This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
move sub attributes before the signature
[perl5.git] / pod / perlsub.pod
CommitLineData
a0d0e21e 1=head1 NAME
d74e8afc 2X<subroutine> X<function>
a0d0e21e
LW
3
4perlsub - Perl subroutines
5
6=head1 SYNOPSIS
7
8To declare subroutines:
d74e8afc 9X<subroutine, declaration> X<sub>
a0d0e21e 10
09bef843
SB
11 sub NAME; # A "forward" declaration.
12 sub NAME(PROTO); # ditto, but with prototypes
13 sub NAME : ATTRS; # with attributes
14 sub NAME(PROTO) : ATTRS; # with attributes and prototypes
cb1a09d0 15
09bef843
SB
16 sub NAME BLOCK # A declaration and a definition.
17 sub NAME(PROTO) BLOCK # ditto, but with prototypes
18 sub NAME : ATTRS BLOCK # with attributes
19 sub NAME(PROTO) : ATTRS BLOCK # with prototypes and attributes
894f226e
DM
20
21 use feature 'signatures';
22 sub NAME(SIG) BLOCK # with signature
23 sub NAME :ATTRS (SIG) BLOCK # with signature, attributes
24 sub NAME :prototype(PROTO) (SIG) BLOCK # with signature, prototype
a0d0e21e 25
748a9306 26To define an anonymous subroutine at runtime:
d74e8afc 27X<subroutine, anonymous>
748a9306 28
09bef843
SB
29 $subref = sub BLOCK; # no proto
30 $subref = sub (PROTO) BLOCK; # with proto
31 $subref = sub : ATTRS BLOCK; # with attributes
32 $subref = sub (PROTO) : ATTRS BLOCK; # with proto and attributes
894f226e
DM
33
34 use feature 'signatures';
35 $subref = sub (SIG) BLOCK; # with signature
36 $subref = sub : ATTRS(SIG) BLOCK; # with signature, attributes
748a9306 37
a0d0e21e 38To import subroutines:
d74e8afc 39X<import>
a0d0e21e 40
19799a22 41 use MODULE qw(NAME1 NAME2 NAME3);
a0d0e21e
LW
42
43To call subroutines:
d74e8afc 44X<subroutine, call> X<call>
a0d0e21e 45
5f05dabc 46 NAME(LIST); # & is optional with parentheses.
54310121 47 NAME LIST; # Parentheses optional if predeclared/imported.
19799a22 48 &NAME(LIST); # Circumvent prototypes.
5a964f20 49 &NAME; # Makes current @_ visible to called subroutine.
a0d0e21e
LW
50
51=head1 DESCRIPTION
52
19799a22
GS
53Like many languages, Perl provides for user-defined subroutines.
54These may be located anywhere in the main program, loaded in from
55other files via the C<do>, C<require>, or C<use> keywords, or
be3174d2 56generated on the fly using C<eval> or anonymous subroutines.
19799a22
GS
57You can even call a function indirectly using a variable containing
58its name or a CODE reference.
cb1a09d0
AD
59
60The Perl model for function call and return values is simple: all
61functions are passed as parameters one single flat list of scalars, and
62all functions likewise return to their caller one single flat list of
63scalars. Any arrays or hashes in these call and return lists will
64collapse, losing their identities--but you may always use
65pass-by-reference instead to avoid this. Both call and return lists may
66contain as many or as few scalar elements as you'd like. (Often a
67function without an explicit return statement is called a subroutine, but
19799a22 68there's really no difference from Perl's perspective.)
d74e8afc 69X<subroutine, parameter> X<parameter>
19799a22 70
30d9c59b
Z
71Any arguments passed in show up in the array C<@_>.
72(They may also show up in lexical variables introduced by a signature;
73see L</Signatures> below.) Therefore, if
19799a22
GS
74you called a function with two arguments, those would be stored in
75C<$_[0]> and C<$_[1]>. The array C<@_> is a local array, but its
76elements are aliases for the actual scalar parameters. In particular,
77if an element C<$_[0]> is updated, the corresponding argument is
78updated (or an error occurs if it is not updatable). If an argument
79is an array or hash element which did not exist when the function
80was called, that element is created only when (and if) it is modified
81or a reference to it is taken. (Some earlier versions of Perl
82created the element whether or not the element was assigned to.)
83Assigning to the whole array C<@_> removes that aliasing, and does
84not update any arguments.
d74e8afc 85X<subroutine, argument> X<argument> X<@_>
19799a22 86
dbb128be
XN
87A C<return> statement may be used to exit a subroutine, optionally
88specifying the returned value, which will be evaluated in the
89appropriate context (list, scalar, or void) depending on the context of
90the subroutine call. If you specify no return value, the subroutine
91returns an empty list in list context, the undefined value in scalar
92context, or nothing in void context. If you return one or more
93aggregates (arrays and hashes), these will be flattened together into
94one large indistinguishable list.
95
96If no C<return> is found and if the last statement is an expression, its
b77865f5
FC
97value is returned. If the last statement is a loop control structure
98like a C<foreach> or a C<while>, the returned value is unspecified. The
9a989771 99empty sub returns the empty list.
d74e8afc 100X<subroutine, return value> X<return value> X<return>
19799a22 101
30d9c59b 102Aside from an experimental facility (see L</Signatures> below),
19799a22
GS
103Perl does not have named formal parameters. In practice all you
104do is assign to a C<my()> list of these. Variables that aren't
105declared to be private are global variables. For gory details
5a0de581
LM
106on creating private variables, see L</"Private Variables via my()">
107and L</"Temporary Values via local()">. To create protected
19799a22
GS
108environments for a set of functions in a separate package (and
109probably a separate file), see L<perlmod/"Packages">.
d74e8afc 110X<formal parameter> X<parameter, formal>
a0d0e21e
LW
111
112Example:
113
cb1a09d0
AD
114 sub max {
115 my $max = shift(@_);
a0d0e21e
LW
116 foreach $foo (@_) {
117 $max = $foo if $max < $foo;
118 }
cb1a09d0 119 return $max;
a0d0e21e 120 }
cb1a09d0 121 $bestday = max($mon,$tue,$wed,$thu,$fri);
a0d0e21e
LW
122
123Example:
124
125 # get a line, combining continuation lines
126 # that start with whitespace
127
128 sub get_line {
19799a22 129 $thisline = $lookahead; # global variables!
54310121 130 LINE: while (defined($lookahead = <STDIN>)) {
a0d0e21e
LW
131 if ($lookahead =~ /^[ \t]/) {
132 $thisline .= $lookahead;
133 }
134 else {
135 last LINE;
136 }
137 }
19799a22 138 return $thisline;
a0d0e21e
LW
139 }
140
141 $lookahead = <STDIN>; # get first line
19799a22 142 while (defined($line = get_line())) {
a0d0e21e
LW
143 ...
144 }
145
09bef843 146Assigning to a list of private variables to name your arguments:
a0d0e21e
LW
147
148 sub maybeset {
149 my($key, $value) = @_;
cb1a09d0 150 $Foo{$key} = $value unless $Foo{$key};
a0d0e21e
LW
151 }
152
19799a22
GS
153Because the assignment copies the values, this also has the effect
154of turning call-by-reference into call-by-value. Otherwise a
155function is free to do in-place modifications of C<@_> and change
156its caller's values.
d74e8afc 157X<call-by-reference> X<call-by-value>
cb1a09d0
AD
158
159 upcase_in($v1, $v2); # this changes $v1 and $v2
160 sub upcase_in {
54310121
PP
161 for (@_) { tr/a-z/A-Z/ }
162 }
cb1a09d0
AD
163
164You aren't allowed to modify constants in this way, of course. If an
165argument were actually literal and you tried to change it, you'd take a
166(presumably fatal) exception. For example, this won't work:
d74e8afc 167X<call-by-reference> X<call-by-value>
cb1a09d0
AD
168
169 upcase_in("frederick");
170
f86cebdf 171It would be much safer if the C<upcase_in()> function
cb1a09d0
AD
172were written to return a copy of its parameters instead
173of changing them in place:
174
19799a22 175 ($v3, $v4) = upcase($v1, $v2); # this doesn't change $v1 and $v2
cb1a09d0 176 sub upcase {
54310121 177 return unless defined wantarray; # void context, do nothing
cb1a09d0 178 my @parms = @_;
54310121 179 for (@parms) { tr/a-z/A-Z/ }
c07a80fd 180 return wantarray ? @parms : $parms[0];
54310121 181 }
cb1a09d0 182
19799a22 183Notice how this (unprototyped) function doesn't care whether it was
a2293a43 184passed real scalars or arrays. Perl sees all arguments as one big,
19799a22
GS
185long, flat parameter list in C<@_>. This is one area where
186Perl's simple argument-passing style shines. The C<upcase()>
187function would work perfectly well without changing the C<upcase()>
188definition even if we fed it things like this:
cb1a09d0
AD
189
190 @newlist = upcase(@list1, @list2);
191 @newlist = upcase( split /:/, $var );
192
193Do not, however, be tempted to do this:
194
195 (@a, @b) = upcase(@list1, @list2);
196
19799a22
GS
197Like the flattened incoming parameter list, the return list is also
198flattened on return. So all you have managed to do here is stored
17b63f68 199everything in C<@a> and made C<@b> empty. See
5a0de581 200L</Pass by Reference> for alternatives.
19799a22
GS
201
202A subroutine may be called using an explicit C<&> prefix. The
203C<&> is optional in modern Perl, as are parentheses if the
204subroutine has been predeclared. The C<&> is I<not> optional
205when just naming the subroutine, such as when it's used as
206an argument to defined() or undef(). Nor is it optional when you
207want to do an indirect subroutine call with a subroutine name or
208reference using the C<&$subref()> or C<&{$subref}()> constructs,
c47ff5f1 209although the C<< $subref->() >> notation solves that problem.
19799a22 210See L<perlref> for more about all that.
d74e8afc 211X<&>
19799a22
GS
212
213Subroutines may be called recursively. If a subroutine is called
214using the C<&> form, the argument list is optional, and if omitted,
215no C<@_> array is set up for the subroutine: the C<@_> array at the
216time of the call is visible to subroutine instead. This is an
217efficiency mechanism that new users may wish to avoid.
d74e8afc 218X<recursion>
a0d0e21e
LW
219
220 &foo(1,2,3); # pass three arguments
221 foo(1,2,3); # the same
222
223 foo(); # pass a null list
224 &foo(); # the same
a0d0e21e 225
cb1a09d0 226 &foo; # foo() get current args, like foo(@_) !!
54310121 227 foo; # like foo() IFF sub foo predeclared, else "foo"
cb1a09d0 228
19799a22
GS
229Not only does the C<&> form make the argument list optional, it also
230disables any prototype checking on arguments you do provide. This
c07a80fd 231is partly for historical reasons, and partly for having a convenient way
9688be67 232to cheat if you know what you're doing. See L</Prototypes> below.
d74e8afc 233X<&>
c07a80fd 234
977616ef
RS
235Since Perl 5.16.0, the C<__SUB__> token is available under C<use feature
236'current_sub'> and C<use 5.16.0>. It will evaluate to a reference to the
906024c7
FC
237currently-running sub, which allows for recursive calls without knowing
238your subroutine's name.
977616ef
RS
239
240 use 5.16.0;
241 my $factorial = sub {
242 my ($x) = @_;
243 return 1 if $x == 1;
244 return($x * __SUB__->( $x - 1 ) );
245 };
246
89d1beed 247The behavior of C<__SUB__> within a regex code block (such as C</(?{...})/>)
a453e28a
DM
248is subject to change.
249
ac90fb77
EM
250Subroutines whose names are in all upper case are reserved to the Perl
251core, as are modules whose names are in all lower case. A subroutine in
252all capitals is a loosely-held convention meaning it will be called
253indirectly by the run-time system itself, usually due to a triggered event.
bf5513e0 254Subroutines whose name start with a left parenthesis are also reserved the
b77865f5 255same way. The following is a list of some subroutines that currently do
bf5513e0
ZA
256special, pre-defined things.
257
258=over
259
260=item documented later in this document
261
262C<AUTOLOAD>
263
264=item documented in L<perlmod>
265
8b7906d1 266C<CLONE>, C<CLONE_SKIP>
bf5513e0
ZA
267
268=item documented in L<perlobj>
269
8b7906d1 270C<DESTROY>, C<DOES>
bf5513e0
ZA
271
272=item documented in L<perltie>
273
274C<BINMODE>, C<CLEAR>, C<CLOSE>, C<DELETE>, C<DESTROY>, C<EOF>, C<EXISTS>,
275C<EXTEND>, C<FETCH>, C<FETCHSIZE>, C<FILENO>, C<FIRSTKEY>, C<GETC>,
276C<NEXTKEY>, C<OPEN>, C<POP>, C<PRINT>, C<PRINTF>, C<PUSH>, C<READ>,
277C<READLINE>, C<SCALAR>, C<SEEK>, C<SHIFT>, C<SPLICE>, C<STORE>,
278C<STORESIZE>, C<TELL>, C<TIEARRAY>, C<TIEHANDLE>, C<TIEHASH>,
279C<TIESCALAR>, C<UNSHIFT>, C<UNTIE>, C<WRITE>
280
281=item documented in L<PerlIO::via>
282
283C<BINMODE>, C<CLEARERR>, C<CLOSE>, C<EOF>, C<ERROR>, C<FDOPEN>, C<FILENO>,
284C<FILL>, C<FLUSH>, C<OPEN>, C<POPPED>, C<PUSHED>, C<READ>, C<SEEK>,
285C<SETLINEBUF>, C<SYSOPEN>, C<TELL>, C<UNREAD>, C<UTF8>, C<WRITE>
286
ec2eb8a9
TC
287=item documented in L<perlfunc>
288
289L<< C<import> | perlfunc/use >>, L<< C<unimport> | perlfunc/use >>,
290L<< C<INC> | perlfunc/require >>
291
292=item documented in L<UNIVERSAL>
293
294C<VERSION>
295
296=item documented in L<perldebguts>
297
298C<DB::DB>, C<DB::sub>, C<DB::lsub>, C<DB::goto>, C<DB::postponed>
299
bf5513e0
ZA
300=item undocumented, used internally by the L<overload> feature
301
302any starting with C<(>
303
304=back
ac90fb77 305
3c10abe3
AG
306The C<BEGIN>, C<UNITCHECK>, C<CHECK>, C<INIT> and C<END> subroutines
307are not so much subroutines as named special code blocks, of which you
308can have more than one in a package, and which you can B<not> call
309explicitly. See L<perlmod/"BEGIN, UNITCHECK, CHECK, INIT and END">
5a964f20 310
30d9c59b
Z
311=head2 Signatures
312
313B<WARNING>: Subroutine signatures are experimental. The feature may be
314modified or removed in future versions of Perl.
315
316Perl has an experimental facility to allow a subroutine's formal
317parameters to be introduced by special syntax, separate from the
318procedural code of the subroutine body. The formal parameter list
319is known as a I<signature>. The facility must be enabled first by a
320pragmatic declaration, C<use feature 'signatures'>, and it will produce
321a warning unless the "experimental::signatures" warnings category is
322disabled.
323
324The signature is part of a subroutine's body. Normally the body of a
894f226e
DM
325subroutine is simply a braced block of code, but when using a signature,
326the signature is a parenthesised list that goes immediately before the
327block, after any name or attributes.
328
329For example,
330
331 sub foo :lvalue ($a, $b = 1, @c) { .... }
332
333The signature declares lexical variables that are
30d9c59b
Z
334in scope for the block. When the subroutine is called, the signature
335takes control first. It populates the signature variables from the
336list of arguments that were passed. If the argument list doesn't meet
337the requirements of the signature, then it will throw an exception.
338When the signature processing is complete, control passes to the block.
339
340Positional parameters are handled by simply naming scalar variables in
341the signature. For example,
342
343 sub foo ($left, $right) {
344 return $left + $right;
345 }
346
347takes two positional parameters, which must be filled at runtime by
348two arguments. By default the parameters are mandatory, and it is
349not permitted to pass more arguments than expected. So the above is
350equivalent to
351
352 sub foo {
353 die "Too many arguments for subroutine" unless @_ <= 2;
354 die "Too few arguments for subroutine" unless @_ >= 2;
355 my $left = $_[0];
356 my $right = $_[1];
357 return $left + $right;
358 }
359
360An argument can be ignored by omitting the main part of the name from
361a parameter declaration, leaving just a bare C<$> sigil. For example,
362
363 sub foo ($first, $, $third) {
364 return "first=$first, third=$third";
365 }
366
367Although the ignored argument doesn't go into a variable, it is still
368mandatory for the caller to pass it.
369
370A positional parameter is made optional by giving a default value,
371separated from the parameter name by C<=>:
372
373 sub foo ($left, $right = 0) {
374 return $left + $right;
375 }
376
377The above subroutine may be called with either one or two arguments.
378The default value expression is evaluated when the subroutine is called,
379so it may provide different default values for different calls. It is
380only evaluated if the argument was actually omitted from the call.
381For example,
382
383 my $auto_id = 0;
384 sub foo ($thing, $id = $auto_id++) {
385 print "$thing has ID $id";
386 }
387
388automatically assigns distinct sequential IDs to things for which no
389ID was supplied by the caller. A default value expression may also
390refer to parameters earlier in the signature, making the default for
391one parameter vary according to the earlier parameters. For example,
392
393 sub foo ($first_name, $surname, $nickname = $first_name) {
394 print "$first_name $surname is known as \"$nickname\"";
395 }
396
397An optional parameter can be nameless just like a mandatory parameter.
398For example,
399
400 sub foo ($thing, $ = 1) {
401 print $thing;
402 }
403
404The parameter's default value will still be evaluated if the corresponding
405argument isn't supplied, even though the value won't be stored anywhere.
406This is in case evaluating it has important side effects. However, it
407will be evaluated in void context, so if it doesn't have side effects
408and is not trivial it will generate a warning if the "void" warning
409category is enabled. If a nameless optional parameter's default value
410is not important, it may be omitted just as the parameter's name was:
411
412 sub foo ($thing, $=) {
413 print $thing;
414 }
415
416Optional positional parameters must come after all mandatory positional
417parameters. (If there are no mandatory positional parameters then an
418optional positional parameters can be the first thing in the signature.)
419If there are multiple optional positional parameters and not enough
420arguments are supplied to fill them all, they will be filled from left
421to right.
422
423After positional parameters, additional arguments may be captured in a
424slurpy parameter. The simplest form of this is just an array variable:
425
426 sub foo ($filter, @inputs) {
427 print $filter->($_) foreach @inputs;
428 }
429
430With a slurpy parameter in the signature, there is no upper limit on how
431many arguments may be passed. A slurpy array parameter may be nameless
432just like a positional parameter, in which case its only effect is to
433turn off the argument limit that would otherwise apply:
434
435 sub foo ($thing, @) {
436 print $thing;
437 }
438
439A slurpy parameter may instead be a hash, in which case the arguments
440available to it are interpreted as alternating keys and values.
441There must be as many keys as values: if there is an odd argument then
442an exception will be thrown. Keys will be stringified, and if there are
443duplicates then the later instance takes precedence over the earlier,
444as with standard hash construction.
445
446 sub foo ($filter, %inputs) {
447 print $filter->($_, $inputs{$_}) foreach sort keys %inputs;
448 }
449
450A slurpy hash parameter may be nameless just like other kinds of
451parameter. It still insists that the number of arguments available to
452it be even, even though they're not being put into a variable.
453
454 sub foo ($thing, %) {
455 print $thing;
456 }
457
458A slurpy parameter, either array or hash, must be the last thing in the
459signature. It may follow mandatory and optional positional parameters;
460it may also be the only thing in the signature. Slurpy parameters cannot
461have default values: if no arguments are supplied for them then you get
462an empty array or empty hash.
463
464A signature may be entirely empty, in which case all it does is check
465that the caller passed no arguments:
466
467 sub foo () {
468 return 123;
469 }
470
8c13e946
RS
471When using a signature, the arguments are still available in the special
472array variable C<@_>, in addition to the lexical variables of the
473signature. There is a difference between the two ways of accessing the
30d9c59b
Z
474arguments: C<@_> I<aliases> the arguments, but the signature variables
475get I<copies> of the arguments. So writing to a signature variable
476only changes that variable, and has no effect on the caller's variables,
477but writing to an element of C<@_> modifies whatever the caller used to
478supply that argument.
479
480There is a potential syntactic ambiguity between signatures and prototypes
481(see L</Prototypes>), because both start with an opening parenthesis and
482both can appear in some of the same places, such as just after the name
483in a subroutine declaration. For historical reasons, when signatures
484are not enabled, any opening parenthesis in such a context will trigger
485very forgiving prototype parsing. Most signatures will be interpreted
486as prototypes in those circumstances, but won't be valid prototypes.
487(A valid prototype cannot contain any alphabetic character.) This will
488lead to somewhat confusing error messages.
489
490To avoid ambiguity, when signatures are enabled the special syntax
491for prototypes is disabled. There is no attempt to guess whether a
492parenthesised group was intended to be a prototype or a signature.
493To give a subroutine a prototype under these circumstances, use a
494L<prototype attribute|attributes/Built-in Attributes>. For example,
495
496 sub foo :prototype($) { $_[0] }
497
498It is entirely possible for a subroutine to have both a prototype and
499a signature. They do different jobs: the prototype affects compilation
500of calls to the subroutine, and the signature puts argument values into
501lexical variables at runtime. You can therefore write
502
894f226e 503 sub foo :prototype($$) ($left, $right) {
30d9c59b
Z
504 return $left + $right;
505 }
506
894f226e
DM
507The prototype attribute, and any other attributes, must come before
508the signature. The signature always immediately precedes the block of
509the subroutine's body.
30d9c59b 510
b687b08b 511=head2 Private Variables via my()
d74e8afc
ITB
512X<my> X<variable, lexical> X<lexical> X<lexical variable> X<scope, lexical>
513X<lexical scope> X<attributes, my>
cb1a09d0
AD
514
515Synopsis:
516
517 my $foo; # declare $foo lexically local
518 my (@wid, %get); # declare list of variables local
519 my $foo = "flurp"; # declare $foo lexical, and init it
520 my @oof = @bar; # declare @oof lexical, and init it
09bef843
SB
521 my $x : Foo = $y; # similar, with an attribute applied
522
a0ae32d3
JH
523B<WARNING>: The use of attribute lists on C<my> declarations is still
524evolving. The current semantics and interface are subject to change.
525See L<attributes> and L<Attribute::Handlers>.
cb1a09d0 526
19799a22 527The C<my> operator declares the listed variables to be lexically
f185f654
KW
528confined to the enclosing block, conditional
529(C<if>/C<unless>/C<elsif>/C<else>), loop
530(C<for>/C<foreach>/C<while>/C<until>/C<continue>), subroutine, C<eval>,
531or C<do>/C<require>/C<use>'d file. If more than one value is listed, the
19799a22
GS
532list must be placed in parentheses. All listed elements must be
533legal lvalues. Only alphanumeric identifiers may be lexically
325192b1 534scoped--magical built-ins like C<$/> must currently be C<local>ized
19799a22
GS
535with C<local> instead.
536
537Unlike dynamic variables created by the C<local> operator, lexical
538variables declared with C<my> are totally hidden from the outside
539world, including any called subroutines. This is true if it's the
540same subroutine called from itself or elsewhere--every call gets
541its own copy.
d74e8afc 542X<local>
19799a22
GS
543
544This doesn't mean that a C<my> variable declared in a statically
545enclosing lexical scope would be invisible. Only dynamic scopes
546are cut off. For example, the C<bumpx()> function below has access
547to the lexical $x variable because both the C<my> and the C<sub>
548occurred at the same scope, presumably file scope.
5a964f20
TC
549
550 my $x = 10;
551 sub bumpx { $x++ }
552
19799a22
GS
553An C<eval()>, however, can see lexical variables of the scope it is
554being evaluated in, so long as the names aren't hidden by declarations within
555the C<eval()> itself. See L<perlref>.
d74e8afc 556X<eval, scope of>
cb1a09d0 557
19799a22 558The parameter list to my() may be assigned to if desired, which allows you
cb1a09d0
AD
559to initialize your variables. (If no initializer is given for a
560particular variable, it is created with the undefined value.) Commonly
19799a22 561this is used to name input parameters to a subroutine. Examples:
cb1a09d0
AD
562
563 $arg = "fred"; # "global" variable
564 $n = cube_root(27);
565 print "$arg thinks the root is $n\n";
566 fred thinks the root is 3
567
568 sub cube_root {
569 my $arg = shift; # name doesn't matter
570 $arg **= 1/3;
571 return $arg;
54310121 572 }
cb1a09d0 573
19799a22
GS
574The C<my> is simply a modifier on something you might assign to. So when
575you do assign to variables in its argument list, C<my> doesn't
6cc33c6d 576change whether those variables are viewed as a scalar or an array. So
cb1a09d0 577
5a964f20 578 my ($foo) = <STDIN>; # WRONG?
cb1a09d0
AD
579 my @FOO = <STDIN>;
580
5f05dabc 581both supply a list context to the right-hand side, while
cb1a09d0
AD
582
583 my $foo = <STDIN>;
584
5f05dabc 585supplies a scalar context. But the following declares only one variable:
748a9306 586
5a964f20 587 my $foo, $bar = 1; # WRONG
748a9306 588
cb1a09d0 589That has the same effect as
748a9306 590
cb1a09d0
AD
591 my $foo;
592 $bar = 1;
a0d0e21e 593
cb1a09d0
AD
594The declared variable is not introduced (is not visible) until after
595the current statement. Thus,
596
597 my $x = $x;
598
19799a22 599can be used to initialize a new $x with the value of the old $x, and
cb1a09d0
AD
600the expression
601
602 my $x = 123 and $x == 123
603
19799a22 604is false unless the old $x happened to have the value C<123>.
cb1a09d0 605
55497cff
PP
606Lexical scopes of control structures are not bounded precisely by the
607braces that delimit their controlled blocks; control expressions are
19799a22 608part of that scope, too. Thus in the loop
55497cff 609
19799a22 610 while (my $line = <>) {
55497cff
PP
611 $line = lc $line;
612 } continue {
613 print $line;
614 }
615
19799a22 616the scope of $line extends from its declaration throughout the rest of
55497cff
PP
617the loop construct (including the C<continue> clause), but not beyond
618it. Similarly, in the conditional
619
620 if ((my $answer = <STDIN>) =~ /^yes$/i) {
621 user_agrees();
622 } elsif ($answer =~ /^no$/i) {
623 user_disagrees();
624 } else {
625 chomp $answer;
626 die "'$answer' is neither 'yes' nor 'no'";
627 }
628
19799a22
GS
629the scope of $answer extends from its declaration through the rest
630of that conditional, including any C<elsif> and C<else> clauses,
96090e4f 631but not beyond it. See L<perlsyn/"Simple Statements"> for information
457b36cb 632on the scope of variables in statements with modifiers.
55497cff 633
5f05dabc 634The C<foreach> loop defaults to scoping its index variable dynamically
19799a22
GS
635in the manner of C<local>. However, if the index variable is
636prefixed with the keyword C<my>, or if there is already a lexical
637by that name in scope, then a new lexical is created instead. Thus
638in the loop
d74e8afc 639X<foreach> X<for>
55497cff
PP
640
641 for my $i (1, 2, 3) {
642 some_function();
643 }
644
19799a22
GS
645the scope of $i extends to the end of the loop, but not beyond it,
646rendering the value of $i inaccessible within C<some_function()>.
d74e8afc 647X<foreach> X<for>
55497cff 648
cb1a09d0 649Some users may wish to encourage the use of lexically scoped variables.
19799a22
GS
650As an aid to catching implicit uses to package variables,
651which are always global, if you say
cb1a09d0
AD
652
653 use strict 'vars';
654
19799a22
GS
655then any variable mentioned from there to the end of the enclosing
656block must either refer to a lexical variable, be predeclared via
77ca0c92 657C<our> or C<use vars>, or else must be fully qualified with the package name.
19799a22
GS
658A compilation error results otherwise. An inner block may countermand
659this with C<no strict 'vars'>.
660
661A C<my> has both a compile-time and a run-time effect. At compile
8593bda5 662time, the compiler takes notice of it. The principal usefulness
19799a22
GS
663of this is to quiet C<use strict 'vars'>, but it is also essential
664for generation of closures as detailed in L<perlref>. Actual
665initialization is delayed until run time, though, so it gets executed
666at the appropriate time, such as each time through a loop, for
667example.
668
669Variables declared with C<my> are not part of any package and are therefore
cb1a09d0
AD
670never fully qualified with the package name. In particular, you're not
671allowed to try to make a package variable (or other global) lexical:
672
673 my $pack::var; # ERROR! Illegal syntax
cb1a09d0
AD
674
675In fact, a dynamic variable (also known as package or global variables)
f86cebdf 676are still accessible using the fully qualified C<::> notation even while a
cb1a09d0
AD
677lexical of the same name is also visible:
678
679 package main;
680 local $x = 10;
681 my $x = 20;
682 print "$x and $::x\n";
683
f86cebdf 684That will print out C<20> and C<10>.
cb1a09d0 685
19799a22
GS
686You may declare C<my> variables at the outermost scope of a file
687to hide any such identifiers from the world outside that file. This
688is similar in spirit to C's static variables when they are used at
689the file level. To do this with a subroutine requires the use of
690a closure (an anonymous function that accesses enclosing lexicals).
691If you want to create a private subroutine that cannot be called
692from outside that block, it can declare a lexical variable containing
693an anonymous sub reference:
cb1a09d0
AD
694
695 my $secret_version = '1.001-beta';
696 my $secret_sub = sub { print $secret_version };
697 &$secret_sub();
698
699As long as the reference is never returned by any function within the
5f05dabc 700module, no outside module can see the subroutine, because its name is not in
cb1a09d0 701any package's symbol table. Remember that it's not I<REALLY> called
19799a22 702C<$some_pack::secret_version> or anything; it's just $secret_version,
cb1a09d0
AD
703unqualified and unqualifiable.
704
19799a22
GS
705This does not work with object methods, however; all object methods
706have to be in the symbol table of some package to be found. See
707L<perlref/"Function Templates"> for something of a work-around to
708this.
cb1a09d0 709
c2611fb3 710=head2 Persistent Private Variables
ba1f8e91
RGS
711X<state> X<state variable> X<static> X<variable, persistent> X<variable, static> X<closure>
712
713There are two ways to build persistent private variables in Perl 5.10.
b77865f5 714First, you can simply use the C<state> feature. Or, you can use closures,
ba1f8e91
RGS
715if you want to stay compatible with releases older than 5.10.
716
717=head3 Persistent variables via state()
718
9d42615f 719Beginning with Perl 5.10.0, you can declare variables with the C<state>
4a904372 720keyword in place of C<my>. For that to work, though, you must have
ba1f8e91 721enabled that feature beforehand, either by using the C<feature> pragma, or
4a904372 722by using C<-E> on one-liners (see L<feature>). Beginning with Perl 5.16,
47d235f1 723the C<CORE::state> form does not require the
4a904372 724C<feature> pragma.
ba1f8e91 725
ad0cc46c
FC
726The C<state> keyword creates a lexical variable (following the same scoping
727rules as C<my>) that persists from one subroutine call to the next. If a
728state variable resides inside an anonymous subroutine, then each copy of
729the subroutine has its own copy of the state variable. However, the value
730of the state variable will still persist between calls to the same copy of
731the anonymous subroutine. (Don't forget that C<sub { ... }> creates a new
732subroutine each time it is executed.)
733
ba1f8e91
RGS
734For example, the following code maintains a private counter, incremented
735each time the gimme_another() function is called:
736
737 use feature 'state';
738 sub gimme_another { state $x; return ++$x }
739
ad0cc46c
FC
740And this example uses anonymous subroutines to create separate counters:
741
742 use feature 'state';
743 sub create_counter {
744 return sub { state $x; return ++$x }
745 }
746
ba1f8e91
RGS
747Also, since C<$x> is lexical, it can't be reached or modified by any Perl
748code outside.
749
f99042c8 750When combined with variable declaration, simple assignment to C<state>
f292fc7a
RS
751variables (as in C<state $x = 42>) is executed only the first time. When such
752statements are evaluated subsequent times, the assignment is ignored. The
f99042c8
Z
753behavior of assignment to C<state> declarations where the left hand side
754of the assignment involves any parentheses is currently undefined.
ba1f8e91
RGS
755
756=head3 Persistent variables with closures
5a964f20
TC
757
758Just because a lexical variable is lexically (also called statically)
f86cebdf 759scoped to its enclosing block, C<eval>, or C<do> FILE, this doesn't mean that
5a964f20
TC
760within a function it works like a C static. It normally works more
761like a C auto, but with implicit garbage collection.
762
763Unlike local variables in C or C++, Perl's lexical variables don't
764necessarily get recycled just because their scope has exited.
765If something more permanent is still aware of the lexical, it will
766stick around. So long as something else references a lexical, that
767lexical won't be freed--which is as it should be. You wouldn't want
768memory being free until you were done using it, or kept around once you
769were done. Automatic garbage collection takes care of this for you.
770
771This means that you can pass back or save away references to lexical
772variables, whereas to return a pointer to a C auto is a grave error.
773It also gives us a way to simulate C's function statics. Here's a
774mechanism for giving a function private variables with both lexical
775scoping and a static lifetime. If you do want to create something like
776C's static variables, just enclose the whole function in an extra block,
777and put the static variable outside the function but in the block.
cb1a09d0
AD
778
779 {
54310121 780 my $secret_val = 0;
cb1a09d0
AD
781 sub gimme_another {
782 return ++$secret_val;
54310121
PP
783 }
784 }
cb1a09d0
AD
785 # $secret_val now becomes unreachable by the outside
786 # world, but retains its value between calls to gimme_another
787
54310121 788If this function is being sourced in from a separate file
cb1a09d0 789via C<require> or C<use>, then this is probably just fine. If it's
19799a22 790all in the main program, you'll need to arrange for the C<my>
cb1a09d0 791to be executed early, either by putting the whole block above
f86cebdf 792your main program, or more likely, placing merely a C<BEGIN>
ac90fb77 793code block around it to make sure it gets executed before your program
cb1a09d0
AD
794starts to run:
795
ac90fb77 796 BEGIN {
54310121 797 my $secret_val = 0;
cb1a09d0
AD
798 sub gimme_another {
799 return ++$secret_val;
54310121
PP
800 }
801 }
cb1a09d0 802
3c10abe3
AG
803See L<perlmod/"BEGIN, UNITCHECK, CHECK, INIT and END"> about the
804special triggered code blocks, C<BEGIN>, C<UNITCHECK>, C<CHECK>,
805C<INIT> and C<END>.
cb1a09d0 806
19799a22
GS
807If declared at the outermost scope (the file scope), then lexicals
808work somewhat like C's file statics. They are available to all
809functions in that same file declared below them, but are inaccessible
810from outside that file. This strategy is sometimes used in modules
811to create private variables that the whole module can see.
5a964f20 812
cb1a09d0 813=head2 Temporary Values via local()
d74e8afc
ITB
814X<local> X<scope, dynamic> X<dynamic scope> X<variable, local>
815X<variable, temporary>
cb1a09d0 816
19799a22 817B<WARNING>: In general, you should be using C<my> instead of C<local>, because
6d28dffb 818it's faster and safer. Exceptions to this include the global punctuation
325192b1
RGS
819variables, global filehandles and formats, and direct manipulation of the
820Perl symbol table itself. C<local> is mostly used when the current value
821of a variable must be visible to called subroutines.
cb1a09d0
AD
822
823Synopsis:
824
325192b1
RGS
825 # localization of values
826
555bd962
BG
827 local $foo; # make $foo dynamically local
828 local (@wid, %get); # make list of variables local
829 local $foo = "flurp"; # make $foo dynamic, and init it
830 local @oof = @bar; # make @oof dynamic, and init it
325192b1 831
555bd962
BG
832 local $hash{key} = "val"; # sets a local value for this hash entry
833 delete local $hash{key}; # delete this entry for the current block
834 local ($cond ? $v1 : $v2); # several types of lvalues support
835 # localization
325192b1
RGS
836
837 # localization of symbols
cb1a09d0 838
555bd962
BG
839 local *FH; # localize $FH, @FH, %FH, &FH ...
840 local *merlyn = *randal; # now $merlyn is really $randal, plus
841 # @merlyn is really @randal, etc
842 local *merlyn = 'randal'; # SAME THING: promote 'randal' to *randal
843 local *merlyn = \$randal; # just alias $merlyn, not @merlyn etc
cb1a09d0 844
19799a22
GS
845A C<local> modifies its listed variables to be "local" to the
846enclosing block, C<eval>, or C<do FILE>--and to I<any subroutine
847called from within that block>. A C<local> just gives temporary
848values to global (meaning package) variables. It does I<not> create
849a local variable. This is known as dynamic scoping. Lexical scoping
850is done with C<my>, which works more like C's auto declarations.
cb1a09d0 851
ceb12f1f 852Some types of lvalues can be localized as well: hash and array elements
325192b1
RGS
853and slices, conditionals (provided that their result is always
854localizable), and symbolic references. As for simple variables, this
855creates new, dynamically scoped values.
856
857If more than one variable or expression is given to C<local>, they must be
858placed in parentheses. This operator works
cb1a09d0 859by saving the current values of those variables in its argument list on a
5f05dabc 860hidden stack and restoring them upon exiting the block, subroutine, or
cb1a09d0
AD
861eval. This means that called subroutines can also reference the local
862variable, but not the global one. The argument list may be assigned to if
863desired, which allows you to initialize your local variables. (If no
864initializer is given for a particular variable, it is created with an
325192b1 865undefined value.)
cb1a09d0 866
19799a22 867Because C<local> is a run-time operator, it gets executed each time
325192b1
RGS
868through a loop. Consequently, it's more efficient to localize your
869variables outside the loop.
870
871=head3 Grammatical note on local()
d74e8afc 872X<local, context>
cb1a09d0 873
f86cebdf
GS
874A C<local> is simply a modifier on an lvalue expression. When you assign to
875a C<local>ized variable, the C<local> doesn't change whether its list is viewed
cb1a09d0
AD
876as a scalar or an array. So
877
878 local($foo) = <STDIN>;
879 local @FOO = <STDIN>;
880
5f05dabc 881both supply a list context to the right-hand side, while
cb1a09d0
AD
882
883 local $foo = <STDIN>;
884
885supplies a scalar context.
886
325192b1 887=head3 Localization of special variables
d74e8afc 888X<local, special variable>
3e3baf6d 889
325192b1
RGS
890If you localize a special variable, you'll be giving a new value to it,
891but its magic won't go away. That means that all side-effects related
892to this magic still work with the localized value.
3e3baf6d 893
325192b1
RGS
894This feature allows code like this to work :
895
896 # Read the whole contents of FILE in $slurp
897 { local $/ = undef; $slurp = <FILE>; }
898
899Note, however, that this restricts localization of some values ; for
9d42615f 900example, the following statement dies, as of perl 5.10.0, with an error
325192b1
RGS
901I<Modification of a read-only value attempted>, because the $1 variable is
902magical and read-only :
903
904 local $1 = 2;
905
658a9f31
JD
906One exception is the default scalar variable: starting with perl 5.14
907C<local($_)> will always strip all magic from $_, to make it possible
908to safely reuse $_ in a subroutine.
325192b1
RGS
909
910B<WARNING>: Localization of tied arrays and hashes does not currently
911work as described.
fd5a896a 912This will be fixed in a future release of Perl; in the meantime, avoid
89d1beed 913code that relies on any particular behavior of localising tied arrays
fd5a896a 914or hashes (localising individual elements is still okay).
325192b1 915See L<perl58delta/"Localising Tied Arrays and Hashes Is Broken"> for more
fd5a896a 916details.
d74e8afc 917X<local, tie>
fd5a896a 918
325192b1 919=head3 Localization of globs
d74e8afc 920X<local, glob> X<glob>
3e3baf6d 921
325192b1
RGS
922The construct
923
924 local *name;
925
926creates a whole new symbol table entry for the glob C<name> in the
927current package. That means that all variables in its glob slot ($name,
928@name, %name, &name, and the C<name> filehandle) are dynamically reset.
929
930This implies, among other things, that any magic eventually carried by
931those variables is locally lost. In other words, saying C<local */>
932will not have any effect on the internal value of the input record
933separator.
934
325192b1 935=head3 Localization of elements of composite types
d74e8afc 936X<local, composite type element> X<local, array element> X<local, hash element>
3e3baf6d 937
6ee623d5 938It's also worth taking a moment to explain what happens when you
f86cebdf 939C<local>ize a member of a composite type (i.e. an array or hash element).
b77865f5 940In this case, the element is C<local>ized I<by name>. This means that
6ee623d5
GS
941when the scope of the C<local()> ends, the saved value will be
942restored to the hash element whose key was named in the C<local()>, or
943the array element whose index was named in the C<local()>. If that
944element was deleted while the C<local()> was in effect (e.g. by a
945C<delete()> from a hash or a C<shift()> of an array), it will spring
946back into existence, possibly extending an array and filling in the
947skipped elements with C<undef>. For instance, if you say
948
949 %hash = ( 'This' => 'is', 'a' => 'test' );
950 @ary = ( 0..5 );
951 {
952 local($ary[5]) = 6;
953 local($hash{'a'}) = 'drill';
954 while (my $e = pop(@ary)) {
955 print "$e . . .\n";
956 last unless $e > 3;
957 }
958 if (@ary) {
959 $hash{'only a'} = 'test';
960 delete $hash{'a'};
961 }
962 }
963 print join(' ', map { "$_ $hash{$_}" } sort keys %hash),".\n";
964 print "The array has ",scalar(@ary)," elements: ",
965 join(', ', map { defined $_ ? $_ : 'undef' } @ary),"\n";
966
967Perl will print
968
969 6 . . .
970 4 . . .
971 3 . . .
972 This is a test only a test.
973 The array has 6 elements: 0, 1, 2, undef, undef, 5
974
19799a22 975The behavior of local() on non-existent members of composite
7185e5cc
GS
976types is subject to change in future.
977
d361fafa
VP
978=head3 Localized deletion of elements of composite types
979X<delete> X<local, composite type element> X<local, array element> X<local, hash element>
980
981You can use the C<delete local $array[$idx]> and C<delete local $hash{key}>
982constructs to delete a composite type entry for the current block and restore
b77865f5 983it when it ends. They return the array/hash value before the localization,
d361fafa
VP
984which means that they are respectively equivalent to
985
986 do {
987 my $val = $array[$idx];
988 local $array[$idx];
989 delete $array[$idx];
990 $val
991 }
992
993and
994
995 do {
996 my $val = $hash{key};
997 local $hash{key};
998 delete $hash{key};
999 $val
1000 }
1001
b77865f5
FC
1002except that for those the C<local> is
1003scoped to the C<do> block. Slices are
d361fafa
VP
1004also accepted.
1005
1006 my %hash = (
1007 a => [ 7, 8, 9 ],
1008 b => 1,
1009 )
1010
1011 {
1012 my $a = delete local $hash{a};
1013 # $a is [ 7, 8, 9 ]
1014 # %hash is (b => 1)
1015
1016 {
1017 my @nums = delete local @$a[0, 2]
1018 # @nums is (7, 9)
1019 # $a is [ undef, 8 ]
1020
1021 $a[0] = 999; # will be erased when the scope ends
1022 }
1023 # $a is back to [ 7, 8, 9 ]
1024
1025 }
1026 # %hash is back to its original state
1027
cd06dffe 1028=head2 Lvalue subroutines
d74e8afc 1029X<lvalue> X<subroutine, lvalue>
cd06dffe 1030
cd06dffe
GS
1031It is possible to return a modifiable value from a subroutine.
1032To do this, you have to declare the subroutine to return an lvalue.
1033
1034 my $val;
1035 sub canmod : lvalue {
4a904372 1036 $val; # or: return $val;
cd06dffe
GS
1037 }
1038 sub nomod {
1039 $val;
1040 }
1041
1042 canmod() = 5; # assigns to $val
1043 nomod() = 5; # ERROR
1044
1045The scalar/list context for the subroutine and for the right-hand
1046side of assignment is determined as if the subroutine call is replaced
b77865f5 1047by a scalar. For example, consider:
cd06dffe
GS
1048
1049 data(2,3) = get_data(3,4);
1050
1051Both subroutines here are called in a scalar context, while in:
1052
1053 (data(2,3)) = get_data(3,4);
1054
1055and in:
1056
1057 (data(2),data(3)) = get_data(3,4);
1058
1059all the subroutines are called in a list context.
1060
771cc755 1061Lvalue subroutines are convenient, but you have to keep in mind that,
b77865f5 1062when used with objects, they may violate encapsulation. A normal
771cc755 1063mutator can check the supplied argument before setting the attribute
b77865f5 1064it is protecting, an lvalue subroutine cannot. If you require any
771cc755
JV
1065special processing when storing and retrieving the values, consider
1066using the CPAN module Sentinel or something similar.
e6a32221 1067
ca40957e
FC
1068=head2 Lexical Subroutines
1069X<my sub> X<state sub> X<our sub> X<subroutine, lexical>
1070
ca40957e
FC
1071Beginning with Perl 5.18, you can declare a private subroutine with C<my>
1072or C<state>. As with state variables, the C<state> keyword is only
1073available under C<use feature 'state'> or C<use 5.010> or higher.
1074
06c4bad0
FC
1075Prior to Perl 5.26, lexical subroutines were deemed experimental and were
1076available only under the C<use feature 'lexical_subs'> pragma. They also
1077produced a warning unless the "experimental::lexical_subs" warnings
1078category was disabled.
1079
ca40957e
FC
1080These subroutines are only visible within the block in which they are
1081declared, and only after that declaration:
1082
06c4bad0
FC
1083 # Include these two lines if your code is intended to run under Perl
1084 # versions earlier than 5.26.
f1d34ca8 1085 no warnings "experimental::lexical_subs";
ca40957e
FC
1086 use feature 'lexical_subs';
1087
67bf5a37 1088 foo(); # calls the package/global subroutine
ca40957e 1089 state sub foo {
67bf5a37 1090 foo(); # also calls the package subroutine
ca40957e 1091 }
67bf5a37
LM
1092 foo(); # calls "state" sub
1093 my $ref = \&foo; # take a reference to "state" sub
ca40957e
FC
1094
1095 my sub bar { ... }
67bf5a37 1096 bar(); # calls "my" sub
ca40957e 1097
67bf5a37 1098You can't (directly) write a recursive lexical subroutine:
ca40957e 1099
67bf5a37
LM
1100 # WRONG
1101 my sub baz {
1102 baz();
ca40957e
FC
1103 }
1104
67bf5a37
LM
1105This example fails because C<baz()> refers to the package/global subroutine
1106C<baz>, not the lexical subroutine currently being defined.
1107
1108The solution is to use L<C<__SUB__>|perlfunc/__SUB__>:
1109
1110 my sub baz {
1111 __SUB__->(); # calls itself
1112 }
1113
1114It is possible to predeclare a lexical subroutine. The C<sub foo {...}>
1115subroutine definition syntax respects any previous C<my sub;> or C<state sub;>
1116declaration. Using this to define recursive subroutines is a bad idea,
1117however:
1118
1119 my sub baz; # predeclaration
1120 sub baz { # define the "my" sub
1121 baz(); # WRONG: calls itself, but leaks memory
1122 }
1123
1124Just like C<< my $f; $f = sub { $f->() } >>, this example leaks memory. The
1125name C<baz> is a reference to the subroutine, and the subroutine uses the name
1126C<baz>; they keep each other alive (see L<perlref/Circular References>).
1127
ca40957e
FC
1128=head3 C<state sub> vs C<my sub>
1129
1130What is the difference between "state" subs and "my" subs? Each time that
1131execution enters a block when "my" subs are declared, a new copy of each
1132sub is created. "State" subroutines persist from one execution of the
1133containing block to the next.
1134
1135So, in general, "state" subroutines are faster. But "my" subs are
1136necessary if you want to create closures:
1137
ca40957e
FC
1138 sub whatever {
1139 my $x = shift;
1140 my sub inner {
1141 ... do something with $x ...
1142 }
1143 inner();
1144 }
1145
1146In this example, a new C<$x> is created when C<whatever> is called, and
1147also a new C<inner>, which can see the new C<$x>. A "state" sub will only
1148see the C<$x> from the first call to C<whatever>.
1149
1150=head3 C<our> subroutines
1151
1152Like C<our $variable>, C<our sub> creates a lexical alias to the package
1153subroutine of the same name.
1154
1155The two main uses for this are to switch back to using the package sub
1156inside an inner scope:
1157
ca40957e
FC
1158 sub foo { ... }
1159
1160 sub bar {
1161 my sub foo { ... }
1162 {
1163 # need to use the outer foo here
1164 our sub foo;
1165 foo();
1166 }
1167 }
1168
1169and to make a subroutine visible to other packages in the same scope:
1170
1171 package MySneakyModule;
1172
ca40957e
FC
1173 our sub do_something { ... }
1174
1175 sub do_something_with_caller {
1176 package DB;
1177 () = caller 1; # sets @DB::args
1178 do_something(@args); # uses MySneakyModule::do_something
1179 }
1180
cb1a09d0 1181=head2 Passing Symbol Table Entries (typeglobs)
d74e8afc 1182X<typeglob> X<*>
cb1a09d0 1183
19799a22
GS
1184B<WARNING>: The mechanism described in this section was originally
1185the only way to simulate pass-by-reference in older versions of
1186Perl. While it still works fine in modern versions, the new reference
1187mechanism is generally easier to work with. See below.
a0d0e21e
LW
1188
1189Sometimes you don't want to pass the value of an array to a subroutine
1190but rather the name of it, so that the subroutine can modify the global
1191copy of it rather than working with a local copy. In perl you can
cb1a09d0 1192refer to all objects of a particular name by prefixing the name
5f05dabc 1193with a star: C<*foo>. This is often known as a "typeglob", because the
a0d0e21e
LW
1194star on the front can be thought of as a wildcard match for all the
1195funny prefix characters on variables and subroutines and such.
1196
55497cff 1197When evaluated, the typeglob produces a scalar value that represents
5f05dabc 1198all the objects of that name, including any filehandle, format, or
a0d0e21e 1199subroutine. When assigned to, it causes the name mentioned to refer to
19799a22 1200whatever C<*> value was assigned to it. Example:
a0d0e21e
LW
1201
1202 sub doubleary {
1203 local(*someary) = @_;
1204 foreach $elem (@someary) {
1205 $elem *= 2;
1206 }
1207 }
1208 doubleary(*foo);
1209 doubleary(*bar);
1210
19799a22 1211Scalars are already passed by reference, so you can modify
a0d0e21e 1212scalar arguments without using this mechanism by referring explicitly
1fef88e7 1213to C<$_[0]> etc. You can modify all the elements of an array by passing
f86cebdf
GS
1214all the elements as scalars, but you have to use the C<*> mechanism (or
1215the equivalent reference mechanism) to C<push>, C<pop>, or change the size of
a0d0e21e
LW
1216an array. It will certainly be faster to pass the typeglob (or reference).
1217
1218Even if you don't want to modify an array, this mechanism is useful for
5f05dabc 1219passing multiple arrays in a single LIST, because normally the LIST
a0d0e21e 1220mechanism will merge all the array values so that you can't extract out
55497cff 1221the individual arrays. For more on typeglobs, see
2ae324a7 1222L<perldata/"Typeglobs and Filehandles">.
cb1a09d0 1223
5a964f20 1224=head2 When to Still Use local()
d74e8afc 1225X<local> X<variable, local>
5a964f20 1226
19799a22
GS
1227Despite the existence of C<my>, there are still three places where the
1228C<local> operator still shines. In fact, in these three places, you
5a964f20
TC
1229I<must> use C<local> instead of C<my>.
1230
13a2d996 1231=over 4
5a964f20 1232
551e1d92
RB
1233=item 1.
1234
1235You need to give a global variable a temporary value, especially $_.
5a964f20 1236
f86cebdf
GS
1237The global variables, like C<@ARGV> or the punctuation variables, must be
1238C<local>ized with C<local()>. This block reads in F</etc/motd>, and splits
5a964f20 1239it up into chunks separated by lines of equal signs, which are placed
f86cebdf 1240in C<@Fields>.
5a964f20
TC
1241
1242 {
1243 local @ARGV = ("/etc/motd");
1244 local $/ = undef;
1245 local $_ = <>;
1246 @Fields = split /^\s*=+\s*$/;
1247 }
1248
19799a22 1249It particular, it's important to C<local>ize $_ in any routine that assigns
5a964f20
TC
1250to it. Look out for implicit assignments in C<while> conditionals.
1251
551e1d92
RB
1252=item 2.
1253
1254You need to create a local file or directory handle or a local function.
5a964f20 1255
09bef843
SB
1256A function that needs a filehandle of its own must use
1257C<local()> on a complete typeglob. This can be used to create new symbol
5a964f20
TC
1258table entries:
1259
1260 sub ioqueue {
1261 local (*READER, *WRITER); # not my!
17b63f68 1262 pipe (READER, WRITER) or die "pipe: $!";
5a964f20
TC
1263 return (*READER, *WRITER);
1264 }
1265 ($head, $tail) = ioqueue();
1266
1267See the Symbol module for a way to create anonymous symbol table
1268entries.
1269
1270Because assignment of a reference to a typeglob creates an alias, this
1271can be used to create what is effectively a local function, or at least,
1272a local alias.
1273
1274 {
4a46e268 1275 local *grow = \&shrink; # only until this block exits
555bd962
BG
1276 grow(); # really calls shrink()
1277 move(); # if move() grow()s, it shrink()s too
5a964f20 1278 }
555bd962 1279 grow(); # get the real grow() again
5a964f20
TC
1280
1281See L<perlref/"Function Templates"> for more about manipulating
1282functions by name in this way.
1283
551e1d92
RB
1284=item 3.
1285
1286You want to temporarily change just one element of an array or hash.
5a964f20 1287
f86cebdf 1288You can C<local>ize just one element of an aggregate. Usually this
5a964f20
TC
1289is done on dynamics:
1290
1291 {
1292 local $SIG{INT} = 'IGNORE';
1293 funct(); # uninterruptible
1294 }
1295 # interruptibility automatically restored here
1296
9d42615f 1297But it also works on lexically declared aggregates.
5a964f20
TC
1298
1299=back
1300
cb1a09d0 1301=head2 Pass by Reference
d74e8afc 1302X<pass by reference> X<pass-by-reference> X<reference>
cb1a09d0 1303
55497cff
PP
1304If you want to pass more than one array or hash into a function--or
1305return them from it--and have them maintain their integrity, then
1306you're going to have to use an explicit pass-by-reference. Before you
1307do that, you need to understand references as detailed in L<perlref>.
c07a80fd 1308This section may not make much sense to you otherwise.
cb1a09d0 1309
19799a22
GS
1310Here are a few simple examples. First, let's pass in several arrays
1311to a function and have it C<pop> all of then, returning a new list
1312of all their former last elements:
cb1a09d0
AD
1313
1314 @tailings = popmany ( \@a, \@b, \@c, \@d );
1315
1316 sub popmany {
1317 my $aref;
8b7906d1 1318 my @retlist;
cb1a09d0
AD
1319 foreach $aref ( @_ ) {
1320 push @retlist, pop @$aref;
54310121 1321 }
cb1a09d0 1322 return @retlist;
54310121 1323 }
cb1a09d0 1324
54310121 1325Here's how you might write a function that returns a
cb1a09d0
AD
1326list of keys occurring in all the hashes passed to it:
1327
54310121 1328 @common = inter( \%foo, \%bar, \%joe );
cb1a09d0
AD
1329 sub inter {
1330 my ($k, $href, %seen); # locals
1331 foreach $href (@_) {
1332 while ( $k = each %$href ) {
1333 $seen{$k}++;
54310121
PP
1334 }
1335 }
cb1a09d0 1336 return grep { $seen{$_} == @_ } keys %seen;
54310121 1337 }
cb1a09d0 1338
5f05dabc 1339So far, we're using just the normal list return mechanism.
54310121
PP
1340What happens if you want to pass or return a hash? Well,
1341if you're using only one of them, or you don't mind them
cb1a09d0 1342concatenating, then the normal calling convention is ok, although
54310121 1343a little expensive.
cb1a09d0
AD
1344
1345Where people get into trouble is here:
1346
1347 (@a, @b) = func(@c, @d);
1348or
1349 (%a, %b) = func(%c, %d);
1350
19799a22
GS
1351That syntax simply won't work. It sets just C<@a> or C<%a> and
1352clears the C<@b> or C<%b>. Plus the function didn't get passed
1353into two separate arrays or hashes: it got one long list in C<@_>,
1354as always.
cb1a09d0
AD
1355
1356If you can arrange for everyone to deal with this through references, it's
1357cleaner code, although not so nice to look at. Here's a function that
1358takes two array references as arguments, returning the two array elements
1359in order of how many elements they have in them:
1360
1361 ($aref, $bref) = func(\@c, \@d);
1362 print "@$aref has more than @$bref\n";
1363 sub func {
1364 my ($cref, $dref) = @_;
1365 if (@$cref > @$dref) {
1366 return ($cref, $dref);
1367 } else {
c07a80fd 1368 return ($dref, $cref);
54310121
PP
1369 }
1370 }
cb1a09d0
AD
1371
1372It turns out that you can actually do this also:
1373
1374 (*a, *b) = func(\@c, \@d);
1375 print "@a has more than @b\n";
1376 sub func {
1377 local (*c, *d) = @_;
1378 if (@c > @d) {
1379 return (\@c, \@d);
1380 } else {
1381 return (\@d, \@c);
54310121
PP
1382 }
1383 }
cb1a09d0
AD
1384
1385Here we're using the typeglobs to do symbol table aliasing. It's
19799a22 1386a tad subtle, though, and also won't work if you're using C<my>
09bef843 1387variables, because only globals (even in disguise as C<local>s)
19799a22 1388are in the symbol table.
5f05dabc
PP
1389
1390If you're passing around filehandles, you could usually just use the bare
19799a22
GS
1391typeglob, like C<*STDOUT>, but typeglobs references work, too.
1392For example:
5f05dabc
PP
1393
1394 splutter(\*STDOUT);
1395 sub splutter {
1396 my $fh = shift;
1397 print $fh "her um well a hmmm\n";
1398 }
1399
1400 $rec = get_rec(\*STDIN);
1401 sub get_rec {
1402 my $fh = shift;
1403 return scalar <$fh>;
1404 }
1405
19799a22
GS
1406If you're planning on generating new filehandles, you could do this.
1407Notice to pass back just the bare *FH, not its reference.
5f05dabc
PP
1408
1409 sub openit {
19799a22 1410 my $path = shift;
5f05dabc 1411 local *FH;
e05a3a1e 1412 return open (FH, $path) ? *FH : undef;
54310121 1413 }
5f05dabc 1414
cb1a09d0 1415=head2 Prototypes
d74e8afc 1416X<prototype> X<subroutine, prototype>
cb1a09d0 1417
19799a22 1418Perl supports a very limited kind of compile-time argument checking
eedb00fa
PM
1419using function prototyping. This can be declared in either the PROTO
1420section or with a L<prototype attribute|attributes/Built-in Attributes>.
30d9c59b 1421If you declare either of
cb1a09d0 1422
26230909
AC
1423 sub mypush (\@@)
1424 sub mypush :prototype(\@@)
30d9c59b
Z
1425
1426then C<mypush()> takes arguments exactly like C<push()> does.
1427
1428If subroutine signatures are enabled (see L</Signatures>), then
1429the shorter PROTO syntax is unavailable, because it would clash with
1430signatures. In that case, a prototype can only be declared in the form
1431of an attribute.
cb1a09d0 1432
30d9c59b 1433The
19799a22
GS
1434function declaration must be visible at compile time. The prototype
1435affects only interpretation of new-style calls to the function,
1436where new-style is defined as not using the C<&> character. In
1437other words, if you call it like a built-in function, then it behaves
1438like a built-in function. If you call it like an old-fashioned
1439subroutine, then it behaves like an old-fashioned subroutine. It
1440naturally falls out from this rule that prototypes have no influence
1441on subroutine references like C<\&foo> or on indirect subroutine
c47ff5f1 1442calls like C<&{$subref}> or C<< $subref->() >>.
c07a80fd
PP
1443
1444Method calls are not influenced by prototypes either, because the
19799a22
GS
1445function to be called is indeterminate at compile time, since
1446the exact code called depends on inheritance.
cb1a09d0 1447
19799a22
GS
1448Because the intent of this feature is primarily to let you define
1449subroutines that work like built-in functions, here are prototypes
1450for some other functions that parse almost exactly like the
1451corresponding built-in.
cb1a09d0 1452
555bd962
BG
1453 Declared as Called as
1454
1455 sub mylink ($$) mylink $old, $new
1456 sub myvec ($$$) myvec $var, $offset, 1
1457 sub myindex ($$;$) myindex &getstring, "substr"
1458 sub mysyswrite ($$$;$) mysyswrite $buf, 0, length($buf) - $off, $off
1459 sub myreverse (@) myreverse $a, $b, $c
1460 sub myjoin ($@) myjoin ":", $a, $b, $c
26230909
AC
1461 sub mypop (\@) mypop @array
1462 sub mysplice (\@$$@) mysplice @array, 0, 2, @pushme
1463 sub mykeys (\[%@]) mykeys %{$hashref}
555bd962
BG
1464 sub myopen (*;$) myopen HANDLE, $name
1465 sub mypipe (**) mypipe READHANDLE, WRITEHANDLE
1466 sub mygrep (&@) mygrep { /foo/ } $a, $b, $c
1467 sub myrand (;$) myrand 42
1468 sub mytime () mytime
cb1a09d0 1469
c07a80fd 1470Any backslashed prototype character represents an actual argument
ae7a3cfa 1471that must start with that character (optionally preceded by C<my>,
b91b7d1a
FC
1472C<our> or C<local>), with the exception of C<$>, which will
1473accept any scalar lvalue expression, such as C<$foo = 7> or
b77865f5 1474C<< my_function()->[0] >>. The value passed as part of C<@_> will be a
ae7a3cfa
FC
1475reference to the actual argument given in the subroutine call,
1476obtained by applying C<\> to that argument.
c07a80fd 1477
c035a075 1478You can use the C<\[]> backslash group notation to specify more than one
b77865f5 1479allowed argument type. For example:
5b794e05
JH
1480
1481 sub myref (\[$@%&*])
1482
1483will allow calling myref() as
1484
1485 myref $var
1486 myref @array
1487 myref %hash
1488 myref &sub
1489 myref *glob
1490
1491and the first argument of myref() will be a reference to
1492a scalar, an array, a hash, a code, or a glob.
1493
c07a80fd 1494Unbackslashed prototype characters have special meanings. Any
19799a22 1495unbackslashed C<@> or C<%> eats all remaining arguments, and forces
f86cebdf
GS
1496list context. An argument represented by C<$> forces scalar context. An
1497C<&> requires an anonymous subroutine, which, if passed as the first
0df79f0c
GS
1498argument, does not require the C<sub> keyword or a subsequent comma.
1499
1500A C<*> allows the subroutine to accept a bareword, constant, scalar expression,
648ca4f7
GS
1501typeglob, or a reference to a typeglob in that slot. The value will be
1502available to the subroutine either as a simple scalar, or (in the latter
0df79f0c
GS
1503two cases) as a reference to the typeglob. If you wish to always convert
1504such arguments to a typeglob reference, use Symbol::qualify_to_ref() as
1505follows:
1506
1507 use Symbol 'qualify_to_ref';
1508
1509 sub foo (*) {
1510 my $fh = qualify_to_ref(shift, caller);
1511 ...
1512 }
c07a80fd 1513
c035a075
DG
1514The C<+> prototype is a special alternative to C<$> that will act like
1515C<\[@%]> when given a literal array or hash variable, but will otherwise
1516force scalar context on the argument. This is useful for functions which
1517should accept either a literal array or an array reference as the argument:
1518
cba5a3b0 1519 sub mypush (+@) {
c035a075
DG
1520 my $aref = shift;
1521 die "Not an array or arrayref" unless ref $aref eq 'ARRAY';
1522 push @$aref, @_;
1523 }
1524
1525When using the C<+> prototype, your function must check that the argument
1526is of an acceptable type.
1527
859a4967 1528A semicolon (C<;>) separates mandatory arguments from optional arguments.
19799a22 1529It is redundant before C<@> or C<%>, which gobble up everything else.
cb1a09d0 1530
34daab0f
RGS
1531As the last character of a prototype, or just before a semicolon, a C<@>
1532or a C<%>, you can use C<_> in place of C<$>: if this argument is not
1533provided, C<$_> will be used instead.
859a4967 1534
19799a22
GS
1535Note how the last three examples in the table above are treated
1536specially by the parser. C<mygrep()> is parsed as a true list
1537operator, C<myrand()> is parsed as a true unary operator with unary
1538precedence the same as C<rand()>, and C<mytime()> is truly without
1539arguments, just like C<time()>. That is, if you say
cb1a09d0
AD
1540
1541 mytime +2;
1542
f86cebdf 1543you'll get C<mytime() + 2>, not C<mytime(2)>, which is how it would be parsed
3a8944db
FC
1544without a prototype. If you want to force a unary function to have the
1545same precedence as a list operator, add C<;> to the end of the prototype:
1546
1547 sub mygetprotobynumber($;);
1548 mygetprotobynumber $a > $b; # parsed as mygetprotobynumber($a > $b)
cb1a09d0 1549
19799a22
GS
1550The interesting thing about C<&> is that you can generate new syntax with it,
1551provided it's in the initial position:
d74e8afc 1552X<&>
cb1a09d0 1553
6d28dffb 1554 sub try (&@) {
cb1a09d0
AD
1555 my($try,$catch) = @_;
1556 eval { &$try };
1557 if ($@) {
1558 local $_ = $@;
1559 &$catch;
1560 }
1561 }
55497cff 1562 sub catch (&) { $_[0] }
cb1a09d0
AD
1563
1564 try {
1565 die "phooey";
1566 } catch {
1567 /phooey/ and print "unphooey\n";
1568 };
1569
f86cebdf 1570That prints C<"unphooey">. (Yes, there are still unresolved
19799a22 1571issues having to do with visibility of C<@_>. I'm ignoring that
f86cebdf 1572question for the moment. (But note that if we make C<@_> lexically
cb1a09d0 1573scoped, those anonymous subroutines can act like closures... (Gee,
5f05dabc 1574is this sounding a little Lispish? (Never mind.))))
cb1a09d0 1575
19799a22 1576And here's a reimplementation of the Perl C<grep> operator:
d74e8afc 1577X<grep>
cb1a09d0
AD
1578
1579 sub mygrep (&@) {
1580 my $code = shift;
1581 my @result;
1582 foreach $_ (@_) {
6e47f808 1583 push(@result, $_) if &$code;
cb1a09d0
AD
1584 }
1585 @result;
1586 }
a0d0e21e 1587
cb1a09d0
AD
1588Some folks would prefer full alphanumeric prototypes. Alphanumerics have
1589been intentionally left out of prototypes for the express purpose of
1590someday in the future adding named, formal parameters. The current
1591mechanism's main goal is to let module writers provide better diagnostics
1592for module users. Larry feels the notation quite understandable to Perl
1593programmers, and that it will not intrude greatly upon the meat of the
1594module, nor make it harder to read. The line noise is visually
1595encapsulated into a small pill that's easy to swallow.
1596
420cdfc1
ST
1597If you try to use an alphanumeric sequence in a prototype you will
1598generate an optional warning - "Illegal character in prototype...".
1599Unfortunately earlier versions of Perl allowed the prototype to be
1600used as long as its prefix was a valid prototype. The warning may be
1601upgraded to a fatal error in a future version of Perl once the
1602majority of offending code is fixed.
1603
cb1a09d0
AD
1604It's probably best to prototype new functions, not retrofit prototyping
1605into older ones. That's because you must be especially careful about
1606silent impositions of differing list versus scalar contexts. For example,
1607if you decide that a function should take just one parameter, like this:
1608
1609 sub func ($) {
1610 my $n = shift;
1611 print "you gave me $n\n";
54310121 1612 }
cb1a09d0
AD
1613
1614and someone has been calling it with an array or expression
1615returning a list:
1616
1617 func(@foo);
f2606479 1618 func( $text =~ /\w+/g );
cb1a09d0 1619
19799a22 1620Then you've just supplied an automatic C<scalar> in front of their
f86cebdf 1621argument, which can be more than a bit surprising. The old C<@foo>
cb1a09d0 1622which used to hold one thing doesn't get passed in. Instead,
19799a22 1623C<func()> now gets passed in a C<1>; that is, the number of elements
f2606479
LM
1624in C<@foo>. And the C<m//g> gets called in scalar context so instead of a
1625list of words it returns a boolean result and advances C<pos($text)>. Ouch!
cb1a09d0 1626
eb40d2ca
PM
1627If a sub has both a PROTO and a BLOCK, the prototype is not applied
1628until after the BLOCK is completely defined. This means that a recursive
1629function with a prototype has to be predeclared for the prototype to take
1630effect, like so:
1631
1632 sub foo($$);
1633 sub foo($$) {
1634 foo 1, 2;
1635 }
1636
5f05dabc 1637This is all very powerful, of course, and should be used only in moderation
54310121 1638to make the world a better place.
44a8e56a
PP
1639
1640=head2 Constant Functions
d74e8afc 1641X<constant>
44a8e56a
PP
1642
1643Functions with a prototype of C<()> are potential candidates for
19799a22
GS
1644inlining. If the result after optimization and constant folding
1645is either a constant or a lexically-scoped scalar which has no other
54310121 1646references, then it will be used in place of function calls made
19799a22
GS
1647without C<&>. Calls made using C<&> are never inlined. (See
1648F<constant.pm> for an easy way to declare most constants.)
44a8e56a 1649
5a964f20 1650The following functions would all be inlined:
44a8e56a 1651
699e6cd4
TP
1652 sub pi () { 3.14159 } # Not exact, but close.
1653 sub PI () { 4 * atan2 1, 1 } # As good as it gets,
1654 # and it's inlined, too!
44a8e56a
PP
1655 sub ST_DEV () { 0 }
1656 sub ST_INO () { 1 }
1657
1658 sub FLAG_FOO () { 1 << 8 }
1659 sub FLAG_BAR () { 1 << 9 }
1660 sub FLAG_MASK () { FLAG_FOO | FLAG_BAR }
54310121
PP
1661
1662 sub OPT_BAZ () { not (0x1B58 & FLAG_MASK) }
88267271 1663
1664 sub N () { int(OPT_BAZ) / 3 }
1665
1666 sub FOO_SET () { 1 if FLAG_MASK & FLAG_FOO }
d3c633ba 1667 sub FOO_SET2 () { if (FLAG_MASK & FLAG_FOO) { 1 } }
88267271 1668
d3c633ba
FC
1669(Be aware that the last example was not always inlined in Perl 5.20 and
1670earlier, which did not behave consistently with subroutines containing
1671inner scopes.) You can countermand inlining by using an explicit
1672C<return>:
88267271 1673
1674 sub baz_val () {
44a8e56a
PP
1675 if (OPT_BAZ) {
1676 return 23;
1677 }
1678 else {
1679 return 42;
1680 }
1681 }
d3c633ba 1682 sub bonk_val () { return 12345 }
cb1a09d0 1683
fe39f0d5
AB
1684As alluded to earlier you can also declare inlined subs dynamically at
1685BEGIN time if their body consists of a lexically-scoped scalar which
b77865f5 1686has no other references. Only the first example here will be inlined:
fe39f0d5
AB
1687
1688 BEGIN {
1689 my $var = 1;
1690 no strict 'refs';
1691 *INLINED = sub () { $var };
1692 }
1693
1694 BEGIN {
1695 my $var = 1;
1696 my $ref = \$var;
1697 no strict 'refs';
1698 *NOT_INLINED = sub () { $var };
1699 }
1700
1701A not so obvious caveat with this (see [RT #79908]) is that the
1702variable will be immediately inlined, and will stop behaving like a
1703normal lexical variable, e.g. this will print C<79907>, not C<79908>:
1704
1705 BEGIN {
1706 my $x = 79907;
1707 *RT_79908 = sub () { $x };
1708 $x++;
1709 }
1710 print RT_79908(); # prints 79907
1711
d3c633ba
FC
1712As of Perl 5.22, this buggy behavior, while preserved for backward
1713compatibility, is detected and emits a deprecation warning. If you want
1714the subroutine to be inlined (with no warning), make sure the variable is
1715not used in a context where it could be modified aside from where it is
1716declared.
1717
1718 # Fine, no warning
1719 BEGIN {
1720 my $x = 54321;
1721 *INLINED = sub () { $x };
1722 }
1723 # Warns. Future Perl versions will stop inlining it.
1724 BEGIN {
1725 my $x;
1726 $x = 54321;
1727 *ALSO_INLINED = sub () { $x };
1728 }
1729
99734069
FC
1730Perl 5.22 also introduces the experimental "const" attribute as an
1731alternative. (Disable the "experimental::const_attr" warnings if you want
1732to use it.) When applied to an anonymous subroutine, it forces the sub to
1733be called when the C<sub> expression is evaluated. The return value is
1734captured and turned into a constant subroutine:
1735
1736 my $x = 54321;
1737 *INLINED = sub : const { $x };
1738 $x++;
1739
1740The return value of C<INLINED> in this example will always be 54321,
1741regardless of later modifications to $x. You can also put any arbitrary
1742code inside the sub, at it will be executed immediately and its return
1743value captured the same way.
1744
fe39f0d5
AB
1745If you really want a subroutine with a C<()> prototype that returns a
1746lexical variable you can easily force it to not be inlined by adding
1747an explicit C<return>:
1748
1749 BEGIN {
1750 my $x = 79907;
1751 *RT_79908 = sub () { return $x };
1752 $x++;
1753 }
1754 print RT_79908(); # prints 79908
1755
1756The easiest way to tell if a subroutine was inlined is by using
d3c633ba 1757L<B::Deparse>. Consider this example of two subroutines returning
fe39f0d5
AB
1758C<1>, one with a C<()> prototype causing it to be inlined, and one
1759without (with deparse output truncated for clarity):
1760
cb07e2f2
KW
1761 $ perl -MO=Deparse -le 'sub ONE { 1 } if (ONE) { print ONE if ONE }'
1762 sub ONE {
1763 1;
1764 }
1765 if (ONE ) {
1766 print ONE() if ONE ;
1767 }
1768 $ perl -MO=Deparse -le 'sub ONE () { 1 } if (ONE) { print ONE if ONE }'
1769 sub ONE () { 1 }
1770 do {
1771 print 1
1772 };
fe39f0d5
AB
1773
1774If you redefine a subroutine that was eligible for inlining, you'll
b77865f5 1775get a warning by default. You can use this warning to tell whether or
fe39f0d5
AB
1776not a particular subroutine is considered inlinable, since it's
1777different than the warning for overriding non-inlined subroutines:
1778
1779 $ perl -e 'sub one () {1} sub one () {2}'
1780 Constant subroutine one redefined at -e line 1.
1781 $ perl -we 'sub one {1} sub one {2}'
1782 Subroutine one redefined at -e line 1.
1783
1784The warning is considered severe enough not to be affected by the
1785B<-w> switch (or its absence) because previously compiled invocations
1786of the function will still be using the old value of the function. If
1787you need to be able to redefine the subroutine, you need to ensure
1788that it isn't inlined, either by dropping the C<()> prototype (which
1789changes calling semantics, so beware) or by thwarting the inlining
d3c633ba
FC
1790mechanism in some other way, e.g. by adding an explicit C<return>, as
1791mentioned above:
fe39f0d5
AB
1792
1793 sub not_inlined () { return 23 }
4cee8e80 1794
19799a22 1795=head2 Overriding Built-in Functions
d74e8afc 1796X<built-in> X<override> X<CORE> X<CORE::GLOBAL>
a0d0e21e 1797
19799a22 1798Many built-in functions may be overridden, though this should be tried
5f05dabc 1799only occasionally and for good reason. Typically this might be
19799a22 1800done by a package attempting to emulate missing built-in functionality
a0d0e21e
LW
1801on a non-Unix system.
1802
163e3a99
JP
1803Overriding may be done only by importing the name from a module at
1804compile time--ordinary predeclaration isn't good enough. However, the
19799a22
GS
1805C<use subs> pragma lets you, in effect, predeclare subs
1806via the import syntax, and these names may then override built-in ones:
a0d0e21e
LW
1807
1808 use subs 'chdir', 'chroot', 'chmod', 'chown';
1809 chdir $somewhere;
1810 sub chdir { ... }
1811
19799a22
GS
1812To unambiguously refer to the built-in form, precede the
1813built-in name with the special package qualifier C<CORE::>. For example,
1814saying C<CORE::open()> always refers to the built-in C<open()>, even
fb73857a 1815if the current package has imported some other subroutine called
19799a22 1816C<&open()> from elsewhere. Even though it looks like a regular
4aaa4757
FC
1817function call, it isn't: the CORE:: prefix in that case is part of Perl's
1818syntax, and works for any keyword, regardless of what is in the CORE
1819package. Taking a reference to it, that is, C<\&CORE::open>, only works
1820for some keywords. See L<CORE>.
fb73857a 1821
19799a22
GS
1822Library modules should not in general export built-in names like C<open>
1823or C<chdir> as part of their default C<@EXPORT> list, because these may
a0d0e21e 1824sneak into someone else's namespace and change the semantics unexpectedly.
19799a22 1825Instead, if the module adds that name to C<@EXPORT_OK>, then it's
a0d0e21e
LW
1826possible for a user to import the name explicitly, but not implicitly.
1827That is, they could say
1828
1829 use Module 'open';
1830
19799a22 1831and it would import the C<open> override. But if they said
a0d0e21e
LW
1832
1833 use Module;
1834
19799a22 1835they would get the default imports without overrides.
a0d0e21e 1836
19799a22 1837The foregoing mechanism for overriding built-in is restricted, quite
95d94a4f 1838deliberately, to the package that requests the import. There is a second
19799a22 1839method that is sometimes applicable when you wish to override a built-in
95d94a4f
GS
1840everywhere, without regard to namespace boundaries. This is achieved by
1841importing a sub into the special namespace C<CORE::GLOBAL::>. Here is an
1842example that quite brazenly replaces the C<glob> operator with something
1843that understands regular expressions.
1844
1845 package REGlob;
1846 require Exporter;
1847 @ISA = 'Exporter';
1848 @EXPORT_OK = 'glob';
1849
1850 sub import {
1851 my $pkg = shift;
1852 return unless @_;
1853 my $sym = shift;
1854 my $where = ($sym =~ s/^GLOBAL_// ? 'CORE::GLOBAL' : caller(0));
1855 $pkg->export($where, $sym, @_);
1856 }
1857
1858 sub glob {
1859 my $pat = shift;
1860 my @got;
7b815c67
RGS
1861 if (opendir my $d, '.') {
1862 @got = grep /$pat/, readdir $d;
1863 closedir $d;
19799a22
GS
1864 }
1865 return @got;
95d94a4f
GS
1866 }
1867 1;
1868
1869And here's how it could be (ab)used:
1870
1871 #use REGlob 'GLOBAL_glob'; # override glob() in ALL namespaces
1872 package Foo;
1873 use REGlob 'glob'; # override glob() in Foo:: only
1874 print for <^[a-z_]+\.pm\$>; # show all pragmatic modules
1875
19799a22 1876The initial comment shows a contrived, even dangerous example.
95d94a4f 1877By overriding C<glob> globally, you would be forcing the new (and
19799a22 1878subversive) behavior for the C<glob> operator for I<every> namespace,
95d94a4f
GS
1879without the complete cognizance or cooperation of the modules that own
1880those namespaces. Naturally, this should be done with extreme caution--if
1881it must be done at all.
1882
1883The C<REGlob> example above does not implement all the support needed to
19799a22 1884cleanly override perl's C<glob> operator. The built-in C<glob> has
95d94a4f 1885different behaviors depending on whether it appears in a scalar or list
19799a22 1886context, but our C<REGlob> doesn't. Indeed, many perl built-in have such
95d94a4f
GS
1887context sensitive behaviors, and these must be adequately supported by
1888a properly written override. For a fully functional example of overriding
1889C<glob>, study the implementation of C<File::DosGlob> in the standard
1890library.
1891
77bc9082
RGS
1892When you override a built-in, your replacement should be consistent (if
1893possible) with the built-in native syntax. You can achieve this by using
1894a suitable prototype. To get the prototype of an overridable built-in,
1895use the C<prototype> function with an argument of C<"CORE::builtin_name">
1896(see L<perlfunc/prototype>).
1897
1898Note however that some built-ins can't have their syntax expressed by a
1899prototype (such as C<system> or C<chomp>). If you override them you won't
1900be able to fully mimic their original syntax.
1901
fe854a6f 1902The built-ins C<do>, C<require> and C<glob> can also be overridden, but due
77bc9082
RGS
1903to special magic, their original syntax is preserved, and you don't have
1904to define a prototype for their replacements. (You can't override the
1905C<do BLOCK> syntax, though).
1906
1907C<require> has special additional dark magic: if you invoke your
1908C<require> replacement as C<require Foo::Bar>, it will actually receive
1909the argument C<"Foo/Bar.pm"> in @_. See L<perlfunc/require>.
1910
1911And, as you'll have noticed from the previous example, if you override
593b9c14 1912C<glob>, the C<< <*> >> glob operator is overridden as well.
77bc9082 1913
9b3023bc 1914In a similar fashion, overriding the C<readline> function also overrides
b77865f5 1915the equivalent I/O operator C<< <FILEHANDLE> >>. Also, overriding
e3f73d4e 1916C<readpipe> also overrides the operators C<``> and C<qx//>.
9b3023bc 1917
fe854a6f 1918Finally, some built-ins (e.g. C<exists> or C<grep>) can't be overridden.
77bc9082 1919
a0d0e21e 1920=head2 Autoloading
d74e8afc 1921X<autoloading> X<AUTOLOAD>
a0d0e21e 1922
19799a22
GS
1923If you call a subroutine that is undefined, you would ordinarily
1924get an immediate, fatal error complaining that the subroutine doesn't
1925exist. (Likewise for subroutines being used as methods, when the
1926method doesn't exist in any base class of the class's package.)
1927However, if an C<AUTOLOAD> subroutine is defined in the package or
1928packages used to locate the original subroutine, then that
1929C<AUTOLOAD> subroutine is called with the arguments that would have
1930been passed to the original subroutine. The fully qualified name
1931of the original subroutine magically appears in the global $AUTOLOAD
1932variable of the same package as the C<AUTOLOAD> routine. The name
1933is not passed as an ordinary argument because, er, well, just
593b9c14 1934because, that's why. (As an exception, a method call to a nonexistent
80ee23cd 1935C<import> or C<unimport> method is just skipped instead. Also, if
5b36e945
FC
1936the AUTOLOAD subroutine is an XSUB, there are other ways to retrieve the
1937subroutine name. See L<perlguts/Autoloading with XSUBs> for details.)
80ee23cd 1938
19799a22
GS
1939
1940Many C<AUTOLOAD> routines load in a definition for the requested
1941subroutine using eval(), then execute that subroutine using a special
1942form of goto() that erases the stack frame of the C<AUTOLOAD> routine
1943without a trace. (See the source to the standard module documented
1944in L<AutoLoader>, for example.) But an C<AUTOLOAD> routine can
1945also just emulate the routine and never define it. For example,
1946let's pretend that a function that wasn't defined should just invoke
1947C<system> with those arguments. All you'd do is:
cb1a09d0
AD
1948
1949 sub AUTOLOAD {
1950 my $program = $AUTOLOAD;
1951 $program =~ s/.*:://;
1952 system($program, @_);
54310121 1953 }
cb1a09d0 1954 date();
6d28dffb 1955 who('am', 'i');
cb1a09d0
AD
1956 ls('-l');
1957
19799a22
GS
1958In fact, if you predeclare functions you want to call that way, you don't
1959even need parentheses:
cb1a09d0
AD
1960
1961 use subs qw(date who ls);
1962 date;
1963 who "am", "i";
593b9c14 1964 ls '-l';
cb1a09d0 1965
13058d67 1966A more complete example of this is the Shell module on CPAN, which
19799a22 1967can treat undefined subroutine calls as calls to external programs.
a0d0e21e 1968
19799a22
GS
1969Mechanisms are available to help modules writers split their modules
1970into autoloadable files. See the standard AutoLoader module
6d28dffb
PP
1971described in L<AutoLoader> and in L<AutoSplit>, the standard
1972SelfLoader modules in L<SelfLoader>, and the document on adding C
19799a22 1973functions to Perl code in L<perlxs>.
cb1a09d0 1974
09bef843 1975=head2 Subroutine Attributes
d74e8afc 1976X<attribute> X<subroutine, attribute> X<attrs>
09bef843
SB
1977
1978A subroutine declaration or definition may have a list of attributes
1979associated with it. If such an attribute list is present, it is
0120eecf 1980broken up at space or colon boundaries and treated as though a
09bef843
SB
1981C<use attributes> had been seen. See L<attributes> for details
1982about what attributes are currently supported.
1983Unlike the limitation with the obsolescent C<use attrs>, the
1984C<sub : ATTRLIST> syntax works to associate the attributes with
1985a pre-declaration, and not just with a subroutine definition.
1986
1987The attributes must be valid as simple identifier names (without any
1988punctuation other than the '_' character). They may have a parameter
1989list appended, which is only checked for whether its parentheses ('(',')')
1990nest properly.
1991
1992Examples of valid syntax (even though the attributes are unknown):
1993
4358a253
SS
1994 sub fnord (&\%) : switch(10,foo(7,3)) : expensive;
1995 sub plugh () : Ugly('\(") :Bad;
09bef843
SB
1996 sub xyzzy : _5x5 { ... }
1997
1998Examples of invalid syntax:
1999
4358a253
SS
2000 sub fnord : switch(10,foo(); # ()-string not balanced
2001 sub snoid : Ugly('('); # ()-string not balanced
2002 sub xyzzy : 5x5; # "5x5" not a valid identifier
2003 sub plugh : Y2::north; # "Y2::north" not a simple identifier
2004 sub snurt : foo + bar; # "+" not a colon or space
09bef843
SB
2005
2006The attribute list is passed as a list of constant strings to the code
2007which associates them with the subroutine. In particular, the second example
2008of valid syntax above currently looks like this in terms of how it's
2009parsed and invoked:
2010
2011 use attributes __PACKAGE__, \&plugh, q[Ugly('\(")], 'Bad';
2012
2013For further details on attribute lists and their manipulation,
a0ae32d3 2014see L<attributes> and L<Attribute::Handlers>.
09bef843 2015
cb1a09d0 2016=head1 SEE ALSO
a0d0e21e 2017
19799a22
GS
2018See L<perlref/"Function Templates"> for more about references and closures.
2019See L<perlxs> if you'd like to learn about calling C subroutines from Perl.
a2293a43 2020See L<perlembed> if you'd like to learn about calling Perl subroutines from C.
19799a22
GS
2021See L<perlmod> to learn about bundling up your functions in separate files.
2022See L<perlmodlib> to learn what library modules come standard on your system.
82e1c0d9 2023See L<perlootut> to learn how to make object method calls.