
530
531 local $foo = <STDIN>;
532
533supplies a scalar context.
534
3e3baf6d
TB
535A note about C<local()> and composite types is in order. Something
536like C<local(%foo)> works by temporarily placing a brand new hash in
537the symbol table. The old hash is left alone, but is hidden "behind"
538the new one.
539
540This means the old variable is completely invisible via the symbol
541table (i.e. the hash entry in the C<*foo> typeglob) for the duration
542of the dynamic scope within which the C<local()> was seen. This
543has the effect of allowing one to temporarily occlude any magic on
544composite types. For instance, this will briefly alter a tied
545hash to some other implementation:
546
547 tie %ahash, 'APackage';
548 [...]
549 {
550 local %ahash;
551 tie %ahash, 'BPackage';
552 [..called code will see %ahash tied to 'BPackage'..]
553 {
554 local %ahash;
555 [..%ahash is a normal (untied) hash here..]
556 }
557 }
558 [..%ahash back to its initial tied self again..]
559
fd5a896a
DM
560B<WARNING> The code example above does not currently work as described.
561This will be fixed in a future release of Perl; in the meantime, avoid
562code that relies on any particular behaviour of localising tied arrays
563or hashes (localising individual elements is still okay).
564See L<perldelta/"Localising Tied Arrays and Hashes Is Broken"> for more
565details.
566
3e3baf6d
TB
567As another example, a custom implementation of C<%ENV> might look
568like this:
569
570 {
571 local %ENV;
572 tie %ENV, 'MyOwnEnv';
573 [..do your own fancy %ENV manipulation here..]
574 }
575 [..normal %ENV behavior here..]
576
6ee623d5 577It's also worth taking a moment to explain what happens when you
f86cebdf
GS
578C<local>ize a member of a composite type (i.e. an array or hash element).
579In this case, the element is C<local>ized I<by name>. This means that
6ee623d5
GS
580when the scope of the C<local()> ends, the saved value will be
581restored to the hash element whose key was named in the C<local()>, or
582the array element whose index was named in the C<local()>. If that
583element was deleted while the C<local()> was in effect (e.g. by a
584C<delete()> from a hash or a C<shift()> of an array), it will spring
585back into existence, possibly extending an array and filling in the
586skipped elements with C<undef>. For instance, if you say
587
588 %hash = ( 'This' => 'is', 'a' => 'test' );
589 @ary = ( 0..5 );
590 {
591 local($ary[5]) = 6;
592 local($hash{'a'}) = 'drill';
593 while (my $e = pop(@ary)) {
594 print "$e . . .\n";
595 last unless $e > 3;
596 }
597 if (@ary) {
598 $hash{'only a'} = 'test';
599 delete $hash{'a'};
600 }
601 }
602 print join(' ', map { "$_ $hash{$_}" } sort keys %hash),".\n";
603 print "The array has ",scalar(@ary)," elements: ",
604 join(', ', map { defined $_ ? $_ : 'undef' } @ary),"\n";
605
606Perl will print
607
608 6 . . .
609 4 . . .
610 3 . . .
611 This is a test only a test.
612 The array has 6 elements: 0, 1, 2, undef, undef, 5
613
19799a22 614The behavior of local() on non-existent members of composite
7185e5cc
GS
615types is subject to change in future.
616
cd06dffe
GS
617=head2 Lvalue subroutines
618
e6a32221
JC
619B<WARNING>: Lvalue subroutines are still experimental and the
620implementation may change in future versions of Perl.
cd06dffe
GS
621
622It is possible to return a modifiable value from a subroutine.
623To do this, you have to declare the subroutine to return an lvalue.
624
625 my $val;
626 sub canmod : lvalue {
e6a32221 627 # return $val; this doesn't work, don't say "return"
cd06dffe
GS
628 $val;
629 }
630 sub nomod {
631 $val;
632 }
633
634 canmod() = 5; # assigns to $val
635 nomod() = 5; # ERROR
636
637The scalar/list context for the subroutine and for the right-hand
638side of assignment is determined as if the subroutine call is replaced
639by a scalar. For example, consider:
640
641 data(2,3) = get_data(3,4);
642
643Both subroutines here are called in a scalar context, while in:
644
645 (data(2,3)) = get_data(3,4);
646
647and in:
648
649 (data(2),data(3)) = get_data(3,4);
650
651all the subroutines are called in a list context.
652
e6a32221
JC
653=over 4
654
655=item Lvalue subroutines are EXPERIMENTAL
656
657They appear to be convenient, but there are several reasons to be
658circumspect.
659
660You can't use the return keyword, you must pass out the value before
661falling out of subroutine scope. (see comment in example above). This
662is usually not a problem, but it disallows an explicit return out of a
663deeply nested loop, which is sometimes a nice way out.
664
665They violate encapsulation. A normal mutator can check the supplied
666argument before setting the attribute it is protecting, an lvalue
667subroutine never gets that chance. Consider;
668
669 my $some_array_ref = []; # protected by mutators ??
670
671 sub set_arr { # normal mutator
672 my $val = shift;
673 die("expected array, you supplied ", ref $val)
674 unless ref $val eq 'ARRAY';
675 $some_array_ref = $val;
676 }
677 sub set_arr_lv : lvalue { # lvalue mutator
678 $some_array_ref;
679 }
680
681 # set_arr_lv cannot stop this !
682 set_arr_lv() = { a => 1 };
818c4caa 683
e6a32221
JC
684=back
685
cb1a09d0
AD
686=head2 Passing Symbol Table Entries (typeglobs)
687
19799a22
GS
688B<WARNING>: The mechanism described in this section was originally
689the only way to simulate pass-by-reference in older versions of
690Perl. While it still works fine in modern versions, the new reference
691mechanism is generally easier to work with. See below.
a0d0e21e
LW
692
693Sometimes you don't want to pass the value of an array to a subroutine
694but rather the name of it, so that the subroutine can modify the global
695copy of it rather than working with a local copy. In perl you can
cb1a09d0 696refer to all objects of a particular name by prefixing the name
5f05dabc 697with a star: C<*foo>. This is often known as a "typeglob", because the
a0d0e21e
LW
698star on the front can be thought of as a wildcard match for all the
699funny prefix characters on variables and subroutines and such.
700
55497cff 701When evaluated, the typeglob produces a scalar value that represents
5f05dabc 702all the objects of that name, including any filehandle, format, or
a0d0e21e 703subroutine. When assigned to, it causes the name mentioned to refer to
19799a22 704whatever C<*> value was assigned to it. Example:
a0d0e21e
LW
705
706 sub doubleary {
707 local(*someary) = @_;
708 foreach $elem (@someary) {
709 $elem *= 2;
710 }
711 }
712 doubleary(*foo);
713 doubleary(*bar);
714
19799a22 715Scalars are already passed by reference, so you can modify
a0d0e21e 716scalar arguments without using this mechanism by referring explicitly
1fef88e7 717to C<$_[0]> etc. You can modify all the elements of an array by passing
f86cebdf
GS
718all the elements as scalars, but you have to use the C<*> mechanism (or
719the equivalent reference mechanism) to C<push>, C<pop>, or change the size of
a0d0e21e
LW
720an array. It will certainly be faster to pass the typeglob (or reference).
721
722Even if you don't want to modify an array, this mechanism is useful for
5f05dabc 723passing multiple arrays in a single LIST, because normally the LIST
a0d0e21e 724mechanism will merge all the array values so that you can't extract out
55497cff 725the individual arrays. For more on typeglobs, see
2ae324a7 726L<perldata/"Typeglobs and Filehandles">.
cb1a09d0 727
5a964f20
TC
728=head2 When to Still Use local()
729
19799a22
GS
730Despite the existence of C<my>, there are still three places where the
731C<local> operator still shines. In fact, in these three places, you
5a964f20
TC
732I<must> use C<local> instead of C<my>.
733
13a2d996 734=over 4
5a964f20 735
551e1d92
RB
736=item 1.
737
738You need to give a global variable a temporary value, especially $_.
5a964f20 739
f86cebdf
GS
740The global variables, like C<@ARGV> or the punctuation variables, must be
741C<local>ized with C<local()>. This block reads in F</etc/motd>, and splits
5a964f20 742it up into chunks separated by lines of equal signs, which are placed
f86cebdf 743in C<@Fields>.
5a964f20
TC
744
745 {
746 local @ARGV = ("/etc/motd");
747 local $/ = undef;
748 local $_ = <>;
749 @Fields = split /^\s*=+\s*$/;
750 }
751
19799a22 752It particular, it's important to C<local>ize $_ in any routine that assigns
5a964f20
TC
753to it. Look out for implicit assignments in C<while> conditionals.
754
551e1d92
RB
755=item 2.
756
757You need to create a local file or directory handle or a local function.
5a964f20 758
09bef843
SB
759A function that needs a filehandle of its own must use
760C<local()> on a complete typeglob. This can be used to create new symbol
5a964f20
TC
761table entries:
762
763 sub ioqueue {
764 local (*READER, *WRITER); # not my!
17b63f68 765 pipe (READER, WRITER) or die "pipe: $!";
5a964f20
TC
766 return (*READER, *WRITER);
767 }
768 ($head, $tail) = ioqueue();
769
770See the Symbol module for a way to create anonymous symbol table
771entries.
772
773Because assignment of a reference to a typeglob creates an alias, this
774can be used to create what is effectively a local function, or at least,
775a local alias.
776
777 {
f86cebdf
GS
778 local *grow = \&shrink; # only until this block exists
779 grow(); # really calls shrink()
780 move(); # if move() grow()s, it shrink()s too
5a964f20 781 }
f86cebdf 782 grow(); # get the real grow() again
5a964f20
TC
783
784See L<perlref/"Function Templates"> for more about manipulating
785functions by name in this way.
786
551e1d92
RB
787=item 3.
788
789You want to temporarily change just one element of an array or hash.
5a964f20 790
f86cebdf 791You can C<local>ize just one element of an aggregate. Usually this
5a964f20
TC
792is done on dynamics:
793
794 {
795 local $SIG{INT} = 'IGNORE';
796 funct(); # uninterruptible
797 }
798 # interruptibility automatically restored here
799
800But it also works on lexically declared aggregates. Prior to 5.005,
801this operation could on occasion misbehave.
802
803=back
804
cb1a09d0
AD
805=head2 Pass by Reference
806
55497cff
PP
807If you want to pass more than one array or hash into a function--or
808return them from it--and have them maintain their integrity, then
809you're going to have to use an explicit pass-by-reference. Before you
810do that, you need to understand references as detailed in L<perlref>.
c07a80fd 811This section may not make much sense to you otherwise.
cb1a09d0 812
19799a22
GS
813Here are a few simple examples. First, let's pass in several arrays
814to a function and have it C<pop> all of then, returning a new list
815of all their former last elements:
cb1a09d0
AD
816
817 @tailings = popmany ( \@a, \@b, \@c, \@d );
818
819 sub popmany {
820 my $aref;
821 my @retlist = ();
822 foreach $aref ( @_ ) {
823 push @retlist, pop @$aref;
54310121 824 }
cb1a09d0 825 return @retlist;
54310121 826 }
cb1a09d0 827
54310121 828Here's how you might write a function that returns a
cb1a09d0
AD
829list of keys occurring in all the hashes passed to it:
830
54310121 831 @common = inter( \%foo, \%bar, \%joe );
cb1a09d0
AD
832 sub inter {
833 my ($k, $href, %seen); # locals
834 foreach $href (@_) {
835 while ( $k = each %$href ) {
836 $seen{$k}++;
54310121
PP
837 }
838 }
cb1a09d0 839 return grep { $seen{$_} == @_ } keys %seen;
54310121 840 }
cb1a09d0 841
5f05dabc 842So far, we're using just the normal list return mechanism.
54310121
PP
843What happens if you want to pass or return a hash? Well,
844if you're using only one of them, or you don't mind them
cb1a09d0 845concatenating, then the normal calling convention is ok, although
54310121 846a little expensive.
cb1a09d0
AD
847
848Where people get into trouble is here:
849
850 (@a, @b) = func(@c, @d);
851or
852 (%a, %b) = func(%c, %d);
853
19799a22
GS
854That syntax simply won't work. It sets just C<@a> or C<%a> and
855clears the C<@b> or C<%b>. Plus the function didn't get passed
856into two separate arrays or hashes: it got one long list in C<@_>,
857as always.
cb1a09d0
AD
858
859If you can arrange for everyone to deal with this through references, it's
860cleaner code, although not so nice to look at. Here's a function that
861takes two array references as arguments, returning the two array elements
862in order of how many elements they have in them:
863
864 ($aref, $bref) = func(\@c, \@d);
865 print "@$aref has more than @$bref\n";
866 sub func {
867 my ($cref, $dref) = @_;
868 if (@$cref > @$dref) {
869 return ($cref, $dref);
870 } else {
c07a80fd 871 return ($dref, $cref);
54310121
PP
872 }
873 }
cb1a09d0
AD
874
875It turns out that you can actually do this also:
876
877 (*a, *b) = func(\@c, \@d);
878 print "@a has more than @b\n";
879 sub func {
880 local (*c, *d) = @_;
881 if (@c > @d) {
882 return (\@c, \@d);
883 } else {
884 return (\@d, \@c);
54310121
PP
885 }
886 }
cb1a09d0
AD
887
888Here we're using the typeglobs to do symbol table aliasing. It's
19799a22 889a tad subtle, though, and also won't work if you're using C<my>
09bef843 890variables, because only globals (even in disguise as C<local>s)
19799a22 891are in the symbol table.
5f05dabc
PP
892
893If you're passing around filehandles, you could usually just use the bare
19799a22
GS
894typeglob, like C<*STDOUT>, but typeglobs references work, too.
895For example:
5f05dabc
PP
896
897 splutter(\*STDOUT);
898 sub splutter {
899 my $fh = shift;
900 print $fh "her um well a hmmm\n";
901 }
902
903 $rec = get_rec(\*STDIN);
904 sub get_rec {
905 my $fh = shift;
906 return scalar <$fh>;
907 }
908
19799a22
GS
909If you're planning on generating new filehandles, you could do this.
910Notice to pass back just the bare *FH, not its reference.
5f05dabc
PP
911
912 sub openit {
19799a22 913 my $path = shift;
5f05dabc 914 local *FH;
e05a3a1e 915 return open (FH, $path) ? *FH : undef;
54310121 916 }
5f05dabc 917
cb1a09d0
AD
918=head2 Prototypes
919
19799a22
GS
920Perl supports a very limited kind of compile-time argument checking
921using function prototyping. If you declare
cb1a09d0
AD
922
923 sub mypush (\@@)
924
19799a22
GS
925then C<mypush()> takes arguments exactly like C<push()> does. The
926function declaration must be visible at compile time. The prototype
927affects only interpretation of new-style calls to the function,
928where new-style is defined as not using the C<&> character. In
929other words, if you call it like a built-in function, then it behaves
930like a built-in function. If you call it like an old-fashioned
931subroutine, then it behaves like an old-fashioned subroutine. It
932naturally falls out from this rule that prototypes have no influence
933on subroutine references like C<\&foo> or on indirect subroutine
c47ff5f1 934calls like C<&{$subref}> or C<< $subref->() >>.
c07a80fd
PP
935
936Method calls are not influenced by prototypes either, because the
19799a22
GS
937function to be called is indeterminate at compile time, since
938the exact code called depends on inheritance.
cb1a09d0 939
19799a22
GS
940Because the intent of this feature is primarily to let you define
941subroutines that work like built-in functions, here are prototypes
942for some other functions that parse almost exactly like the
943corresponding built-in.
cb1a09d0
AD
944
945 Declared as Called as
946
f86cebdf
GS
947 sub mylink ($$) mylink $old, $new
948 sub myvec ($$$) myvec $var, $offset, 1
949 sub myindex ($$;$) myindex &getstring, "substr"
950 sub mysyswrite ($$$;$) mysyswrite $buf, 0, length($buf) - $off, $off
951 sub myreverse (@) myreverse $a, $b, $c
952 sub myjoin ($@) myjoin ":", $a, $b, $c
953 sub mypop (\@) mypop @array
954 sub mysplice (\@$$@) mysplice @array, @array, 0, @pushme
955 sub mykeys (\%) mykeys %{$hashref}
956 sub myopen (*;$) myopen HANDLE, $name
957 sub mypipe (**) mypipe READHANDLE, WRITEHANDLE
958 sub mygrep (&@) mygrep { /foo/ } $a, $b, $c
959 sub myrand ($) myrand 42
960 sub mytime () mytime
cb1a09d0 961
c07a80fd 962Any backslashed prototype character represents an actual argument
6e47f808 963that absolutely must start with that character. The value passed
19799a22
GS
964as part of C<@_> will be a reference to the actual argument given
965in the subroutine call, obtained by applying C<\> to that argument.
c07a80fd 966
5b794e05
JH
967You can also backslash several argument types simultaneously by using
968the C<\[]> notation:
969
970 sub myref (\[$@%&*])
971
972will allow calling myref() as
973
974 myref $var
975 myref @array
976 myref %hash
977 myref &sub
978 myref *glob
979
980and the first argument of myref() will be a reference to
981a scalar, an array, a hash, a code, or a glob.
982
c07a80fd 983Unbackslashed prototype characters have special meanings. Any
19799a22 984unbackslashed C<@> or C<%> eats all remaining arguments, and forces
f86cebdf
GS
985list context. An argument represented by C<$> forces scalar context. An
986C<&> requires an anonymous subroutine, which, if passed as the first
0df79f0c
GS
987argument, does not require the C<sub> keyword or a subsequent comma.
988
989A C<*> allows the subroutine to accept a bareword, constant, scalar expression,
648ca4f7
GS
990typeglob, or a reference to a typeglob in that slot. The value will be
991available to the subroutine either as a simple scalar, or (in the latter
0df79f0c
GS
992two cases) as a reference to the typeglob. If you wish to always convert
993such arguments to a typeglob reference, use Symbol::qualify_to_ref() as
994follows:
995
996 use Symbol 'qualify_to_ref';
997
998 sub foo (*) {
999 my $fh = qualify_to_ref(shift, caller);
1000 ...
1001 }
c07a80fd
PP
1002
1003A semicolon separates mandatory arguments from optional arguments.
19799a22 1004It is redundant before C<@> or C<%>, which gobble up everything else.
cb1a09d0 1005
19799a22
GS
1006Note how the last three examples in the table above are treated
1007specially by the parser. C<mygrep()> is parsed as a true list
1008operator, C<myrand()> is parsed as a true unary operator with unary
1009precedence the same as C<rand()>, and C<mytime()> is truly without
1010arguments, just like C<time()>. That is, if you say
cb1a09d0
AD
1011
1012 mytime +2;
1013
f86cebdf 1014you'll get C<mytime() + 2>, not C<mytime(2)>, which is how it would be parsed
19799a22 1015without a prototype.
cb1a09d0 1016
19799a22
GS
1017The interesting thing about C<&> is that you can generate new syntax with it,
1018provided it's in the initial position:
cb1a09d0 1019
6d28dffb 1020 sub try (&@) {
cb1a09d0
AD
1021 my($try,$catch) = @_;
1022 eval { &$try };
1023 if ($@) {
1024 local $_ = $@;
1025 &$catch;
1026 }
1027 }
55497cff 1028 sub catch (&) { $_[0] }
cb1a09d0
AD
1029
1030 try {
1031 die "phooey";
1032 } catch {
1033 /phooey/ and print "unphooey\n";
1034 };
1035
f86cebdf 1036That prints C<"unphooey">. (Yes, there are still unresolved
19799a22 1037issues having to do with visibility of C<@_>. I'm ignoring that
f86cebdf 1038question for the moment. (But note that if we make C<@_> lexically
cb1a09d0 1039scoped, those anonymous subroutines can act like closures... (Gee,
5f05dabc 1040is this sounding a little Lispish? (Never mind.))))
cb1a09d0 1041
19799a22 1042And here's a reimplementation of the Perl C<grep> operator:
cb1a09d0
AD
1043
1044 sub mygrep (&@) {
1045 my $code = shift;
1046 my @result;
1047 foreach $_ (@_) {
6e47f808 1048 push(@result, $_) if &$code;
cb1a09d0
AD
1049 }
1050 @result;
1051 }
a0d0e21e 1052
cb1a09d0
AD
1053Some folks would prefer full alphanumeric prototypes. Alphanumerics have
1054been intentionally left out of prototypes for the express purpose of
1055someday in the future adding named, formal parameters. The current
1056mechanism's main goal is to let module writers provide better diagnostics
1057for module users. Larry feels the notation quite understandable to Perl
1058programmers, and that it will not intrude greatly upon the meat of the
1059module, nor make it harder to read. The line noise is visually
1060encapsulated into a small pill that's easy to swallow.
1061
420cdfc1
ST
1062If you try to use an alphanumeric sequence in a prototype you will
1063generate an optional warning - "Illegal character in prototype...".
1064Unfortunately earlier versions of Perl allowed the prototype to be
1065used as long as its prefix was a valid prototype. The warning may be
1066upgraded to a fatal error in a future version of Perl once the
1067majority of offending code is fixed.
1068
cb1a09d0
AD
1069It's probably best to prototype new functions, not retrofit prototyping
1070into older ones. That's because you must be especially careful about
1071silent impositions of differing list versus scalar contexts. For example,
1072if you decide that a function should take just one parameter, like this:
1073
1074 sub func ($) {
1075 my $n = shift;
1076 print "you gave me $n\n";
54310121 1077 }
cb1a09d0
AD
1078
1079and someone has been calling it with an array or expression
1080returning a list:
1081
1082 func(@foo);
1083 func( split /:/ );
1084
19799a22 1085Then you've just supplied an automatic C<scalar> in front of their
f86cebdf 1086argument, which can be more than a bit surprising. The old C<@foo>
cb1a09d0 1087which used to hold one thing doesn't get passed in. Instead,
19799a22
GS
1088C<func()> now gets passed in a C<1>; that is, the number of elements
1089in C<@foo>. And the C<split> gets called in scalar context so it
1090starts scribbling on your C<@_> parameter list. Ouch!
cb1a09d0 1091
5f05dabc 1092This is all very powerful, of course, and should be used only in moderation
54310121 1093to make the world a better place.
44a8e56a
PP
1094
1095=head2 Constant Functions
1096
1097Functions with a prototype of C<()> are potential candidates for
19799a22
GS
1098inlining. If the result after optimization and constant folding
1099is either a constant or a lexically-scoped scalar which has no other
54310121 1100references, then it will be used in place of function calls made
19799a22
GS
1101without C<&>. Calls made using C<&> are never inlined. (See
1102F<constant.pm> for an easy way to declare most constants.)
44a8e56a 1103
5a964f20 1104The following functions would all be inlined:
44a8e56a 1105
699e6cd4
TP
1106 sub pi () { 3.14159 } # Not exact, but close.
1107 sub PI () { 4 * atan2 1, 1 } # As good as it gets,
1108 # and it's inlined, too!
44a8e56a
PP
1109 sub ST_DEV () { 0 }
1110 sub ST_INO () { 1 }
1111
1112 sub FLAG_FOO () { 1 << 8 }
1113 sub FLAG_BAR () { 1 << 9 }
1114 sub FLAG_MASK () { FLAG_FOO | FLAG_BAR }
54310121
PP
1115
1116 sub OPT_BAZ () { not (0x1B58 & FLAG_MASK) }
44a8e56a
PP
1117 sub BAZ_VAL () {
1118 if (OPT_BAZ) {
1119 return 23;
1120 }
1121 else {
1122 return 42;
1123 }
1124 }
cb1a09d0 1125
54310121
PP
1126 sub N () { int(BAZ_VAL) / 3 }
1127 BEGIN {
1128 my $prod = 1;
1129 for (1..N) { $prod *= $_ }
1130 sub N_FACTORIAL () { $prod }
1131 }
1132
5a964f20 1133If you redefine a subroutine that was eligible for inlining, you'll get
4cee8e80
CS
1134a mandatory warning. (You can use this warning to tell whether or not a
1135particular subroutine is considered constant.) The warning is
1136considered severe enough not to be optional because previously compiled
1137invocations of the function will still be using the old value of the
19799a22 1138function. If you need to be able to redefine the subroutine, you need to
4cee8e80 1139ensure that it isn't inlined, either by dropping the C<()> prototype
19799a22 1140(which changes calling semantics, so beware) or by thwarting the
4cee8e80
CS
1141inlining mechanism in some other way, such as
1142
4cee8e80 1143 sub not_inlined () {
54310121 1144 23 if $];
4cee8e80
CS
1145 }
1146
19799a22 1147=head2 Overriding Built-in Functions
a0d0e21e 1148
19799a22 1149Many built-in functions may be overridden, though this should be tried
5f05dabc 1150only occasionally and for good reason. Typically this might be
19799a22 1151done by a package attempting to emulate missing built-in functionality
a0d0e21e
LW
1152on a non-Unix system.
1153
163e3a99
JP
1154Overriding may be done only by importing the name from a module at
1155compile time--ordinary predeclaration isn't good enough. However, the
19799a22
GS
1156C<use subs> pragma lets you, in effect, predeclare subs
1157via the import syntax, and these names may then override built-in ones:
a0d0e21e
LW
1158
1159 use subs 'chdir', 'chroot', 'chmod', 'chown';
1160 chdir $somewhere;
1161 sub chdir { ... }
1162
19799a22
GS
1163To unambiguously refer to the built-in form, precede the
1164built-in name with the special package qualifier C<CORE::>. For example,
1165saying C<CORE::open()> always refers to the built-in C<open()>, even
fb73857a 1166if the current package has imported some other subroutine called
19799a22 1167C<&open()> from elsewhere. Even though it looks like a regular
09bef843 1168function call, it isn't: you can't take a reference to it, such as
19799a22 1169the incorrect C<\&CORE::open> might appear to produce.
fb73857a 1170
19799a22
GS
1171Library modules should not in general export built-in names like C<open>
1172or C<chdir> as part of their default C<@EXPORT> list, because these may
a0d0e21e 1173sneak into someone else's namespace and change the semantics unexpectedly.
19799a22 1174Instead, if the module adds that name to C<@EXPORT_OK>, then it's
a0d0e21e
LW
1175possible for a user to import the name explicitly, but not implicitly.
1176That is, they could say
1177
1178 use Module 'open';
1179
19799a22 1180and it would import the C<open> override. But if they said
a0d0e21e
LW
1181
1182 use Module;
1183
19799a22 1184they would get the default imports without overrides.
a0d0e21e 1185
19799a22 1186The foregoing mechanism for overriding built-in is restricted, quite
95d94a4f 1187deliberately, to the package that requests the import. There is a second
19799a22 1188method that is sometimes applicable when you wish to override a built-in
95d94a4f
GS
1189everywhere, without regard to namespace boundaries. This is achieved by
1190importing a sub into the special namespace C<CORE::GLOBAL::>. Here is an
1191example that quite brazenly replaces the C<glob> operator with something
1192that understands regular expressions.
1193
1194 package REGlob;
1195 require Exporter;
1196 @ISA = 'Exporter';
1197 @EXPORT_OK = 'glob';
1198
1199 sub import {
1200 my $pkg = shift;
1201 return unless @_;
1202 my $sym = shift;
1203 my $where = ($sym =~ s/^GLOBAL_// ? 'CORE::GLOBAL' : caller(0));
1204 $pkg->export($where, $sym, @_);
1205 }
1206
1207 sub glob {
1208 my $pat = shift;
1209 my @got;
19799a22
GS
1210 local *D;
1211 if (opendir D, '.') {
1212 @got = grep /$pat/, readdir D;
1213 closedir D;
1214 }
1215 return @got;
95d94a4f
GS
1216 }
1217 1;
1218
1219And here's how it could be (ab)used:
1220
1221 #use REGlob 'GLOBAL_glob'; # override glob() in ALL namespaces
1222 package Foo;
1223 use REGlob 'glob'; # override glob() in Foo:: only
1224 print for <^[a-z_]+\.pm\$>; # show all pragmatic modules
1225
19799a22 1226The initial comment shows a contrived, even dangerous example.
95d94a4f 1227By overriding C<glob> globally, you would be forcing the new (and
19799a22 1228subversive) behavior for the C<glob> operator for I<every> namespace,
95d94a4f
GS
1229without the complete cognizance or cooperation of the modules that own
1230those namespaces. Naturally, this should be done with extreme caution--if
1231it must be done at all.
1232
1233The C<REGlob> example above does not implement all the support needed to
19799a22 1234cleanly override perl's C<glob> operator. The built-in C<glob> has
95d94a4f 1235different behaviors depending on whether it appears in a scalar or list
19799a22 1236context, but our C<REGlob> doesn't. Indeed, many perl built-in have such
95d94a4f
GS
1237context sensitive behaviors, and these must be adequately supported by
1238a properly written override. For a fully functional example of overriding
1239C<glob>, study the implementation of C<File::DosGlob> in the standard
1240library.
1241
77bc9082
RGS
1242When you override a built-in, your replacement should be consistent (if
1243possible) with the built-in native syntax. You can achieve this by using
1244a suitable prototype. To get the prototype of an overridable built-in,
1245use the C<prototype> function with an argument of C<"CORE::builtin_name">
1246(see L<perlfunc/prototype>).
1247
1248Note however that some built-ins can't have their syntax expressed by a
1249prototype (such as C<system> or C<chomp>). If you override them you won't
1250be able to fully mimic their original syntax.
1251
fe854a6f 1252The built-ins C<do>, C<require> and C<glob> can also be overridden, but due
77bc9082
RGS
1253to special magic, their original syntax is preserved, and you don't have
1254to define a prototype for their replacements. (You can't override the
1255C<do BLOCK> syntax, though).
1256
1257C<require> has special additional dark magic: if you invoke your
1258C<require> replacement as C<require Foo::Bar>, it will actually receive
1259the argument C<"Foo/Bar.pm"> in @_. See L<perlfunc/require>.
1260
1261And, as you'll have noticed from the previous example, if you override
593b9c14 1262C<glob>, the C<< <*> >> glob operator is overridden as well.
77bc9082 1263
9b3023bc
RGS
1264In a similar fashion, overriding the C<readline> function also overrides
1265the equivalent I/O operator C<< <FILEHANDLE> >>.
1266
fe854a6f 1267Finally, some built-ins (e.g. C<exists> or C<grep>) can't be overridden.
77bc9082 1268
a0d0e21e
LW
1269=head2 Autoloading
1270
19799a22
GS
1271If you call a subroutine that is undefined, you would ordinarily
1272get an immediate, fatal error complaining that the subroutine doesn't
1273exist. (Likewise for subroutines being used as methods, when the
1274method doesn't exist in any base class of the class's package.)
1275However, if an C<AUTOLOAD> subroutine is defined in the package or
1276packages used to locate the original subroutine, then that
1277C<AUTOLOAD> subroutine is called with the arguments that would have
1278been passed to the original subroutine. The fully qualified name
1279of the original subroutine magically appears in the global $AUTOLOAD
1280variable of the same package as the C<AUTOLOAD> routine. The name
1281is not passed as an ordinary argument because, er, well, just
593b9c14
YST
1282because, that's why. (As an exception, a method call to a nonexistent
1283C<import> or C<unimport> method is just skipped instead.)
19799a22
GS
1284
1285Many C<AUTOLOAD> routines load in a definition for the requested
1286subroutine using eval(), then execute that subroutine using a special
1287form of goto() that erases the stack frame of the C<AUTOLOAD> routine
1288without a trace. (See the source to the standard module documented
1289in L<AutoLoader>, for example.) But an C<AUTOLOAD> routine can
1290also just emulate the routine and never define it. For example,
1291let's pretend that a function that wasn't defined should just invoke
1292C<system> with those arguments. All you'd do is:
cb1a09d0
AD
1293
1294 sub AUTOLOAD {
1295 my $program = $AUTOLOAD;
1296 $program =~ s/.*:://;
1297 system($program, @_);
54310121 1298 }
cb1a09d0 1299 date();
6d28dffb 1300 who('am', 'i');
cb1a09d0
AD
1301 ls('-l');
1302
19799a22
GS
1303In fact, if you predeclare functions you want to call that way, you don't
1304even need parentheses:
cb1a09d0
AD
1305
1306 use subs qw(date who ls);
1307 date;
1308 who "am", "i";
593b9c14 1309 ls '-l';
cb1a09d0
AD
1310
1311A more complete example of this is the standard Shell module, which
19799a22 1312can treat undefined subroutine calls as calls to external programs.
a0d0e21e 1313
19799a22
GS
1314Mechanisms are available to help modules writers split their modules
1315into autoloadable files. See the standard AutoLoader module
6d28dffb
PP
1316described in L<AutoLoader> and in L<AutoSplit>, the standard
1317SelfLoader modules in L<SelfLoader>, and the document on adding C
19799a22 1318functions to Perl code in L<perlxs>.
cb1a09d0 1319
09bef843
SB
1320=head2 Subroutine Attributes
1321
1322A subroutine declaration or definition may have a list of attributes
1323associated with it. If such an attribute list is present, it is
0120eecf 1324broken up at space or colon boundaries and treated as though a
09bef843
SB
1325C<use attributes> had been seen. See L<attributes> for details
1326about what attributes are currently supported.
1327Unlike the limitation with the obsolescent C<use attrs>, the
1328C<sub : ATTRLIST> syntax works to associate the attributes with
1329a pre-declaration, and not just with a subroutine definition.
1330
1331The attributes must be valid as simple identifier names (without any
1332punctuation other than the '_' character). They may have a parameter
1333list appended, which is only checked for whether its parentheses ('(',')')
1334nest properly.
1335
1336Examples of valid syntax (even though the attributes are unknown):
1337
0120eecf
GS
1338 sub fnord (&\%) : switch(10,foo(7,3)) : expensive ;
1339 sub plugh () : Ugly('\(") :Bad ;
09bef843
SB
1340 sub xyzzy : _5x5 { ... }
1341
1342Examples of invalid syntax:
1343
1344 sub fnord : switch(10,foo() ; # ()-string not balanced
1345 sub snoid : Ugly('(') ; # ()-string not balanced
1346 sub xyzzy : 5x5 ; # "5x5" not a valid identifier
1347 sub plugh : Y2::north ; # "Y2::north" not a simple identifier
0120eecf 1348 sub snurt : foo + bar ; # "+" not a colon or space
09bef843
SB
1349
1350The attribute list is passed as a list of constant strings to the code
1351which associates them with the subroutine. In particular, the second example
1352of valid syntax above currently looks like this in terms of how it's
1353parsed and invoked:
1354
1355 use attributes __PACKAGE__, \&plugh, q[Ugly('\(")], 'Bad';
1356
1357For further details on attribute lists and their manipulation,
a0ae32d3 1358see L<attributes> and L<Attribute::Handlers>.
09bef843 1359
cb1a09d0 1360=head1 SEE ALSO
a0d0e21e 1361
19799a22
GS
1362See L<perlref/"Function Templates"> for more about references and closures.
1363See L<perlxs> if you'd like to learn about calling C subroutines from Perl.
a2293a43 1364See L<perlembed> if you'd like to learn about calling Perl subroutines from C.
19799a22
GS
1365See L<perlmod> to learn about bundling up your functions in separate files.
1366See L<perlmodlib> to learn what library modules come standard on your system.
1367See L<perltoot> to learn how to make object method calls.